Journal of Mammalogy, 93(4):1129–1138, 2012
Group dynamics of Yellowstone pronghorn P. J. WHITE,* CLAIRE N. GOWER, TROY L. DAVIS, JENNIFER W. SHELDON,
AND
JESSE R. WHITE
* Correspondent:
[email protected] Understanding mechanisms that influence the grouping tendencies of large herbivores is necessary to predict the influence of environmental and human factors on threatened populations. Locations of 53 adult female pronghorn (Antilocapra americana) in Yellowstone National Park during June 1999–April 2005 indicated that mean and typical group sizes and the variation in group size decreased during fawning when females secluded themselves, but became larger and more dynamic during fawn rearing and the rut and winter. Mixed-effects models indicated a strong effect of time of year on mean group sizes, with some evidence that predators negatively affected group sizes during winter. Within-animal variability (0.64) was substantially higher than between-animal variability (0.02). Pronghorn density, snow water equivalent, and predation apparently influenced variations in group size. Multiple regressions indicated effects of pronghorn density and snow water equivalent on typical group size, the size of the group in which the average animal finds itself. Overall, there was fluidity in group cohesion, with female associations changing within and among days. The behavioral plasticity of pronghorn with respect to grouping and social cohesion might confer resilience to changes in environmental conditions, but often makes it difficult to predict the consequences of conservation actions to control disease, protect or restore key habitat, regulate harvests, and limit adverse effects of development and recreation. Key words: Park
Antilocapra americana, association, group size, predation, pronghorn, social behavior, Yellowstone National
Ó 2012 American Society of Mammalogists
DOI: 10.1644/10-MAMM-A-257.1
Intraspecific variations in the grouping tendencies of wild vertebrates are common and have been systematically and adaptively related to differences in ecological or demographic circumstances, including density, dispersion of resources, environmental heterogeneity, experience, kinship, nutrition and physiology, predation pressure, and social interactions (Lott 1991). Understanding the proximate mechanisms that produce changes in group composition and size is critical for predicting the consequences of environmental changes and management strategies to control disease, protect or restore key habitat, regulate harvests, limit adverse effects of development and recreation, and maintain ecologically effective populations (Deblinger and Alldredge 1989; Lott 1991). Information on spatial and temporal changes in grouping tendencies also is important for assessing methods used to obtain unbiased estimates of population size that are based on assumptions (e.g., random marking or sampling) influenced by the grouping tendencies of animals (Eberhardt et al. 1998). Groupings of individuals in many taxa is a complex behavioral process that is driven by a combination of
environmental, population, and community processes that change with ecological conditions over relatively short spatial and temporal scales (Chapman and Chapman 2000). Dynamic group behavior is typical of social animals inhabiting a heterogeneous landscape (both temporally and spatially) where available food patches differ in size, quantity, and quality throughout the year (Berger et al. 1983; Kitchen 1974; Telfer and Kelsall 1984). In some ungulate species (e.g., deer [Odocoileus], elk [Cervus elaphus], and pronghorn [Antilocapra americana]), group sizes are typically smaller during calving, fawning, and the neonatal hiding period when females become semi-isolated. Conversely, group sizes are larger during the natal and rutting period when females and their young coalesce into groups, and largest during winter when individuals congregate on their winter range (Byers 1997;
www.mammalogy.org 1129
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
National Park Service, P.O. Box 168, Yellowstone National Park, WY 82190, USA (PJW, TLD) Montana Department of Fish, Wildlife, and Parks, 1400 South 19th Avenue, Bozeman, MT 59718, USA (CNG) Yellowstone Ecological Research Center, 2048 Analysis Drive, Bozeman, MT 59718, USA (JWS) Gardiner Public School, Gardiner, MT 59030, USA (JRW)
1130
Vol. 93, No. 4
JOURNAL OF MAMMALOGY
We conducted research to improve our understanding of various abiotic and biotic factors influencing the grouping tendencies of Yellowstone pronghorn. We predicted that animals would switch groups frequently to coincide with mating behaviors and variation associated with annual movement patterns (Fryxell et al. 2004); pronghorn density and predation risk positively affect group size and its variation (Brun 1977; Gower et al. 2009a; Pulliam 1973); group sizes would be larger in open grassland habitats with higher vegetation greenness, which strongly correlates with green biomass (Fryxell 1991; Huete et al. 2002); and group sizes would become larger as the depth and density of snow increased and pronghorn congregated on windswept areas, but less variable due to the increased energetic costs associated with moving between windswept areas (Heard 1992; Je˛drzejewski et al. 1992; Sweeny and Sweeny 1984).
