Planetary health and food systems

TL;DR. The global food system is the largest single driver of environmental change on Earth — roughly 25-30% of all greenhouse-gas emissions, ~70% of freshwater withdrawal, ~78% of freshwater and coastal eutrophication, and the dominant driver of biodiversity loss and land-use change. The 2019 EAT-Lancet Commission, co-chaired by Walter Willett and Johan Rockström, gave the first integrated framework: a Planetary Health Diet that is roughly half vegetables and fruit, dominated by plant proteins, with red meat capped at about 14 grams a day. Beef sits at the center of the problem — a kilogram emits roughly 60 kg CO₂-equivalent against ~6 kg for poultry and under 1 kg for legumes. Aquaculture has overtaken wild-caught seafood as the dominant source of fish, and 35% of wild stocks are overfished. The honest news from Poore-Nemecek's 2018 Science synthesis: type of food beats food miles by roughly an order of magnitude, transport averages under 10% of emissions, and 73% of food emissions happen at the farm — which is where consumer substitution has its leverage. About 30-40% of U.S. food is wasted, and a third of that loss is at the household. Switching from a typical Western pattern to the Planetary Health Diet would cut diet-related GHG emissions, land use, and eutrophication by roughly 50%. Local-sourcing, plastic-straw bans, and "carbon-offset" beef are not in the same league.

What you'll learn

• Why food is the single largest driver of planetary environmental change, and the specific share of each major impact category.
• Which six of the nine planetary boundaries food crosses, and how.
• What the EAT-Lancet Planetary Health Diet actually prescribes.
• Why beef's footprint is roughly ten times poultry's and a hundred times legumes' — and why "grass-fed" is not the climate fix it is marketed as.
• The arithmetic of dairy, aquaculture, monoculture, and food waste.
• Why "buy local" loses to "eat plants" by a factor of ten, and where local does matter.
• Where in the food chain emissions actually occur — and where that locates the consumer's leverage point.

1. The food system's planetary load

The numbers come from the 2019 EAT-Lancet Commission, the Poore-Nemecek 2018 synthesis of 38,700 farms and 1,600 processors across 119 countries, and FAO statistical work. Food is responsible for:

• 25-30% of global greenhouse-gas emissions (EAT-Lancet gives 25%; Poore-Nemecek 26%; IPCC AR6 ranges 23-42% depending on whether you include fertilizer manufacture, refrigeration, and land-use change).
• About 70% of global freshwater withdrawals, almost entirely irrigation.
• About 78% of global freshwater and marine eutrophication — the nitrogen and phosphorus runoff that creates dead zones, including the Gulf of Mexico's seasonal hypoxic patch.
• The dominant single driver of biodiversity loss and tropical deforestation; roughly half of habitable land on Earth is used to grow food, with grazing alone occupying ~26% of ice-free land.

The EAT-Lancet headline finding: we cannot keep warming below 2°C without changing what we eat, even if every other sector decarbonized perfectly tomorrow. Marion Nestle's What to Eat Now (2025) renames this overlap triple-duty diets — largely plant-based eating reduces chronic-disease risk, reduces emissions, and reduces grain diverted to feed livestock. Walter Willett, who co-chaired EAT-Lancet, notes in Eat, Drink, and Be Healthy that U.S. agriculture alone produces 8-9% of national emissions, and beef accounts for 36% of diet-related emissions while supplying only 4% of the food supply.

2. Six planetary boundaries food crosses

The planetary boundaries framework — Rockström et al. (Nature 2009), Steffen et al. (Science 2015), Richardson et al. (Science Advances 2023) — identifies nine biophysical limits for safe human operation. The 2023 update found six transgressed. Food is the dominant driver of all six:

