Does certification improve biodiversity conservation in Brazilian coffee farms?

•Certified coffee farms already had a different conservation profile.•Certification produced subtle changes in habitat availability and connectivity.•Certified farms had better conservation profile changes than the surrounding landscape.•Certification was important to control deforestation and agricultural expansion.•Coffee certification has a prominent importance in conservation of Cerrado biome. Socio-environmental certification uses evaluation criteria to promote the conservation of the natural environment and landscape connectivity, with the aim of constructing agricultural landscapes more suitable for biodiversity conservation. To test this, we examine whether socio-environmental certification of Brazilian coffee farms contributes to local conservation, particularly in terms of deforestation control, habitat protection and regeneration, and connectivity. The analysis compared changes in landscape structure and connectivity in certified farms before (1995–2002) and after nine years from the beginning of the certification process (2002–2011), using as a reference the surrounding landscape and a control group of non-certified farms. To quantify changes in landscape connectivity we used probabilistic indices of functional connectivity based on graph theory, and two species of terrestrial mammals with contrasting dispersal capacities and habitat requirements: Priodontes maximus (giant armadillo) and Marmosops incanus (gray slender mouse opossum). Our results show that changes in the last decade have been subtle, but that certified farms differ from surrounding areas for the greater deforestation control and habitat availability for both land cover types, and for the greater connectivity for P. maximus. The difference between certified and non-certified farms is not clear-cut, however, we have evidence that the certified farms contributed more than the surrounding areas to the conservation of the studied species when the balance of gains and losses of connectivity is considered. The subtle differences in temporal changes and groups might be partially explained by the fact that certified farms already had a different conservation profile at the beginning of the certification process. Despite the limitations in the sampling size (small number) and time scale (only nine years after certification) which may hinders the detection of certification effects, our findings indicate that certification was important in controlling deforestation and the conversion of ne