Sociality Increases Juvenile Survival after a Catastrophic Event in the Feral Horse (Equus caballus)

Transcription

1 Iowa State University From the SelectedWorks of Cassandra M.V. Nuñez 2015 Sociality Increases Juvenile Survival after a Catastrophic Event in the Feral Horse (Equus caballus) Cassandra M.V. Nuñez, Princeton University James S. Adelman, Virginia Polytechnic Institute and State University Daniel I. Rubenstein, Princeton University Available at:

2 This is a manuscript of an article in Behavioral Ecology 26 (2015): 138. Posted with permission Full title: Sociality increases juvenile survival after a catastrophic event in the feral horse (Equus caballus) Short title: Sociality increases juvenile survival in feral horses Cassandra M.V. Nuñez, corresponding author, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ, 08544, USA. Current address: Department of Biological Sciences, Virginia Polytechnic Institute and State University, Derring Hall Room 2125, 1405 Perry Street, Mail Code 0406, Blacksburg, VA, 24061, USA. nunezcmv@vt.edu Phone: James S. Adelman, Department of Biological Sciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA Daniel I. Rubenstein, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ, 08544, USA

3 Lay Summary We know group living benefits social animals. But could the number of friends an animal has help it survive a catastrophic event? We asked whether sociality affected foals ability to survive the removal of close associates. Foals with more associates after the removal were more likely to survive. Moreover, associations formed before the removal, were critical to the survival of foals left without parents. Our results demonstrate another important link between sociality and animal survival. Abstract In several social species, adult associations have been linked to individual fitness. Less is known about offspring associations and the mechanisms by which they may influence fitness. We investigate the effects of sociality on juvenile survival in feral horses (Equus caballus). We use foal degree (number of associates) and foal weight (number of interactions) to assess sociality s importance to foal survival of a catastrophic event, the gathering and removal of 40% of the horse population. We found that 1) foal degree was a better predictor of foal survival than was foal weight; 2) following the gather, foals with access to at least one parent and some members of their natal groups were more likely to survive than were foals left with no such access; 3) foals with a higher degree both preand post-gather were more likely to survive; 4) the influence of pre-gather degree appeared to be more pronounced among foals without access to parent(s) and natal group members, and; 5) the influence of foal degree post-gather was important to the survival of all foals, regardless of parental and natal group access. These results add to our understanding of the benefits afforded by juvenile social networks. Moreover, our study 2

4 is unique in that foals were separated from parental support at a crucial point in development. We were therefore able to separate the effects of parental behavior on juvenile response, allowing us to more directly ask the question: do the behaviors exhibited by young mammals confer immediate and/or future benefits? Key Words Equus caballus, feral horse, juvenile, network degree, social networks, survival Introduction Social mammals typically organize themselves into coherent groups that feed, rest, and travel together (Alexander, 1974). Social living affords several direct benefits, including decreased predation (Powell, 1974; Lima and Bednekoff, 1999), increased reproductive opportunities (White and Galef Jr, 2000), enhanced foraging efficiency (Creel and Creel, 1995), and in some species, group members will aid in the rearing of young (Pusey and Packer, 1994). Within these social groups, however, there is often variation in individual levels of sociality: some animals will have more/fewer associates than others and will spend more/less time with certain individuals. These social links have been shown to affect individual fitness and characteristics associated with fitness. For example, predictable social relationships are associated with longer life in rock hyraxes (Procavia capensis) (Barocas et al., 2011). Similarly, female chacma baboons (Papio cynocephalus ursinus) that concentrate grooming activity to a few associates exhibit lower baseline glucocorticoid levels and smaller increases in glucocorticoids in response to psychological stressors than do females with more diffuse networks (Crockford et al., 2008; Wittig et al., 2008). Moreover, the formation of strong social bonds in adult 3

5 female baboons (Silk et al., 2009) and horses (Equus caballus) (Cameron et al., 2009) can confer survival benefits to their young. Few studies have addressed the social relationships that young animals develop; in fact, many studies exclude juveniles from their analysis, as juvenile networks will often closely mirror those of their mothers (Lusseau, 2003; Rubenstein et al., 2007; Sundaresan et al., 2007). However, studies exploring the relationships of young animals have demonstrated that these relationships can have direct effects on their fitness. For instance, in bottlenose dolphins (Tursiops sp.), individuals that are more centrally connected early in life are more likely to survive post-weaning (Stanton and Mann, 2012). Similarly, juvenile song sparrows (Melospiza melodia) with more associates during the fledgling period ultimately establish larger home ranges, often near these associates. These individuals may benefit from higher foraging success, decreased predation, reduced aggression from territorial males, and/or increased opportunities for social learning (Templeton et al., 2012). Though it is clear that social relationships formed during the juvenile stage can increase fitness, we know little about precisely how these relationships benefit young animals. For instance, are young animals with more associates more likely to weather day to day challenges? Do these associations provide a sociality baseline upon which young animals are able to build later in development or in adulthood? Do they help young animals survive unforeseen events? A more complete understanding of how young animals benefit from their social environment is critical if we are to truly comprehend sociality s role in the demography of gregarious species. In addition, such information is integral to the successful management and conservation of social animals, 4

