Functional redundancy in assemblages may insure ecosystem processes after perturbation potentially causing temporary or permanent local species extinctions. Yet, functional redundancy has only been inferred by indirect evidence or measured by methods that may not be the most appropriate. Here, we apply an existing method to measure functional redundancy, which is the fraction of species diversity not expressed by functional diversity, to assess whether functional redundancy affects community resilience after disturbance.
Subtropical grassland, south Brazil (30°05′46″S, 51°40′37″W).
Species traits and community composition were assessed in quadrats before grazing and after community recovery. Grazing intensity (G) was measured in each quadrat. We used traits linked to grazing intensity to define functional redundancy (FR) as the difference of Gini–Simpson index of species diversity (D) and Rao's quadratic entropy (Q). Also, with the same traits, we defined community functional stability (S) as the similarity between trait-based community composition before grazing and 47 and 180 d after grazing ending. Using path analysis we assessed different postulated causal models linking functional diversity (Q), functional redundancy (FR), grazing intensity (G) and community-weighted mean traits to community stability (S) under grazing.
Path analysis revealed the most valid causal model FR → S ← G, with a significant positive path coefficient for FR → S and a marginally significant negative one for S ← G. Since FR and G were independent in their covariation and in their effects on S, the model discriminated community resistance to grazing (the effect of G on S) from community resilience after grazing caused by functional redundancy (indicated by the effect of FR on S).
We show that expressing functional redundancy mathematically is a useful tool for testing causal models linking diversity to community stability. The results support the conclusion that functional redundancy enhanced community resilience, therefore corroborating the insurance hypothesis.