Model‐based decomposition of environmental, spatial and species‐interaction effects on the community structure of common fish species in 772 European lakes

Abstract

Aim We tested whether there is a strong effect of species interactions on assembly of local lake fish communities, in addition to environmental filters and dispersal. Location Seven hundred and seventy‐two European lakes and reservoirs. Time period 1993–2012. Major taxa studied Nineteen species of freshwater fishes. Methods We applied a latent variable approach using Bayesian Markov chain Monte Carlo algorithms (R package “BORAL”). We compared the contributions of six environmental predictors and the spatial organization of 772 European lakes in 209 river basins on the presence/absence of the 19 most frequent fish species and on the biomass and mean mass of the six dominant species. We inspected the residual correlation matrix for positive and negative correlations between species. Results Environmental (50%) and spatial (10%) predictors contributed to the presence/absence assembly of lake fish communities, whereas lake size and productivity contributed strongly to the biomass and mean mass structures. We found highly significant negative correlations between predator and prey fish species pairs in the presence/absence, biomass and mean mass datasets. There were more significantly positive than negative correlations between species pairs in all three datasets. In addition, unmeasured abiotic predictors might explain some of the correlations between species. Main conclusions Strong effects of species interactions on assembly of lake fish communities are very likely. We admit that our approach is of a correlational nature and does not generate mechanistic evidence that interactions strongly shape fish community structures; however, the results fit with present knowledge about the interactions between the most frequent fish species in European lakes and they support the assumption that, in particular, the mean masses of fish species in lakes are modified by species interactions.

Signals of Potential Species Associations Offer Clues about Community Organisation of Stream Fish across Seasons

“…We found that signals of potential species associations (>70% of residual variance) outweighed both the abiotic environmental component and spatial processes in structuring stream fish communities. In this sense, Mehner et al [35] originally proposed that the unexplained residuals in Bayesian Markov chain Monte Carlo algorithms potentially indicate the imprints of species associations; they also found that species interactions have strong effects on fish community structure across European lakes. Although residual correlations are not equivalent to species interactions that are detected under manipulative in-stream experiments per se [32], these species-to-species associations can still provide ecological evidence of biotic constraints and their role in community organisation [35,48].…”

“…In this sense, Mehner et al [35] originally proposed that the unexplained residuals in Bayesian Markov chain Monte Carlo algorithms potentially indicate the imprints of species associations; they also found that species interactions have strong effects on fish community structure across European lakes. Although residual correlations are not equivalent to species interactions that are detected under manipulative in-stream experiments per se [32], these species-to-species associations can still provide ecological evidence of biotic constraints and their role in community organisation [35,48]. Despite this, we re-emphasize that caution is still needed when interpreting these biologically driven signals [49,50].…”

“…Our results broadly agree with Darwin's [51] original idea that biotic interactions contribute to species distributional patterns at subtropical latitudes. For comparison purposes, the influence of these associations on stream fish communities should be further assessed and replicated in experiments and empirical studies at the community level, and across different environmental and biogeographical contexts [35]. An additional reason to anticipate further applications of this approach elsewhere is the relatively weak performance of constrained ordinations explaining compositional variation [52], particularly across freshwater ecosystems where more traditional analytical approaches account for barely 50% of the total variation [53].…”

“…For multiple comparisons between stream orders, the p-values were adjusted using Holm's method [34]. We used the predicted variance of the local environment, spatial processes and unexplained residuals from the MCMC model to infer the relative role of the environmental, spatial and potential biotic associations on fish community organization [35]. All analyses were performed in R 4.1.3 [36].…”

Long‐term effects of rewilding on species composition: 22 years of raptor monitoring in the Chernobyl Exclusion Zone

“…To examine how raptor species composition changed in the CEZ over time, we modeled abundance of breeding pairs for each species for each site (year of data collection), with time as an explanatory variable ( abundance ~ time ). The model was fitted with a negative binomial distribution with two latent variables (the latent variables can be interpreted as a way of accounting for any residual covariation not explained by the covariates; Mehner et al 2021) and an offset (log [ area ]) to account for the different area of the study plot sampled between years.…”