6 particularly when such measures involve the removal and/or translocation of individuals (Williams and Lusseau, 2006). For example, catastrophes such as droughts, fires, tornadoes, floods, and hurricanes can have major impacts on individual survival and reproductive fitness (Shaffer, 1981). For many species, resilience to these largely unpredictable events is crucial, and for social species, the relationships formed with other group members may be important. The removal or translocation of individuals for management or conservation purposes can serve as a proxy for such events: for the animals left, the removal of associates is akin to the death of those individuals. Here, we assess the importance of relationships formed early in life on surviving such an event in a highly social ungulate, the feral horse (Equus caballus). In November, 1996, 40% of the horse population living on Shackelford Banks, NC, USA was found to have contracted equine infectious anemia. Equine infectious anemia is a chronic disease caused by a virus from the family Retroviridae, genus Lentivirus; it is limited to horses, donkeys and mules, and is characterized by periodic episodes of fever, hemolytic anemia, jaundice, depression, edema, and weight loss (Franco and Paes, 2011). In response, the National Park Service performed a gather in which these animals were removed and euthanized off-site. Six foals (aged 6-7 months) subsequently experienced what is akin to a catastrophic event the loss of their natal groups (i.e. their entire social network). Ten additional individuals (aged 4-7 months) had access to at least one parent and some remnant of their natal groups, enabling an evaluation of social network importance to juvenile survival in more typical and atypical situations. 5

7 Typically, foals, both male and female, will remain with their natal groups for 2-4 years (Tyler, 1972). These groups consist of usually one stallion (but sometimes more than one), one to several mare(s) and their offspring (Rubenstein, 1981). In this more characteristic scenario, the social connections juveniles form early in life will be highly correlated with the parents (primarily the mother s) social connections (Tyler, 1972; Rubenstein et al., 2007; Sundaresan et al., 2007). As such, in normal situations it is difficult to disentangle how juvenile survival is influenced by its own sociality versus its parents sociality. For the six individuals left alone, the Park Service removal provided a mechanism by which we were able to control for any current effects of the natal group (including maternal social networks) on juvenile sociality. Consequently, we were able to measure the importance of juvenile social networks to subsequent survival more directly than has been possible in past studies (Stanton and Mann, 2012), and more accurately determine whether prior behavior predicts current behavior. In addition, the removal of foals natal groups allowed us to ask the more general question: do the behaviors exhibited by young mammals confer immediate and/or future benefits? Specifically, does juvenile social behavior enhance current survival, future survival, or both? To our knowledge, no other study has been able to separate the effects of mother vs. infant social behavior in this way (Fagen, 1977; Bateson and Young, 1981; Lee, 1983; Martin and Caro, 1985; Gomendio, 1988). Here, we use social network analyses to measure foal sociality at different stages of development. Specifically, we use degree (the number of close associates to which an individual is connected) and weighted degree (the number of interactions in which an 6

8 individual engages) to investigate the effects of prior and current sociality on the ability of 4-7 month-old foals to survive after the gathering and removal of 40% of the horse population. In addition, we compare the effects of sociality on juvenile survival for foals left with at least one parent and some remnant of their natal groups versus those left without any social support. Methods Study area Shackleford Banks is a barrier island located at 34º N and 76º W, approximately 3 km off the coast of North Carolina, USA. The island is 15 km in length, and varies between 0.5 and 3 km in width. At the time of this study, feral horses had lived on the island for approximately 400 years (Conant et al., 2012). Study subjects We studied the feral horses of Shackleford Banks from 1995 through 1997 to determine the effects of variation in the mother-infant relationship on foal development. At the time of the study, the social groups of Shackleford horses were typical of feral equids. They consisted of coherent bands of usually one, but sometimes up to three stallion(s) with one to several mare(s) and their offspring (Rubenstein, 1981). Although the bands were predominantly non-territorial and the animals moved freely within overlapping home ranges, individual bands were spatially distinct from one another and individuals of particular bands rarely interacted (Feist and McCullough, 1976; Rubenstein, 1981; Rubenstein, 1986). Historically, males sometimes fought to acquire mares from other groups; band stallions almost always retained their mares, but changes to group 7

