I can't define the niche but i know it when i see it: A formal link between statistical theory and the ecological niche

Abstract

The niche is one of the most important concepts in ecology. However, there has been a persistent controversy on how to define, measure, and predict the ecological niche of an organism. Here I argue that these problems arise in part because the niche is defined by the set of all possible environments, many of which do not exist in nature. A complete description of the niche would require knowledge of a large number of environments that do not exist in nature. Given this, I propose that ecologists should not focus on the niche itself but instead on determining if a particular environment is a part of the niche. I then demonstrate that such an analysis has a natural interpretation as an estimate of the probability that an environment is suitable and that either experimental investigations or analyses of presence data can estimate this quantity. Depending on the way that resources interact to shape environmental requirements, the probability that an environment is a part of the niche behaves like published descriptions of causal inferences and distribution models. However, in some cases the probability that an environment is suitable can be strongly influenced by unmeasured aspects of the environment. When this is true, experimental and distribution models have complimentary strengths and weaknesses. © 2009 Oikos.