MATERIALS
AND
METHODS
Yellowstone pronghorn inhabited foothills, mountain slopes, and valley bottoms along the Lamar and Yellowstone rivers in the northern portion of Yellowstone National Park, Wyoming, and adjacent areas of southwestern Montana. All pronghorn spent the winter in the Gardiner Basin of Montana near the northern boundary of the park during November through March, after which 60–80% of the animals migrated 15–50 km east to summer at higher elevations (White et al. 2007b). The climate was characterized by short, cool summers and long, cold winters with a mean annual temperature of 1.88C. Mean annual precipitation varied from 25 to 35 cm as elevation increased from 1,500 m in river drainages to 3,400 m on mountains. Average snow water equivalents (i.e., SWEs; amount of water in snow) ranged from 2 to 30 cm along this elevation gradient (Watson et al. 2009). Severe drought occurred during 1998–2004 (White et al. 2007a). Habitat used by pronghorn included open grasslandsagebrush steppe (sagebrush [Artemisia tridentata], greasewood [Sarcobatus vermiculatus], and rabbitbrush [Chrysothamnus spp.]), upland grasslands and meadows (bluebunch wheatgrass [Elymus spicatus], Idaho fescue [Festuca idahoensis], and prairie junegrass [Koeleria macrantha]), old agricultural fields (e.g., crested wheatgrass [Agropyron cristatum]), agricultural fields and pastures on private land (alfalfa [Medicago sativa]), and nonvegetated areas (Boccadori et al. 2008). Predators included black bears (Ursus americanus), coyotes (Canis latrans), grizzly bears (U. arctos), mountain lions (Puma concolor), and wolves (Canis lupus—BarnoweMeyer et al. 2009). Sympatric ungulates included elk, bighorn sheep (Ovis canadensis), bison (Bison bison), and mule deer (Odocoileus hemionus—White and Garrott 2005). Coyotes were the primary predator of adult and fawn pronghorn during all seasons, but wolves and mountain lions accounted for additional predation of adults and fawns (Barnowe-Meyer et al. 2009). Adult female pronghorn were captured from groups distributed across their winter range inside Yellowstone
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
Sweeny and Sweeny 1984). Herding behavior in ungulates reduces the time each member must spend scanning for danger, provides cooperative defense, and dilutes individual risk (Illius and Fitzgibbon 1994; Lipetz and Bekoff 1982; Pulliam 1973). However, grouping behavior should become more dynamic as the density of animals on the landscape increases (Brun 1977; Clutton-Brock et al. 1982; Hebblewhite and Pletscher 2002). Also, animals must balance competing demands between nutrition and predation by forming groups of different sizes as the level of predation and starvation risk on the landscape changes (Gower et al. 2009a). Variability in forage biomass due to heterogeneous landscapes, changes in plant phenology, and changes in availability of forage due to herbivory all result in group sizes varying predictably over seasons and among different types of plant communities. Thus, grouping behavior in ungulates varies spatially and temporally and is generally influenced by habitat type and forage production, with larger groups generally occurring in open areas with more green biomass that promotes a foraging response (Fryxell 1991; Hirth 1977; Jarman 1974). In addition, snow pack influences grouping behavior by reducing forage availability and causing ungulates to remain at available foraging sites in areas with less snow due to the increased energetic costs required to move between patches (Heard 1992; Je˛drzejewski et al. 1992; O’Gara 2004; Sweeny and Sweeny 1984). The American pronghorn is a highly gregarious species for which behavioral plasticity has been widely reported. Changes in grouping tendencies and mating systems of pronghorn have been related to variation in food distribution and richness (Bromley 1969; Kitchen 1974), density and demographic changes (Byers and Kitchen 1988), interaction rates (Byers 1997), and human disturbance such as fencing and hunting pressure (Bromley 1977; Copeland 1980). However, few studies (e.g., Byers 1997) have quantified the proximate determinants responsible for particular variations in grouping tendencies, especially when considered at multiple timescales. This gap in understanding constrains our ability to predict the consequences of management manipulations or habitat alterations on behavior and population dynamics (Je˛drzejewski et al. 2006). Since 1970, numbers of pronghorn in Yellowstone National Park (Wyoming and Montana) have exhibited periods of relative stability for 10–15 years, punctuated by relatively rapid, dramatic fluctuations in numbers (White et al. 2007a). Also, there have been dynamic and rapid changes in mating behaviors and migration tendencies during this period (White et al. 2007b). Understanding the influence of environmental and human factors on these changes is essential for developing feasible management strategies to conserve this population (,300 animals) of a native species of special concern that retains one of only a few pronghorn migrations remaining in the greater Yellowstone region (White et al. 2007b). Additionally, understanding how pronghorn aggregate and behaviorally respond to predators will provide a more comprehensive understanding of multispecies interactions at the community level.
August 2012
WHITE ET AL.—GROUP DYNAMICS OF YELLOWSTONE PRONGHORN
generally ,15 cm deep (Yoakum 2004). Snow data for the Gardiner Basin winter range for Yellowstone pronghorn were not available. There is typically little snow accumulation at lower elevations in the basin, but snow buildup could restrict use of higher-elevation mountain foothills. Thus, we used a validated snow pack simulation model (Watson et al. 2009) to construct a covariate that represented the average SWE on the nearby Blacktail Deer Plateau area of the pronghorn range for each 14-day period through the year (see map in White et al. [2007b]). We constructed a covariate that indexed vegetation greenness using 14-day averaged normalized difference vegetation index (NDVI) values (Bartlette et al. 2006) for grassland and meadow areas used by pronghorn across their range. NDVI is a remotely sensed measure of vegetation that is derived from daily satellite imagery (near infrared and visible red reflectance values) and varies linearly with the fraction of photosynthetically active radiation (Huete et al. 2002; Pettorelli et al. 2005). We constructed a covariate, HABITAT, by classifying each pronghorn group location into 1 of 3 vegetation cover types (grassland, shrub, or forest) based on the potential for vegetation to provide hiding cover or obscure visibility. We considered the influence of additional covariates on pronghorn grouping tendencies using time of year and estimates of pronghorn and wolf numbers. We developed the categorical covariate, PERIOD, which categorized 3 biologically relevant time periods: fawning (1 May–30 June; including neonatal rearing and lactation), summer (1 July–30 September; characterized by high food resources, natal rearing, and the rut), and winter (1 October–30 April; characterized by low food resources and pregnancy). For the analyses of typical group size, we could not use this covariate because the 14-day interval often encompassed more than 1 time period. Therefore, we defined SEASON as a continuous variable from 1 to 26 representing the 14-day intervals through each year. We used analysis of variance (ANOVA) and Tukey’s multiple comparisons with unequal sample size (Kutner et al. 2005) to evaluate the change in pronghorn group size, variability, and composition, and to identify differences in the means among fawning, summer, and winter periods. We defined DENSITY as the number of pronghorn per unit area on the winter range (30 km2), migratory summer range (244 km2), and resident summer range (57 km2—White et al. 2007b). Airplane counts were conducted in late March or early April before pronghorn migrated. We multiplied each count (corrected for an 89% sightability—White et al. 2007a) by the proportion of radiocollared females migrating that year (range ¼ 0.68–0.81; n ¼ 12–28 collars—White et al. 2007b) to estimate the density of migratory pronghorn on their summer range. The proportions of the population estimated as migratory during aerial composition flights in July–September were similar (0.65–0.69) to the proportions of radiocollared animals migrating during 2003–2006 (White et al. 2007b). For typical group size analyses that summed groups over 14-day intervals, we estimated the density of pronghorn on the winter (30 km2) and summer (301 km2) ranges. Wolf numbers were