1. Climate change. Food produces ~13.7 Gt CO₂-e per year. Livestock alone is ~14.5% of global emissions (FAO Gerber et al., 2013).
2. Biosphere integrity. Agricultural expansion is the leading direct cause of vertebrate population decline tracked in the WWF Living Planet Index.
3. Biogeochemical flows. Synthetic fertilizers (Haber-Bosch nitrogen, mined phosphorus) have roughly doubled the natural nitrogen flux into ecosystems. Both N and P are outside the safe range.
4. Land-system change. Forty percent of habitable land is cropland or grazing; tropical deforestation continues for cattle pasture, soy (mostly animal feed), and palm oil.
5. Freshwater change. Agriculture withdraws 70% of freshwater; the 2023 update placed both "blue water" (rivers/lakes) and "green water" (soil moisture) outside the safe range.
6. Novel entities — synthetic chemicals new to biosystems. Pesticides, PFAS, microplastics, and pharmaceutical residues now contaminate ecosystems globally. Nestle reports PFAS detected in 44% of common U.S. fish samples.

The three boundaries food does not dominate are ocean acidification, atmospheric aerosols, and stratospheric ozone. On the others, food is the largest single human pressure.

3. The Planetary Health Diet

EAT-Lancet's operational answer to "what should 10 billion people eat by 2050?" is the Planetary Health Diet. For a 2,500-calorie reference adult, daily targets are roughly: vegetables ~300 g (half the plate), fruit ~200 g, whole grains ~232 g dry; plant protein dominant — legumes ~75 g, nuts ~50 g, soy ~25 g; animal protein modest — fish ~28 g, poultry ~29 g, eggs ~13 g, red meat ~14 g (about one serving per week, not per day), dairy ~250 g; unsaturated oils ~40 g, added sugars ≤31 g, starchy vegetables ≤50 g.

Not a vegan prescription. It is the diet that simultaneously meets nutritional needs for 10 billion people, stays within the six food-crossed boundaries, and reduces global diet-related mortality by 19-24% in EAT-Lancet's modeling. It looks remarkably like the Mediterranean diet the Seven Countries Study identified in the 1960s. Willett's Harvard Healthy Eating Plate, the 2020-2025 Dietary Guidelines for Americans, and EAT-Lancet sit within ~10% of each other across most categories.

The politically inconvenient piece in wealthy countries is the roughly 50% cut in red meat and doubling of fruits and vegetables the Commission identifies as the wealthy-country benchmark. The 2015 U.S. Dietary Guidelines Advisory Committee tried to include this. They were censored at industry behest via the appropriations bill — documented in Food Politics and re-litigated in What to Eat Now (2025).

4. Beef's outsized footprint

Across every LCA since 2010, beef is the highest-emitting common food. Poore-Nemecek's 2018 Science synthesis gives the standard numbers, in kg CO₂-e per kg of product: beef (beef herd) ~60; lamb ~24; cheese ~21; beef (dairy byproduct) ~17; pork ~7; poultry ~6; eggs ~4.5; tofu ~3; beans/lentils ~0.4-0.9; root vegetables ~0.4. A kilogram of dedicated-herd beef is 100-150x more emissions-intensive than a kilogram of legumes delivering similar usable protein. Four mechanisms:

• Enteric methane. Ruminants ferment cellulose in their four-chambered stomachs and burp methane — 28-80x CO₂'s 100-year warming potential. About 40% of beef's footprint is enteric methane.
• Feed conversion. Cattle need ~25 kg of feed per kg of edible beef, mostly corn and soy that could be eaten directly — a ten-to-one calorie loss.
• Land use and deforestation. Beef has the largest land footprint per gram of protein. Amazonian deforestation has been driven primarily by cattle pasture and secondarily by soy for cattle feed.
• Manure and nitrous oxide. Manure lagoons emit N₂O at ~265x the warming potential of CO₂.

The grass-fed loophole. Pasture-raised or "regeneratively grazed" beef is often defended as sequestering enough soil carbon to neutralize its emissions. The evidence doesn't support that. Willett reports grass-fed cattle live ~2x as long en route to slaughter, emitting methane the entire time, so lifetime per-kg footprint is comparable to feedlot. Grazed and Confused? (Garnett et al., 2017) found that under optimistic grazing systems, soil-carbon sequestration offsets only 20-60% of grazing emissions and saturates within 20-40 years. Animal-welfare and biodiversity benefits are real. Climate neutrality is not.

