Sustainable high-yield farming is essential for bending the curve of biodiversity loss

Abstract

Food production does more damage to wild species than any other sector of human activity, yet how best to limit its growing impact is greatly contested. Reviewing progress to date in interventions that encourage less damaging diets or cut food loss and waste, we conclude that both are essential but far from sufficient. In terms of production, field studies from five continents quantifying the population-level impacts of land sharing, land sparing, intermediate and mixed approaches for almost 2000 individually assessed species show that implementing high-yield farming to spare natural habitats consistently outperforms land sharing, particularly for species of highest conservation concern. Sparing also offers considerable potential for mitigating climate change. Delivering land sparing nevertheless raises several important challenges-in particular, identifying and promoting higher yielding farm systems that are less environmentally harmful than current industrial agriculture, and devising mechanisms to limit rebound effects and instead tie yield gains to habitat conservation. Progress will depend on conservationists forging novel collaborations with the agriculture sector. While this may be challenging, we suggest that without it there is no realistic prospect of slowing biodiversity loss.This article is part of the discussion meeting issue 'Bending the curve towards nature recovery: building on Georgina Mace's legacy for a biodiverse future'.

Keywords: biodiversity; farming; food production; land sparing; land use; yield.

Figure 1.

The pre-eminent importance of farming for conservation. (a) Agriculture threatens a greater percentage of threatened and near-threatened birds (1773 species), mammals (1280 species), reptiles (1468 species) and amphibians (2523 species) than any other human activity (data from IUCN [5]). (b) The area of cropland estimated to be needed by each country by 2060 compared to its cropland area in 2010, based on business-as-usual projections of population size, diet, trade and yield (from Tilman et al. [8]).

Figure 2.

Promising interventions for reducing meat consumption and food waste. (a) Changes in the relative sales of vegetarian meals as their relative availability increases, with diners split by quartiles based on how often they chose vegetarian meals before the study [52]. Data based on 32 687 meal choices by 1394 individual diners at a Cambridge college. Note that total meal sales were unaffected by changes in meal availability and that ‘vegetarian’ here includes vegan meals. (b) In all-you-can-eat buffets diners using 26.5 cm diameter plates served themselves roughly 50% more food, ate 45% more and wasted more than twice as much as those using 21 cm plates [53]. The authors suggest customers choosing larger plates take and waste more food because they perceive a given serving to be smaller if it is on a bigger plate––a manifestation of the Delboeuf illusion, in which the black circle on the right appears smaller than that on the left. Food amounts were assessed visually, compared with plate area, and so are reported in cm².

Figure 3.

Summary of results from all sharing/sparing studies using the Green et al. [75] framework. Pie charts show proportions of species whose landscape-wide populations would be greatest under extreme land sparing (red), extreme sharing (blue) or any intermediate (purple); for most species their population density exhibits a negative convex relationship with increasing yield (inset), so their populations are largest under land sparing. Data are plotted for a total of 1766 individually assessed species; calculations assume present-day production levels. Image produced by Tom Finch.

Figure 4.

The likely effects of a typical European country increasing land sharing or organic production (re-drawn from Binner et al. [118]). This would probably increase the abundance of farmland species (woodpigeon icon) because of using fewer fertilizers and pesticides and reinstating some small habitat features (white circles), but would have little impact on domestic habitat specialists associated with large habitat blocks (woodpecker icon, white square). However, yields and hence food output (cereal icon) would fall. Forgone domestic production would then necessarily be met by increased production overseas, causing damage to habitats and specialist species overseas (anteater and tree icons). Note that the offshore impact of food imported into richer countries is typically far greater, on a per tonne basis, than that of domestic production (see text).

Figure 5.

Delivering sustainable yield increases. (a,b) One framework for assessing the potential harms of higher yielding production methods is to plot externalities or other costs of a tonne of production under contrasting systems against how much land each requires [27]. Most such analyses to date reveal positive cost : cost associations—(a) as illustrated in a comparison of soil loss and land cost across conventional and organic UK dairy systems (from Balmford [27]). Negative associations, such as that between antimicrobial use and land cost across 74 UK pig production systems (b), instead indicate trade-offs [148]. These might potentially be addressed by identifying exceptional systems that are characterized by low costs in both domains and so lie in the bottom left of the plot; note these are not predicted by labelling schemes (colours). (c,d) Contrasting ways of achieving marked yield increases—the effects of (c) adopting GM soybean, maize and cotton, showing the percentage difference in outcomes compared with matched non-GM crops (from a meta-analysis by [149]); and (d) long-term international assistance to African farmers enabling them to access improved seeds and mineral fertilizers [150].

Figure 6.

Potential cost savings from land sparing. Lines show estimated overall costs to the UK taxpayer (combining compensation payments and capital, administration and monitoring costs) of 20 year sharing and sparing schemes scaled to deliver a combined biodiversity and carbon target. Ninety-five per cent bootstrapped confidence intervals reflect uncertainty in compensation payments [201].