Molecular phylogeny and evolution of inflorescence types in Eperua.

PREMISE: The Amazonian hyperdominant genus Eperua (Fabaceae) currently holds 20 described species and has two strongly different inflorescence and flower types, with corresponding different pollination syndrome. The evolution of these vastly different inflorescence types within this genus was unknown and the main topic in this study. METHODS: We constructed a molecular phylogeny, based on the full nuclear ribosomal DNA and partial plastome, using Bayesian inference and maximum likelihood methods, to test whether the genus is monophyletic, whether all species are monophyletic and if the shift from bat to bee pollination (or vice versa) occurred once in this genus. RESULTS: All but two species are well supported by the nuclear ribosomal phylogeny. The plastome phylogeny, however, shows a strong geographic signal suggesting strong local hybridization or chloroplast capture, rendering chloroplast barcodes meaningless in this genus. CONCLUSIONS: With our data, we cannot fully resolve the backbone of the tree to clarify sister genera relationships and confirm monophyly of the genus Eperua. Within the genus, the shift from bat to bee and bee to bat pollination has occurred several times but, with the bee to bat not always leading to a pendant inflorescence.

Estimating species richness in hyper-diverse large tree communities.

Species richness estimation is one of the most widely used analyses carried out by ecologists, and nonparametric estimators are probably the most used techniques to carry out such estimations. We tested the assumptions and results of nonparametric estimators and those of a logseries approach to species richness estimation for simulated tropical forests and five data sets from the field. We conclude that nonparametric estimators are not suitable to estimate species richness in tropical forests, where sampling intensity is usually low and richness is high, because the assumptions of the methods do not meet the sampling strategy used in most studies. The logseries, while also requiring substantial sampling, is much more effective in estimating species richness than commonly used nonparametric estimators, and its assumptions better match the way field data is being collected.