Open AccessA network approach for inferring species associations from co‐occurrence data
Author(s) -
MoruetaHolme Naia,
Blonder Benjamin,
Sandel Brody,
McGill Brian J.,
Peet Robert K.,
Ott Jeffrey E.,
Violle Cyrille,
Enquist Brian J.,
Jørgensen Peter M.,
Svenning JensChristian
Publication year - 2016
Publication title -
ecography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.973
H-Index - 128
eISSN - 1600-0587
pISSN - 0906-7590
DOI - 10.1111/ecog.01892
Subject(s) - spurious relationship , biological dispersal , ecology , species distribution , context (archaeology) , range (aeronautics) , nestedness , community , geography , biology , ecosystem , computer science , species richness , habitat , machine learning , population , materials science , archaeology , sociology , composite material , demography
Positive and negative associations between species are a key outcome of community assembly from regional species pools. These associations are difficult to detect and can be caused by a range of processes such as species interactions, local environmental constraints and dispersal. We integrate new ideas around species distribution modeling, covariance matrix estimation, and network analysis to provide an approach to inferring non‐random species associations from local‐ and regional‐scale occurrence data. Specifically, we provide a novel framework for identifying species associations that overcomes three challenges: 1) correcting for indirect effects from other species, 2) avoiding spurious associations driven by regional‐scale distributions, and 3) describing these associations in a multi‐species context. We highlight a range of research questions and analyses that this framework is able to address. We show that the approach is statistically robust using simulated data. In addition, we present an empirical analysis of > 1000 North American tree communities that gives evidence for weak positive associations among small groups of species. Finally, we discuss several possible extensions for identifying drivers of associations, predicting community assembly, and better linking biogeography and community ecology.