SEXUAL SEGREGATION IN DESERT BIGHORN SHEEP(OVIS CANADENSIS MEXICANA)

by

MICHAEL S MOORING13) THOMAS A FITZPATRICK1)JILL E BENJAMIN1) IAN C FRASER1) TARA T NISHIHIRA1)

DOMINIC D REISIG1) and ERIC M ROMINGER24)

(1 Department of Biology Point Loma Nazarene University San Diego CA USA 2 NewMexico Department of Game and Fish Santa Fe NM USA)

(Acc 15-I-2003)

Summary

One or several factors could explain sexual segregation in which males and females ofpolygynoussexually dimorphic species form separateherds during most of the year Bighornsheep (Ovis canadensis) are polygynous ungulates that exhibit extreme sexual dimorphismand segregate into ram and ewe herds outside of the rutting season Four major hypothesesfor sexual segregation were tested in a population of desert bighorn (O c mexicana) atthe Red Rock Wildlife Area New Mexico from 1999-2001 We collected data on the size

3) Corresponding author Department of Biology Point Loma Nazarene University 3900Lomaland Drive San Diego CA 92106 USA e-mail mikemooringptlomaedu4) We thank the New Mexico Department of Game and Fish for permission to study thebighorn sheep of Red Rock and the use of a quad ATV during summer 2000 NMDGFBighorn sheep Program Coordinator Heather Whitlaw provided logistical and administrativesupport Red Rock caretaker Don Graves gave much valued assistance in the eld Dr JamesRominger spent many hours identifyingplants for us Mrs Ruth Anderson provided NationalWeather Service data for Redrock and contractorsAlton Ford and Steve Harvill freely sharedtheir knowledge with us We are especially grateful to Betsy and Louie Cabral who openedtheir ranch to us allowing us to set up tents and live in their caboose lsquoguest housersquo andfreely offered their friendshipFunding for this study was provided by the PLNU Departmentof Biology a PLNU Alumni Association award a PLNU Wesleyan Center Scholarshipand Fellowship a PLNU Research and Special Projects grant and by the New MexicoDepartment of Game and Fish The Wesleyan Center for 21st Century Studies providedof ce space away from distractions for the preparation of the manuscript We dedicate thispublication to the memory of the late Louie Cabral an old cowboy who had a way withhorses

copy Koninklijke Brill NV Leiden 2003 Behaviour 140 183-207Also available online -

184 MOORING ET AL

composition and location of ram and ewe groups during the summer period of segregationActivity budgets were recorded for males in ram herds and females in ewe herds and foragingselectivitywas measured for males and females in mixed groups during early rut Habitat wasevaluated by measuring forage availability ruggedness and visibility at sites utilized by ramand ewe groups Ram herds utilized areas with more available forage compared with ewesites while ewe groups preferred more rugged terrain than that used by ram groups Ewegroups occurred much closer to free water sources than did ram groups Bighorns in ram andewe groups did not differ in foraging time or selectivitynor did time spent moving recliningor ruminating differ between the sexes as predicted by the lsquoactivity budget hypothesisrsquo Theresults support the predictions of the lsquoreproductive strategy-predationrisk hypothesisrsquo whichproposes that males seek more abundant forage in order to build up body condition neededto maximize mating success (even if exposing themselves to greater predation risk) whilefemales choose rugged terrain that minimizes predation risk to themselves and their offspring(even if sacri cing forage abundance) Female bighorns chose sites that provided accessto water also predicted by the lsquoreproductive strategy-predation risk hypothesisrsquo indicatingthat lactation-related water requirements may constrain the movements of ewe groups andcontribute to patterns of sexual segregation in desert bighorn

Introduction

Sexual segregation in which males and females live separately outside ofthe breeding season is an important problem in behavioral ecology thathas received much attention recently (eg Main et al 1996 Bleich et al1997 Ruckstuhl amp Neuhaus 2002) Sexual segregation may occur at differ-ent ecological scales and within different lsquoaggregation unitsrsquo (Main et al1996 Perez-Barberia amp Gordon 1998) and thus has been dif cult to de- ne We will de ne sexual segregation as the separation of males and fe-males into different groups for most of the year (lsquosocial segregationrsquo Con-radt 1998) whether those segregated groups occur in geographically dis-tinct regions (lsquoecologicalrsquo or lsquospatial segregationrsquo) or use the same area non-synchronously (lsquotemporal segregationrsquo)

Sexual body size dimorphism which is associated with polygynous mat-ing systems in ungulates (Weckerly 1998 Loison et al 1999 Perez-Barberia et al 2002) is likely to be a key factor favoring sexual segrega-tion (Mysterud 2000 Ruckstuhl amp Neuhaus 2002) Bighorn sheep (Oviscanadensis) are highly sexually dimorphic Sexual selection has favoredlarge powerful males with massive horns that compete for dominance statusand access to estrus females during the rut Fully grown desert bighorn rams( Nx D 90 kg) are about twice the mass of adult ewes ( Nx D 48 kg) makingbighorn sheep an excellent species in which to investigate sexual segregation

SEXUAL SEGREGATION IN BIGHORN 185

(Hanson amp Deming 1980) Because of its impact on population structuredetermining the causal bases of sexual segregation is important for the con-servation of desert bighorn populations which have been reduced in numbersby disease predation habitat loss and introduced species (Valdez amp Kraus-man 1999 Krausman amp Shackleton 2000) occur in naturally fragmentedterrain (Bleich et al 1990) and usually number less than 50 individuals andare therefore at risk of rapid extinction (Berger 1990)

Predation risk is likely to vary between the sexes and thus may in uencesexual segregation Bighorns evade predation through a combination of ex-cellent eyesight climbing ability and use of open rugged terrain Ruggedprecipitous escape terrain is a major habitat requirement that is preferen-tially selected it provides refuge from predators and is especially impor-tant for females giving birth (Krausman et al 1999 Krausman amp Shackle-ton 2000) Predation is probably the main selective force for group livingin bighorn sheep (Krausman amp Shackleton 2000 Shackleton et al 1999)Bighorns live in 3 types of groups that may differ in group size and spac-ing (1) all male (ram) groups usually of older mature males (2) maternal(ewe) groups consisting of females lambs and yearlings and (3) mixed mat-ing groups comprised of sexually mature males and females with offspringMales and females in most bighorn populations occupy separate seasonalranges throughout most of the year although there is temporal and spatialoverlap (Krausman et al 1999 Shackleton et al 1999) During the matingseason mature males join maternal groups for the duration of the rut

A plethora of hypotheses have been suggested to explain the general phe-nomenon of sexual segregation of which four hypotheses have been iden-ti ed in recent reviews (Main et al 1996 Bleich et al 1997 Ruckstuhlamp Neuhaus 2000 2002) as the most likely to provide useful explanationsof sexual segregation in polygynous ungulates (Table 1) These hypothesesare not necessarily mutually exclusive and one or more factors could be re-sponsible for the evolution of sexual segregation The hypotheses and theirattendant predictions relate to sexual dimorphism in reproductive strategyand body size and subsequent in uences on energetics foraging ecologypredation risk activity budgets and requirements of lactation (Table 1)

The purpose of this study was to gather information to test the predic-tions of these hypotheses taking advantage of the unique opportunities toobserve desert bighorn sheep at the Red Rock Wildlife Area Redrock NewMexico As pointed out by Main et al (1996) if sexual segregation confers

186 MOORING ET AL

TABLE 1 Hypotheses to explain sexual segregation in ungulates

1 Reproductive strategy-predation risk hypothesis (Main et al 1996 Bleich et al 1997Ruckstuhl amp Neuhaus 2000 2002) Males and females pursue different strategies to maxi-mize reproductive success with males maximizing body condition and females maximizingoffspring survival Because males are less vulnerable to predation than females they can ex-ploit areas of greater predation risk with more abundant forage whereas females use areasof increased security that contain predictable sources of food and water for offspring and tosupport lactation

Prediction1 Males will choose sites with more abundant food (even with greater predationrisk)

Prediction 2 Females will choose sites of reduced predation risk (even with lower forageabundance)

Prediction 3 Females with young will occur closer to water sources than mature males

2 Sexual dimorphism-body size hypothesis (Main et al 1996 Bleich et al 1997 Ruck-stuhl amp Neuhaus 2000 2002) Metabolic and digestive differences between the sexes enablelarger-bodied males to exploit more abundant lower-quality forage than smaller-bodied fe-males who must be more selective for less common high-quality forage

Prediction 1 Males will select more abundant and lower-quality foragePrediction 2 Females will selectively feed on less abundant higher-quality forage

3 Activity budget hypothesis (Conradt 1998 Ruckstuhl 1998 Ruckstuhl amp Neuhaus 20002002) Larger males cannot forage with smaller females due to differences in activity budgetsthat result from body size differences in digestive physiology

Prediction 1 Males will spend more time lying down or ruminating to digest higher- berdiet

Prediction 2 Females will spend more time foraging and moving to obtain a high-qualitydiet

Prediction 3 Females will be more selective (eg taking more steps while foraging)Prediction 4 Subadult males will forage more like females than mature males

4 Competition avoidance hypothesis (Geist amp Petocz 1977 Bleich et al 1997) Maturemales avoid competition with potential mates and offspring by segregating themselves fromfemale herds

Prediction 1 Females will inhabit sites with more abundant forage than that found at malesites

Prediction 2 Large dominant males are more likely to segregate than subordinate males

advantages to reproductive success by improving physical condition thensexual segregation should be most pronounced during the time of year whenphysical condition is most in uenced by habitat choice and when energyrequirements differ most between the sexes In desert bighorn sheep this pe-riod corresponds to the spring-summer when males are building up energystores in preparation for rut females are giving birth lactating and raising

SEXUAL SEGREGATION IN BIGHORN 187

offspring and forage availability is in many years limited by the failure ofwinter rains Thus we chose to observe bighorn sheep during the summer of3 consecutive years We collected data on size composition and location ofmale and female groups and evaluated the habitat characteristics where thesegroups were sighted (forage availability visibility ruggedness distance towater) Unlike previous studies of sexual segregation in desert bighorn wewere able construct activity budgets for males and females from hundreds ofhours of observations These activity data enabled us to test the predictionsof the lsquoactivity budget hypothesisrsquo (Ruckstuhl amp Neuhaus 2000 2002) notpreviously tested in desert bighorn

