Abstract
The challenges associated with sampling rare species or populations can limit our ability to make accurate and informed estimates of biodiversity for clades or ecosystems. This may be particularly true for tropical trees, which tend to be poorly sampled, and are thought to harbor extensive cryptic diversity. Here, we integrate genomics, morphology, and geography to estimate the number of species in a clade of dioecious tropical trees (Canarium L.; Burseraceae) endemic to Madagascar, for which previous taxonomic treatments have recognized between one and 33 species. By sampling genomic data from even a limited number of individuals per taxon, we were able to clearly reject both previous hypotheses, and support instead an intermediate number of taxa. We recognize at least six distinct clades based on genetic structure and species delimitation analyses that correspond clearly with geographic and discrete morphological differences. Two widespread clades co-occur broadly throughout eastern wet forests, one clade is endemic to western dry forests, and several slightly admixed clades are more narrowly distributed in mountainous regions in the north. Multiple previously described taxa were recovered as paraphyletic in our analyses, some of which were associated with admixed individuals, suggesting that hybridization contributes to taxonomic difficulties in Canarium. An improved understanding of Canarium species diversity has important implications for conservation efforts and understanding the origins of diversity in Madagascar. Our study shows that even limited genomic sampling, when combined with geography and morphology, can greatly improve estimates of species diversity for difficult tropical clades.