10
Individual traits inuence vigilance in wild female eastern grey kangaroos A. M. Edwards A,B,C , E. C. Best A , S. P. Blomberg A and A. W. Goldizen A A The University of Queensland, School of Biological Sciences, Brisbane, Qld 4072, Australia. B Present address: The University of Tasmania, School of Zoology, Sandy Bay, Tas. 7005, Australia. C Corresponding author. Email: [email protected] Abstract. Vigilance is an essential component of antipredator behaviour and is also used to monitor conspecics, but is traded off against feeding in herbivores. This trade-off can be inuenced by variation in many environmental, social and individual traits. Our aim was to test the relationship between individual-level traits, including boldness, body condition and reproductive state, and vigilance, while controlling for environmental and social variables. Using multiple 5-min video samples of 30 foraging, individually recognisable, female eastern grey kangaroos (Macropus giganteus) at Sundown National Park in Queensland, we investigated individual-level variation in the duration, intensity and target of vigilance behaviour during foraging. On separate occasions, we used ight-initiation distance tests to measure boldness in our kangaroos. Females with longer ight-initiation distances (shyer females) spent more time vigilant, providing preliminary support for studies of animal personality that have suggested that boldness may covary with vigilance. Body condition did not affect the total time spent vigilant, but females in poorer body condition spent more of their vigilance time in low-intensity vigilance. Vigilance patterns were not related to reproductive state, but varied among months and differed between mornings and afternoons, and females spent more time in high-intensity vigilance when further from cover. Even after accounting for all our variables we found that 7% of the variation in total time vigilant and 14% of the variation in vigilance intensity was explained by individual identity. This highlights the importance of individual-level variation in vigilance behaviour. Additional keywords: antipredator behaviour, boldness, group-size effect, personality. Received 19 March 2013, accepted 8 September 2013, published online 26 September 2013 Introduction Vigilance behaviour is generally assumed to act as a method of predator detection in prey species (Elgar 1989; Roberts 1996); however, vigilance also facilitates the monitoring of conspecics, known as social vigilance. Social vigilance can provide information on reproductive and feeding opportunities, potential agonistic interactions, and conspecicsperceptions of danger (Elgar 1989; Cameron and du Toit 2005; Gaynor and Cords 2012). Most vigilance involves the interruption of feeding and thus incurs a cost, and animals are therefore expected to carefully optimise this feeding/vigilance trade-off. The optimisation of this trade-off for an individual should depend on its characteristics and those of the group and environment in which it is located at any given time. Indeed, vigilance behaviour is very exible and many factors are known to affect individualsvigilance behaviour (e.g. Roberts 1996; Carter and Goldizen 2003; Fairbanks and Dobson 2007; Pays and Jarman 2008; reviewed by Elgar 1989). Despite the exibility of vigilance behaviour, there is increasing evidence of signicant differences in vigilance among individuals. Individual variation has been shown in many species (yellow-bellied marmots (Marmota aviventris): Blumstein et al. 2004; adult spotted hyaenas (Crocuta crocuta): Pangle and Holekamp 2010; nutmeg manikins (Lonchura punctulata): Rieucau et al. 2010; white-tailed prairie dogs (Cynomys leucurus): Hoogland et al. 2013), with individual identity explaining as much as 12.7% of the variation in vigilance rates (female bighorn sheep (Ovis canadensis): Rieucau and Martin 2008). In eastern grey kangaroos (Macropus giganteus), Dannock et al.(2013) documented individual differences in the proportions of time that individuals spent vigilant, and in the vigilance postures used, while Carter et al.(2009b) showed that, individuals within the same population showed very different relationships between vigilance and group size, even though a pattern of decreasing vigilance with increasing group sizes was found across the population. Given that vigilance is important for defence against predation, and should thus be under strong selection, differences among individuals in vigilance require explanation. Many characteristics of individuals could affect vigilance and at least partially explain individual variation in vigilance. Males and female often exhibit different patterns of vigilance (e.g. Burger and Gochfeld 1994; Shorrocks and Cockayne 2005; Journal compilation Ó CSIRO 2013 www.publish.csiro.au/journals/ajz CSIRO PUBLISHING Australian Journal of Zoology, 2013, 61, 332341 http://dx.doi.org/10.1071/ZO13025

Individual traits influence vigilance in wild female eastern grey kangaroos

  • Upload
    uq

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Individual traits influence vigilance in wild female easterngrey kangaroos

A. M. EdwardsA,B,C, E. C. BestA, S. P. BlombergA and A. W. GoldizenA

AThe University of Queensland, School of Biological Sciences, Brisbane, Qld 4072, Australia.BPresent address: The University of Tasmania, School of Zoology, Sandy Bay, Tas. 7005, Australia.CCorresponding author. Email: [email protected]

Abstract. Vigilance is an essential component of antipredator behaviour and is also used to monitor conspecifics, but istraded off against feeding in herbivores. This trade-off can be influenced by variation in many environmental, social andindividual traits. Our aim was to test the relationship between individual-level traits, including boldness, body conditionand reproductive state, and vigilance, while controlling for environmental and social variables. Using multiple 5-min videosamples of 30 foraging, individually recognisable, female eastern grey kangaroos (Macropus giganteus) at SundownNational Park in Queensland, we investigated individual-level variation in the duration, intensity and target of vigilancebehaviour during foraging. On separate occasions, we used flight-initiation distance tests to measure boldness in ourkangaroos. Females with longer flight-initiation distances (shyer females) spent more time vigilant, providing preliminarysupport for studies of animal personality that have suggested that boldness may covary with vigilance. Body condition didnot affect the total time spent vigilant, but females in poorer body condition spentmore of their vigilance time in low-intensityvigilance. Vigilance patterns were not related to reproductive state, but varied among months and differed betweenmornings and afternoons, and females spent more time in high-intensity vigilance when further from cover. Even afteraccounting for all our variables we found that 7% of the variation in total time vigilant and 14% of the variation in vigilanceintensity was explained by individual identity. This highlights the importance of individual-level variation in vigilancebehaviour.

Additional keywords: antipredator behaviour, boldness, group-size effect, personality.

Received 19 March 2013, accepted 8 September 2013, published online 26 September 2013

Introduction

Vigilance behaviour is generally assumed to act as a method ofpredator detection in prey species (Elgar 1989; Roberts 1996);however, vigilance also facilitates themonitoring of conspecifics,known as social vigilance. Social vigilance can provideinformation on reproductive and feeding opportunities, potentialagonistic interactions, and conspecifics’ perceptions of danger(Elgar 1989; Cameron and du Toit 2005; Gaynor and Cords2012). Most vigilance involves the interruption of feeding andthus incurs a cost, and animals are therefore expected to carefullyoptimise this feeding/vigilance trade-off. The optimisation of thistrade-off for an individual shoulddependon its characteristics andthose of the group and environment in which it is located at anygiven time. Indeed, vigilance behaviour is very flexible andmanyfactors are known to affect individuals’ vigilance behaviour (e.g.Roberts 1996; Carter and Goldizen 2003; Fairbanks and Dobson2007; Pays and Jarman 2008; reviewed by Elgar 1989).

Despite the flexibility of vigilance behaviour, there isincreasing evidence of significant differences in vigilanceamong individuals. Individual variation has been shown inmany species (yellow-bellied marmots (Marmota flaviventris):

Blumstein et al. 2004; adult spotted hyaenas (Crocuta crocuta):Pangle and Holekamp 2010; nutmeg manikins (Lonchurapunctulata): Rieucau et al. 2010; white-tailed prairie dogs(Cynomys leucurus): Hoogland et al. 2013), with individualidentity explaining asmuch as 12.7% of the variation in vigilancerates (female bighorn sheep (Ovis canadensis): Rieucau andMartin 2008). In eastern grey kangaroos (Macropus giganteus),Dannock et al. (2013) documented individual differences in theproportions of time that individuals spent vigilant, and in thevigilance postures used, while Carter et al. (2009b) showed that,individuals within the same population showed very differentrelationships between vigilance and group size, even though apattern of decreasing vigilance with increasing group sizes wasfound across the population. Given that vigilance is importantfor defence against predation, and should thus be under strongselection, differences among individuals in vigilance requireexplanation.

