TAXONOMIC CO-OCCURRENCE PATTERNS AS A TOOL FOR DETECTING COMPETITIVE INTERACTIONS IN MODERN AND FOSSIL MARINE BENTHIC ECOSYSTEMS

Several lines of evidence suggest that the frequency and intensity of biotic interactions in marine ecosystems has increased through the Phanerozoic. However, direct evidence of biotic interactions, especially competitive interactions, is extremely sparse. A recently developed approach (Lyons et al., 2015, Nature) examined co-occurrence patterns among continental species in a species pool to determine whether there are species pairs that occur together more often than expected by chance (aggregated pairs) or less often than expected by chance (segregated pairs). Here we apply this approach to modern and fossil marine benthic invertebrate datasets. Each dataset consists of a set of collections from the same geographic area (modern datasets) or from the same formation, facies, and geographic area (fossil datasets) that are assumed to sample the same species pool. We hypothesized that competitive exclusion, manifested as segregation, should be more frequent among congeneric species than among more distantly related species that are less likely to compete for the same resources. We find that this is indeed the case for modern benthic datasets. However, no such pattern is evident for the fossil datasets. This difference may be explained by lateral mixing due to transport obliterating the segregation of taxa in the fossil record and/or by larger amount of time that is incorporated in fossil assemblages relative to the modern datasets. We also assessed species co-occurrence patterns from the Cambrian to Recent, hypothesizing that the percentage of all possible species pairs that are significantly aggregated or segregated should increase through time if biotic interactions have indeed become more frequent and intense. However, we did not find Phanerozoic-scale trends as the median percentage of significant pairs of all possible pairs per time unit consistently remains below 10%. Moreover, the great majority of significant pairs are aggregated rather than segregated, and there is no Phanerozoic trend in the ratio of aggregated to segregated pairs. The modern datasets consistently showed similar results for these analyses, suggesting that this pattern may be genuine.