Agriculture is the greatest threat to biodiversity across the world and a major contributor to anthropogenic greenhouse gas emissions. Both pressures will increase over coming decades as populations and per capita consumption rise. How we choose to produce food will, to a large extent, determine the state of biodiversity and the wider environment in the 21st century. Balancing livestock production and environmental concerns is of particular importance: rangelands cover approximately one quarter of the world’s ice-free land and livestock consume over one third of all calories from crops. In addition, livestock, particularly ruminants are extremely inefficient and use more land, nitrogen and water than other foodstuffs, whilst producing more CO2. Finally, there is a strong relationship between wealth and meat consumption. Combined with increasing populations, this means that demand for meat is likely to continue to increase. Two alternative strategies have been proposed to minimise the environmental cost of food production: land sharing attempts to maximise biodiversity within the farmed landscape by keeping yields artificially low. Alternatively, land sparing links high yield agriculture with increased habitat protection or restoration. Previous studies have examined the relative benefits of the two strategies in different agricultural systems and regions. However, my research is the first to look at a wide range of alternative livestock production systems in a highly disturbed part of the tropics. To investigate this issue I collected data on the production and requirements of different cattle ranching systems in Yucatán, Mexico, as well as on the population densities of birds, trees and dung beetles, and carbon stocks in both natural habitats and the different ranching systems. I used novel methods to estimate the yields of my study sites and applied both previously developed density-yield functions and new scenario building methods to model how species and carbon stocks responded to increasing agricultural yields. I found that all taxa, and carbon stocks, show similar responses: rapidly declining with conversion from natural habitats to agricultural land. The populations of most species, and regional carbon stocks, were therefore maximised with a land-sparing strategy that combines high yields with forest protection or restoration. Such as strategy is broadly aligned with the goals of producers, environmental organisations and policy makers in Yucatán. However, mechanisms for active land sparing, which can link yield increases with habitat protection, will be needed to ensure that the benefits of land sparing can be realised. In addition, even with land sparing, probable 2030 production targets still resulted in forest loss, highlighting the need for demand reduction as an important part of a sustainable food strategy.