23
MURDOCH RESEARCH REPOSITORY This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination. The definitive version is available at http://dx.doi.org/10.1111/mms.12260 Sprogis, K.R., Raudino, H.C., Rankin, R., MacLeod, C.D. and Bejder, L. (2016) Home range size of adult Indo-Pacific bottlenose dolphins (Tursiops aduncus) in a coastal and estuarine system is habitat and sex-specific. Marine Mammal Science, 32 (1). pp. 287-308. http://researchrepository.murdoch.edu.au/28451 Copyright: © 2015 Society for Marine Mammalogy It is posted here for your personal use. No further distribution is permitted.

Home range size of adult Indo-Pacific bottlenose dolphins (Tursiops

  • Upload
    buingoc

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

MURDOCH RESEARCH REPOSITORY

This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination.

The definitive version is available at http://dx.doi.org/10.1111/mms.12260

Sprogis, K.R., Raudino, H.C., Rankin, R., MacLeod, C.D. and Bejder, L. (2016) Home range size of adult Indo-Pacific bottlenose dolphins (Tursiops aduncus) in a coastal and

estuarine system is habitat and sex-specific. Marine Mammal Science, 32 (1). pp. 287-308.

http://researchrepository.murdoch.edu.au/28451

Copyright: © 2015 Society for Marine Mammalogy

It is posted here for your personal use. No further distribution is permitted.

Home range size of adult Indo-Pacific bottlenose dolphins(Tursiops aduncus) in a coastal and estuarine system is

habitat and sex-specific

KATE R. SPROGIS,1 HOLLY C. RAUDINO,2 and ROBERT RANKIN, Murdoch University

Cetacean Research Unit, School of Veterinary and Life Sciences, Murdoch University, 90 South

Street, Murdoch 6150, Western Australia, Australia; COLIN D. MACLEOD, GIS In Ecology,

120 Churchill Drive, Glasgow, G11 7EZ, United Kingdom; LARS BEJDER,Murdoch Univer-

sity Cetacean Research Unit, School of Veterinary and Life Sciences, Murdoch University, 90

South Street, Murdoch 6150, Western Australia, Australia.

Abstract

This study examined sex-specific differences in home range size of adult Indo-Pacific bottlenose dolphins off Bunbury, Western Australia. We applied a new ker-nel density estimation approach that accounted for physical barriers to movements.A Bayesian mixture model was developed to estimate a sex effect in home range sizewith latent group partitioning constrained by association data. A post hoc analysisinvestigated group partitioning relating to the proportion of time spent in open vs.sheltered waters. From 2007 to 2013, photographic-identification data were col-lected along boat-based systematic transect lines (n = 586). Analyses focused onadult dolphins of known sex (sighted ≥ 30 times; n = 22 males and 34 females). The95% utilization distributions of males varied between 27 and 187 km2 (�x� SD;94.8� 48.15) and for females between 20 and 133 km2 (65.6� 30.9). The mixturemodel indicated a 99% probability that males had larger home ranges than females.Dolphins mostly sighted in open waters had larger home ranges than those in shel-tered waters. Home ranges of dolphins sighted in sheltered waters overlapped withareas of highest human activity. We suggest that sex differences in home ranges aredriven by male mating strategies, and home range size differences between habitatsmay be influenced by prey availability and predation risk.

Key words: home range, barriers, utilization distribution, kernel density estimation,Bayesian mixture model, sex-specific, bottlenose dolphin, Tursiops aduncus.

Understanding the characteristics of an animal’s home range provides insights intothe species’ ecology (Worton 1989). The concept of home range was originallydescribed as “the area traversed by the individual in its normal activities of food gath-ering, mating, and caring for young” (Burt 1943). While this definition has beenwidely applied to a range of taxa, it does not incorporate an animal’s intensity of usewithin its home range, i.e., it assumes that every location within a home range is ofequal importance to the individual (Don 1949, Seaman and Powell 1996). The study

1Corresponding author (e-mail: [email protected]).2nee Smith.

1

of home range has progressed from the early attempts that identified distributions viasimple outlines encompassing the area of use, i.e., minimum convex polygon (Mohr1947), to methods that describe utilization distribution, which examine the intensityof use of different locations within a home range or study area (Van Winkle 1975,Kie et al. 2010). In particular, kernel density estimation (KDE; Silverman 1986) isone of the most common methods to estimate utilization distribution and was intro-duced to ecology through Worton’s (1989) seminal paper.Originally, KDE was developed for species that move freely throughout a landscape

where barriers to movement were not encountered (Knight et al. 2009). However, thepresence of a barrier, such as a river for terrestrial species or a coastline for marine spe-cies, can physically prevent movement. This means that for such species, the conven-tional home range approach potentially includes unavailable areas, thus providing anoverestimate of an animal’s home range. Therefore, using methods that account forbarriers is important to eliminate this potential bias and improve the accuracy of theestimated home range (e.g., Getz and Wilmers 2004, Knight et al. 2009, Benhamouand Cornelis 2010). Accurate estimates of an animal’s home range are of biologicalinterest and important for conservation applications, such as reserve design (Maxwellet al. 2011), management of threatened populations (Seminoff et al. 2002), and iden-tification of overlap with anthropogenic impacts (Rayment et al. 2009).Home range size can be sex-specific for both marine (e.g., spottail shark, Carcharhi-

nus sorrah, Knip et al. 2012; gray seal, Halichoerus grypus, Austin et al. 2004) and ter-restrial species (e.g., chimpanzee, Pan troglodytes, Chapman and Wrangham 1993;grizzly bear, Ursus arctos horribilis, Mace and Waller 1997). Sex-specific differences inhome range size and location can have implications for conservation (Wearmouth andSims 2008). For instance, differences can render a particular sex more vulnerable tohuman impacts. New Zealand sea lions (Phocarctos hookeri) off the Auckland Islands,New Zealand (Leung et al. 2012) and wandering albatross (Diomedea exulans) offSouth Georgia (Xavier et al. 2004) are examples where female foraging ranges havelarger overlap with fisheries than males, resulting in higher female bycatch mortality.Reduced survival of females by fishing activity may lead to reduced reproductive out-put and result in population decline (Wearmouth and Sims 2008, Leung et al.2012). Sex-biased impacts therefore emphasize the importance of estimating homerange characteristics separately for each sex.For the common bottlenose dolphin (Tursiops truncatus), sex-specific home range

characteristics vary between geographic locations. In some locations, there are noapparent differences between sexes (e.g., Wilson et al. 1997, Gubbins 2002, Lynn andW€ursig 2002, Silva et al. 2008). In the Azores Archipelago, for example, the lack ofsexual differences and large home range sizes of dolphins are suggested to be relatedto the patchy prey distribution and lower productivity of the oceanic waters com-pared to coastal areas (Silva et al. 2008). In contrast, in other locations adult malesrange farther than adult females (e.g., Scott et al. 1990, Owen et al. 2002, Urian et al.2009). For example, in Sarasota Bay, Florida, males range farther than females toincrease mating opportunities (Owen et al. 2002). Differences in the spatial distribu-tion between and within populations may result from differing habitat characteristics(Ballance 1992, Defran and Weller 1999, Martinez-Serrano et al. 2011) and basicbiological priorities, such as, mating strategies, prey availability and predation risk(Matthiopoulos and Aarts 2010).In the closely related Indo-Pacific bottlenose dolphin (T. aduncus), however, less is

known about sex-specific differences in home range sizes (if any). In the ClarenceRiver, Australia, Fury et al. (2013) explored differences of space use between the sexes

2

of T. aduncus, although due to a small sample size of males they compared femalehome ranges with mixed sex groups. In Shark Bay, Australia, ranges for males andfemales are highly variable, however, on average males range further than females (av-erage 90% KDE for males = 65.9 km2 and females = 52.9 km2; Watson-Capps2005, Randic et al. 2012). In Bunbury, Western Australia, Smith et al. (2013) sug-gested that adult male dolphins range over larger areas than adult females whensearching for prey and mating opportunities.This study tested whether there were sex-specific differences in home range size for

adult T. aduncus in coastal and sheltered waters off Bunbury, Western Australia. Toestimate home range size, we applied a new KDE method that specifically accountsfor physical barriers to movements. This method is based around the “kernel interpo-lation with barriers” tool found in Esri’s ArcGIS 10 GIS software, and is used toinspect the 95% utilization distribution of individuals within the study area. In addi-tion, we developed a Bayesian mixture model to (1) test whether there was a sex effectin home range size and (2) explore whether dolphins could be partitioned intogroups, based on home range size, associations with conspecifics, and by habitat (openvs. sheltered waters). The latter was carried out to identify patterns that may haveimportant ecological and conservation implications.