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
National Park using net guns from helicopters or ground darting and fitted with very-high-frequency and global positioning system radiocollars (models LMRT-2 and GPS 3300S; Lotek Wireless, Newmarket, Ontario, Canada). We handled all pronghorn in compliance with guidelines recommended by the American Society of Mammalogists (Sikes et al. 2011). Telemetry homing techniques were used to locate radiocollared pronghorn at least twice monthly from the ground. We recorded the number, sex, and composition (i.e., doe, buck, or fawn) of pronghorn in groups (1 animal) containing radiocollared female pronghorn during 1999–2005. We did not sample groups without radiocollared animals, including groups composed solely of males. Dynamic group composition has previously been reported in pronghorn (Kitchen 1974). We calculated correlation coefficients from successive observations of groups containing individual radiocollared pronghorn during summer (excluding the June parturition period) and winter. We used observations at least 5 days apart in these analyses to allow adequate time for changes in group composition and calculated correlations between group sizes of successive locations according to Eberhardt et al. (1998). We developed 3 response variables related to different elements of ungulate grouping behavior: mean group size, group size variation, and typical group size (Gower et al. 2009a). Group size was defined as a single animal or individuals ‘‘that remain together for a period of time while interacting with one another to a distinctly greater degree than with other conspecifics’’ (Wilson 1975:585). Groups were counted from the ground using binoculars or spotting scopes. Group size variation was derived by calculating the absolute difference between a given group size and the mean group size for that particular year, and was used to assess the frequency of joining and dispersion from groups. Because this metric takes into account all groups observed during a given year, it is not dependent on the interval between relocations or the difference in group size between successive locations. Typical group size was defined as the size of the group the average P P in 2 which animal finds itself and calculated as Gi / Gi, where Gi is the size of the ith group (Jarman 1974). This metric is a descriptive statistic, which was used to complement mean group size and provide more information and insight into pronghorn grouping tendencies. Typical group size has advantages over mean group size because the metric is less sensitive to the number of records of solitary animals than mean group size. Consequently, typical group size provides a group count representative of one that is actually experienced by each herd member (Heard 1992; Lingle 2003). Because typical group size is derived from the sum of an assemblage of groups over a determined time period, we calculated it using 14-day sampling intervals. We explored the influence of landscape attributes on pronghorn grouping tendencies using estimates of snow pack, vegetation greenness, and habitat type. The winter distribution of pronghorn is generally restricted to lower-elevation, windswept areas where food is exposed and snows are
1131
1132
Vol. 93, No. 4
JOURNAL OF MAMMALOGY
hypothesized constant grouping behavior for the respective response variables. Because of strong correlation between some covariates, we did not include combinations of SEASON þ SWE (0.63), DENSITY þ PERIOD (0.93), NDVI þ PERIOD (0.76), NDVI þ DENSITY (0.85), or SWE þ NDVI (0.71) in the same models. We used Akaike’s information criterion (AIC) to rank models given the data and compare the relative ability of each model to explain variation in the data (Burnham and Anderson 2002, 2004). We used AIC corrected for small sample size (AICc) for analyses of typical group size because this response variable was derived from the sum of an assemblage of groups over a determined time period and, therefore, was a smaller data set than the one used to analyze mean group size and group size variation. Akaike’s model weights (wis) were used to address model selection uncertainty when no clear support for a single model was found (Burnham and Anderson 2002). Predictor weights were used to assess the relative importance of an individual covariate within a model (Burnham and Anderson 2001). Predictor weights were quantified by summing the Akaike weights for all models containing a given covariate. We computed the predictor weights for each covariate considered in the model suite. The covariate with the largest predictor weight was considered the most important, whereas the covariate with the smallest sum was considered least important. This procedure allowed us to make inferences concerning the relative importance of individual covariates, in addition to the relative importance of the model within which the covariates were contained (Burnham and Anderson 2001). We also used evidence ratios, the ratio of Akaike’s weights wi/ wj, calculated as (L(gijx)/L(gjjx)), to determine the relative likelihood of model pairs given the data (Burnham and Anderson 2002, 2004). Covariate coefficients and variance of the random effects were estimated using restricted maximum likelihood. Comparable AIC values were calculated using maximum-likelihood estimation (Loison et al. 2004; Pinheiro and Bates 2000). All statistical analyses were performed using the R environment for statistical computing and the nlme package (Pinheiro and Bates 2000; R Development Core Team 2006).
RESULTS We collected a total of 2,846 group sizes by sampling 53 adult, female, radio-marked pronghorn during June 1999–May 2005. Mean group sizes were relatively small (X¯ ¼ 4.8, variance [s2] ¼ 12.6; 95% confidence interval [95% CI] ¼ 4.5– 5.2) during fawning in May and June, but larger (X¯ ¼ 10.2, s2 ¼ 54.3; 95% CI ¼ 9.4–10.9) in the summer during July– September, and through the winter during October–April (X¯ ¼ 11.9, s2 ¼ 107.2; 95% CI ¼ 11.5–12.4). Transformed group sizes differed between the 3 seasonal periods (ANOVA: F2,2,843 ¼ 211.4, P , 0.0001), with the largest differences ˆ ln(summer)ln(fawning) occurring between fawning and summer (D ¼ 0.75; 95% CI ¼ 0.62–0.89; P , 0.001) and fawning and ˆ ln(winter)ln(fawning) ¼ 0.87; 95% CI ¼ 0.77–0.97; P , winter (D
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
estimated annually in December based on repeated counts of packs, the majority of which contained radiocollared animals (Smith 2005). We used wolf numbers (PREDATION) as an index for overall predation risk because precise estimates for other predators (e.g., coyotes and bears) were unavailable. We used mixed-effects linear models (Pinheiro and Bates 2000) to evaluate competing hypotheses in the form of a priori models for assessing group size and group size variation, which treated the covariates as fixed effects. We treated individual animal identity (ID) as a random effect (i.e., intercept-only) because these data sets included repeated observations of the same individuals through time. By including a random effect associated with an individual, we accounted for any tendencies by individuals to repeatedly associate with similar-sized groups. This covariate allowed us to partition the variation in group size between individuals (intercept; rˆ 2B) and within an individual (residual; rˆ 2W), and to determine how much of the variation in group size was accounted for by the fixed and random effects. We randomly selected the marked pronghorn to associate with a particular group when more than 1 radiocollared pronghorn was observed in a group. We used multiple regression analyses (not mixed-effects linear models) to evaluate competing a priori models for assessing typical group size because this data set did not include repeatedmeasures sampling. For both the ANOVA and mixed-effects linear models, we transformed the group size response variables using the natural log to conform to linear model assumptions and used diagnostic residual plots to assess whether the assumptions of normality and constant variance had been met. We analyzed the group size variation and typical group size response variables using both the logged and untransformed data sets. Similar results were obtained, so for ease of interpretation, we chose to use the untransformed data set for all analyses of group size variation and typical group size. Combinations of the covariates were included in the additive form or as an interaction for all analyses. We predicted a fixed rate of change in the response variable per unit change in the predictor variable (i.e., linear form). We centered and scaled all continuous covariates to facilitate comparisons and interpretations of covariate coefficients. Variance inflation factors (VIFs), which measure multicollinearity among covariates, were calculated for all additive and interactive combinations of predictors. Those models that included predictor combinations with VIF , 6 were retained in the model list (Kutner et al. 2005). Correlation coefficients also were calculated to further check for multicollinearity between the predictor variables, with correlation coefficients , 0.6 and . 0.6 retained in the model list. For the analyses of mean group size and group size variation, we constructed a suite of 38 models that included the covariates PERIOD, HABITAT, DENSITY, NDVI, SWE, and PREDATION (Appendix I). For analyses of the typical group size, we constructed a suite of 28 models, substituting the covariate PERIOD with SEASON (Appendix I). We also included a NULL model into each of the 3 model suites that
