Abstract
Some animal societies, particularly in primate and corvid species, are based on enduring social bonds, based on sophisticated cognitive skills. A key unresolved issue for understanding the evolution of complex sociality and the associated advanced cognition is to uncover the fitness advantages that social relationships convey to individuals. Here, we investigated whether the social environment and the quality of individuals’ social relationships in captive groups of carrion crows affect excretion patterns of coccidian oocysts, which have been shown to exert detrimental consequences on bird health. In females, variables relating to the social environment and the quality of individuals’ social relationships did not have a strong influence on the occurrence of coccidian oocysts. However, male crows with strong social bonds and with a large number of relatives in their group were less likely to excrete feces containing coccidian oocysts. Thus, our data present evidence that high quality social relationships help male carrion crows to keep endo-parasite burden low, suggesting that the social life might have potential fitness advantages by reducing parasite burden. Ultimately, these advantages might be important in driving the evolution of complex group living.
Significance statement Uncovering the fitness benefits of close social bonds is a key factor in understanding why close relationships between individuals have evolved in many distantly related taxa, such as mammals and birds. Here, we investigated potential health benefits of close social bonds in carrion crows. Carrion crows live in complex male-biased social groups and possess high cognitive skills that allow managing social relationships. Male crows with strong social bonds and with a large number of relatives in their group were less likely to excrete feces containing coccidian oocysts. Our results show that social bonds might have direct fitness implications in male crows, which ultimately might drive the evolution of complex group living.
Introduction
In many animals, the benefits of conspecific aggregations (e.g. protection against predators, increased access to food through shared defense or knowledge of resource locations) can arise in unstructured assemblages, where anonymous individuals show minimal interactions that they can manage with a relatively small brain and unsophisticated cognitive abilities (e.g. fish schools, ungulate grazing herds). However, some animal societies, notably those in some primate species, are based on enduring social bonds and according to the ‘social brain hypothesis’, the management of these complex relationships requires sophisticated cognition [1].
Currently, the key issue for understanding the emergence of this kind of complex sociality and the associated cognitive skills is to uncover the fitness advantages that bonding conveys to individuals [2]. In spite of much interest, this question is largely unresolved. Research in primates has shown that the quality of social relationships may affect individuals’ longevity and reproductive success. The largest body of evidence supporting this hypothesis has been found in humans, where a rich social network and social support yields measurable positive consequences for health [3]. Other primate species show similar patterns. Social relationships help female chacma baboons (Papio hamadryas ursinus) cope with stressful events and increase longevity [4], social bonds in male Assamese macaques (Macaca assamensis) are directly linked to the number of offspring sired [5], and in Barbary macaques strong social bonds have a stress reducing effect [6]. These studies, together with more limited information on other mammalian species [1, 7], suggest that the human tendency to engage in enduring social bonds may have deep evolutionary roots [8]. To test this fascinating hypothesis, however, we crucially need to investigate the fitness consequences of social bonds in a larger variety of taxa and types of social organizations [2].
Recent research has shown that advanced sociality associated with sophisticated cognition is not unique to humans and closely related primate species, but can be found in a variety of taxa [9, 10]. Among birds, corvids show complex social relationships, proving a suitable model to investigate the fitness consequences of social bonding. Common ravens (Corvus corax), for example, are capable of forming valuable relationships, not only within reproductive pairs but also between short/long-term coalition and alliance partners in their social group [11, 12]. Such bonds are characterized by low levels of aggressive and high levels of affiliative behaviors exchanged in a reciprocal way over extended periods of time [13]. Within such valuable relationships, individuals support each other in agonistic encounters [12, 14] and share information and resources [12, 15, 16], whereas competition dominates in other social contexts [17, 18].
Carrion crows (Corvus corone) are also capable of developing enduring non-reproductive social bonds, especially among males, which in some populations can share and defend a territory, and cooperate in rearing a communal brood [19, 20]. Within-groups, linear dominance hierarchies help reduce conflicts and keep the group cohesive [21, 22]. As expected, cognitive skills mirror the social complexity in these species. For example, crows can discriminate between reliable and unreliable conspecifics [23] and show inequity aversion, being able to quantify other individuals’ efforts and payoff in comparison with their own in token exchange tasks [24]. Carrion crows are therefore a suitable model to investigate the fitness correlates of social bonding and, ultimately, the evolution of the social brain [25].
