A parsimonious combination of functional traits predicting plant response to disturbance and soil fertility

摘要

Questions: 1. How many traits associated with persistence and regeneration are necessary to predict the response of plants to soil fertility and disturbance? 2. Are correlated changes in trait expressions linked to the response of functional groups tofertility and disturbance? Location: Lower Frankonia, Germany. Methods: On 120 plots located in managed and abandoned grasslands, fields, and vineyards, we recorded species composition, disturbance intensity, soil water and nutrients, and ten candidatetraits for 75 species. We used a novel method which is based on three steps: (1) logistic regression to separate responsive from non-responsive species; (2) iterative clustering of all possible combinations of the candidate traits including regression ofeach cluster in response to the environmental variables; (3) selection of the trait combination that performed best with respect to goodness of fit of all clusters from this combination. Bivariate trait relationships across functional groups were analysed with reduced major axis regression (RMA). Results: The parsimonious trait combination consisted of life span, specific leaf area (SLA), canopy height, and seed number. The 'acquisitive' functional groups in terms of SLA and height were linked to higher fertility and earlier disturbance, while the 'retentive' groups related to lower fertility and later disturbance. Investment in reproduction, however, displayed a reverse relation. SLA and canopy height showed correlated shifts in two pairs of co-occurring functional groups. Conclusions: A small number of traits is sufficient to predict the response of species. Plants need a higher increment in SLA to reach the same height, if start of disturbance is earlier. Linkages between traits shift from generative to vegetative with increasing fertility and earlier start of disturbance. Functional groups enable shifts in scaling relationships between traits to be analysed, in contrast to the analysis of single traits.