Image

ABSTRACT
Species distribution models (SDMs) are widely used across fields as diverse as ecology, archaeology, epidemiology, and conservation biology. Despite their broad utility, fundamental questions remain about their ability to estimate underlying biological processes, and the field still struggles with statistical issues that are inherent to the data that is used to build them. In this talk, I will discuss a new synthetic framework for conceptualizing randomization tests that are meant to deal with these issues, and demonstrate how useful it can be to ask seemingly inane questions like:

• What does it look like if nothing interesting is going on?
• What if we only see what we look for?
• What are the effects of climate change on Australia's only endemic Pokemon?

Through exploring the behavior of SDMs in such artificial scenarios, I will show how we can shed light on some deeply problematic statistical issues, and how that same carefully crafted nonsense can be used to address those problems and allow us to make better decisions for our ecological future.

BIOGRAPHY
Dan Warren specializes in quantitative approaches to population biology, particularly in fields related to biodiversity and evolution.  Originally from the USA, he has worked in Australia (ANU, Macquarie), Germany (Senckenberg Biodiversity and Climate Research Center), and Japan (OIST).  In 2024 he started a new position as a Senior Research Fellow at the Gulbali Institute in Albury-Wodonga.  He is the author of several popular R packages including ENMTools and RWTY, and has developed quantitative approaches that are now standard tools in ecology, conservation, and evolutionary biology.