The full shelf
PARADOX
What the abundance of our supermarkets hides about the true meaning of a sustainable food system
THE SUPERMARKET ILLUSION
A supermarket shelf is a remarkably convincing picture of stability.
Walk into one in London in the middle of winter and much of the world appears to be within reach. Tomatoes and strawberries sit beneath artificial light regardless of the season outside. Bananas have travelled from tropical climates; rice from fields thousands of miles away. Coffee, avocados, olive oil and salmon occupy their familiar places, replenished so routinely that their presence rarely seems remarkable. What appears before us is less a harvest than an expectation: that almost everything should be available, almost all of the time.
We tend to judge the food system by the one part of it we can see: whether the shelves are full.
But the supermarket shows us the end of the food system, not the system itself. Behind those shelves lies a network of farms, fisheries, processing plants, warehouses, ports and roads, connected by refrigeration, energy, fertiliser, animal feed, water, labour and international trade. In the UK, this network extends particularly far beyond national borders. Around 40 per cent of the country's food supply is imported, while dependence is considerably higher for some parts of the diet: in 2023, domestic production supplied only 53 per cent of vegetables and 16 per cent of fruit (Defra, 2024).
This dependence is not necessarily a weakness. In fact, it has long been one of the system's strengths. Combining domestic production with food sourced from different countries allows shortages in one place to be compensated for by supply from another. The UK's National Audit Office notes that this ability to source food globally has helped the food sector absorb recent disruptions and maintain supplies (NAO, 2026). A failed harvest, transport interruption or regional weather event does not automatically translate into an empty shelf.
Yet availability tells us something about whether a food system is functioning. It tells us much less about whether that system is sustainable.
Food production ultimately depends on resources that are easier to overlook than the products they create: fertile soils, reliable water, functioning ecosystems and pollinators, alongside energy, fertilisers, feed and human labour. Some of these foundations can deteriorate gradually even while production continues. The UK Food Security Report, for example, identifies the long-term degradation of natural capital as a risk to domestic food production, noting that changes affecting soil, water and ecosystem services can accumulate until they begin to undermine agricultural fertility and yields (Defra, 2024).
The environmental consequences can also occur far from the people consuming the food. Recent high-resolution research tracing carbon-storage and biodiversity losses through global agricultural supply chains found that food consumption accounted for 83 per cent of the losses mapped by the study. Those impacts were unevenly distributed geographically, with international trade allowing environmental pressures associated with consumption to move across national borders (Liu et al., 2026). The place where food is eaten, in other words, may reveal remarkably little about where its ecological cost has been paid.
This is where the apparently simple word sustainable becomes difficult.
A widely used definition developed by the High Level Panel of Experts on Food Security and Nutrition describes a sustainable food system as one that provides food security and nutrition for everyone without compromising the economic, social and environmental foundations required to provide them for future generations (HLPE, 2014). Sustainability therefore asks something more demanding than whether enough food can be produced today. It asks whether the conditions that make production possible can endure.
A food system might be efficient without being sustainable. It might withstand a particular disruption without being sustainable. It might even keep supermarket shelves impressively full while gradually eroding the soil, water, biodiversity, livelihoods or climatic stability upon which future production depends.
That distinction matters increasingly as climate change begins to alter not simply individual harvests but the conditions under which the global food system operates. The question is no longer only whether food can continue to reach us when one farm, region or supply route fails. It is whether a system built to compensate for failure in one place remains resilient when pressures begin to accumulate across many places at once.
The full supermarket shelf may therefore be evidence of an extraordinary logistical achievement. But it is not, by itself, evidence of sustainability.
It tells us that the system is working today.
The harder question is what it takes for it to keep working tomorrow.
The same network that distributes risk across countries also connects them. Wheat grown in one region may be consumed on another continent; fertiliser manufactured in one country can support harvests in several others; livestock production may depend on feed imported across an ocean. A disruption can therefore be absorbed through the network, but it can also travel through it.
Climate change makes this distinction increasingly important because agricultural risk is not simply about whether extreme events become more severe in one particular place. It is also about the possibility that significant stresses occur across several important producing regions within the same period.
Recent research on severe water scarcity illustrates the problem. Trnka et al. (2026) examined the relationship between water shortages across major cereal-growing regions and global grain prices. For wheat, they found a clear association between the extent of severe water scarcity and higher global prices. Their climate projections suggest that, under a scenario in which global mean temperature reaches around 3°C above the 1951–1980 baseline, average global wheat prices could reach roughly three times their 2010 level after adjustment for inflation (Trnka et al., 2026).