9 composition did sometimes occur, with reproductive females leaving their current group(s) to join new one(s) (Rubenstein, 1981; Rubenstein, 1986). These behaviors have proven typical in other feral horse populations (Feist and McCullough, 1976). At the outset of the study, approximately 183 animals (~19 individuals/km 2 ) occupied the island, consistent with population densities for the previous five years (Rubenstein and Nuñez, 2009). A cohort of 16 foals was observed in 1996 (see Table 1). The data presented here were collected during the first year of the foals development from Gathers and Removals In November, 1996, the National Park Service began managing the Shackleford Banks horse population. They gathered the majority of the animals (approximately 180 animals) and 76 individuals testing positive for equine infectious anemia were removed and euthanized off-site. Animals remaining on the island tested negative for the disease. All but one of the foals studied experienced some change to their bands social structure (see Table 1). Six foals (aged 6-7 months) were separated from their natal groups via the removal of their entire natal group (n=2 foals) or their permanent estrangement from any individuals remaining on the island (n=4 foals). Ten other foals (aged 4-7 months) remained on the island with at least one parent (nine with their mothers and one with its father genetic testing verified paternity in this case (Conant et al., 2012)) and some members of their natal groups. All but one of these foals had at least one individual removed from their natal groups (see Table 1). For one of these ten foals, logistical 8

10 issues prevented the collection of sufficient data after the gather. This foal was excluded from all analyses (see Table 1). Behavioral and demographic sampling This study was conducted by one observer (C.M.V. Nuñez) totaling over 232 hours of behavioral observation. Horses were identified individually by color, sex, age, physical condition, and other distinguishing markings. Ages were known from long-term records for the identified horses of Shackleford Banks (Rubenstein and Nuñez, 2009). On average, each individual was located (mean + SE) (range = ) times per week for 33 weeks pre-management and (range = ) times per week for 35 weeks post-management. Foal survival has been tracked to the present day (for a total of 18 years (S. Stuska, unpublished data)). However, as we were interested in determining foal survival of the gather specifically, foals were considered to have survived if they lived to one year of age. Additionally, one year is generally accepted as the age of independence for feral horses (Tyler, 1972; Cameron et al., 2000). During focal sampling, the mares and foals nearest neighbors were recorded at 5-minute intervals (Altmann, 1974). Mares and foals without neighbors within 15m were noted as having no near neighbor. We did not collect focal data on the near neighbor(s) of other group members, limiting our analyses to that of mare and foal degree and foal weighted degree only (Wey et al., 2008). Mare rank was determined via ad libitum sampling (Altmann, 1974) of agonistic interactions between mares and supplants given and received by mares; lower ranks indicated more dominant placement in the hierarchy. Stallion rank was similarly determined via ad libitum sampling of agonistic interactions 9

11 between stallions. Mare and stallion physical condition was assessed via rump scoring, examining the curvature of the line between the tailbone and the point of the hip. Scores were based on a scale from 1 to 5 (a score of 1 being the poorest (Pollock, 1980)) and were recorded monthly during the study. Group size was determined during each sampling period. Changes in group composition were rare, but did occur (see Study subjects). To account for such behavior, we calculated total group size as the total number of animals that lived with individual foals and therefore, to which foals had access during the study period. All sampling was conducted in accordance with the laws of the United States government and the National Research Council standards (National Research Council, 2011). Given the non-invasive nature of this study, neither the Princeton University Institutional Animal Care and Use Committee nor the National Park Service deemed permitting necessary. Statistics Social networks We calculated the number of associates (degree) and the number of interactions (weight) for each subject from their near neighbor data. The number of times each focal animal was observed affected the likelihood that we would detect changes and/or additions to that animal s social network (see Fig. 1 a-d) (Gotelli and Colwell, 2001; Lusseau, 2007). To control for this effect on foal degree, we calculated residuals from separate non-linear regressions of foal degree and the number of focal samples conducted on each individual both pre- and post-gather (see Fig. 1 a-b). To control for this effect on foal weight, we 10

12 calculated residuals from linear regressions of foal weight and the number of focal samples for each individual (see Fig. 1 c-d). We heretofore refer to these metrics as residual foal degree and residual foal weight, respectively. As residual weight proved a worse predictor of foal survival (see Table 2), all subsequent analyses focused on residual foal degree. To assess which social and ecological factors determined residual foal degree (pre -gather), we regressed pre-gather residual foal degree against condensed metrics of the foals social and abiotic environment from a Principal Components Analysis in R version (R Development Core Team, 2011). The PCA included the following factors: mother degree, mother residual rank (rank regressed against number of resident mares, to control for group size), mother physical condition, band stallion rank, band stallion physical condition, number of adult females in the group, total group size, and home range area (see Table 3). Non-normal distributions were log-transformed. The data determining these factors were collected pre-gather from April 1996 to mid-november, To assess whether early experience influenced future sociality, we performed a separate analysis to determine whether residual foal degree pre-gather predicted residual foal degree post-gather. Foal survival To determine how foal survival varied with foal sociality (pre- and post-gather) and parental/natal group access (post-gather), we performed survival analyses with parametric accelerated failure models assuming a Weibull distribution using the survreg function 11