Methods

Study site and animals

Desert bighorn sheep include those subspecies which typically inhabit the arid and rugged re-gions of the American Southwest and Mexico The Mexican bighorn (Ovis canadensis mexi-cana) inhabits the Chihuahuan Desert including New Mexico where native populationshavedeclined everywhere due to drought habitat loss livestock competition predation diseaseand human disturbance (Rominger 1998) The population inhabiting the Red Rock WildlifeArea (RRWA) was established in 1972 by the New Mexico Department of Game and Fish(NMDGF) as a breeding population to augment free-ranging populations and reintroducesheep to traditional habitat The RRWA reserve (32plusmn440N 108plusmn410E) consists of 54 km2 ofrolling hills and steep cliffs along the western banks of the Gila River at 1280-1495 m eleva-tion The facility is surrounded by a 26 m woven game fence with interior fences that dividethe reserve into 5 pastures bighorn pass freely though gates in the interior fencing and thushave access to all 5 pastures

The desert scrub habitat at RRWA is dominated by scattered trees of juniper (Juniperusmonosperma) and velvet mesquite (Prosopis velutina) shrubs of whitethorn (Acacia con-stricta) catclaw (A gregii) oreganillo (Aloysia wrightii) senna (Cassia bauhinoides) so-tol (Dasylirion wheeleri) and ocotillo (Fouquieria splendens) various forbs including fourorsquoclock (Mirabilis multiora M pumila Boerhaavia spp) spreading sida (Sida licaulis)and bush peavine (Lathyrun eucosmus) grasses of black grama (Bouteloua eripoda) deer-grass (Muhlenbergia rigens) panicgrass (Panicum hirtiaaule) and sand dropseed (Sporobo-lus crytandrus) and prickly pear cacti (Opuntia spp)

Most populations of desert bighorn range over vast rugged areas in small groups that areintolerant of human presence and extremely dif cult to nd and observe (Krausman amp Shack-leton 2000) Because RRWA sheep are con ned to the reserve by the perimeter game fencepopulation density tends to be high for desert bighorn (14-20 sheep per km2 in recent years)This allowed us to locate and observe a signi cant proportion (gt50) of the population ona daily basis without radio telemetry in contrast to most wild populations and provided aunique opportunity to collect ne-grained behavioral data on desert bighorn Although thepresence of the game fence prevents bighorns from leaving the reserve it does not preventother animals from enteringPeccaries (Pecari tajacu) regularlydig under the perimeter fence

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

184 MOORING ET AL

composition and location of ram and ewe groups during the summer period of segregationActivity budgets were recorded for males in ram herds and females in ewe herds and foragingselectivitywas measured for males and females in mixed groups during early rut Habitat wasevaluated by measuring forage availability ruggedness and visibility at sites utilized by ramand ewe groups Ram herds utilized areas with more available forage compared with ewesites while ewe groups preferred more rugged terrain than that used by ram groups Ewegroups occurred much closer to free water sources than did ram groups Bighorns in ram andewe groups did not differ in foraging time or selectivitynor did time spent moving recliningor ruminating differ between the sexes as predicted by the lsquoactivity budget hypothesisrsquo Theresults support the predictions of the lsquoreproductive strategy-predationrisk hypothesisrsquo whichproposes that males seek more abundant forage in order to build up body condition neededto maximize mating success (even if exposing themselves to greater predation risk) whilefemales choose rugged terrain that minimizes predation risk to themselves and their offspring(even if sacri cing forage abundance) Female bighorns chose sites that provided accessto water also predicted by the lsquoreproductive strategy-predation risk hypothesisrsquo indicatingthat lactation-related water requirements may constrain the movements of ewe groups andcontribute to patterns of sexual segregation in desert bighorn

Introduction

Sexual segregation in which males and females live separately outside ofthe breeding season is an important problem in behavioral ecology thathas received much attention recently (eg Main et al 1996 Bleich et al1997 Ruckstuhl amp Neuhaus 2002) Sexual segregation may occur at differ-ent ecological scales and within different lsquoaggregation unitsrsquo (Main et al1996 Perez-Barberia amp Gordon 1998) and thus has been dif cult to de- ne We will de ne sexual segregation as the separation of males and fe-males into different groups for most of the year (lsquosocial segregationrsquo Con-radt 1998) whether those segregated groups occur in geographically dis-tinct regions (lsquoecologicalrsquo or lsquospatial segregationrsquo) or use the same area non-synchronously (lsquotemporal segregationrsquo)

Sexual body size dimorphism which is associated with polygynous mat-ing systems in ungulates (Weckerly 1998 Loison et al 1999 Perez-Barberia et al 2002) is likely to be a key factor favoring sexual segrega-tion (Mysterud 2000 Ruckstuhl amp Neuhaus 2002) Bighorn sheep (Oviscanadensis) are highly sexually dimorphic Sexual selection has favoredlarge powerful males with massive horns that compete for dominance statusand access to estrus females during the rut Fully grown desert bighorn rams( Nx D 90 kg) are about twice the mass of adult ewes ( Nx D 48 kg) makingbighorn sheep an excellent species in which to investigate sexual segregation

SEXUAL SEGREGATION IN BIGHORN 185

(Hanson amp Deming 1980) Because of its impact on population structuredetermining the causal bases of sexual segregation is important for the con-servation of desert bighorn populations which have been reduced in numbersby disease predation habitat loss and introduced species (Valdez amp Kraus-man 1999 Krausman amp Shackleton 2000) occur in naturally fragmentedterrain (Bleich et al 1990) and usually number less than 50 individuals andare therefore at risk of rapid extinction (Berger 1990)

Predation risk is likely to vary between the sexes and thus may in uencesexual segregation Bighorns evade predation through a combination of ex-cellent eyesight climbing ability and use of open rugged terrain Ruggedprecipitous escape terrain is a major habitat requirement that is preferen-tially selected it provides refuge from predators and is especially impor-tant for females giving birth (Krausman et al 1999 Krausman amp Shackle-ton 2000) Predation is probably the main selective force for group livingin bighorn sheep (Krausman amp Shackleton 2000 Shackleton et al 1999)Bighorns live in 3 types of groups that may differ in group size and spac-ing (1) all male (ram) groups usually of older mature males (2) maternal(ewe) groups consisting of females lambs and yearlings and (3) mixed mat-ing groups comprised of sexually mature males and females with offspringMales and females in most bighorn populations occupy separate seasonalranges throughout most of the year although there is temporal and spatialoverlap (Krausman et al 1999 Shackleton et al 1999) During the matingseason mature males join maternal groups for the duration of the rut

A plethora of hypotheses have been suggested to explain the general phe-nomenon of sexual segregation of which four hypotheses have been iden-ti ed in recent reviews (Main et al 1996 Bleich et al 1997 Ruckstuhlamp Neuhaus 2000 2002) as the most likely to provide useful explanationsof sexual segregation in polygynous ungulates (Table 1) These hypothesesare not necessarily mutually exclusive and one or more factors could be re-sponsible for the evolution of sexual segregation The hypotheses and theirattendant predictions relate to sexual dimorphism in reproductive strategyand body size and subsequent in uences on energetics foraging ecologypredation risk activity budgets and requirements of lactation (Table 1)

The purpose of this study was to gather information to test the predic-tions of these hypotheses taking advantage of the unique opportunities toobserve desert bighorn sheep at the Red Rock Wildlife Area Redrock NewMexico As pointed out by Main et al (1996) if sexual segregation confers

186 MOORING ET AL

TABLE 1 Hypotheses to explain sexual segregation in ungulates

1 Reproductive strategy-predation risk hypothesis (Main et al 1996 Bleich et al 1997Ruckstuhl amp Neuhaus 2000 2002) Males and females pursue different strategies to maxi-mize reproductive success with males maximizing body condition and females maximizingoffspring survival Because males are less vulnerable to predation than females they can ex-ploit areas of greater predation risk with more abundant forage whereas females use areasof increased security that contain predictable sources of food and water for offspring and tosupport lactation

Prediction1 Males will choose sites with more abundant food (even with greater predationrisk)

Prediction 2 Females will choose sites of reduced predation risk (even with lower forageabundance)

Prediction 3 Females with young will occur closer to water sources than mature males

2 Sexual dimorphism-body size hypothesis (Main et al 1996 Bleich et al 1997 Ruck-stuhl amp Neuhaus 2000 2002) Metabolic and digestive differences between the sexes enablelarger-bodied males to exploit more abundant lower-quality forage than smaller-bodied fe-males who must be more selective for less common high-quality forage

Prediction 1 Males will select more abundant and lower-quality foragePrediction 2 Females will selectively feed on less abundant higher-quality forage

3 Activity budget hypothesis (Conradt 1998 Ruckstuhl 1998 Ruckstuhl amp Neuhaus 20002002) Larger males cannot forage with smaller females due to differences in activity budgetsthat result from body size differences in digestive physiology

Prediction 1 Males will spend more time lying down or ruminating to digest higher- berdiet

Prediction 2 Females will spend more time foraging and moving to obtain a high-qualitydiet

Prediction 3 Females will be more selective (eg taking more steps while foraging)Prediction 4 Subadult males will forage more like females than mature males

4 Competition avoidance hypothesis (Geist amp Petocz 1977 Bleich et al 1997) Maturemales avoid competition with potential mates and offspring by segregating themselves fromfemale herds

Prediction 1 Females will inhabit sites with more abundant forage than that found at malesites

Prediction 2 Large dominant males are more likely to segregate than subordinate males

advantages to reproductive success by improving physical condition thensexual segregation should be most pronounced during the time of year whenphysical condition is most in uenced by habitat choice and when energyrequirements differ most between the sexes In desert bighorn sheep this pe-riod corresponds to the spring-summer when males are building up energystores in preparation for rut females are giving birth lactating and raising

SEXUAL SEGREGATION IN BIGHORN 187

offspring and forage availability is in many years limited by the failure ofwinter rains Thus we chose to observe bighorn sheep during the summer of3 consecutive years We collected data on size composition and location ofmale and female groups and evaluated the habitat characteristics where thesegroups were sighted (forage availability visibility ruggedness distance towater) Unlike previous studies of sexual segregation in desert bighorn wewere able construct activity budgets for males and females from hundreds ofhours of observations These activity data enabled us to test the predictionsof the lsquoactivity budget hypothesisrsquo (Ruckstuhl amp Neuhaus 2000 2002) notpreviously tested in desert bighorn