That projection should not be read as a prediction that a loaf of bread will simply become three times more expensive. Commodity prices are only one component of retail food prices, and future markets will also respond to changes in technology, production, trade and policy. What the study exposes instead is a structural vulnerability: when water stress expands across enough wheat-growing land at the same time, the ability to compensate for failure by sourcing from somewhere else begins to narrow.
The geography that once spread risk can begin to synchronise it.
This helps explain why resilience and sustainability, although closely related, should not be treated as synonyms.
A system can be resilient to a particular shock while remaining dependent on practices that undermine its longer-term foundations. Irrigation can protect crops from rainfall variability while depleting groundwater. Fertiliser can maintain high yields while contributing to nutrient pollution if poorly managed. International sourcing can compensate for domestic shortages while shifting pressure onto land and water elsewhere. Emergency measures can keep food moving through a crisis without addressing the conditions, making future crises more likely.
Resilience asks, in essence: can the system withstand disruption?
Sustainability asks a different question: can it continue doing so without exhausting the resources, ecosystems and social conditions on which it depends?
The difference is partly one of time. A resilient response may help a system survive tomorrow. A sustainable system must also consider what that response makes possible — or impossible — ten, twenty or fifty years later.
There is another complication. Resilience is not automatically desirable simply because a system persists. A highly concentrated or environmentally damaging food system might prove remarkably capable of surviving shocks. If resilience means only returning the system to the way it operated before the disruption, then resilience can preserve weaknesses as effectively as strengths.
The more useful question is therefore not simply how quickly a food system can recover, but what exactly we are trying to make resilient.
Its ability to move commodities?
Its ability to keep prices affordable?
Farmers' livelihoods?
Access to nutritious food?
The ecosystems that make future harvests possible?
These goals can overlap, but they are not identical. Protecting one can sometimes place pressure on another. A policy that maximises production may damage biodiversity; a measure that improves environmental performance may increase costs in the short term; a supply chain optimised for efficiency may remove the redundancy that becomes valuable during a crisis.
A sustainable food system has to negotiate these tensions rather than pretend they do not exist.
That makes sustainability less like a destination and more like a balancing act: producing enough, remaining affordable, adapting to disruption and protecting the ecological foundations of production at the same time.
The supermarket shelf may remain full through a drought, a failed harvest or a disrupted trade route. That is resilience, and it matters.
But if keeping it full today makes it harder to fill tomorrow, resilience alone is not enough.
RESILIENT IS NOT THE SAME AS SUSTAINABLE
Food systems have always had to deal with disruption. Harvests fail. Rivers flood. Droughts arrive. Prices rise, trade routes change and political crises interrupt supply. What distinguishes a resilient food system is not the absence of these shocks, but its ability to continue providing food despite them.
That distinction is important. In food-system research, resilience is generally understood as the capacity of a system, operating across different levels, to continue providing sufficient, appropriate and accessible food in the face of disturbances, including those that cannot easily be anticipated (Tendall et al., 2015). A resilient system therefore does not need to prevent every failure. It needs enough flexibility, diversity and adaptive capacity to absorb one without allowing the consequences to spread through the whole system.
In Britain, international trade provides part of that flexibility.
A supermarket does not depend on a single farm, region or even country. When poor weather damages production in one place, buyers can turn to another. When a transport route is interrupted, supply chains can sometimes be redirected. When domestic production falls short, imports can fill part of the gap. The National Audit Office argues that this capacity to source food globally has been one of the reasons the UK food system has been able to manage recent disruptions without severe interruptions to supply (NAO, 2026).
Seen from this perspective, globalisation can function as a form of insurance.
But insurance works best when risks are not all realised at once.
The geography that once spread risk can begin to synchronise it.
ADVERTISEMENT
THE HIDDEN INPUTS
A loaf of bread does not begin with flour.
Before the wheat is milled, before it reaches a bakery and long before it appears on a supermarket shelf, its production has already drawn on a much larger system: soil capable of supporting a crop, water arriving at the right time, nutrients available to the plant, energy to power machinery, fuel and infrastructure to move the harvest, and human labour throughout the chain.
These foundations are easy to overlook because most of them disappear from the product we eventually buy. We see the tomato, not the water used to grow it. We see the steak, not the land and feed behind the animal. We see a bag of flour, not the nitrogen and phosphorus that helped produce the wheat.
Food therefore carries an invisible material history.
Perhaps the most fundamental part of that history lies beneath our feet. Soil is not simply a surface on which agriculture takes place. It stores and cycles nutrients, regulates water, supports immense biological communities and contributes to carbon storage. Yet agricultural production can also alter these functions through erosion, compaction, loss of organic matter, contamination and other forms of degradation. The FAO has estimated that erosion alone removes between 20 and 37 billion tonnes of topsoil globally each year, carrying nutrients away from agricultural land and reducing its productive capacity (FAO and ITPS, 2015).
