Are redundancy indices redundant? An evaluation based on parameterized simulations

Functional redundancy is considered a major component of the insurance mechanism, which theoretically maintains ecosystem stability by preventing the loss of ecosystem functions with species loss. Over the past decades, examination of functional trait patterns to elucidate processes of community stability and ecosystem functioning have stimulated considerable amount of research in ecology. As a result, a multitude of indices have been developed, describing community functional structure with various levels of overlap in their methodology. Here, we review the set of indices that have been suggested to measure the level of redundancy in traits among species in ecological communities. We first evaluate the correlations among redundancy indices and classical indices of community taxonomic and functional structure (species richness, Simpson diversity, functional richness, evenness and divergence). Second, we estimate the predictive power of these indices in terms of community vulnerability to species loss. Finally, we assess the sensitivity of the results to scenarios with different species loss orders. We simulated communities with different levels of taxonomic and functional structure (richness, evenness and divergence). Then, we simulated four scenarios of species loss order (abundance, functional uniqueness, environmental sensibility and random). The vulnerability of communities was estimated by the changes in community structural parameters (functional richness, functional divergence and biomass) as species were progressively removed from the initial communities. Our results showed that four out of the five redundancy indices tested were strongly correlated (Pearson R > ∣0.6∣) with at least one of the classical indices of community structure. Those correlations partly explained why the redundancy indices did not outperform classical indices in predicting community vulnerability to species loss. The fifth redundancy index (FredD) was the least correlated with classical indices of community structure (Pearson R < ∣0.24∣), however it also reached the lowest performance in predicting community vulnerability to species loss (R2 < 0.07). The order in which species – and their functional traits – are lost, and the community parameters assessed had a strong impact on communities’ vulnerability to species loss. In contradiction to theoretical predictions of the insurance hypothesis, the current redundancy indices might not be good indicators of vulnerability to species loss.We are grateful to the editor and the two anonymous reviewers for their valuable comments that contributed to improve the manuscript. The study was supported by Czech Science Foundation grant GA16‐15012S. F.d.B. was supported by the Agencia Estatal de Investigacion (Plan Nacional de I+D+i, project PGC2018-099027-B-I00). C.P.C. was supported by the Estonian Research Council (project PSG293). E.V. was funded by the 2017 program for attracting and retaining talent of Comunidad de Madrid (no. 2017‐T2/ AMB‐5406).Peer reviewe