UC Santa Cruz

Understanding the distribution of lifes variety has driven naturalists and scientists for centuries, yet this has been constrained both by the available data and the models needed for their analysis. Here we compiled data for over 67,000 marine and terrestrial species and used artificial neural networks to model species richness with the state and variability of climate, productivity, and multiple other environmental variables. We find terrestrial diversity is better predicted by the available environmental drivers than is marine diversity, and that marine diversity can be predicted with a smaller set of variables. Ecological mechanisms such as geographic isolation and structural complexity appear to explain model residuals and also identify regions and processes that deserve further attention at the global scale. Improving estimates of the relationships between the patterns of global biodiversity, and the environmental mechanisms that support them, should help in efforts to mitigate the impacts of climate change and provide guidance for adapting to life in the Anthropocene.