Calibration of Ellenberg indicator values for the Faroe Islands

Abstract

Ellenberg's bio-indication system for soil moisture (F), soil nitrogen (N) and soil reaction (R) was examined, based on 559 vegetation samples and environmental charac- teristics (vegetation cover, soil depth, soil moisture, chemical soil properties) from four Faroe islands. The original indicator values from central Europe were used for the calculation of weighted community indicator values of F, N and R. These were regressed with respect to environmental data, applying standard curvilinear regression and generalized linear model- ling (GLM) and new predicted values of community indicator values were obtained from the best model. Faroe species optima values of 162 taxa for one or more of the three Ellenberg scales were derived from fitting Huisman-Olff-Fresco (HOF) models of species abundance with respect to predicted com- munity indicator values and are proposed as new Ellenberg species indicator values to be used in the Faroe Islands. F was best correlated with a GLM model containing soil moisture, organic soil fraction, soil depth and total vegetation cover, R with a GLM model containing pH and calcium in % organic soil fraction, N with total phosphorus in % organic soil fraction. The calibrated species indicator scales are much truncated, as compared with the original values, resulting in significantly different overall distributions of the original and new species indicator values. The recalculated community indicator values are much better correlated to environmental measurements. Several species do not have clear optima, but linear or monotone relationships to the examined indicator scales. This probably indicates that the occurrence of some species in the Faroe Islands are either determined by factors other than moisture, pH or soil nutrient status or, given the young age and environ- mental instability of the islands, are governed by stochastic mechanisms. Extension of Ellenberg indicator values outside central Europe should always be carefully calibrated by means of adequate environmental data and adequate statistical models, such as HOF models, should be applied.