Water presents a similar paradox. Agriculture depends upon it, but agriculture is also by far the largest user of freshwater withdrawn by humans. Globally, the sector accounts for more than 70 per cent of freshwater withdrawals, although the proportion varies substantially between countries and regions (FAO, 2024). Irrigation has allowed food production to expand and made harvests less dependent on rainfall, but where withdrawals consistently exceed replenishment, the technology that creates resilience in the short term can contribute to scarcity in the longer term.
Then there are nutrients.
Modern agriculture has been transformed by the ability to supply crops with nitrogen, phosphorus and potassium in concentrated forms. Synthetic nitrogen fertiliser in particular has played an enormous role in increasing agricultural productivity. But nitrogen does not remain neatly within the crop for which it was intended. When applied inefficiently or in excess, reactive nitrogen can escape into water and the atmosphere, contributing to eutrophication, air pollution, greenhouse-gas emissions and ecosystem change (UNEP, 2019).
Phosphorus presents almost the opposite problem. It is essential for plant growth and has no substitute in food production, yet the mineral phosphate rock from which most fertiliser phosphorus is derived is a finite geological resource concentrated in a relatively small number of countries. At the same time, large quantities of phosphorus are lost from agricultural and food systems into waste streams and waterways, where the same nutrient that is valuable in a field can become a pollutant elsewhere (Brownlie et al., 2022).
This reveals one of the strange characteristics of the modern food system: scarcity and excess can exist at the same time.
We manufacture or extract nutrients to sustain agricultural production, transport them through farms, animals, food processors and cities, and then spend further resources trying to prevent those nutrients from becoming pollution once they leave the useful part of the system.
Researchers are increasingly asking whether some of those flows can be closed.
A recent life-cycle assessment of a green biorefinery integrated into dairy farming, for example, examined whether locally grown forage could be separated into protein concentrate for animal feed and a fibrous pulp co-product, potentially reducing dependence on imported protein feeds. The results were not a simple endorsement of the technology. Its environmental performance depended strongly on how the remaining pulp was used, what products the system displaced and how agricultural land was allocated (Elshani et al., 2026).
That qualification matters.
A material being local, recycled, bio-based or circular does not automatically make the system surrounding it sustainable. Moving a burden from one part of a life cycle to another is not the same as removing it. Recovering a nutrient may require energy. Replacing an imported feed may require additional land. Using an agricultural residue for one purpose can make it unavailable for another.
Sustainability therefore depends partly on where we draw the boundary around the problem.
And the boundary cannot stop at natural resources.
Food systems also depend on human work: planting, harvesting, processing, packing, transporting, stocking and selling. The International Labour Organization estimates that agriculture employs hundreds of millions of people worldwide and remains one of the sectors in which informal and vulnerable forms of employment are particularly widespread (ILO, 2025). A system capable of producing food with a small environmental footprint but unable to provide safe work or viable livelihoods would solve only part of the sustainability problem.
This is why describing soil, water, nutrients, energy and labour simply as inputs can itself be misleading. An input sounds like something that enters a production process and can be replaced when necessary. But many of the foundations of food production are not interchangeable components.
A degraded soil cannot always be quickly rebuilt. A depleted aquifer cannot necessarily be refilled on a human timescale. Lost biodiversity cannot simply be ordered back into a field. Agricultural knowledge, communities and livelihoods can disappear too.
The food system does not merely use these things.
It depends on their continued existence.
And that changes the meaning of efficiency. Producing more food from fewer resources can certainly be valuable. But if efficiency is achieved by drawing down resources faster than they can recover, the apparent gain is partly borrowed from the future.
A sustainable food system therefore has to account not only for what comes out of it but also for what must continually go in — and what condition those foundations are left in afterwards.
The product on the shelf may look remarkably simple.
The system required to put it there is anything but.
WHEN THE SOLUTION EATS THE PROBLEM
Some sustainability solutions become complicated only when we ask what they depend on.
Aquaculture offers an unusually clear example.
At first, the logic appears straightforward. If demand for seafood continues to grow while many wild fish populations are already under pressure, producing fish in farms rather than catching them from the ocean seems to offer a way of separating consumption from extraction. And in one important sense, aquaculture has already transformed the global seafood system. In 2022, for the first time, aquaculture surpassed capture fisheries as the main producer of aquatic animals, accounting for 51 per cent of global production (FAO, 2024).
But farming a fish does not necessarily mean that wild fish have disappeared from the equation.