Many characteristics of individuals could affect vigilanceand at least partially explain individual variation in vigilance.Males and female often exhibit different patterns of vigilance(e.g. Burger and Gochfeld 1994; Shorrocks and Cockayne 2005;

Journal compilation � CSIRO 2013 www.publish.csiro.au/journals/ajz

CSIRO PUBLISHING

Australian Journal of Zoology, 2013, 61, 332–341http://dx.doi.org/10.1071/ZO13025

Lung and Childress 2007; Pays and Jarman 2008; Lea andBlumstein 2011), due to either differing vulnerability topredators, or to different priorities for selecting alternativeactivities. Females with young are more vigilant than non-reproductive females in many species (Colagross and Cockburn1993; Burger and Gochfeld 1994; Hunter and Skinner 1998;Childress and Lung 2003), presumably because of thevulnerability of their young to predators (Steenbeek et al. 1999;Treves et al. 2003). Late pregnancy or the carrying of young canalso make females themselves more vulnerable. However, thereproductive state of female mammals also influences nutritionalrequirements and thus would be expected to influence females’foraging/vigilance trade-offs independently of effects on theirpredation risk. Peak lactation has been estimated to increaseenergetic requirements by ~75% for female macropods (Cork1991). This suggests that lactating females may be less able toafford to be vigilant than other females, despite the fact thattheir young may be particularly vulnerable to predation.

Body condition and nutritional status may also affectvigilance, by acting on individuals’ foraging/vigilance trade-offs.Since animals at risk of predation have to trade off increasingtheir food intake against antipredator behaviour and thus safety(Houston et al. 1993; Brown 1999), the quality and quantity offood available and an individual’s body condition should affectthe time that it needs to spend foraging, and thus the time it canafford to spend vigilant. This predicts that animals in poorcondition will trade off vigilance for feeding to increase theirenergy reserves; this pattern was indeed found in yellow-belliedmarmots (Lea and Blumstein 2011) and in food-supplementationexperiments on Belding’s ground squirrels (Spermophilusbeldingi) (Bachman 1993).

Another possible explanation for among-individual variationin vigilance is that vigilance patterns are related to individuals’positions along the bold/shy continuum. Much recent work hasfocussed on personality differences among animals; personality(also called temperament) is defined as consistent interindividualvariation in traits such as boldness, sociability and exploration(Sih et al. 2004; Bell 2007; Réale et al. 2007). Variation alongthe shy–bold axis relates to individuals’ propensities to takerisks. Boldness has been shown to affect foraging decisions inseveral studies of captive animals (guppies (Poecilia reticulata):Dyer et al. 2009; fallow deer (Dama dama): Bergvall et al. 2011;barnacle geese (Branta leucopsis): Kurvers et al. 2010) andlocomotion patterns in bluegill sunfish (Lepomis macrochirus)(Wilson and Godin 2010). It seems likely that the boldness ofindividuals would also affect their vigilance behaviour.Individual differences in vigilance could be maintained due totrade-offs with other behaviours (e.g. Wolf et al. 2007; Patricket al. 2012). Since the vigilance of foraging animals is traded offagainst food intake, at least to some extent, bolder individualsmight incur a higher risk of predation but eat more, while shyerindividuals might eat less but be less likely to be eatenthemselves, with these selection pressures potentially balancingeach other. To our knowledge, only one test of the relationshipbetween boldness and vigilance behaviour of wild animalshas been presented in the literature; in that test, vigilance anda measure of boldness, flight initiation distance, were notsignificantly related in yellow-bellied marmots (Blumstein et al.2004).

Because somany factors, including individuals’ characteristics,but also predictors of predation risk, group size and composition,personality, reproductive status and nutritional status may allaffect vigilance, it has been difficult to untangle their separateeffects. Themultitude of factors affecting vigilance, as well as thegrowing acceptance that a significant portion of vigilance isdirected at conspecifics rather than predators, complicate the taskof understanding how individual animals trade off vigilance andfeeding at any point in time. To better understand how thesemanyfactors affect vigilance, we have taken advantage of the ability todistinguish (1) antipredator from social vigilance (the targets ofvigilance) in kangaroos, by quantifying the gaze direction ofindividuals who are on the periphery of their groups (Favreauet al. 2010), and (2) low-intensity and high-intensity vigilance,based on postures. Dissecting the vigilance of individuals bytarget and intensity of vigilance, in addition to knowing theidentity of our focal animals, allowed us to achieve a deeperunderstanding of the factors that affect individuals’ vigilancepatterns than has been previously achieved for wild herbivores.

Our main objective was to better understand individualvariation in vigilance in eastern grey kangaroos, by testing how arange of individuals’ characteristics affected their vigilance,while controlling for a range of other possible predictors. Westudied only adult female kangaroos, to remove variation causedby age/sex class, and collected data only during periods whenfemales were foraging. Foraging female kangaroos spend almost100% of their time either feeding or vigilant; understanding howdifferent factors affect vigilance thus also tells us how thesefactors influence the trade-off that the animals make betweenmaximising food intake and staying safe. We modelled threedifferent aspects ofvigilancebehaviour: (1) theoverall proportionof time that females spent vigilant, (2) the proportion of theirvigilant time that they spent in low-intensity vigilance (lower-vigilance posture) and (3) the proportion of their vigilance timethat females on the periphery of groups spent facing out from theirgroups, and thus were likely engaging in antipredator vigilance.We also aimed to determine for each of these measures ofvigilance how much of the variation measured was due toindividual identity after consideration of our predictor variables.

Methods

Study species

The eastern grey kangaroo is a large member of the familyMacropodidae,with awide distribution reaching fromCapeYorkPeninsula in the north of Australia to Tasmania in the south. Thekangaroos’ predators include dingos and wild dogs (Canisfamiliaris) (Jarman 1987), red foxes (Vulpes vulpes) and wedge-tailed eagles (Aquila audax) (Banks 2001). Breeding occursall year round, with most births in the summer months. Theeastern grey kangaroo is the most social species in the family(Clarke et al. 1995), exhibiting a fission–fusion social system inwhich individuals constantly move among temporary groups(Southwell 1984a, 1984b; Heathcote 1987; Jarman 1994; Carteret al. 2009a).

Previous studies on vigilance in eastern grey kangaroos havefound differing patterns. For example, some populations showeddecreasing vigilance with increasing group size (Carter et al.2009b; Pays et al. 2009), following the classic group-size effect

Effect of individual traits on vigilance in kangaroos Australian Journal of Zoology 333

(Pulliam 1973; Lima 1995; Roberts 1996), while others did not(Colagross andCockburn 1993; Favreau et al. 2010). An increasein social vigilance with larger group sizes may counteract thedecrease in antipredator vigilance and explain the lack of arelationship between group size and overall vigilance found insome studies (Favreau et al. 2010). The intensity of vigilance hasalso been shown to decrease with an increase in group size(Jarman 1987), also suggesting that animals in larger groupsperceive a lowered predation risk. We also take these variablesinto consideration in ourmethods and, in addition, investigate theinfluence of variation in individual traits and identity. However,individuals have been found to vary in their pattern of vigilance inrelation to group size (Carter et al. 2009b); therefore, in our studyof wild eastern grey kangaroos we recorded the identities of allfocal kangaroos to include as a random effect in our statisticalmodels.

Study site and population

This study was undertaken at the southern end of SundownNational Park, Queensland, Australia (28�9106700S, 151�5804900E).The 37.4-ha site consisted of grassy fields and openeucalypt–secondary-growth pine forests and was home to 200adult females, all of whom had been individually identified forother ongoing studies. The kangaroos were semihabituated tohumans due to the presence of researchers and occasionalcampers. However, they fled when approached closely (within5–10m) by humans and were never fed, caught or manipulatedin any way. Vigilance observations were undertaken over acontinuous 10-day period each month between May and August2010.

