Checkerboard metacommunity structure: an incoherent concept.

Checkerboards have emerged as a metaphor to (1) describe mutually exclusive patterns of co-occurrence for ecologically similar species that are geographically interspersed (i.e., checkerboard distributions), and (2) characterize relationships among species distributions along gradients that involve entire metacommunities (i.e., checkerboard metacommunity structure). Critical differences exist in the conceptual foundations that characterize these patterns. Checkerboard distributions are characterized by mutual exclusion of geographically interspersed species, usually pairs of ecologically similar species for which competition prevents syntopy. In contrast, checkerboard metacommunity structures are more restrictive: groups of species must exhibit mutually exclusive distributions, and each of these groups must be spatially independent of all other groups. Consequently, in a checkerboard metacommunity, competition defines one relationship for each species (i.e., that with its mutually exclusive partner), whereas independence characterizes all other interspecific associations. Consequently, a structure designed to be consistent with this concept will conclude that the metacommunity has random rather than checkerboard structure. Indeed, empirical checkerboard metacommunities are quite rare (7 of 766 reported empirical structures), and likely arise because of poor characterization of species ranges due to detection errors (i.e., a preponderance of rare or hard-to-detect species), rather than from underlying ecological mechanisms. Importantly, no ecological mechanism has been identified that is consistent with the concept of negative coherence. Consequently, the evaluation of checkerboards should be restricted to small sets of ecologically similar species for which interspecific interactions may lead to mutual exclusion, and coherence should be used only to evaluate if species distributions are more coherent than expected by chance (i.e., one-tailed tests).