In this study, we investigated whether the social environment and the quality of an individual’s social relationships affects health in captive groups of carrion crows that varied in size and composition. In particular, we focused on the pattern of excretion of coccidian oocysts, as a proxy of health in carrion crows. Coccidian protozoans are among the most common endoparasites of birds [26, 27] and are widespread among corvids, including the carrion crows [28–30]; this study, see below. Infections with large numbers of coccidia clinically manifest as ‘coccidiosis’, causing infections in the intestinal tract [31]. This severely affects individual body condition, longevity and fecundity, by inhibiting the uptake of essential dietary components [32, 33]. The pathogenicity is well documented in poultry [34–36], and evidence for it is further accumulating in wild bird species (e.g. [37, 38]). Coccidia are transmitted via the fecal- oral route. Coccidian oocysts excreted in the feces of host individuals become infectious by undergoing sporogony within 24 hours [35].
In this study, we address three non-mutually exclusive hypotheses: 1. Group size hypothesis: parasite excretion patterns depend on the size of the group and/or the number of related individuals. We expect infections to spread more easily in large groups, although the presence of relatives in the group may counteract the effect of crowding by reducing the occurrence of stressful conflicts and competition [39] that can affect vulnerability to parasite infection. 2. Social bonds hypothesis: parasite excretion patterns depend on the quality of the social relationships that an individual maintains, with individuals with strong social bonds being less likely to be infected than individuals with weak social bonds. 3. Intrinsic factors: parasite excretion rates are explained by an individual’s sex and age. We did expect sex differences, because males are more competitive than females within their groups [21] and are probably more exposed to the negative consequences of stress, which is expected to impair their ability to fight against parasites [40]. Also, previous results in wild and captive carrion crows suggest that males are more sensitive to the presence/absence of relatives [41] and therefore are expected to be more likely to show a health response to the composition of their social group compared to females. Age must also be controlled for, as it could affect parasite infection rate, although no specific predictions can be formulated based on the current knowledge of the host species.
Methods
Study subjects and ethics statement
We collected data over four different periods, from November 2008 to January 2009, from January to February 2013, from May to June 2013 and from January to February 2014, from a total of 34 crows, which were housed in a large outdoor aviary (30 x 12 x 6 m) in Northern Spain (Navafria, León, 42°36’33 N 5°26’56 W). The aviary comprised four major compartments (12 x 6 x 3 m), connected through eight smaller testing compartments (3 x 3 x 3 m). The major compartments were equipped with wooden perches, natural vegetation and rocks. An enriched diet consisting of fruit, vegetables, bread, meat and milk products was provided on a daily basis. Water and dry food were available ad libitum. Keeping crows in captivity was authorized by Junta de Castilla y León (núcleo zoológico 005074).
Birds were kept in groups that mirrored the social aggregations that naturally occur in the wild: (A) ‘Flock’: three or more juvenile individuals; in 2008 some juveniles were initially kept in ‘sibling pairs’ and subsequently merged into a juvenile flock; (B) ‘Pairs’: adult male and adult female; in some cases adults were also kept in trios; (C) ‘Cooperative Family’: a reproductive pair with its own already independent offspring. Group composition, however, varied due to the death of some individuals and reproduction of breeding pairs, as summarized in Table S1.
Behavioral protocol data
A total of 1075 individual focal protocols have been recorded. Each focal protocol lasted 5 minutes, and all occurring behaviors where recorded. However, for the present study, we focused on frequencies of agonistic behavior (threat, chase flight, and fight) and affiliative behaviors (approach, allopreening and sitting in contact). The identity of the interacting individuals was noted, as well as their role (initiator/receiver). All behavioral protocols (see sample size in Table S1) were recorded on video and analyzed by CW.
Composite sociality index
For each period of data recording, we calculated a composite sociality index (CSI) for each crow dyad within a group according to Silk, et al. 2010 [4]. A CSI was constructed for five affiliative behaviors as follows: (AIij/AIave + ARij/ARave + CSij/CSave + APIij/APIave + APRij/APRave)/5. AIij represents the mean number of approaches initiated by individual i towards individual j per focal protocol, and AIave is the mean number of approaches initiated by all members of the group per social protocol. Similarly, AR refers to approaches received, CS to contact sit, API to allopreening initiated and APR to allopreening received. Note that we used mean values instead of absolute numbers because the sample of focal protocols varied among individuals (see Table S1). A second CSI for four agonistic behaviors (aggression initiated won, aggression initiated lost, aggression received won, aggression received lost) was calculated as follows: (AIWij/AIWave + AILij/AILave + ARWij/ARWave + ARLij/ARLave)/4. For each period of data collection, we determined the average and maximum affiliative CSI for each dyad. Further, over the entire study period and all observed dyads, we calculated a mean affiliative and aggressive CSI. When an individual fulfilled both conditions of having a higher than average affiliative CSI and lower than average aggressive CSI with another individual, they were considered closely bonded. In each period of data taking we determined the number of group members to which each individual was closely bonded.