Thirty individually identified adult female kangaroos wereused as focal individuals for this study. Identification was basedon females’ facial features, scarring, colouration and patches ofdifferent coloured fur, along with size and reproductive state(Jarman et al. 1989; Carter et al. 2009b). Photographs of sampledfemaleswere taken using aNikonD5000digital SLRcamera, andused to verify the females’ identities by comparison with ourphoto database of all females present in the area. Consistency incorrectly identifying individuals was measured by the two fieldworkers, who undertook joint surveys of the study area duringthe first day of each month of observations and separatelyidentified target females. These data were used to create anoverall interobserver percentage agreement. A total of 68sightings of the 30 focal females were made over five sessionsfor these tests, with an interobserver agreement of 97.1%. Allvigilance data for this paper were collected by AME.

Data collection

Characteristics of focal females were determined as follows.Each time a femalewas sampled her reproductive statewas noted.Reproductive state was divided into five categories: no visiblepouch young, small or medium pouch young, large pouch young,young-at-foot, young-at-foot plus a small or medium pouchyoung. The latter category reflects the fact that female kangarooscan simultaneously provide milk to two young of different ages.Young-at-foot are young kangaroos who are permanentlyexcluded from their mother’s pouch but still take milk from her.Monthly body condition scores were assigned to each female

basedon four categories (from1 = poor condition to 4 = verygoodcondition) that reflected subjective estimates of the amount of fatand muscle covering the hips and ribs. Photographs were takenfrom directly in front of each female when she was feeding withher head down directly in front of her body and comparedamong individuals and between months to ensure consistencyin assigning these condition scores. All body-condition scoreswere estimated by the same observer (ECB).

The boldness of each female was measured using flight-initiation distance (FID) tests (Stankowich and Blumstein 2005;Carter et al. 2010) undertaken in alternatingmorning and eveningfield sessions during five 2-week field trips between March 2010and December 2011, using a different assistant each month toreduce the chance of habituation. Each kangaroo was tested amaximumof four timesbyeachassistant. The assistant positionedthemselves 10m in front of the feeding kangaroo and thenapproached the kangaroo at a steady walking pace until thekangaroo moved her feet from her feeding location. The distancebetween the assistant and the kangaroo at the point when thekangaroo moved away was recorded, using a 30-m measuringtape; this was the FID. Generally, the kangaroos moved only ashort distance away from their feeding positions or moved to aneighbouring group, but occasionally they fled to cover (mean(+s.d.) flight length = 11.83 + 0.56m). Samples were discardedif the target female responded to the reaction of other groupmembers. Between 3 and 21 repeat samples were taken for eachfemale with a mean (�s.d.) of 10.9 repeats� 6.5 repeats. Samplesizes differed among females as FID tests could only be donewhen females were found on the periphery of groups, so thatthey would react to the observer before others did. Each femalewas tested multiple times, and in a range of group sizes, butvariability in FID group sizes was 2.86–4.27. Taking manyfactors, including group size, into consideration (and controllingfor a significant effect of habituation to the test), femalekangaroosin this population have been shown to have consistent individualdifference in their FIDs. The number of times that each individualhad been observed during daily surveys of association patterns ina separate study (this measure should closely approximate theproportion of the time that an individual was present in the studyarea and thus exposed to our presence) was not related to her FID,demonstrating that this is a robust individual trait rather than areflection of levels of habituation to researchers (Best 2013). Wefound large variation among females in their mean FIDs (meanFIDs ranged between 0.67 and 8.11m).

Observations of vigilance behaviour were collected onlyduring active foraging periods in daylight, for ~1.5-h periods aftersunrise and before sunset, so that our data reflected vigilance/foraging trade-offs. The exact times and durations of theseobservation sessions varied, mostly based on temperature; datawere collected only when more than 50% of the population wereactively feeding and only on individuals in groups in which allmembers were still feeding. Observations were not made duringvery low light conditions (i.e. before dawn or after dusk) becauseof the reduced ability to make accurate observations. Each focalfemalewas sampled amaximumof twice per day, once during thedawn feeding session and once during the dusk feeding session.This ensured independence of data as the midday resting periodsand the nights allowed sufficient time for multiple fission–fusionevents in groups (Carter et al. 2009a). The entire field site was

334 Australian Journal of Zoology A. M. Edwards et al.

covered during each session, beginning in different locations tospread out the times and locations at which individuals weresampled. Data on vigilance were not collected during the samesessions as data on females’ FIDs, as vigilance data had to becollected on undisturbed females.

Each time a group containing one of the 30 focal femalekangaroos was encountered a sample was taken. A group wasdefined as a collection of kangaroos in which the most peripheralindividual was no more than 15m from its nearest neighbour(Jarman 1987). Young-at-foot were included in group size countsas adult females were often seen responding to a young-at-foot’svigilance or escape behaviours as they would to those of anotheradult kangaroo. All samples were taken at a distance greater than20m and after all kangaroos in the group had returned to naturalfeeding and vigilance behaviours andwere ignoring the observer.Each sample lasted for 5min and was videorecorded on a SonyHandycam DCR SR68 S. If the observer, another person or avehicle disturbed the groupmembers the samplewas discarded; ifthe group members then began feeding again a new sample wastaken. Sampleswere also discarded if the group size changed or ifthe focal individual moved out of sight of the observer.

At each 5-min sample, the date, time, identity of the focalfemale and the following variables were recorded. An individualwas considered to be on the periphery of its group if no othergroup members were within a continuous arc of at least 180�

around the individual. Groups of six or fewer tend to feed in aline or string rather than in a circular formation; therefore allanimals in groups of six or fewer were considered to beperipheral (Colagross and Cockburn 1993). The distance tocover was recorded as less than or greater than 10m, or 0m if theindividual was in cover. Wind speed was taken as an averageover 10 s, recorded using a Kestrel 3000 Pocket Wind Meter.Windhas been shown to affect the vigilance ofmacropods (Carterand Goldizen 2003). Whether a group contained a mature maleand/or young-at-foot was also recorded, as these could influencefemales’ vigilance.

The videoswere later analysed using the programVLC (http://www.videolan.org/vlc/index.html). Each time the focal animalinterrupted its previous behaviour to scan the environment, avigilance eventwas recorded, aswas the stance used anddirectionof gaze. Stance was divided into two categories: ‘low-intensityvigilance’ in which the individual raised her head to view thesurroundings but left her forepaws on the ground, and ‘high-intensity vigilance’ where the individual raised her head withher forepaws lifted off the ground (Jarman 1987). Direction ofgaze was determined as per Favreau et al. (2010). A female wasconsidered to be exhibiting social vigilance if there was anothermember of the group within 180� of her gaze, or antipredatorvigilance if there were no group members within 180� of hergaze. The direction of gaze was only determined for females whowere located on the periphery of a group as distinguishing thetarget of vigilance for central animals is difficult. While werecognise that we can never be sure of the function of a particularvigilant act, we assume that animals who face out of their groupwhile vigilant are primarily showing antipredator vigilance whilethose who face towards group members are at least capable ofvisual social vigilance (Favreau et al. 2010). Favreau et al. (2010)showed that as group size increased in our kangaroo population,time spent facing conspecifics also increased, supporting the

prediction that vigilance when facing towards conspecifics wasmostly antipredator in function, as shown by a decrease withincreasing group size. Once the videos were analysed, the totalproportion of time spent in vigilance, and the proportions of thetime spent vigilant that were spent in each stance and directionwere calculated.

