Pyron R. A. and Burbrink F.T. 2013. Phylogenetic estimates of speciation and extinction rates for testing ecological and evolutionary hypotheses. Trends in Ecology and Evolution. In press.
In this paper, the authors discussed the use of phylogenies for estimating speciation and extinction rates in order to answer questions related to evolution, ecology and biogeography. The review focuses on how speciation and extinction rates are influenced by three models: (1) “time-dependent”, considered as the influence of time on diversification, (2) “diversity-dependent”, interpreted as the relationship between the number of species in a clade and the availability of ecological niches, and (3) “traits-dependent”, based on the idea that the character states effect the success of a clade. The paper summarizes methods used to estimate speciation and extinction rates, the use of branch length and tree type (gene trees versus species trees), models that allow a “negative net diversification” when extinction is greater than speciation, the use of hidden Markov models (HMMs) to estimate the likelihood of the speciation and extinction rates, and the development of the R package ‘diversitree”, which includes models for character-state change. However, the authors state that the lack of accurate mathematical models is still a limitation for obtaining accurate parameter estimates. To test the influence of the three models (time, diversity, and traits) the authors performed simulation analyses and reported if the model was capable of detecting and estimating the true speciation and extinction rates. They found that the success depends on different factors for each model: (1) “time-dependent” detected both extinction and speciation rates concordant with the fossil data, (2) “diversity-dependent” estimated extinction rates with less accuracy in recent clades than in older clades, and (3) “traits-dependent” detected in more than 70% of the cases the relationship between speciation rates and traits when the trees are large, and the effects of a trait on extinction were recovered in 40% of the cases when the trees had at least 500 tips. In addition, the authors warn users about the factors that can affect phylogenetic inference: “gene tree discordance, unsampled taxa and cryptic species”, and recommended the integration of the time, diversity and trait dependent models in a single framework. Finally the authors emphasized that extinction is hard to estimate, and that more simulations are needed to determine the conditions when the extinction and speciation rates can be estimated accurately.
[UW phylogenetics seminar, 10/24/2013, Itzue Caviedes-Solis]