The land sparing-sharing framework originated as a model for quantifying and understanding the implications for wild species of using land in different ways to produce food [1
]. It is based on the idea that there are two main ways to reduce the impacts of farming on wild species—making farmland itself more wildlife-friendly, or making more space for unfarmed habitats—and on the observation that there is a tension between these two sorts of interventions. There is a limit to how friendly one can make farmland for wild species without reducing yields, and lower yields mean that more land is needed to produce each tonne of food, making it harder to create more space for nature. The land sparing-sharing model (“the model”) provides a framework for researchers to collect and analyse data to understand to what extent this tension exists in the real world, and to estimate how large populations of species would be under land sparing, land sharing, or intermediate scenarios.
Since the publication of the model in 2005, it has been applied to field data from Ghana, India, Uganda, Mexico, Brazil, Uruguay, Kazakhstan and Colombia [2
]. These empirical, place-based papers, however, have been outnumbered by reviews and other conceptual papers, some positive, but many of them critical of the framework [8
]. Some have suggested abandoning the framework entirely. I believe that these critiques were written in good faith, and motivated by a genuine desire to correct what the authors perceived as deficiencies of the model. Some criticisms have helped to clarify where further work is needed, and to ensure that issues of social justice and food security, for example, are widely discussed. However, many of the criticisms of the model are misplaced, and the discussion has become unnecessarily polarised. To move from antagonism towards better conversations about conservation and farming, it seems worthwhile to explore these points of misunderstanding. As with other contentious issues, developing a shared understanding of concepts, methods and perspectives is important for constructive interactions [27
Here, I review definitions of land sparing and related concepts. I summarise what the model does, what it does not do, and what has been learned so far from empirical studies. I discuss the ethical underpinnings of the model, and explore eight contested issues. I conclude that the value and relevance of the model depends on interpreting it appropriately, alongside insights from other disciplines.
1.1. Definitions and Related Concepts
One of the most frequent problems in the discussion is that the term “land sparing” is understood by different people in different ways, and is sometimes confounded with the Borlaug hypothesis, forest transition, sustainable intensification or even eco-modernism. In fact, each of these terms describes something different, and distinguishing between them is crucial for engaging in a meaningful conversation about land sparing and land sharing. I outline here my understanding of these and other terms, the most important of which are defined in Table 1
. The land sparing-sharing framework was developed to inform conservation actions, and thus both land sparing and land sharing describe broad categories of what conservation interventions might seek to do.
“Land sparing” is a compound noun, with “sparing” as the gerund. It is thus grammatically equivalent to “thought provoking” or “bird watching” and can also be used as an adjective, as in “land-sparing strategies”. It refers to the outcome of intentional conservation interventions that combine increasing yields on farmed land with sparing native vegetation or freeing up land for habitat restoration elsewhere. If land is not spared for nature, then it is not land sparing. “Land sharing” is also a compound noun, and refers to the outcome of intentional conservation interventions that enhance the value of farmlands for wild species. Wild species are those who are not cultivated or domesticated. Land sharing implies purposeful efforts to make farmlands more wildlife-friendly than they would otherwise be. Land sparing and sharing are often abbreviated to simply “sparing” and “sharing”.