Methods

Study Site

The study took place in Bunbury (33�320S, 115�630E), southwestern Australia(Fig. 1). Bunbury has one of the largest shipping ports in the state (Bunbury PortAuthority 2013). From 2007 to 2011, the study area encompassed 120 km2 whichwas surveyed along three transect routes: Buffalo Beach, Back Beach, and Inner watertransects (Fig. 1). In August 2011, the study area was extended 9.3 km from shoreand increased to 540 km2 with the addition of three new transect routes: BuffaloBeach offshore, Back Beach offshore, and Busselton (Fig. 1). Water depth rangedfrom ≤1 m to 24 m, with a low tidal range generally <1 m. The benthic habitat con-sists of temperate limestone reefs, seagrass, macroalgae communities, sand and mudflats (Smith 2012; KRS, unpublished data).

Sampling Design

Dolphin identities and sighting location were documented during systematic, boatphotographic-identification (photo-ID) surveys, in all austral seasons (summer [De-cember–February], autumn [March–May], winter [June–August], and spring[September–November]) between March 2007 and August 2013. Surveys were con-ducted along predetermined transect routes (Fig. 1) at 10 kn using a 5 m researchvessel with an 80 hp engine. Traversing a transect was defined as a survey, with eachdolphin group encounter during a survey termed a sighting. Surveys were undertakenin weather conditions with Beaufort sea states ≤3, and with two to five observers (me-dian = 4) on-board. Using the naked eye or occasionally 7 9 50 binoculars, observersscanned for dolphins out to 250 m on either side of the vessel while on transect. Thethree original transects were aimed to be completed six times within a season, whilecompleting the additional three transects three times within a season. Transects wererun in open and sheltered water habitats. The open water habitat consisted of the

3

Figure 1. The study site (Bunbury, Western Australia; 540 km2) was divided into six tran-sects and categorized into open and sheltered water habitats. The open water habitat consistedof Buffalo Beach, Buffalo Beach offshore, Back Beach, Back Beach offshore, and Busselton tran-sects. The sheltered water habitat consisted of the Inner water transect, encompassing Koom-bana Bay, Leschenault Inlet and Estuary, Inner and Outer Harbours, and the lower reaches ofthe Collie River.

4

coastal transects, while the sheltered water habitat consisted of only the Inner watertransect (Fig. 1).

Photographic-identification and Sex Determination of Study Animals

When a dolphin group was encountered the vessel departed the transect line andapproached the dolphins to a suitable sighting distance for observations (typically10–30 m). Using a Nikon D300s camera with either 300 or 400 mm lenses, animage of every dolphin dorsal fin was aimed to be photographed for the purpose ofidentification (W€ursig and W€ursig 1977). Sightings lasted a minimum of 5 min(and a maximum of 30 min) to determine group composition and obtain sufficientphoto-ID images. For each group encounter, we recorded location (GPS position),time, group composition, and group size. A group was defined as one or more dolphinswithin 100 m of any other member involved in the same or similar behavioral activ-ity (Smith 2012). Occasionally, sightings were conducted when dolphin groups wereencountered on route to or from a transect commencement or end point.The sex of dolphins was confirmed through one of three methods: genetic analyses

from biopsy samples that were collected as part of a separate research project (Danielet al., unpublished data3 ), visual confirmation of genital areas or, for adult females,repeated and consistent observations in the presence of a dependent calf. FollowingSmith (2012), individuals were assigned to one of three mutually exclusive age cate-gories: calf, juvenile, or adult, based on physical traits such as body length and behavior.Dolphin body length (as a proxy for age) was estimated in the field and reconfirmed posthoc through dolphin group images from our long-term photo-ID data set.Only good quality photo-ID images were used in subsequent analyses (Smith et al.

2013). Images of each dorsal fin were used to identify dolphins by unique nicks andnotches (W€ursig and W€ursig 1977). Secondary markings (such as tooth rake scars)were not used, as individuals may only be sighted from one side and scars fade overtime (Lockyer and Morris 1990). Each identifiable individual was given a uniquethree-letter code and added to the database and dorsal fin catalog. Dolphin resight-ings were matched to the catalog following the protocols from the Sarasota DolphinResearch Program (2006). To ensure correct identification of individuals, photo-IDof each individual was double-checked by a minimum of two researchers.

Kernel Density Estimates

The KDE analyses were limited to adult individuals observed on ≥30 occasions toensure a reasonable representation of their ranging area (Seaman et al. 1999). Onlyadult individuals of known sex that were classified as being of the adult age at thebeginning of the study were included in analyses. Juveniles were excluded from theanalysis to avoid potential ontogenetic shifts in home range characteristics (Welshet al. 2013), as newly independent dolphins exhibit a high degree of site fidelity to asubset of their natal area before expanding or shifting their range (M€oller and Behere-garay 2004, McHugh et al. 2011). Dependent calves were also excluded from analy-sis. To avoid temporal autocorrelation, only the first sighting of an individual on agiven day was used in this analysis.

3Daniel, Claire, et al., Evolution and Ecology Research Centre, University of New South Wales, Kens-ington, Australia, unpublished data.

5

Kernel density estimates for each dolphin was calculated following the protocolsby MacLeod (2014) for “estimating a home range in an environment where there arebarriers to movements.” MacLeod (2014) outlines a series of steps using ArcGIS toolswhich includes details from file preparation to implementation of the final kernel foreach individual. With this method, rather than choosing the more traditional “kerneldensity estimate” tool from the spatial analyst toolbox, the “kernel interpolation withbarriers” tool was selected (available from the Geostatistical analyst toolbox in Arc-GIS 10.1; Esri, Redlands, CA). The kernel interpolation with barriers tool uses theshortest distance between points without intersecting the barrier, allowing the con-tour to change abruptly at the edge of the barrier (Gribov and Krivoruchko 2011).All subsequent steps were calculated in the Universal Transverse Mercator (UTM)Zone 50 South projection and based on the WGS 1984 datum.The key user-defined parameters for the “kernel interpolation with barriers” tool

were the output cell size and bandwidth value. The output grid cell size was set to2009 200 m, which allowed sufficient information to be included in narrow areas ofthe study site, such as rivers and estuaries in the sheltered water habitat. The kernelfunction was set to a first order polynomial and the ridge parameter retained thedefault value of 50. The bandwidth is a smoothing value that determines the widthof the kernel, i.e., it is the search radius that determines which surrounding locationpoints will contribute to the KDE. There is currently no best method for bandwidthselection (Worton 1989, Gitzen et al. 2006). The choice of a bandwidth selectionmethod may vary depending on the study goals, sample size and patterns of space useby the study species (Gitzen et al. 2006). For the “kernel interpolation with barriers”tool the value can be chosen by visual inspection (Wand and Jones 1995). In thisstudy, the bandwidth value was chosen by running successive trials and selecting theestimate that was most in accordance with our prior knowledge about individual dol-phins’ space-use and was fixed to 6,000 for each individual to ensure comparableresults between individuals and sexes. The bandwidth value was held constant acrossthe plane for a fixed kernel, rather than changing the value at different densities foran adaptive kernel (Seaman and Powell 1996, Wood et al. 2000). Adaptive kernelstend to perform poorly, often over-estimating home range areas (Powell 2000).The KDE represents values for the estimated number of sightings per km2 that are

likely to occur within each grid cell. From these values, utilization distributionwithin the study area was defined as the minimum area in which an individual had a95% probability of being located (Worton 1995). Each 95% utilization distributionwas extracted from the KDE by calculating the threshold value that enclosed 95% ofall observations used to create the KDE (MacLeod 2014). Hereafter, the 95% utiliza-tion distribution for each individual was referred to as home range within the studyarea. Thus, the estimated home range only applied to usage within the study areaitself and was relative to the survey effort, which was equal for all individuals. Wenote that the full expanse of each individual’s true home range may not have beencaptured, as the range may be larger than the study area. However, this same limita-tion was applied to all individuals and so is unlikely to cause a bias in comparison ofestimated home range sizes between sexes.

