Aim: Assess the potential of new predictors (land use, edaphic factors and high-resolution topographic and climatic variables, i.e., topo-climatic) to improve the prediction of plant community functional traits (specific leaf area, vegetative height and seed mass) and species richness in models of mountain grasslands.
Location: The western Swiss Alps
Methods: Using 912 grassland plots, we constructed predictive models for community-weighted means of plant traits and species richness using high resolution (25 m) topo-climatic predictors traditionally used in previous modelling studies in this area. In addition, 78 new plots were sampled for evaluation and error assessment in four narrower sets of homogenous conditions based on predictions by the topo-climatic models within two elevation belts (montane and alpine). New, finer-scale predictors were generated from direct field measurements or very high-resolution (5 m) numerical data. We then used multimodel inference to test the capacity of these finer predictors to explain part of the residual variance in the initial topo-climatic models.
Results: We showed that the finer-scale predictors explained up to 44% of the residual variance in the classical topo-climatic models. The very high-resolution topographic position, soil C/N ratio and pH performed notably well in our analysis. Land use (farming intensity) was highlighted as potentially important in montane grasslands, but improvements were only significant for species richness predictions.
Main conclusions: Compared with classical topo-climatic models, the new, finer-scale predictors significantly improved the prediction of all traits and species richness in alpine plant communities and that of specific leaf area and richness in montane grasslands. The differences in the importance of the predictors, dependent on both trait and position along the elevation gradient, highlight the different factors that shape the distribution of species and communities along elevation gradients.