Land sparing is sometimes conflated with the Borlaug hypothesis [9
], attributed to plant scientist Norman Borlaug [31
]. Borlaug [32
] wrote “that by producing more food per unit of cultivated area more land would be available for other uses, including recreation and wildlife.” Land sparing thus refers to a situation where food is produced using higher yields on a smaller area and native vegetation is conserved, while the Borlaug hypothesis posits one way of achieving that outcome: that yield increases will result in what has been termed “passive” land sparing [33
]. Yield increases that result in land being spared are also often referred to as “land saving” [28
]. Evidence for and against the Borlaug hypotheses is reviewed in Section 2.3
Each of the terms sustainable intensification, ecological intensification and agroecological intensification refer to actions to increase agricultural yields while reducing environmental harm, but each has a distinct meaning [35
]. Sustainable intensification refers to actions to “increase food production from existing farmland in ways that place far less pressure on the environment and that do not undermine our capacity to continue producing food in the future” [36
]. The focus is on reducing the environmental impacts of food production. Whereas a land-sparing approach situates higher yields as part of the solution for sparing more land for nature, and could be implemented in the context of declining, stable or increasing food production, sustainable intensification situates higher yields as part of the solution for increasing the global food supply, and is usually associated with the imperative to produce more but in ways that are less harmful to the environment [37
]. Ecological intensification refers to actions that replace anthropogenic inputs or increase yields by harnessing ecosystem services [38
]. Agroecological intensification builds on ecological intensification, with an emphasis on local knowledge and empowerment of smallholders [39
]. It is often linked with calls for food sovereignty—for local people to control how food is produced and distributed [40
The idea of a forest transition has been linked to the concept of land sparing [9
]. It refers to the observation that in many countries, forest cover first declines and later increases alongside economic development [45
]. The causes of forest transitions are not restricted to the agricultural sector, but include industrialisation and urbanisation. While the occurrence of forest transitions might appear to offer some hope for reconciling conservation and economic development, there are two important caveats. First, most of the “new” forests in the later stages of the transition are plantations, secondary forests or managed forests, which typically have lower value for biodiversity than primary forests [46
]. Second, a focus on national-level patterns ignores international trade. Wealthier countries may experience a forest transition while outsourcing their demand for wood, food and other land-demanding products to other countries [47
]. Evidence for forest transitions is thus not evidence that overall, land has been spared for nature.
Land sparing has been associated with ecomodernism [48
], an environmental philosophy which argues that to save nature, humanity must use technology to decouple human development from environmental impact [49
]. But while Phalan et al. [10
] described land sparing as a way to minimise the inevitable harm of food production, ecomodernists are more positive, and view decoupling as a pathway towards a “good Anthropocene”. Those using the land sparing-sharing framework see reductions in consumption as an important part of the solution [2
], whereas ecomodernists do not. Those using the sparing-sharing framework are agnostic about how yields should be increased, and welcome initiatives involving farmer knowledge networks and ecological intensification; ecomodernists focus almost exclusively on modern, technological solutions [51
To summarise, land sparing and land sharing describe two ends of a continuum of intentional spatial organisation of food production and biodiversity conservation: whether separated or integrated. The Borlaug hypothesis describes one possible way in which land sparing might be achieved: as a consequence of increasing yields. Forest transitions describe increases in countries’ forest cover as a consequence of economic development, but they do not necessarily ensure that land is spared for nature. Ecomodernism is a philosophical stance adopted by some technological optimists and humanists. Sustainable intensification, ecological intensification and agroecological intensification refer to methods for increasing yields while reducing the environmental impacts of farming, and may or may not incorporate some land sparing and/or sharing.
1.2. What Does the Model Do?
The land sparing-sharing model provides an analytical framework for asking questions about the value of agricultural land for wild species. The model requires data on the densities of wild species across a gradient of agricultural yields, from unfarmed land to high-yielding farmland. Density–yield curves are fitted to the data, and are used to project the population-level outcomes of different combinations of land use that meet a series of specified production levels. The model is described in detail elsewhere [1
]. Here, I describe its conceptual advantages over previous methods. Some forms of agricultural land management, such as agroforestry or organic farming, are commonly described as being “good for biodiversity”. The model, in simple terms, helps to clarify:
how good they are,
for what species,
in relation to what reference, and
at what cost.
Understanding how good farmed landscapes are for wild species is important, because suitability varies on a continuous, not binary, scale. The land sparing-sharing model made an important advance on previous work by using the population density of each individual species as a metric for habitat quality. Previously, the most common metric used in the literature was sample- or site-scale species richness [52
]. Biodiversity, however, is much more than species richness. Although species richness continues to be used, it is rarely, if ever, an appropriate focus for land managers, because it fails to detect changes in species composition (for example, the replacement of restricted-range species and grassland specialists by widespread generalists) and is blind to changes in population size (a species can decline by 99% at a site and still count towards richness) [54
]. Commonly-used diversity indices, such as Shannon richness, have similar weaknesses.
Analysing how population densities of each species respond to land conversion and yield increases allows us to identify which species are negatively affected by agriculture, and which species benefit from it. If we assume that humans have an equal responsibility to each species not to eliminate it, then we should focus most attention and resources on those in greatest danger of extinction. By analysing population outcomes for each species separately, the land sparing-sharing framework enables researchers to identify which species are harmed and which benefit from farming, and to understand effects on species which are already threatened or which have restricted ranges.
