Co-Occurrence Patterns of Aquatic Macroinvertebrates in Laurentian Great Lakes Coastal Wetlands.

In niche-based community assembly theory, it is presumed that communities in habitats with high natural disturbance regimes are less likely to be structured by competitive mechanisms. Laurentian Great Lakes (hereafter Great Lakes) coastal wetlands can experience drastic diel fluctuations in dissolved oxygen levels, severe wave action, ice scour, and near complete freezing during the winter such that conditions are inhospitable for most organisms. The high natural disturbance levels are thought to cause high interannual turnover for aquatic macroinvertebrate communities and support the hypothesis that these communities are less likely to experience less competitive interactions and negative co-occurrence structure. We hypothesize that non-random co-occurrence patterns will be rare in Great Lake coastal wetlands and non-competitive processes (e.g., through shared or differential microhabitat affinities, pollution tolerances, or biotic homogenization) will be more common than competitively driven negative co-occurrence patterns. Null model analysis was performed on 134 macroinvertebrate communities sampled from across the Great Lakes basin from 2000 to 2013. To disentangle the effects of alternative structuring mechanisms (i.e., shared/differential habitat affinities, shared/differential pollution tolerance, and biological homogenization/competitive exclusion), communities were parsed based on the year sampled, the vegetation type from which community samples were collected, and lastly species' functional feeding group assignment or taxonomic group. As expected, very few communities were non-randomly structured; however, all of those that were non-random exhibited showed more negative co-occurrences than by chance. Upon further investigation, these communities consisted of species that are known to overwinter in wetlands, and therefore, avoid having to recolonize after each spring thaw. With expected changes in habitat conditions due to climate change, we propose that null model analyses can be used as an early warning system for community change.