Zusammenfassung
Despite recent modelling approaches integrating the effects of niche-based processes and dispersal-based processes on local plant species composition, their relative importance is still not clear. We test whether the predictability of local species composition from environmental conditions is influenced by dispersal traits. We analyzed a large database with co-occurrence data, using ordination ...
Zusammenfassung
Despite recent modelling approaches integrating the effects of niche-based processes and dispersal-based processes on local plant species composition, their relative importance is still not clear. We test whether the predictability of local species composition from environmental conditions is influenced by dispersal traits. We analyzed a large database with co-occurrence data, using ordination techniques (DCA and CCA) to identify the major environmental determinants of species composition. The percentage of explained variance in occurrence was quantified for individual species with CCA. Effects of life-history traits on the predictability of occurrence patterns were tested by means of regression analysis, using a generalized linear models approach. The results reveal close correlations between species composition and environmental conditions, implying that the predictability of the set of species that might occur in a given environmental setting ("habitat species pool") is high. The habitat species pool, however, reflects the potential species composition, and not the actual local situation. At the level of individual species, a large proportion (>90%) of the variation in occurrence remained unexplained. Predictability of species occurrence patterns was increased by a greater capacity for long-distance dispersal, greater adult longevity and the capacity to build a persistent seed bank. The results indicate that the predictability of species composition from environmental conditions is reduced by a few orders of magnitude by dispersal limitation and that poor dispersers are underrepresented.
Altmetric