Thu, Aug 18, 2022: 3:30 PM-3:45 PM
518B
Background/Question/MethodsTo discern community assembly processes driving ecological communities, ecologists have traditionally relied on null models based on pairwise species co-occurrence indices in conjunction with permutation algorithms (that randomize community presence-absence matrices subject to some constraints). However, pairwise co-occurrence metrics ignore potential species interactions involving multiple species, which could reflect functional guilds or motifs composed of multiple species within the community. As a result, null model testing based on pairwise co-occurrence indices may fail to detect these multi-species interactions and lead to Type II error (i.e. false negatives). Here we introduce a novel joint occupancy metric that measures the number of sites occupied by multiple species simultaneously. This metric is applied to combinations of two to n species, where n is the total number of species, and generates a suite of values capturing the whole range of potential species interactions. The resulting joint occupancy values can then be compared to null expectations from those randomized community presence-absence matrices and offer more robust test of the presence of direct pairwise vs. higher-order interactions in community assembly processes.
Results/ConclusionsNull model testing using standard permutation algorithms in conjunction with the proposed joint occupancy metric revealed nine theoretical multi-species co-occurrence archetypes. Two of these theoretical archetypes correspond to Type II errors when null model testing is performed using a pairwise co-occurrence index instead of the whole suite of joint occupancy values (i.e. joint occupancy lies beyond null expectations only when more than two species are considered simultaneously). These null model tests performed on the empirical 289 presence-absence community matrices led to the discernment of more frequent archetypes than others, with three of the theoretical archetypes not reflected in the analyzed community. One archetype in particular revealed Type II errors, with species co-occurring more than expected by chance for more than two species. This suggests the existence of complex facilitative interactions between multiple species in these communities. By avoiding Type II errors, the proposed joint occupancy metric can therefore improve the robustness of inferences derived from null model testing. The identified nine archetypes may further enhance the ecological interpretations of outcomes from such tests. We therefore advocate for the use of joint occupancy metric in inferring structures of biotic interactions that drive ecological community patterns of (multi-)species co-occurrence using null models.
Results/ConclusionsNull model testing using standard permutation algorithms in conjunction with the proposed joint occupancy metric revealed nine theoretical multi-species co-occurrence archetypes. Two of these theoretical archetypes correspond to Type II errors when null model testing is performed using a pairwise co-occurrence index instead of the whole suite of joint occupancy values (i.e. joint occupancy lies beyond null expectations only when more than two species are considered simultaneously). These null model tests performed on the empirical 289 presence-absence community matrices led to the discernment of more frequent archetypes than others, with three of the theoretical archetypes not reflected in the analyzed community. One archetype in particular revealed Type II errors, with species co-occurring more than expected by chance for more than two species. This suggests the existence of complex facilitative interactions between multiple species in these communities. By avoiding Type II errors, the proposed joint occupancy metric can therefore improve the robustness of inferences derived from null model testing. The identified nine archetypes may further enhance the ecological interpretations of outcomes from such tests. We therefore advocate for the use of joint occupancy metric in inferring structures of biotic interactions that drive ecological community patterns of (multi-)species co-occurrence using null models.