Methods

Study site and animals

Desert bighorn sheep include those subspecies which typically inhabit the arid and rugged re-gions of the American Southwest and Mexico The Mexican bighorn (Ovis canadensis mexi-cana) inhabits the Chihuahuan Desert including New Mexico where native populationshavedeclined everywhere due to drought habitat loss livestock competition predation diseaseand human disturbance (Rominger 1998) The population inhabiting the Red Rock WildlifeArea (RRWA) was established in 1972 by the New Mexico Department of Game and Fish(NMDGF) as a breeding population to augment free-ranging populations and reintroducesheep to traditional habitat The RRWA reserve (32plusmn440N 108plusmn410E) consists of 54 km2 ofrolling hills and steep cliffs along the western banks of the Gila River at 1280-1495 m eleva-tion The facility is surrounded by a 26 m woven game fence with interior fences that dividethe reserve into 5 pastures bighorn pass freely though gates in the interior fencing and thushave access to all 5 pastures

The desert scrub habitat at RRWA is dominated by scattered trees of juniper (Juniperusmonosperma) and velvet mesquite (Prosopis velutina) shrubs of whitethorn (Acacia con-stricta) catclaw (A gregii) oreganillo (Aloysia wrightii) senna (Cassia bauhinoides) so-tol (Dasylirion wheeleri) and ocotillo (Fouquieria splendens) various forbs including fourorsquoclock (Mirabilis multiora M pumila Boerhaavia spp) spreading sida (Sida licaulis)and bush peavine (Lathyrun eucosmus) grasses of black grama (Bouteloua eripoda) deer-grass (Muhlenbergia rigens) panicgrass (Panicum hirtiaaule) and sand dropseed (Sporobo-lus crytandrus) and prickly pear cacti (Opuntia spp)

Most populations of desert bighorn range over vast rugged areas in small groups that areintolerant of human presence and extremely dif cult to nd and observe (Krausman amp Shack-leton 2000) Because RRWA sheep are con ned to the reserve by the perimeter game fencepopulation density tends to be high for desert bighorn (14-20 sheep per km2 in recent years)This allowed us to locate and observe a signi cant proportion (gt50) of the population ona daily basis without radio telemetry in contrast to most wild populations and provided aunique opportunity to collect ne-grained behavioral data on desert bighorn Although thepresence of the game fence prevents bighorns from leaving the reserve it does not preventother animals from enteringPeccaries (Pecari tajacu) regularlydig under the perimeter fence

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 185

(Hanson amp Deming 1980) Because of its impact on population structuredetermining the causal bases of sexual segregation is important for the con-servation of desert bighorn populations which have been reduced in numbersby disease predation habitat loss and introduced species (Valdez amp Kraus-man 1999 Krausman amp Shackleton 2000) occur in naturally fragmentedterrain (Bleich et al 1990) and usually number less than 50 individuals andare therefore at risk of rapid extinction (Berger 1990)

Predation risk is likely to vary between the sexes and thus may in uencesexual segregation Bighorns evade predation through a combination of ex-cellent eyesight climbing ability and use of open rugged terrain Ruggedprecipitous escape terrain is a major habitat requirement that is preferen-tially selected it provides refuge from predators and is especially impor-tant for females giving birth (Krausman et al 1999 Krausman amp Shackle-ton 2000) Predation is probably the main selective force for group livingin bighorn sheep (Krausman amp Shackleton 2000 Shackleton et al 1999)Bighorns live in 3 types of groups that may differ in group size and spac-ing (1) all male (ram) groups usually of older mature males (2) maternal(ewe) groups consisting of females lambs and yearlings and (3) mixed mat-ing groups comprised of sexually mature males and females with offspringMales and females in most bighorn populations occupy separate seasonalranges throughout most of the year although there is temporal and spatialoverlap (Krausman et al 1999 Shackleton et al 1999) During the matingseason mature males join maternal groups for the duration of the rut

A plethora of hypotheses have been suggested to explain the general phe-nomenon of sexual segregation of which four hypotheses have been iden-ti ed in recent reviews (Main et al 1996 Bleich et al 1997 Ruckstuhlamp Neuhaus 2000 2002) as the most likely to provide useful explanationsof sexual segregation in polygynous ungulates (Table 1) These hypothesesare not necessarily mutually exclusive and one or more factors could be re-sponsible for the evolution of sexual segregation The hypotheses and theirattendant predictions relate to sexual dimorphism in reproductive strategyand body size and subsequent in uences on energetics foraging ecologypredation risk activity budgets and requirements of lactation (Table 1)

The purpose of this study was to gather information to test the predic-tions of these hypotheses taking advantage of the unique opportunities toobserve desert bighorn sheep at the Red Rock Wildlife Area Redrock NewMexico As pointed out by Main et al (1996) if sexual segregation confers

186 MOORING ET AL

TABLE 1 Hypotheses to explain sexual segregation in ungulates

1 Reproductive strategy-predation risk hypothesis (Main et al 1996 Bleich et al 1997Ruckstuhl amp Neuhaus 2000 2002) Males and females pursue different strategies to maxi-mize reproductive success with males maximizing body condition and females maximizingoffspring survival Because males are less vulnerable to predation than females they can ex-ploit areas of greater predation risk with more abundant forage whereas females use areasof increased security that contain predictable sources of food and water for offspring and tosupport lactation

Prediction1 Males will choose sites with more abundant food (even with greater predationrisk)

Prediction 2 Females will choose sites of reduced predation risk (even with lower forageabundance)

Prediction 3 Females with young will occur closer to water sources than mature males

2 Sexual dimorphism-body size hypothesis (Main et al 1996 Bleich et al 1997 Ruck-stuhl amp Neuhaus 2000 2002) Metabolic and digestive differences between the sexes enablelarger-bodied males to exploit more abundant lower-quality forage than smaller-bodied fe-males who must be more selective for less common high-quality forage

Prediction 1 Males will select more abundant and lower-quality foragePrediction 2 Females will selectively feed on less abundant higher-quality forage

3 Activity budget hypothesis (Conradt 1998 Ruckstuhl 1998 Ruckstuhl amp Neuhaus 20002002) Larger males cannot forage with smaller females due to differences in activity budgetsthat result from body size differences in digestive physiology

Prediction 1 Males will spend more time lying down or ruminating to digest higher- berdiet

Prediction 2 Females will spend more time foraging and moving to obtain a high-qualitydiet

Prediction 3 Females will be more selective (eg taking more steps while foraging)Prediction 4 Subadult males will forage more like females than mature males

4 Competition avoidance hypothesis (Geist amp Petocz 1977 Bleich et al 1997) Maturemales avoid competition with potential mates and offspring by segregating themselves fromfemale herds

Prediction 1 Females will inhabit sites with more abundant forage than that found at malesites

Prediction 2 Large dominant males are more likely to segregate than subordinate males

advantages to reproductive success by improving physical condition thensexual segregation should be most pronounced during the time of year whenphysical condition is most in uenced by habitat choice and when energyrequirements differ most between the sexes In desert bighorn sheep this pe-riod corresponds to the spring-summer when males are building up energystores in preparation for rut females are giving birth lactating and raising

SEXUAL SEGREGATION IN BIGHORN 187

offspring and forage availability is in many years limited by the failure ofwinter rains Thus we chose to observe bighorn sheep during the summer of3 consecutive years We collected data on size composition and location ofmale and female groups and evaluated the habitat characteristics where thesegroups were sighted (forage availability visibility ruggedness distance towater) Unlike previous studies of sexual segregation in desert bighorn wewere able construct activity budgets for males and females from hundreds ofhours of observations These activity data enabled us to test the predictionsof the lsquoactivity budget hypothesisrsquo (Ruckstuhl amp Neuhaus 2000 2002) notpreviously tested in desert bighorn

Methods

Study site and animals

Desert bighorn sheep include those subspecies which typically inhabit the arid and rugged re-gions of the American Southwest and Mexico The Mexican bighorn (Ovis canadensis mexi-cana) inhabits the Chihuahuan Desert including New Mexico where native populationshavedeclined everywhere due to drought habitat loss livestock competition predation diseaseand human disturbance (Rominger 1998) The population inhabiting the Red Rock WildlifeArea (RRWA) was established in 1972 by the New Mexico Department of Game and Fish(NMDGF) as a breeding population to augment free-ranging populations and reintroducesheep to traditional habitat The RRWA reserve (32plusmn440N 108plusmn410E) consists of 54 km2 ofrolling hills and steep cliffs along the western banks of the Gila River at 1280-1495 m eleva-tion The facility is surrounded by a 26 m woven game fence with interior fences that dividethe reserve into 5 pastures bighorn pass freely though gates in the interior fencing and thushave access to all 5 pastures

The desert scrub habitat at RRWA is dominated by scattered trees of juniper (Juniperusmonosperma) and velvet mesquite (Prosopis velutina) shrubs of whitethorn (Acacia con-stricta) catclaw (A gregii) oreganillo (Aloysia wrightii) senna (Cassia bauhinoides) so-tol (Dasylirion wheeleri) and ocotillo (Fouquieria splendens) various forbs including fourorsquoclock (Mirabilis multiora M pumila Boerhaavia spp) spreading sida (Sida licaulis)and bush peavine (Lathyrun eucosmus) grasses of black grama (Bouteloua eripoda) deer-grass (Muhlenbergia rigens) panicgrass (Panicum hirtiaaule) and sand dropseed (Sporobo-lus crytandrus) and prickly pear cacti (Opuntia spp)

Most populations of desert bighorn range over vast rugged areas in small groups that areintolerant of human presence and extremely dif cult to nd and observe (Krausman amp Shack-leton 2000) Because RRWA sheep are con ned to the reserve by the perimeter game fencepopulation density tends to be high for desert bighorn (14-20 sheep per km2 in recent years)This allowed us to locate and observe a signi cant proportion (gt50) of the population ona daily basis without radio telemetry in contrast to most wild populations and provided aunique opportunity to collect ne-grained behavioral data on desert bighorn Although thepresence of the game fence prevents bighorns from leaving the reserve it does not preventother animals from enteringPeccaries (Pecari tajacu) regularlydig under the perimeter fence

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

186 MOORING ET AL

TABLE 1 Hypotheses to explain sexual segregation in ungulates

1 Reproductive strategy-predation risk hypothesis (Main et al 1996 Bleich et al 1997Ruckstuhl amp Neuhaus 2000 2002) Males and females pursue different strategies to maxi-mize reproductive success with males maximizing body condition and females maximizingoffspring survival Because males are less vulnerable to predation than females they can ex-ploit areas of greater predation risk with more abundant forage whereas females use areasof increased security that contain predictable sources of food and water for offspring and tosupport lactation

Prediction1 Males will choose sites with more abundant food (even with greater predationrisk)

Prediction 2 Females will choose sites of reduced predation risk (even with lower forageabundance)

Prediction 3 Females with young will occur closer to water sources than mature males