Studies purporting to show “high” biodiversity or “high” yields within an agricultural system must define “high” in relation to some reference. If we are interested in the impacts of agriculture, then the appropriate reference for population density is that in unfarmed native vegetation. For yields, it is the maximum attainable yield that can be sustained in the long term. The land sparing-sharing framework provides a way to incorporate both of these reference states.
Finally, all conservation comes at a cost. One of the main costs in an agricultural context is opportunity cost: food not produced because land is kept uncultivated or farmed using lower-yield methods [56
]. The particular relevance of opportunity cost for land sparing and land sharing is that if yields are reduced in one place, some of the forgone production is likely to be displaced (and have impacts on biodiversity) elsewhere. This effect is illustrated by yield reductions caused by crop diseases, which have prompted greater reliance on imports [57
]. Assessments of the outcomes of wildlife-friendly farming or protected areas have often overlooked this leakage problem by ignoring the land-sparing potential of higher-yield farming. While the land sparing-sharing framework is a model of biophysical, not economic, relationships, and is not designed to predict land-use change, it provides an approximate way to measure opportunity cost and internalise leakage. The implicit assumption is that displacement of food production elsewhere would result in the same amount of harm to biodiversity as if it were retained within the focal landscape.
1.3. What Does the Model Not Do?
A good way to think about the land sparing-sharing model is that it is intersectional (distinct but overlapping) with other questions about food, land use and conservation. There are many such questions, and the model does not address all, or even most, of them. Like all models, this one provides a partial and simplified view of reality. Like all models, its purpose is not to provide a complete description of the world, but to provide useful information about a specific question. Using density–yield curves to understand how species respond to agriculture can be conceptualised as one part of a metaphorical elephant (Figure 1
). We can put what we learn from the model together with what we learn by asking other questions, as we search for answers as to how to reconcile human demand for food, and conservation.
To give an example, the model does not address the social organisation of farming. Among the complex web of factors that have driven smallholder farmers into poverty traps are trade liberalisation, unequal access to land and capital, and development policies focused on commercial export crops [58
]. Meanwhile, the results of land sparing-sharing studies indicate that protecting as much native vegetation as possible, even if that means higher-yield farming, is likely to be the best way to conserve wild species (see Section 1.4
). Are these two findings in conflict? No. We must take both into account if conservation and human development are to succeed. Taken together, what they imply is a need to spare land for nature while at the same time supporting policies that level the playing field for smallholders. How to do so will depend on local circumstances [59
], but could involve strengthening land rights and other institutions, making it easier for smallholders to access and share information, technology and resources to increase yields and restore habitats, and funding research into agronomic and agroecological methods appropriate to the local context.
The original formulation of the land sparing-sharing framework omits some complexities, but it can be and already has been modified to incorporate many of these, including spatial configuration and the influence of edge effects [60
]; the influence of changing diets and reducing food waste [50
]; inclusion of some ecosystem services [4
]; prediction of the effects of specific public policies [63
]; and application to forestry [64
], urban planning [66
] and marine conservation [70
]. It is a model, and so all assumptions can be varied and tested. Further modifications have been suggested to incorporate species dispersal and source-sink dynamics [71
], heterogeneity at multiple scales [15
], analysis of outcomes for food security and human wellbeing [72
], use of optimisation techniques and more complex mixes of land use [73
] and replacement of imprecise terms such as “win-wins” with more clearly-defined concepts such as Pareto improvement [74
]. An important challenge is to move beyond vote-counting towards developing conservation plans that ensure populations of all species in a region are sustained.
