Species richness of vascular plants in the subarctic landscape of northern Finland: modelling relationships to the environment

This paper presents models based on empirical data which can be used to predict the patterns of species richness of vascular plants at the poorly explored mesoscale. Using generalized linear modelling, multiple regression models of species richness in the Kevo Nature Reserve, North Finland, are built with a training set of 257 grid squares and 33 environmental variables. We validated the accuracy of the derived models with an independent test set of 100 grid squares. Two different modelling approaches are used: one where species richness is treated straightforwardly as the response variable, and another where it is tentatively stratified into two groups according to taxon types, i.e. alpine taxa versus wide-spread and silvine (forest) taxa. However, the latter approach only marginally improved the accuracy of the predictions of total number of species. Linear altitudinal variables were among the best predictors of vascular plant richness at the mesoscale. Bs variables involving altitude are crude surrogates for energy-related factors, the results support the available energy hypothesis and advocate its significance in richness-environment relationships. Other important predictors of species richness included length of rivers and brooks, abundance of cliff walls, occurrences of steep-sided gorges and valleys, and relative abundance of gabbro in bedrock. However, the accuracy of the predictions in the derived models is relatively modest. This points towards the necessity of field work as a final guarantee to identify local hotspots of vascular plant species in a subarctic landscape.