SEARCH

SEARCH BY CITATION

Reliably predicting vegetation distribution requires habitat distribution models (HDMs) that are ecologically sound. Current correlative HDMs are increasingly criticized because they lack sufficient functional basis. To include functional information into these models, we integrated two concepts from community ecology into a new type of HDM. We incorporated: 1) species selection by their traits in which only those species that pass the environmental filter can be part of the community (assembly theory); 2) that the occurrence probability of a community is determined by the extent to which the community mean traits fit the required traits as set by the environment. In this paper, our trait-based HDM is presented and its predictive capacity explored.

Our approach consists of two steps. In step 1, four plant traits (stem-specific density and indicator values for nutrients, moisture and acidity) are predicted from four dominant environmental drivers (disturbance, nutrient supply, moisture supply and acidity) using regression. In step 2, these traits are used to predict the occurrence probability of 13 vegetation types, covering the majority of vegetation types across the Netherlands. The model was validated by comparison to the observed vegetation type for 263 plots in the Netherlands. Model performance was within the range of conventional HDMs and decreased with increasing uncertainty in the environment-trait relationships and with an increasing number of vegetation types.

This study shows that including functionality into HDMs is not necessarily at the cost of model performance, while it has several conceptual advantages among including an increased insight in the functional characteristics of the vegetation and sources of unpredictability in community assembly. As such it is a promising first step towards more functional HDMs. Further development of a trait-based HDM hinges on replacing indicator values by truly functional traits and the translation of these relationships into mechanistic relationships.