Many of the species raised in aquaculture require feed. For carnivorous and omnivorous species, that feed has historically included fishmeal and fish oil produced from small pelagic fish and other marine organisms. Anchovies, sardines, herring and similar species can therefore be caught, processed and transformed into ingredients used to feed salmon, shrimp and other farmed animals.
The apparent alternative to fishing can still contain fishing inside it.
This does not mean that aquaculture simply consumes more wild fish than it produces. That interpretation would ignore several decades of substantial change. Feed formulations have become more efficient, inclusion rates of fishmeal and fish oil have declined for many farmed species, and plant proteins, processing by-products, algae, insects and other ingredients are increasingly being explored or used as substitutes. Research has shown that the amount of wild fish required to produce many farmed aquatic species has fallen considerably as the sector has expanded (Naylor et al., 2021).
Yet the underlying dependency has not disappeared.
Fishmeal and particularly fish oil remain valuable feed ingredients, while the continued expansion of aquaculture increases the total volume of feed that must be supplied. This creates a problem familiar from other parts of the sustainability debate: an improvement in efficiency does not necessarily produce a reduction in total pressure if the system itself grows faster than the efficiency gain.
The question, then, is not simply whether farmed fish use less wild fish than they once did.
It is what happens when aquaculture becomes the dominant way the world produces aquatic food.
This question matters because the fish used for feed are not merely industrial raw materials. Small pelagic species occupy important positions in marine food webs, providing prey for larger fish, seabirds and marine mammals. They can also be eaten directly by people. In parts of the world where fish is an important source of affordable protein and micronutrients, diverting edible fish into animal feed raises a second sustainability question alongside the ecological one: not only how much food a system produces, but also who has access to the nutritional value moving through it.
Recent analysis has brought this tension back into focus. An investigation by the Changing Markets Foundation, supported by modelling of future aquaculture-feed demand, argues that rapid expansion of feed-intensive aquaculture could substantially increase pressure on wild fish resources if current production patterns continue. The organisation contrasts this pathway with greater expansion of species such as bivalve shellfish and some low-trophic aquaculture, which require little or no fishmeal derived from wild capture (Changing Markets Foundation, 2026).
The precise scale of future demand depends on assumptions about growth, species, feed conversion, substitution and technological change, so projections should not be mistaken for inevitable outcomes. But the underlying question is difficult to avoid.
What exactly are we trying to replace?
If the objective is simply to produce more tonnes of seafood without increasing the direct capture of the species consumers prefer to eat, intensive aquaculture can appear highly successful. If the boundary expands to include the fish caught for feed, pressure on marine ecosystems, competition for nutritious resources, land used to produce alternative feed ingredients and the environmental impacts of farming itself, the picture becomes more complicated.
This is not an argument against aquaculture.
In fact, treating all aquaculture as one system would reproduce exactly the mistake that sustainability analysis should avoid. A mussel attached to a rope, a carp raised in a pond and a salmon raised on formulated feed are all described as aquaculture, but their resource requirements and environmental relationships can be profoundly different. Some forms of aquaculture can provide nutritious food with relatively low environmental pressures, while others involve substantial feed, energy, pollution or ecosystem-management challenges (Gephart et al., 2021).
The distinction matters because sustainability is often attached to categories when it should be attached to systems.
Farmed does not automatically mean sustainable any more than wild, local, organic or plant-based does. Each label tells us something about how a product was produced, but none tells us everything about the resources, trade-offs and consequences behind it.
This is where the idea of a solution begins to change.
A sustainability intervention can solve the problem visible directly in front of it while moving another problem somewhere less visible. Replacing marine ingredients with crops may reduce demand for wild fish but increase demand for agricultural land. Improving feed efficiency may reduce the amount required per kilogram of farmed fish while industry growth increases total feed consumption. Turning fisheries’ by-products into feed can recover nutrients that might otherwise be wasted, but it cannot create an unlimited resource.
The problem has not necessarily disappeared.
Its location may simply have changed.
This is why the boundaries we draw around a food system matter so much. A narrow boundary asks whether aquaculture can produce fish without catching that same fish from the ocean. A wider one asks what the farmed fish eats, where those ingredients originate, what else could have consumed them, what ecosystems support them and what happens to the wastes that leave the farm.
Widen the boundary far enough and sustainability stops looking like a property of an individual product.
It becomes a property of relationships.
Between predator and prey. Between farm and ocean. Between efficiency and scale. Between food and feed. Between what one population consumes and what another might have eaten.
The most sustainable solution, then, may not always be the one that replaces an existing system most efficiently. Sometimes it is the one that changes what the system asks for in the first place.
And sometimes, before asking whether a solution works, we need to ask what the solution is eating.
THE GEOGRAPHY WE DON’T SEE
Every food system has a geography.