Data analysis

Four hundred 5-min samples were used in the analyses of thefactors affecting the overall proportion of time spent vigilant andthe proportions of vigilance time spent in the low-intensityvigilance stance. For the analysis of the proportion of vigilancetime spent in antipredator vigilance, only the 247 samples takenon females who were on the periphery of their groups were used.The proportions of vigilance time spent in the two stances addedup to 100%, thus vigilant animals not exhibiting low-intensityvigilance can be assumed to have been showing high-intensityvigilance. The samewas true for antipredator and social vigilancefor peripheral females.

Analysis was undertaken using full linear mixed-effectsregression models fitted using REML (Pinheiro and Bates 2000).An arcsin-square-root transformation (sin–1(Hx)) was performedon all three response variables (overall proportion of time spentin vigilance, proportion of vigilance time spent in low-intensityvigilance, proportion of vigilance time spent in antipredatorvigilance). For the model of the overall proportion of time spentvigilant, this transformation was followed by a Yeo–Johnsontransformation (Yeo and Johnson 2000) as this helped to improvenormality of the residuals. Normality was assessed usingquantile–quantile plots of the residuals. Explanatory variablesincluded wind, group size, the interaction between these twovariables and the mean FID for each individual across all thetests done on it. Presence/absence of males, presence/absenceof young-at-foot, time of day (morning versus afternoon),reproductive state (categories defined above), position in thegroup (internal versus peripheral) and month were included ascategorical variables, while distance to cover (in cover, <10mand �10m) and body condition were included as orderedfactors, and individual ID as a random effect. All variables wereincluded in each of the three models except that position in thegroup was not included in the model investigating the proportionof time spent in antipredator vigilance, as this model used dataonly from females who were on the periphery of groups. Theproportions of variance due to individual differences in theresponse variables were calculated, compared with the totalvariance left unexplained by the fixed effects, to assess the effectof idiosyncratic individual differences on the proportion of timespent vigilant.

Markov Chain Monte Carlo (MCMC) analyses were thenperformed on allmodels, using 100 000 simulations, to determinethe 95% highest probability density (HPD) intervals (Baayenet al. 2008) to establish whether the fixed effects influenced theresponse variables. When the HPD interval did not include zero(i.e. all values were negative or all positive) the fixed effect wasconsidered to have a significant effect on the response variable.P-values for these results were calculated based on MCMCestimates of posterior distributions for the parameters. P-valuesfor fixed factors with more than two levels were calculated from

Effect of individual traits on vigilance in kangaroos Australian Journal of Zoology 335

the posterior distributions of the parameters, assuming that themarginal distributions of the parameters have elliptical contours(Baayen et al. 2008).

All statistical analyses were conducted in R (R Core Team2012), using the packages lme4 (Bates et al. 2011) and languageR(Baayen 2011). All means are presented with standard errors. Apositive HPD interval was interpreted as a positive relationshipwith vigilance, with the opposite being the case for a negativeHPD interval.

Results

The mean (�s.e.) proportion of time spent vigilant for the 400samples was 16.49� 0.79%. On average, over the 400 samplestaken, females spent 49.72� 2.12% of their vigilant time in thelower-intensity stance and 50.28� 2.12% in the higher-intensitystance. Over the 247 samples taken on females that were onthe periphery of groups, females spent a mean of 64.88� 0.02%of their vigilant time facing away from conspecifics and35.12� 0.02% facing towards conspecifics.

Aim 1 – Factors that affected the total proportion of timespent vigilant

The proportion of time that females spent in vigilance increasedsignificantly with their mean FID (Pmcmc = 0.0343); thus shyerfemalesweremorevigilant (Table1;Fig. 1). In addition, vigilancetime varied with month (Pmcmc < 0.0001) and with the time

of day (Pmcmc= 0.0002, with females more vigilant in themorning than the afternoon (mean� s.e. proportions of timevigilant for 30 females = 0.201� 0.015 for morning sessions and0.133� 0.012 for afternoon sessions). There was a trend forvigilance to increase with distance to cover (Pmcmc = 0.0533).The percentage of the variation in the proportion of time spentvigilant that was explained by individual identity was 7.19.

Aim 2 – Factors affecting intensity of vigilance

Females in poorer body condition spent more of their vigilancetime in the lower-intensity posture than did those in bettercondition (Pmcmc = 0.0392) (Table 2; Fig. 2). Time of day anddistance to cover also affected the vigilance stance used(Pmcmc = 0.0051 for time of day; Pmcmc= 0.0109 for distanceto cover), with females spending more of their vigilance timein lower-intensity vigilance in the morning than the afternoon(mean� s.e. proportions of vigilant time in low-intensityvigilance for 30 females = 0.545� 0.037 for morning sessionsand 0.425� 0.043 for afternoon sessions) and when in the forestcompared with out in the open (Fig. 3). Individual identityexplained 14.23% of the variation in this measure of vigilance.

Aim 3 – Factors affecting use of antipredator versussocial vigilance

The only variable that affected the proportion of their vigilanttime that females on the periphery of groups spent facing outfrom the group (assumed to be showing antipredator vigilance)was time of day (Pmcmc = 0.0438) (Table 3). They spent moreof their vigilance time in antipredator vigilance during theafternoon sessions (mean� s.e. proportions of vigilant time inantipredator vigilance = 0.627� 0.041 for morning sessions and0.671� 0.033 for afternoon sessions; n = 29 females for eachsession). For this measure of vigilance, none of the variationwas explained by individual identity.

Table 1. Statistical results for analyses of total proportion of time spentin vigilance

‘Male’ relates towhether there was an adultmale in the group, ‘young’ relatesto whether there was a young-at-foot in the group, ‘reprod’ stands forreproductive states as defined in the Methods section, ‘bci’ means bodycondition index (also defined in Methods section), ‘position’ relates toposition in the group (central versus peripheral), and ‘cover’ relates to distancefrom cover (as defined in the Methods). n= 400 observations for 30 females

Variable Estimate HPD 95 PmcmcLower Upper

(Intercept) 0.2919 0.2162 0.3651 0.000time of day1 –0.0416 –0.0636 –0.0194 0.0002wind 0.0050 –0.0015 0.0109 0.1332size –0.0031 –0.0070 0.0005 0.0909male1 –0.0223 –0.0571 0.0131 0.2196young1 –0.0119 –0.0372 0.0183 0.4955reprod1 0.0028 –0.0420 0.0537 0.7803reprod2 0.0144 –0.0298 0.0666 0.4347reprod3 –0.0162 –0.0795 0.0448 0.5659reprod4 –0.0267 –0.0737 0.0238 0.3142bci.L 0.0348 –0.0090 0.0797 0.1175bci.Q –0.0265 –0.0597 0.0100 0.1584bci.C –0.0050 –0.0252 0.0167 0.6735fid 0.0126 0.0010 0.0239 0.0343position1 0.0031 –0.0353 0.0392 0.9251cover.L 0.0146 –0.0070 0.0356 0.1834cover.Q 0.0165 –0.0028 0.0349 0.0910month2 –0.0493 –0.0840 –0.0100 0.0122month3 –0.0762 –0.1108 –0.0386 0.000month4 –0.0435 –0.0759 –0.0102 0.0110month5 –0.0887 –0.1265 –0.0491 0.000wind:size 0.0000 –0.0006 0.0007 0.8867

0.25

Mea

n pr

opor

tion

time

vigi

lant

0.20

0.15

0.10

2

Mean flight initiation distance (m)

4 6 8

Fig. 1. Meanproportionof time spent vigilant by each femaleplotted againsther mean flight-initiation distance in metres.

