Effects of northern pike on patterns of nest use and reproductive

Behavioral Ecology Vol. 8 No. 6: 655-662

Effects of northern pike on patterns of nestuse and reproductive behavior of male fatheadminnows in a boreal lake

Hilary M. Jones and Cynthia A. PaszkowskiDepartment of Biological Sciences, University of Alberta, Edmonton, Alberta, T6G 2E9, Canada

We conducted a two-part study to assets predator avoidance by reproductive male fathead minnows (Pimephates prxmelas)subjected to predation threat from northern pike (Esox hicxus). First, we determined if patterns of nest use by egg-guardingmale minnows in a boreal lake were related to pike densities. We sampled northern pike and identified four areas of "highpike-density" and three areas of "low pike-density." We censused natural nests and placed nest boards in these areas. We foundeggs on natural nests more frequently in areas with low densities of pike than in areas with high densities of pike. However, wecould not fuDy explain the distribution of nests by predation risk. Second, we evaluated the behavioral response of egg-guardingmales to a control stimulus (a piece of wood) or a live pike in a wire cage. We used time to return to the nest after a stimulusas a measure of risk taking. Males took different amounts of risk based on predation threat; males in the predator treatmenttook longer to return to their nests than control males. Risk taking was not related to the number or age of the eggs but todistance to nearest egg-guarding neighbor; males with close neighbors returned sooner than more isolated males. Males in thepredator treatment had lower total activity and egg rubbing than control males after they returned to their nests. We concludethat male fathead minnows altered their reproductive behavior in ways that reduced predation risk, but the cost of predatoravoidance might include egg predation, lost mating opportunities, or usurpation of nests. Kty words: egg-guarding, encounterrate, Esox hichis, Pimephalts promelas, predation. [Behav Ecol 8:655-662 (1997)]

T)n;dator avoidance is a significant factor influencing ani-X. mal decision making. It affects habitat use, feeding pat-terns, morphology, and growth of prey (Craig, 1994; Dill andFraser, 1984; Godin and Crossman, 1994; Helfinan, 1986;lima and Dill, 1990; Sih, 1980). The reduced vulnerabilitygained by avoidance may sacrifice food or mating opportu-nities (e.g.,Werner and Hall, 1988). Predator-mediated habitatshifts made while foraging have been documented (Werneret al., 198S), but there is a paucity of data on how predatorsinfluence choice of breeding territory and nest use by prey(but see McKaye, 1984). Some studies have documentedchanges in reproductive behavior in the presence of predators(Magurran and Seghers, 1990; Reznick and Endler, 1982; Sih,1988), but data are sparse here as well (Magnhagen, 1990,1991).

During reproduction, an animal may be more vulnerableto predation because of mate attraction tactics (e.g., callingor displaying), egg-bearing, courtship, and parental behaviors(Magnhagen, 1991; Peckarsky et aL, 199S; Ryan, 1985; Svens-son, 1988; Wing, 1988). Increased vulnerability to predatorsduring reproduction may lead to mortality, but die nonlethaleffects of predators are not so obvious. They may lead tochanges in reproductive success of individuals and prey pop-ulations as a whole (Dill, 1987; Fraser and Gilliam, 1992; Rez-nick and Endler, 1982; Sih et aL, 1990).

The fathead minnow (PimtphaUs pnmdas) is a common,small-bodied fish found in lakes of western Panada and is es-pecially vulnerable to piscivores (Gillen et al., 1981; Moody etal, 1983; Robinson, 1989; Robinson and Tonn, 1989). Fromlate May to August, male fathead minnows move inshore(Price et aL, 1991) and establish territories around nests inthe littoral zone. Males use the undersides of logs, rocks, ordebris as nests. At this time, males become relatively site at-

Received 28 December 1995; accepted 17 April 1997.1045-2S49/97/J5.00 O 1997 International Society for Behariora] Ecology

tached, while females and juveniles remain mobile. Malesmate with one or more females and defend eggs until hatch-ing (7-30 days depending on water temperature). Males de-fend territories and eggs from other males that compete fornests and from females and juveniles that are egg predators.During this period, males develop dark nuptial coloration andtubercles (McMillan and Smith, 1974). After the eggs hatch,males die. Because nests are easily found and observed andbecause males exhibit a variety of nesting behaviors, the fat-head minnow is an ideal species for determining the influ-ence of predation on location of nests and on reproductivebehavior.

The northern pike {Esox tudus) is a dominant pisovore inboreal lakes. Pike occupy shallow, vegetated areas and are sol-itary, ambush predators that preferentially eat fathead min-nows (Mauck and Coble, 1971; Moody et aL, 1983; Robinson,1989; Wahl and Stein, 1988). Presumably, site attachment, in-creased coloration, and conspicuous territorial behaviorsmake egg-guarding male fathead minnows vulnerable tonorthern pike. Further, territorial male fathead minnows areusually aggregated. Northern pike, which are efficient huntersaround prey refuges in the littoral zone (Ekldv and Diehl,1994), may cue in on local aggregations and exploit them, aspike consume concentrated patches of food more intensivelythan food that is evenly distributed (Ivlev, 1961).

When an egg-guarding male fathead minnow is confrontedby a northern pike, his situation differs from that experiencedby many other vertebrates with parental care. The directthreat of predation applies only to the adult minnow; the pikewill not eat die eggs. In this respect, die situation is differentfrom that facing many species of squamate reptiles, birds, ormimnnli where the same predator will not only injure or killnesting adults but will also take eggs or young. A hypotheticalfathead minnow nest containing several clutches of eggs thatare 5 days old likely represents the lifetime reproductive out-put of a male minnow. If a male fathead minnow abandonshis nest to move beyond the striking distance of a pike or to

Dow

nloaded from https://academ

ic.oup.com/beheco/article/8/6/655/208378 by guest on 02 January 2022

696 Behavioral Ecology Vol. 8 No. 6

conceal himself in a highly structured patch of habitat, theseirreplaceable eggs become temporally vulnerable to predationby invertebrates and by other small fishes, including othermale fathead minnows that may attempt to usurp the nest site(Unger, 1985). Why should a male leave his nest at all? Thesurvival of the eggs to hatching requires his survival. The eggsof a male eaten by a pike will be exposed continuously to thepredation threats described above, as well as to bacterial andfungal infections normally kept in check by the male's clean-ing behavior and removal of dead eggs.

