A Practical Integer Linear Program Formulation for the Cophylogeny Reconstruction

David Bryant University of Otago, New Zealand Phylogenetics with Nets and a Lasso Joint work with Alethea Rea The (positive) lasso is a widely used method in statistics which selects variables and improves least squares estimates. The key idea is to find the best non-negative least squares estimate min β≥0 ‖Xβ − y‖2 while slapping on the constraint ‖β‖1 ≤ λ. Increasing λ gives (usually) solutions with more non-zero variables. Efron et al (2004) present an algorithm which efficiently explores lasso solutions for all λ, but it doesn’t always work. We describe an algorithm which does, and look at applications to phylogenetic network methods like NeighborNet. Chris Burridge University of Tasmania Divergence of island biotas when they were not always islands Joint work with Danielle Nankervis, Lindi Olivier, Jessica Wadley, Bill Brown, and Jeremy Austin Continental shelf islands are islands that possessed terrestrial connections to other landmasses via exposed continental shelf during low sea stands. These islands often contain lineages that are related but distinct from those on the adjacent mainland, and their divergence is usually assumed to reflect isolation initiated by the post LGM marine transgression. However, a range of alternate divergence scenarios exists. Divergence may have been initiated by an earlier marine transgression and was retained through subsequent glacial periods of terrestrial connection, or may have been dissociated from sea level changes, involving marine dispersal. Here we test these alternate scenarios for a species that exhibits morphologically distinguishable lineages between Tasmania and the Australian mainland, but also has marine dispersal potential – the wedge-tailed eagle Aquila audax. Genetic variation was surveyed at 20 microsatellite loci and in mitochondrial DNA, from 148 Tasmanian and 48 mainland individuals. The populations differed significantly in allele frequencies, and gene flow subsequent to the initiation of their divergence was rejected. While this is consistent with marine transgression initiating lineage divergence, estimates of age of this divergence substantially post-dated the most recent marine transgression, rejecting this mechanism. Instead, the data are consistent with a small number of mainland individuals colonising Tasmania by marine dispersal, and subsequent isolation. This study highlights that divergence of continental shelf island biotas from those of the adjacent mainland do not necessarily follow the paradigm of initiation by marine transgression, and have bearing on the conservation status applied to putatively ‘long-isolated’ lineages on such islands. Mike Charleston University of Sydney, Australia A Practical Integer Linear Program Formulation for the Cophylogeny Reconstruction Problem The problem of uncovering ancestral relationships between ecologically linked species is of fundamental interest in evolutionary biology. Given a pair of evolutionary trees, which in the treatment here are rooted and binary, and the observed links between their tips such as “this pathogen strain infects that host species,” an explanation is sought of how the two trees have grown in evolutionary time. Such