Robustness driven model development in island biogeography

Activity: Talk and presentationAcademic presentationAcademic


Islands are model systems for understanding biodiversity, ecology and evolution. Tackling outstanding questions on the colonisation and diversification histories of species in island communities requires statistical inference models to understand the rates of (macro)evolutionary processes and select among different models to determine the most probable explanation. This can be done either by deriving new models tailored for specific questions or by applying models that already exist but may not be ideally suited for the question at hand. Here we test whether a general model of island biogeography is applicable to new areas of island biogeography, by examining how it performs when its assumptions are violated. The model (DAISIE) can estimate rates of colonisation, speciation and extinction (CES rates) on islands and detect whether these rates are diversity-dependent. DAISIE was developed for studying phylogenetic data from oceanic islands (never connected to the mainland) and assumes island area is constant through time. We tested whether the model is robust to changes in island area and connectivity through time, either via geological changes to the island, or sea-level oscillations. We identify that the model fails to reliably estimate CES rates when islands were previously connected to the mainland (continental islands), in which cases a new inference model is required. We then develop this continental model of island biogeography and apply it to simulated data to show whether initial species presence on the island can be detected and whether these signatures exist in empirical data from continental islands.
Event titleMathematical Models in Ecology and Evolution 2022
Event typeConference
LocationReading, United KingdomShow on map
Degree of RecognitionInternational