Parasitological examination
During the entire study period, 399 individual fecal samples were collected, directly after defecation (for a detailed overview see Table S1). Further, from April 2013 to October 2014, 178 fecal samples were collected from 30 different groups belonging to a long term studied population of wild carrion crows [42] in the rural area ‘La Sobarriba’, Northern Spain (42°37’ N, 5°26’ W). For the purpose of this study, after observing a crow group for up to 30 min, we approached the area where the crows were foraging and looked for fresh fecal samples on the ground. We could assign the samples to a given group but not to individual crows. Sex and age of the wild crows were unknown.
Eggs and oocysts of intestinal parasites were searched in fecal samples. In 2008, we used a modified version of the flotation method [43] to examine the occurrence/abundance of parasite products in the fecal samples. Fresh feces (0.1g) were suspended in a 2 ml collecting tube with 1 ml saturated saline. Collection tubes were shaken for 10 seconds and afterwards centrifuged for 5 minutes at 3000 rpm. After centrifugation, the collection tubes were filled with saline solution and a cover slip (18 x 18 mm) was positioned onto the tube. The high density of the saline solution causes the parasite eggs and oocysts to float up and be caught on the cover slip [44]. After 10 minutes, the cover slip was moved onto an object slide and the parasite eggs and oocysts were identified (by size and shape) and counted. From January 2013 onwards, we used a McMaster counting chamber to examine the occurrence of parasite products in the fecal samples. The entire fecal sample was weighed, then diluted with 3 ml saturated NaCl solution per 0.1 g of feces and thoroughly mixed. Afterwards, the solution was filled into both McMaster counting chambers. After 10 minutes of resting period we counted parasite eggs and oocysts in both chambers and calculated the number of parasite products per 1 ml of feces.
We used a binocular with 100-fold and 400-fold amplification to identify parasite products. We found Coccidian oocysts, eggs of several nematode species (Capillaria sp., Ascarida sp., Syngamus sp. and Heterakis sp., Trichostrongylus tenius) and cestode eggs to a varying degree. The proportion of positive samples was highest for coccidian oocysts (24.7 %, n = 336), whereas the prevalence of eggs of nematode and cestode species ranged between 1 and 17 %. We therefore restricted our analyses to coccidia, the frequency of which allowed for robust statistical testing. Coccidia are a diverse group of protozoan parasites which frequently infect the avian intestinal tract, causing lesions [45]. Because the abundance of parasite cocciadian oocysts in the feces depends on the stadium of the parasite cycle, which is difficult to control, we analyzed presence/absence instead of absolute numbers.
Data analysis
Parasites with different life cycles and virulence may exert diverse and synergetic effects on birds [43]. Exploring the causes and consequences of multiple parasitism exceeded the aims of this study, which focused on the most common infection found in crows at the study place, i.e. coccidia, and that represents a threat for many species of wild birds [46, 32, 47]. Thus, to reduce the environmental noise associated with the presence of multiple parasites, we restricted our analyses to samples infected only by coccidia, i.e. with no indications of other parasitic taxa (nematode eggs, cestodes). Likewise, we considered control (‘healthy’) samples those that were found free from any of the parasitic taxa described above.
