Assessing trade-offs between milk production and greenhouse gas emissions on French dairy farms using vine-copula-based regressions

In the context of climate change, livestock systems have several challenges to meet, including improving their environmental performances and ensuring food security. Because the many factors on farms interact, strategies to decrease one emission may increase another emission or decrease farm production. Since the effectiveness of combined strategies to mitigate emissions at the farm scale cannot be assessed by considering one strategy at a time, whole-farm mitigation plans are preferred. In this context, we aimed to assess the influence of simultaneous changes in management practices on farm production and emissions by considering the multivariate dependence structure among their descriptive components. To this end, we investigated copula-based regressions to explain outputs of farms, considering complex dependence structures among variables. The aim was to assess trade-offs between milk production and emissions as a function of management practices. The method was applied to a dataset of management practices, production, and emissions of 2523 French dairy farms surveyed in 2013. We first fitted copula-based regressions to milk production per cow and total greenhouse gas (GHG) emissions per livestock unit separately, and then assuming dependence of both of them on farm characteristics. Subsequently, we explored whole-farm mitigation scenarios to decrease GHG emissions while maintaining milk production. In these scenarios, the production and environmental goals were always much less likely to be attained when considered together (median probabilities of 0.08-0.20) than when considered independently (median probabilities of 0.27-0.77), which highlights the utility of capturing interactions among practices and outputs of farms to help develop effective mitigation scenarios and identify practices that can optimize trade-offs between production and emissions. • Complex interactions on farms cause the effectiveness of mitigation options to vary. • Vine regressions capture the conditional dependence of farm outputs on practices. • Goals are less likely to be attained when considered together than individually. • Farms with high performances provided more energy to cows while decreasing inputs. • A tool based on the method could assess changes in practices before implementing them.