August 2012
WHITE ET AL.—GROUP DYNAMICS OF YELLOWSTONE PRONGHORN
during all seasons, but males comprised a relatively low percentage of groups during May and June (X¯ ¼ 8.9%, s2 ¼ 1.9%) and July–September (X¯ ¼ 10.6%, s2 ¼ 1.9%; Fig. 1). Males comprised a higher percentage of groups during October–March (X¯ ¼ 21.1%, s2 ¼ 3.2%; ANOVA: F2,2,843 ¼ 142.1, P , 0.0001). Model selection results for the group size analysis supported one model that received 0.88 of the model weight and contained the covariates PERIOD and PREDATION, and the interaction between PERIOD and PREDATION (PERIOD 3 PREDATION; Table 1). The evidence ratio for the top model versus the 2nd-ranked model (wi ¼ 0.04) was 22, which suggests strong support for the top model and low model selection uncertainty. PERIOD (1.0) and PREDATION (0.97) received almost all of the predictor weight, whereas DENSITY, SWE, HABITAT, and NDVI accounted for an insignificant amount of the predictor weight. PERIOD was the most influential predictor of mean group size for pronghorn, with groups being smaller during fawning compared to the slightly larger groups in summer and winter. Removal of this predictor resulted in the DAIC value increasing by .100. Although PREDATION appeared in the highest ranked model that received most (0.88) of the model weight, the confidence limits spanned zero for both the fawning and the summer periods (Table 2). Thus, we did not find substantial support for our prediction that group sizes would be larger as the presence of predators increased during these 2 periods of the year. However, there was some evidence that predation may negatively affect pronghorn group sizes in the winter (Table 2). Results of the random effects from the mixed modeling indicated that within-animal variability (rˆ 2W ¼ 0.64) was substantially higher than between-animal variability (rˆ 2B ¼ 0.023). Model selection results for the group size variation analysis supported one model that received 0.74 of the model weight and contained the covariates DENSITY, SWE, and PREDATION (Table 1). The evidence ratio for the top model versus the 2nd-ranked model (wi ¼ 0.20) was 4.4, suggesting strong support for the top model and low model selection uncertainty. Overall, models that contained the covariates DENSITY and SWE accounted for nearly all of the model weight, whereas models that included the covariates HABITAT, PERIOD, and NDVI were not well supported and accounted for an insignificant amount of the predictor weight (Table 2). As predicted, pronghorn density had a positive effect on group size variation, whereas increasing SWE reduced group size variation. Contrary to our predictions, PREDATION was negatively correlated to group size variation indicating that as predators on the landscape increased, grouping became more stable (Table 2). Results of the random effects from the mixed modeling indicated that the between animal variability was relatively low, suggesting no animals had a specific tendency to change groups more frequently than others (rˆ 2B ¼ 1.2). Conversely, variation in grouping behavior was inconsistent within individual pronghorn, as evidenced by the substantially high within-animal variability (rˆ 2W ¼ 44.4).
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
ˆ defines the difference in transformed mean group 0.001). D size between 2 seasonal periods, based on Tukey’s multiple comparison tests. Group sizes differed only slightly between ˆ ln(winter)ln(summer) ¼ 0.12; 95% CI ¼ summer and winter (D 0.01–0.23; P ¼ 0.02). An advantage of log transformations is that the results can be interpreted on the original scale of the variable (Ramsey and Schafer 2002). By taking the exponent of the difference in the mean log between 2 time periods, we can obtain a multiplicative value for the change in the median between the time periods. Consequently, the median group size during winter was 2.4 times larger than during fawning (95% CI ¼ 2.2–2.7) and 1.1 times larger than during summer (95% CI ¼ 1.0–1.3). The median group size during summer was 2.1 times greater than during fawning (95% CI ¼ 1.9–2.4). These observed changes in mean group size between the temporal periods on the untransformed scale represented a 112% increase in mean pronghorn group size from fawning to summer, a 148% increase in mean group size from fawning to winter, and a 17% increase in mean group size from summer to winter. There were also observable differences in the stability of pronghorn groups through the year (ANOVA: F2,2,843 ¼ 12.76, P , 0.0001). Variation in mean group size was relatively small during the fawning period (mean absolute difference, which is the absolute difference between a given group size and the mean group size for that particular year ¼ 5.5; 95% CI ¼ 5.2– 5.8; n ¼ 452) and summer (mean absolute difference ¼ 5.8; 95% CI ¼ 5.4–6.3; n ¼ 367). There was more instability in grouping behavior during winter (mean absolute difference ¼ 7.1; 95% CI ¼ 6.7–7.3; n ¼ 2,027). Typical group size was substantially larger than mean group size, which is always the case when the variance of mean group size is larger than zero (Heard 1992). Typical group size ranged from 4.7 to 42.0 (X¯ ¼ 15.8, n ¼ one hundred twenty-four 14day sampling intervals), with a mean typical group size of 7.2 (95% CI ¼ 6.6–7.8; n ¼ 25) during fawning, 14.7 (95% CI ¼ 13.0–16.5; n ¼ 31) during summer, and 19.5 (95% CI ¼ 17.7– 21.4; n ¼ 68) during winter (ANOVA: F2,2,843 ¼ 211.4, P , 0.0001). Typical group size differed among the 3 biological ˆ ln(summer)ln(fawning) ¼ 7.52, 95% CI ¼ 3.4–11.57, time periods: D ˆ ln(winter)ln(fawning) ¼ 12.34, 95% CI ¼ 8.82–15.86, P , 0.001; D ˆ ln(winter)ln(summer) ¼ 4.81, 95% CI ¼ 1.55– P , 0.001; and D ˆ defines the difference in mean typical 8.08, P , 0.001. D group size between 2 seasonal periods, based on Tukey’s multiple comparison tests. This represented a 106% increase in mean typical group size from fawning to summer, 181% increase from fawning to winter, and 36% increase from summer to winter. Radiocollared individuals switched groups frequently and mixed with other animals during both summer and winter. The numbers of animals in groups with collared females often changed substantially between locations . 5 days apart and ranged between 1 and 87 individuals. Correlations between successive locations were largely ,0.40 and not statistically significant (winter: X¯ ¼ 0.24 6 0.06 SE; summer: X¯ ¼ 0.26 6 0.03). Mixed groups of females and males were observed
1133
1134
Vol. 93, No. 4
JOURNAL OF MAMMALOGY
Model selection results for the typical group size analysis supported 3 top models that were within 2 DAIC of one another and, in combination, received 0.91 of the model weight (Table 1). Each of these models contained the covariates DENSITY and SWE, which accounted for 0.99 and 0.90 of the predictor weights, respectively (Table 2). As predicted, typical group size was positively correlated with pronghorn density, whereas there was a negative relationship between typical group size and SWE (Table 2). An interaction between DENSITY and SWE appeared in one of the top-ranked models, but confidence intervals spanned zero. Similarly PREDATION appeared in one of the top-ranked models, but it only accounted for 0.19 of the predictor weight and also had confidence intervals that spanned zero.
