Bayesian estimation of competitiveness in male house finches: small-billed males are more competitive

摘要

A common method for assessing the competitiveness of animals, especially in birds, is to pit pairs of unfamiliar individuals against each other in contests for limited resources under controlled conditions. Although this approach can clarify dominant-subordinate relationships within dyads, it is often difficult to determine competitiveness for a large group of individuals. Here, by using Bayesian statistical inference and 'hypothetical competition groups', which are formed when individuals experience a series of paired contests, we estimated social competitiveness of male house finches, Haemorhous mexicanus. First, Bayesian competitiveness estimates from paired contests successfully predicted future contest outcomes among four unfamiliar individuals (i.e. social dominance). When data of all rank combinations were pooled, future dominant males had, on average, higher competitiveness estimates than future subordinate males. Similarly, Bayesian statistical inference and hypothetical competition groups identified accurately the relative competitiveness of four subgroups of males (i.e. colourful and drab males from urban and rural sites), which matched the result of direct contests when they were all put into the same cage. This consistency reinforces the validity of Bayesian competitiveness estimation based on hypothetical competition groups. Moreover, we found that the competitiveness estimate was negatively linked to male beak size in the Bayesian framework. Males with smaller bills were more competitive than those with larger bills, perhaps due to their elevated foraging motivation (i.e. limited ability to consume or husk large, valuable seeds). We argue that Bayesian competitiveness estimations, together with a series of paired contests, is a sophisticated approach for acquiring a broad understanding of social and individual competitiveness. (C) 2015 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.