This chapter describes multi-type trees and various extensions to the basic phylogenetic model that can account for population structure, geographical heterogeneity and epidemiological population dynamics.

Statistical phylogeography

Phylogeography can be viewed as an approach that brings together phylogenetics and biogeography (Avise 2000). Phylogeography has a long historyand the methods employed are very diverse. Phylogeographical patterns can be explored within a single species or between closely related species, and these patterns can be used to address questions of geographical origins and expansions, island biogeography (e.g. Canary Islands (Sanmartín et al. 2008)), range expansions/contractions and effects of environmental and climate changes on geographical dispersal and extent. As with other chapters in this book, we don't attempt to be comprehensive, but instead point to some relevant material and focus on approaches that we are familiar with and believe to have promise. We mainly consider methods that attempt to directly reconcile geographic data with the phylogenetic relationships of the sampled taxa. Until recently, due to its simplicity, the most popular method reconciling discrete geographical locations with phylogenetic relationships (and as a result inferring ancestral locations on the phylogenetic tree) was maximum parsimony (Maddison and Maddison 2005; Slatkin and Maddison 1989; Swofford 2003; Wallace et al. 2007). However, this method doesn't allow for a probabilistic assessment of the uncertainty associated with the reconstruction of ancestral locations. As with the field of phylogenetics, phylogeography recently experienced a transition to model-based inference approaches. When it finally came, the struggle for authority between model-based approaches and the alternatives was relatively short compared to the protracted ‘troubled growth of statistical phylogenetics’ recounted by Felsenstein (2001). History will probably find that the argument in defence of model-based approaches for phylogeography mounted in the journal Molecular Ecology by Mark Beaumont and a swath of fellow proponents of Approximate Bayesian Computation (ABC) and other forms of model-based inference (Beaumont et al. 2010) was the decisive moment. That paper probably marks the analogous transition in phylogeographic research to the transition made in phylogenetic research a decade or more before. We will only consider model-based methods in this chapter. Statistical methods for phylogeography include those that perform inferences conditional on a tree, and those that jointly estimate the phylogeny, the phylogeographic state and the parameters of interest.