tLand use effects are considered among the main stressors on freshwater biodiversity, with up to 80%of land in Europe under intensive use. Here, we address the impact of arable and urban landscapes ontaxon richness, Shannon–Wiener diversity, taxon rareness and taxonomic distinctness of eleven organ-ism groups encompassing vertebrates, invertebrates and plants, occurring in five freshwater ecosystemtypes across Europe: rivers, floodplains, lakes, ponds and groundwater. In addition, nine geo-climaticdescriptors (e.g. latitude, longitude, precipitation) were used to disentangle land use effects from thoseof natural drivers of biodiversity. Using a variance partitioning scheme based on boosted regression treesand generalised linear regression modelling, we sought: (i) to partition the unique, shared and unex-plained variation in the metrics explained by both groups of descriptor variables, (ii) to quantify thecontribution of each descriptor variable to biodiversity variation in the most parsimonious regressionmodel and (iii) to identify interactions of land use and natural descriptors. The variation in biodiver-sity uniquely described by land use was consistently low across both ecosystem types and organismgroups. In contrast, geo-climatic descriptors uniquely, and jointly with land use, explained significantlymore variance in all 39 biodiversity metrics tested. Regression models revealed significant interactionsbetween geo-climatic descriptors and land use for a third of the models, with interactions accountingfor up to 17% of the model’s deviance. However, no consistent patterns were observed related to thetype of biodiversity metric and organism group considered. Subdividing data according to the strongestgeo-climatic gradient in each dataset aimed to reduce the strength of natural descriptors relative to landuse. Although data sub-setting can highlight land use effects on freshwater biodiversity, sub-setting ourdata often failed to produce stronger land use effects. There was no increase in spatial congruence in thesubsets, suggesting that the observed land use effects were not dependent on the spatial extent of thesubsets. Our results confirm significant joint effects of, and interactions between, land use and naturalenvironmental descriptors on freshwater biodiversity, across ecosystem types and organism groups. Thishas implications for biodiversity monitoring. First, the combined analysis of anthropogenic and naturaldescriptors is a prerequisite for the analysis of human threats to biodiversity. Second, geo-climatically, butnot necessarily geographically more homogeneous datasets can help unmask the role of anthropogenicdescriptors. And third, whole community-based biodiversity metrics (including taxon richness) are notideal indicators of anthropogenic effects on biodiversity at broad scales. ; JRC.H.1-Water Resources