Biome type and availability of natural vegetation dictate whether farmland intensification or expansion is worse for biodiversity

Abstract To reduce the biodiversity impact of agriculture, increasing yields on existing farmland has been proposed as an alternative to farmland expansion. However, the relative effects of yield increases versus agricultural expansion have mostly been examined regionally, and measured in terms of species persistence—a metric relevant to extinction risk but limited in describing ecological communities and their support for ecosystem services. Without a thorough analysis, the lower biodiversity impacts of agricultural intensification remain largely speculative. This study provides a global assessment of biodiversity responses to land conversion and yield increases, including closing yield gaps. We also compare the biodiversity impacts of expanding farmland versus intensifying yields in agricultural landscapes to achieve a 1% increase in total production. Utilizing a large biodiversity database, natural vegetation data, and agricultural yield estimates at the landscape scale, we assess three biodiversity metrics: species richness, total abundance, and relative community abundance-weighted average range-size (RCAR), which provides a proxy for biotic homogenisation. Our models highlight that land conversion is associated with significant biodiversity loss at both local and landscape scales, emphasizing the importance of avoiding farmland expansion into new landscapes. However, yield also lead to significant biodiversity loss; closing yield gaps is associated with a median species loss of nearly 11%, and median abundance loss of almost 13%, with some agricultural landscapes losing almost 90% of species and more than 90% in abundance. Additionally, 30% of global agricultural landscapes, predominantly in the tropics, are likely to experience increased biotic homogenization. Neither expansion nor intensification is consistently better for biodiversity, with biome type, crop, biodiversity metric, and percentage of natural vegetation influencing which approach is less harmful. Our results suggest that minimising the biodiversity cost of agriculture requires a context-dependent balance between intensification and expansion in agricultural landscapes.