We used generalized linear mixed models (GLMMs) with binomial error distributions and logit link functions in the package lme4 [48] to examine the variation in the occurrence of coccidian oocysts (presence of oocyst in the feces = 1, absence = 0). We used an information-theoretic approach to assess which variables and models best explained the variation in the data [49]. The information-theoretic approach recognizes the uncertainty associated with biological data, and allows calculating model-averaged parameter estimates and standard errors by considering non-exclusive candidate models. Such analytical approach has been shown to be particularly suitable for the comparison of alternative hypotheses – i.e. competing models – when multiple variables may depict the data [50, 51], which is the case in our study. We built up and compared models composed of explanatory variables associated with the three sets of hypotheses that may influence the occurrence of coccidia infection as described in the introduction section. We considered three age categories (first-year, second-year and adult) and assessed the differences by Tukey post-hoc testing in the multcomp package [52]. We conducted separate analyses for each sex because of different expectations regarding the influence of social factors on the occurrence of coccidian oocysts between males and females (see introduction and supplementary materials). Apart from sex, we did not include highly correlated variables (see supplementary materials) simultaneously in the same model to avoid potential spurious results due to multicollinearity. For each model, we fitted individual identity as a random term to control for the potential dependence associated with multiple samples from the same individuals. Given the low number of individuals per group for each sex (19 males in 14 groups, 15 females in 12 groups), we did not fit group identity as a random term in the models to avoid redundancy. Overall, we ran two groups of models (16 and 11 models for males and females, respectively) that included combinations of low-correlated explanatory variables as well as the intercept-only models, i.e. the null models. We ranked the models for each candidate group of models according to Akaike’s Information Criterion corrected for small sample sizes (AICc) and Akaike model weights [49, 53] by using the MuMIn package [54]. Models with lower AICc values and higher Akaike model weights are better supported by the data, with the sum of all model weights being one [49]. We calculated model-averaged parameter estimates and standard errors for each parameter by considering all candidate models within each group of models. Statistical analyses were performed in R 3.2.2 [55].
Results
Social bonds and parasitological examination
We observed 57 bonded dyads, 31 of which were male-female dyads (19 between related individuals and 12 between unrelated individuals), 18 were male-male (5 between related and 13 between unrelated individuals) and 8 female-female (6 related, 2 unrelated). On average, males and females had 1.7 ± 1.71 and 1.64 ± 1.60 bonded partners respectively. The mean CSI of related male-female bonds was 2.135 ± 0.846 and 3.034 ± 2.165 for unrelated male-female bonds. The mean CSI of male-male bonds was 2.345 ± 0.952 for related and 2.754 ± 1.798 for unrelated individuals. In female-female bonds, the mean CSI was 4.806 ± 1.549 and 1.556 ± 0.101 for related and unrelated partners respectively.
Overall, 83 out of 336 samples contained coccidian oocysts, whereas 253 samples were not infected (i.e., free of parasite products of any of the considered taxa). Coccidian oocysts occurred in 57 of 198 samples in males, and in 26 of 138 samples in females. In the droppings collected from wild carrion crows, 84 out of 178 samples contained coccidian oocysts, showing that coccidia are a common parasite in groups of wild carrion crows.
(a) Occurrence of coccidian oocysts in males
The selection of models that examined the occurrence of coccidian oocysts in males yielded a high-ranking model that accounted for a weight of 0.41 (Table 1). The maximum CSI value and the number of related individuals in the group were included in the best model. Crows with a strong social partner and living with a larger number of related individuals were less likely to contain coccidia (Table 2, Figures 1a and 1b).
The addition of the factor group size to the model with the maximum CSI value and the number of related individuals did not improve model performance, as indicated by the higher AICc value and the similar log-likelihood value of the second-ranking model compared to the best model (Table 1). Further indication of the low influence of group size on the occurrence of oocysts arises from the fact that the 95% confidence interval of the model-averaged parameter estimate for group size included zero (- 0.187, + 0.433; see Table 2).
Model averaging results also suggested a tendency for males with fewer affiliated individuals to be more likely to host coccidia compared to males with more affiliated individuals (Table 2). However, the number of affiliated individuals was fitted in models with rather low weights (0.01 - 0.06, Table 1), which suggests a low influence of this variable. Likewise, the average CSI value was fitted in low-ranking models (Table 1) and its model-averaged parameter estimate clearly included zero (Table 2), suggesting low explanatory power.
The third-ranking model included age as a fixed term (Tables 1 and 2). Post-hoc analyses by Tukey contrasts showed that first-year males were less likely to host coccidia than second-year birds (second-year – first-year: estimate ± SE = 2.44 ± 0.89, z = 2.733, p = 0.016) and adults (first-year – adult: estimate ± SE = - 2.62 ± 0.90, z = - 2.896, p = 0.010), whereas no difference was found between the latter two age categories (second-year – adults: estimate ± SE = -0.17 ± 0.55, z = -0.312, p = 0.946). Although age was highly correlated with the maximum CSI value and the number of related individuals (see Methods section), the addition of these two variables significantly improved the performance of the model based on age (Anova test: AICc = 203.14 vs 208.08 respectively for the model with the three variables and the model with the age only, Chi-squared = 9.172, df = 2, p = 0.010). Thus, both the age, the maximum CSI value, and the number of related individuals had a strong influence on the occurrence of coccidian oocysts in males.