336 Australian Journal of Zoology A. M. Edwards et al.

Table 2. Statistical results for analyses of proportion of vigilance timespent in low-intensity vigilance stance

‘Male’ relates towhether therewas an adult male in the group, ‘young’ relatesto whether there was a young-at-foot in the group, ‘reprod’ stands forreproductive states as defined in the Methods section, ‘bci’ means bodycondition index (also defined in Methods section), ‘position’ relates toposition in the group (central versus peripheral), and ‘cover’ relates to distancefrom cover (as defined in the Methods). n= 400 observations for 30 females

Variable Estimate HPD 95 PmcmcLower Upper

(Intercept) 1.2774 0.8727 1.6747 0.0000time of day1 –0.1576 –0.2647 –0.0466 0.0051wind –0.0319 –0.0616 0.0007 0.0466size 0.0013 –0.0163 0.0215 0.792male1 0.0904 –0.0841 0.2631 0.3078young1 0.0526 –0.092 0.1785 0.5238reprod1 –0.1248 –0.3759 0.1153 0.3051reprod2 –0.0076 –0.2827 0.2142 0.8067reprod3 –0.2942 –0.6381 0.0305 0.0761reprod4 0.0167 –0.2556 0.2643 0.9728bci.L –0.3339 –0.6083 –0.1177 0.0062bci.Q –0.0393 –0.2208 0.1368 0.6113bci.C –0.0592 –0.1628 0.0531 0.2905fid –0.0546 –0.1184 0.0154 0.1156position1 –0.0191 –0.1914 0.1781 0.9302cover.L –0.1487 –0.2574 –0.0405 0.0080cover.Q 0.0872 –0.0014 0.1857 0.0567month2 –0.0992 –0.2839 0.0766 0.2742month3 –0.0679 –0.2452 0.1077 0.4456month4 –0.1184 –0.2817 0.0418 0.1455month5 –0.2325 –0.4243 –0.0363 0.0207wind:size 0.0017 –0.0016 0.0046 0.3273

0.8

Mea

n pr

opor

tion

vigi

lanc

e tim

ein

low

inte

nsity

vig

ilanc

e

0.6

0.4

0.2

1.0 1.5 2.0 2.5

Mean body condition index

3.0 3.5 4.0

Fig. 2. Mean proportion of total vigilance time spent in low-intensityvigilance posture for each female plotted against her mean body conditionindex (estimates on a scale of 1 to 4 with 1 meaning very poor and 4 meaningvery good condition).

0.8

1.0

0.6

0.4

0.2

0.0

In cover <10 m from cover

Cover index

>10 m from cover

Mea

n pr

opor

tion

vigi

lanc

e tim

ein

low

inte

nsity

vig

ilanc

e

Fig. 3. Proportions of total vigilance time spent in low-intensity vigilancefor females in cover, <10m from cover or >10m from cover. n= 400observations. The thick lines representmedians, while the tops and bottoms ofboxes show the 25% and 75% quartiles of the data.

Table 3. Statistical results for analyses of the proportion of vigilancetime that was spent facing away from group members and thus

presumably engaged in antipredator vigilance‘Male’ relates towhether therewas an adult male in the group, ‘young’ relatesto whether there was a young–at–foot in the group, ‘reprod’ stands forreproductive states as defined in the Methods section, ‘bci’ means bodycondition index (also defined in Methods section), and ‘cover’ relates todistance from cover (as defined in the Methods). n= 247 observations for

30 females

Variable Estimate HPD 95 PmcmcLower Upper

(Intercept) 1.1695 0.7839 1.5442 0.0000time of day1 0.1615 0.0057 0.3183 0.0438wind –0.0308 –0.0718 0.0098 0.1394size –0.0174 –0.045 0.0093 0.2001male1 –0.0399 –0.2844 0.1801 0.6853young1 –0.0648 –0.2404 0.1157 0.4988reprod1 –0.1557 –0.4364 0.1611 0.3550reprod2 –0.0245 –0.3017 0.2860 0.9506reprod3 0.1244 –0.3056 0.5648 0.5742reprod4 –0.0824 –0.3891 0.2413 0.6413bci.L 0.1821 –0.1004 0.4615 0.2108bci.Q –0.2531 –0.4932 –0.0138 0.0411bci.C 0.0442 –0.1043 0.1742 0.6033fid 0.0190 –0.0515 0.0839 0.6106cover.L 0.0154 –0.1362 0.1595 0.8350cover.Q 0.0725 –0.0592 0.2009 0.2811month2 –0.3186 –0.5694 –0.0786 0.0111month3 –0.0993 –0.3276 0.1302 0.3822month4 –0.1781 –0.3936 0.0413 0.1074month5 –0.2871 –0.5408 –0.0348 0.0292wind:size 0.0014 –0.0024 0.0052 0.4761

Effect of individual traits on vigilance in kangaroos Australian Journal of Zoology 337

DiscussionWe found that some characteristics of individual females,including boldness and body condition, had significant thoughweak effects on vigilance patterns, while reproductive state hadno effect. Different aspects of vigilance (i.e. targets and postures)were affected by these variables in different ways. Month, timeof day and distance to cover also affected females’ vigilance.Even after taking all these predictors into account, 7% and 14%respectively, of the total time spent vigilant and of the proportionof vigilance time spent in the lower-intensity vigilance posture,were explained by individual identity. Thus, vigilance is abehaviour that is affected by many different factors. There aretwo reasons why it is probably not surprising that effects ofindividuals’ characteristics on vigilance, while present, wereweak. The first is that so many different factors affect animals’vigilance (reviewed by Elgar 1989); thus any single factor shouldhave only a weak effect on vigilance patterns. The second is thatpredation risk was low for adults at our study site, due to theabsence of dingos/wild dogs; in such a low-risk situation,differences in individuals’behavioural strategiesmaybe reduced.

Our analysis suggests that shyer, or more risk-averse,females in our population of wild eastern grey kangaroos spentsignificantly more time vigilant than did bolder kangaroos. Toour knowledge this is the first finding of a link betweenboldness and vigilance in wild animals. Individuals differedsignificantly in their flight-initiation distances (controlling for asignificant effect of habituation to the test). However, the numberof times that each individual had been observed during dailysurveys of association patterns in a separate study (this measureshould closely approximate the proportion of time that anindividual was present in the study area and thus exposed to ourpresence) was not related to its flight-initiation distance (Best2013). This provides evidence that interindividual differencesin flight-initiation distances are a good measure of differencesin individual’s behaviour, rather than a reflection of levels ofhabituation to researchers.

We found that females in better body condition spent no moreor less total time vigilant than those in poorer condition butspentmoreof their vigilance time inhigher-vigilanceposture.Theuse of a higher-vigilance posture by kangaroos presumablyincreases the chance of detecting danger and important socialoccurrences, but at a greater cost to energy expenditure, as itinvolves the kangaroo lifting its front feet off the ground andstanding straighter. There has been little research done on theeffect of body condition on vigilance. An exception is aninvestigation of the effects of body condition on the vigilanceof European rabbits (Oryctolagus cuniculus) (Monclús andRödel 2009), which found that vigilance was related to bodycondition in juvenile rabbits in field enclosures, although not inadults. The authors proposed that young rabbits in poor conditionreduced their vigilance levels in favour of feeding and thusgrowth. Other aspects of antipredator behaviour have been foundto vary with body condition in a similar way. A study on yellow-bellied marmots (Marmota flaviventris) found that animals inbetter condition suppressed foraging for longer after playbacksof alarm calls (Lea and Blumstein 2011). Likewise, food-supplemented Arabian babblers (Turdoides squamiceps) spentmore time in sentinel behaviour than did non-supplementedindividuals (Wright et al. 2001), suggesting that individuals in

better condition could afford to spend more time in thisantipredator behaviour. Animals in poor condition often havelower energetic reserves (Dewitt et al. 1999); vigilance behaviouris expected to be more costly for them because they need toincrease feeding in order to survive (Preisser et al. 2005;Monclúsand Rödel 2009). While we did not find an effect of bodycondition on the total amount of time females spent vigilant,females in poorer condition exhibited significantly less of thelikely more energetically costly high-intensity vigilance thandid females in better condition. It would be interesting to furtherinvestigate vigilance in relation to body condition, to test whetherindividuals differ in this relationship; some individuals mayadjust their use of different vigilance postures whereas othersmay adjust their time commitments to vigilance.