2 Sexual dimorphism-body size hypothesis (Main et al 1996 Bleich et al 1997 Ruck-stuhl amp Neuhaus 2000 2002) Metabolic and digestive differences between the sexes enablelarger-bodied males to exploit more abundant lower-quality forage than smaller-bodied fe-males who must be more selective for less common high-quality forage

Prediction 1 Males will select more abundant and lower-quality foragePrediction 2 Females will selectively feed on less abundant higher-quality forage

3 Activity budget hypothesis (Conradt 1998 Ruckstuhl 1998 Ruckstuhl amp Neuhaus 20002002) Larger males cannot forage with smaller females due to differences in activity budgetsthat result from body size differences in digestive physiology

Prediction 1 Males will spend more time lying down or ruminating to digest higher- berdiet

Prediction 2 Females will spend more time foraging and moving to obtain a high-qualitydiet

Prediction 3 Females will be more selective (eg taking more steps while foraging)Prediction 4 Subadult males will forage more like females than mature males

4 Competition avoidance hypothesis (Geist amp Petocz 1977 Bleich et al 1997) Maturemales avoid competition with potential mates and offspring by segregating themselves fromfemale herds

Prediction 1 Females will inhabit sites with more abundant forage than that found at malesites

Prediction 2 Large dominant males are more likely to segregate than subordinate males

advantages to reproductive success by improving physical condition thensexual segregation should be most pronounced during the time of year whenphysical condition is most in uenced by habitat choice and when energyrequirements differ most between the sexes In desert bighorn sheep this pe-riod corresponds to the spring-summer when males are building up energystores in preparation for rut females are giving birth lactating and raising

SEXUAL SEGREGATION IN BIGHORN 187

offspring and forage availability is in many years limited by the failure ofwinter rains Thus we chose to observe bighorn sheep during the summer of3 consecutive years We collected data on size composition and location ofmale and female groups and evaluated the habitat characteristics where thesegroups were sighted (forage availability visibility ruggedness distance towater) Unlike previous studies of sexual segregation in desert bighorn wewere able construct activity budgets for males and females from hundreds ofhours of observations These activity data enabled us to test the predictionsof the lsquoactivity budget hypothesisrsquo (Ruckstuhl amp Neuhaus 2000 2002) notpreviously tested in desert bighorn

Methods

Study site and animals

Desert bighorn sheep include those subspecies which typically inhabit the arid and rugged re-gions of the American Southwest and Mexico The Mexican bighorn (Ovis canadensis mexi-cana) inhabits the Chihuahuan Desert including New Mexico where native populationshavedeclined everywhere due to drought habitat loss livestock competition predation diseaseand human disturbance (Rominger 1998) The population inhabiting the Red Rock WildlifeArea (RRWA) was established in 1972 by the New Mexico Department of Game and Fish(NMDGF) as a breeding population to augment free-ranging populations and reintroducesheep to traditional habitat The RRWA reserve (32plusmn440N 108plusmn410E) consists of 54 km2 ofrolling hills and steep cliffs along the western banks of the Gila River at 1280-1495 m eleva-tion The facility is surrounded by a 26 m woven game fence with interior fences that dividethe reserve into 5 pastures bighorn pass freely though gates in the interior fencing and thushave access to all 5 pastures

The desert scrub habitat at RRWA is dominated by scattered trees of juniper (Juniperusmonosperma) and velvet mesquite (Prosopis velutina) shrubs of whitethorn (Acacia con-stricta) catclaw (A gregii) oreganillo (Aloysia wrightii) senna (Cassia bauhinoides) so-tol (Dasylirion wheeleri) and ocotillo (Fouquieria splendens) various forbs including fourorsquoclock (Mirabilis multiora M pumila Boerhaavia spp) spreading sida (Sida licaulis)and bush peavine (Lathyrun eucosmus) grasses of black grama (Bouteloua eripoda) deer-grass (Muhlenbergia rigens) panicgrass (Panicum hirtiaaule) and sand dropseed (Sporobo-lus crytandrus) and prickly pear cacti (Opuntia spp)

Most populations of desert bighorn range over vast rugged areas in small groups that areintolerant of human presence and extremely dif cult to nd and observe (Krausman amp Shack-leton 2000) Because RRWA sheep are con ned to the reserve by the perimeter game fencepopulation density tends to be high for desert bighorn (14-20 sheep per km2 in recent years)This allowed us to locate and observe a signi cant proportion (gt50) of the population ona daily basis without radio telemetry in contrast to most wild populations and provided aunique opportunity to collect ne-grained behavioral data on desert bighorn Although thepresence of the game fence prevents bighorns from leaving the reserve it does not preventother animals from enteringPeccaries (Pecari tajacu) regularlydig under the perimeter fence

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 187

offspring and forage availability is in many years limited by the failure ofwinter rains Thus we chose to observe bighorn sheep during the summer of3 consecutive years We collected data on size composition and location ofmale and female groups and evaluated the habitat characteristics where thesegroups were sighted (forage availability visibility ruggedness distance towater) Unlike previous studies of sexual segregation in desert bighorn wewere able construct activity budgets for males and females from hundreds ofhours of observations These activity data enabled us to test the predictionsof the lsquoactivity budget hypothesisrsquo (Ruckstuhl amp Neuhaus 2000 2002) notpreviously tested in desert bighorn

Methods

Study site and animals

Desert bighorn sheep include those subspecies which typically inhabit the arid and rugged re-gions of the American Southwest and Mexico The Mexican bighorn (Ovis canadensis mexi-cana) inhabits the Chihuahuan Desert including New Mexico where native populationshavedeclined everywhere due to drought habitat loss livestock competition predation diseaseand human disturbance (Rominger 1998) The population inhabiting the Red Rock WildlifeArea (RRWA) was established in 1972 by the New Mexico Department of Game and Fish(NMDGF) as a breeding population to augment free-ranging populations and reintroducesheep to traditional habitat The RRWA reserve (32plusmn440N 108plusmn410E) consists of 54 km2 ofrolling hills and steep cliffs along the western banks of the Gila River at 1280-1495 m eleva-tion The facility is surrounded by a 26 m woven game fence with interior fences that dividethe reserve into 5 pastures bighorn pass freely though gates in the interior fencing and thushave access to all 5 pastures

The desert scrub habitat at RRWA is dominated by scattered trees of juniper (Juniperusmonosperma) and velvet mesquite (Prosopis velutina) shrubs of whitethorn (Acacia con-stricta) catclaw (A gregii) oreganillo (Aloysia wrightii) senna (Cassia bauhinoides) so-tol (Dasylirion wheeleri) and ocotillo (Fouquieria splendens) various forbs including fourorsquoclock (Mirabilis multiora M pumila Boerhaavia spp) spreading sida (Sida licaulis)and bush peavine (Lathyrun eucosmus) grasses of black grama (Bouteloua eripoda) deer-grass (Muhlenbergia rigens) panicgrass (Panicum hirtiaaule) and sand dropseed (Sporobo-lus crytandrus) and prickly pear cacti (Opuntia spp)

Most populations of desert bighorn range over vast rugged areas in small groups that areintolerant of human presence and extremely dif cult to nd and observe (Krausman amp Shack-leton 2000) Because RRWA sheep are con ned to the reserve by the perimeter game fencepopulation density tends to be high for desert bighorn (14-20 sheep per km2 in recent years)This allowed us to locate and observe a signi cant proportion (gt50) of the population ona daily basis without radio telemetry in contrast to most wild populations and provided aunique opportunity to collect ne-grained behavioral data on desert bighorn Although thepresence of the game fence prevents bighorns from leaving the reserve it does not preventother animals from enteringPeccaries (Pecari tajacu) regularlydig under the perimeter fence

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

188 MOORING ET AL

to the reserve providing entrance for predators such as coyote (Canis latrans) bobcat (Lynxrufus) and mountain lion (Felis concolor) mountain lions can also enter by climbing thewooden fence posts Despite attempted predator control by NMDGF signi cant mortalitiesattributable to predation occur regularly Mountain lions have been the most important preda-tors of RRWA adult bighorn in recent years of the 18 con rmed mountain lion predationssince 1982 8 occurred over our study period Thus predation risk is very real in this popula-tion and there is no question that predator avoidance strategies represent current threat ratherthan adaptations to the past (Byers 1997 Ruckstuhl amp Neuhaus 2000)

The Red Rock population size varied from year to year due to predation and other naturalmortality and the periodic capture and removal of animals to augment other populationsAn-nual ground surveys conducted in late April of each year by NMDGF indicated a populationof 108 bighorns in 1999 (55 males 39 females 14 juveniles) 76 bighorn in 2000 followingtransplant of 27 rams and 5 mortalities (20 males 39 females 17 juveniles) and 79 bighornsin 2001 (29 males 31 females 19 juveniles) Measurements taken on captured sheep during1995 and 1997 provided a mean mass of 60 kg for females (N D 37) and 84 kg for males(N D 44) indicating that the average male weight was about 140 that of the average femaleweight

Behavioral observations

We conducted eldwork during June-July of 1999 and 2000 and early August 2001 Workwas done primarily on foot using the 30 km of primitive road system or hiking cross-countrywe also used a 4-wheel drive vehicle and a quad all-terrain vehicle (ATV) Behavioral ob-servations were conducted on unmarked individuals with 10pound binoculars and 15-60pound tele-scopes we were usually far enough away (500-1000 m) for telescopes to be required Weconducted observations primarily during the morning hours when sheep were most activeDesert bighorns use an energy-minimizing strategy to avoid heat stress during the hot sum-mer months spending most of the afternoonsbedded down in shade (Simmons 1980 Kraus-man et al 1999 Krausman amp Shackleton 2000) To test the lsquoactivity budget hypothesisrsquowe were interested in recording foraging moving and ruminating preliminary observationsindicated that sheep in the afternoons seldom foraged or moved and resting individuals wereoften hidden from view in the shade of junipers We observed mature rams and adult eweschosen randomly from the herd during 10-min focal animal samples We recorded activityby instantaneous sampling at 1-min intervals (Altmann 1974) using wrist watches with arepeating alarm function to time scans The number of scans devoted to different activitieswas used to compute the mean percentage of scans that focal animals spent engaged in thoseactivities The percentage of time spent feeding (grazing and browsing) standing moving(walking or running) and lying down was analyzed for both 1999 and 2000 activity data onrumination and vigilance were only available for 2000 Activity observations totaled 347 hover 3 summers

Fieldwork conducted in August 2001 examined the degree of foraging selectivity exhib-ited by rams and ewes based on the number of steps taken per min foraging A test of selectiv-ity is confounded if vegetation cover varies between ewe and ram group sites because sheepwould have to walk more in search of food in areas with a sparser vegetation cover Becauserams do not usually associate with ewes outside of the rut observations were made at the be-ginning of the rutting season when both sexes were found together in large mixed groups inthe same vegetationcommunitiesObservationswere conductedon foraging ewes and mature