DISCUSSION This study was unique in attempting to identify predictors associated with the grouping tendencies of pronghorn using multivariate statistical and model selection methods. Radiocollared females switched groups frequently and mean group size was strongly influenced by reproductive and social factors because groups were smaller during parturition and the neonatal rearing period. Pronghorn density, snow pack, and predator numbers apparently influenced group sizes and cohesion, whereas habitat type and vegetation greenness were not good predictors of group sizes or group stability. Group cohesion increased as snow pack at higher elevations and predator numbers increased and constrained pronghorn to a smaller area on the winter range. Conversely, groups exhibited
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
FIG. 1.—Seasonal variation in the sizes and sex composition of groups containing radiocollared, adult, female pronghorn in Yellowstone National Park, Montana and Wyoming, during June 1999–April 2005.
August 2012
1135
WHITE ET AL.—GROUP DYNAMICS OF YELLOWSTONE PRONGHORN
Model structure ln(group size) PERIOD þ PREDATION þ (PERIOD 3 PREDATION) PERIOD þ HABITAT þ SWE þ PREDATION PERIOD þ SWE þ PREDATION PERIOD þ HABITAT þ PREDATION PERIOD þ PREDATION Group size variation DENSITY þ SWE þ PREDATION HABITAT þ DENSITY þ SWE þ PREDATION DENSITY þ SWE PERIOD þ SWE þ PREDATION DENSITY þ SWE þ (DENSITY 3 SWE) Typical group size DENSITY þ SWE DENSITY þ SWE þ (DENSITY 3 SWE) DENSITY þ SWE þ PREDATION DENSITY þ SEASON þ (DENSITY 3 SEASON) DENSITY þ SEASON
k DAIC
wi
8 9 7 8 6
0.00 6.20 6.94 7.16 7.50
0.88 0.04 0.03 0.02 0.02
6 8 5 7 6
0.00 2.95 6.78 6.87 7.40
0.74 0.17 0.02 0.02 0.02
4 5 5 5 4
0.00 0.64 1.53 3.03 8.95
0.41 0.30 0.19 0.09 0.00
fracturing and rejoining at a greater intensity as pronghorn density increased. Yellowstone pronghorn were in significantly smaller groups during May and June when females dispersed to isolated sites to give birth and hide their newborn fawns and males became solitary or formed small bachelor groups. Fawns joined their mothers in nursery groups after they were 3–6 weeks old and capable of rapid, sustained running. Group sizes increased during July–August when mature males began defending groups of females (harems) from other males as the rut approached. Pronghorn formed larger, mixed-sex groups during November–March when all pronghorn congregated on the winter range. Similar seasonal changes in grouping tendencies were reported for pronghorn at the nearby National Bison Range in western Montana, where males remained separate from females (,10% mixed-sex groups), and groups were relatively small (X¯ 5) during the entire spring and summer (April–September—Byers 1997). Surprisingly, mean group sizes were not much larger in winter than in summer, even though migratory pronghorn that used different areas and did not interact during summer aggregated and mixed on a 30-km2 range during winter. This finding likely reflects the fact that about one-third of the groups we observed with radiocollared females through the year had fewer than 5 animals, which would decrease mean group size. In contrast, typical group size, which is the size of the group in which the average animal finds itself, increased from fawning through summer and was highest during winter. Thus, the
TABLE 2.—Coefficient estimates and 95% confidence limits (CIs) from the top models for factors influencing the mean group size, group size variation, and typical group size of pronghorn in Yellowstone National Park during 1995–2005. Bold font denotes estimates with confidence limits that do not span zero. SWE ¼ snow water equivalent. Covariate ln(group size) Intercept—PERIOD–Fawning PERIOD–Summer PERIOD–Winter PREDATION–Fawning PERIOD–Summer 3 PREDATION PERIOD–Winter 3 PREDATION Group size variation Intercept DENSITY SWE PREDATION Typical group size Intercept DENSITY SWE
Estimate
Lower CI
Upper CI
1.31 0.75 0.86 0.00 0.02 0.10
1.22 0.64 0.77 0.07 0.08 0.18
1.40 0.86 0.94 0.07 0.13 0.02
6.85 0.99 0.55 0.41
6.38 0.68 0.86 0.69
7.31 1.30 0.23 0.14
15.83 6.28 2.04
14.83 5.11 3.21
16.84 7.44 0.87
typical group size metric, which is less sensitive to the number of records of small groups found in this study, was likely a better metric for evaluating and describing seasonal trends in grouping through the year. Larger aggregations of pronghorn during summer and winter were coupled with decreased group cohesion. During these periods, pronghorn aggregated and dispersed more frequently than during the fawning period. This can be explained by higher densities of animals during summer and winter compared to during fawning, and might also be influenced by mating behavior and the constant fracturing and rejoining of groups as males repeatedly attempted to keep groups of females together. Larger aggregations of pronghorn in winter also showed dynamic grouping behavior, but these dynamics were dampened as winter severity increased. This finding has been reported for other ungulates, including elk that inhabit the deep-snow conditions of central Yellowstone, where the depth and density of the local snow inhibited movement among foraging patches (Gower et al. 2009a). Larger aggregations are likely a result of pronghorn congregating in areas with reduced snow accumulations and available food resources. Using microhabitats that have shallow or no snow and provide protection from the wind is typical for animals trying to conserve energy in a harsh winter environment (Moen 1973). Contrary to our expectations, a slight decrease in mean group size was observed as the number of predators increased on the landscape. This particular behavioral response did not appear to be universally adopted throughout the year, and it is unclear if the decrease was biologically significant. The possible reduction in group size in relation to predators appeared to be a response employed only during winter, which could suggest that grouping behavior in pronghorn might have different drivers in different seasons. Incongruent with other studies that have documented reduced stability in group dynamics with increased predators on the landscape (e.g.,
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
TABLE 1.—Model selection results for pronghorn group size and group size variation mixed-effects model analyses, and typical group size multiple regression analyses. Residual error for the mixed modeling accounts for 1 parameter value for the ln(group size) and group size variation analyses. All models are ranked according to Akaike information criterion (AIC) and presented along with the number of parameters (k), DAIC value (i.e., change in AIC relative to the best model), and Akaike weights (wis). The AIC values for the top models were 6,875.4 for ln(group size), 18,921.2 for group size variation, and 789.2 for typical group size. SWE ¼ snow water equivalent.