We also found that the amount of time females spent vigilantvaried significantly with month and time of day, with femalessignificantly more vigilant in the morning, but more likely to usethe high-intensity vigilance stance in the afternoon and whenaway from cover. Time-of-day effects on vigilance could eitherrelate to the activity times of predators (Smith et al. 2004; Valeixet al. 2009) or to trade-offs with feeding if animals are hungrierat certain times of the day. There is some evidence that bothdingos (Harden 1985) and red foxes are more active in themorning than at other times (Heptner and Naumov 1998), butother work suggests that the temporal patterns of activity ofpredators inAustralia arequite variable (Brook et al. 2012).Whiledingos were not apparent at our study site, many young weretaken by foxes (unpubl. data). It is also possible that if kangaroosfeed during the night, they may be hungrier in the afternoon,after resting for 6–8 h, than in the morning, and thus be able to bemore vigilant in the morning. The greater use of high-intensityvigilance in the afternoon may help them compensate forspending less time vigilant then, particularly as kangaroos canmoderate the cost of vigilance by continuing to chew their foodwhile vigilant (Unck et al. 2009). Understanding these patternswill require more knowledge of the behaviours of kangaroos’predators and of their patterns of hunger. We cannot explain atthis point why vigilance patterns varied among months, butthe quality and quantity of grass varied greatly, depending onrainfall patterns, and this could have affected the animals’vigilance–feeding trade-offs. The greater use of low-intensityvigilance in the wooded areas than out in the open suggeststhat the kangaroos perceived a lower risk of predation whenin cover.

We found fewer relationships between social factors andvigilance than was predicted, with no significant relationshipsbetween measures of vigilance and group size or the presence ofadult males or young in groups. Increases in social vigilancewith group size may cancel out expected decreases inantipredator vigilance (Favreau et al. 2010). In addition, ourvigilance observations were done in winter, when groups tendedto be smaller, animals were in poorer condition, adult malesrarely associated with groups of females and most young werestill in their mothers’ pouches (unpubl. data). The seasonality ofour study may thus explain why reproductive state and variablesrelating to group size and composition did not relate significantlyto any measures of vigilance in our study.

The final aspect of vigilance that we investigated was whetherfemales on the periphery of groups faced into the group (‘social

338 Australian Journal of Zoology A. M. Edwards et al.

vigilance’) or out (‘antipredator vigilance’). Apart from timeof day, with females more likely to be vigilant facing out ofthe group in the afternoon than in the morning, none of thevariables had a significant effect. Our sample sizes were smallerfor this part of our analysis because we could only use samplesdone on females located on the periphery of their groups. Wesuggest that, given the large number of variables that affectvigilance, a larger study may be required to understand thedeterminants of these different kinds of vigilance. However, itis perhaps not surprising that boldness was not related to howmuch of an individual’s total vigilance was social versusantipredator in nature, as shyer animals might be expected toshow more of both forms of vigilance.

An ideal analysis of the relationship between boldness andvigilance would take into consideration the effects of fixedeffects on both vigilance and flight-initiation distance measuresin order to separate within- and between-individual correlationsin behavioural traits. We have considered these other fixedeffects for vigilance in our analysis, but could not also do so forthe FID measures. This is because our FID tests were done atdifferent times from the focal samples of vigilance behaviour;thus we have two different datasets, one on vigilance and oneon FIDs. This precluded the use of methods such as bivariatemixed-effects models, recommended by Wilson et al. (2011)and Dingemanse et al. (2012). Garamszegi and Herczeg (2012)discuss the statistical problems involved in considering within-and between-individual behavioural correlations together. Theyparticularly point to two problems that apply to our study. Thefirst is that data on two behavioural traits cannot usually becollected at the same time, particularly in the case of field studies.For example, inour study itwould clearlyhavebeen inappropriateto carry out flight-distance tests and vigilance observations atthe same time; there is thus no obvious way around the problemof having two separate datasets. Garamszegi and Herczeg (2012)also carried out simulations to estimate the sorts of sample sizesthat would be required for approaches based on mixed modelsto be useful for this kind of situation; their conclusion was thatthe necessary sample sizes would rarely be achievable in fieldstudies such as ours. They thus suggest that, while not perfect,the best approach currently available for studying between-individual correlations in behaviour is to establish that eachbehaviour is repeatable, then correlate individual-specificmeasures of the two behaviours. Because so many things affectanimals’ vigilance behaviour, even moment to moment, wewould not necessarily expect patterns of vigilance to be highlyrepeatable. Indeed, the percentages of the variance in vigilancepatterns that were explained by individuals’ identities in ourstudy ranged from 0 to 14.23. We thus took a different approach,which was to test for the effects of a variety of variables onvigilance in our modelling of the relationships betweenmeasures of vigilance and flight-initiation distances. While weaccept that our statistical approach is not perfect, we agree withGaramszegi and Herczeg (2012) that a better approach is notyet available and encourage further development of statisticalapproaches to this problem.

We also investigated the relationships between vigilance anda large range of other factors, in addition to personality, inwild herbivores while controlling for individual identity andtherefore avoiding the pseudoreplication that has likely occurred

in many studies of vigilance. We found several significantpredictors of vigilance, with some of these variables predictingtotal time in vigilance, and others predicting vigilance postureor the direction (target) of vigilance. We suspect that ourresults were strongly affected by season, which affects foodavailability, group size, body condition and reproductive state inkangaroos; thus a similar study over the summer period isnecessary for a fuller understanding of vigilance patterns in thisspecies.

Acknowledgements

We thank Peter Hasselgrove and Ian Elms, Queensland Parks and WildlifeService rangers at Sundown National Park, for their support of our work.This research was approved by the University of Queensland’s AnimalExperimentation Ethics Committee and conducted under a ScientificPurposes Permit from Queensland’s Environmental Protection Agency. Wethank Julien Martin, Niels Dingemanse and an anonymous reviewer fortheir helpful critiques of an earlier version of this manuscript.This research complied with the laws of Australia and was carried out withclearance from the University of Queensland’s Native and Exotic Wildlifeand Marine Animals Animal Experimentation Ethics Committee and aScientific Purposes Permit from Queensland Parks and Wildlife Service.The authors declare that they have no conflict of interest.

References

Baayen, R. H. (2011). LanguageR: Data sets and functions with ‘AnalyzingLinguistic Data: A Practical Introduction to Statistics’. R package version1.2. Available at: http://CRAN.R-project.org/package=languageR

Baayen, R., Davidson, D., and Bates, D. (2008). Mixed-effects modelingwith crossed random effects for subjects and items. Journal of Memoryand Language 59, 390–412. doi:10.1016/j.jml.2007.12.005

Bachman, G. C. (1993). The effect of body condition on the trade-offbetween vigilance and foraging in Belding’s ground squirrels. AnimalBehaviour 46, 233–244. doi:10.1006/anbe.1993.1185

Banks, P. B. (2001). Predation-sensitive grouping and habitat use by easterngrey kangaroos: a field experiment. Animal Behaviour 61, 1013–1021.doi:10.1006/anbe.2001.1686

Bates, D., Maechler, M., and Bolker, B. (2011). lme4: Linear mixed-effectsmodels using S4 classes. R package version 0.999375-42. Available at:http://CRAN.R-project.org/package=lme4

Bell, A. M. (2007). Future directions in behavioural syndromes research.Proceedings of the Royal Society of London. Series B, Biological Sciences274, 755–761. doi:10.1098/rspb.2006.0199

Bergvall, U. A., Schäpers, A., Kjellander, P., and Weiss, A. (2011).Personality and foraging decisions in fallow deer (Dama dama). AnimalBehaviour 81, 101–112. doi:10.1016/j.anbehav.2010.09.018

Best, E. C. (2013). Network analysis of the social structure of female easterngrey kangaroos (Macropus giganteus). Ph.D. Thesis. Univeristy ofQueensland, Brisbane.