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 189

rams (Class 2 3 4) during 10-min focal animal samples Because sheep rarely foraged with-out interruption for the entire 10 minutes the actual time spent foraging (grazing browsingor searching)was recordedwith a stopwatch to the nearest second Whenever the focal animalstopped foraging to engage in another activity (eg vigilance chasing interacting) the stop-watch was stopped until the focal animal resumed foraging Following Risenhoover amp Bailey(1985) and Ruckstuhl (1998) the number of steps made by both forelegs was counted with ahand tally (hindleg steps were counted when the forelegs were not visible) The measurementresulting was the number of steps per minute of actual foraging

Inter-observer reliability tests were conducted in 1999 and 2000 to assess the level ofagreement among observers (observations in 2001 were made by MSM only) For focalobservation reliability tests in 1999 an experiencedobserver (MSM) observed the same focalanimal with one other observer at a time (TAF JEB ICF) for 15 10-min samples In 2000 all4 observers (MSM TAF TTN DDR) collecteddata on the same focal animal at the same timefor 15 10-min samples Pearson correlations showed that inter-observer reliability exceeded090 for both summers For summers 1999 and 2000 the mean correlation coef cient forfocal samples was r D 091 and r D 093 respectively

Group locations and habitat evaluation

Individuals that were located on average lt10 body lengths from one another and movedtogether in a coordinated fashion were considered to be members of the same group The fol-lowing agesex classes were recorded based upon body size and horn con guration (Geist1971 Hansen amp Deming 1980) lambs (juveniles lt1 year) yearling ewes (1-year old fe-males) yearling rams (1-year old males D Class 1 rams) ewes (adult females gt 2 years)and rams (adult males gt 2 years D Class 2 3 and 4 rams) Rams were further categorizedby horn curl as either Class 2 (12 curl) Class 3 (34 curl) or Class 4 (34-full curl) withhorn thickness and brooming (wear) of tips assisting in identi cation (Geist 1971 Hansenamp Deming 1980) Because of overlap between Class 3 and 4 horn structure we later com-bined these oldest rams into one category We catalogued bighorn herds as ewe groups ramgroups or mixed groups (see Introduction) During summer 2000 whenever we sighted aherd of bighorns we recorded group size composition and the map reference location Torecord group locations we used a map of the refuge overlaid with a grid reference system inwhich each quadrat corresponded to 100 pound 100 m on the ground Reference grids could bespeci ed to 10 m increments north-south or east-west with error not likely to exceed 50 mThis grid system could later be overlaid on a topographic map to measure terrain ruggednessand distance to water sources

Locations where male and female herds were observed were later measured in the eld forhabitat quality vegetation slope and visibilityGiven the limitations of fecal dietary analysiswhen applied to mixed feeders of graze and browse (eg Holecheck amp Valdez 1985 Hobbs1987 Bartolome et al 1995) we chose to measure forage availability which is reliableand easy to interpret To characterize forage availability we sampled eleven 40-m vegetationtransects using the Daubenmire frame technique (Daubenmire 1959) in representative areasof the reserve during July of 2000 Many studies have indicated that bighorn sheep selectforage opportunistically on the basis of availability not nutritional value (Krausman et al1989 Krausman amp Shackleton 2000) Forage types (browse grasses forbs) are eaten in pro-portions similar to their availability although bighorns may show preferences for individual

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

190 MOORING ET AL

species within a forage class (Miller amp Gaud 1989 Wikeem amp Pitt 1991 1992) Thereforeavailabilityof major forage types may be used as an index of bighorn diet

In addition to vegetation sampling 46 quadrats throughout the reserve were sampled forruggedness and visibilityRuggedness was estimated in two ways (1) For each 100 pound 100 mquadrat we used a random number table to choose a starting point and with a clinometermeasured the slope at 40 m in each cardinal direction The slope values were averaged to givethe mean slope (2) The second method was performed by placing the map grid system over aUSGS 75-minute quadrangle map (3-D TopoQuad DeLorme Yarmouth ME) and countingthe number of arcs forming the contour lines within each quadrat as an index of ruggedness(Bleich et al 1997) with greater number of contour lines indicating more rugged terrainVisibility was measured using a modi cation of the technique of Risenhoover amp Bailey(1985) we placed a lsquovisibility stickrsquo (yard stick demarcated into 10-cm segments) at thestarting point used for slope measurement and recorded the percentage of the stick visible toan observer at 20-m and 40-m along the 4 cardinal compass directions The mean visibilitywas then computed for each quadrat

Water availability

Proximity to water sources was measured during summer 2000 The location of standingwater in the reserve was identi ed on a USGS 75-minute quadrangle map covering the RedRock reserve overlaid by a tranparency of the research map with the reference grid Thesummer of 2000 was exceptionally dry with only 99 mm of rain falling in June-July andso ephemeral pools following rains would not have been a signi cant source of drinkingwater for sheep During this time period the 2 water catchment devices in the reserve werenot operating and the two windmill-powered pumps were hardly ever used by bighorns(Steve Harvill personal communication) perhaps because they are located away from escapeterrain and are surrounded by dense vegetation that obscures visibility Water was availablethroughout summer 2000 at a spring on the Right Fork of Jackrsquos Canyon and at several poolsfound along Ash Creek Box Canyon We measured the distance from the map reference forevery herd sighted to the closest source of water using calipers and distance scale

Statistical analysis

We used the SPSS 80 statisticalpackage (Norusis 1998) to conduct nonparametric statisticaltests for correlation analysis and comparison of means (Siegel amp Castellan 1988) The levelof signi cance was set at 005 and all tests were two-tailed Analysis could not be based onfocal samples because this would have in ated the sample size and created a pseudorepli-cation problem by repeat sampling of the same individuals Individual means could not becalculatedbecause most bighorns were not individually recognizable at the distances that ob-servationswere conducted (usually gt500 m) Desert bighorn sheep are typically intolerant toclose approach unlike Rocky Mountain bighorns which can be approached to within 5-20 m(eg Hogg 1987 Hass 1990) Therefore following the procedure of Mooring (1995) weused mean grooming rates for each agesex class as our sampling unit For analysis of activitybudget data we computed means for each observationday for rams ewes and yearling malesin different group types Analysis using daily means avoided pseudoreplicationby producinga sample size that was no greater than the number of individuals of each agesex class presentin the population with the sampling unit a composite measure representing the characteristic

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 191

activity budget of each class However by limiting the sample size in this way we increasedthe possibility of committing a Type II error This was especially a concern because the mainresult of the daily means analysis was a nding of no signi cant difference between ramsand ewes in activities that are predicted to differ between the sexes by the lsquoactivity budgethypothesisrsquo Therefore to con rm the results we re-analyzed the activity budget data usingeach 10-min focal animal sample as a data point thereby increasing the chance of obtaininga signi cant difference (if one actually existed) by in ating the sample size To evaluate thepower of these analyses to detect a real difference in activity budgets we conducted post-hocpower analysis using GPOWER Version 20 (Faul amp Erdfelder 1992) Cohen (1988) has sug-gested that a power level of gt 080 be used when reg D 005 and no other basis for selecting apower level exists

Results

Group size and composition

Group size differed signi cantly between males in ram groups and femalesin ewe groups during both summers During summer 1999 ram groupsize ( Nx D 6) was smaller than ewe group size ( Nx D 15) (Mann-Whitneyz D 108 N D 633 p D 00001 with 9 of ram groups consisting ofsolitary males (N D 301 total sightings) During summer 2000 ram groups( Nx D 4) were also smaller than ewe groups ( Nx D 21) (z D 1153 N D 273p D 00001 and 10 of ram groups were solitary males (N D 96 Solitaryrams were always Class 2 or older In 1999 39 of solitary rams were Class2 and 61 of were Class 3 or 4 (N D 28I in 2000 100 of solitary ramswere Class 3 or 4 (N D 10 Data from 2000 indicated that ram groups weremade up primarily of older rams with yearling rams (D Class 1) making uponly 7 of all ram groups sighted (N D 50 The percentage of Class 2 andClass 34 rams in ram groups was 12 and 81 respectively

Vegetation transects

We identi ed 72 species of grasses forbs shrubs and cacti on the vegeta-tion transects in most cases to the species level Of total vegetation sampledthe mean percentage cover (sectSEM) per transect was 581 (sect62 shrubs256 (sect45 grass 114 (sect41 forbs and 49 (sect26 cacti We com-pared locations where ram groups and ewe groups were sighted (Fig 1)using the nearest vegetation transect as representative of forage availabil-ity (Fig 2) Rams selected sites with 125 greater total vegetation covercompared with sites selected by ewe groups (Mann-Whitney z D 449

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

192 MOORING ET AL

Fig

1L

ocat

ion

ofra

m(l

)an

dew

e(m

)gr

oup

sigh

ting

sat

the

Red

Roc

kW

ildl

ife

Are

aN

ewM

exic

oJu

ne-J

uly

2000

Not

eth

epr

esen

ceof

ram

grou

psin

the

sout

hwes

tpo

rtio

nof

the

rese

rve

(low

erri

ght

ofm

ap)

and

the

conc

entr

atio

nof

ewe

grou

psar

ound

Ash

Cre

ekB

oxca

nyon

(lef

tcen

ter

ofm

ap)

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 193

Fig 2 Percent cover of total vegetation grasses shrubs forbs and cactus as estimated byDaubenmire transects at sites utilizedby ram (n) versus ewe (u ) groups at Red Rock WildlifeArea summer 2000 Sites selected by ram groups had a signi cantly higher cover of totalvegetation grasses and shrubs in comparison with sites utilized by ewe groups Astericksover the bars indicate a signi cant difference between the percent cover of ram versus ewe

sites by the Mann-Whitney test (indicates p lt 00001 indicates p lt 003

N D 119 p D 00001 The greater vegetation cover at ram sites was dueto greater coverage of grasses (z D 428 N D 119 p D 00001 and shrubs(z D 212 N D 119 p D 003 Grass and shrub cover at ram sites was200 and 108 respectively of that at ewe sites (Fig 2) Ewe groups usedless vegetated sites including rocks bare dirt and litter (z D 196 N D 119p D 005 There was no signi cant difference between ram and ewe groupsin the coverage of forbs and cactus (p gt 017