5. Dairy

Dairy carries the methane problem at smaller scale — cows are the same ruminants, and dairy herds produce ~4% of global emissions. A kilogram of cow's milk averages ~3 kg CO₂-e; cheese is ~21 kg. Industrial dairy concentrates manure pollution, antibiotic use, and water draw at a scale the rural midwestern landscape registers in impaired waterways.

The almond-vs-dairy water debate is misleading without LCA. A 2018 Science graphic comparing one liter of dairy milk to almond, oat, rice, and soy milk found dairy worse on every category — emissions, land, freshwater, eutrophication. Almond milk uses more water per liter than oat or soy (almonds are thirsty California crops in a drought-prone state) but still less than dairy. Oat milk has the lowest land use and second-lowest emissions. Soy has the best protein profile and a small footprint. Any plant milk is environmentally better than dairy; oat or soy are usually best. Almond is the worst plant milk and the dominant plant milk in California for marketing reasons.

6. Aquaculture and the collapse of wild stocks

The FAO's State of World Fisheries and Aquaculture (SOFIA 2022) reports 35.4% of assessed marine stocks fished at biologically unsustainable levels and another 57.3% at the maximum. Only ~7% have unused capacity. Aquaculture passed wild-caught as the dominant seafood source around 2014 and now supplies more than half of all seafood consumed.

Aquaculture is not automatically the green answer:

• Farmed salmon's feed problem. Salmon are obligate carnivores fed fishmeal and fish oil made primarily from wild-caught anchovies, sardines, and menhaden. Farmed salmon still consume roughly 1-1.5 kg of wild fish per kg produced.
• Sea lice, escapes, and disease. Net-pen farms concentrate parasites and pathogens that spill into wild populations; escapees interbreed with and outcompete wild fish.
• PCBs, dioxins, and PFAS concentrate in feed and accumulate in farmed flesh.

Shellfish (mussels, oysters, clams) and seaweed are the genuinely low-impact ocean foods — filter feeders that improve water quality, requiring no feed. Their CO₂ footprint per kg of protein is among the lowest of any food.

The reliable consumer signal is Monterey Bay Aquarium's Seafood Watch. MSC and ASC labels are useful but have documented gaps. No single seafood guide is fully sufficient.

7. Monoculture and biodiversity

The world grows ~6,000 food plants, but three crops — wheat, rice, corn — supply about 60% of human dietary calories. Add soybeans, sugarcane, and palm oil and the figure exceeds 75%. Most corn and soy is not eaten directly — it is fed to livestock or processed into oils, sweeteners, and biofuels.

Pollan's In Defense of Food documents the same collapse on the U.S. farm: a typical 1900 Iowa farm grew cattle, chickens, corn, hogs, apples, hay, oats, potatoes, cherries, wheat, plums, grapes, and pears. Today's grows corn and soy in rotation. Three-quarters of broiler chickens are a single Cornish-cross genotype. More than 99% of U.S. turkeys are Broad-Breasted Whites. Half of U.S. commercial broccoli is one variety, Marathon.

Three consequences:

• Nutrient density. Donald Davis at UT Austin documented (JACN, 2004) that 43 USDA-tracked crops lost 15-38% of measured iron, riboflavin, calcium, vitamin C, and zinc between 1950 and 1999. Modern wheat has ~28% less iron than 130-year-old varieties. Schatzker's The Dorito Effect extends the argument to flavor — aroma compounds are built from essential nutrients, and breeding for yield bred both out together.
• Seed-bank collapse. The FAO estimates 75% of plant genetic diversity has been lost since 1900. The gene bank is climate-adaptation insurance, and it is shrinking.
• Pest and disease vulnerability. Genetic uniformity invites catastrophic loss. The Cavendish banana, which displaced the Gros Michel after Panama disease in the 1950s, is now collapsing under TR4 fungus across plantations worldwide.

Monoculture is not inevitable. It is the outcome of yield-only commodity pricing, corn and soy subsidies, and an absent market signal for nutrient density, flavor, or diversity.

