On Integrative Species Delimitation…

Accurate delimitation of species is a fundamental first step that underlies much of what we do in biology. But this can prove challenging in many situations. Why? Let me count the ways. Incomplete lineage sorting, hybridization, morphological conservatism, and niche conservatism, to name a few. Of course, access to complete sampling from all OTUs across their geographic ranges is very often an issue as well.

Furthermore, consider the fact that, for well-studied faunas and floras, we desire to illuminate species boundaries in the hardest-to-delimit clades. That is, most of the clearest species boundaries have been identified already. Thus, when applying delimitation methods to modern empirical data, we are judging them based on their performance on the most recalcitrant of datasets.

Given that ambiguity can exist for species boundaries in multiple types of data, a holistic approach to species delimitation makes good sense. Among the plethora of delimitation methods described over the past 7 years or so, several accommodate data from gene- or species trees, as well as phenotypes and even geography (e.g., see here and here). You can see Melissa DeBiasse’s review of one of these methods (iBPP; Solís‐Lemus et al. 2015) from last year.

An interesting philosophical aspect of holistic delimitation methods is their formal integration of phenotypic data with our modern, coalescent-based framework for analyzing multilocus data. Phenotypic data have a long (and ongoing) history in systematics. However, they have occasionally been eschewed in phylogeny reconstruction, based in part on the difficulty of modeling evolution of some morphological characters.

No recent delimitation methods seek to model morphological evolution in a way analogous to the way we model molecular evolution. Instead, they rely (variously) on modeling trait variances within- and among species. In iBPP specifically, these variances are assumed a priori to have resulted from a Brownian motion model of evolution, which may or may not be accurate for the trait(s) under consideration. Indeed, many morphological traits fail to conform to strict Brownian expectations, which is a reminder of the problems associated with modeling trait evolution. Still, the authors of iBPP suggest their method is somewhat robust to violations of a Brownian model.

Have holistic methods such as these become a gold standard for species delimitation? The jury is still out. Answering that question will require further assessment of the methods across a greater diversity of clades and in more complex speciation scenarios. It will also depend on the extent to which current approaches to modeling trait evolution actually capture the dynamics of that process.


Solís‐Lemus, C., Knowles, L. L., & Ané, C. (2015). Bayesian species delimitation combining multiple genes and traits in a unified framework. Evolution 69:492-507.


About Bryan McLean

Biologist studying mammalian phylogeny, morphology, comparative biology, and macroevolution.
This entry was posted in evolution, methods, phylogeography, population genetics, software, species delimitation. Bookmark the permalink.