Some of it is obvious. Fields occupy landscapes, fishing boats leave ports, cattle graze on pasture and orchards follow climates in which particular fruits can grow. But the geography of the food we consume is much larger than the geography immediately around us.
International trade has stretched the distance between consumption and consequence.
A cup of coffee drunk in London requires no British coffee plantation. A bar of chocolate contains no visible trace of the land on which its cocoa was grown. The water embodied in a tomato does not arrive in the supermarket as water, nor does the habitat altered to produce soy appear alongside it on the ingredients label.
The commodity moves.
Much of its environmental history stays behind.
For a country such as the United Kingdom, this hidden geography is substantial. Recent research estimating the global land footprint of UK food and feed imports found that around 11 million hectares of agricultural land outside the country are currently associated with supplying those imports (Alexander et al., 2026). For comparison, roughly 17 million hectares within the UK itself are used for agriculture. The two figures measure different things and should not simply be added together, but their scale reveals something easily missed when we think about a national food system: a significant part of the land that sustains British consumption is not British land at all.
Imports therefore do more than move food between countries.
They connect one country's consumption to another country's landscape.
This connection becomes especially important when environmental impacts are measured where they occur rather than where the final product is consumed. The UK's own Food Security Report tracks pressures associated with British consumption beyond national borders, including greenhouse-gas emissions, deforestation, biodiversity loss and water scarcity. Its indicators show, for example, that the scarcity-weighted use of surface and groundwater associated with UK food consumption increased between 2019 and 2021, with wheat, rice, maize, sugar cane and olives among the important contributors (Defra, 2024).
Calling this virtual water or an environmental footprint can make it sound almost abstract. But the consequences are physical.
Water is withdrawn from an actual river or aquifer. Agricultural land occupies a real ecosystem. Carbon stored in vegetation and soils can be lost through land-use change. Habitat alteration affects species living in particular places.
The distance between producer and consumer does not make those pressures less real.
It makes them less visible.
New research is beginning to map that separation with increasing precision. Liu et al. (2026) traced long-term carbon-storage and biodiversity losses associated with land use through global agricultural supply chains. Rather than treating environmental damage as evenly distributed across agricultural land, the study found strong geographical concentration: approximately two-thirds of the combined mapped losses occurred within only one-third of the agricultural area considered.
Food consumption drove 83 per cent of the losses identified by the study, but those losses did not necessarily occur where the food was consumed. International trade redistributed their economic drivers across borders. Brazil emerged as the largest net exporter of both carbon-storage and biodiversity losses in the analysis, while China was the largest net importer, illustrating how consumption in one part of the world can become connected to ecological change in another (Liu et al., 2026).
This does not mean that international trade is inherently environmentally damaging.
Trade can allow food to be produced in places where land, climate or other resources make production more efficient. It can diversify supply, improve access to foods that cannot be produced locally and, as we have already seen, provide resilience when harvests fail in individual regions. Producing everything domestically would not automatically reduce environmental pressure; in some cases, relocating production to a less suitable environment could increase it.
The question is therefore not whether food should cross borders.
It is whether environmental responsibility should disappear when it does.
National sustainability statistics can become misleading if improvements within a country's borders are achieved partly by moving resource-intensive production elsewhere. A country might restore land, reduce agricultural pollution or lower territorial emissions at home while continuing to consume products whose impacts occur abroad. From the perspective of the national landscape, environmental performance may appear to improve. From the perspective of the global system, part of the pressure may simply have changed address.
This is sometimes described through consumption-based accounting: instead of asking only what environmental impacts occur inside a territory, it asks what impacts are associated with the goods and services consumed by the people within it.
For food, that change in perspective can be profound.
It turns a supermarket into something resembling a map.
Coffee connects a cup to a growing region. Animal feed connects livestock production to cropland elsewhere. Fruit connects a refrigerated aisle to water availability in another climate. Meat and dairy connect diets not only to the pasture we can see but also to land used to grow feed that may have crossed several borders before reaching the animal.
The supply chain is therefore also a chain of landscapes.
And those landscapes do not experience environmental pressure equally.
This is one reason why a kilogram of the same agricultural commodity can carry very different ecological consequences depending on where and how it was produced. Biodiversity is particularly sensitive to geography. Converting land in an ecosystem containing many endemic or threatened species is not ecologically equivalent to using the same area in a landscape with a different biological composition. Water has a similar geography: consuming a cubic metre where water is abundant is not equivalent to consuming it in a basin already experiencing severe scarcity.
A global food system can obscure these differences because commodities become increasingly interchangeable as they move through markets.
A tonne becomes a tonne.
A price becomes a price.
But a landscape is not interchangeable.
