Controlling for the effects of environmental availability when testing how the environment determines community compositional uniqueness

Abstract

Identifying the drivers that promote unique species compositions (i.e. ecological uniqueness) is crucial to understanding the mechanisms underpinning diversity patterns and for effective conservation planning. Environmental conditions are often sampled differentially in datasets, which can lead to rarer environments having unique species compositions, provided that environmental differences increase compositional differences. This effect, however, will be undesirable when a study aims to test the direct impact of environments rather than their availability. We developed an approach to reduce the effects of environmental availability in ecological uniqueness analyses through calculating expected pairwise compositional dissimilarities for each unique environmental condition. We further used simulations to assess the performance of our methods by randomly generating communities from two hypothetical environments with non‐overlapping species pools. Additionally, we used a dataset of 50 tree communities to demonstrate how environmental availability could impact relationships between environmental conditions and ecological uniqueness in empirical studies. Our simulations revealed that uniqueness metrics based on observed values are sensitive to environmental availability, while our approach correctly concluded that there were no differences among environments under the unbalanced design. Our analysis of tree communities produced divergent conclusions between the two approaches, as increasing slope reduced ecological uniqueness after controlling for their low availability in the dataset only. This suggests that low environmental availability inflated the ecological uniqueness of sites with high slope, which opposed its direct negative effects on ecological uniqueness, leading to a weak relationship based on observed values. To achieve a more mechanistic understanding of ecological uniqueness patterns, the effects of environmental availability must be considered. We recommend that studies use both uncorrected and corrected analyses to identify not only the direct effects of environmental conditions but also the degree to which their availability influences the observed relationships between ecological uniqueness and environment conditions. Our approach should be most necessary and applicable in unbalanced designs, which is a common characteristic in empirical studies.