The land sparing-sharing model has been criticised for “ignoring” politics, rural livelihoods and governance [11
]. This criticism reflects a misunderstanding of the purpose of the model. Just as climate models serve to predict the effects on global and regional climate of adding certain amounts of greenhouse gases to atmosphere, the land sparing-sharing model serves to predict the effects on wild species of using land in different ways to produce food. And just as social, economic and political factors will determine what action is taken to limit climate change, similar factors will determine what action is taken to limit the impacts of land-use change and food production on biodiversity. Predictions of global climate forcing are apolitical, and yet the insights they provide are essential for informed advocacy and activism. Similarly, an accurate quantitative understanding of how land-use decisions affect wild species is an essential starting point for informed conservation action. If our aims are climate justice and just conservation [76
], we must integrate model insights with social and political knowledge as we strive towards both social and inter-species justice [77
1.4. What Have We Learned So Far?
A relatively small number of empirical studies have collected all of the data necessary to apply the land sparing-sharing model. Many other empirical and theoretical studies have referenced the concepts [26
], but have failed to include data on an unfarmed baseline, to measure the densities of large numbers of individual species, to quantify farm yields, or to calculate the total population effects of different land-use choices [10
]. Those other studies can sometimes provide useful information. However, any conclusions they draw about the merits of sparing and sharing from the shape of curves with species richness on the Y
axis, for example, are unreliable. Here, I focus on what we have learned from studies that used the land sparing-sharing model proposed by Green et al. [1
], or which modified it [4
] while adhering to the critical points mentioned above.
The key finding from all empirical studies to date, covering >1500 species, is that most species would have larger populations if a given amount of food is produced on as small an area as possible, while sparing as large an area of native vegetation as possible. This is true for birds and trees in the Upper Guinea forests of Ghana [2
], birds and trees in the Upper Gangetic Plain of India [2
], birds in Uganda’s banana-coffee arc [3
], birds and dung beetles in the Colombian Chocó-Andes [4
], birds in the Kazakhstan steppe [5
], birds in the Pampas grasslands of Brazil and Uruguay [6
], and birds, trees and dung beetles in the Yucatán, Mexico [7
]. It is especially true for species with small global ranges, which are often those of most conservation concern. The majority of species—specialists and generalists alike—are negatively affected when their habitats are converted to agricultural use, even to apparently benign uses such as diverse landscape mosaics with agroforestry plots and fallows. A typical example is the Chocolate-backed Kingfisher Halcyon badia
), for which even the most complex agroforests offer a poor substitute for its native forest habitat. A few species benefit from conversion, including some not found in the original native vegetation types. So, the species in most need of help are those who would be most favoured by land sparing, while land sharing is of little benefit to such species.
Why might we see such consistent responses across taxa, continents and biomes? One part of the reason is that most species are relatively specialised and rare [79
]. We might expect that the habitat structure and resources they depend on are easily disrupted, even by relatively subtle changes such as understorey clearance to grow coffee or cocoa. Farming eliminates niches by diverting land, water, light and nutrients for human consumption [80
]. Adopting more benign methods may not help much; at the same yields, organic farming is no better for invertebrate diversity than conventional farming [81
]. It is not that wildlife-friendly farming systems fail to provide habitat for many species—they do [82
]. But they fail to provide good quality habitat for most of the species that would occur in the absence of farming.
There are some interesting nuances to these results. The farming systems with most conservation value, such as lightly-grazed semi-natural grasslands, are so low-yielding as to make little meaningful contribution to food supply [5
]. Although some species peak in density at low or intermediate yields—in high-nature-value farmland, as it is termed in Europe [83
]—they do so in farmland with yields too low to maintain current food production if scaled up. Such systems are often also uneconomic, even with subsidies, and are prone to abandonment or intensification [84
]. If we wish to preserve these systems, we are better to do so primarily for their ecological and cultural values, rather than positioning them as a model of production and conservation in harmony. This could be supported by reforms to agricultural and agri-environmental subsidies [86
]. Sparing of low-yield farmland necessarily competes with the sparing of unfarmed land, and may require yields to be increased elsewhere [6
]. The sparing-sharing model can be readily adapted to examine the sparing of high-nature-value farmland alongside native vegetation, using mixed scenarios with multiple compartments rather than two as in the original model [73
Although such mixed scenarios merit further investigation, the results of sparing-sharing studies indicate that wildlife-friendly farming has serious limitations for conservation. Narratives proposing that countryside habitats offer a “win-win” for people and nature are appealing. However, such narratives have often relied on species richness to measure biodiversity, have failed to test any alternative configurations of land use, have omitted reference systems, or have not specified what evidence it would take to show that countryside habitats offer a poor deal for wild species [10
]. Instead, the data collected so far suggest that protecting and restoring as much native vegetation as possible would be more beneficial for wild species than making farmlands more wildlife-friendly. The data show that even the most wildlife-friendly systems are missing many species from the native ecosystem, and that most of these farming systems are low-yielding.
