The missing theory of species co-occurrence in ecology
2018-01-11T18:59:36Z (GMT) by
Talk for the "150 Years of <i>The American Naturalist</i>" Symposium at the 2018 Stand Alone Conference of the American Society of Naturalists.<br><br>Abstract: In 1983, <i>The American Naturalist</i> published a special issue, the innocuously titled “A Round Table on Research in Ecology and Evolutionary Biology.” Embedded in this broad title was an attempt to address a roiling debate in ecology at the time: can the spatial arrangement of species on a landscape be used to infer the underlying community structuring mechanism? Specifically, can co-occurrence patterns signal underlying competition? The seven papers in this issue magnify the surprisingly expansive questions at the heart of this debate through a profound exploration of ecology, evolution, inference, and philosophy. In recent years, similar competition-inference methods have resurfaced into prominence, as ecologists increasingly estimate species interactions using large databases of spatial and temporal occurrences. However, the important debates of the 20th century have been largely ignored in the rapid adoption and current widespread implementation of new machine learning and network inferential methods. Further, foundational concepts shaped through debate have significant bearing not only on the inference of species interactions, but also on the many pattern-process inference methods that dominate modern community assembly theory. In this symposium, I will trace the origins of the idea that species interactions can be inferred from spatial patterns, highlighting the role of this often-overlooked 1983 special issue. Further, I will explore the modern landscape of these debates about pattern-process inference, highlighting empirical ground-truthing, simulation models, and new theoretical frameworks.