RESUMO
PREMISE OF THE STUDY: The North American Cercis clade spans dry to mesic climates and exhibits complex morphological variation. We tested various proposed species classifications of this group and whether aspects of leaf morphology, particularly the "drip-tip" in some regional populations, are adaptive and/or linked with phylogeny. METHODS: We made measurements on over 1100 herbarium specimens from throughout North America and analyzed the data with univariate and multivariate approaches. We analyzed phylogenetically DNA sequence data from nuclear ITS and three plastid regions from 40 samples, and estimated divergence times with a relaxed-clock Bayesian analysis. We used climate and geographic position data to predict the variation observed in leaf size and shape by using stepwise multiple linear regressions. KEY RESULTS: Morphometric analyses yielded a pattern of continuous and often clinal character variation across North America, without correlated gaps in character states. Conversely, phylogenetic and divergence time analyses yielded distinct clades from California, the interior west, and eastern North America separated by between ~12 and 16 million years. Multiple regressions yielded highly significant correlations between leaf apex shape and precipitation of the warmest quarter. CONCLUSIONS: Despite a pattern of continuous morphological character variation, the long period of geographic and presumably genetic isolation warrants the delimitation of three species. Predictive modeling supports the adaptive value of acuminate apices or "drip-tips" in mesic habitats. This suggests that Cercis leaves change more rapidly than inferred from parsimony reconstruction, which has implications for the evolution of the dry floras of North America and Eurasia.