Bayesian Mixture Model

The home range of each dolphin in the study population was likely to be depen-dent upon the home range of its close associates (Fr�ere et al. 2010). As such, giventhe inherent nonindependence of data collected from a highly social species, we used

6

a method that aims to accommodate for such nonindependence, thus avoiding statis-tical tests that assume independence.When exploring sex-specific differences in home range size, it was suspected that

individuals were not from a homogeneous population and that different communitiesof individuals might be present within the study area. Each community may havediffering ranging patterns, perhaps dependent upon the amount of time an individualspent in open vs. sheltered water habitats. We did not, however, have a measure forthis suspected dimension of open vs. sheltered water effect.One established method for dealing with heterogeneous populations is to use mix-

ture modeling where samples of individuals are probabilistically assigned to two ormore groups (“latent” groups; Melnykov and Maitra 2010). Often, mixture modelsare used to increase/add population heterogeneity and mixing weights are not directlyinterpreted after model fitting. Here, the fitted mixing-probabilities were plotted togain insights into the spatial-separation of different social groups. Thus, the modelestimated the mixing probabilities, whereas our interpretation of such values in refer-ence to spatial covariates were entirely post hoc. To investigate ecological drivers thatthe two latent groups might correspond to, the proportion of time spent by individu-als in open vs. sheltered water habitats were inspected. The proportion of time anindividual was sighted within the open water and sheltered water habitat was calcu-lated for each individual by dividing the total number of sightings of a particularindividual in open or sheltered waters by the total number of sightings for that indi-vidual.A Bayesian mixture model was therefore developed, hereafter termed mixture

model, to address two aims: (1) to estimate a sex-effect on dolphin home range whileaccounting for heteroskedasticity (unequal variance across different subpopulations)and correlated error distribution, and (2) to employ a data-driven partitioning of thedolphins into two latent groups. These ideas were already technically well-developed(Melnykov and Maitra 2010). For latent group partitioning, the partitioning wasconstrained by association data while recognizing the above-mentioned nonindepen-dence of associations between individuals. This approach was our innovation by usinga prior on the latent group probabilities according to a multivariate probit distribu-tion (MVP) with a fixed and known correlation matrix R of dimension n 9 n. R wasestimated according to the Half-Weight Index; a measure of association between twoindividuals (Ginsberg and Young 1992). We calculated the Half-Weight Index onthe complete data set using SOCPROG 2.4 (Whitehead 2009). Two individuals wereassumed to be associated if they were sighted in the same group.The mixture model was run in JAGS (Plummer 2008) using the package rjags

(Plummer 2014) through R v3.0.3 software (R Development Core Team 2011).Uninformative priors were specified for the latent group means (lz), sex-effect (bM),and variances (rz

2), while the prior on k (a vector) was informative and driven by thematrix of social affiliation R (also see JAGS code in Appendix S1). The steps in theMarkov chain Monte Carlo (MCMC) sampler were as follows:

pðlzÞ ¼ Normð0; 105Þ for z ¼ 1; 2

pðbMÞ ¼ Normð0; 105Þpðr2z Þ ¼ Unifð0; 105Þ for z ¼ 1; 2

pðkÞ ¼ MVPnðO;RÞ

7

The hierarchical distribution of the random variables is as follows:

zi ¼ Multinomialðk; 1� kiÞgi ¼ lz¼1Iðzi ¼ 1Þ þ l2¼2Iðzi ¼ 2Þ þ xibM

r2i ¼ r2z¼1Iðzi ¼ 1Þ þ r2z¼2Iðzi ¼ 2Þyi ¼ Normðgi; r2i Þ

Where k is a vector of correlated probabilities of being in Latent Group 1 drawn fromthe MVP, zi is an index of the Latent Group (1 or 2), Ι() is the step function, x is avector of sexes (0 for female, 1 for male), and yi are the observed home ranges.Two versions of the mixture model were run, with the difference being whether

males and females had different variances in their home range distributions. Model#1 had a single variance for both male and females, for a total of two variance parame-ters. Model #2 had a total of four variance parameters for all male/female and latentgroup combinations. Both models were set in the common Bayesian framework toallow heterogeneity in an outcome with different group means and variances. Bothmodels partitioned groups based on network analyses, which was driven by both dol-phin associations and home ranges (Handcock et al. 2007).The most parsimonious model was selected based on a special variant of the

deviance information criterion (DIC), developed specifically for high-dimensionalmixture models (Plummer 2008). DIC is an estimate of expected predictive error,where lower deviance is better; it is a Bayesian form of other information-theoreticmodel-selection criteria (Plummer 2008). Additionally, to confirm adequate modelfit a goodness-of-fit P-value was computed through a posterior predictive check basedon sum of squared Pearson residuals (Gelman 2003). Values close to 0.5 suggest noevidence of poor model fit, whereas P-values close to 0 or 1 suggests a poor model fit(Martin et al. 2011).

Results

Survey Effort and Individual Sighting Frequencies

From March 2007 to August 2013, 586 surveys were conducted (Table 1), includ-ing 177 surveys in sheltered waters and 409 surveys in open waters (Table 2). A total

Table 1. Summary of annual survey effort from March 2007 to August 2013.

Year # of surveys # of days # of months # of sightings Mean group size (SD)

2007 51 48 10 219 4.75 (4.29)2008 94 90 12 298 5.21 (4.69)2009 93 90 12 307 6.47 (5.92)2010 96 73 12 216 6.72 (5.87)2011 91 75 12 238 6.31 (6.92)2012 113 89 12 266 5.76 (5.51)2013 48 37 8 106 7.16 (7.75)Total 586 502 78 1,650 5.98 (0.75)

8

of 1,650 dolphin groups were sighted and photographed. Group sizes ranged fromone to 60 dolphins (�x� SD; 5.98 � 0.75). A total of 463 individual dolphins wereidentified: 263 adults, 80 juveniles, and 120 calves. Of these, sex was confirmed for215 individuals (83 males and 132 females). A total of 56 adult individuals of knownsex (22 males and 34 females) were observed on ≥30 occasions and thus included inhome range analyses. The sighting frequency of adult males ranged from 30 to 68(46.7 � 12.35 SD) occasions, while the female sighting frequency ranged from 30 to115 (62.7� 25.2 SD) occasions.

Utilization Distributions of Adult Male and Female Dolphins

Home ranges varied in size, shape, and location for the 56 individuals (see Fig. 2for examples). Male 95% utilization distribution ranged from 27 to 187 km2 (94.8� 48.15 SD, median = 75.8) and female 95% utilization distribution ranged from20 to 133 km2 (65.6� 30.9 SD, median = 71.6).

Mixture Modeling

Model selection and goodness-of-fit—The MCMC algorithm was used to estimate theposterior distributions of model parameters for Model #1 and Model #2. Both modelshad an initial discarded “burn-in” phase of one million iterations, i.e., discarded val-ues from the Markov chain before convergence was reached (McCarthy 2007), andwere run for 800,000 iterations on three MCMC. For inference about the posterior-distributions, results were thinned to 6,000 total draws (2,000 per chain = 6,000total). Based on DIC selection, the DDIC = –11.64 (SE = 16.50) suggested that therewas more support for Model #1 relative to Model #2.Model #1 had a single variance for both male and females, for a total of two vari-

ance parameters. Chains were visually inspected for convergence and adequate mix-ing, as well as ensuring all univariate Gelman-Rubin scale reduction factors wereclose to 1 (Gelman and Rubin 1992). Results indicated convergence with scale reduc-tion factors ranging from 0.10 to 1.0025. The model had an average explained vari-ance, R2, of 0.43. The Bayesian P-value from the posterior predictive check statisticwas 0.607, suggesting no evidence for lack of fit.

