Abstract
For large herbivores, predation-risk, habitat structure and population density are often reported as major determinants of group size variation within and between species. However, whether the underlying causes of these relationships imply an ecological adaptation or are the result of a purely mechanistic process in which fusion and fragmentation events only depend on the rate of group meeting, is still under debate. The aim of this study was to model guanaco family and bachelor group sizes in contrasting ecological settings in order to test hypotheses regarding the adaptive significance of group-size variation. We surveyed guanaco group sizes within three wildlife reserves located in eastern Patagonia where guanacos occupy a mosaic of grasslands and shrublands. Two of these reserves have been free from predators for decades while in the third, pumas often prey on guanacos. All locations have experienced important changes in guanaco abundance throughout the study offering the opportunity to test for density effects. We found that bachelor group size increased with increasing density, as expected by the mechanistic approach, but was independent of habitat structure or predation risk. In contrast, the smaller and territorial family groups were larger in the predator-exposed than in the predator-free locations, and were larger in open grasslands than in shrublands. However, the influence of population density on these social units was very weak. Therefore, family group data supported the adaptive significance of group-size variation but did not support the mechanistic idea. Yet, the magnitude of the effects was small and between-population variation in family group size after controlling for habitat and predation was negligible, suggesting that plasticity of these social units is considerably low. Our results showed that different social units might respond differentially to local ecological conditions, supporting two contrasting hypotheses in a single species, and highlight the importance of taking into account the proximate interests and constraints to which group members may be exposed to when deriving predictions about group-size variation.