We conducted a two-part study to assess predator avoidanceby fathead minnows at two scales in a lake in Alberta, Canada.At the population level, we predicted that the threat of pre-dadon by pike was an important factor determining lakewidepatterns of nest location by male fathead minnows. To testthis, we determined if the number of nests with eggs foundon natural and artificial sites was related to local density ofpike. At the individual level, we predicted that a male fatheadminnow guarding eggs would alter his behavior to reduce pre-dation risk when confronted with northern pike near his nest.For example, a male may reduce the frequency of egg clean-ing in the presence of a pike. To test this, we presented egg-guarding minnows with caged pike and recorded the behav-ioral response of males to predation threat. We also measuredphysical variables (e.g., pH) and nest characteristics (e.g., ageof eggs) to determine how external factors influenced males'decisions to remain or abandon their eggs.

METHODS

Predation risk and nest locations

Density and distribution of northern pike in Armstrong LakeArmstrong Lake is a 230-ha, shallow, day-bottomed lake (max-imum depth 4.5 m) in central Alberta, Canada (54°24' N119*39' W) containing pike, fathead minnow, brook stickle-back (Culea inconstans), and white sucker (Calastomus ccm-mersom). The lake has an extensive littoral zone with emer-gent vegetation including cattail (Typha latifotia), bulrush(Sctrpus spp.), giant bur-reed (Sporangium eucarpum), andyellow pond lily (Nuphar variegatum). Zones deeper than 1m are dominated by pondweed (Potamogeton spp.) and coon-tail (Ceratophylhtm dmersum).

Local residents reported that densities of northern pikewere very low in Armstrong Lake owing to overwinter mor-tality caused by low oxygen conditions. Because low numbersof pike would have translated into a minimal predation threatfor minnows in the lake, we wanted to estimate the density ofpike. To determine the overall density of northern pike inArmstrong Lake and to identify areas of relatively high andlow densities of pike for experimental work (see below), weused a mark, release, and recapture technique (Blower et al.,1981). In 1994, we captured fish by gill netting from the firstday the lake had open water (27 April) to the first day of icecover (21 October).

We used a random stratified design of sampling locationsfor pike to minimize variability between samples (Schaeffer etal., 1986). Stratified sampling is a common technique for es-timating space use and abundance of northern pike (Hilbornand Walters, 1992). We chose 13 locations for sampling [min-imum of 300 m apart, 1-3 m depth, 3 m from shore, 70 X 10m across (area required to set one gill net)]. Locations werechosen based on their accessibility and presence of represen-tative foatopos such as substrate type and vegetation; we at-tempted to include representatives of all habitat types foundin Armstrong Lake (e.g., fallen trees, beaver lodges, sandyshore). These locations maximized information on pike hab-itat use within the lake relative to structural features. We mea-

sured pH, depth, dissolved oxygen, and temperature in eachlocation every month to determine if locations were physicallyand chemically similar.

Northern pike are thought to be crepuscular (Christiansen,1976; Ivanova, 1969), so we set nets twice daily, at dawn anddusk, for 3 h. A short set was desirable because it minimizedmortality of northern pike; minimal loss of the populationduring sampling is essential for mark, release, and recapturemethods (Krebs, 1989). Of the IS locations, we randomlychose four per 3-h sampling period. We placed one gill net(50-m net, 1.3,1.9, 23 , 3.1, 8.1 an mesh sizes) parallel to theshoreline in each of the four locations. Blight locations weresampled per day. We gill-netted 3 days weekly in late April andMay because we expected high mobility among pike imme-diately after their spring spawning ( Diana et aL, 1977; Iva-nova, 1969); this offered increased opportunities for markingfish. For June through October, we gill-netted 3 days per weekevery 2 weeks. Our sampling effort was less intensive in August(only seven nets set) to avoid catching waterfowL Samplingwas random, and all locations were sampled once before anywere sampled again. We marked captured fish with a caudalfin clip and measured total length (centimeters) before re-lease. Mark-recapture data were analyzed with the Schumach-er-Eschmeyer method, and Schumacher-Eschmeyer confi-dence intervals were estimated (Krebs, 1989).

Encounter ratesWe conducted more intensive sampling of pike on the eastshore of the lake (location 6) with one gill net twice a weekin late May, June, and July. Minnows nest on this shore, andthe additional sampling of pike in this location allowed us toestimate average encounter rate between site-attached (egg-guarding) male fathead minnows and pike. We did not esti-mate the encounter rate for August because we did not wantto disturb egg-guarding male minnows at this location thatwere being used for another experiment (see below).

Us* of natural and artificial nestsTo determine if patterns of nest use by male fathead minnowswere related to levels of predation threat on a lakewide basis,we selected 7 locations, from the 13 originally examined, thatpresented a range of pike densities and could be easily andreliably sampled. Based on the number of pike caught perhour per net in May and early June, we identified areas of thelake that yielded consistently high catches (overall mean •1.5 pike/h per net), termed "high pike-density" locations (lo-cations 1-4), or low catches (overall mean 0.6 pike/h pernet), termed "low pike-density" locations (locations 5-7). Weexcluded locations 8-13 (Figure 1) from the investigation ofnesting patterns because they were difficult to sample owingto heavy stands of macrophytes or shallow water. We contin-ued to monitor these locations throughout the summer todetermine if patterns of high and low densities of pike re-mained constant.

To determine if minnows were present at all high and lowpike-density locations and thus determine if minnows wereavailable to establish nesting territories (males) and lay eggs(females), we set unbaited minnow traps (mesh 13 cm1) for12 h in the littoral zone at each location. We set one to threetraps per location once a month in June and July (high pike-density: n >= 13 traps per month; low pike-density: n = 8 trapsper month ).

