Partitioning phylogenetic and adaptive components of the geographical body-size pattern of New World birds

Aim To evaluate seasonal body‐size patterns for New World birds in geographical space, to develop environmental models to explain the gradients, and to estimate phylogenetic and adaptive contributions. Location The Western Hemisphere. Methods We used range maps to generate gridded geometric mean body masses. Summer and winter patterns were distinguished based on breeding and non‐breeding ranges. We first generated the geographical gradients, followed by phylogenetic eigenvector regression to generate body sizes predicted by the birds’ positions in a phylogenetic tree, which were used to generate the expected phylogenetic gradient. Subtracting the expected pattern from the observed pattern isolated the adaptive component. Ordinary least squares multiple‐regression models examined factors influencing the phylogenetic, adaptive and combined components of the seasonal body‐size patterns, and non‐spatial and spatial models were compared. Results Birds are larger in the temperate zones than in the tropics. The gradient is quantitatively stronger in winter than in summer. Regression models explained 66.6% of the variance in summer mass and 45.9% of the variance in winter mass. In summer, phylogenetic and adaptive responses of birds contribute equally to the gradient. In winter, the gradient in North America is much stronger than that expected by taxonomic turnover, and responses of species independent of their family membership drive the overall pattern. Main conclusions We confirm Bergmann's rule in New World birds and conclude that winter temperatures ultimately drive the pattern, exerting selection pressures on birds that overwhelm patterns expected by phylogenetic inertia at the family level. However, in summer, the movement of migratory species into the temperate zone weakens the gradient and generates a pattern more congruent with that expected from the taxonomic composition of the fauna. The analytical method we develop here represents a useful tool for partitioning the phylogenetic and non‐phylogenetic components of spatially explicit macroecological data.