Inferring the drivers of species diversification: Using statistical network science

Understanding the processes that drive diversification of species is essential and urgent. Although detailed data collection and modern genetics have helped enormously in being able to map existing species back to their common ancestor, extinct species are blatantly missing in these analyses, leaving us with only a very partial and biased picture of the extinction process. Nevertheless, extinct species can leave behind a signal in the phylogeny the currently extant species. Over the past decades simple macro-evolutionary models have been proposed for this signal.
Unfortunately, these models lack biological plausibility and more complex versions of such models, which incorporate species interactions, are computationally almost intractable, because a coherent description of these interactions results in an extremely large set of coupled differential equations. Moreover, fitting this complex model to the molecular phylogeny of extant species requires integration over a high-dimensional space of partially extinct tree topologies, which is computationally unfeasible with current methods.
In this thesis we present a novel quantitative framework, combining statistical methodologies with data augmentation techniques, that allows the integration, combination and incorporation of biological realism into a general class of models making it possible to consider the complexity of ecological interactions. We illustrate our framework by studying novel species diversification models that include dependence of diversification on phylogenetic diversity.
We conclude that our methodology has great potential to overcome and quantify a broad variety of hypotheses in theoretical and evolutionary community ecology.