Equilibrium modeling of alpine plant distribution: how far can we go?

Predictive distribution modeling of species and communities has gained much importance in recent years. In this paper, generalized linear models (GLM) are implemented in a Geographical Information System to mimic the spatial distribution of alpine and subalpine species habitat and diversity in the s tudy area of Belalp (Aletsch region, Wallis, Swiss Alps). Quantitative pred ictors used to quantify environmental requirements of species are: annual m ean temperature, slope angle, topographic position, solar radiation, snow c over indices and the three spectral bands of a color infrared aerial photog raph, as well as disjunctive classes of qualitative substrate-related predi ctors. Presence-absence logistic GLM are adjusted for 63 species. Percent g round cover measured on an ordinal scale is additionally modeled using a sp ecial case of GLM for 26 species with significant variation of abundance in the field. Both ordinal abundance and presence/absence at each spatial loc ation are successfully modeled for some species, as shown by quantitative e valuation using an independent data set. Finally, species richness (SR) is modeled by (i) using a Poisson GLM and (ii) summing up single species predi ctions by presence/absence models. Successful models are finally used to mi mic potential impact of climatic change on plant distribution and diversity . Results from these scenarios suggest (i) an overall trend toward a reduct ion of suitable habitat for alpine species and (ii) two different responses for the distribution of SR, namely: (a) a serious shift of the optimal SR elevation belt upward in elevation or (b) the SR optimal belt shifting only slightly upward in elevation, accompanied by a parallel spatial spread out of high SR patches at, these elevations. Limitations of both species and d iversity models are discussed and some suggestions for future research are proposed.