Plant trait diversity in many vegetation models is crudely represented using a discrete classification of a handful of 'plant types' (named plant functional types; PFTs). The parameterization of PFTs reflects mean properties of observed plant traits over broad categories ignoring most of the inter- and intraspecific plant trait variability. Taking advantage of a multivariate leaf-trait distribution (leaf economics spectrum), as well as documented plant drought strategies, we generate an ensemble of hypothetical species with coordinated attributes, rather than using few PFTs. The behavior of these proxy species is tested using a mechanistic ecohydrological model that translates plant traits into plant performance. Simulations are carried out for a range of climates representative of different elevations and wetness conditions in the European Alps. Using this framework we investigate the sensitivity of ecosystem response to plant trait diversity and compare it with the sensitivity to climate variability. Plant trait diversity leads to highly divergent vegetation carbon dynamics (fluxes and pools) and to a lesser extent water fluxes (transpiration). Abiotic variables, such as soil water content and evaporation, are only marginally affected. These results highlight the need for revising the representation of plant attributes in vegetation models. Probabilistic approaches, based on observed multivariate whole-plant trait distributions, provide a viable alternative.
Keywords: aggregation biases; carbon and water dynamics; fallacy of averages; inter- and intraspecific plant trait diversity; plant functional types (PFTs); plant trait spectra; vegetation modeling.
© 2015 The Authors. New Phytologist © 2015 New Phytologist Trust.