Forest plot shows the parameter estimates of maximum elevation on wing morphology, based on separate phylogenetic models for each 3-km-wide elevational band (starting with 0-3 km and ending with 5-8 km above sea level). The sample sizes within elevational bands refer to flapping species only because the models included an interaction term (elevation: flight mode) such that the main effect of elevation was estimated with flight mode at the reference level. Central points show the mean, and error bars show 95% CI. Results were generated using 100 randomly selected trees and averaged via Rubins rules. Positive values indicate the strength of positive correlation between maximum elevation and the corresponding wing metric. Significant effects are inferred when estimated 95% CI does not span zero (highlighted in a darker shade). The findings were robust to the removal of seabirds and long-distance migrants and the use of alternative bandwidths (2 and 4 km;Figure S4). Models testing the absolute changes in wing morphology (without size correction) showed similar trends with larger effect sizes (Figure S4). Silhouettes illustrate tendency of wings to become larger and more elongated with increasing elevation.