Table 2. The number of surveys for the sheltered water habitat (Inner water transect) andopen water habitat (Back Beach and Buffalo Beach transects) from March 2007 to April 2013.The number of surveys for the additional transects in the open water habitat (Back Beach off-shore, Buffalo Beach offshore and Busselton transects) from August 2011 to August 2013.

YearInnerwaters

BackBeach

BuffaloBeach

Back Beachoffshore

Buffalo Beachoffshore Busselton Total

2007 22 16 13 NA NA NA 512008 34 28 32 NA NA NA 942009 30 27 36 NA NA NA 932010 30 30 36 NA NA NA 962011 25 27 24 5 4 6 912012 23 23 25 15 13 14 1132013 13 8 9 9 7 2 48Total 177 159 175 29 24 22 586

9

Figure 2. Examples of home ranges within the study area (95% utilization distribution;light grey polygons) from KDE including land barriers (dark gray) for two adult male (A, B)and two adult female dolphins (C, D).

10

Sex effect and latent group partitioning—The mixture model was used to estimate asex effect in home range size while probabilistically assigning individual dolphinsinto two latent groups. Probability values close to 0 suggested a high probability ofbeing in Latent Group 1, while values close to 1 suggested high probability of beingin Latent Group 2 (Fig. 3). After inspecting the mixing weights, males had moreextreme values closer to 0 or 1, suggesting a strong separation. For females, someindividuals had values ~0.5, i.e., they were in between one latent group or the other,while 30 out of 34 females appeared to have mixing weights that favored one latentgroup over the other (Fig. 3).For the sex effect (bM) there was a median difference of 1.7 (95% CI: 1.4–2.08),

i.e., the difference between male and female median home range sizes was a factor of1.7. The posterior distribution of females and males in Latent Group 1 had a medianof 35 km2 (95% CI: 28–44.5) and 59.5 km2 (95% CI: 39.2–92.56), respectively.The posterior distribution of females and males in Latent Group 2 had a median of86 km2 (95% CI: 75–95) and 146.2 km2 (95% CI: 105–197.6), respectively. Addi-tionally, the probability that the sex effect was >0 was 0.99, i.e., adult males had a99% probability of having a larger 95% utilization distribution than adult females.The relationship between latent group probabilities, home range size and the pro-

portion of time individuals spent in open vs. sheltered waters was explored (Fig. 4).Dolphins that were sighted more frequently in the open water habitat had larger

Figure 3. The probability of individual dolphins being in Latent Group 2 based on 95%utilization distribution while also accounting for their association patterns. Values close to 1suggested high probability of being in Latent Group 2. Inversely, values close to 0 suggested ahigh probability of being in Latent Group 1. Closed circles represent adult female dolphins(n = 34) and open circles represent adult male dolphins (n = 22).

11

home ranges compared to dolphins that were sighted more often in the shelteredwater habitat. This analysis also highlighted that the dolphins with the smallesthome ranges (<40 km2) were adult females (with one exception).

Discussion

We examined sex-specific differences in home range size of adult T. aduncus resid-ing in open and sheltered waters off Bunbury, Western Australia. A new GIS-basedapproach for kernel density estimation produced detailed representations of utiliza-tion distributions by accounting for physical barriers to movements throughout thestudy area. To test for sex-specific differences in home range size, we developed aBayesian mixture model that documented a 99% probability that adult male homeranges were larger than adult females. From model results, we inferred that a majorsource of home range heterogeneity was due to a division of the community into atleast two “latent groups”; these latent groups seemed to strongly correspond to thehabitat in which individuals were most often sighted (open vs. sheltered waters). Indi-viduals with larger home ranges were more frequently sighted along the open waterhabitat. In contrast, individuals with the smallest home ranges were most often

Figure 4. The relationship between latent group probabilities and 95% utilization distribu-tion home range size, and the proportion of time an individual was sighted within the openwater habitat. Latent group probabilities are represented by the size of the circle. Larger circlessuggest membership in Latent Group 2 and, inversely, smaller circles suggest membership inLatent Group 1 (Fig. 3). Closed circles represent adult female dolphins (n = 34) and open cir-cles represent adult male dolphins (n = 22).

12

sighted in the sheltered water habitat (bay, estuary, and riverine waters), which repre-sents the area of highest human usage.

Incorporating Complex Barriers into Analyses of Home Ranges

Previous studies aimed at quantifying cetacean home ranges that have accountedfor unavailable areas (such as land barriers) have done so by clipping land out prior tofinal home range calculations (e.g., McHugh et al. 2011) or removing land post hoc(e.g., Urian et al. 2009). However, density estimation is defined by the distribution ofthe locations regardless of clipping (Knight et al. 2009), and therefore, the kernelwill be smoothed across the barrier. A kernel density approach to detect and correctfor barriers was developed using the R interface (Benhamou and Cornelis 2010).However, to date, the method cannot be applied to study areas with complex barriers,such as a detailed coastline (Calenge 2006).The method presented here included complex barriers throughout the interpola-

tion of kernel density estimation. By allowing the kernel to change abruptly at theedge of the barrier (Gribov and Krivoruchko 2011) this method refined previous ana-lytical techniques, ultimately improving the accuracy of home range size estimatesand structure. This method was easily applied through a user-friendly GIS interfaceand should make it possible to accurately estimate individual home ranges of any spe-cies or population which occur in areas with barriers to movement.

Adult Dolphin Home Range Size Was Sex-specific

Most coastal bottlenose dolphin populations live in fission-fusion societies wheresex-specific bonds exist (Wells et al. 1987; Connor et al. 2000, 2001). Adult femalesform associations with other females with similar reproductive states and overlappinghome ranges (Wells et al. 1987, Smolker et al. 1992, M€oller and Harcourt 2008,Fr�ere et al. 2010). Adult males form alliances as a strategy to cooperatively gainaccess to adult females in order to optimize mating opportunities (Connor et al.1992). Home ranges of alliance members overlap with other alliances and the femaleswith which they consort (Randic et al. 2012). Alliance formation in male bottlenosedolphins supports a polygynous mating system. Typical for polygynous mating sys-tems, males generally have larger home ranges than females to maximize matingopportunities with multiple females (Greenwood 1980, Gaulin and Fitzgerald 1989).In Bunbury, adult male-male dolphin relationships are stronger and consistentlymore stable than adult female-female social relationships (Smith 2012). While notspecifically tested for, the strength and stability of male-male social relationships inthis population may reflect the presence of male alliances as have been identified inother bottlenose dolphin populations (e.g., Wells et al. 1987, Connor et al. 1992,M€oller et al. 2001, Parsons et al. 2003). As such, it is likely that male mating strate-gies strongly influence the sex-specific differences in home range size that we havefound in this population.Male home ranges contract or expand in response to breeding and/or birthing sea-

sons (Greenwood 1980, Clutton-Brock 1989). Bottlenose dolphins exhibit diffusebreeding seasons (Perrin and Reilly 1984), usually peaking in summer months. InBunbury, the peak is late summer/early autumn (Smith 2012), during which timethere is an increase in dolphin abundance within the study area (Smith et al. 2013;Sprogis et al., unpublished data). We suggest that during the breeding season, adultmale home ranges are mainly driven by the distribution of reproductive females. In

13

contrast, we suggest that during the nonbreeding season, males adjust their homeranges accordingly to optimize prey intake resulting in larger home ranges. Unfortu-nately, seasonal home range size could not be tested due to low sample sizes.