The southwest corner of the reserve (lower right side of Fig 1) appeared tobe used only by rams as ewes were never sighted in this area We comparedvegetation cover from the two transects within this lsquorams-onlyrsquo area with theother nine transects which were located in areas used by both ram and ewegroups Total vegetation cover in the lsquorams-onlyrsquo transects ( Nx D 19) wasgreater than that found in the other transects ( Nx D 13) (Mann-Whitneyz D 212 N D 11 p D 003 This was mainly due to the lsquorams-onlyrsquo tran-sects having higher grass cover ( Nx D 8) than the other transects ( Nx D 3)(z D 213 N D 11 p D 003 There was no signi cant difference betweenthe lsquorams-onlyrsquo area and the other transects in coverage of forbs shrubs orcactus

Habitat ruggedness and visibility

Ewe groups chose more rugged sites whether ruggedness was estimatedfrom slope or from contour lines on the topographic map (Fig 3) Ewe

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

194 MOORING ET AL

Fig 3 Mean (sectSEM) ruggednessmeasures for terrain inhabited by ram (n) versus ewe (u )groups at Red Rock Wildlife Area summer 2000 lsquoSlopersquo (left) is in degrees from horizontallsquoContoursrsquo (right) are the number of contour lines per 100 pound 100 m quadrat Ewe groupschose more rugged sites whether ruggedness was estimated from slope or from contour lines

(p lt 0007

Fig 4 Mean (sectSEM) visibility (in centimeters) at 20 and 40 m for terrain inhabited byram (n) versus ewe (u ) groups at Red Rock Wildlife Area summer 2000 There was nodifference between ram and ewe groups in habitat visibility at 20 m but rams were found in

areas with greater visibility at 40 m (p lt 005

sites had steeper slopes than those utilized by rams (Ewes Nx D 234plusmnRams Nx D 198plusmn Mann-Whitney z D 286 N D 119 p D 0004 andquadrats where ewe groups were sighted had more contour lines bisectingthem (Ewes Nx D 41 Rams Nx D 34 z D 270 N D 119 p D 0007Although there was no difference between the sexes in habitat visibility at20 m rams were found in areas with signi cantly greater visibility at 40 m(Fig 4) compared with sites used by ewe groups (Ewes Nx D 27 cm RamsNx D 34 cm z D 197 N D 119 p D 005 There was however asigni cant negative correlation between slope and visibility for the 46 habitatquadrats (Spearman rank correlation rs D iexcl030 N D 46 p D 004indicating that more rugged sites tended to have less visibility

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 195

Distance to water

On average ram groups were found more than twice as far from standingwater as ewe groups (Mann-Whitney z D 457 N D 119 p D 00001 Themean distance to water for rams was over 1 km ( Nx D 1053 m) compared toless than half a kilometer for ewe groups ( Nx D 486 m) When mixed groupswere included in the analysis there was also a highly signi cant difference inthe distance to water among ram ewe and mixed groups (Ranked ANOVAF2160 D 142 p D 00001 but the distance between mixed groups and freewater ( Nx D 490 m) did not differ from that of ewe groups (Scheffe multiplecomparisons p D 097 Even when the windmills were entered into theanalysis as a possible water source ram groups were still farther from water( Nx D 496 m) compared with ewe groups ( Nx D 347 m) (Mann-Whitneyz D 335 N D 119 p D 0001 and mixed groups ( Nx D 387 m) were notsigni cantly different from ewe groups (p D 071

Activity budgets

Males and females in segregated groups

Based upon the conservative analysis of daily means the amount of timedevoted to foraging moving and lying down did not differ signi cantly be-tween males in ram groups and females in ewe groups within 1999 and 2000(Table 2) Foraging and standing time were much different between 1999 and2000 probably because heavy pre-rut activity in 1999 (when the populationcontained 55 males and the sex ratio of 141100 was strongly skewed in favorof males) resulted in harassment of females Males in ram groups spent twiceas much time vigilant compared with females in ewe groups The proportionof time devoted to ruminating did not differ signi cantly between males inram groups and females in ewe groups nor was there any difference whenmales and females in mixed groups were included (z D 057 N D 56p D 057 Post hoc power analysis indicated that the effect size (d) forthe observed mean activity differences ranged from d D 12 to d D 50 Thepower of this test to detect a large effect size (d D 080) for the actual samplesizes of N D 25 45 and 73 was 047 074 and 092 respectively (Cohen1988)

Because most of the activity budget analyses resulted in no signi cantdifference between rams and ewes we repeated the tests treating each focalobservation as a data point thus in ating the sample size so as to avoid a

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

196 MOORING ET AL

TABLE 2 Mean (sectSEM) activity budgets by mature males in ram groupsand adult females in ewe groups at Red Rock Wildlife Area during summer

1999 and 2000

Percent Scans Year Rams1 Ewes z2 p2

Foraging 1999 111 (sect17 105 (sect17 003 0982000 394 (sect64 379 (sect52 027 078

Standing 1999 143 (sect19 119 (sect14 087 0392000 29 (sect11 72 (sect26 203 004

Moving 1999 71 (sect10 77 (sect10 039 0702000 57 (sect20 58 (sect09 090 037

Lying down 1999 271 (sect38 296 (sect36 060 0552000 200 (sect68 346 (sect69 191 006

Ruminate 2000 292 (sect136 454 (sect125 038 043Vigilant 2000 319 (sect71 146 (sect34 200 004

1 1999 Rams N D 37I Ewes N D 36 2000 Rams N D 20I Ewes N D 25I RuminateRams N D 10I Ewes N D 152 Mann-Whitney test

Type II error The power of these tests to detect a large effect size (d D 080)was 099 (Cohen 1988) Even with this more liberal analysis the resultswere very similar to the analysis based on daily means For both summersthe amount of time devoted to foraging did not differ signi cantly betweenrams and ewes (Mann-Whitney 1999 z D 027 N D 636 p D 078 2000z D 042 N D 273 p D 067 Rams moved more than ewes in 1999(z D 213 N D 636 p D 003 but not in 2000 (z D 059 N D 273p D 056I ewes spent more time lying down than rams in 1999 (z D 250N D 636 p D 001 but not in 2000 (z D 126 N D 273 p D 021I andewes stood more than rams in 2000 (z D 192 N D 273 p D 005) butnot in 1999 (z D 112 N D 636 p D 026 Again during summer 2000rams were twice as vigilant as ewes (z D 378 N D 273 p D 00001 andrumination time did not differ between rams and ewes (z D 021 N D 424p D 084 Thus using both a very conservative and a more liberal analysismales and females in segregated groups did not differ in activity budgets inany of the ways predicted by the lsquoactivity budget hypothesisrsquo (Table 1) Timeforaging and ruminating was non-signi cant by all analyses and differencesin time spent moving or reclining were in the opposite direction predicted bythe hypothesis

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 197

Yearling males

The lsquoactivity budget hypothesisrsquo predicts that smaller-bodied subadult maleswill forage more like females than like mature males (Table 1) Bighornyearling males in ram groups did not differ from mature males in ram groupsin time spent foraging moving or lying down (Feed z D 086 N D 25p D 040I Move z D 059 N D 25 p D 060I Lying down z D 062N D 25 p D 060I nor did yearling rams in ewe groups differ fromadult females in ewe groups in those activities (Feed z D 054 N D 44p D 059I Move z D 144 N D 44 p D 015I Lying down z D 069N D 44 p D 049 There was no effect of group type on the time devotedto foraging moving or lying down by yearling males (Ranked ANOVAFeed F240 D 061 p D 055I Move F240 D 224 p D 012I Lying downF240 D 078 p D 047 The power of these conservative tests to detecta large effect size ranged between 037 to 073 depending on sample size(Cohen 1988)

When activity data of yearling males was analyzed with all focal samplesyearlings again did not differ from older males in ram groups or adult femalesin ewe groups in the time spent foraging or moving (Male yearlings vs ramsFeed z D 104 N D 237 p D 030I Move z D 051 N D 237 p D 061IMale yearlings vs ewes Feed z D 064 N D 334 p D 052I Movez D 005 N D 334 p D 096 nor did they differ from ewes in timelying down (z D 077 N D 334 p D 044 Yearling males did spend moretime lying down than mature males in ram groups (z D 194 N D 237p D 005 Again there was no effect of group type on the time devoted toforaging moving or lying down by yearling males (Ranked ANOVA FeedF2146 D 009 p D 091I Move F2146 D 122 p D 030I Lying downF2146 D 062 p D 054 The power of these more liberal tests to detecta large effect size was 099 for all sample sizes (Cohen 1988) Thereforeaccording to both the conservative and liberal analyses yearling males didnot differ from mature males (or adult females) in activities predicted by thelsquoactivity budget hypothesisrsquo (Table 1)

Foraging selectivity

The number of steps taken per minute foraging (a measure of foraging selec-tivity) was recorded for rams and ewes in mixed groups feeding in the samevegetation communities Comparison of ewes with all mature rams (Class

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

198 MOORING ET AL

2 3 4) indicated no signi cant difference in steps per foraging min (Mann-Whitney z D 121 N D 65 p D 023 In fact the mean number of stepsper min made by foraging rams ( Nx D 14) was numerically greater than thatfor ewes ( Nx D 11) When the analysis was con ned to the largest Class 3and 4 rams there was still no signi cant difference in foraging selectivitybetween rams and ewes (z D 093 N D 55 p D 035 with mean stepsper min remaining the same for rams and ewes The power of these tests todetect a large effect size (d D 080) was 083-089 depending on samplesize (Cohen 1988) It thus appears unlikely that ewes were more selectivethan rams

Discussion

Sexual segregation

Sexual segregation occurred in bighorn sheep on social temporal and spatiallevels Prior to the rut during summers 1999 and 2000 mature males wereusually found in ram groups that were temporally or spatially segregatedfrom ewe groups Although there was a good deal of spatial overlap betweenram and ewe groups ram groups ranged more widely and often occurred inlsquoexclusiversquo male areas where ewe groups were never sighted When ramswere found in mixed groups they appeared to be primarily interested inassessing the reproductive status of ewes as indicated by frequent snif ngand ehmen behavior and attempted mounting When ewes failed to exhibitsexually receptive behavior rams left the mixed groups and segregated onceagain into ram groups (for summers 1999 and 2000 rutting did not occuruntil after observations were completed)

Reproductive strategy-predation risk hypothesis

The lsquoreproductive strategy-predation risk hypothesisrsquo was supported by ourdata As expected by the rst prediction ram groups used sites with moreabundant grasses and shrubs compared to sites utilized by ewe groups Ramsites had 25 greater vegetation cover including more grass and shrubsGiven that ram groups were 2-5 times smaller than ewe groups segregatedmales had about 50-125 more forage available per individual than did fe-males in ewe groups Rams may also have had more high-quality forage

