Estimating confidence intervals on reconstructed cophylogenies using bootstrap resampling

Cophylogenies represent coevolutionary histories of two or more sets of coevolved taxa, and are used to study coevolution and other fundamental evolutionary processes. As with traditional phylogenies, cophylogenies are primarily reconstructed using computational analysis of DNA and other biomolecular sequence data. An essential question concerns the reliability of reconstructed phylogenies and cophylogenies.Statistical resampling offers a principled approach to evaluate statistical confidence for these tasks. We therefore apply bootstrap resampling – one of the most widely used non-parametric resampling techniques – to place confidence intervals on a reconstructed cophylogeny, which is the first such method to our knowledge. We validate the performance of the resulting reliability estimates in a simulation study as well as an empirical case study of bobtail squid and its bioluminescent endosymbionts. The utility of statistical resampling to assess cophylogenetic reconstruction reliability in an automated and data-driven manner points the way forward – both for future methods development and wider adoption in studies of symbioses and other forms of coevolution.