Home Range Size Was Dependent on Time Spent in Open vs. Sheltered Water Habitats

Based on mixture model results, we suggest that dolphin home range sizes cor-responded to the habitat in which they were most often sighted (open vs. shelteredwaters). Dolphins most frequently sighted within the open water habitat exhibitedlarger home ranges than dolphins most frequently sighted in the sheltered waterhabitat (bay, estuary, and riverine waters). Habitat partitioning between open andsheltered waters occur in T. truncatus populations, for example off North Carolina(Gannon and Waples 2004) and Florida (Fazioli et al. 2006). Movement betweenhabitats also occurs within populations of T. aduncus, for example, some individu-als move between estuaries and adjacent coastal habitats (e.g., Chabanne et al.2012).The interaction between prey availability and predation risk may be important

factors contributing to dolphin home range size differences among open and shel-tered waters (Heithaus and Dill 2002). Prey availability and distribution contributeto shaping dolphin distributions (e.g., Allen et al. 2001, Hastie et al. 2004, Elwenet al. 2010, Degrati et al. 2012). As per optimal foraging theory, dolphins shoulddistribute themselves for optimal foraging efficiency in order to maximize netenergy gain (MacArthur 1966, Schoener 1971). To assist in maximum energygained, the selection of high quality prey rather than quantity is suggested as amajor determinant of foraging strategies employed (Spitz et al. 2012). If animalshave access to concentrated high quality prey, they do not have to search far forfood and can have small home ranges (Harestad and Bunnell 1979). Estuaries, suchas the Leschenault Estuary in the sheltered water habitat of our study, are highlyproductive systems (Semeniuk et al. 2000, Elliott and Whitfield 2011). Estuariesprovide a rich source of organic matter and nutrients (Elliott and Whitfield 2011),which sustain numerous fish species at different stages of their life cycle (Potteret al. 2013). Marine and estuarine fish inhabit the Leschenault Estuary and are inhigh abundance, biomass, and quality compared to open waters (Potter et al. 2000;Veale et al. 2014; McCluskey et al., unpublished data4). The concentrated highquality prey support dolphins and optimizes foraging, likely allowing for smallhome ranges in these habitats.In contrast, if animals live in habitats with patchy prey distribution they should

have correspondingly larger home ranges as they must travel further in order to findadequate food (Wiens 1976, Ford 1983, Fauchald 1999). The large home ranges ofT. truncatus along the open coast of the Gulf of California and the Southern CaliforniaBight have been linked to the patchy and ephemeral distribution of prey resources(Ballance 1992, Defran et al. 1999, respectively). Similarly, T. truncatus in openwaters off the Azores undertake long-distance movements to cover the lower densityand patchy distribution of prey, and hence have considerably large home ranges (Silvaet al. 2008). Prey in open waters is patchily distributed and largely dictated by physi-cal oceanic processes (Caputi et al. 1996, Silva et al. 2008). Offshore from our studyarea the dominant oceanographic feature is the Leeuwin Current, which transports

4McCluskey, Shannon, et al., Murdoch University Cetacean Research Unit, Murdoch, Australia,unpublished data.

14

warm, oligotrophic waters pole-wards along the shelf break (Godfrey and Ridgway1985, Pearce and Griffiths 1991). The interannual variations in the strength of theLeeuwin Current and associated countercurrents affect the distribution and recruit-ment of fish species (Lenanton et al. 1991, Caputi et al. 1996, Pearce and Pattiaratchi1999). We suggest that the larger home ranges of the dolphins in open waters areinfluenced by the variability of prey distribution in this highly dynamic environ-ment.Dolphins will not necessarily select habitats based solely on the energetic return of

their prey, particularly if predation risk varies among habitats (Heithaus and Dill2002). The home range size of dolphins within open and sheltered water habitatsmay be influenced by a trade-off between minimizing predation risk and the benefitsgained from foraging (Lima and Dill 1990). As demonstrated by Heithaus and Dill(2002) in Shark Bay, T. aduncus feed on high biomass prey in shallow waters, how-ever, during warmer months when tiger shark (Galeocerdo cuvier) density in shallowwaters increases, dolphin habitat use deviates from that of the productive shallows. Incontrast, in Sarasota Bay, T. truncatus are more prevalent in shallower waters duringwarmer months to decrease predation risk from bull sharks (Carcharhinus leucas) pre-sent in deeper passes (Wells et al. 1980). In our study, evidence of shark predationattempts ondolphins is clear from bite wounds and scars on dolphins (KRS, personalobservations). Evaluation of bite marks suggest that the species responsible are likelywhite (Carcharodon carcharias), tiger, and several smaller Carcharhinid species (King2014). Detailed studies of the interaction between predation risk and prey availabilitywill be required in our study area to understand what role they play on shaping dol-phin home range differences between habitats.

Conservation Implications for Dolphins in Sheltered Waters

Dolphins with the smallest home ranges were predominately female and fre-quently sighted within the sheltered water habitat. This area (Koombana Bay,Leschenault Inlet and Estuary, Inner and Outer Harbours) consists of waterways ofhigh human usage. Specifically, these waters include one of Western Australia’sbusiest commercial shipping ports, are popular for recreational water activities, andare the focus of a viable dolphin-targeted tourism industry (Arcangeli and Crosti2009, Jensen et al. 2009). The dolphins in the sheltered water habitat are themajor tourism icon for the region (dolphin-watch and swim-with industry),attracting around 60,000 visitors per year and generating over AUS$5.6 millioninto the local economy (Bunbury Dolphin Discovery Inc. 2008). As such, the innerwater dolphins are disproportionally exposed to human disturbance from shipping,recreational and tourism boating activities, and coastal development. These pres-sures can result in cumulative threats, in particular from vessel disturbance (Bejderet al. 2006, Christiansen et al. 2010), vessel strikes (Wells and Scott 1997), entan-glement in fishing gear (Wells et al. 2008), exposure to contaminants(Balmer et al.2011), and illegal food-provisioning (Donaldson et al. 2010, 2012). Further, maleand female dolphins may differ in their susceptibility to threats (Lusseau 2003,Symons et al. 2014) and be impacted disproportionately (Crespo et al. 1997, Bairdet al. 2015). As such, females with small home ranges residing in the inner watersmay be more likely to encounter threats, which could lead to population conse-quences, such as reduced reproductive output. Overall, we recommend that man-agement efforts focus on minimizing human impacts to dolphins frequenting thesewaters.

15

Acknowledgments

We thank numerous research assistants who assisted with field research and data processing.Thank you to M. Cannon, D. Chabanne, and K. Nicholson for coordinating previous field sea-sons. We are grateful to S. Allen, D. McElligott, and A. Sellas who obtained biopsies forgenetic analyses; C. Daniel, C. Fr�ere, A. Kopps, and O. Manlik for sex results from geneticanalyses. We thank R. Wells for comments on the manuscript. We are appreciative to thefunding partners from the South West Marine Research Program; Bemax Cable Sands, BHPBilliton Worsley Alumina Ltd., the Bunbury Dolphin Discovery Centre, Bunbury PortAuthority, City of Bunbury, Cristal Global, the Western Australian Department of Parks andWildlife, Iluka, Millard Marine, Naturaliste Charters, Newmont Boddington Gold, SouthWest Development Commission, andWA Plantation Resources. KRS was supported through-out her Ph.D. by an Australian Postgraduate Award, a Murdoch University Research Excel-lence Scholarship, and a SWMRP Scholarship. HCS was supported during her Ph.D. by aMurdoch University Scholarship. This study was carried out with approval from the MurdochUniversity Animal Ethics committee (W2009/06, W2342/10) and was licensed by theDepartment of Environment and Conservation (SF005811, SF008624). We are grateful to theanonymous reviewers and Guido Parra, whose comments were much appreciated and improvedthe manuscript greatly.

Literature Cited

Allen, M. C., A. J. Read, J. Gaudet and L. S. Sayigh. 2001. Fine-scale habitat selection offoraging bottlenose dolphins Tursiops truncatus near Clearwater, Florida. Marine EcologyProgress Series 222:253–264.

Arcangeli, A., and R. Crosti. 2009. The short-term impact of dolphin-watching on thebehavior of bottlenose dolphins (Tursiops truncatus) in Western Australia. Journal ofMarine Animals and their Ecology 2:3–9.