We visually surveyed natural nests of fathead minnows atthe seven lecaseas with high or low densities of pike twice amonth in June, July, and August either by wading or from acanoe. We defined a natural nest as a substrate (rock, log,twig, or piece of bark) on which we found eggs. Our defini-tion thus excluded sites where males defended an area and a

Dow



Jone* and Paszkowski • Predator avoidance by egg-guarding male fathead minnows 657

L f

1J

1

|

S8

High pOcs-denslty locations

QJLowpOa-dmsRyl

Otim locations

I I H 1* 11

Location

Figure 1Mean catch (per hour per gillnet) of northern pike for Mayand June 1994 in ArmstrongLake, Alberta. Error bars indi-cate 1 SE. Location* 1, 2, 3,and 4 were deiignated a*"high pike-density" locationsfor experiments on selectionof nest location by male fat-head minnow Locations 5, 6,and 7 were designated "lowpike-density" locations (n - 4nets per location). Some loca-tions were not used used as lo-cations for experiments be-cause they could not be sam-pled consistently or becausethey became inaccessible be-cause of vegetation growth.

potential oviposition site but eggs were absent. We could notdocument the total number of potential natural nests (sub-strates that appeared appropriate as nests but lacked eggs) inany location because high winds continually deposited andremoved substrates that minnows might use as nests.

Each of the seven focal locations had some suitable sub-strate, but because potential substrates for nests were notequally and predictably available to minnows at each of thelocation at any one time, we used artifical nests to control fornesting opportunity. At each location in early June, we placednine nest boards on the water surface [mean depth to thebottom of lake from the board was 23 ± 3 (SE) cm]. Boards(IS X 25 cm) were anchored to rocks with nylon cord. Ourpreliminary observations in 1993 indicated that male fatheadminnows would readily use these boards as nests (Jones,1995). We defined a nest as the presence of one or moreclusters of eggs on the underside of the board. One boardaccommodated up to three males at one time. The length ofthe nesting period (time for eggs to hatch) was typically 10days, so by sampling every 12-14 days, we did not resamplethe same nesting attempt on natural or artificial substrates.

AnalysesWe compared the use of natural nests between low and highpike-density locations and artificial nests between low andhigh pike-density locations with a t test using Systat 5.2.1 (Wil-kinson, 1992). Data were pooled among low- or high-density

locations by month [n =« 18 for low-density locations (2searches/month X 3 months X 3 locations); n - 24 for high-density locations (2 searches/month X 3 months X 4 loca-tions)]. We square-root transformed the data (square root+3/8); this is a common transformation when the observa-tions include zero (Zar, 1984).

Predation risk and nesting behavior

To examine the response of individual male fathead minnowsto a potential predator, we exposed males using natural nestsat location 6 (an area with low pike-density) to either a controlstimulus or a predator stimulus. The stimulus was a 53-cmpiece of driftwood in a cage (control) or a live pike (54 ± 2cm; n = 4) in a cage (predator treatment). The rectangularcage (57 cm X 8 cm X 10 cm) was made from a metal framewrapped in chicken wire. The interior of the cage was partiallylined with clear plastic to prevent injury to enclosed pike.Water flowed through the cage when it was set on the bottomof the lake, ensuring that minnows experienced both a visualand chemical stimulus when a pike was presented.

Experiments were conducted 4-8 August 1994, during day-light hours (1000-1700 h) when it was calm. Each day, weplotted sites of males guarding eggs in natural nests on a map.From these sites, we randomly chose five males and designat-ed them as treatment or control. Of the 20 males manipulat-ed, we used results from only 14 in analyses (n = 7 control,

Dow



658 Behavioral Ecology Vol. 8 No. 6

7 treatment), as the remaining males exhibited fungal infec-tions and disease which might have affected behavior. Individ-ual males were tested only once to avoid habituation to thestimulus (Magnhagcn and Vestergaard, 1991; Martin andKraemer, 1987).

We observed males for 10-min intervals before introductionof a stimulus (prestimuhis interval), 10 min while the stimuluswas present (stimulus interval), and 10 min after the stimuluswas removed (poststimulus interval). The behavior of malefathead minnows appeared to be unaffected by the presenceof an observer (Jones, 1995). When presenting a stimulus,observers placed the cage at the periphery of the male's ter-ritory (approximately IS cm from the nest, based on labora-tory and field measurements of territory size; Jones, 1995)and recorded behavior while standing in the water or onshore, 1 m away from the nest.

For each male, we recorded return time (in seconds), thetime for the male to return to his nest after introduction ofthe cage, as a measure of risk taking (Lachance and FitzGer-ald, 1992). We recorded the frequency of occurrence of re-productive behaviors and analyzed five that were reliably de-tected. These were chase, rub, nibble, tight circle, and widecircle. Chases were vigorous lunges at conspecific and hetero-specific fish and invertebrates. Rubs were a conspicuousabrading motion of the dorsal pad on the underside of thenest (McMillan, 1972). Nibble occurred when a male assumeda vertical position in the water column underneath the nestand placed his mouth, nostrils, and tubercles in contact withthe ceiling of tile nest and individual eggs (McMillan, 1972).A tight circle involved a revolution around the nest (up to 5cm from the nest), whereas wide circles were revolution* frominside the nest to the perimeter of a male's territory (up to15 cm from the nest). Because all males had eggs in theirnests, behaviors other than rubs were infrequent. We calcu-lated total activity of each male by summing the number ofchases, rubs, nibbles, and circles performed by a male in each10-min time interval.

To determine how external factors influenced reproductivebehavior and return time, we measured physical variables andnest characteristics after behavioral observations were com-pleted. Physical variables were nest type (log, twig, rock), wa-ter depth (centimeters) at the nest, distance to nearest neigh-bor with-a nest (centimeters), distance to shore (centimeters)from the nest, and percent cover of the nest, defined as thepercentage of the nest visible to an adjacent observer 1 maway from a male's territory. Water temperature [20°C ± 1 . 0(SE)], dissolved oxygen (7.9 ± 1.4 mg/1) and pH (8.6 ± 0.2)varied minimally among nests and thus were not included inanalyses. We measured two characteristics of nests: numberand age of eggs. We counted individual eggs to determinenumber of eggs; age of eggs was determined by visual inspec-tion (Jones, 1995).

