Résumé
Phylogenetic conservatism or heritability of the geographic range sizes of species has been predicted to occur because of the phylogenetic conservatism of niche traits. However, evidence for range size conservatism is mixed, and even when statistically significant is often rather weak and of questionable biological importance. Here, we test the prediction that such conservatism will be more strongly expressed when the amount of spatial overlap between sister species increases. We used the global distributions of 1136 avian species (> 10% of extant members of this Class), and tested the conservatism of geographic range sizes using the coefficients of correlation between values for pairs of sister species. We used a null model to test whether the range sizes of sister species were more similar to one another than expected by chance. We found that sister species showed a significant positive relationship between their geographic range sizes whatever the degree of spatial overlap. However, as predicted, the level of conservatism increases with the level of range overlap between sister species. More precisely, the strong increase in the coefficient of correlation between sister species' range sizes when we add species with some range overlap to the pool of pairs without any such overlap indicates an important threshold effect. These results suggest that niche conservatism is more likely to lead to marked heritability of the range sizes of species when similar niche traits are expressed under more similar environmental conditions. These results have significant implications because they suggest 1) that previous analyses of conservatism of range sizes have been confounded by the level of spatial overlap, and 2) that closely related species experiencing similar conditions may tend to expand or restrict their geographic ranges in parallel when faced with climate change.