Bailey D McKay, Herman L Mays Jr, Yuchun Wu, Hui Li, Cheng-Te Yao, Isao Nishiumi, and Fasheng Zou. 2013. An empirical comparison of character-based and coalescent-based approaches to species delimitation in a young avian complex. Molecular Ecology 22: 4943-4957.
This study assesses the differences in how coalescent-based and character-based approaches delimit species, and uses the light-vented/Taiwan bulbul (Pycnonotus sinensis/Pycnonotus taivanus) complex as a model. Population Aggregate Analysis (PAA) was used to find fixed plumage, mitochondrial (ND2), and nuclear (13 introns) characters among a population consisting of 81 specimens spanning the distribution of P. sinensis (of which there are 4 subspecies) and P. taivanus. The cumulative dataset suggested a 4 group complex, while Bayesian Phylogenetics and Phylogeography (BPP) seemed to favor a 5 lineage split overall. The authors observed that increasing or decreasing the θ prior led to significant changes in speciation probability, which seemed to contribute to their ultimately supporting the 4 species hypothesis based off of character-based analysis (despite statistically higher support for BPP’s 5 lineage hypothesis).
We noticed that sampling of P. s. hainanus was limited to an island off the mainland, and it is unclear whether the absence of representation among a potential mainland hybrid zone (between it and P. sinensis) affected the hypotheses generated by PAA. There was also issue with what constitutes an appropriate cutoff for statistical support in Bayesian analyses, with the consensus being that 0.76, a value which the study indicates as statistical support, as being too low. The frequency of poor support in the BPP analysis led to the question of whether BPP is inherently problematic with respect to more recent divergences, and how priors should be adjusted going forward. [UW Phyloseminar, Fall 2013, Leonard Jones]