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 199

available to them During the summer browse leaves have the highest crudeprotein and in vivo digestibility values followed by forbs with grasses hav-ing the lowest protein and digestibility values (Schwartz amp Hobbs 1985)Although this picture is complicated by the fact that secondary plant com-pounds (highest in browse and forbs) have the potential to reduce digestibil-ity and protein availability while biogenic silica (found in grasses) can re-duce digestibility and lengthen rumen retention times (Hobbs 1987 Robbinset al 1987) it is generally true that browse leaves and forbs represent thehighest quality forage (Cook 1972 Schwartz amp Hobbs 1985 Seegmilleret al 1990) Our results indicated that males in ram groups had more high-quality browse available to them than did females in ewe groups and giventhat ewes were not more selective than rams it is likely that rams consumedmore high-quality browse Although ram sites contained on average about8 more shrubs than ewe sites even small differences in diet quality canhave a profound effect on ruminant productivity through a multiplier effect(White 1983)

The second prediction of the lsquoreproductive strategy-predation risk hypoth-esisrsquo was that females should minimize predation risk to themselves and theiroffspring by choosing rugged escape terrain According to this hypothesismales can better exploit areas of greater predation risk with more abundantforage than females because their larger size makes them less vulnerable topredators That bighorn ewes are more vulnerable to predation is indicated bythe fact that females were more likely to ee from coyotes than males (Ble-ich 1999) Whether terrain ruggedness was measured by mean slope or bythe number of contour lines per quadrat ewe groups were found in areas thatwere about 20 more rugged than ram group sites Visibility is an impor-tant habitat feature for bighorn because predators are detected visually Ewesites had less visibility at 40 m than ram groups thus reducing the abilityof ewes to detect a predator Given the negative correlation between rugged-ness and visibility it would appear that ewes selected habitat on the basisof ruggedness rather than visibility indicating that rugged escape terrain ismore important in reducing predation risk (Fairbanks et al 1987) Becauseforaging ef ciency is reduced when sheep are in habitats with obstructedvisibility (Risenhoover amp Bailey 1985) members of ewe groups probablyincurred a cost in forage intake in the less open terrain they selected

Overall ewes were found in safer terrain than males and thus our resultssupport the prediction that ewes seek to minimize predation risk By choos-ing more precipitous terrain ewes and their accompanying lambs would be

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

200 MOORING ET AL

likely to increase survivorship Mountain lion predation was found to be themajor mortality factor for desert bighorn populations in California and NewMexico (Hoban 1990 Wehausen 1996 Hayes et al 2000 Rominger ampWeisenberger 2000) Most bighorn lambs die during their rst few weeksoften to predation and mortality during the rst year of life can be as highas 40 to 90 (Krausman amp Shackleton 2000) Of the 179 mortalities atRRWA from 1972-1999 80 (45) were lambs many of whom are presumedto have suffered predation During the summers of 1999 and 2000 we per-sonally examined carcasses of 6 bighorns that were killed by mountain lionsat RRWA (4 rams 1 ewe and 1 lamb) and all of them were found gt100 maway from cliffs in less rugged terrain Two additional sheep (both rams)are known to have succumbed to mountain lion predation outside of summerduring 1999-2001 making rams 75 of lion kills during our study period

Several previous studies of bighorn sheep support the contention of thelsquoreproductive strategy-predation risk hypothesisrsquo that males can compromisesecurity in favor of food while females can compromise food in favor ofsecurity Male desert bighorn (O c nelsoni) in the Mojave Desert of Cal-ifornia occurred on ranges with greater forage availability and obtained ahigher quality diet made up of more shrubs while females and young se-lected steeper more open habitat and ate more low-quality grasses than didmales (Bleich et al 1997) Also in agreement with our data predator den-sity was highest in areas used predominately by desert bighorn rams in theMojave Desert and males predominated among predator kills (Bleich et al1997) Pregnant Rocky Mountain bighorn ewes (O c canadensis) in Albertamigrated to high-elevation lambing areas with lower predation risk despitehaving to forfeit high-quality forage (Festa-Bianchet 1988) thus trading offfood quality for increased security

In agreement with the third prediction of the lsquoreproductive strategy-predation risk hypothesisrsquo bighorn females occurred closer to water sourcesthan males in ram groups The nding that mixed groups were not signif-icantly farther from free water than ewe groups indicates that distance towater was controlled by ewes During July and August bighorn ewes andyoung of the Santa Rosa Mountains in California stayed closer to waterholesthan did rams (Blong amp Pollard 1968 cited in Simmons 1980) Desertbighorn ewes in the Mojave Desert of California were also found signi -cantly closer to sources of free water during sexual segregation than rams(Bleich et al 1997) Bighorn ewes begin using water earlier in the season

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 201

than rams and lactating ewes drink more frequently than dry ewes (Turneramp Weaver 1980) indicating that both lactation and summer conditions in-crease the water requirements of ewes Bowyer (1984) concluded that sexualsegregation in mule deer (Odocoileus hemionus) may be the result of theincreased water requirements of lactating females combined with decreasedwater loss in larger males That ewes are consistently found closer to waterthan rams during the summer suggests that females have a greater require-ment for free water which constrains the movement of ewes more than ramsand encourages spatial segregation of the sexes

Sexual dimorphism-body size hypothesis

Large-bodied ruminant males are capable of exploiting more brous lower-quality forage than smaller females due to larger rumen volume and lowermass-speci c metabolic rate (Van Soest 1982) This aspect of rumen physi-ology forms the basis for the predictions of the lsquosexual dimorphism-body sizehypothesisrsquo that males will consume abundant lower-quality forage while fe-males will selectively feed on sparse higher-quality forage Although we didnot measure diet selection there was no indication that rams selected lower-quality forage compared with ewes In fact the low-quality grasses (26 ofvegetation transects) were less abundant than the high-quality browse (58)Based on the greater abundance of shrubs in the vegetation transects of ramgroups it is likely that rams consumed more high-quality browse If ramsdid not consume more brous low-quality forage this would explain whythey did not engage in longer ruminating bouts than ewes

Why were the predictions of the lsquosexual dimorphism-body size hypothe-sisrsquo not supported although rumen physiology allows larger-bodied animalsto eat higher- ber foods That rams can eat lower-quality forage does notmean that they do so or should be expected to In fact given the importanceof body size and condition for maximizing mating success in male bighornsit would be maladaptive for rams to consume less nutritious forage whenhigh-quality browse is available In order to obtain the size strength andenergy reserves required to win high dominant status and thus gain accessto estrous ewes rams would be expected to select the best-quality forageavailable Males should only select low-quality forage when the abundanceof high-quality food is inadequate but high-quality browse from shrubs wasabundant at our study site during summer months

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

202 MOORING ET AL

Activity budget hypothesis

The lsquoactivity budget hypothesisrsquo argues that large males cannot remain withsmaller females due to differences in activity budgets that result from bodysize differences in digestive physiology Speci cally it predicts that maleswill spend more time lying down and ruminating (to digest a high- ber diet)while females will spend more time foraging and walking (because they aremore selective) We have no indication from the behavioral data that desertbighorn rams and ewes could not synchronize their activities The proportionof the activity budget devoted to foraging and moving did not differ betweenrams and ewes in sexually segregated groups nor was there a difference inforaging selectivity by males and females when together in mixed groupsThere was no difference in time spent lying down and ruminating betweenthe sexes

These results differ from those of Ruckstuhl (1998) who found that RockyMountain bighorn ewes (O c canadensis) in Alberta during spring-summerspent a greater percentage of time foraging and walking compared with ramswhile rams devoted more of their activity budget to lying down than ewesVan Dyke (1978) also found that California bighorn rams (O c californiana)in Oregon spent less time foraging and more time resting in the summer thanewes However as in this study Ruckstuhl (1998) failed to nd any differ-ence between the sexes in step rate while foraging Differences in the veg-etational communities available to desert and northern bighorn in the sum-mer might explain the activity budget differences between these populationsThe biomass of forage plants available is less in the arid regions inhabitedby desert bighorn compared with northern mountains If desert bighorn ramscan eat as much or more high quality forage as ewes signi cant differencesin foraging and ruminating time would not be expected between desert ramsand ewes

If body size dimorphism between males and females leads to sexual dif-ferences in foraging behavior that cause incompatible activity budgets thelsquoactivity budget hypothesisrsquo predicts that the larger the difference in bodysize the more activity budgets should differ between the sexes (Ruckstuhlamp Neuhaus 2000) Thus a further prediction is that subadult males (similarin size to adult females) should exhibit foraging behavior closer to that ofewes than mature rams Our data revealed that yearling rams at RRWA didnot differ from mature males in ram groups or adult females in ewe groups

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 203

in the time spent foraging or moving similar to the ndings of Ruckstuhl(1998) There was no effect of group type on the percentage of time devotedto foraging moving or lying down by yearling males whether in ewe ramor mixed groups

Competition avoidance hypothesis

The lsquocompetition avoidance hypothesisrsquo proposes that males avoid competi-tion with potential mates and offspring by segregating themselves from fe-male herds The rst prediction that females will inhabit sites with moreabundant forage than that found at male sites was not supported The secondprediction that the largest most dominant males are more likely to segregatethan smaller subordinate males was supported by our data Ram groups con-sisted primarily of Class 2 3 or 4 rams with yearling (Class 1) rams beingrarely sighted (7 of ram group sightings) Class 3 and 4 rams which are thelargest and most dominant made up 81 of all ram groups in 2000 How-ever other hypotheses would also predict the largest males be most likely tosegregate The largest rams would be least vulnerable to predation and mosttolerant of foraging sites away from escape terrain (lsquoreproductive strategy-predation risk hypothesisrsquo) and the largest rams would have the greatest abil-ity to consume brous low-quality forage (lsquosexual dimorphism-body sizehypothesisrsquo) In short our results do not support competition avoidance as amechanism for sexual segregation in bighorn sheep

Conclusions

Our results support the lsquoreproductive strategy-predation riskrsquo explanationof sexual segregation During the summer months desert bighorn femalesappeared to maximize their tness by minimizing the risks of predationto themselves and their offspring by restricting their movements to morerugged areas with access to adequate food and water for lactation On theother hand males maximized their tness by roaming widely and feedingin forage-rich habitats that enhanced their body condition and probabilityof mating success As suggested by Bleich et al (1997) sexual segregationin bighorn sheep may best be seen as the result of a compromise betweenoptimal foraging and predator avoidance Our results did not support thelsquosexual dimorphism-body size hypothesisrsquo lsquoactivity budget hypothesisrsquo or