13 in the survival package (R package version ) (Therneau, 2012), in R version (R Development Core Team, 2011). These models allow for the inclusion of continuous variables and the Weibull distribution makes fewer assumptions about the risk of death over time than do other distributions (Klein and Moeschberger, 2003). Using Akaike s Information Criterion adjusted for small sample size (AICc) (Burnham and Anderson, 2002), we compared models that incorporated various combinations of sociality variables pre- and post-gather, and parental/natal group access post-gather. Specifically, we constructed models using all combinations of residual foal degree preand post-gather, parental/natal group access, and two-way interactions between social network metrics and parental/natal group access. In addition, we ran analogous models substituting residual foal weight pre- and post-gather for residual foal degree (see Table 2). For the candidate model set outlined above, we calculated AICc weights of each model (Burnham and Anderson, 2002). Using these metrics, we calculated the relative importance of each explanatory variable (Burnham and Anderson, 2002) (see Appendix A). In addition, we calculated model-averaged parameter estimates for each explanatory variable, including 95% confidence intervals (Burnham and Anderson, 2002) (see Appendix A). To determine if foal sociality per se, and not social and ecological factors correlating with residual foal degree, influenced animal survival, we ran eight separate survival models. For this analysis, we replaced pre-gather residual foal degree with PC1 from the principal components analysis discussed above (also see Results). We used 12

14 AICc values to compare these alternative models. Given the limited sample size, we had minimal statistical power to detect the potential effects of other factors such as foal sex and age at the time of the gather on foal survival. The distribution of males and females was nearly equivalent: six foals were female (40%); nine were male (60%). At the time of the gather, all foals were between 4 and 7 months of age, see Table 1. Results Predictors of residual foal degree pre-gather Mother degree before the gather, the total number of adult females in the group, and total group size all loaded positively onto PC1 (see Table 3). Stallion condition, which was associated with both the number of females in the group (Linear Model: F 1, 13 = 6.03, r 2 = 0.32, P = 0.03) and total group size (Linear Model: F 1, 13 = 10.00, r 2 = 0.43, P = 0.007) also loaded positively onto PC1. Home range area, which tended to decrease with group size (Linear Model: F 1, 13 = 3.80, r 2 = 0.23, P = 0.07), loaded positively and more weakly onto PC1 (see Table 3). These results suggest that PC1 primarily captures a mare s (and possibly the stallion s) level of sociality. Residual mother rank and stallion rank loaded positively and mother condition loaded negatively onto PC2 (see Table 3), suggesting that PC2 captures parental quality. PCs 1 and 2 explained 66% of the variation in these variables. PC1 was predictive of residual foal degree pre-gather (Overall Linear Model: F 2, 13 = 9.39, r 2 = 0.53, P = 0.003; PC1: estimate = 1.31, t = 4.33, P = ); however, PC2 was not (PC2: estimate = -0.09, t = -0.23, P = 0.82). Neither PC1 nor PC2 predicted residual foal degree post-gather (Overall Linear Model: F 2, 12 = 1.88, r 2 = 0.11, P = 0.19; PC1: estimate = 0.41, t = -1.27, P = 0.23; PC2: estimate = -0.65, t = -1.52, P = 0.15). 13

15 Residual foal degree pre-gather did not predict residual foal degree post-gather (Linear Model: F 1, 13 = 1.80, r 2 = 0.12, estimate = 0.44, P = 0.20); this result seems to be largely driven by foals left alone after the gather (Linear Model, Foal parental/natal group access- no access: F 1, 4 = 0.39, r 2 = 0.09, estimate = 0.19, P = 0.57; Foal parental/natal group access- access: F 1, 7 = 4.81, r 2 = 0.41, estimate = 0.55, P = 0.06). Predictors of foal survival Our analyses suggest that three of the explanatory variables considered played the largest role in explaining variation in foal survival: residual foal degree pre-gather (relative importance = 0.27), residual foal degree post-gather (relative importance = 0.76), and parental/natal group access (relative importance = 0.38), see Table 4. The interaction between residual foal degree pre-gather and parental/natal group access had considerably less support than the variables listed above (relative importance = 0.06), but had more support than did the interaction between residual foal degree post-gather and parental/natal group access (relative importance = 0.02), see Table 4. The precision of the parameter estimates for the above variables suggests that higher residual foal degree post-gather is consistently related to longer survival (estimate = 0.44; 95% CI = ), see Table 4. The relationship between residual foal degree pre-gather and survival, however, appears to depend on the interaction with parental/natal group access. Specifically, among foals with no access to parents or natal group members, residual foal degree pre-gather was positively associated with survival (estimate = 0.36; 95% CI = ). Among foals with parental and natal group 14