Austin, D., W. D. Bowen and J. I. McMillan. 2004. Intraspecific variation in movementpatterns: Modeling individual behaviour in a large marine predator. Oikos 105:15–30.

Baird, R. W., S. D. Mahaffy, A. M. Gorgone, et al. 2015. False killer whales and fisheriesinteractions in Hawaiian waters: Evidence for sex bias and variation among populationsand social groups. Marine Mammal Science 31:579–590.

Ballance, L. T. 1992. Habitat use patterns and ranges of the bottlenose dolphin in the Gulf ofCalifornia, Mexico. Marine Mammal Science 8:262–274.

Balmer, B. C., L. H. Schwacke, R. S. Wells, et al. 2011. Relationship between persistentorganic pollutants (POPs) and ranging patterns in common bottlenose dolphins (Tursiopstruncatus) from coastal Georgia, USA. Science of the Total Environment 409:2094–2101.

Bejder, L., A. Samuels, H. Whitehead, et al. 2006. Decline in relative abundance of bottlenosedolphins exposed to long-term disturbance. Conservation Biology 20:1791–1798.

Benhamou, S., and D. Cornelis. 2010. Incorporating movement behavior and barriers toimprove kernel home range space use estimates. Journal of Wildlife Management74:1353–1360.

Bunbury Dolphin Discovery Inc. 2008. Investment case. Bunbury Dolphin Discovery Centre,Bunbury, Australia. 21 pp.

Bunbury Port Authority. 2013. Annual report. Bunbury Port Authority, Bunbury, Australia77 pp.

Burt, W. H. 1943. Territoriality and home range concepts as applied to mammals. Journal ofMammalogy 24:346–352.

Calenge, C. 2006. The package “adehabitat” for the R software: A tool for the analysis of spaceand habitat use by animals. Ecological Modelling 197:516–519.

16

Caputi, N., W. J. Fletcher, A. Pearce and C. F. Chubb. 1996. Effect of the Leeuwin Currenton the recruitment of fish and invertebrates along the Western Australian coast. Marineand Freshwater Research 47:147–155.

Chabanne, D., H. Finn, C. Salgado-Kent and L. Bejder. 2012. Identification of a residentcommunity of bottlenose dolphins (Tursiops spp.) in the Swan-Canning Estuary, WesternAustralia, using behavioural information. Pacific Conservation Biology 18:247–262.

Chapman, C. A., and R. W. Wrangham. 1993. Range use of the forest chimpanzees of Kibale:Implications for the understanding of the chimpanzee social-organisation. AmericanJournal of Primatology 31:263–273.

Christiansen, F., D. Lusseau, E. Stensland and P. Berggren. 2010. Effects of tourist boats onthe behaviour of Indo-Pacific bottlenose dolphins off the south coast of Zanzibar.Endangered Species Research 11:91–99.

Clutton-Brock, T. H. 1989. Mammalian mating systems. Proceedings of the Royal SocietySeries B-Biological Sciences 236:339–372.

Connor, R. C., R. A. Smolker and A. F. Richards. 1992. Two levels of alliance formationamong male bottlenose dolphins (Tursiops sp.). Proceedings of the National Academy ofSciences of the United States of America 89:987–990.

Connor, R. C., R. S. Wells, J. Mann and A. J. Read. 2000. The bottlenose dolphin: Socialrelationships in a fission-fusion society. Pages 91–126 in J. Mann, R. C. Connor, P. L.Tyack and H. Whitehead, eds. Cetacean societies: Field studies of dolphins and whales.The University of Chicago Press, Chicago, IL.

Connor, R. C., M. R. Heithaus and L. M. Barre. 2001. Complex social structure, alliancestability and mating access in a bottlenose dolphin ‘super-alliance’. Proceedings of theRoyal Society of London Series B-Biological Sciences 268:263–267.

Crespo, E. A., S. N. Pedraza and S. L. Dans, et al. 1997. Direct and indirect effects of thehighseas fisheries on the marine mammal populations in the northern and centralPatagonian coast. Journal of Northwest Atlantic Fishery Science:189–207.

Defran, R. H., and D. W. Weller. 1999. Occurrence, distribution, site fidelity, and school sizeof bottlenose dolphins (Tursiops truncatus) off San Diego, California. Marine MammalScience 15:366–380.

Defran, R. H., D. W. Weller, D. L. Kelly and M. A. Espinosa. 1999. Range characteristics ofPacific coast bottlenose dolphins (Tursiops truncatus) in the Southern California Bight.Marine Mammal Science 15:381–393.

Degrati, M., S. L. Dans, G. V. Garaffo and E. A. Crespo. 2012. Diving for food: A switch offoraging strategy of dusky dolphins in Argentina. Journal of Ethology 30:361–367.

Don, W. H. 1949. Calculation of size of home range. Journal of Mammalogy 30:1–18.Donaldson, R., H. Finn and M. Calver. 2010. Illegal feeding increases risk of boat-strike and

entanglement in bottlenose dolphins in Perth, Western Australia. Pacific ConservationBiology 16:157–161.

Donaldson, R., H. Finn, L. Bejder, D. Lusseau and M. Calver. 2012. The social side of human-wildlife interaction: Wildlife can learn harmful behaviours from each other. AnimalConservation 15:427–435.

Elliott, M., and A. K. Whitfield. 2011. Challenging paradigms in estuarine ecology andmanagement. Estuarine Coastal and Shelf Science 94:306–314.

Elwen, S. H., M. Thornton, D. Reeb and P. B. Best. 2010. Near-shore distribution ofHeaviside’s (Cephalorhynchus heavisidii) and dusky dolphins (Lagenorhynchus obscurus) atthe southern limit of their range in South Africa. African Zoology 45:78–91.

Fauchald, P. 1999. Foraging in a hierarchical patch system. American Naturalist 153:603–613.

Fazioli, K. L., S. Hofmann and R. S. Wells. 2006. Use of gulf of Mexico coastal waters bydistinct assemblages of bottlenose dolphins (Tursiops truncatus). Aquatic Mammals32:212–222.

Ford, R. G. 1983. Home range in a patchy environment: Optimal foraging predictions.American Zoologist 23:315–326.

17

Fr�ere, C. H., M. Kr€utzen, J. Mann, et al. 2010. Home range overlap, matrilineal andbiparental kinship drive female associations in bottlenose dolphins. Animal Behaviour80:481–486.

Fury, C. A., K. E. Ruckstuhl and P. L. Harrison. 2013. Spatial and social sexual segregationpatterns in Indo-Pacific bottlenose dolphins (Tursiops aduncus). PLOS ONE 8:e52987.

Gannon, D. P., and D. M. Waples. 2004. Diets of coastal bottlenose dolphins from the U.S.mid-Atlantic coast differ by habitat. Marine Mammal Science 20:527–545.

Gaulin, S. J. C., and R. W. Fitzgerald. 1989. Sexual selection for spatial-learning ability.Animal Behaviour 37:322–331.

Gelman, A. 2003. A Bayesian formulation of exploratory data analysis and goodness-of-fittesting. International Statistical Review 71:369–382.

Gelman, A. R., and D. B. Rubin. 1992. Inference from iterative simulation using multiplesequences. Statistical Science 7:457–511.

Getz, W. M., and C. C. Wilmers. 2004. A local nearest-neighbor convex-hull construction ofhome ranges and utilization distributions. Ecography 27:489–505.

Ginsberg, J. R., and T. P. Young. 1992. Measuring association between individuals or groupsin behavioral-studies. Animal Behaviour 44:377–379.

Gitzen, R. A., J. J. Millspaugh and B. J. Kernohan. 2006. Bandwidth selection for fixed-kernel analysis of animal utilization distributions. Journal of Wildlife Management70:1334–1344.

Godfrey, J. S., and K. R. Ridgway. 1985. The large-scale environment of the poleward-flowing Leeuwin Current, Western Australia: Longshore steric height gradients, windstresses and geostrophic flow. Journal of Physical Oceanography 15:481–495.

Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals.Animal Behaviour 28:1140–1162.

Gribov, A., and K. Krivoruchko. 2011. Local polynomials for data detrending andinterpolation in the presence of barriers. Stochastic Environmental Research and RiskAssessment 25:1057–1063.