Blumstein, D. T., Runyan, A., Seymour, M., Nicodemus, A., Ozgul, A.,Ransler, F., Im, S., Stark, T., Zugmeyer, C., and Daniel, J. C. (2004).Locomotor ability and wariness in yellow-bellied marmots. Journal ofEthology 110, 615–634. doi:10.1111/j.1439-0310.2004.01000.x

Brook, L. A., Johnson, C. N., and Ritchie, E. G. (2012). Effects of predatorcontrol on behaviour of an apex predator and indirect consequences formesopredator suppression. Journal of Applied Ecology 49, 1278–1286.doi:10.1111/j.1365-2664.2012.02207.x

Brown, J. S. (1999). Vigilance, patch use and habitat selection: foragingunder predation risk. Evolutionary Ecology Research 1, 49–71.

Burger, J., and Gochfeld, M. (1994). Vigilance in African mammals –

differences among mothers, other females, and males. Behaviour 131,153–169. doi:10.1163/156853994X00415

Effect of individual traits on vigilance in kangaroos Australian Journal of Zoology 339

Cameron, E. Z., and du Toit, J. T. (2005). Social influences on vigilancebehaviour in giraffes, Giraffa camelopardalis. Animal Behaviour 69,1337–1344. doi:10.1016/j.anbehav.2004.08.015

Carter, K., and Goldizen, A. W. (2003). Habitat choice and vigilancebehaviour of brush-tailed rock-wallabies (Petrogale penicillata) withintheir nocturnal foraging ranges. Wildlife Research 30, 355–364.doi:10.1071/WR02095

Carter,A. J.,Macdonald,S.L., Thomson,V.A., andGoldizen,A.W. (2009a).Structured association patterns and their energetic benefits in femaleeastern grey kangaroos, Macropus giganteus. Animal Behaviour 77,839–846. doi:10.1016/j.anbehav.2008.12.007

Carter, A. J., Pays, O., and Goldizen, A. W. (2009b). Individual variation inthe relationship between vigilance and group size in eastern greykangaroos. Behavioral Ecology and Sociobiology 64, 237–245.doi:10.1007/s00265-009-0840-4

Carter, A. J., Goldizen, A. W., and Tromp, S. A. (2010). Agamas exhibitbehavioural syndromes: bolder males bask and feed more but may sufferhigher predation. Behavioral Ecology 21, 655–661. doi:10.1093/beheco/arq036

Childress, M. J., and Lung, M. A. (2003). Predation risk, gender andthe group size effect: does elk vigilance depend upon the behaviourof conspecifics? Animal Behaviour 66, 389–398. doi:10.1006/anbe.2003.2217

Clarke, J. L., Jones, M. E., and Jarman, P. J. (1995). Diurnal and nocturnalgrouping and foraging behaviours of free-ranging eastern grey kangaroos.Australian Journal of Zoology 43, 519–529. doi:10.1071/ZO9950519

Colagross, A. M. L., and Cockburn, A. (1993). Vigilance and grouping inthe eastern gray kangaroo, Macropus giganteus. Australian Journal ofZoology 41, 325–334. doi:10.1071/ZO9930325

Cork, S. J. (1991). Meeting the energy requirements for lactation in amacropodid marsupial: current nutrition versus stored body reserves.Journal of Zoology 225, 567–576. doi:10.1111/j.1469-7998.1991.tb04325.x

Dannock, R. J., Blomberg, S. P., and Goldizen, A. W. (2013). Individualvariation in vigilance in female eastern grey kangaroos. AustralianJournal of Zoology 61, 312–319. doi:10.1071/ZO12122

Dewitt, T. J., Sih, A., and Hucko, J. A. (1999). Trait compensation andcospecialization in a freshwater snail: size, shape and antipredatorbehaviour.Animal Behaviour 58, 397–407. doi:10.1006/anbe.1999.1158

Dingemanse, N. J., Dochtermann, N. A., and Nakagawa, S. (2012). Definingbehavioural syndromes and the role of ‘syndromedeviation’ in explainingtheir evolution. Behavioral Ecology and Sociobiology 66, 1543–1548.doi:10.1007/s00265-012-1416-2

Dyer, J. R. G., Croft, D. P., Morrell, L. J., and Krause, J. (2009). Shoalcomposition determines foraging success in the guppy. BehavioralEcology 20, 165–171. doi:10.1093/beheco/arn129

Elgar, M. A. (1989). Predator vigilance and group-size in mammals andbirds – a critical review of the empirical evidence. Biological Reviewsof the Cambridge Philosophical Society 64, 13–33. doi:10.1111/j.1469-185X.1989.tb00636.x

Fairbanks, B., and Dobson, F. S. (2007). Mechanisms of the group-sizeeffect on vigilance in Columbian ground squirrels: dilution versusdetection. Animal Behaviour 73, 115–123. doi:10.1016/j.anbehav.2006.07.002

Favreau, F. R., Goldizen, A. W., and Pays, O. (2010). Interactions amongsocial monitoring, anti-predator vigilance and group size in eastern greykangaroos. Proceedings of the Royal Society of London. Series B,Biological Sciences 277, 2089–2095. doi:10.1098/rspb.2009.2337

Garamszegi, L. Z., and Herczeg, G. (2012). Behavioural syndromes,syndrome deviation and the within- and between-individual componentsof phenotypic correlations: when reality does not meet statistics.Behavioral Ecology and Sociobiology 66, 1651–1658. doi:10.1007/s00265-012-1439-8

Gaynor, K. M., and Cords, M. (2012). Antipredator and social monitoringfunctions of vigilance behaviour in blue monkeys. Animal Behaviour 84,531–537. doi:10.1016/j.anbehav.2012.06.003

Harden, R. H. (1985). The ecology of the dingo in north-eastern New SouthWales. 1. Movements and home range. Australian Wildlife Research 12,25–37. doi:10.1071/WR9850025

Heathcote, C. F. (1987). Grouping of eastern grey kangaroos in openhabitat. Australian Wildlife Research 14, 343–348. doi:10.1071/WR9870343

Heptner, V., and Naumov, N. (1998). Sirenia and Carnivora (seacows, bearsandwolves). In ‘Mammals of the Soviet Union. Vol. II, Part 1a’. (SciencePublishers, Inc.: USA.)

Hoogland, J. L., Hale, S. L., Kirk, A. D., and Sui, Y. D. (2013). Individualvariation in vigilance among white-tailed prairie dogs (Cynomysleucurus). The Southwestern Naturalist, in press.

Houston, A. I., McNamara, J. M., and Hutchinson, J. M. C. (1993). Generalresults concerning the trade-off between gaining energy and avoidingpredation. Philosophical Transactions of the Royal Society of London.Series B, Biological Sciences 341, 375–397. doi:10.1098/rstb.1993.0123

Hunter, L. T. B., and Skinner, J. D. (1998). Vigilance behaviour in Africanungulates: the role of predation pressure. Behaviour 135, 195–211.doi:10.1163/156853998793066320

Jarman, P. J. (1987). Group-size and activity in eastern grey kangaroos.Animal Behaviour 35, 1044–1050. doi:10.1016/S0003-3472(87)80161-6

Jarman, P. J. (1994). Individual behaviour and social organisation ofkangaroos. In ‘Animal Societies: Individuals, Interactions andOrganisation’. (Eds P. J. Jarman, and A. Rossiter.) pp. 69–85. (KyotoUniversity Press: Kyoto.)

