Fridley et al. (2007) introduced an algorithm estimating habitat specialization of plant species using information from large species co-occurrence datasets (typically composed of vegetation samples). The measure reflects betadiversity among vegetation samples containing given species: lower betadiversity is considered as indication of habitat specialist (species is restricted to rather narrow range of habitats), while higher betadiversity as indicator of habitat generalist (species occupies wide range of habitats). I argue that proposed metric is influenced not only by real width of species niche along environmental gradient (species habitat specialization), but also by species dispersal ability. To support this idea, I used 1) simulated data, with model including information about species niche width and species dispersal ability, and 2) real data from forest vegetation (in the Czech Republic) combined with information about plant functional traits related to dispersal. The main question is: how is the measure of species habitat specialization proposed by Fridley et al. (2007) related to real species habitat requirements on one side and to species dispersal ability on the other?

Results/Conclusions

Results of both simulated and real data analysis indicate that studied measure of habitat specialization based on co-occurrence data is related to both real species niche along ecological gradient and species dispersal ability simultaneously. In forest understory, habitat specialists, as defined by the metric, are mainly shade tolerant species with low dispersal ability and niche limited distribution. Contrary to this, habitat generalists are light demanding species with high dispersal ability and dispersal limited distribution.The studied metric of species habitat specialization more than habitat specialization itself reflects the position of species along a virtual gradient between deterministic and neutral species behavior, or between niche limited and dispersal limited species distribution, respectively.