Gubbins, C. 2002. Use of home ranges by resident bottlenose dolphins (Tursiops truncatus) in aSouth Carolina estuary. Journal of Mammalogy 83:178–187.

Handcock, M. S., A. E. Raftery and J. M. Tantrum. 2007. Model-based clustering for socialnetworks. Journal of the Royal Statistical Society Series A-Statistics in Society 170:301–322.

Harestad, A. S., and F. L. Bunnell. 1979. Home range and body-weight: Re-evaluation.Ecology 60:389–402.

Hastie, G. D., B. Wilson, L. J. Wilson, K. M. Parsons and P. M. Thompson. 2004.Functional mechanisms underlying cetacean distribution patterns: Hotspots forbottlenose dolphins are linked to foraging. Marine Biology 144:397–403.

Heithaus, M. R., and L. M. Dill. 2002. Food availability and tiger shark predation riskinfluence bottlenose dolphin habitat use. Ecology 83:480–491.

Jensen, F. H., L. Bejder, M. Wahlberg, N. A. Soto, M. Johnson and P. T. Madsen. 2009.Vessel noise effects on delphinid communication. Marine Ecology Progress Series395:161–175.

Kie, J. G., J. Matthiopoulos, J. Fieberg, et al. 2010. The home-range concept: Are traditionalestimators still relevant with modern telemetry technology? Philosophical Transactionsof the Royal Society B-Biological Sciences 365:2221–2231.

King, C. 2014. Investigating the movement patterns of sharks and the significance ofpotential shark predation attempts on bottlenose dolphins (Tursiops aduncus) in thewaters of south-western Australia. Honours thesis, Murdoch University, Perth,Australia. 84 pp.

Knight, C. M., R. E. Kenward, R. E. Gozlan, K. H. Hodder, S. S. Walls and M. C. Lucas.2009. Home-range estimation within complex restricted environments: Importance ofmethod selection in detecting seasonal change. Wildlife Research 36:213–224.

18

Knip, D. M., M. R. Heupel and C. A. Simpfendorfer. 2012. Habitat use and spatialsegregation of adult spottail sharks Carcharhinus sorrah in tropical nearshore waters.Journal of Fish Biology 80:767–784.

Lenanton, R. C., L. Joll, J. Penn and K. Jones. 1991. The influence of the Leeuwin Current oncoastal fisheries of Western Australia. Journal of the Royal Society of Western Australia74:101–114.

Leung, E. S., B. L. Chilvers, S. Nakagawa, A. B. Moore and B. C. Robertson. 2012. Sexualsegregation in juvenile New Zealand sea lion foraging ranges: Implications forintraspecific competition, population dynamics and conservation. PLOS ONE 7:e45389.

Lima, S. L., and L. M. Dill. 1990. Behavioral decisions made under the risk of predation: Areview and prospectus. Canadian Journal of Zoology 68:619–640.

Lockyer, C. H., and R. J. Morris. 1990. Some observations on wound healing and persistenceof scars in Tursiops truncatus. Report of the International Whaling Commission (SpecialIssue 12):113–118.

Lusseau, D. 2003. Male and female bottlenose dolphins Tursiops spp. have different strategiesto avoid interactions with tour boats in Doubtful Sound. New Zealand. Marine EcologyProgress Series 257:267–274.

Lynn, S. K., and B. W€ursig. 2002. Summer movement patterns of bottlenose dolphins in aTexas Bay. Gulf of Mexico Science 20:25–37.

MacArthur, R. H., and E. R. Pianka. 1966. On optimal use of a patchy environment.American Naturalist 100:603–609.

Mace, R. D., and J. S. Waller. 1997. Spatial and temporal interaction of male and femalegrizzly bears in northwestern Montana. Journal of Wildlife Management 61:39–52.

MacLeod, C. D. 2014. An introduction to using GIS in marine biology. Supplementaryworkbook four. Investigating home ranges of individual animals. Pictish BeastPublications, Glasgow, U.K.

Martin, J., J. A. Royle, D. I. Mackenzie, H. H. Edwards, M. Kery and B. Gardner. 2011.Accounting for non-independent detection when estimating abundance of organismswith a Bayesian approach. Methods in Ecology and Evolution 2:595–601.

Martinez-Serrano, I., A. Serrano, G. Heckel and Y. Schramm. 2011. Distribution and homerange of bottlenose dolphins (Tursiops truncatus) off Veracruz, Mexico. Ciencias Marinas37:379–392.

Matthiopoulos, J., and G. Aarts. 2010. The spatial analysis of marine mammal abundance.Pages 68–97 in I. L. Boyd, W. D. Bowen and S. Iverson, eds. Marine mammal ecologyand conservation: A handbook of techniques. Oxford University Press, Oxford, U.K.

Maxwell, S. M., G. A. Breed, B. A. Nickel, et al. 2011. Using satellite tracking to optimizeprotection of long-lived marine species: Olive ridley sea turtle conservation in CentralAfrica. PLOS ONE 6:e19905.

McCarthy, M. A. 2007. Bayesian methods for ecology. Cambridge University Press,Cambridge, U.K..

McHugh, K. A., J. B. Allen, A. A. Barleycorn and R. S. Wells. 2011. Natal philopatry,ranging behavior, and habitat selection of juvenile bottlenose dolphins in Sarasota Bay,Florida. Journal of Mammalogy 92:1298–1313.

Melnykov, V., and R. Maitra. 2010. Finite mixture models and model-based clustering.Statistics Surveys 4:80–116.

Mohr, C. O. 1947. Table of equivalent populations of North American small mammals.American Midland Naturalist 37:223–249.

M€oller, L. M., and L. B. Beheregaray. 2004. Genetic evidence for sex-biased dispersal inresident bottlenose dolphins (Tursiops aduncus). Molecular Ecology 13:1607–1612.

M€oller, L. M., and R. G. Harcourt. 2008. Shared reproductive state enhances femaleassociations in dolphins. Research Letters in Ecology 2008:5.

M€oller, L. M., L. B. Beheregaray, R. G. Harcourt and M. Kr€utzen. 2001. Alliancemembership and kinship in wild male bottlenose dolphins (Tursiops aduncus) of

19

southeastern Australia. Proceedings of the Royal Society of London Series B-BiologicalSciences 268:1941–1947.

Owen, E. C. G., R. S. Wells and S. Hofmann. 2002. Ranging and association patterns ofpaired and unpaired adult male Atlantic bottlenose dolphins, Tursiops truncatus, inSarasota, Florida, provide no evidence for alternative male strategies. Canadian Journal ofZoology 80:2072–2089.

Parsons, K. M., J. W. Durban, D. E. Claridge, K. C. Balcomb, L. R. Noble and P. M.Thompson. 2003. Kinship as a basis for alliance formation between male bottlenosedolphins, Tursiops truncatus, in the Bahamas. Animal Behaviour 66:185–194.

Pearce, A. F., and R. W. Griffiths. 1991. The mesoscale structure of the Leeuwin Current: Acomparison of laboratory models and satellite imagery. Journal of Geophysical Research-Oceans 96:16739–16757.

Pearce, A., and C. Pattiaratchi. 1999. The Capes Current: A summer countercurrent flowingpast Cape Leeuwin and Cape Naturaliste, Western Australia. Continental Shelf Research19:401–420.

Perrin, W. F., and S. B. Reilly. 1984. Reproductive parameters of dolphins and small whalesof the family Delphinidae. Report of the International Whaling Commission (SpecialIssue 6):181–185.

Plummer, M. 2008. Penalized loss functions for Bayesian model comparison. Biostatistics9:523–539.

Plummer, M. 2014. rjags: Bayesian graphical models using MCMC, in R package version 3-13. Available at http://cran.r-project.org/package=rjags.

Potter, I., P. Chalmer, D. Tiivel, R. Steckis, M. Platell and R. Lenanton. 2000. The fish faunaand finfish fishery of the Leschenault peninsula estuary in south-western Australia.Journal of the Royal Society of Western Australia 83:481–501.