Once this is recognised, the meaning of a sustainable food system expands again. It cannot be judged solely by the condition of the country in which the food is eaten or even solely by the environmental performance of its domestic agriculture. Its boundaries have to extend along the supply chains that sustain consumption.
That creates an uncomfortable question for wealthy importing countries.
Can a food system be called sustainable at home if a significant part of the environmental pressure required to sustain it occurs somewhere else?
There is no simple line separating legitimate trade from outsourced environmental damage. Nor would retreating into complete national self-sufficiency necessarily solve the problem. The more useful principle is one of visibility: if consumption depends upon land, water, ecosystems and labour elsewhere, those dependencies should remain part of how sustainability is measured.
The geography of food does not end at the border.
Neither should the geography of responsibility.
ADVERTISEMENT
This creates one of the central tensions of food-system transformation.
Food must reflect more of its real environmental and social costs without becoming less accessible to people with the least ability to absorb higher prices.
There is no single mechanism capable of resolving that tension. Agricultural policy, social protection, wages, public procurement, infrastructure, research and food environments all influence what is produced, what it costs and what people are able to choose. This is why placing responsibility primarily on the individual consumer is inadequate.
Consumers make choices, but they make them inside systems that have already made many choices for them.
What crops receive support. Which foods are available nearby. What appears inexpensive. How products are marketed. What information appears on packaging. Whether someone has time to cook. Whether a school serves nutritious meals. Whether a farmer can afford to change production practices. Whether a supermarket accepts the imperfect appearance of a crop.
The sustainable choice cannot depend on every person becoming an expert in agricultural supply chains before buying dinner.
This does not make individual behaviour irrelevant. Diets matter, particularly in high-consuming populations where shifts away from resource-intensive foods can substantially reduce environmental pressures. But consumer behaviour becomes much more powerful when the surrounding system makes lower-impact, nutritious choices normal, accessible and affordable rather than exceptional.
Production has to change too.
A sustainable system would protect the productive capacity of soil rather than treating fertility as an unlimited resource. It would manage water according to local scarcity rather than assuming every litre has the same environmental cost. Nutrients would be used more efficiently and, where possible, recovered rather than continually lost. Agricultural landscapes would have to produce food while retaining enough ecological function to support biodiversity, pollination, water regulation and resilience.
None of this requires rejecting technology.
Improved crop varieties, precision agriculture, alternative proteins, better irrigation, nutrient recovery, food-storage technologies and new feed ingredients may all contribute. But the lesson from the systems we have already examined is that technologies should be evaluated by what they change across the wider system, not simply by the problem immediately in front of them.
Efficiency remains valuable.
It is simply not the final measure.
Nor does sustainability require rejecting global trade. A resilient food system will almost certainly continue to exchange food across borders. Trade can diversify supply and allow production to occur where conditions are favourable. But the environmental and social consequences embedded in traded food cannot disappear from accounting merely because they occur elsewhere.
A more sustainable system would therefore be neither completely local nor completely global.
It would be diverse.
Diverse in crops, sources, production methods and supply routes. Diverse enough that the failure of one region, species, technology or supplier does not become the failure of the whole. Diversity can carry costs in a system optimised for maximum efficiency, but it can also provide something efficiency tends to remove: redundancy.
And redundancy is often what resilience looks like before it is needed.
Perhaps, then, a sustainable food system would not look radically different in every visible detail from the one we have now. Supermarkets would still exist. Food would still cross oceans. Farmers would still use technology. Trade, processing and convenience would remain part of how billions of people eat.
The deeper difference would lie in what the system is designed to preserve.
Not only output, but productive capacity.
Not only low prices but also access to nutritious food and viable livelihoods.
Not only efficiency, but resilience.
Not only domestic environmental improvements, but also the landscapes and communities embedded throughout supply chains.
Not only today's harvest, but also the possibility of another harvest tomorrow.
That is why sustainability cannot be reduced to a single metric. Carbon matters, but so do biodiversity and water. Yield matters, but so do nutrition and affordability. Resilience matters, but so does the condition of the system being made resilient.
The difficult part is not choosing which one matters.
It is building a food system capable of holding them together.
SO WHAT WOULD SUSTAINABLE ACTUALLY LOOK LIKE?
By this point, sustainability may seem to have become almost impossibly demanding.
A food system must produce enough food, but production alone is not sufficient. It must withstand disruption without becoming dependent on responses that weaken its future. It must use soil, water and nutrients without exhausting them. It must recognise environmental pressures beyond national borders. It must consider livelihoods as well as landscapes, nutrition as well as yield, and the consequences of solutions as well as the problems they were designed to solve.
If everything is connected, where do we begin?
Perhaps by abandoning the idea that a sustainable food system should look like one particular thing.
There is no single crop, farming method, diet or supply chain that can carry the meaning of sustainability on its own. A tomato grown locally in a heated greenhouse may involve different trade-offs from one grown outdoors and transported from another country. Organic production can reduce reliance on synthetic pesticides and fertilisers while requiring more land for some crops. A plant-based ingredient can have a relatively low climate impact while still being associated with water stress or biodiversity pressure in a particular place.
Sustainability is therefore not a label attached to a product.
It is a set of conditions that the system must continue to satisfy.
One of the most influential definitions describes a sustainable food system as one that provides food security and nutrition for all while preserving the economic, social and environmental foundations required to provide them for future generations (HLPE, 2014). Hidden inside that definition are three demands that cannot easily be separated: the system must work economically, socially and environmentally.
More recent research has attempted to make those broad ambitions measurable. A global food-systems assessment published in Nature Food in 2026 developed targets and benchmarks across 44 indicators, spanning diets and nutrition, environment and production, livelihoods and equity, resilience and governance. Its conclusion was not that every country should converge on an identical food system. Rather, the framework makes visible how performance in one dimension can coexist with weakness in another — and how far current trajectories remain from many of the targets associated with healthier and more sustainable food systems (Carducci et al., 2026).
This multidimensional view matters because food systems exist to do more than minimise environmental damage.
They have to feed people.
A system that protects biodiversity but leaves nutritious food unaffordable is not sustainable. Nor is there one that produces abundant calories while contributing to poor diets, or one that achieves low consumer prices by making farming economically unviable for the people producing the food.
Nutrition therefore belongs inside the sustainability question, not beside it.
The same is true of affordability. The world's food system produces enormous quantities of food, yet access to a healthy diet remains deeply unequal. FAO and its partner organisations estimated that approximately 2.6 billion people were unable to afford a healthy diet in 2024 (FAO et al., 2025). Sustainability cannot mean designing an environmentally ideal diet that large parts of the population cannot realistically obtain.
Consumers make choices, but they make them inside systems that have already made many choices for them.
BEYOND THE FULL SHELF
Return to the supermarket shelf.
It looks much the same as it did when we began. The tomatoes are still there. So are the rice, coffee, bread, fruit and fish. Prices may have changed, origins may shift with the season, but the basic promise remains remarkably consistent: food will be here when we come looking for it.
What has changed is what the shelf contains.
Not physically, but conceptually.
Behind it now sits a much larger landscape. Soil and water. Fertiliser and feed. Farms and fisheries. Energy, labour and transport. Rivers, aquifers and ecosystems that may lie thousands of miles from the place where the final product is sold.
The shelf is not the food system.
It is where the food system becomes visible.
That distinction matters because visibility has shaped much of the way we think about sustainability. We notice packaging because we hold it in our hands. We notice food waste because we throw it away. We notice the price because we pay it. But many of the conditions that determine whether food can continue to reach us remain distant from the moment of purchase.
A depleted aquifer does not appear beside the price.
Lost biodiversity has no place on the receipt.
A farmer's shrinking margin is not listed among the ingredients.
Nor is the resilience created by having another supplier, another crop or another route available when the first one fails.
None of this makes the individual choice meaningless. What we eat matters. What we waste matters. Demand influences what is produced, and millions of individual decisions can collectively reshape markets.
But the journey through the food system reveals the limits of treating sustainability primarily as a question of choosing the correct product.
There may be no correct product in isolation.
The same food can carry different consequences depending on where it was grown, how water was managed, what fed the animal, what energy powered production, how workers were treated, what ecosystem was altered and what happened to the nutrients and wastes left behind.
Sustainability emerges from those relationships.
This is why a resilient food system cannot simply be one that survives the next disruption. Nor can a sustainable one simply be the system with the lowest carbon emissions, the shortest supply chains or the greatest agricultural output.
It has to hold several things together that are easier to optimise separately.
Enough food.
Nutritious food.
Affordable food.
Viable livelihoods.
Functioning ecosystems.
Resources that can recover.
Supply chains capable of adapting when conditions change.
And a distribution of environmental costs that does not allow one place to appear sustainable simply because another carries its burden.
There will be trade-offs between these goals. There will be technologies that solve one problem while creating another, production methods that work well in one landscape but poorly in another, and policies whose benefits become visible only over time.
That uncertainty is not evidence that sustainability has no meaning.
It is evidence that sustainability cannot be reduced to a shortcut.
Not local.
Not organic.
Not efficient.
Not even resilient.
Each can be part of the answer. None is the answer by itself.
