Abstract Aim Anthropogenic changes in land use may have major consequences for global biodiversity. However, species diversity is determined by a suite of factors that may affect species differently at different spatial scales. We tested the combined effects of land use and spatial scale on α, β and γ diversity in the tropics using experimental communities of cavity-nesting bees and waSPS (Hymenoptera: Aculeata). We aimed to determine whether: (1) land-use intensity negatively affects species richness of cavity-nesting Hymenoptera, (2) β diversity, both within and between plots, is higher in more natural systems, (3) species richness of flowering herbs correlates positively with species richness of Hymenoptera within and across habitats, (4) richness of cavity-nesting Hymenoptera in highly modified habitats declines with increasing distance from natural or semi-natural habitats, (5) the effects of land use, herb diversity and forest distance on Hymenoptera α and β diversity vary at different spatial scales, and (6) bees and waSPS respond to land use in a similar way. Location Manabi, south-west Ecuador. Methods We examined diversity (species richness) within 48 plots of five habitat types that comprised a gradient of decreasing agricultural intensity from rice and pasture to coffee agroforests, unmanaged abandoned agroforests and forest fragments, using standardized nesting resources for reproducing communities of cavity-nesting bees and waSPS. Results (1) Land use significantly affected α diversity of trap-nesting bees and waSPS at the subplot (per trap) scale, but not subplot β diversity or plot-scale species richness (γ diversity). (2) Beta diversity was surprisingly higher between plots within a land-use type than between land-use types. (3) Species richness of bees and waSPS increased with diversity of flowering herbs at the subplot (trap) scale only. (4) Forest distance correlated positively with bee species richness at the plot scale only. (5) Land use, herb diversity and forest distance each showed significant correlations with bee and wasp diversity at only one spatial scale. (6) Despite differences in life history, bees and waSPS responded to land-use intensity in a similar way. Main conclusions The effects of land use on species richness were highly dependent on spatial scale. Subplot-scale analyses showed that rice and pasture contained the highest species diversity, whereas plot-scale analyses showed no significant difference in the diversity of different land-use types. We emphasize caution in the estimation of biodiversity at only one spatial scale, and highlight the surprisingly large contribution of managed land to the regional biodiversity of these species.