16 access, residual foal degree pre-gather did not show a consistent relationship with survival (estimate = 0.15; 95% CI = ), see Table 4. The relative importance of residual foal weight was minimal when compared to that of residual foal degree (see Table 4). In addition, confidence intervals of modelaveraged parameter estimates for all models including residual foal weight overlapped zero, further indicating that residual foal weight did not significantly predict foal survival (see Table 4). To determine whether the number of foals close associates or a broader range of social and ecological factors were more important in directly affecting survival, we substituted PC1 for pre-gather residual foal degree in all survival models considering pre-gather residual foal degree (see above). Substituting PC1 yielded a worse fit to the data in 7 of 8 cases (see Table 5), suggesting that foal residual degree itself, and not its correlates, better explained foal survival. Discussion In many environments, maintaining individual fitness requires buffering the effects of catastrophic events. Here, we investigated whether the social networks of juvenile feral horses affected their ability to survive a catastrophic event. Our study is unique in that the removal of these foals associates, including their parents, allowed us to separate the effects of prior and current social environment on foal survival (Stanton and Mann, 2012). We demonstrate five important findings: 1) that foal degree was a better predictor of foal survival than was foal weight; 2) that following a catastrophic event, access to parent(s) and natal group members is important to juvenile survival; 3) that when 15

17 controlling for parental and natal group access, both pre- and post-gather sociality were important to foal survival; 4) that the influence of pre-gather sociality appeared most pronounced for foals left without access to parent(s) or natal group members, and; 5) that the influence of post-gather sociality was equally important to foals regardless of their access to parent(s) or natal group members. These results contribute to our understanding of the benefits afforded by sociality; they indicate that social animals not only benefit from the factors we typically think of (such as decreased predation risk, increased foraging efficiency, etc.), but also, that they are shielded from at least some of the potentially devastating effects of catastrophic events. For example, the foals more likely to survive the gather either 1) had access to at least one parent and some members of their original natal groups or 2) were left alone, but were able to form more associations and build sufficient social networks. Moreover, our results contribute to our understanding of whether sociality s benefits are conferred immediately and/or in the future. The fact that foals with parental and natal group access were more likely to survive is not particularly surprising. In feral horses, foals will often nurse from their mothers for one full year; longer if the mare does not conceive during post-partum estrus (Tyler, 1972). In addition, foals remaining with their fathers may benefit from consistencies in home range quality and from their fathers rank (Rubenstein and Nuñez, 2009). Finally, all animals with parental and natal group access likely also benefitted socially from the presence of these individuals (Bourjade et al., 2009; Shannon et al., 16

18 ). Conversely, foals denied such access suffered nutritional, hierarchical, and social losses during a critical period of development. More interesting is the fact that even when we controlled for parental and natal group access, increased foal sociality was associated with increased survival. Moreover, our relative importance analysis indicates that there were varying effects of pre- and postgather sociality on foal survival. Post-gather sociality seemed important for all foals, regardless of their access to parent(s) and natal group members; this may reflect the universal importance of sociality following a catastrophic event. Small sample size may have limited our power to detect an interaction between residual foal degree post-gather and parental/natal group access. However, the fact that all but one of the foals considered here experienced the loss of at least one group member (see Table 1) is noteworthy. Such losses have been shown to have significant impacts on social species (Shannon et al., 2013). It may be that the gather and subsequent removals were, at least to some extent, catastrophic for all foals, and that increased sociality became equally important for survival, even for foals with access to their parent(s) and natal group members. In addition, all of the subjects were gathered, held in one paddock, and were kept in close quarters with unfamiliar animals for a period lasting between 2 and 5 days (personal observation). Though the effects of these measures were likely short-lived, they were undoubtedly stress-inducing (personal observation). Conversely, our data suggest that pre-gather sociality affected the future survival probability for foals left alone after the gather, but not for foals with access to their parent(s) and natal group members. Although the relative importance for the interaction 17

19 between pre-gather residual degree and parental/natal group access was lower than the relative importance of several main-effect terms (see Table 4), relative importance partly depends upon the number of models that include a given variable (Burnham and Anderson, 2002). Interactions are included in fewer models than are main effects. Therefore, an interaction s relative importance will appear low when compared to the relative importance of main effects (see Table 4). However, when one considers that the relative importance for the interaction between pre-gather residual degree and parental/natal group access was three times greater than was the analogous interaction with post-gather residual degree, (see Table 4), the significance of the results is made clear; pre-gather sociality affected the future survival probability for foals left alone after the gather, but not for foals with access to their parent(s) and natal group members. Foals with parental and natal group access after the gather experienced a more typical upbringing. Although all but one of these animals experienced some social loss, the nutritional, hierarchical, and social support provided by parents and remaining group members clearly increased these foals chances of survival. The development of social skills in these foals was undoubtedly important; feral horses are a social species and developing such abilities is paramount to coexisting with other group members (Tyler, 1972). However, increased sociality early in development was not paramount to these foals subsequent survival. This is in contrast to foals left alone after the gather; for these foals, sociality pre-gather was associated with increased survival post-gather. These results demonstrate the importance of sociality in this species and indicate that sociality early in development may act as a buffer against future stressors. 18