In addition, we scored the color (1—5) of the egg-guardingmale in each of the three time intervals to determine if malesremain conspicuously colored in the presence of a threat.Scores were based on McMillan and Smith (1974) and Unger(198S); males that were pale were scored 1, and those thatwere progressively darker with a pronounced dorsal pad werescored 2, 3, or 4. Males assigned a score of 5 were not onlydark with a developed dorsal pad, but had two white or goldenbands around the body.

AnalysesTo determine differences between return times in control andtreatment males after exposure to the cage, we used a Wil-coxon's test (Sokal and Rohlf, 1981). Multiple regression wasused to relate return time to nest characteristics, physical vari-ables, and male color.

To determine if total activity differed between control andtreatment and across sampling intervals, we used a repeated-measures MANCOVA (Morrison, 1990) with a nested split-plotdesign, where treatment was the between-cubject effect andsampling interval was the within-subject effect Individualmales were nested within treatment, and number and age ofeggs were the covariate*. We examined the data and residualplots for normality to ensure that the assumptions of the re-peated-measures model were not violated (von Ende, 1993).Behavior was analyzed as number of acts per 10-min interval.

Because rubs performed during the stimulus interval werenon-QormaDy distributed, we analyzed them nonparametrical-h/ with a Mann-Whitney U test, examining if the number ofrubs per minute in the nest was significantly different betweencontrol and treatment males. We compared rubs per minutein the nest, not per 10-min observation interval, to accountfor the time males spent away from the nest after a stimuluswas introduced. To analyze rubs during pre- and poststimulusintervals, we used a repeated-measures ANCOVA.

RESULTS

Predation ride and nest location

Mark-ncttptun analysisBased on our April-October sampling, we estimated the num-ber of pike in the lake to be 421 (95% a , 265, 1077). This isequivalent to 1.7 pike/ha with a range of 1.1—4.1 pike/habased on the upper and lower confidence intervals. We cap-tured and marked 122 northern pike and recaptured 7.

Encounter ratesEstimated mean encounter rates of northern pike with egg-guarding minnows at location 6 differed little between sum-mer months. In May, we caught an estimated 0.30 pike/h (n» 2 nets set), hi June, we set 8 nets and caught 0.33 ± 0.04pike/h. Injury, when northern pike mobility was low (Dianaet aL, 1977), the catch was 0.10 pike/h (n = 6 nets).

Natural and artificial nest ustMinnows were present at all locations in early June and inearly July, hi June, in locations with high pike-density, wetrapped on average 111 ± 75 (SD) males/trap per 12-h setand 51 ± 41 females. At low pike-density locations we trapped37 ± 22 males/trap and 32±26 females/trap. In July, in highpike-density locations, we trapped 37±17 males and 66±38females/trap, whereas at low density locations, we trapped13±7 males and 48±12 females/trap. Most of the trappedfemales were gravid.

More natural nests were found in areas of low pike-densitythan in areas of high pike-density (t - -2.14 , p - .038; Table1). Low-density locations had a mean 5.7*9.7 (SE) nests/month, whereas high-density locations had 1.4±2.1 nests/month. Although there appeared to be more potential naturalnest substrates at location 6 than other locations, a muchhigher proportion of nest substrates were unused in locations2, 3, and 4. We never found eggs in locations 1 (high pike-density) or 5 (low pike-density). When we repeated the sameanalysis but removed location 6 (because it contained so manymore nests than any other location), differences between lowand high pike-density areas were not significant (f = 1.731, p= .092).

Similarly, artificial nest boards placed in locations with lowdensities of pike had more nests than beards ia higb-donsitylocations, but the difference was not significant (t m —1.404,p » .168; Table 1). For the three low-density locations com-bined, mean board use was 1.7 ± 3.0 nests/month. In highpike-density locations, mean use was 0.7 ± 1 . 6 nests/month.

Dow



Jones and Paszkowski • Predator avoidance by egg-guarding male fathead minnows 659

Table 1Mean number of natural substrates and artificial nest boards found with eggs of fatheadpike

Mean number of nests (SE)

in rriipon to loou Hrn"ty of uoruxern

Nest type

NaturalArtificial

Higb-densitjr locations

1

00

2

2 3 (1.0)0.2 (0.2)

3

2.1 (1.1)2.7 (1.0)

4

0.7 (0.4)0

Low-density locations

5 6

0 16.6 (4.2)0 5.1 (1.3)

7

0.7 (0.4)0

We defined a nest as a substrate that bore eggs. We surveyed each of the seven locations twice a month in June, July, and August 1994 (n '18 low pike-density locations; n = 24 high pike-density locations).

Minnows did not use the boards at locations 1 and 4 (highpike-density), and 5 and 7 (low pike-density). Variability washigh for both artificial and natural sites because there werevery few nests at any of the locations in June and early Julydue to high winds that created wave action which preventednest defense.

Predatkm risk and nest behavior

When the cage containing driftwood (control) or pike (treat-ment) was introduced, all males left their nests and lost breed-ing coloration. Most males decreased two color scores in thestimulus interval and usually left their nests to hide in nearbyvegetation up to 1 m away.

Return timeMales in the predator treatment took significantly longer toreturn to their nests than control males (Wilcoxon C **-2.82, p - .010). Control males returned to the nest after61.9 ± 23.7 s (range 1-159 s) and treatment males returnedafter 149.6 ± 82.9 s (range 1-600 s). One control and onetreatment male returned to the nest only after the cage wasremoved.

Control males had on average 280 ± 165 eggs while treat-ment males had 150 ± 94. Egg age varied between < 1 and6 days (control 5.4 ±1.0; treatment 4 3 ± 2.2). The time forthe male to return to the nest was not related to the numberor age of eggs (F = 1.95, p - .189).