8. Food miles vs. food type

The most consequential consumer-leverage finding is the food miles myth. Hannah Ritchie's 2020 analysis of the Poore-Nemecek dataset found transport accounts for under 10% of food's total emissions on average, and under 1% for beef. Type of food beats food miles by roughly an order of magnitude.

• Local beef has a larger footprint than air-freighted lentils. Even with worst-case air freight, the difference is dwarfed by the beef-vs-legumes type effect.
• A vegetable grown in a heated European greenhouse can have a larger footprint than one shipped from open-field Spain, because heated greenhouses are energy-intensive. Production system beats distance.
• The exception: air-freighted produce. Air freight is ~50x as carbon-intensive per kg-mile as container shipping. Out-of-season berries, asparagus, and fresh tuna are the cases where "local and seasonal" matters — a small share of the food supply.

Where local does matter: traceability, freshness and flavor, regional farm economies, short food chains that restore accountability (Pollan), avoiding palm-oil and deforestation-linked imports, and seed-variety diversity. Local matters culturally and economically. It is not a climate strategy on its own.

9. Food waste

The U.S. wastes 30-40% of its food supply between farm and plate, per USDA and EPA. Globally, FAO puts the figure near 33%. About a third of that waste is at the household.

Project Drawdown ranks reducing food waste as the single highest-impact climate solution available, ahead of plant-rich diets and refrigerant management. The arithmetic: if the food system is ~25-30% of global emissions and a third of food is wasted, food waste alone is ~8-10% of global emissions.

Household practice that moves the number:

• Plan meals before shopping — most household waste is over-buying.
• Use the freezer (Pollan's "buy a freezer" rule).
• Trust eyes and nose more than the date label. "Best by" is a quality estimate, not a safety date; FDA does not require a date on most foods.
• Eat leftovers as the default lunch.
• Compost what you can't eat so carbon goes back to soil, not landfill methane.

This is the most consequential single change a household can make without any change in eating pattern.

10. Regenerative agriculture

Regenerative agriculture is the most contested term in current food-system discourse. The premise — that grazing or no-till cropping can sequester soil carbon at scale — is plausible at small scale and consistently oversold at large scale.

The strong claim: Allan Savory's 2013 TED talk, with ~9 million views, argued that "holistic planned grazing" could sequester enough carbon to reverse climate change. Ranchers, certifiers, and the marketing arms of several major beef companies have repeated it.

The evidence: WRI's Creating a Sustainable Food Future (Searchinger et al., 2019) and FCRN's Grazed and Confused? (Garnett et al., 2017) reviewed the literature on grazing-soil sequestration. Both concluded that under optimistic assumptions, sequestration offsets 20-60% of grazing emissions, the effect saturates within 20-40 years, and no peer-reviewed analysis supports the strong Savory claim. Sequestration is real. It is not large enough to neutralize beef.

"Regenerative" is not a certified term. Regenerative Organic Certified (Rodale, Patagonia, Dr. Bronner's, since 2018) layers soil, animal-welfare, and labor standards on USDA Organic but is uncommon. Most "regenerative" claims on packaging are unaudited marketing.

The honest position: regenerative practices likely improve soil, water-holding capacity, and biodiversity on managed grasslands. They do not make beef carbon-neutral. Buying "regenerative" beef is better than feedlot; eating less beef is much better than either.

11. What individual choice can shift — and what it can't

Poore-Nemecek 2018 published the figure that locates the consumer's leverage: producers (farms and supply chain to the farm gate) account for ~73% of food's emissions; transport, retail, and packaging account for the remaining ~27%. Substitution at the type of food level changes what is produced and is the highest-leverage move.

Their modeled scenario: shifting a typical Western diet to the Planetary Health Diet cuts food's GHG emissions by ~49%, land use by ~76%, freshwater withdrawals by ~19%, and eutrophication by ~49%. The same diet shift is associated with a ~19-24% reduction in adult mortality (EAT-Lancet). The two outcomes are coupled because the same patterns drive both.