20 In addition, our data may shed light on the process by which juvenile animals develop and maintain social skills. Foals that socialized with other individuals of varying sex and age early in development were likely better able to develop social skills (Bourjade et al., 2008; Bourjade et al., 2009). Later in development, these same foals may have been more able to form new relationships, outside of their natal groups. The fact that foal residual degree pre-gather did not predict foal residual degree post-gather does not support this hypothesis. However, foals left alone seem to drive this result, indicating that the massive social shifts occurring during and after the gather may have dampened any correlation between associations pre- and post-gather. Regardless, our results suggest that in social species, increased sociality early in development can provide an important safeguard against future stressors. The nature of our data did not allow us to investigate the potential effects of more complex social measures on foal survival. However, our research highlights the importance of degree, conceptually the simplest measure of animal sociality (Wey et al., 2008), to the survival of young animals. Other studies have found the strength or quality of associations rather than their number to be important to animal fitness and/or to factors affecting fitness (Crockford et al., 2008; Wittig et al., 2008; Barocas et al., 2011; Stanton and Mann, 2012). Here, we found the opposite: that while residual foal degree proved crucial to foal survival after a catastrophic event, residual foal weight did not. To our knowledge, no other study has been able to test the importance of sociality to offspring survival in the manner that we have done here. It may be that, in catastrophic conditions, 19

21 the most basic metrics of sociality become more important; that is, the more associations you have, the better are your chances. Whether the behaviors exhibited during the juvenile period provide immediate or long-term benefits has been a point of debate (Bateson and Young, 1981; Fagen, 1981; Lee, 1983; Martin and Caro, 1985; Gomendio, 1988). The fact that increased sociality post-gather was important for all foals in this study, regardless of their access to parent(s) and natal group members (Fig. 2 b), is consistent with the notion that juvenile behavior confers immediate benefits. On the other hand, the fact that increased sociality early in development was more predictive the survival for foals left alone, than for foals with access to parent(s) and natal group members (Fig. 2 a), is consistent with the notion that juvenile behavior confers future benefits. Taken together, these data suggest that although early behavior can affect future survival, appropriate responses to current circumstances (enabled by increased behavioral flexibility and adaptability) are also crucial (Byers, 1998). Finally, our results show that although the social and ecological environment (i.e., home range area, mare sociality, number of adult females in the group, and total group size) may affect the level of sociality developed early in life, it is this sociality, not the social and ecological environment, that more strongly influences future survival probability (see Table 5). Though the initial social and ecological environment affects the potential for increased foal sociality, it does not in and itself guarantee the realization of this potential. Less gregarious individuals will typically form fewer associations regardless of their social and ecological circumstances (Cameron et al., 2009; Silk et al., 20

22 ). In the present study, mare (and possibly stallion) sociality and band ecology may help shape foal experience and the potential for foal associations early in development; however, this initial environment is not equivalent to foal sociality and therefore, does not affect the likelihood that foals will survive a catastrophic event (see Table 5). For several species, early social behavior can affect the subsequent social competence of individuals, and it has been suggested that an environment rich in social interaction may provide benefits to animals (Ballen et al., 2014). Here, we show that in feral horses, foals with more associates were more likely to survive a catastrophic event. But by which mechanism? The lessons foals learned from these associates may have been critical to their subsequent survival. Such social learning has been shown to confer benefits in several taxa including insects, (Chittka and Leadbeater, 2005; Coolen et al., 2005), fish (Dugatkin, 1992; Brown, 2003), reptiles (Ballen et al., 2014), birds, (Galef Jr. and White, 1998; Freeberg, 2000; Hunt and Gray, 2003), and mammals (Cook and Mineka, 1989; Byrne and Byrne, 1993; Nicol and Pope, 1994; Suárez and Kravetz, 1998; Whiten et al., 1999). Social learning helps species to attain and manipulate food, find appropriate mates, avoid predators, and use tools. In addition, social learning can help young animals form associations with new individuals and can teach them appropriate social behavior. In domestic horses, for example, the introduction of adults to juvenile groups results in the expression of new behaviors, the emergence of preferential social associations, increases in positive social behavior (including affiliative behaviors and social investigation), and decreases in agonistic interactions by the young animals (Bourjade et al., 2008). In addition, research with Przewalski s horses (Equus f. 21