Physical variables did not influence return time (Table 2).There was, however, a significant relationship between returntime, distance to nearest neighbor, and number of prestimu-lus rubs (F = 10.74, p = .001, FT = .86; Table 2). There wasa significant interaction between rubs and distance to nearestnesting neighbor (t ~ -2.81, p «• .017). The distances be-tween neighboring males that were guarding eggs rangedfrom 12 cm to 700 cm (mean 333 ± 52 cm). A male with a

Table 1Multipl

fegresrio of ictmu thne of male fathead itiiinM't—» to the

nest after an encounter with a caged pike (treatment; a *> 7) or acontrol stimulus (a • 7) with physical variables

Physical variable P p

Nest typeDepth of nestDistance to shore% Cower of nestDistance to nearest neighbor

* Only distance of male minnow to nearest nesting neighbor was asignificant variable predicting time for males to return to the nest(see text).

1.4650.3142.6040.6151.419

JOS.595.158.463.003*

dose neighbor rubbed more (prestimulus) and returned tothe nest sooner than males with distant neighbors.

Noting behaviorMales in the control and treatment did not differ in totalactivity during prestimulus [F » 3.71, p - .080) or stimulusintervals (F = 1.82, p - .205). The total activity of controlmales was significantly greater than treatment males in thepoststimuhis interval (F = 531, p «• .042). During the pred-ator treatment, 1 male spawned and received 250 eggs. Whenhe is included in • the analysis, age of eggs was a significantcovariate explaining the total activity (F = 9.47, p •» .011).Total activity lessened as eggs got older. The number of eggsdid not influence the total activity {F - 3.24, p " .097). Malesincreased their total activity from prestimulus intervals to post-stimulus intervals but the increase was not significant for con-trol or treatment individuals (F = 1.19, p = 322; Figure 2a;Table 3).

The most frequent behavior was rubs (Table 3). During thestimulus interval, mean rubs per minute in the nest was 6.1±1 .0 for control males and 6.6 ± 2.2 rubs/min for treatmentmales; differences were not significant (U » 24.0, p «• .949).Males increased the number of rubs from prestimulus to post-stimulus intervals, but the difference was not significant (F «•0.29, p - .602; Figure 2b). Control males performed signifi-cantly more rubs in the poststimuhis interval than treatmentmales (F = 8.79, p •* .013; Figure 2b). The number of eggsdid not influence the number of rubs performed (F = 0.14,p - .719). Age of eggs was a significant covariate explainingthe number of rubs (F = 11.82, p - .006); the number ofrubs decreased as eggs got older. There was no interactionbetween age of eggs and treatment (F « .83, p •= .449).

TablesMeans (SE») per 10-min interval of the total activity and —™K»r ofrubs of egg-guarding male ftthr*** minnows before (presthnuhn),tu insr (stimulus)* a td after (Doststx^nulusj es^posure to

luieal from a caged northern pike

Treatment Prestimulus Stimulus Poststimulus

Total activityPredator (n «• 7)Control (n «= 7)

Number of rubsPredatorControl

513 (8.9) 59.7 (213) 66.0 (12.7)73.9 (15.5) 693(8.7) 91.6(16.1)

33.0 (7.0)44.7 (6.4)

45.4 (173)52.0 (7.9)

52.4 (13.1)693 (8.7)

See Figure 2 for significant differences between means. Because allmales had eggs, behaviors other than rubs were infrequent; thus wesummed additional reproductive behaviors other than rubs (chases,etc.) and reported them as one behavioral measure—total activity.

Dow



660 Behavioral Ecology VoL 8 No. 6

Contn

(b)

Figure 2(a) The total activity and (b)number of rub* of male fat-head minnow in ArmstrongLake, Alberta, confronted witha control stimului or a live,caged northern pike. Ban in-dicate 1 SE. * Control maleshad greater total activity andperfomed significantly morerubs in the poststimulus inter-val than treatment males.

Eo

I

75•s

M -

Eo

I«.O

20 "

Stimulus Post

Sampling Interval

Pre Post

DISCUSSION

A northern pike represents a complex, dual risk to an egg-guarding male fathead minnow. Pike are a direct threat to themale himself and an indirect threat to his offspring if he exitsthe nest in the presence of pike and leaves the eggs open topredation by other fish and invertebrates. In our experiment,the behavior of male fathead minnows was influenced by bothof these conflicting threats acting simultaneously. To look atboth threats independently, one could compare males de-fending nests with eggs and without eggs. The egg-guardingmales of our study responded to disturbance (i.e., introduc-tion of the cage with or without pike) by leaving their nestsimmediately, seeking shelter, and becoming less conspicuouslycolored. Males thus behaved initially in a manner that re-duced their own risk of mortality. Control males returned totheir nests sooner than treatment males, indicating that thelatter males had detected pike either visually or chemically(Mathis and Smith, 1993) and were increasing die risk to theiruntended eggs in response to a greater risk to themselves.

We expected males to take greater risks, Le., return sooner,for larger numbers of eggs than for smaller numbers and forolder eggs versus younger eggs. In fact, return time was notrelated to number of eggs or age of eggs. Lachance and Fltz-Gerald (1992) similarly found no relation between clutch sizeand parental investment in the three-spined stickleback (Gas-terosteus acuUaius). Magnhagen and Vestergaard (1991)found only a weak relationship between egg age and risk tak-ing in common gobies (Pomatoschistus microps), where timeaway from the nest after a disturbance decreased throughoutthe brood cycle.

