Assessing the functional turnover of species assemblages with tailored dissimilarity matrices

It is often suggested that community functional diversity is an appropriate predictive measure of ecosystem functioning, particularly if relevant species traits for the ecological property of interest are carefully selected. However, methods for selecting traits are often based on expert knowledge or on theoretical models of ecosystem functioning, but usually do not include explicitly developed quantitative procedures. Here we propose to construct a so‐called ‘tailored dissimilarity matrix’ between species assemblages to emphasize their functional turnover in response to some user‐defined ecological property. First, a subset of community weighted mean trait values (CWM) is selected by stepwise regression on the ecological process of interest. The selected CWM values are then replaced by the residuals of the least‐squares regressions of each single CWM on the ecological process of interest and pairwise Euclidean distances between the residual values at each sampling site are calculated. We illustrate the advantages of the tailored approach using two distinct plant and bee communities under contrasting fire regimes in temperate forests of southern Switzerland. Our results demonstrated that, unlike for the original CWM values, the tailored approach optimized the degree of functional differentiation among bee and plant species assemblages, i.e. the species functional turnover, with respect to different fire regimes.