Doing the same thing over and over again and getting the same result: Assessing variance in wetland invertebrate assemblages

Past efforts to explain variation of invertebrate assemblages in freshwater wetlands have been less productive than anticipated. To explore why efforts are disappointing, we assembled large invertebrate data sets from North Dakota prairie potholes, California rock pools, and Georgia Carolina bay wetlands that addressed spatial (among wetlands) and temporal (among seasons and years) variation. We anticipated that these large data-set sizes would enable robust conclusions to be drawn, and each place had unique environmental conditions that might contribute to greater explanatory power. We used statistical techniques that partitioned variation in invertebrate assemblages into spatial and/or temporal components, and that also yielded a measure of the amount of unexplained variation; Permutational Multivariate Analysis of Variation and Principal Coordinates Analysis assessed whole assemblage variation, and Analysis of Variance or Analysis of Covariance assessed variation in taxon richness, total abundances, and abundances of wide-spread individual taxa. Across all locations, variation explained by spatial and temporal factors, and unexplained variation were of comparable magnitudes (i.e., similar R² values of ~ 50%). Review of other published studies indicate that this pattern is widespread. The 50% or more unexplained variation is typically ignored by researchers, who instead focus on explained fractions. We argue that, besides addressing explained spatial and temporal variation in invertebrate assemblages (e.g., control by hydrology, resources, predation), efforts to understand what contributes to currently unexplained variation, that is unrelated to local spatial or temporal controls (e.g., broad climatic and biogeographic patterns, organism physiology and behavior), will lead to a fuller comprehension of how invertebrates in freshwater wetlands are controlled.