Tropical mountain cradles of dry forest diversity.

O f all ecosystems in tropical America, seasonally dry forests, patchily distributed where relatively fertile soils, annual rainfall 1,600 mm, and strong seasonal drought coincide, are perhaps the least understood and most endangered (1, 2). Mesoamerican dry forest diversity peaks in southwest Mexico, where up to 16% of tree species are local endemics (3). Judith Becerra, in this issue of PNAS (4), presents a novel historical analysis of Mexican dry forests based on a time-calibrated phylogeny of the dry forest tree genus Bursera, which was the source of the sacred Mayan incense known as copal. Becerra found that peak diversification of Bursera lineages [34–17 million years ago (Ma)] in northwest Mexico followed the uplift of the Sierra Madre Occidental (34–15 Ma), whereas diversification of a southern lineage (peaking at 13.5 Ma) tracked the rise of the Neovolcanic axis (23–2.5 Ma) (Fig. 1). The rising mountains, through their influence on regional climate, permitted dry forests to take hold in Mexico and to spread into Central America. Becerra’s study (4) is an exciting contribution to historical biogeography, which, in the absence of adequate fossils, turns increasingly to plant phylogeny to infer biome histories (5). The dry forest study raises challenging methodological questions. Can the age of a single clade be used to infer the age of a biome? In the neighboring rain forest biome, for example, major clades have radiated at vastly different times, so focus on a single taxon would be misleading. On the side of rain forest youth, the common ancestor of the species-rich (n 300 species) neotropical tree genus Inga is apparently less than 6 million years old (6). Inga shares the rain forest with trees from several families in the Malpighiales, whose ancestors derive from mid-Cretaceous (94–112 Ma) proto-rain forests (7). As another example, the dry forest legume clade Leucaena underwent endemic radiation in southwest Mexico beginning 10 million years ago (8), making it younger than Bursera but still compatible with Becerra’s biogeographic model.