On the quality of soil biodiversity indicators: abiotic and biotic parameters as predictors of soil faunal richness at different spatial scales

Direct measurement of soil biodiversity is expensive, and therefore a substitution of measurement by indication is desirable. We analysed three large datasets for the potential to predict the diversity of soil faunal groups from other parameters. The datasets represent different spatial scales, namely grassland nematodes on an European scale, forest collembola on a regional scale, and grassland ants on a local scale. We tested two groups of parameters as possible surrogates for species richness: (1) environmental parameters, such as climate, soil and vegetation characteristics, and (2) community parameters, such as higher taxon richness, indicator taxa, and maximum dominance. Climate and soil parameters were significantly correlated with biodiversity in all datasets. However, in spite of the large variety of measurement types analysed, prediction quality of environmental variables was weak and the explained proportion of variance ranged generally below 50%. Richness was subject to considerable stochastic variation among subsamples (CV=20-60%) thereby evading a narrow correlation with environmental parameters. Higher taxon richness, based on taxa of intermediate hierarchical order, proved the best predictor of richness in collembola and nematodes explaining 55 and 89% of total variance, respectively. A combination of two ant species was the best predictor of ant richness explaining 59% of total variance. The authors conclude that a rough guess of soil faunal diversity can be cost-effectively derived from environmental data while an estimate of moderate quality can be obtained with reduced taxonomic effort. The precise richness of a soil community, however, is subject to autogeneous community dynamics, to biotic interactions with other populations, and to conditions in the past, and can therefore only be retrieved by immediate investigation of the community itself. Criteria for the quality of indicator parameters are discussed.