High-leverage:

• Eat less red meat and dairy — the single largest dietary lever. Partial shifts (Meatless Monday, "vegan till 6," weekend-only beef) produce measurable reductions.
• Don't waste food. Eliminating a third of household waste is roughly equivalent to switching a third of meat consumption to plants.
• Eat more legumes, whole grains, and seasonal vegetables — active replacement, not passive avoidance.
• Choose lower-impact seafood: bivalves, small pelagics, ASC/MSC-certified options. Avoid farmed salmon as a default.
• Avoid palm-oil-driven deforestation. Read the ingredient list; most ultra-processed food contains palm oil.

Low-leverage but heavily marketed:

• Buying local without changing diet composition — small at best.
• Plastic straw and bag avoidance as a food-system intervention — real but tiny.
• "Carbon-offsetting" beef — offset markets are unverifiable at scale.
• Buying "regenerative" beef instead of eating less beef — a confused trade-off.

Individual leverage is real but bounded. The food system delivering ultra-processed food at low prices is a regulated political economy — corn and soy subsidies, USDA checkoff programs funding "Beef. It's What's for Dinner," tax-deductible marketing, externalized environmental costs. Demand change at scale shifts producer behavior, which at scale shifts policy. The political will to remove the regulatory tilt toward meat, dairy, sugar, and refined grain in place since the 1977 McGovern Goals is the larger lever behind the personal one.

FAQ

Q: Is the Mediterranean diet planetary-aligned?

Largely yes. The traditional Mediterranean pattern — vegetables, fruit, whole grains, legumes, olive oil, modest fish, modest fermented dairy, small red meat — sits within ~10% of EAT-Lancet on every category. The modernized Italian or Spanish diet (more meat, more refined grain, more sugar) drifts toward the Western pattern and loses most of the planetary advantage.

Q: Is "grass-fed" beef carbon-neutral?

No. The most generous published analysis (Rowntree et al., 2020) found one specific multi-paddock system net-zero over its assessment window. Grazed and Confused? and Creating a Sustainable Food Future concluded sequestration offsets 20-60% of grazing emissions at best, saturates within decades, and does not scale. Grass-fed beef typically has a lifetime per-kg footprint comparable to or higher than feedlot. Animal-welfare and biodiversity benefits are real; climate neutrality is not.

Q: Are oat-milk and almond-milk actually better than dairy?

Yes, both, on every category in the published LCA work. Oat is the best plant default across emissions, land, and water. Almond uses more water than other plant milks (still less than dairy) and is grown in water-stressed regions. Soy has the strongest protein profile per liter. Don't use almond milk as a proxy for "plant milk."

Q: What about palm oil?

It is the most productive oilseed by area, which is its trap — efficient enough to be profitable to plant in cleared rainforest. About 85% comes from Indonesia and Malaysia, where peatland and rainforest conversion is the single largest driver of deforestation and orangutan habitat loss. The RSPO certification is contested. Cutting ultra-processed intake cuts palm-oil exposure by default.

Q: Is "regenerative" certified?

Not federally. USDA Organic is the only federally enforceable agriculture certification. Regenerative Organic Certified (since 2018) is a private third-party standard layered on USDA Organic. Most "regenerative" claims on packaging are unaudited marketing — treat regenerative-without-ROC the way you'd treat "natural" without a definition.

Q: Can a vegan diet be unhealthy?

Yes. Ultra-processed vegan food (Beyond/Impossible burgers, vegan cookies, sugary plant milks, vegan ice cream) is still ultra-processed. NOVA 4 is NOVA 4 whether the protein came from a cow or a soybean. A vegan diet built around whole grains, legumes, nuts, seeds, vegetables, and fruit is among the most protective patterns measured. The category "plant-based" obscures more than it clarifies.

Q: Does buying organic help the planet?

Mixed. Per-acre, organic has lower impact on biodiversity, soil, and water pollution. Per-kg produced, organic yields are 20-25% lower, so the per-kg climate footprint is similar or slightly higher than conventional. Organic is a clear win on local biodiversity and farm-worker chemical exposure; not on global emissions per calorie. The high-leverage move remains what you eat, not how it was farmed.