23 przewalskii) demonstrates that in groups with lower adult-young ratios, juveniles tend to associate more strongly with animals of similar age, largely avoiding older, more experienced individuals (Bourjade et al., 2009). Juveniles in these groups exhibit aggression rates up to four times higher than those in groups containing more adults (Bourjade et al., 2009). These data indicate that the presence of adults may provide more opportunities for social learning and that more varied social groups are beneficial for young animals. Our data take these results one step further and indicate that the social ties and lessons learned from associates may actually facilitate offspring survival. Understanding the importance of sociality to the survival of species is important if we are to effectively conserve and minimize human impacts to social species. Social network analysis with orca whales, for example, shows that all individuals are not created equal when it comes to maintaining social cohesion. Certain whale matrilines are more central than others and juvenile whales play an important role in maintaining network cohesion (Williams and Lusseau, 2006). Simulated removals of individuals demonstrate that the social network is more likely to fragment under targeted removals (mimicking historic live-captures from the northeastern Pacific) than under random removals. With a fuller understanding of the benefits afforded by animals social networks (and the individuals within them), we can potentially minimize management practices that disrupt the social environment; this may be particularly important in regions frequently subject to catastrophic events such as coasts, high altitudes, and drought-prone areas, as these events may already affect species social environment on a relatively large scale (Shaffer, 1981). 22

24 Future studies should continue to examine the details behind sociality s effect on juvenile survival, particularly in species subject to catastrophic events. Through such studies, we will undoubtedly learn more about the specific benefits sociality affords young animals. Only when we determine the true nature of these benefits will we fully understand its contribution to animal fitness. Acknowledgements We want to thank S. C. Moyers for her insightful comments on an earlier draft of this manuscript and Dr. D. M. Hawley for the use of her facilities. In addition, we thank Drs. S.R. Sundaresan and I.R. Fischhoff for their statistics advice. This study was funded by Princeton University and the National Science Foundation (IIS to D. I. Rubenstein). References Alexander RD, The evolution of social behaviour. Annual Review of Ecology and Systematics 5: Altmann J, Observational study of behavior: Sampling methods. Behaviour 49: Ballen C, Shine R, Olsson M, Effects of early social isolation on the behaviour and performance of juvenile lizards, Chamaeleo calyptratus. Animal Behaviour 88:1-6. Barocas A, Ilany A, Koren L, Kam M, Geffen E, Variance in centrality within rock hyrax social networks predicts adult longevity. PLoS ONE 6:e

25 Bateson P, Young M, Separation from the mother and the development of play in cats. Animal Behaviour 29: Bourjade M, des Roches AdB, Hausberger M, Adult-young ratio, a major factor regulating social behaviour of young: A horse study. PLoS ONE 4:e4888. Bourjade M, Moulinot M, Henry S, Richard-Yris M-A, Hausberger M, Could adults be used to improve social skills of young horses, Equus caballus? Developmental Psychobiology 50: Brown GE, Learning about danger: Chemical alarm cues and local risk assessment in prey fishes. Fish and Fisheries 4: Burnham KP, Anderson DR, Model selection and multimodel inference: a practical information-theoretic approach, Second ed. New York, NY, USA: Springer- Verlag. Byers JA, Animal play: evolutionary, comparative and ecological perspectives. Cambridge: Cambridge University Press. Byrne RW, Byrne JME, Complex leaf-gathering skills of mountain gorillas (Gorilla g. beringei): Variability and standardization. American Journal of Primatology 31: Cameron EZ, Linklater WL, Stafford KJ, Minot EO, Aging and improving reproductive success in horses: declining residual reproductive value or just older and wiser? Behavioral Ecology and Sociobiology 47:

26 Cameron EZ, Setsaas TH, Linklater WL, Social bonds between unrelated females increase reproductive success in feral horses. Proceedings of the National Academy of Sciences of the United States of America 106: Chittka L, Leadbeater E, Social Learning: Public Information in Insects. Current Biology 15:R869-R871. Conant E, Juras R, Cothran E, A microsatellite analysis of five Colonial Spanish horse populations of the southeastern United States. Animal Genetics 43: Cook M, Mineka S, Observational conditioning of fear to fear-relevant versus fearirrelevant stimuli in rhesus monkeys. In. US: American Psychological Association; Coolen I, Dangles O, Casas J, Social learning in noncolonial insects? Current Biology 15: Creel S, Creel NM, Communal hunting and pack size in African wild dogs, Lycaon pictus. Animal Behaviour 50: Crockford C, Wittig RM, Whitten PL, Seyfarth RM, Cheney DL, Social stressors and coping mechanisms in wild female baboons (Papio hamadryas ursinus). Hormones and Behavior 53: Dugatkin LA, Sexual selection and imitation: Females copy the mate choice of others. The American Naturalist 139: Fagen RM, Selection for optimal age-dependent schedules of play-behavior. American Naturalist 111: Fagen RM, Animal play behavior. New York, NY: Oxford University Press 25