1136
Vol. 93, No. 4
JOURNAL OF MAMMALOGY
frequently rather than staying with the same group. During winter, radiocollared female pronghorn that were year-round residents of Gardiner Basin were frequently located in groups with pronghorn that migrated to higher elevations in summer. Similar fluidity in associations has been reported in several other pronghorn populations (Folker 1956; Gregg 1955; Kitchen 1974; Swanger 1977) and related to changes in density and rates of agonistic interactions and the lack of matrilineal social bonds (Byers 1997; Fichter 1972). These findings suggest that radiocollared female pronghorn in Yellowstone become well mixed and move freely within the population. Thus, it appears reasonable to use population estimation methods (e.g., Petersen mark–recapture) that assume marked animals are randomly distributed in the population. The fact that visibility bias causes an observer to see a higher proportion of larger groups does not necessarily bias estimates if marked animals are randomly distributed in the population (Eberhardt et al. 1998). Grouping tendencies of Yellowstone pronghorn appear to be strongly influenced by environmental, reproductive, and social factors. Thus, biologists should consider the effects of grouping behavior when planning the timing and location of conservation actions to control disease, protect or restore key habitat, regulate harvests, or limit adverse effects of development and recreation on pronghorn. For example, Yellowstone pronghorn tend to avoid areas with deep snow and aggregate during winter on lower elevations of Gardiner Basin. Unfortunately, this winter range has degraded over decades due to decreased sagebrush and because migration routes to historic wintering habitat outside the park have been fragmented by development, fencing, and other land-use practices (Boccadori et al. 2008; White et al. 2007b). Thus, the National Park Service has initiated restoration to reestablish native vegetation and provide more habitats for larger groups of pronghorn. Also, the State of Montana eliminated harvest of pronghorn in the area adjacent to the northern boundary of the park, which allows pronghorn to feed in and migrate through this area with fewer disturbances to groups. In addition, park personnel worked with the United States Forest Service, State of Montana, private landowners, and the National Parks Conservation Association to improve connectivity between the park and historic winter ranges to the north. Similar problems and remedial actions are facing managers of migratory ungulates worldwide.
ACKNOWLEDGMENTS Financial support for this project was provided by the Bernice Barbour Foundation, Montana State University, National Park Service, University of Idaho, Yellowstone Association, Yellowstone Ecological Research Center, and the Yellowstone Park Foundation. We thank K. Barnowe-Meyer, V. Boccadori, J. Byers, K. Nittinger, D. Thompson, and J. Treanor for data collection efforts; the dozens of technicians and volunteers that assisted with the project; Helicopter Capture Services, Montana Aircraft, Hawkins & Powers Aviation, Leading Edge Aviation, J. Powers, T. Roffe, and S. Sweeney for capture and aerial telemetry support; B. Crabtree and D. Smith for
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
Gower et al. 2009a), pronghorn exhibited a slight increase in group stability, with less group size variation as the number of predators on the landscape increased. One potential benefit of dynamic grouping behavior is that prey animals can balance the conflicting demands of minimizing predation risk and maximizing food acquisition (Gower et al. 2009a). In the case of pronghorn, specialized physiological and cursorial abilities that make them successful at escaping predators via fleeing might be a more beneficial approach than grouping (Byers 1997). Fleeing might not be an effective strategy for some ungulate species, such as elk, that reside at higher elevations with deeper snow, or for individuals that reside in heterogeneous landscapes where landscape features present obstacles that would hinder escape. However, there is little snow on the lower elevations of the Gardiner Basin winter range and the habitat consists of large expanses of sagebrush-scrub and grassland that rarely inhibits a pronghorn’s ability to flee from predators. In addition to mitigating predation risk by adopting an antipredator behavior of grouping, elk in Yellowstone also changed their movement patterns (Gower et al. 2009b). Similarly, trade-offs between predation risk and foraging differed between migrant strategies for elk (Hebblewhite and Merrill 2007). Thus, it appears that different antipredator responses are adopted to take advantage of, or accommodate, environmental, behavioral, or physiological capabilities, or a combination of these. Therefore, we cannot expect that different prey species would universally adopt the same behavioral response. Another explanation for the lack of observable changes in grouping tendencies in response to predators could be that our annual index of predation was too coarse a metric for evaluating short-duration grouping tendencies of prey. Costly antipredator behaviors may be adopted during bouts of elevated predation threat, but relaxed when the imminent threat of predation subsides (Gower et al. 2009a; White and Berger 2001; Wolff and Van Horn 2003). Therefore, if pronghorn were changing their grouping behavior at a short timescale, we might have missed these fine-scale behavioral responses. For example, pronghorn might not be selecting for smaller groups as an antipredator strategy, but smaller groups could be a consequence of groups fracturing after repeated encounters with predators such as coyotes and wolves. Often there is extreme spatial and temporal variation in predator presence, so this increased intensity of fracturing and rejoining of individuals as a consequence of predators roaming the landscape is similar to the responses of ungulates in some areas to relatively intense human hunting pressure (Gude et al. 2006; Je˛drzejewski et al. 2006). Additionally, we used wolf presence as a metric of predation. This metric might not adequately reflect predation threat on the landscape, and a metric that describes the risk of predation by coyotes might provide additional insights at certain times of the year, such as during fawning. Group sizes changed substantially among successive locations of Yellowstone pronghorn, suggesting that females were mixing and switching associations with other animals
August 2012
WHITE ET AL.—GROUP DYNAMICS OF YELLOWSTONE PRONGHORN
information and discussions regarding the predator guild; K. Tonnessen and the Rocky Mountains Cooperative Ecosystem Studies Unit for facilitating cooperative funding agreements; and R. Garrott, K. Manlove, and G. Plumb for critiques of earlier versions of this manuscript.
