Most of this website is written for a specialist scientific audience; folks who aren’t scientists are very welcome to look around, but this page might be a more accessible introduction.

I am an associate professor at U of Tennessee, Knoxville, in the department of Ecology & Evolutionary Biology. In my lab, we develop and apply phylogenetic tools to address evolutionary questions. They are usually generated by a direct research need: how can we tell whether this group is evolving at a different rate? How can we choose between phylogeographic models without limiting ourselves to a pre-selected small set? Is there hidden variation in states that lets some herbaceous plants retain the ability to make wood while others have lost this ability? By developing techniques to address these questions, we both solve the original question and enable other biologists to use these new techniques to answer more questions. Broadly, the areas covered include trait evolution, species delimitation, phylogeography, dating trees, and more work in progress.

I work on empirical systems in collaboration with others as well as other kinds of studies, but many people coming to this website are probably most interested in methods developed here. Some highlights:

  • Species delimitationBrownie (O’Meara 2009), which has nonparametric and parametric species delimitation: it returns a species tree and species assignments without first requiring assignment of individuals to populations. We are currently working on a different approach (but which requires pre-assignment) in phraplbut it is not quite ready (expect submission of manuscript Aug 2015).
  • Continuous trait evolutionary rates: Different rates for Brownian motion were first in Brownie (O’Meara, Ané, Sanderson, and Wainwright, 2006), then RBrownie (no longer on CRAN), and are available in R in Liam Revell’s phytools. We later made flexible Ornstein-Uhlenbeck models that allow the mean, variance, and attraction parameters to change on the tree. This is in the R package OUwie (Beaulieu, Jhwueng, Boettiger, and O’Meara, 2012).
  • Discrete trait models: Sometimes all you see is a binary state, but this hides a lot of complexity. For example, we can code plants as “woody” or “herbaceous”, but are all herbaceous plants the same? It’s possible that some still have the genes for producing wood and can readily re-evolve woodiness, but others cannot. Most methods treat these the same, but we allow hidden states: in this case, the hidden state might be “having genes for woodiness”. This is in the R package corHMM (Beaulieu, O’Meara, and Donoghue 2013)
  • Discrete trait diversification models: BiSSE models are popular but can have issues with spurious correlations, as well as treating all taxa with a given observed trait as the same. We have what we think is a solution in HiSSE. This is in review at Systematic Biology (since June 2015), but a preprint is available here: Beaulieu and O’Meara in review
  • Phylogeography: Often in phylogeography we look at just a handful of models. In other cases, models must be restricted in some way, by, say, prohibiting migration. What if there were something like ModeTest but for phylogeography rather than nucleotide models? We have developed software, phrapl, for this. The manuscript has been in review at Evolution for the past six months.
  • Dating: Stephen Smith and I worked on a reimplementation of Mike Sanderson’s penalized likelihood to scale to large trees: treepl (Smith and O’Meara 2012). If you can get it installed, it works well. I am also working on datelife.org, a repository of chronograms inspired by TimeTree but with an eye for reusability (though it’s not very useful yet, we just got funded to improve it).
  • DNA/AA evolution: Read a recent phylogeny paper: they probably used GTR+G or a similar model, perhaps partitioned. We know far more about how DNA evolves than can be fit in a few parameter model, but a full, free codon model is generally too parameter-rich to fit, either. We have a method selac that can combine models for mutation on nucleotides with selection on amino acids. Still not published, but stay tuned (O’Meara, Beaulieu, Chai, Gilchrist, in prep).
  • In the pipeline: We have an improvement to the DEC model of Ree and Smith, approaches for doing comparative method on networks, rather than trees, ways to flexibly make complex models, and more, in prep. You can get a sense of this by looking at what people in the lab are working on.

See more info on research here.

In the last six years, work in the lab has been funded by five NSF grants to me as PI or Co-PI as well as awards from iPlant, Encyclopedia of Life, and Google Summer of Code to me or people doing work in the lab (see more info here). Grad students in the lab have been supported by teaching assistantships as well as a PEER fellowship. The UT Knoxville-based National Institute for Mathematical and Biological Synthesis (NIMBioS) remains critically important for my work, whether by funding independent postdocs (I have mentored seven NIMBioS postdocs, in addition to three additional postdocs in my lab with other funds), sponsoring workshops, or organizing working groups.