Geographic range size predicts species’ responses to land-use change and intensification, but the reason why is not well established because many correlates of larger geographic ranges, such as realized niche breadth, may mediate species’ responses to environmental change. Agricultural land uses (hereafter 'agroecosystems') have warm, dry and more variable microclimates than do cooler and wetter mature forests, so are predicted to filter for species that have warmer, drier and broader fundamental and realized niches. To test these predictions, we estimated species' realized niches, for temperature and precipitation, and geographic range sizes of 764 insect species by matching GBIF occurrence records to global climate layers, and modelled how species presence/absence in mature forest and nearby agroecosystems depend on species’ realized niches or geographic ranges. The predicted species niche effects consistently matched the expected direction of microclimatic transition from mature forest to agroecosystems. We found a clear signal that species with preference for warmer and drier climates were more likely to be present in agroecosystems. In addition, the probability that species occurred in different land-use types was predicted better by species’ realized niche than their geographic range size. However, niche effects are often context-dependent and varied amongst studies, taxonomic groups and regions used in this analysis: predicting which particular aspects of species’ realized niche cause sensitivity to land-use change, and the underpinning mechanisms, remains a major challenge for future research and multiple components of species’ realized niches may be important to consider. Using realized niches derived from open-source occurrence records can be a simple and widely applicable tool to help identify when biodiversity responds to the microclimate component of land-use change.