Lashance and FioGerald (1992) and Whorisfcey and Fitz-Gerald (1985) reported that depth and distance to shore weresignificant predictors of return time in three-spined stickle-backs threatened by an artificial heron. In our experiment,depth, distance to shore, and all other physical variables did

not influence behavioral decisions of fathead minnows. Malesdid, however, return to dieir nests sooner when they were clos-er to other egg-guarding males. Unger (1983) found that soli-tary males lost more weight and remained paler in color com-pared to males guarding in a competitive setting. In ArmstrongLake, males nested close to one another (range 12-700 cm; n*• 33). This may be a function of the distribution of appropriatenest sites, but guarding a nest near neighbors may relieve pre-dation pressure on an individual (sensu Krause, 199S). If thelatter is true, males guarding a nest in areas with high numbersof near neighbors may return to the nest sooner, regardless ofegg number or age, because the probability of being eaten isreduced, whereas the probability of egg predation and nestusurpation by adjacent males a high.

Increased levels of activity can increase die probability thatan animal is detected by predators, particularly a visual pred-ator like northern pike (Diana et al., 1977; Endler, 1987).Matity et aL (1994) suggested that die conspicuous behaviorof breeding male fathead minnows makes diem more vulner-able than juvenile or female minnows to predation by gartersnakes (Thamnophis radix). Fathead minnows disturbed byour experimental cage abandoned their nests immediately fol-lowing die intrusion, but most males (12 of 14) returned totheir nests while die cage, widi or widiout pike, was still pres-ent, demonstrating that their site attachment was indeedstrong. Despite being absent for longer periods of time, malesexposed to pike did not increase their activity upon returningto their nests as much as control males did. Presumably, malesin our experiment increased their rates of egg rubbing afterbeing away from the nest to compensate for care lost whilethey wore absoat, as well as to indicate nost ownership to near-by males and females. Males exposed to pike were less activeand thus less conspicuous. Our observations agree widi thoseof Sargent (1988), who found that fathead minnow males sub-ject to predation risk from crayfish (Orcomctes spp.) exhibited

Dow



Jones and Paszkowiki * Predator avoidance by egg-guarding male fathead minnowt 661

lower rates of egg rubbing and, in turn, had lower rate* ofegg survival than those not at risk.

In our study, rub* and total activity were not influenced bythe number of eggs, but the total number of each behaviordecreased as eggs got older. Sargent (1988, 1989), however,showed a positive correlation between number of rubs andnumber of eggs and between number of rubs and age of eggsin filth carl minnows. A larger sample size of males with youn-ger eggs could more convincingly define our pattern.

Our experiment showed that individual male fathead min-nows were able to distinguish between two different distur-bances at their nest sites and to exhibit immediate and longerterm behavioral responses that corresponded to the severityof threat directly to themselves and indirectly to their eggs.Behavioral avoidance tactics using visual and chemical pred-ator recognition systems (Mathis et aL, 1993) may compensatefor lack of morphological defenses (e.g., spines) in fatheadminnows that often protect small fishes from predators. Thecost, however, for a male responding to frequent encounterswith pike may be reduced mating opportunities and higheregg mortality due to predation and disease resulting from low-er levels of care. An encounter with a pike, especially a pikethat has recently captured a fathead minnow and is associatedwith alarm pheromone, can have much more prolonged ef-fects on fathead minnow behavior than documented in ourfield experiments (Mathis and Smith, 1992). In a complemen-tary laboratory study, we found that fathead minnow malesconfronted with a caged pike decreased their reproductiveactivities for up to 24 h and abandoned the nest for up to 17min (Jones and Paszkowski, 1997). Similarly, Iowa darters(Etheostoma adit) decreased reproductive behavior when ex-posed to northern pike in a laboratory setting (Olivers et aL,1995).

Based on our estimates of encounter rates from a singlelocation, male minnows guarding a nest with eggs in Arm-strong Lake experienced pervasive predation threat. Our es-timates are actually conservative because they do not includedaily multiple encounters between a minnow and the sameindividual pike. Granted, not every contact with a pike trans-lates into a lethal predation threat to a male; however, anycontact could disrupt male reproductive activities. For males,the benefits of establishing a nest in areas where pike densitiesare low are obvious. The trend toward greater occurrence ofnatural nests in areas with low densities of pike that we ob-served in Armstrong Lake is consistent with the explanationthat males chose breeding territories and nests in areas wherethey assessed local predation risk to be low. Local predation"hot spots" may be sites where alarm pheromone is common-ly detected, and minnows may recognize and avoid these areasbased on chemical cues (Mathis and Smith, 1992). Similarly,relative frequency of visual contact with pike among locationsmight influence minnows' decisions regarding where to guarda nest. High rates of disturbance at a potential nest site, trig-gering repeated retreats and returns, might eventually lead tocomplete abandonment (Jones and Paszkowski, 1997).

The distribution of egg-guarding fathead minnows couldalso be influenced by features related to environmental con-ditions and habitat structure. We found no significant abioticdifferences among locations (e.g., temperature); however, be-cause fathead minnows require substrates to reproduce, butdo not construct nests themselves, the distribution of naturalnests in Armstrong Lake could simply have been an outcomeof the distribution of substrates such as rocks and woody de-bris. Our study addressed this possibility by employing artifi-cial nests; the fact that two locations never contained nestseven when we provided artificial substrates (Table 1), suggeststhat males were selecting nest locations based on multiple cri-teria. Similarly, one location (location 6) contained many

more natural nests and nests on artificial substrates than otherlocations. A low risk of predation may have been one factorthat made this location exceptionally attractive; however, wecannot conclude that predation risk was the definitive factorinfluencing the distribution of nests.

Additional habitat-related criteria could include the visibil-ity of egg-guarding males to females. A cryptic nest, surround-ed by vegetation or deadfall, might hide a displaying malefrom visual predators but also from potential mates. Anotherexplanation for the distribution of nests with eggs is that maleswithout territories, females, or both sexes of fathead minnowsimply did not occur in areas of high pike-density. Absence ofminnows could again reflect avoidance of pike per se or avoid-ance of areas with inappropriate environmental or feedingconditions. Sampling of minnows, however, showed that bothsexes were active in locations with low and high pike-density.Male minnows appeared to frequent high-risk areas but didnot establish nests there, although such nests would have in-tercepted females; we trapped gravid female minnows at highpike-density locations. A final possibility for observed patternsis that comparable numbers of males established nests in highpike-density locations as in low pike-density locations, butthese former males were seldom detected in surveys becausethey suffered high rates of predation or their nests sufferedhigh rates of predation as males were constantly displaced. Ifdiis were true, our results reflect nesting success rather thannest site selection. We cannot eliminate this possibility, but wedid not detect exceptionally high turnover rates for nests athigh pike-density locations to indicate that males were contin-ually ^ff^blithing nests then quickly being replaced as they

h ddg q y gwere eaten or as their nests were destroyed.