How should these findings be used by decision-makers? They help to clarify that what wild species need, much more than they need wildlife-friendly farming, is large areas of native vegetation. This insight does not exclude the possibility that decision-makers might prefer intermediate strategies for social or political reasons, or for conserving certain species, but they clarify that such strategies represent a poor compromise overall for biodiversity conservation. The findings do not imply that land sparing will be optimal for other objectives too, such as ecosystem service provision or food security. It may (or may not) be that more complex strategies are optimal when other objectives are included. Some objectives can be readily combined on the same land, while others cannot. What is clear—at least in the growing set of places studied so far—is that separation of food production and conservation on different areas of land, rather than integration, would be the most effective way of reconciling these two objectives.
This tells us something useful about how conservation efforts might be focused. In Ghana, for example, actions to prevent any further forest loss and degradation in forest reserves are likely to be more beneficial for biodiversity than promoting wildlife-friendly cocoa agroforestry. In Kazakhstan, concentrating food production on existing croplands will cause less harm to biodiversity than converting intact and grazed steppes. The land sparing-sharing model is best thought of as an aid to understanding the technical potential of different modes of agriculture to support wild species. It cannot predict land-use change, but can be used as a tool for predicting the consequences of land-use change for species populations. As one part of an elephant, information on density–yield responses should be integrated with social, political and other information when making decisions.
1.5. Choices about Farming and Land-Use Are Underpinned by Ethics
The land sparing-sharing model was developed with the implicit recognition that non-human species have their own intrinsic value, independent of any contribution they make to human wellbeing. Its aim is to identify how to meet human needs (in this case, food production) with least harm to non-human organisms. There is increasing recognition that humans have various ethical responsibilities towards other species, not least an obligation not to eliminate them [78
]. The model does not provide information on how to weight the objectives of producing more food or conserving species populations, and it does not attempt to combine them into a single objective function. While the model could be adapted to trade off one anthropocentric objective against another [24
], that is not its original purpose, and in my view, it would be a mistake to sideline the intrinsic value of wild species.
This emphasis on the intrinsic value of wild species is perhaps one of the most important, yet overlooked, reasons that different authors have come to different conclusions about sparing and sharing [17
]. Broadly speaking, concern with preventing the disappearance of wild species underpins the logic and analyses of the land sparing-sharing framework: how do we meet a human need at least cost to nonhuman species? Those favouring land sharing tend to gravitate towards a different question: how do we design landscapes that meet a range of human needs? Biodiversity conservation is conceptualised as one small component, and replacement of some species by others is seen as undesirable only to the extent that it undermines human interests. Emphasising the needs of wild species is consistent with an ecocentric ethic, which sees humans as one species among many, and values non-human species for their own sakes [90
]. A risk with ecocentrism is that by seeing humans as a single species, we may overlook injustice between different groups of people [91
]. Focusing on human wellbeing is more consistent with an anthropocentric ethic, where non-human species have value only in relation to humanity [92
], and where “biodiversity” is used to refer primarily to the diversity of crops, livestock and farmland [93
]. A risk with anthropocentrism is that by focusing on human wellbeing, we may ignore the needs of species who are not useful or valuable to us. Attention to both the needs of vulnerable groups of people, and to those of non-human species, is needed if we are to pursue “just conservation” [76
The sparing-sharing model places most emphasis on understanding what wild species need, given the constraint of food production for humans. It is thus informed primarily by an ecocentric ethic, but this should in no way preclude additional consideration of social justice and nature’s contributions to people too [94
]. Better understanding of one part of the elephant should be seen as complementary, not competitive, with understanding other parts of the elephant.
Nevertheless, conserving farmland biodiversity for the services it provides (ecological intensification) is not the same as conserving on-farm biodiversity for its own sake (land sharing). Ecosystem services from on-farm biodiversity can contribute to food production, and can in many cases support sustainable increases in yields [38
]. However, actions to promote service-providing biodiversity are insufficient to conserve those species that have little direct service value [95
]. Such species—probably the majority of life on Earth—will benefit if humanity can produce the same amount of food on less land (including through ecological intensification), while conserving and restoring native vegetation elsewhere.