Disturbance reinforces community assembly processes differentially across spatial scales.

ABSTRACT

BACKGROUND AND AIMS: There is a paucity of empirical research and a lack of predictive models concerning the interplay between spatial scale and disturbance as they affect the structure and assembly of plant communities. We proposed and tested a trait dispersion-based conceptual model hypothesizing that disturbance reinforces assembly processes differentially across spatial scales. Disturbance would reinforce functional divergence at the small scale (neighbourhood), would not affect functional dispersion at the intermediate scale (patch) and would reinforce functional convergence at the large scale (site). We also evaluated functional and species richness of native and exotic plants to infer underlying processes. Native and exotic species richness were expected to increase and decrease with disturbance, respectively, at the neighbourhood scale, and to show similar associations with disturbance at the patch (concave) and site (negative) scales. METHODS: In an arid shrubland, we estimated species richness and functional dispersion and richness within 1 m2 quadrats (neighbourhood) nested within 100 m2 plots (patch) along a small-scale natural disturbance gradient caused by an endemic fossorial rodent. Data for the site scale (2500 m2 plots) were taken from a previous study. We also tested the conceptual model through a quantitative literature review and a meta-analysis. KEY RESULTS: As spatial scale increased, disturbance sequentially promoted functional divergence, random trait dispersion and functional convergence. Functional richness was unaffected by disturbance across spatial scales. Disturbance favoured natives over exotics at the neighbourhood scale, while both decreased under high disturbance at the patch and site scales. CONCLUSIONS: The results supported the hypothesis that disturbance reinforces assembly processes differentially across scales and hampers plant invasion. The quantitative literature review and the meta-analysis supported most of the model predictions.