The negative relationship between the maximum CSI value and the occurrence of oocysts may be difficult to interpret. On the one hand, it may indicate that birds with a highly affiliated group member are less likely to excrete fecal samples containing coccidian oocysts, but, on the other hand, it may also suggest that only healthy crows are able to maintain strong social bonds. To disentangle the two possibilities, we ran two additional analyses. First, we fitted a binomial GLMM with the maximum CSI value at a given day t as a predictor of the occurrence of coccidian oocysts later in time (t + n days), including individual identity as a random term and two additional explanatory terms, namely the occurrence of parasites at time t and the time elapsed between measuring maximum CSI values and assessing the occurrence of oocysts (n days). Neither of the two latter variables were significant (occurrence of parasites at time t: estimate ± SE = 0.611 ± 0.520, z = 1.175, p = 0.240; number of days elapsed: estimate ± SE = 0.001 ± 0.004, z = 0.320, p = 0.749), whereas crows with higher maximum CSI values proved less likely to host coccidia later in life (estimate ± SE = - 0.191± 0.077, z = -2.477, p = 0.013). This significant relationship supports a causal effect of maximum CSI value on the occurrence of the parasites. In a second analysis, we addressed the possibility that infected birds were less able to maintain strong social bonds. In this case, we expected that the presence of oocysts in the fecal samples of a given individual at time t would predict a low maximum CSI value later in life (t + n days). However, a linear mixed-effects model, where individual identity was fitted as random term, showed that the presence of coccidia at time t did not significantly reduce the maximum CSI value at time t + n days (estimate ± SE = -0.104± 0.126, t = -0.834, p = 0.406), after controlling for the maximum CSI value at t (estimate ± SE = 0.981 ± 0.018, t = 55.038,p < 0.001) and the number of days elapsed between assessing the occurrence of oocysts and measuring maximum CSI values (estimate ± SE = -0.005 ± 0.001, t = -4.456, p < 0.001). The results of this second model therefore also supports the conclusion that a strong social bond enhanced resistance against coccidian endoparasites.
(b) Occurrence of coccidian oocysts in females
The selection of models that examined the occurrence of coccidian oocysts in females yielded five high-ranking models that accounted for more model weight than the intercept-only model (Table 1). The best model included age as a fixed term and was approximately 10 times more likely to be the best model than the null model (i.e., model weights: 0.49 vs 0.05). Post-hoc analysis by Tukey contrasts showed that first- year females were less likely to host coccidia than adult females (first-year – adult: estimate ± SE = - 1.37 ± 0.50, z = -2.729, p = 0.017), whereas no difference was found between second-year and adults (second-year – adult: estimate ± SE = - 1.27 ± 0.69, z = -1.847, p = 0.151), nor between second-year and first-year (second-year – first-year: estimate ± SE = 0.11 ± 0.73, z = 0.143, p = 0.989).
The second-ranking model included group size as an explanatory variable (Table 1). Despite model-averaging analyses suggesting that females in smaller groups tended to be less likely to host coccidia (Table 2), this model had a ΔAICc value of only 1.68 compared to the null model. Moreover, the second-ranking model was only around two times as likely to be the best model as compared to the null model (model weights: 0.11 vs 0.06), suggesting a limited influence of group size on coccidian infection in female crows.
The average and maximum CSI values, as well as the number of affiliated individuals and the number of relatives, were fitted in rather low-ranking models (ΔAICc to the null model < 0.75, Table 1). Examinations of model-averaging parameter estimates confirmed a low influence of these explanatory variables on the occurrence of coccidian oocysts in females (Table 2).
Discussion
In the present study, we have shown that carrion crow males with closely bonded social partners (high CSI scores) proved less likely to excrete coccidian oocysts. The fact that high CSI values predicted low occurrence of coccidia at a later point in time and that, conversely, the presence of oocysts in the feces did not predict the strength of future social bonds, confirms the causal link between the quality of social relationships and reduced occurrence of coccidian oocysts, which can be considered an estimate of health [32, 33].
