Background/Question/Methods Both stochastic and deterministic ecological filters act upon regional species pool and leave their imprints on local community assembly. Recurrent and non-random co-occurrence patterns among species should reflect common and distinct interactions of species to their biotic and abiotic environment. For example, segregated co-occurrence between species pairs might indicate species’ preference for segregated habitat types or antagonistic interactions between species within the same habitat. Several studies investigating co-occurrence patterns apply null models that randomize species distribution with constraints that maintain only observed species frequency and site-richness. Such “unconstrained” null models, however, do not distinguish the effect of habitat from biotic interactions. In this study we apply a suite of null models that exclude and explicitly include habitat constraints to explain assembly of Saproxylic beetles emerging from logs (n=360) retrieved from 72 stands following forest fire. The logs represent a range of habitat gradients: tree species (2 classes), burn severity (3 classes) and size/dbh (4 classes). Larval development occurs entirely in the host-tree, thus the successful beetle emergence is contingent on properties of local factors, i.e., suitability of a particular log quality and interspecific interactions.

Results/Conclusions Under “unconstrained” null model, 28% of the species pairwise tests (among 22 species that occur in ≥ 5% of stands) were significantly aggregated (46 pairs) or segregated (22 pairs). The habitat-constrained null models indicated that 72% of the significant aggregated associations were attributed to habitat variables, with tree size/dbh (proxy for resource) alone accounting for 39%; and additional 15% shared with tree type or burn severity. Habitat variables explained 60% of the segregated associations, with tree-type alone accounting for 23% of the associations, and additional 22% shared with the other habitat variables. In addition, 17 of the non-significant pairs were significantly aggregated, but were confounded by segregating effect of tree (11), burn (3) and dbh (3). Results of species-habitat associations corroborated these findings.  41% of the segregated and 28% of the aggregated associations could not be explained by habitat variables, and might be attributed to interspecific interactions. In addition, five segregated pairs were revealed within tree-constrained models.  We conclude that while habitat is the principal factor for fine-scale co-occurrence patterns of Saproxylic beetles, there is also strong evidence for interspecific interactions. We suggest habitat-constrained null models could form a basis to distinguish importance of habitat effects from interspecific interactions for community assembly.