WE AGREE WITH Levis et al. that humans have influenced dry forests since their first arrival in the neotropics. This long interaction has had major effects, not least in leading to widespread destruction of this now highly threatened vegetation (1). It is also possible, as pointed out by Levis et al., that humans modified the distributions of useful woody plant species in dry forests and that this human influence could be partly responsible for their wide geographic distribution. However, the number of such species that are or were cultivated as a proportion of the overall flora of neotropical dry forests is small (8 out of the 4660 species in our data set, according to Levis et al.). We found high floristic turnover among major geographic areas of dry forest. This pattern is driven by the large numbers of range-restricted species in our data set: 3115 species are restricted to only one of the 12 regional floristic groups we identified. Therefore, the effect of geographically widespread species on our conclusions is negligible, whatever the reasons underlying their broad ranges.