Martin (1993) proposed that investigations of nest preda-tion and nest site selection could provide new perspectives oncommunity-level patterns of habitat use and species coexis-tence in birds. Our study indicates that the effects of piscivoreson the behavior of individuals of a species of prey fish, andlinks between predation risk and reproductive success, mayinfluence larger scale, lakewide patterns of habitat use by thatprey species. There may also be community-level impacts.Brook stickleback, which inhabit Armstrong Lake and manyother boreal lakes with fathead minnows (Robinson andTonn, 1989), also exhibit paternal care of eggs within nests inthe littoral zone. Both species are small as adults, highly vul-nerable to piscivorous fishes (Robinson 1989), and share acommon set of alarm substances (Mathis and Smith, 1993).Investigation of the location of stickleback nests relative tolocal pike-densities and the location of fathead minnow nestscould provide novel insights into habitat-use patterns in lakefish communities.

We thank Meanook Biological Research Station for accommodationand L. Turner and M. Janowicz for field assistance. Thanks to J. Murieand N. Stacey for comments on the manuscript. This research wassupported by grants from the Canadian Circumpolar Institute to H.J.,and the National Sciences and Engineering Research Council of Can-ada to CP.

REFERENCES

Blower JC, Cook LM. Bishop JA. 1981. Estimating the size of animalpopulations. London: Allen and Urrwin.

Olivers P, BD Wisenden, RJF Smith, 1995. Predation risk influencesreproductive behaviour of Iowa Darters, Ethmtowui edit (Osteich-thyes, Percidae). Ethology 99:278-285.

Christiansen DC, 1976. Feeding and behavior of northern pike (MScthesis). Edmonton: University of Alberta.

Craig C, 1994. Predator foraging behavior in response to perceptionand learning by its prey: interactions between orb-«pinning spidersand stingtess bees. Behav Ecol Sociobiol 35:45-52.

Dow



662 Behavioral Ecology VoL 8 No. 6

Diana JS, Mackay WS, Ehrman M, 1977. Movement* and habitat pref-erence of northern pike in Lac Ste Anne, Alberta. Trans Am FishSoc 106^60-565.

Dill LM, 1987. Animal decision making and its consequences: thefuture of aquatic ecology and behavior. Can J Zool 65:803-811.

Dill LM, Fraser AHC, 1984. Risk of predation and the feeding behav-ior of Juvenile coho salmon (Oncorhytuhus hsuXch). Behav Ecol So-ciobiol 1&65-71.

Ekldv P, Diehl S, 1994. Puorore efficiency and refuging prey: theimportance of predator search mode. Oecologia 98:544-353.

Endler JA, 1987. Predation, light intensity and courtship behavior inPocaHa nticulata (Pisces: Poedlidae). Anim Behav 35:1376-1385.

Fraser DF, Gilliam JF, 1992. Nonlethal impacts of predator invasion:facultative suppression of growth and reproduction. Ecology 73:959-970.

Gfllen AL, Stein RA, Carlene RF, 1981. Predatkm by pellet-reared ti-ger muskellunge on minnows and bhiegiUs in experimental systems.Trans Am Fish Soc 110:197-209.

Godin JJ, Crossman SL, 1994. Hunger dependent predator inspectionand foraging behaviors in the threee-apine tiir\\*\«tr-y (GasUnatnuacultatus) under predadon risk. Behav Ecol Sociobiol 34:359-366.

Helfman G, 1986. Behavioral responses of prey fishes during predator-prey interactions. In: Predator-prey relationships: perspectives andapproaches from the study of lower vertebrates (Feder ME, LauderCV, eds). Chicago: University of Chicago Press, 135-156.

Hilborn R, Walters CJ, 1992. Quantitative fisheries stock assessment:choice, dynamics and uncertainty. New Ybrk: Chapman and Hall.

Ivanova MN, 1969. The behavior of predatory fish during feeding. JIchthycd 9-.574-S77.

Ivlev VS. 1961. Experimental ecology of the feeding of fishes. NewHaven, Connecticut: Yale University Press.

Jones HMJ, 1995. Effects of predadon on nest-guarding behavior ofmale fathead minnows (MSc thesis). Edmonton: University of Al-berta.

Jones HMJ, Paszkowski CA, 1997. Effects of exposure to predatorycues on territorial behavior of male fathead minnows. Environ BiolFish 4937-109.

Krause J, 1993. The effect of 'SchrechstofT on the shoaling behaviorof the minnows: a test of Hamilton's selfish herd theory. Anim Be-hav 45:1019-1024.

Krebs CJ, 1989. Ecological methodology. New York: Harper and Col-lins.

Lachance S, FltzGerald CJ, 1992. Parental care tactics of three-spinedsticklebacks living in a harsh environment. Behav Ecol 3:360-366.

lima SL, Dill LM, 1990. Behavioral decisions made under the risk ofpredation: a review and prospectus. Can J Zool 68:619-640.

Magnhagen C, 1990. Reproduction under predation risk in the sandgoby, Pomototthutus mtoiutus, and the black goby, Gobnu nigrr. theeffect of age and longevity. Behav Ecol Sociobiol 26331-335.

Magnhagen C, 1991. Predation risk as a cost of reproduction. TrendsEcol Evol 6:183-186.

Magnhagen C, Vestergaard K, 1991. Risk taking in relation to repro-ductive investments and future reproductive opportunities: field ex-periments on new-guarding common gobies, Pomolosdnttus micnps.Behav Ecol 2351-359.

