Imperfect detection of terrestrial mammals in environmental impact assessment (EIA) baseline surveys may compromise decision-making and mitigation measures

Abstract

The Environmental Impact Assessment (EIA) is critical for managing human impacts on biodiversity. Reliable baseline data is essential to assess potential development effects, while inaccurate information about species presence or absence can lead to poor decisions. We examined how methodological choices, such as sampling methods (i.e., camera traps, census, indirect sign surveys, interviews with locals), affect species detection in baseline biodiversity surveys for EIA in mining projects and scientific inventories, focusing on the Iron Quadrangle region of Minas Gerais, Brazil. We employed occupancy models, which consider imperfect detections, to assess how study type and methodological attributes influence false-positive and true detections of medium to large-sized terrestrial mammals. Our analysis revealed that study type strongly predicted false positives, with a potential additive effect with sampling method. In EIA baseline surveys, sign surveys registered 2.1 % false positives, rising to 4.4 % for interviews, while scientific studies had nearly zero false positives. For true detections, we found an interaction between study type and sampling method, where species census, camera traps, and sign surveys were up to three times less likely to detect species in EIA surveys compared to scientific studies. This suggests that EIA characteristics may reduce correct species detection. Both false-positive and true detections may be influenced by the inadequate quality of EIA baseline surveys. This underscores the need to incorporate detection estimates into biodiversity surveys. If studies fail to account for detection probability they can lead to biased and misleading results, which in the case of baseline surveys, could result in unfounded decisions within the EIA process.