27 Feist JD, McCullough DR, Behavior patterns and communication in feral horses. Zeitschrift fu r Tierpsychologie 41: Franco MMJ, Paes AC, Equine infectious anemia. Veterinaria e Zootecnia 18: Freeberg TM, Culture and courtship in vertebrates: A review of social learning and transmission of courtship systems and mating patterns. Behavioural Processes 51: Galef Jr. BG, White DJ, Mate-choice copying in Japanese quail, Coturnix coturnix japonica. Animal Behaviour 55: Gomendio M, The development of different types of play in gazelles: implications for the nature and functions of play. Animal Behaviour 36: Gotelli NJ, Colwell RK, Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4: Hunt GR, Gray RD, Diversification and cumulative evolution in New Caledonian crow tool manufacture. Proceedings of the Royal Society B-Biological Sciences 270: Klein JP, Moeschberger ML, Survival Analysis: Techniques for Censored and Truncated Data. New York, NY: Springer. Lee P, Play as a means for developing relationships. Primate social relationships: Lima SL, Bednekoff PA, Back to the basics of antipredatory vigilance: Can nonvigilant animals detect attack? Animal Behaviour 58:

28 Lusseau D, The emergent properties of a dolphin social network. Proc Biol Sci 7:S Lusseau D, Evidence for social role in a dolphin social network. Evolutionary Ecology 21: Martin P, Caro TM, On the functions of play and its role in behavioral development. Advances in the Study of Behavior 15: National Research Council, Guide for the Care and Use of Laboratory Animals: Eighth Edition. Washington, D.C.: The National Academies Press. Nicol CJ, Pope SJ, Social learning in sibling pigs. Applied Animal Behaviour Science 40: Pollock J, Behavioural ecology and body condition changes in New Forest ponies. Scientific Publications of the RSPCA 6: Powell GVN, Experimental analysis of the social value of flocking by starlings (Sturnus vulgaris) in relation to predation and foraging. Animal Behaviour 22: Pusey AE, Packer C, Non-offspring nursing in social carnivores-minimizing the costs. Behavioral Ecology 5: R Development Core Team, R: A Language and Environment for Statistical Computing. In, ed. Vienna, Austria: R Foundation for Statistical Computing. 27

29 Rubenstein D, Sundaresan S, Fischhoff I, Saltz D, Social networks in wild asses: Comparing patterns and processes among populations. Exploration into the biological resources of Mongolia 10: Rubenstein DI, Behavioural ecology of island feral horses. Equine Veterinary Journal 13: Rubenstein DI, Ecology and sociality in horses and zebras. In: Ecological Aspects of Social Evolution, Birds and Mammals (Rubenstein DI, Wrangham, R.W., ed). Princeton: Princeton University Press; Rubenstein DI, Nuñez CMV, Sociality and reproductive skew in horses and zebras. In: Reproductive Skew in Vertebrates: Proximate and Ultimate Causes (Hager R, Jones CB, eds). Cambridge: Cambridge University Press; Shaffer ML, Minimum population sizes for species conservation. Bioscience 31: Shannon G, Slotow R, Durant S, Sayialel K, Poole J, Moss C, McComb K, Effects of social disruption in elephants persist decades after culling. Frontiers in Zoology 10:62. Silk JB, Beehner JC, Bergman TJ, Crockford C, Engh AL, Moscovice LR, Wittig RM, Seyfarth RM, Cheney DL, The benefits of social capital: Close social bonds among female baboons enhance offspring survival. Proceedings of the Royal Society B: Biological Sciences 276: Stanton MA, Mann J, Early social networks predict survival in wild bottlenose dolphins. PLoS ONE 7:e

30 Suárez OV, Kravetz FO, Transmission of food selectivity from mothers to offspring in Akodon azarae (Rodentia, Muridae). Behaviour 135: Sundaresan S, Fischhoff I, Dushoff J, Rubenstein D, Network metrics reveal differences in social organization between two fission fusion species, Grevy s zebra and onager. Oecologia 151: Templeton CN, Reed VA, Campbell SE, Beecher MD, Spatial movements and social networks in juvenile male song sparrows. Behavioral Ecology 23: Therneau T, A Package for Survival Analysis in S. R package version Tyler SJ, The behaviour and social organization of the New Forest ponies. Animal Behaviour Monographs 5: Wey T, Blumstein DT, Shen W, Jordan F, Social network analysis of animal behaviour: A promising tool for the study of sociality. Animal Behaviour 75: White DJ, Galef Jr BG, Differences between the sexes in direction and duration of response to seeing a potential sex partner mate with another. Animal Behaviour 59: Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, Wrangham RW, Boesch C, Cultures in chimpanzees. Nature 399: Williams R, Lusseau D, A killer whale social network is vulnerable to targeted removals. Biology Letters 2: