Reconciling farming and biodiversity

Just some thoughts on land-sparing and land-sharing.

>>The problem

Agriculture is a leading threat to biodiversity: not only does it drive the conversion of more-or-less pristine habitats, but the increasingly intense management of agricultural landscapes is resulting in the decline of farmland wildlife.

As human population size and consumption rates increase, it seems likely that global food production will have to increase too (though reducing waste and over-consumption, and making the food supply chain more equitable will also be important). We have a decision to make about how and where food production should take place. The crucial question (for me) is: how can we meet future food production needs (whether higher, lower, or equal to current production) whilst maximising wildlife populations?

>>The solutions – land-sparing or land-sharing?

Broadly, there are two opposing strategies we might use to trade-off the (generally competing) interests of food production and wildlife conservation – through (i) their integration or (ii) their compartmentalisation.


The former strategy (wildlife-friendly farming, or land-sharing) involves managing the food production landscape in a way which is sympathetic to wildlife, in order to maximise on-farm biodiversity. Essentially, this requires an increase in ‘heterogeneity’. This might be achieved through maintaining semi-natural features such as hedgerows, field margins and fallows, or reducing chemical inputs.

This strategy is initially attractive. However, wildlife-friendly farms are often lower-yielding, such that in order to maintain the same overall level of food production, a larger area of land needs to be farmed. So, whilst organic farming may reduce (e.g. chemical) externalities per unit area, it may simply ‘export’ these externalities to another farm / region / nation, such that per unit product the negative impacts are actually amplified.


An alternative strategy (land-sparing) involves maximising the productivity of farmed land so that land elsewhere can be protected from agricultural expansion (or restored back to ‘natural’ habitat). Land sparing – potentially promoting a farmed landscape which is relatively hostile to wildlife – goes against conservation orthodoxy which typically focuses on making farming more wildlife friendly.

However, it addresses head-on the fact that any in-farm conservation measure which reduces yield is essentially ‘leaky’, displacing food production elsewhere.


A key strength of the land-sparing concept is that it addresses the food production ‘leakage’ from in-farm conservation measures.

With remarkable consistency, basically every empirical test of the ‘land-sparing/sharing framework’ has come up in favour of land-sparing. That is, most species (from birds in Kazakhstan and Uganda to trees in Ghana and India) tend to do quite badly as soon as natural habitats get converted to farmland, such that even low-yielding, wildlife-friendly farming is pretty bad for biodiversity. Instead, the majority of species are expected to benefit from the intensification of farmland (which wasn’t doing much good for biodiversity anyway), combined with the protection of natural habitat.

Predictably, the framework has come up against a lot of criticism – some of which is justified, some of which derives, I think, from misunderstandings. (Actually, I don’t much like the terms land-sparing and land-sharing, but unfortunately they’ve stuck.)

>>The definition problem

I think one of the first problems the land-sparing/sharing framework runs into is a problem of definition.


Does a hedgerow count as ‘spared’ or ‘shared’ land? The answer comes down to spatial scale, or ‘grain size’. I think it makes sense to conceptualise a landscape as a regular grid, with each cell representing a ‘management unit’. (It’s worth considering whether this grid-based approach is sensible – maybe another time). Food production decisions are made at the cell level, such that cells are managed along a gradient from zero- to high-yield farming.

How big should these cells be? There are two considerations – at what scale are management decisions made, and over what scale do individual organisms live out their lives.

Whilst there’s clearly no right answer, it seems that a 1km grid is sensible (and convenient). A 1km² (100ha) patch of habitat can probably support a pretty independently functioning community of plants, insects, small mammals, or birds. Any smaller than this, and you’re getting below the size at which most organisms of interest live their lives. Any bigger, and the scale at which management decisions need to be made becomes impractically large.

So, a lone tree wouldn’t usually count as spared land, whereas a woodlot probably would.


The land-sparing/sharing framework is often illustrated as a continuum: at one end, land-sparing is presented as a discontinuous landscape with large blocks of farmed and un-farmed land. At the other, land-sharing represents a more continuous landscape.

An extremely (absurdly) aggregated form of land-sparing might involve East Anglia being left devoid of natural habitat, whilst Wales becomes transformed into a rewilded wilderness (well, they did vote for Brexit..). I don’t think anyone is seriously advocating such a scenario, except, perhaps, our Environment Secretary:

“It would make so much more sense if those with the big fields do the sheep, and those with the hill farms do the butterflies”

Whilst it’s true that wildlife-friendly farming will tend to increase heterogeneity, this need act only at the level of ‘cell’. At the landscape level, I don’t think there is any prerequisite that land-sparing should be more discontinuous. Grid-cells can be managed independently, such that land-sparing could result in a continuous patchwork of high-yielding farmland interspersed with patches of natural habitat (e.g. bottom left panel, below).


My take on the land-sparing/sharing framework. All four landscapes produce the same total amount of food, have the same overall size (100 x 100km) and the same grain size (1km²). In the top row (land-sharing), wildlife-friendly farming (orange) results in lower yields per km², so the area available for natural habitat (green) is reduced. In the bottom row (land-sparing), high-yield farming (yellow) means that more land can be preserved for nature. The configuration of natural habitat within the landscape is flexible, and I don’t think that either land-sparing or land-sharing necessitates a more (left) or less (right) ‘aggregated’ landscape.

Mutual exclusivity?

The framing of land-sparing and land-sharing at opposite ends of a continuum suggests, perhaps, that they are mutually exclusive. This is exemplified, I suppose, by the ‘binary’ landscapes I illustrated above (i.e. with only 2 possible levels of agricultural production in a given cell). But this need not be the case.

A combination of high-yield farming, low-yield wildlife-friendly farming, and natural habitat is obviously possible too, and perhaps desirable given that conservationists are generally trying to manage the landscape for species and communities with different ecological requirements.

>>The ideology problem

The land-sparing/sharing framework is a model, and all models are wrong. But I think it’s a pretty powerful and persuasive one, acknowledging as it does the importance of food production.

Not everyone agrees though. Some have argued – in quite strong terms – that the way the problem is framed predisposes us (researchers) to come out in favour of land-sparing. For example, John Vandermeer argues that:

“The land/sparing versus land/sharing framework is a marvellous example of setting the terms of the debate so that one side is almost certainly going to win… The framework of land sparing versus land sharing is more obfuscating than enlightening”

He also suggests that the framework is somehow related to Monsanto’s anxiousness “to obtain access to European markets for their GMO products”, and that by re-branding low-yield farming as “high quality agro-ecology” and high-yield farming as “destructive industrial agriculture”, the outcome would necessarily be reversed.

I’m not at all convinced by this argument (but maybe I’m blinded by ideological predisposition, or simply a desire to please my superiors. If anyone can help me understand John’s point of view, I’d be interested). Whatever name we choose to give a certain agricultural practice, the empirical evidence so far suggests that most species would benefit from higher-yield farming coupled with protection of natural habitat. The framework does not require high-yield farming to be socially and environmentally destructive – it’s possible (hopefully) and desirable (definitely) to intensify in a sustainable way.

One fair criticism, though, is the way in which the models have been evaluated to date. As discussed below, this has perhaps predisposed previous studies to favour land-sparing.

>>The outcome problem

How should we evaluate the biodiversity outcomes of different food production scenarios? Different suites of species will achieve maximum population size under different scenarios, such that any scenario will favour some species at the expense of others.


Winners and losers. Extreme land-sharing (promoting wildlife-friendly farming) will probably be good for Whinchats (left), but not so good for Redstarts (right), which would do better under extreme land-sparing (promoting woodland). To improve the conservation status of both species, a mixed strategy would probably be necessary.

Previous empirical tests of the land-sparing/sharing framework have classified species whose density increases with agricultural yield as ‘winners’. These are the open-habitat specialists and generalists which can cope with farming, and which generally ‘prefer’ a land-sharing strategy. The way the framework has generally been evaluated has – as far as I can tell – essentially ignored these species; they are not of sufficient ‘conservation concern’, and are outnumbered by species which always decline with agriculture (e.g. forest specialists). These ‘loser’ species, perhaps unsurprisingly, tend to prefer land-sparing, because this is the strategy under which forest area (for example) is maximised (see figure, below).

This approach perhaps makes sense in the tropics, where forest clearance represents a key threat to (forest-adapted) biodiversity. But in places like the UK, with a long history of land clearance, much of our wildlife is now intimately associated with (low intensity) farming and the semi-natural open habitats which it maintains. My guess would be that the UK’s ‘farmland birds’ would be winners (because they’re generally absent from e.g. woodland habitats) who would, on the whole, do best under land-sharing.

These farmland birds – threatened by agricultural intensification – are currently of major conservation concern. Clearly, ignoring them would not be satisfactory. Instead, we should look for optimal solutions which maximise the population size of a suite of species – both winners and losers. As mentioned above, my guess is that a mixed strategy would come out top, providing habitat for a range of species (from skylarks to sedge warblers to spotted flycatchers).


Number of bird species (Kazakhstan) which are winners (light colours) or losers (dark colours) from agriculture, coloured according to which land-use scenario maximises their population size, across a range of production targets. Although only a few species ‘win’ from agriculture (including the critically endangered Sociable Lapwing), these tend to do best under land-sharing. Losers are more prevalent, and do best under land-sparing. Figure 1 from

>>Some more caveats

There are other issues with the land-sparing/sharing framework, which I’ll maybe tackle in another post:

  • Other ecosystem services (e.g. carbon storage, soil conservation, human recreation) need to be considered.
  • Yield increases must be explicitly coupled with habitat protection / restoration through suitable policy mechanisms.
  • The framework kind of assumes that farming does not depend on biodiversity – but it’s likely that wildlife provides some important services to agriculture (e.g. pollination, pest control). Though I do wonder how much this has been exaggerated.
  • The framework also assumes that current high-yield farm practices can be sustained in the future – though this seems unlikely given the rates of soil loss in parts of south-east England.
  • Finally – and, for me, most interestingly – we must consider habitat fragmentation and edge effects. I’m currently assuming that biodiversity is additive – that we can sum up the populations of individual cells across the entire landscape. However, isolated fragments of natural habitat may sustain much smaller populations than larger, or better connected ones. Additionally, if the agricultural ‘matrix’ between habitat fragments is super hostile, population viability might be reduced.
  • Finally (2) – what does natural habitat even look like, anyway?

3 thoughts on “Reconciling farming and biodiversity

  1. stevecjjones says:

    Great piece, Tom. I’m definitely no scientist so my comments / questions will likely strike you as naive but here goes:

    Regarding land-sparing scale, at one extreme one might ask: ‘What scale of land would need to be spared to support viable populations of a region’s entire historical species pool, including extirpated species?’ So, in Europe, we have species that cannot survive in farmland without presumably insurmountable conflict – e.g. European bison (which will eat arable crops/grassland intended for domestic stock and may spread diseases to livestock). We’d need at least some areas to be ‘spared’ for such species, and the scale of ‘spared’ land would presumably need to be large. Maybe it’s pointless having in mind the historical species pool in Europe’s long-settled cultural landscapes but, in these time of rewilding, maybe not, and, if not, does that affect one’s idea of scale?

    Regarding whinchat, isn’t it true that, in south-west England at least, the last place they’re holding on in good numbers is on the 28,000ha Salisbury Plain Training Area? I gather that even skylarks reach higher densities there than on any farmland (according to an RSPB/DE BBS survey conducted in the 90s) . Maybe you could argue that this area is extraordinarily low yielding farmland: I’d argue that it’s land spared and is being kept open by a disturbance regime (caused by tanks) similar to what we might see if we had lots of wild megaherbivores roaming about! I’d not assume that whinchat and skylark would reach only low densities in land spared and maintained as open habitat.

    In the UK, I’ve often wondered how the Hope Farm results – that indicate it’s possible to increase both yields and the density of farmland birds – would affect the results of a land-sparing modeling exercise? There’s only one Hope Farm, but if there were lots, and they figured significantly in your sample, would the model indicate that you could have high-yield, high-farm-bird-density farmland on the one hand, and free-up land on the other to be spared for species simply unable to persist in farmland without conflict?


    • tommfinch says:

      Steve, thanks very much for your comments. They’re not at all naive, and why should they be? I fear scientists aren’t all they’re cracked up to be…

      Good point about scale – it’s what I was trying to get at under the ‘grain size’ section. You’re right that something like a bison would require a larger area to spared. I suppose this would apply to things like Great Bustards too, which aren’t very well suited to human features in the landscape – 100ha of habitat just isn’t enough. The ‘historical species pool’ is an interesting idea, and one which I might return to.

      It’s funny that you mention Salisbury Plain, as it’s one of my study sites (along with The Fens). The report I have gives a density of 1.6 pairs per km2 for Whinchat and 39 (!) for Skylark. We are treating the downland as a ‘natural’ baseline habitat, though some livestock production occurs here. It comes down to whether we sit in the rewilding or traditional management camp. This is another idea I’d like to come back to in a future post. You may find this paper of interest I suppose what I meant was that ‘farmland birds’ are generally absent from climax habitats, though maybe this is a gross exaggeration – we’ll see what the data show.

      And yes, win-win scenarios appear possible, and should of course be encouraged. Under the land sparing-sharing framework, these win-wins would free up land for habitat specialists, without the need for extensive, land-hungry, low-yielding farms to provide for agricultural ‘winners’.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s