Potter, I. C., J. R. Tweedley, M. Elliott and A. K. Whitfield. 2013. The ways in which fishuse estuaries: A refinement and expansion of the guild approach. Fish and Fisheries16:230–239.

Powell, R. A. 2000. Animal home ranges and territories and home range estimators. Pages65–110 in L. Boitani and T. K. Fuller, eds. Research techniques in animal ecology:Controversies and consequences. Columbia University Press, New York, NY.

R Development Core Team. 2011. R: A language and environment for statistical computing.R Foundation for Statistical Computing, Vienna, Austria.

Randic, S., R. C. Connor, W. B. Sherwin and M. Kr€utzen. 2012. A novel mammalian socialstructure in Indo-Pacific bottlenose dolphins (Tursiops sp.): Complex male alliances in anopen social network. Proceedings of the Royal Society B-Biological Sciences 279:3083–3090.

Rayment, W., S. Dawson, E. Slooten, S. Br€ager, S. D. Fresne and T. Webster. 2009. Kerneldensity estimates of alongshore home range of Hector’s dolphins at Banks Peninsula,New Zealand. Marine Mammal Science 25:537–556.

Sarasota Dolphin Research Program. 2006. Manual for field research and laboratory activities.Chicago Zoological Society and Dolphin Biology Research Institute. 62 pp.

Schoener, T. W. 1971. Theory of feeding strategies. Annual Review of Ecology andSystematics 2:369–404.

Scott, M. D., R. S. Wells and B. A. Irvine. 1990. A long term study of bottlenose dolphins onthe west coast of Florida. Pages 235–244 in S. R. Leatherwood and R. R. Reeves, eds.The bottlenose dolphin. Academic Press, San Diego, CA.

Seaman, D. E., J. J. Millspaugh, B. J. Kernohan, G. C. Brundige, K. J. Raedeke and R. A.Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal of WildlifeManagement 63:739–747.

Seaman, D. E., and R. A. Powell. 1996. An evaluation of the accuracy of kernel densityestimators for home range analysis. Ecology 77:2075–2085.

Semeniuk, V., T. A. Semeniuk and J. Unno. 2000. The Leschenault Inlet estuary: Anoverview. Journal of the Royal Society of Western Australia 83:207–228.

20

Seminoff, J. A., A. Resendiz and W. J. Nichols. 2002. Home range of green turtles Cheloniamydas at a coastal foraging area in the Gulf of California, Mexico. Marine EcologyProgress Series 242:253–265.

Silva, M. A., R. Prieto, S. Magalhaes, M. I. Seabra, R. S. Santos and P. S. Hammond. 2008.Ranging patterns of bottlenose dolphins living in oceanic waters: Implications forpopulation structure. Marine Biology 156:179–192.

Silverman, B. W. 1986. Density estimation for statistics and data analysis. Chapman & Hall,London, U.K..

Smith, H. C. 2012. Population dynamics and habitat use of bottlenose dolphins (Tursiopsaduncus), Bunbury, Western Australia. Ph.D. thesis, Murdoch University, Perth,Australia. 180 pp.

Smith, H. C., K. Pollock, K. Waples, S. Bradley and L. Bejder. 2013. Use of the robust designto estimate seasonal abundance and demographic parameters of a coastal bottlenosedolphin (Tursiops aduncus) population. PLOS ONE 8:e76574.

Smolker, R. A., A. F. Richards, R. C. Connor and J. W. Pepper. 1992. Sex-differences inpatterns of association among Indian Ocean bottlenose dolphins. Behaviour 123:38–69.

Spitz, J., A. W. Trites, V. Becquet, A. Brind’Amour, Y. Cherel, R. Galois and V. Ridoux.2012. Cost of living dictates what whales, dolphins and porpoises eat: The importance ofprey quality on predator foraging strategies. PLOS ONE 7:e50096.

Symons, J., E. Pirotta and D. Lusseau. 2014. Sex differences in risk perception in deep-divingbottlenose dolphins leads to decreased foraging efficiency when exposed to humandisturbance. Journal of Applied Ecology 51:1584–1592.

Urian, K. W., S. Hofmann, R. S. Wells and A. J. Read. 2009. Fine-scale population structureof bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida. Marine MammalScience 25:619–638.

Van Winkle, W. 1975. Comparison of several probabilistic home-range models. Journal ofWildlife Management 39:118–123.

Veale, L., J. R. Tweedley, K. R. Clarke, C. S. Hallett and I. C. Potter. 2014. Characteristics ofthe ichthyofauna of a temperate microtidal estuary with a reverse salinity gradient,including inter-decadal comparisons. Journal of Fish Biology 85:1320–1354.

Wand, M. P., and M. C. Jones. 1995. Kernel smoothing. Chapman and Hall, London, U.K.Watson-Capps, J. J. 2005. Female mating behavior in the context of sexual coercion and

female ranging behavior of bottlenose dolphins (Tursiops sp.) in Shark Bay, WesternAustralia. Ph.D. thesis, Georgetown University, Washington, DC. 195 pp.

Wearmouth, V. J., and D. W. Sims. 2008. Sexual segregation in marine fish, reptiles, birdsand mammals: Behaviour patterns, mechanisms and conservation implications. Pages107–170 in D. W. Sims, ed. Advances in marine biology. Elsevier Academic Press Inc.,San Diego, CA.

Wells, R. S., and M. D. Scott. 1997. Seasonal incidence of boat strikes on bottlenose dolphinsnear Sarasota, Florida. Marine Mammal Science 13:475–480.

Wells, R. S., A. B. Irvine and M. D. Scott. 1980. The social ecology of inshore odontocetes.John Wiley and Sons Inc., New York, NY.

Wells, R. S., M. D. Scott and A. B. Irvine. 1987. The social structure of free-rangingbottlenose dolphins. Pages 247–305 in H. H. Genoways, ed. Current mammalogy.Plenum Press, New York, NY.

Wells, R. S., J. B. Allen, S. Hofmann, et al. 2008. Consequences of injuries on survival andreproduction of common bottlenose dolphins (Tursiops truncatus) along the west coast ofFlorida. Marine Mammal Science 24:774–794.

Welsh, J. Q., C. H. R. Goatley and D. R. Bellwood. 2013. The ontogeny of home ranges:Evidence from coral reef fishes. Proceedings of the Royal Society B-Biological Sciences280:32066–32066.

Whitehead, H. 2009. SOCPROG programs: Analysing animal social structures. BehavioralEcology and Sociobiology 63:765–778.

21

Wiens, J. A. 1976. Population responses to patchy environments. Annual Review of Ecologyand Systematics 7:81–120.

Wilson, B., P. M. Thompson and P. S. Hammond. 1997. Habitat use by bottlenose dolphins:Seasonal distribution and stratified movement patterns in the Moray Firth, Scotland.Journal of Applied Ecology 34:1365–1374.

Wood, A. G., B. Naef-Daenzer, P. A. Prince and J. P. Croxall. 2000. Quantifying habitat usein satellite-tracked pelagic seabirds: Application of kernel estimation to albatrosslocations. Journal of Avian Biology 31:278–286.

Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-rangestudies. Ecology 70:164–168.

Worton, B. J. 1995. Using Monte-Carlo simulation to evaluate kernel-based home-rangeestimators. Journal of Wildlife Management 59:794–800.

W€ursig, B., and M. W€ursig. 1977. The photographic determination of group size,composition, and stability of coastal porpoises (Tursiops truncatus). Science 198:755–756.

Xavier, J. C., P. N. Trathan, J. P. Croxall, A. G. Wood, G. Podesta and P. G. Rodhouse.2004. Foraging ecology and interactions with fisheries of wandering albatrosses(Diomedea exulans) breeding at South Georgia. Fisheries Oceanography 13:324–344.

Received: 1 February 2015Accepted: 17 July 2015

Supporting Information

The following supporting information is available for this article online at http://onlinelibrary.wiley.com/doi/10.1111/mms.12260/suppinfo.Appendix S1. JAGS code for Bayesian mixture model.

22