1. Quantifying how biological diversity is distributed in the landscape is one of the central themes of conservation ecology. For this purpose, landscape classifications are being intensively used in conservation planning and biodiversity management, although there is still little information about their efficacy.
2. I used data from 158 running water sites in Hungary to examine the contribution of six a priori established habitat types to regional level diversity of fish assemblages. Three community measures [species richness, diversity (Shannon, Simpson indices), assemblage composition] were examined at two assemblage levels (entire assemblage, the native assemblage). The relative role of non-native species was quantified to examine their contribution to patterns in diversity in this strongly human influenced landscape.
3. Additive diversity partitioning revealed the primary importance of beta diversity (i.e. among-site factors) to patterns in species richness. Landscape-scale patterns in species richness were best explained by between-habitat type (beta2: 41.2%), followed by within-habitat type (beta1: 37.7%) and finally within-site (alpha: 21.1%) diversity. Diversity indices showed patterns different from species richness, indicating the importance of relative abundance distributions on the results. Exclusion of non-natives from the analysis gave similar results to the entire-assemblage level analysis.
4. Canonical analysis of principal coordinates, complemented with indicator species analysis justified the separation of fish assemblages among the habitat types, although classification error was high. Multivariate dispersion, a measure of compositional beta diversity, showed significant differences among the habitat types. Contrary to species diversity (i.e. richness, diversity indices), patterns in compositional diversity were strongly influenced by the exclusion of non-natives from the analyses.
5. This study is the first to quantify how running water habitat types contribute to fish diversity at the landscape scale and how non-native species influence this pattern. These results on riverine fish assemblages support the hypothesis that environmental variability (i.e. the diversity of habitat types) is an indication of biodiversity and can be used in large-scale conservation designs. The study emphasises the joint application of additive diversity partitioning and multivariate statistics when exploring the contribution of landscape components to the overall biodiversity of the landscape mosaic.