LITERATURE CITED
FICHTER, E. 1972. On the nature of pronghorn groups. Antelope States Workshop Proceedings 5:222–230. FOLKER, R. V. 1956. A preliminary study of an antelope herd in Owyhee County, Idaho. M.S. thesis, University of Idaho, Moscow. FRYXELL, J. M. 1991. Forage quality and aggregation by large herbivores. American Naturalist 138:478–498. FRYXELL, J. M., J. F. WILMSHURST, AND A. R. E. SINCLAIR. 2004. Predictive models of movement by Serengeti grazers. Ecology 85:2429–2435. GOWER, C. N., R. A. GARROTT, P. J. WHITE, S. CHERRY, AND N. G. YOCCOZ. 2009a. Elk group size and wolf predation: a flexible strategy when faced with variable risk. Pp. 401–422 in The ecology of large mammals in central Yellowstone: sixteen years of integrated field studies (R. A. Garrott, P. J. White, and F. G. R. Watson, eds.). Elsevier, San Diego, California. GOWER, C. N., R. A. GARROTT, P. J. WHITE, F. G. R. WATSON, S. S. CORNISH, AND M. S. BECKER. 2009b. Spatial responses of elk to wolf predation risk: using the landscape to balance multiple demands. Pp. 373–399 in The ecology of large mammals in central Yellowstone: sixteen years of integrated field studies (R. A. Garrott, P. J. White, and F. G. R. Watson, eds.). Elsevier, San Diego, California. GREGG, H. A. 1955. Summer habits of Wyoming antelope. Ph.D. dissertation, Cornell University, Ithaca, New York. GUDE, J. A., R. A. GARROTT, J. J. BORKOWSKI, AND F. KING. 2006. Prey risk allocation in a grazing environment. Ecological Applications 16:285–298. HEARD, D. C. 1992. The effect of wolf predation and snow cover on musk-ox group size. American Naturalist 139:190–204. HEBBLEWHITE, M., AND E. MERRILL. 2007. Multiscale wolf predation risk for elk: does migration reduce risk? Oecologia 152:377–387. HEBBLEWHITE, M., AND D. H. PLETSCHER. 2002. Effects of elk group size on predation by wolves. Canadian Journal of Zoology 80:800– 809. HIRTH, D. H. 1977. Social behavior in white-tailed deer in relation to habitat. Wildlife Monographs 53:1–55. HUETE, A., K. DIDAN, T. MIURA, E. P. RODRIGUEZ, X. GAO, AND L. G. FERREIRA. 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment 83:195–213. ILLIUS, A. W., AND C. FITZGIBBON. 1994. Cost of vigilance in foraging ungulates. Animal Behaviour 47:481–484. JARMAN, P. J. 1974. Social organization of antelope in relation to their ecology. Behaviour 48:215–267. JE˛DRZEJEWSKI, W., B. JE˛DRZEJEWSKA, H. OKARMA, AND A. L. RUPRECHT. 1992. Wolf predation and snow cover as mortality factors in the ungulate community of the Białowiez˙a National Park, Poland. Oecologia 90:27–36. JE˛DRZEJEWSKI, W., J. SPAEDTKE, J. F. KAMLER, B. JE˛DRZEJEWSKA, AND U. STENKEWITZ. 2006. Group size dynamics of red deer in Białowiez˙a Primeval Forest, Poland. Journal of Wildlife Management 70:1054– 1059. KITCHEN, D. W. 1974. Social behavior and ecology of the pronghorn. Wildlife Monographs 38:1–96. KUTNER, M. H., C. J. NACHTSHEIM, J. NETER, AND W. LI. 2005. Applied linear statistical models. McGraw-Hill, New York. LINGLE, S. 2003. Group composition and cohesion in sympatric whitetailed deer and mule deer. Canadian Journal of Zoology 81:1119– 1130. LIPETZ, V. E., AND M. BEKOFF. 1982. Group size and vigilance in pronghorn. Zeitschrift f¨ur Tierpsychologie 58:203–216.
