The impact of fish predators and dewatering on stochastic vs. deterministic assembly processes in aquatic invertebrate communities

Background/Question/Methods

Community assembly studies show that competition, predation, and disturbance events (deterministic processes) create “ecological filters” that eliminate species lacking adaptive traits.  This leads to species assemblages with low variation among similar habitats (low β-diversity). In contrast, random colonization events (stochastic processes) may increase species variation among similar habitats (high β-diversity) because most individuals that colonize could survive. Furthermore, deterministic processes may be more important when they have strong selective pressure (i.e. harsh environments, effective predators), but stochastic processes may become more important if deterministic processes are weak. However, most studies test only a single factor, making it impossible to compare factors directly or test interactions.

We used a factorial design to compare effects of fish predation and dewatering on aquatic invertebrates community assembly in wetland mesocosms. Five mesocosms (10mX20m) were dewatered for 2-3 weeks twice each year (=Harsh treatment), and five were always flooded (=Benign treatment). We also sampled fish exclosures (=Fishless treatment) and unmanipulated areas (=Fish treatment) in each mesocosm.  Patterns of β diversity were examined with non-metric dimensional scaling (NMDS) ordinations, permutation analysis of variance (PermANOVA), and analysis of multivariate homogeneity of group dispersions. Indicator species analysis (ISA) was also used to assess taxa association with treatments.

Results/Conclusions

NMDS ordinations displayed a prominent difference in invertebrate communities between Harsh vs. Benign treatments. Communities in Fish vs. Fishless treatments were also different, but only in Benign treatments. ISA showed Bezzia, Chydoridae, Cyclopidae, Daphniidae, Hydrachnidia and Tanypodinae were associated with Benign treatment, Chironomini was associated with Fish treatment, and no taxa were associated with Harsh or Fishless treatments. Two-way PermANOVA found that the communities differed between Harsh vs. Benign but were not different between Fish vs. Fishless treatments. However, when we used PermANOVA to test the fish treatment effect in Benign treatments alone, its effect was significant. Therefore, fish treatment effect was determined by environment condition, probably because fish populations were decreased by intermittent water-level drawdowns. This also suggests that the environmental factors had a greater influence on community assembly than biotic factors in our wetlands. Thus it would be expected that differences in multivariate dispersion (β-diversity) between the fish treatments would be greater than between the environmental treatments.  Surprisingly, NMDS ordinations showed that multivariate dispersion was not greatly different among any treatments. This suggests that variability due to stochastic factors did not change in response to the intensity of the deterministic factor in our mesocosms.