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

204 MOORING ET AL

lsquocompetition avoidance hypothesisrsquo as explanations of sexual segregation indesert bighorns

Because animals have evolved to deal successfully with many differentselection pressures it is unlikely that sexual segregation as a general phe-nomenon can be explained by a single hypothesis except perhaps for theoverarching in uence of sexual body size dimorphism on energetics pre-dation risk and behavior The difference in activity budgets between con-speci c populations of northern and desert bighorn suggests that a unitaryexplanation for sexual segregation in all ungulates should not be expectedDifferent hypotheses or predictions may apply to varying degrees at differ-ent stages of the reproductive cycle or in different seasons For example theneed for females to have access to water should be most crucial during lacta-tion while the trade-off between food and security for ewe groups should bemost favored during the period when young are most vulnerable to predationEcological parameters such as predation pressure and forage qualityquantitymay vary seasonally The factors driving sexual segregation in the summercould in principle be different in other seasons Future studies of sexual seg-regation in polygynous ungulates will need to address the multiple selectionfactors that differentially in uence the sexes while taking into considerationthe dynamic nature of these factors across populations habitats reproductivecycles and seasons of the year

References

Altmann J (1974) Observational study of behavior sampling methods mdash Behaviour 49p 227-267

Bartolome J Franch J Gutman M amp Seligman NG (1995) Physical factors that in u-ence fecal analysis estimates of herbivore diets mdash J Range Manag 48 p 267-270

Berger J (1990) Persistence of different-sizedpopulationsan empirical assessment of rapidextinctions in bighorn sheep mdash Conserv Biol 4 p 91-98

Bleich VC Wehausen JD amp Holl SA (1990) Desert-dwelling mountain sheep conser-vation implications of a naturally fragmented distribution mdash Conserv Biol 4 p 383-390

mdash mdash (1999) Mountain sheep and coyotes patterns of predator evasion in a mountain ungu-late mdash J Mammal 80 p 283-289

mdash mdash Bowyer RT amp Wehausen JD (1997) Sexual segregation in mountain sheep re-sources or predation mdash Wildlife Monogr 134 p 1-50

Bowyer RT (1984) Sexual segregation in southern mule deer mdash J Mammal 65 p 410-417

SEXUAL SEGREGATION IN BIGHORN 205

Byers JA (1997) American pronghorn social adaptations and the ghost of predators pastmdash University of Chicago Press Chicago

Cohen J (1988) Statisticalpower analysis for the behavioral sciences2nd edn mdash LawrenceErlbaum Associates Hillsdale New Jersey

Conradt L (1998) Could asynchrony in activity between the sexes cause intersexual socialsegregation in ruminants mdash Proc Roy Soc London B 265 p 1359-1363

Cook CW (1972)Comparativenutritivevalues of forbs grasses and shrubs mdash In Wildlandshrubs mdash their biology and utilization(CM McKell JP Blaisdell amp JR Goodin eds)USDA Forest and Range Experiment Station Technical Report INT-1 Ogden Utahp 303-310

Daubenmire R (1959) A canopy-coverage method of vegetational analysis mdash NorthwestScience 33 p 43-64

Fairbanks WS Bailey JA amp Cook RS (1987) Habitat use by a low elevation semi-captive bighorn sheep population mdash J Wildl Manag 51 p 912-915

Faul F amp Erdfelder E (1992) GPOWER A priori post-hoc and compromise power analy-ses for MS-DOS [Computer program] mdash Bonn University Department of PsychologyBonn FRG

Festa-Bianchet M (1988) Seasonal range selection in bighorn sheep con icts betweenforage quality forage quantity and predator avoidance mdash Oecologia 75 p 580-586

Geist V (1971) Mountain sheep a study in behaviorand evolutionmdash Universityof ChicagoPress Chicago

mdash mdash amp Petocz RG (1977) Bighorn sheep in winter do rams maximize reproductive tnessby spatial and habitat segregation from ewes mdash Can J Zool 55 p 1802-1810

Hansen CG amp Deming OV (1980) Growth and development mdash In The desert bighornits life history ecology and management (G Monson amp L Sumner eds) University ofArizona Press Tucson p 152-171

Hass CC (1990) Alternative maternal-care patterns in two herds of bighorn sheep mdashJ Mammal 71 p 24-35

Hayes CL Rubin ES JorgensonMC Botta RA amp Boyce WM (2000)Mountain lionpredation of bighorn sheep in the PeninsularRanges Californiamdash J Wildl Manag 64p 954-959

Hoban PA (1990) A review of desert bighorn sheep in the San Andres Mountains NewMexico mdash Trans Desert Bighorn Council 34 p 14-22

Hobbs NT (1987) Fecal indices to dietary quality a critiquemdash J Wildl Manag 51 p 317-320

Hogg JT (1987) Intrasexual competition in mate choice in Rocky Mountain bighorn sheepmdash Ethology 75 p 119-144

Holechek JL amp Valdez R (1985) Magni cation and shrub stemmy material in uences onfecal analysis accuracy mdash J Range Manag 38 p 350-352

Krausman PR Leopold BD Seegmiller RF amp Torres SG (1989) Relationships be-tween desert bighorn sheep and habitat in western Arizona mdash Wildlife Monogr 102p 1-66

mdash mdash Sandoval AV amp Etchberger RC (1999) Natural history of desert bighorn sheep mdashIn Mountain sheep of North America (R Valdez amp PR Krausman eds) University ofArizona Press Tucson p 139-191

mdash mdash amp Shackleton DM (2000) Bighorn sheep mdash In Ecology and management of largemammals in North America (S Demarais amp PR Krausman eds) Prentice-Hall UpperSaddle River p 517-544

206 MOORING ET AL

Loison A Gaillard J-M Pelabon C amp Yoccoz NG (1999) What factors shape sexualsize dimorphism in ungulates mdash Evol Ecol Res 1 p 611-633

Main MB Weckerly FW amp Bleich VC (1996) Sexual segregation in ungulates newdirections for research mdash J Mammal 77 p 449-461

Miller GD amp Gaud WS (1989) Composition and variabilityof desert bighorn sheep dietsmdash J Wildl Manag 53 p 597-606

Mooring MS (1995) Effect of tick challenge on grooming rate by impala mdash Anim Behav50 377-392

Mysterud A (2000) The relationshipbetween ecological segregationand sexual size dimor-phism in large herbivores mdash Oecologia 124 p 40-54

Norusis MJ (1998) SPSS 80 guide to data analysis mdash Prentice-Hall Upper Saddle RiverPerez-Barberia FJ amp Gordon IJ (1998) The in uence of sexual dimorphism in body size

and mouth morphology on diet selection and sexual segregation in cervids mdash Acta VetHung 46 p 357-367

mdash mdash mdash mdash amp Pagel M (2002) The origins of sexual dimorphism in body size in ungulatesmdash Evolution 56 p 1276-1285

Risenhoover KL amp Bailey JA (1985) Foraging ecology of mountain sheep implicationsfor habitat management mdash J Wildl Manag 49 p 797-804

Robbins CT Hanley TA Hagerman AE Hjeljord O Baker DL Schwartz C ampMautz WW (1987) Role of tannins in defending plants against ruminants reductionin protein availability mdash Ecology 68 p 98-107

Rominger EM (1998) Status of desert bighorn sheep in New Mexico 1997 mdash TransDesert Bighorn Council 42 p 50-52

mdash mdash amp Weisenberger ME (2000) Biological extinction and a test of the ldquoconspicuousindividual hypothesisrdquo in the San Andres Mountains New Mexico mdash Transactions ofthe 2nd North American Wild Sheep Conference April 6-9 1999 Reno NV p 293-307

Ruckstuhl KE (1998) Foraging behaviour and sexual segregation in bighorn sheep mdashAnim Behav 56 p 99-106

mdash mdash amp Neuhaus P (2000) Sexual segregation in ungulates a new approach mdash Behaviour137 p 361-377

mdash mdash amp mdash mdash (2002) Sexual segregation in ungulates a comparative test of three hypothesesmdash Biol Rev 77 p 77-96

Schwartz CC amp Hobbs NT (1985) Forage and range evaluation mdash In Bioenergetics ofwild herbivores (RJ Hudson amp RG White eds) CRC Press Boca Raton p 25-51

Seegmiller RF Krausman PR Brown WH amp Whiting FM (1990) Nutritional compo-sition of desert bighorn sheep forage in the Harquhala Mountains Arizona mdash DesertPlants 10 p 87-90

Shackleton DM Shank CC amp Wikeem BM (1999) Natural history of Rocky Mountainand California bighorn sheep mdash In Mountain sheep of North America (R Valdez ampPR Krausman eds) University of Arizona Press Tucson p 78-138

Siegel S amp Castellan NJ (1988) Nonparametric statistics for the behavioral sciences 2ndedn mdash McGraw-Hill New York

Simmons NM (1980) Behavior mdash In The desert bighorn its life history ecology andmanagement (G Monson amp L Sumner eds) University of Arizona Press Tucsonp 124-144

SEXUAL SEGREGATION IN BIGHORN 207

Turner JC amp Weaver RA (1980) Water mdash In The desert bighorn its life history ecologyand management (G Monson amp L Sumner eds) University of Arizona Press Tucsonp 100-112

Valdez R amp Krausman PR (1999) Description distribution and abundance of moun-tain sheep in North America mdash In Mountain sheep of North America (R Valdez ampPR Krausman eds) University of Arizona Press Tucson p 3-22

Van Dyke WA (1978) Population characteristics and habitat utilization of bighorn sheepSteens Mountain Oregon mdash MSc thesis Oregon State University Corvallis

Van Soest PJ (1982) Nutritional ecology of the ruminant mdash O amp B Books CorvallisWeckerly FW (1998) Sexual-size dimorphism in uence of mass and mating systems in the

most dimorphic mammals mdash J Mammal 79 p 33-52Wehausen JD (1996) Effects of mountain lion predation on bighorn sheep in the Sierra

Nevada and Granite Mountains of California mdash Wildl Soc Bull 24 p 471-479White RG (1983) Foraging patterns and their multiplier effects on productivity of northern

ungulates mdash Oikos 40 p 377-384Wikeem BM amp Pitt MD (1991) Grazing effects and range trend assessment on California

bighorn sheep range mdash J Wildl Manag 44 p 466-470mdash mdash amp mdash mdash (1992) Diet of California bighorn sheep Ovis canadensis californiana in

British Columbia assessing optimal foraging habitat mdash Can Field-Nat 106 p 327-335