NullSens: Partitioning Abotic and Biotic Contributions to Community Variation

Contributors: Steven D. Essinger, Gail L. Rosen, Christopher B. Blackwood

Description:

It is well known that both abiotic factors and biotic interactions structure ecological communities. However, the relative importance of these factors in influencing community composition is openly debated and may have profound implications for management of communities and our ability to predict community responses to perturbations. Statistical methods typically used for analysis of abiotic factors search for correlation between community composition and abiotic gradients using ordination and regression techniques, whereas methods for detection of biotic interactions primarily rely on assessing patterns of presence-absence. Each of these analyses is carried out independently because there is a lack of unified statistical methods for simultaneous analysis of biotic and abiotic factors influencing community composition. This paper presents a unified method that may be used to test hypotheses about species interactions when environmental gradients are present. The method employs a null model to assess community-wide biotic covariation after removing individual species responses to environmental gradients (i.e. abiotic factors) so that apparent species interactions are not masked nor augmented by the abiotic responses. The method can be used to calculate percentages of variation explained due to abiotic, biotic and unexplained factors. Within the biotic variation results, the algorithm provides output indicating if covariation among species due to biotic interactions is predominantly positive or negative within the community. We demonstrate the performance of the method by a) sensitivity analysis on simulated communities and b) published leaf-litter ecological dataset. The method is publicly available via the R package, NullSens.