Résumé
We analysed the degree to which spatial, ecological and phylogenetic factors influenced geographic gradients in range size for all trees native to North America. We compared observed range sizes, measured in either one (latitudinal or longitudinal extent) or two dimensions (range area), with three null models that respected constraints imposed by the geometry of the continent; we tested whether climatic gradients could account for increasing range size with latitude a Rapoport effect in North American trees; and whether variation in range size was associated with phylogeny or ecological characteristics of species. We documented an excess of species with small ranges and a lack of widely distributed species compared with null expectations both with and without environmental constraints. We found evidence for a Rapoport effect in North American trees, at different taxonomic levels and for different groups defined by ecological characteristics. This pattern is not an outcome constrained by continental shape and does not depend on the metric for range size, but it is consistent with the climatic variability hypothesis proposed to explain the Rapoport pattern. Finally we showed that trees east of the Rocky Mountains have larger ranges on average than trees to the west or in Mexico and that the proportion of deciduous compared to evergreen species increases with range size. Variation in range size among North American trees is not spatially random, and has a phylogenetic and ecological signal. Consistent with the climatic variability hypothesis, range size of North American trees increases with increasing variability in annual temperature range at higher latitudes.