Predicting future biodiversity with species distribution models: current applications, persistent issues, and where to go from here

There is much current interest in macroecology to make predictions of future biodiversity patterns in order to inform both regional and global priorities for conservation and sustainability of ecosystem functions and services. Species distribution models use data on species' occurrence records, environmental predictor variables, and sometimes other data sources to estimate niche relationships and distribution extents—these models can also be combined to make biodiversity estimates. As the field of species distribution modeling has grown considerably over the past two decades, many approaches now exist to build models, evaluate their performance, and use them to make predictions for unsampled areas and times. I will provide an overview of current techniques to predict future distributions of species and biodiversity, detail some issues with these techniques concerning uncertainty and realism of predictions, and contribute my humble thoughts on where the field should go from here.