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
BARNOWE-MEYER, K. K., P. J. WHITE, T. L. DAVIS, AND J. A. BYERS. 2009. Predator-specific mortality of pronghorn on Yellowstone’s northern range. Western North American Naturalist 69:186–194. BARTLETTE, R. A., D. A. TIMERSTEIN, AND J. C. EIDENSHINK. 2006. NDVI and derived image data: 1989–2005. United States Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, General Technical Report RMRS-GTR-179-DVD:1–6. BERGER, J., D. DANEKE, J. JOHNSON, AND S. H. BERWICK. 1983. Pronghorn foraging economy and predator avoidance in a desert ecosystem: implications for the conservation of large mammalian herbivores. Biological Conservation 25:193–208. BOCCADORI, S. J., P. J. WHITE, R. A. GARROTT, J. J. BORKOWSKI, AND T. DAVIS. 2008. Yellowstone pronghorn alter resource selection after sagebrush decline. Journal of Mammalogy 89:1031–1040. BROMLEY, P. T. 1969. Territoriality in pronghorn bucks on the National Bison Range, Moiese, Montana. Journal of Mammalogy 50:81–89. BROMLEY, P. T. 1977. Aspects of the behavioral ecology and sociobiology of the pronghorn (Antilocapra americana). Ph.D. dissertation, University of Calgary, Calgary, Alberta, Canada. BRUN, E. H. 1977. Winter behavior of pronghorns in relation to habitat. Journal of Wildlife Management 41:560–571. BURNHAM, K. P., AND D. R. ANDERSON. 2001. Kullback–Leibler information as a basis for strong inference in ecological studies. Wildlife Research 28:111–119. BURNHAM, K. P., AND D. R. ANDERSON. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Springer-Verlag, New York. BURNHAM, K. P., AND D. R. ANDERSON. 2004. Multimodel inference— understanding AIC and BIC in model selection. Sociological Methods and Research 33:261–304. BYERS, J. A. 1997. American pronghorn: social adaptations and the ghosts of predators past. University of Chicago Press, Chicago, Illinois. BYERS, J. A., AND D. W. KITCHEN. 1988. Mating system shift in a pronghorn population. Behavioral Ecology and Sociobiology 22:355–360. CHAPMAN, C. A., AND L. J. CHAPMAN. 2000. Constraints on group size in red colobus and red-tailed guenos: examining generality of the ecological constraints model. International Journal of Primatology 21:565–585. CLUTTON-BROCK, T. H., F. E. GUINNESS, AND S. D. ALBON. 1982. Red deer: behavior and ecology of two sexes. University of Chicago Press, Chicago, Illinois. COPELAND, G. L. 1980. Antelope buck breeding behavior, habitat selection and hunting impact. Idaho Department of Fish and Game Wildlife Bulletin 8:1–45. DEBLINGER, R. D., AND A. W. ALLDREDGE. 1989. Management implications of variations in pronghorn social behavior. Wildlife Society Bulletin 17:82–87. EBERHARDT, L. L., R. A. GARROTT, P. J. WHITE, AND P. J. GOGAN. 1998. Alternative approaches to aerial censusing of elk. Journal of Wildlife Management 62:1046–1055.
1137
1138
JOURNAL OF MAMMALOGY
SWEENY, J. M., AND J. R. SWEENY. 1984. Snow depths influencing winter movements of elk. Journal of Mammalogy 65:524–526. TELFER, E. S., AND J. P. KELSALL. 1984. Adaptions of some large North American mammals for survival in snow. Ecology 65:1828–1834. WATSON, F. G. R., T. N. ANDERSON, W. B. NEWMAN, S. S. CORNISH, AND T. R. THEIN. 2009. Modeling spatial snow pack dynamics. Pp. 85– 112 in The ecology of large mammals in central Yellowstone: sixteen years of integrated field studies (R. A. Garrott, P. J. White, and F. G. R. Watson, eds.). Elsevier, San Diego, California. WHITE, K. S., AND J. BERGER. 2001. Antipredator strategies of Alaskan moose: are maternal trade-offs influenced by offspring activity? Canadian Journal of Zoology 79:2055–2062. WHITE, P. J., J. E. BRUGGEMAN, AND R. A. GARROTT. 2007a. Irruptive population dynamics in Yellowstone pronghorn. Ecological Applications 17:1598–1606. WHITE, P. J., T. L. DAVIS, K. K. BARNOWE-MEYER, R. L. CRABTREE, AND R. A. GARROTT. 2007b. Partial migration and philopatry of Yellowstone pronghorn. Biological Conservation 135:518–526. WHITE, P. J., AND R. A. GARROTT. 2005. Northern Yellowstone elk after wolf restoration. Wildlife Society Bulletin 33:942–955. WILSON, E. O. 1975. Sociobiology: the new synthesis. Harvard University Press, Cambridge, Massachusetts. WOLFF, J. O., AND T. VAN HORN. 2003. Vigilance and foraging patterns of American elk during the rut in habitats with and without predators. Canadian Journal of Zoology 81:266–271. YOAKUM, J. D. 2004. Habitat characteristics and requirements. Pp. 409–445 in Pronghorn: ecology and management (B. W. O’Gara and J. D. Yoakum, eds.). Wildlife Management Institute, University Press of Colorado, Boulder.
Submitted 20 July 2011. Accepted 17 January 2012. Associate Editor was Janet L. Rachlow.
Downloaded from https://academic.oup.com/jmammal/article-abstract/93/4/1129/960158 by guest on 11 January 2019
LOISON, A., E. J. SOLBERG, N. G. YOCCOZ, AND R. LANGVATN. 2004. Sex differences in the interplay of cohort and mother quality on body mass of red deer calves. Ecology 85:1992–2002. LOTT, D. F. 1991. Intraspecific variation in the social systems of wild vertebrates. Cambridge University Press, New York. MOEN, A. N. 1973. Wildlife ecology: an analytical approach. W. H. Freeman, San Franciso, California. O’GARA, B. W. 2004. Behavior. Pp. 145–194 in Pronghorn: ecology and management (B. W. O’Gara and J. D. Yoakum, eds.). Wildlife Management Institute, University Press of Colorado, Boulder. PETTORELLI, N., J. O. VIK, A. MYSTERUD, J. GAILLARD, C. J. TUCKER, AND N. C. STENSETH. 2005. Using the satellite-derived NDVI to assess ecological responses to environmental change. Trends in Ecology & Evolution 20:503–510. PINHEIRO, J. C., AND D. M. BATES. 2000. Mixed-effects models in S and S-Plus. Springer, New York. PULLIAM, H. R. 1973. On the advantage of flocking. Journal of Theoretical Biology 38:419–422. RAMSEY, F., AND D. SCHAFER. 2002. The statistical sleuth: a course in methods of data analysis. Duxbury Press, Pacific Grove, California. R DEVELOPMENT CORE TEAM. 2006. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. Accessed December 2009. Sikes, R. S., Gannon, W. L. and THE ANIMAL CARE AND USE COMMITTEE OF THE AMERICAN SOCIETY OF MAMMALOGISTS. 2011. Guidelines of the American Society of Mammalogists for the use of wild mammals in research. Journal of Mammalogy 92:235–253. SMITH, D. W. 2005. Ten years of Yellowstone wolves. Yellowstone Science 13:7–33. SWANGER, J. L. 1977. Aspects of doe pronghorn social structure on the National Bison Range. M.S. thesis, University of Montana, Missoula.
Vol. 93, No. 4
