Can richness patterns of rarities be predicted from mesoscale atlas data? A case study of vascular plants in the Kevo Reserve

This paper presents an empirical model of the mesoscale patterns of the number of rare vascular plant taxa in a Finnish subarctic landscape. A multiple regression model relating the species richness of rarities to several environmental variables is developed using generalized linear models and data from 362 1 x 1 km grid squares. The final model accounts for 60% of the variation in the species data. The results suggest that the local hotspots of rare flora (squares with greater than or equal to 3 rare taxa) are mainly found in topographically heterogeneous grid squares, where high cliffs occur in deep gorges. However, it seems that the empirical models derived from mesoscale atlas data and environmental variables can provide only moderately accurate surrogates for extensive field surveys and fine-scale observations on the distributions of rare taxa. The squares with neither-recorded-nor-expected rarities coincide best, and half of the observed and predicted rarity hotspots match. Predictions are least accurate in squares where one or two rare species have been recorded or are predicted to occur. Potential reasons far the moderate performance of the model and the ecology and habitats of the species in concern are discussed. (C) 1998 Elsevier Science Ltd.