Although agriculture contributes to four main drivers of biodiversity loss, the impact assessment of food products remains limited to either in situ measurements that preclude generalization or to systematic models that are not calibrated on in situ data. Here we describe the BVIAS (Biodiversity Value Increment from Agricultural Statistics) model, which estimates the biodiversity impact of food products using accountancy data and public statistics for use in environmental labeling schemes or other purposes. Going beyond existing methods, BVIAS accounts for the main drivers of biodiversity loss related to agriculture, relies on a large farm dataset (>5000 farms), and is calibrated using in situ data from the literature. We apply it to compare major Food Quality Schemes (FQSs) to their conventional counterparts. We show that only explicit requirements (e.g., ban on pesticides, grass-fed content) in FQS specifications lead to significant differences in practice. Consistent with the literature, we find that organic farms have a lower biodiversity impact on a per-hectare basis, as well as those producing Comté (Protected Designation of Origin), but lower yields offset this local benefit, resulting in a higher impact per ton. However, the biodiversity impact gap between product types (here milk vs. cereals) is far greater than the difference between FQS and conventional versions of the same product. This study highlights that, for environmental labeling, the distinction between product types is more important than the distinction between FQS and conventional. • BVIAS estimates biodiversity impact of 10 products across 5528 farms using FADN. • Model calibrated using in situ data covering the main drivers of biodiversity loss. • Only explicit FQS requirements lead to notable practice differences. • Organic and some PDO farms exhibit lower per-hectare but higher per-ton impact. • Impact gap is much greater across product types than between FQS and conventional.
Huet et al. (Thu,) studied this question.