Intensive, large-scale agriculture promotes the conversion of natural habitats and diversified crops into monocultures, decreasing both native vegetation cover and landscape heterogeneity, leading to landscape simplification. Yet, a key knowledge gap persists on the relative impacts of the loss of native vegetation and landscape heterogeneity on biodiversity. Addressing this gap is pressing to support policies that conciliate agricultural production and biodiversity conservation and to move forward some scientific controversies, as the “land sharing versus land sparing” and “habitat loss versus fragmentation” debates.Through a hierarchical sampling design that maximized variation, while minimizing correlation, between landscape heterogeneity and native vegetation cover, we recorded the occurrence of medium and large-bodied mammals in native vegetation and agricultural areas of 55 landscapes in a global conservation hotspot and a key commodity production area – the Brazilian savanna, Cerrado. We compared simple, additive, and interactive models to investigate the effects of landscape heterogeneity and native vegetation cover on richness and composition of native and invasive mammals. Native and invasive mammal communities were affected by both native vegetation cover and landscape heterogeneity, although the effects of the first was stronger than the later. Both aspects had positive effects on native species richness and negative on invasive species richness, indicating that the loss of native vegetation and the reduction in landscape heterogeneity lead to biotic homogenization. Yet, while landscape heterogeneity benefited most native species, the direction of its effect varied among invasive species and depended on native vegetation cover.Synthesis and applications: Besides reducing habitat loss, avoiding landscape homogenization is key for conciliating agricultural production and biodiversity conservation, pointing to the relevance of policies encouraging crop diversification. As increasing landscape heterogeneity can in part compensate the negative effects of losing native habitat on biodiversity in agroecosystems, policies can gain feasibility by adjusting the balance between native vegetation cover and landscape heterogeneity according to what best suits local restraints and demands.

# Dataset of the article Landscape heterogeneity can partially offset negative effects of habitat loss on mammalian biodiversity in agroecosystems [https://doi.org/10.5061/dryad.f7m0cfz38](https://doi.org/10.5061/dryad.f7m0cfz38) Mammal's species detection in 55 landscapes in the Cerrado Biome in Brazil. ## Description of the data and file structure The dataset has 55 lines, and each sampled landscape (200 ha) is in one line. The columns represent each mammal's species detection (0 or 1) and the native and invasive species richness (number of native or invasive species detected per landscape). The scientific names of the species are indicated in the column names. Three species are invasive: Sus scrofa, Canis lupus familiaris, and Lepus europaeus. The data was collected using camera traps and transects (3 camera stations and 9 transects per landscape). ## Code/Software Analyzes were performed in R environment with 'MuMIn' package (Barton, 2022) and 'psych' package (Revelle, 2022).