Q: Where does plant-based meat (Beyond, Impossible) fit?

Better than beef on every environmental metric — roughly one-tenth the emissions, one-twentieth the land use, half the water — but still ultra-processed (NOVA 4). For someone replacing beef burgers, the planetary swap is real; for someone replacing beans, the swap is sideways. Treat plant-based meat as a transition food, not a destination.

Sources

• Willett, W. et al. (EAT-Lancet Commission). Food in the Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food systems. The Lancet, 2019. DOI: 10.1016/S0140-6736(18)31788-4.
• Poore, J. & Nemecek, T. Reducing food's environmental impacts through producers and consumers. Science 360, 987-992, 2018. DOI: 10.1126/science.aaq0216.
• Rockström, J. et al. A safe operating space for humanity. Nature 461, 472-475, 2009. Steffen, W. et al. Planetary boundaries: guiding human development on a changing planet. Science 347, 1259855, 2015. Richardson, K. et al. Earth beyond six of nine planetary boundaries. Science Advances 9, eadh2458, 2023.
• FAO. The State of World Fisheries and Aquaculture (SOFIA) 2022.
• Gerber, P.J. et al. (FAO). Tackling climate change through livestock: a global assessment of emissions and mitigation opportunities. FAO, 2013.
• Garnett, T. et al. Grazed and Confused? Ruminating on cattle, grazing systems, methane, nitrous oxide, the soil carbon sequestration question — and what it all means for greenhouse gas emissions. Food Climate Research Network (Oxford), 2017.
• Searchinger, T. et al. Creating a Sustainable Food Future. World Resources Institute, 2019.
• Davis, D.R. et al. Changes in USDA Food Composition Data for 43 Garden Crops, 1950 to 1999. Journal of the American College of Nutrition 23(6), 2004.
• Willett, W. Eat, Drink, and Be Healthy — Chapter 12, "The Planet's Health Matters Too." Free Press, revised 2017.
• Nestle, M. What to Eat Now (2025) — chapters on meat, dairy, fish, and triple-duty diets.
• Nestle, M. Food Politics — on USDA dual mandate, checkoff programs, and dietary-guidelines capture.
• Pollan, M. In Defense of Food — Part II, "The Western Diet and the Diseases of Civilization," on monoculture, the leaves-to-seeds shift, and pasture-based animal production.
• Schatzker, M. The Dorito Effect — on the dilution effect, breeding for yield, and lost flavor-and-nutrient density.
• Duyff, R.L. Complete Food and Nutrition Guide — on sustainability, GMOs, organic, aquaculture, and food waste.

Related modules

• How food gets engineered — the industrial substitution that drives the leaves-to-seeds shift and the rise of palm oil and seed oils.
• Big Food vs. public health — the regulated political economy that funds beef and dairy marketing and removed environmental advice from the 2015 Dietary Guidelines.
• Real food vs. edible foodlike substance — Pollan's frame for the consumer-side reorientation that aligns chronic-disease risk and planetary footprint.

Related glossary

• EAT-Lancet — the 2019 Commission and its Planetary Health Diet.
• Planetary boundaries — the nine biophysical limits framework (Rockström, Steffen, Richardson).
• GHG emissions — greenhouse-gas emissions, measured in CO₂-equivalent.
• Monoculture — single-species cropping, with its yield and biodiversity tradeoffs.
• CAFO — concentrated animal feeding operation, the dominant U.S. beef, pork, dairy, and poultry model.
• Aquaculture — fish and shellfish farming, now the majority of seafood produced.
• Food miles — the distance food travels, a smaller component of footprint than food type.
• Biodiversity loss — the dominant single driver is agricultural land conversion.
• Regenerative agriculture — uncertified umbrella term for soil-building grazing and cropping practices.
• Food waste — the 30-40% of U.S. food supply lost between farm and plate.
• Mediterranean diet — the closest pre-existing dietary pattern to the Planetary Health Diet.
• Planetary Health Diet — EAT-Lancet's operational dietary prescription for 10 billion people by 2050.