Jarman, P. J., Jones, M. E., Johnson, C. N., Southwell, C. J., Stuartdick, R. I.,Higginbottom, K. B., and Clarke, J. L. (1989). Macropod studies atWallaby Creek. 8. Individual recognition of kangaroos and wallabies.Australian Wildlife Research 16, 179–185. doi:10.1071/WR9890179

Kurvers, R. H. J. M., Prins, H. H. T., van Wieren, S. E., van Oers, K., Nolet,B. A., and Ydenberg, R. C. (2010). The effect of personality on socialforaging: shy barnacle geese scrounge more. Proceedings of the RoyalSociety of London. Series B, Biological Sciences 277, 601–608.doi:10.1098/rspb.2009.1474

Lea, A. J., and Blumstein, D. T. (2011). Age and sex influence marmotantipredator behaviour during periods of heightened risk. BehavioralEcology and Sociobiology 65, 1525–1533. doi:10.1007/s00265-011-1162-x

Lima, S. L. (1995). Back to the basics of antipredatory vigilance – the group-size effect. Animal Behaviour 49, 11–20. doi:10.1016/0003-3472(95)80149-9

Lung, M. A., and Childress, M. J. (2007). The influence of conspecifics andpredation risk on the vigilance of elk (Cervus elaphus) in YellowstoneNational Park. Behavioral Ecology 18, 12–20. doi:10.1093/beheco/arl066

Monclús, R., and Rödel, H. G. (2009). Influence of different individual traitson vigilance behaviour in European rabbits. Journal of Ethology 115,758–766. doi:10.1111/j.1439-0310.2009.01661.x

Pangle,W.M., andHolekamp,K. E. (2010). Functions of vigilance behaviourin a social carnivore, the spotted hyaena, Crocuta crocuta. AnimalBehaviour 80, 257–267. doi:10.1016/j.anbehav.2010.04.026

Patrick, S.C., Chapman, J.R.,Dugdale,H.L.,Quinn, J. L., andSheldon,B.C.(2012). Promiscuity, paternity and personality in the great tit.Proceedingsof the Royal Society of London. Series B, Biological Sciences 279,1724–1730. doi:10.1098/rspb.2011.1820

Pays, O., and Jarman, P. J. (2008). Does sex affect both individual andcollective vigilance in social mammalian herbivores: the case of theeastern grey kangaroo? Behavioral Ecology and Sociobiology 62,757–767. doi:10.1007/s00265-007-0501-4

340 Australian Journal of Zoology A. M. Edwards et al.

Pays, O., Goulard, M., Blomberg, S. P., Goldizen, A. W., Sirot, E., andJarman, P. J. (2009). The effect of social facilitation on vigilance in theeastern gray kangaroo, Macropus giganteus. Behavioral Ecology 20,469–477. doi:10.1093/beheco/arp019

Pinheiro, J. C., and Bates, D. M. (2000). ‘Mixed-effects Models in S andS-PLUS.’ (Springer.)

Preisser, E. L., Bolnick, D. I., and Benard,M. F. (2005). Scared to death? Theeffects of intimidation and consumption in predator–prey interactions.Ecology 86, 501–509. doi:10.1890/04-0719

Pulliam, H. R. (1973). Advantages in flocking. Journal of TheoreticalBiology 38, 419–422. doi:10.1016/0022-5193(73)90184-7

R Core Team (2012). R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria.Available at: http://www.R-project.org

Réale, D., Reader, S. M., Sol, D., McDougall, P. T., and Dingemanse, N. J.(2007). Integrating animal temperament within ecology and evolution.Biological Reviews of the Cambridge Philosophical Society 82,291–318. doi:10.1111/j.1469-185X.2007.00010.x

Rieucau, G., andMartin, J. G. A. (2008). Many eyes or many ewes: vigilancetactics in female bighorn sheep Ovis canadensis vary according toreproductive status.Oikos 117, 501–506. doi:10.1111/j.0030-1299.2008.16274.x

Rieucau, G., Morand-Ferron, J., and Giraldeau, L.-A. (2010). Group sizeeffect in nutmeg mannikin: between-individuals behavioural differencesbut same plasticity. Behavioral Ecology 21, 684–689. doi:10.1093/beheco/arq039

Roberts,G. (1996).Why individual vigilance declines as group size increases.Animal Behaviour 51, 1077–1086. doi:10.1006/anbe.1996.0109

Shorrocks, B., and Cockayne, A. (2005). Vigilance and group size in impala(Aepyceros melampus Lichtenstein): a study in Nairobi National Park,Kenya. African Journal of Ecology 43, 91–96. doi:10.1111/j.1365-2028.2005.00541.x

Sih, A., Bell, A., and Johnson, J. C. (2004). Behavioural syndromes: anecological and evolutionary overview.Trends in Ecology&Evolution 19,372–378. doi:10.1016/j.tree.2004.04.009

Smith, A. C., Kelez, S., and Buchanan-Smith, H.M. (2004). Factors affectingvigilance within wild mixed-species troops of saddleback (Saguinusfuscicollis) andmoustached tamarins (S.mystax).BehavioralEcologyandSociobiology 56, 18–25. doi:10.1007/s00265-003-0753-6

Southwell, C. J. (1984a). Variability in grouping in the eastern greykangaroo, Macropus giganteus. 1. Group density and group-size.Australian Wildlife Research 11, 423–435. doi:10.1071/WR9840423

Southwell, C. J. (1984b). Variability in grouping in the eastern grey kangaroo,Macropusgiganteus. 2.Dynamicsofgroup formation.AustralianWildlifeResearch 11, 437–449. doi:10.1071/WR9840437

Stankowich, T., andBlumstein, D. T. (2005). Fear in animals: ameta-analysisand review of risk assessment. Proceedings of the Royal Society ofLondon. Series B, Biological Sciences 272, 2627–2634. doi:10.1098/rspb.2005.3251

Steenbeek,R., Piek,R.C., vanBuul,M., andvanHooff, J. (1999).Vigilance inwild Thomas’s langurs (Presbytis thomasi): the importance of infanticiderisk. Behavioral Ecology and Sociobiology 45, 137–150. doi:10.1007/s002650050547

Treves, A., Drescher, A., and Snowdon, C. T. (2003). Maternal watchfulnessin black howler monkeys (Alouatta pigra). Journal of Ethology 109,135–146. doi:10.1046/j.1439-0310.2003.00853.x

Unck, C. E., Waterman, J. M., Verburgt, L., and Bateman, P. W. (2009).Quantity versus quality: how does level of predation threat affect Capeground squirrel vigilance? Animal Behaviour 78, 625–632. doi:10.1016/j.anbehav.2009.05.028

Valeix, M., Fritz, H., Loveridge, A. J., Davidson, Z., Hunt, J. E.,Murindagomo, F., and Macdonald, D. W. (2009). Does the risk ofencountering lions influence African herbivore behaviour at waterholes?Behavioral Ecology and Sociobiology 63, 1483–1494. doi:10.1007/s00265-009-0760-3

Wilson, A. D. M., and Godin, J.-G. J. (2010). Boldness and intermittentlocomotion in the bluegill sunfish, Lepomis macrochirus. BehavioralEcology 21, 57–62. doi:10.1093/beheco/arp157

Wilson, A. J., de Boer, M., Arnott, G., and Grimmer, A. (2011). Integratingpersonality research and animal contest theory: aggressiveness inthe green swordtail Xiphophorus helleri. PLoS ONE 6(11), e28024.doi:10.1371/journal.pone.0028024

Wolf, M., van Doorn, G. S., Leimar, O., and Weissing, F. J. (2007). Life-history trade-offs favour the evolution of animal personalities. Nature447, 581–584. doi:10.1038/nature05835

Wright, J., Maklakov, A. A., and Khazin, V. (2001). State-dependentsentinels: an experimental study in the Arabian babbler. Proceedingsof the Royal Society of London. Series B, Biological Sciences 268,821–826. doi:10.1098/rspb.2000.1574

Yeo, I. K., and Johnson, R. A. (2000). A new family of power transformationsto improve normality or symmetry. Biometrika 87, 954–959.doi:10.1093/biomet/87.4.954

Handling Editor: Paul Cooper

Effect of individual traits on vigilance in kangaroos Australian Journal of Zoology 341

www.publish.csiro.au/journals/ajz