Current Research

Speciation

Biotechnology and Biological Sciences Research Council
We are interested in the question of whether speciation accelerates evolution at the molecular level, causing punctuational bursts of evolution. We are interested in the mechanisms by which this may occur, including sampling effects or bottlenecks, rapid adaptive evolution, and discrete genetic mechanisms that may promote rapid speciation of small groups. These include hybridization, polyploidy, chromosomal rearrangements or karyotypic change, and behavioural mechanisms such as pollinator or host switching.

Linguistics and Cultural Evolution

Leverhulme Trust and the European Research Council
We are working on methods for inferring phylogenetic trees of languages and using those trees to measure rates of word evolution over time. In collaboration with Russell Gray at the University of Auckland, we are applying these methods to Indo-European, Bantu, Austronesian, Mayan and Uto-Aztecan languages.
We are particularly interested in why some words evolve at higher rates than others, and have developed the idea of the linguistic half-life. We have identified some meanings whose words evolve slowly enough to have time-depths of at least 20,000 years, making them candidates for deep reconstruction of languages including perhaps proto-IndoEuropean or earlier such as Nostratic.
Our studies of cultural evolution investigate the idea that human cultures behave as if they were distinct biological species.

Phylogenetic Inference

Natural Environment Research Council
We are investigating models of sequence evolution suitable for data that can vary their tempo or rate and mode of evolution from site to site or throughout the phylogenetic tree. We have devised mixture models and covarion based models for gene-sequence, morphological or other data. The techniques are implemented as Markov Chain Monte Carlo methods (including reversible-jump) that find Bayesian posterior distributions of model parameters. They can detect rate variation, heterotachy, so-called invariant sites, and identify the model of sequence evolution that best describes a given site's evolution.

Comparative Methods

Biotechnology and Biological Sciences Research Council
We are working on methods for the analysis discrete data and methods for analysing continuously varying data. The methods can be sued to reconstruct ancestral states, measure rates of evolution, and test for correlated evolution among pairs of traits, or in the case of continuously varying traits it is possible to conduct multiple regressions. The techniques are implemented as Markov Chain Monte Carlo methods (including reversible-jump) that find Bayesian posterior distributions of model parameters. They take into account phylogenetic uncertainty, accommodate unknow states or missing data and give estimates of model uncertainty.
Recently we have devised a Bayesian posterior predictive method that allows investigators to derive predicted values of traits for taxa with unknown values.