We deployed camera traps in private reserves with a matrix of coffee plantations and forest to determine how land use influences small-scale spatial ecology and species richness patterns of medium-and large-sized mammals. We were interested in three questions: 1) do mammals in the area use forested sites more than coffee plantation sites? 2) how do landscape features (i.e., asphalt roads and distance to protected areas) influence mammalian use of sites and 3) do configuration and management of coffee plantations influence mammalian relative abundances and species richness? We used hierarchical multi-species Bayesian abundance models that account for imperfect detection to model the influence of landscape features and species local relative abundance and occupancy probabilities. We found a strong influence of land use on mammalian occupancy probabilities and relative abundances, as well as on species richness. More species used areas around the camera traps in forest than in coffee plantations and far from asphalt roads. Local relative abundances and species richness estimates were higher in reserves with coffee crops mixed with forests, as opposed to coffee crop surrounded by forests. Our results show to maintain mammalian diversity and abundances, coffee crops should be mixed in with forests and the presence of asphalt roads within these should be avoided.