Temporal co-occurrence in a community of subtropical Australian birds

Abstract

Abstract With a view to creating a biodiversity baseline to monitor a rewilding effort, I estimated the community structure of birdlife in subtropical South-East Queensland, Australia. Community structure was derived from patterns of temporal species co-occurrence and principles from network theory. In summary, 88 avian species were identified at Hiddenvale Research Station on 10 dates spanning 3 seasons in 2019. Rarefaction-extrapolation and Chao2 coverage estimation indicates comprehensive (94%) sampling, with the resultant assemblage being highly segregated (checkerboard C-score SES = + 4.8 relative to a fixed–fixed null model). The inferred co-occurrence network possesses a non-random highly dispersed degree structure that significantly departs from not only random Erdős-Rényi graphs but also the ‘scale-free’ connectivity distribution that is sometimes observed in co-occurrence networks. Raptors were significantly over-represented among the most negative pairwise correlation scores (hypergeometric enrichment P < 0.001). This indicates their presence is typically associated with reduced occurrence of many other species. On the other hand, analysis of the topology of the inferred co-occurrence network found that Silvereyes ( Zosterops lateralis ) possess the highest number of significant positive pairwise connections and the highest radiality (or ‘network influence’). Collectively, this suggests the presence of Z. lateralis indicates an enhanced likelihood of observing a wide diversity of other bird species, particularly diminutive, non-threatening species such as little friarbirds ( Philemon citreogularis ) and brown honeyeaters ( Lichmera indistincta ). This study illustrates the value of applying the Partial Correlation and Information Theory (PCIT) algorithm to ecological co-occurrence analysis, enabling inference of network topology without assuming scale-free structure and providing an unbiased representation of community organisation.