A meta-analysis of the relationship between color and aggression

Color varies in pattern and degree across the tree of life. In animals, genetic variation in color is hypothesized to have pleiotropic effects on a variety of behaviors, due to shared dependence on underlying biochemical pathways. Such pleiotropy can constrain the independent evolution of color and behavior. Although associations between color and behavior have been reported, this relationship has not yet been addressed across a broad taxonomic scale with a formal meta-analysis. We used a phylogenetic meta-analytic approach to examine the relationship between individual variation in aggressive behavior and variation in multiple colors. Seventy studies met our inclusion criteria (vertebrates = 66; invertebrates = 4). After accounting for phylogeny and correcting for publication bias, there was a positive association between measures of aggression and degree or area of coloration (mean = 0.274, 95% CI = (0.041, 0.481)). However, this positive association was not influenced by type of color or by several other variables that we tested. Because the data supports a positive association between aggression and degree or area of coloration, irrespective of whether color is melanin-based, carotenoid-based, or structural, we conclude that this pattern does not strongly support the melanin-pleiotropy hypothesis. The relationship was also unaffected by moderators accounting for individual condition, social rank, or age; thus, the results do not strongly support hypotheses that condition-dependence accounts for relationships between color and aggressive behavior. We propose that the moderate positive correlation between aggression and coloration across Animalia that we observed is underlain by genetic covariation between behavior and color traits that serve as badges of status.