Expansion of land area used for agriculture is a leading cause of biodiversity loss and greenhouse gas emissions, particularly in the tropics. One potential way to reduce these impacts is to increase food production per unit area (yield) on existing farmland, so as to minimize farmland area and to spare land for habitat conservation or restoration. There is now widespread evidence that such a strategy could benefit a large proportion of wild species, provided that spared land is conserved as natural habitat (1). However, the scope for yield growth to spare land by lowering food prices and, hence, incentives for clearance (“passive” land sparing) can be undermined if lower prices stimulate demand and if higher yields raise profits, encouraging agricultural expansion and increasing the opportunity cost of conservation (2, 3). We offer a first description of four categories of “active” land-sparing mechanisms that could overcome these rebound effects by linking yield increases with habitat protection or restoration (table S1). The effectiveness, limitations, and potential for unintended consequences of these mechanisms have yet to be systematically tested, but in each case, we describe real-world interventions that illustrate how intentional links between yield increases and land sparing might be developed

Greenhouse gas emissions from global agriculture are increasing at around 1% perannum, yet substantial cuts in emissions are needed across all sectors 1 . The challenge of reducing agricultural emissions is particularly acute, because the reductions achievable by changing farming practices are limited 2,3 and are hampered by rapidly rising food demand 4,5 . Here we assess the technical mitigation potential offered by land sparingincreasing agricultural yields, reducing farmland area and actively restoring natural habitats on the land spared 6 . Restored habitats can sequester carbon and can offset emissions from agriculture. Using the United Kingdom as an example, we estimate net emissions in 2050 under a range of future agricultural scenarios. We find that a landsparing strategy has the technical potential to achieve significant reductions in net emissions from agriculture and land-use change. Coupling land sparing with demandside strategies to reduce meat consumption and food waste can further increase the technical mitigation potential, however economic and implementation considerations might limit the degree to which this technical potential could be realised in practice.We projected the mitigation potential of land sparing in the United Kingdom with reference to its binding commitment to reduce emissions by 80% by 2050 (relative to 1990 levels) 7 . We began by identifying a technically plausible range in the future yields of all major crop and livestock commodities produced in the UK, based on historic trends and future potential. We define yields as the annual tonnage of production per hectare for crops and the feed conversion ratio (feed consumed per kilogram of production) for livestock. Future yields could vary across a wide range, driven by a number of biophysical, technical and socioeconomic factors [8][9][10][11] . We assessed the likely bounds of this range based on an assessment of technical potential and reflect this in our projections, which span yield declines through to sustained long-term growth averaging 1.3% per annum across all commodities 3 (Table 1; Supplementary Fig. 1; Supplementary Discussion). For the avoidance of doubt, we do not equate our lower yielding scenarios with 'land sharing'.We next projected emissions attributable to UK agricultural production out to 2050, quantifying all sources of emissions that would be affected by a land-sparing strategy. We therefore quantified not only emissions reported under 'Agriculture' in the UK's greenhouse gas inventory 12 , but also emissions related to agriculture but reported in other sectors (e.g. farm energy use, agro-chemical production and land-use change), and emissions arising overseas due to imported feed for livestock (see Supplementary Table 2 for all emissions sources quantified). Our projections assumed that agricultural production increases from present levels in proportion to projected demand growth (Supplementary Table 1). In certain scenarios, projected UK farming capacity does not keep pace with demand growth. In such cases ...

1. The break-up of the Soviet Union in 1991 led to the abandonment of >40 million ha of cropland, a collapse in livestock numbers and the recovery of depleted biodiversity on the steppe grasslands of Kazakhstan and southern Russia. More recently, large-scale reclamation of abandoned cropland and intensification of agriculture are observed, highlighting a need for strategies to reconcile agricultural development and biodiversity. 2. We related bird densities along a land-use gradient to yield estimates from arable and livestock systems in central Kazakhstan to decide whether a land-sparing, a land-sharing or an intermediate strategy would result in the largest benefits for biodiversity. 3. For 'loser species' (whose population size is reduced by farming), land sparing was predicted to support higher total populations of more species than was land sharing, at all production targets. 'Winners' (species benefitting from agriculture) profited from land sharing when judged from food energy or protein. Intermediate yields were best for very few species. Heavily grazed steppe grassland was important for several globally threatened and biome-restricted species. 4. Government statistics suggested that over 50% of abandoned cropland has been reclaimed since 2000 and crop yields have increased. In the same period, there was significant progress towards the designation of new protected areas, but the total area in Kazakhstan still falls short of the Convention on Biological Diversity's 17% target. 5. Policy implications. Further increases in agricultural production are likely to reduce populations of most birds, especially if they are achieved by conversion of abandoned cropland, or grassland. Our results suggest that production increases would do least harm if they resulted from increasing the output of existing cropland, using approaches such as snow accumulation, no-till and more efficient grain harvesting and storage, rather than from further reclamation of abandoned land that is now reverting back to steppe. Production increases should be offset by improved conservation planning through the designation of protected areas on land potentially suitable for cropland expansion.