Perhaps the better way to think about a sustainable food system is not as a finished model waiting to be implemented, but as a system capable of continually asking what its own success depends upon — and whether those foundations are becoming stronger or weaker as a result.
That changes the question we ask of the supermarket shelf.
Instead of only asking whether there is enough food on it today, we begin to ask what had to remain healthy, productive, viable and available for that food to arrive there — and whether those same conditions will still exist for the people who come after us.
A full shelf is an extraordinary achievement.
But perhaps a sustainable food system is not one that simply keeps the shelves full.
It is one that can keep filling them without emptying the systems behind them.
REFERENCES
Arendarczyk, B., Rabin, S. et al. (2026) ‘Global land footprint of UK food and feed imports under future socioeconomic scenarios’, PLOS ONE, 21(6), e0352499. doi: 10.1371/journal.pone.0352499. PubMed Central (PMC)
Brownlie, W.J., Sutton, M.A., Heal, K.V., Reay, D.S. and Spears, B.M. (eds.) (2022) Our Phosphorus Future. Edinburgh: UK Centre for Ecology & Hydrology. doi: 10.13140/RG.2.2.17834.08645. The OPF Report
Carducci, B., Schneider Lecy, K., Nordhagen, S. et al. (2026) ‘Food systems performance evaluated against targets and benchmarks reveals urgent gaps and a path to 2050’, Nature Food, 7, pp. 802–812. doi: 10.1038/s43016-026-01379-0. Nature
Changing Markets Foundation (2026) Fishing the Feed. Changing Markets Foundation. Published 3 September 2026. Changing Markets
Department for Environment, Food & Rural Affairs (Defra) (2024) United Kingdom Food Security Report 2024. London: UK Government. GOV.UK
Elshani, N., Adler, S., Tidåker, P., Sommerseth, J.K., Koesling, M. et al. (2026) ‘Environmental impacts of a biorefinery integrated into dairy farming system’, npj Sustainable Agriculture, 4, Article 76. doi: 10.1038/s44264-026-00189-y. Nature
FAO (2024) The State of World Fisheries and Aquaculture 2024: Blue Transformation in Action. Rome: Food and Agriculture Organization of the United Nations.
FAO and ITPS (2015) Status of the World's Soil Resources: Main Report. Rome: Food and Agriculture Organization of the United Nations and Intergovernmental Technical Panel on Soils. Agris
AO, IFAD, UNICEF, WFP and WHO (2025) The State of Food Security and Nutrition in the World 2025: Addressing High Food Price Inflation for Food Security and Nutrition. Rome: FAO. FAOHome
Gephart, J.A., Henriksson, P.J.G., Parker, R.W.R. et al. (2021) ‘Environmental performance of blue foods’, Nature, 597, pp. 360–365. doi: 10.1038/s41586-021-03889-2. Nature
HLPE (2014) Food Losses and Waste in the Context of Sustainable Food Systems. A report by the High Level Panel of Experts on Food Security and Nutrition of the Committee on World Food Security. Rome: Committee on World Food Security. FAOHome
Liu, B., Behrens, P., Sun, Z. and Scherer, L. (2026) ‘High-resolution carbon and biodiversity mapping shows correlated losses across space and agricultural products’, Nature Food, 7, pp. 792–801. doi: 10.1038/s43016-026-01387-0. Nature
National Audit Office (NAO) (2026) Resilience of the Food Supply Chain to Disruptions. HC 587, Session 2026–27. London: National Audit Office. Published 4 September 2026. National Audit Office (NAO)
Naylor, R.L., Hardy, R.W., Buschmann, A.H., Bush, S.R., Cao, L., Klinger, D.H., Little, D.C., Lubchenco, J., Shumway, S.E. and Troell, M. (2021) ‘A 20-year retrospective review of global aquaculture’, Nature, 591, pp. 551–563. doi: 10.1038/s41586-021-03308-6. DOI
Tendall, D.M., Joerin, J., Kopainsky, B., Edwards, P., Shreck, A., Le, Q.B., Kruetli, P., Grant, M. and Six, J. (2015) ‘Food system resilience: Defining the concept’, Global Food Security, 6, pp. 17–23. doi: 10.1016/j.gfs.2015.08.001. ScienceDirect
Trnka, M., Meitner, J., Balek, J. et al. (2026) ‘Climate-induced severe water scarcity events as harbingers of global wheat price’, Earth's Future, 14(8). doi: 10.1029/2025EF006095. AGU Journals
United Nations Environment Programme (UNEP) (2019) Frontiers 2018/19: Emerging Issues of Environmental Concern. Nairobi: UNEP. See ‘The Nitrogen Fix: From Nitrogen Cycle Pollution to Nitrogen Circular Economy’.