Health correlates of high quality social ties have been reported in humans (e.g. [56]) and a large body of research has focused on the causal psychological and biological processes that underlie this strong link [3, 57]. Similar evidence has been recently found in non-human primates [4, 5], suggesting that the tendency to form strong social ties may have deep phylogenetic roots [58]. Furthermore, as Silk [2] pointed out, it is important to recognize that social complexity is not limited to primate species. Indeed, coalition formation has been reported in several mammalian and bird species, e.g. African elephants Loxodonta africanus [59], spotted hyenas Crocutta crocutta [60], bottlenosed dolphins Tursiops aduncus [61], ravens [62], and carrion crows [20]. Our study uncovered for the first time, potential health consequences of high quality social relationships in a bird species, showing similarities with primates and suggesting that the direct fitness benefits of strong social bonds may be a key factor that triggers advanced sociality in a variety of taxa.
Cooperatively breeding groups of carrion crows form through two different mechanisms, namely delayed dispersal of offspring, which remain in the natal territory with their parents and siblings [63], and formation of long lasting social bonds among distant relatives, most often males, that share all purpose territories and frequently mate polyandrously [19, 20]. These bonds typically form when an immigrant male joins an established family group and allies with the same sex resident breeder, to whom it is related [20]. Cooperative breeding, with group members working together to raise the brood, can only arise once stable groups have formed [64]. Both offspring delayed dispersal and bonding between adults males are therefore necessary preceding steps that eventually lead to cooperation, which has to be considered a consequence rather than the cause for the formation of the social group [65–67]. In other words, to understand why carrion crows form groups we need to understand the fitness advantages that sociality conveys, independently of the payoff derived from cooperation at the nest. In carrion crows, as well as in many other cooperatively breeding species, the benefits of delayed dispersal for offspring are well studied [42], but little is known about the fitness advantage of forming long lasting social bonds. We suggest that the health benefits of high quality relationships may be an important factor for establishing stable alliances between males.
Besides the positive effect of strong social bonds, we found that the number of relatives in the group also affected parasite excretion patterns in carrion crows, with individuals living in groups with more related individuals excreting less samples containing coccidian oocysts. Sociality is based on kinship in wild cooperatively breeding carrion crows. Offspring remain in the natal territory with their parents for years and more distant relatives are actively recruited to form cooperative alliances with the resident breeders [42]. As a result, social groups in cooperatively breeding carrion crow are extended families, comprising members with different degree of relatedness. Indirect fitness benefits are known to be a primary driver of kin-based sociality in many taxa of animals [69]. However, our results now show that kin living can also accrue immediate direct benefits to carrion crows in terms of improved health, as estimated via parasite product excretion.
Male versus female sociality
In cooperatively breeding carrion crows, group sex ratio is male biased [68]. Within the social groups, conflicts are more likely among males, which form stable linear hierarchies that are much stronger than those of females. This sex difference mirrors the level of competition for resources, such as territory and reproduction, which is higher in males than females [21, 68]. Females of a group also generally gain unrestricted access to food sources, whereas males often interfere with each other and fight [39]. In their groups, males are therefore likely to undergo higher levels of social stress than females. This may explain why the observed benefits of social bonding in carrion crows arise in males but not in females. First year males were also less likely to excrete feces containing coccidian oocysts than second year and adult males. This may also be due to the increasing level of competiveness that males experience when they age and face the challenge of acquiring a territory and breeding position in the population.
Conclusions
Social complexity, associated with high cognitive abilities, has deep phylogenetic roots, being widespread not only among primate species but also in a variety of other mammals and birds. Understanding its evolution, however, is difficult because of our scant knowledge of the fitness consequences of sophisticated social behavior across taxa. We are just starting to fill the gaps, and this study shows a measurable positive health effect of the quality of social relationships in a bird species that may trigger social bonding, setting the stage for a complex form of cooperation, i.e. cooperative breeding, to arise. Importantly, our results parallel previous findings in primate species, suggesting a general pattern in vertebrates for the evolution of sociality.
Acknowledgments
We thank Kurt Kotrschal for feedback and discussion, Jenna Kiddie for editing the manuscript and Friederike Hillemann for help with fecal sample collection. This work was supported by the Spanish Ministry of Economy and Competitiveness, project CGL2011-27260, to VB and a KWA (‘kurzfristiges wissenschaftliches Auslandsstipendium’) stipend of the University of Vienna to CAFW.Cooperative Vertebrates. Science (80- ) 296:69-72. doi: 10.1126/science.296.5565.69