Magurran AE, Seghers BH, 1990. Risk aensitive courtship in the guppy(PotcUut nticulata). Behavior 112:194-201.

Martin P, Kraemer HC, 1987. Individual differences in behavior andtheir M?ri«ti/-al consecfuenees. Anim Behav 35:1366-1375.

Martin TE, 1993. Nest predation and nest sites. Biosdence 43:523-532.

Mathis A, Olivers DP, Smith RJF, 1993. Population differences in re-sponses of fathead minnows to visual and chemical stimuli frompredators. Ethology 93:31-40.

Mathis A, Smith RJF, 1992. Avoidance of areas marked with a chemicalalarm substance by fathead minnows (PiwuphaUs pnmtlas) in a nat-ural habitat CanJ Zool 70:1473-1476.

Mathis A, Smith RJF, 1993. Chemical alarm signals increase the sur-vival tune of fathi»aH minnows (Pivtrphaln primulas) during en-counters with northern pike (Etta tucmj). Behav Ecol 4:260-265.

Matity JC, Olivers DP, Smith RJF, 1994. Population and sex differ-ences in andpredator responses of breeding fathead minnows (Pi-mephala promdtu) to chemical stimuli from garter snakes (Thorn-nophis radix and T. artahi).] Otera Ecol 20:2111-2121.

Mauck WL, Coble DW, 1971. Vulnerability of some fishes to northernpike (Esox huhu) predation. J Fish Res Board Can 28:957-969.

McKaye KR, 1984. Behavioral aspects of dchlid reproductive strate-gies pattems of territoriality and brood defense in Central Amer-ican substratum ipiwneri and African mouth brooders. In: Fish re-production (Potts GW, Wootton RJ, eds). Orlando, Florida: Aca-demic Press; 245-273.

McMillan V. 1972. Mating of the fathead. Nat Hist 81:72-78.McMillan VE, Smith RJF, 1974. Agonistic and reproductive behavior

of the fathead minnow. Z Tierpjychol 34:25-58.Moody RC. Helland JM, Stein RA, 1983. Escape tactics used by blue-

gills and fathead minnows to avoid predation by tiger muskeDunge.Environ Biol Fish 8:61-65.

Morrison DF, 1990. Multivariate statistical methods. New York: Mc-GrawHUL

Peckarsky BL, Cowan CA, Penton MA, Anderson C, 1993. Sublethalconsequences of stream-dwelling predatory stoneflies on mayflygrowth and fecundity. Ecology 74:1836-1846.

Price CJ. Tonn WM, Paszkowski CA, 1991. Incraspecific pattems ofresource use by fathead minnows in a small boreal lake. CanJ Zool69:2109-2115.

Reznkk D, Endler JA, 1982. The impact of predation on life historyevolution in Trinidadian guppies (PotdHa nticulata) Evolution 36:160-177.

Robinson CLK, 1989. Laboratory survival of four prey in the presenceof northern pike. CanJ Zool 67:418-420.

Robinson CLK, Tonn WM, 1989. Influence of environmental factorsand piscivory in structuring fish assemblages of small Alberta lakes.Can J Fish Aquat Sq 46:81-89.

Ryan MJ, 1985. The Tungara frog: a study in sexual selection andcommunication Chicago: University of Chicago Press.

Sargent RC, 1988. Parental care and egg survival both increase withdutch size in the fathead minnow, Pimtphala prvmtlas. Behav EcolSodobiol 23:33-37.

Sargent RC, 1989. AUoparental care in the fathead minnow, Pimt-phala pnmdos. stepfathers discriminate against their adopted eggs.Behav Ecol Sociobiol 25:379-385.

Schaeffer RL, Mendenhall W, Ott L, 1986. Elementary survey sam-pling. Boston: Duxbury Press.

Sih A, 1980. Optimal behavior can foragers balance conflicting de-mands? Science 210:1041-1043.

Sih A, 1988. The effects of predators on habitat use, activity and mat-ing behavior of a semi-aquatic bug. Anim Behav 36:1846-1848.

Sih A, Krupa J, Travers S, 1990. An experimental study on the effectsof predation risk and feeding regime on the mating behavior ofthe water strider, Ctrris nrntgts. Am Nat 135:284-290.

Sokal RR, Rohlf FJ, 1981. Biometry, 2nd ecLSan Francisco: WH Free-

Svensson I, 1988. Reproductive costs in two sex-role reversed pipefishspecies, j Anim Ecol 57:929-942.

Unger LM, 1983. Nest defense by deceit in the fathead minnow, Pi-mtphaia pnmtuu. Behav Ecol Sodobiol 13:125-130.

von Ende CN, 1993. Repeated measures analysis: growth and othertime-dependent measures. In: Design and analysis of ecological ex-periments (Scheiner, SM, Gurevitch J, eds). New York: Chapmanand Hall; 113-137.

Wahl DH, Stein RA, 1988. Selecdve predation by three esodds: therole of prey behavior and morphology. Trans Am Fish Soc 117:142-151.

Werner EE, Gflllam JF, Hall DJ, Mittlebach GG, 1983. An experimen-tal test of the effects of predation risk on habitat use in fish. Ecology64:1540-1548.

Werner EE, Hall DJ, 1988. Ontogenetic habitat shifts in bluegill: theforaging rate-predation risk tradeoff. Ecology 69:1352-1366.

Whoriskey FG, FltzGerald GJ, 1985. Nest-sites of the threespine stick-leback: can site characters alone protect the neat against egg pred-ators and are nest sites a limiting resource? Can J Zool 63:1991-1994.

Wilkinson L, 1992. SYSTAT: the system for statistics. Evanston, Illinois:Syttatlnc

Wing SR, 1988. Cost of mating for female insects: risk of predationin Photinus athutnnu (Coleoptera: Lampyridae). Am Nat 31:139-142.

Zar JH, 1984. Btostatistical analysis, 2nd ed. Toronto: Prentice Hall.

Dow



Documents

Effects of northern pike on patterns of nest use and reproductive