Concordance of Ecomorphological Relationships in Three Assemblages of Passerine Birds

Abstract

We investigated the generality of the correlation between feeding behavior and morphology in assemblages of insectivorous passerine birds. We outlined an approach that satisfies two criteria necessary for demonstrating the concordance of ecological relationships among communities. First, we determined that ecomorphological correlations were consistently strong within assemblages under comparison by canonical-correlation analysis. Second, the similarity of the ecomorphological-distance relationships among assemblages (i.e., in analogy to the similarity of slopes in a regression analysis) was analyzed by multiple-regression analysis and analysis of covariance (ANCOVA). Our analysis was based on foraging and morphological measurements obtained from passerine birds from localities in New Hampshire [USA], Costa Rice, and St. Kitts. Our within-assemblage analyses demonstrated a strong association between ecology and morphology for the Hubbard Brook site and for the pooled sample of Costa Rica and St. Kitts. Although the correlations were strong, the temperate and tropical localities differed in the foraging variables and morphological traits that described the relationship. These results suggest a consistent relationship between morphology and feeding behavior, but we were not able to distinguish among-locality differences. A pooled canonical analysis of all three assemblages revealed two strong, significant correlations to which species in all three localities contributed, suggesting that species in the three localities share a common ecomorphological pattern. Because canonical-correlation analysis lacks the power to distinguish the contribution of each assemblage to the overall correlation, we used on ANCOVA to test for heterogeneity in the ecomorphological association. We employed a stepwise multiple-regression procedure to reduce the number of morphological measurements used in the ANCOVA. Significant models were found only for the foraging variables summarized by the first two reciprocal-averaging axes. Two ANCOVA''s were subsequently performed, revealing significant unadjusted locality effects. In both analyses, however, the locality effects vanished when adjusted for differences in the morphological covariates. With one exception, both models lacked a significant interaction term between the morphological covariates and locality. Overall, the results indicate a homogeneity in ecomorphological correlations among passerine assemblages.