CTW Studio

Brief 002Global food system

What does eating animals cost—beyond the price tag?

Count the lives. Trace the land, emissions, forests, water and health effects. Keep the system boundaries—and the uncertainty—visible.

Data checked —Global unless labelledNo advocacy math
Follow the evidence

01 Scale before opinion

How many animals are killed for food every day?

Official global data can answer this for major land animals. For fish, it mostly records tonnes—so researchers must estimate individual lives from body weights.

Seven tracked land species · ——animals per average day
That is about—every minute

Source: FAOSTAT via Our World in Data, annual slaughter counts ↗ · Scope & caveats ↓

Observed · FAO / OWID

Daily slaughter, by species

World
Bars use a common linear scale. Counts divide the annual total by the number of days in the source year. Fish and species absent from the selected series are excluded.
View the values as a table

Source: FAOSTAT via OWID, seven land species, annual ↗ · Scope & caveats ↓

Estimated · farmed finfish · 2019

≈340M / day

Peer-reviewed midpoint: 124B a year. Plausible range: 214–468M/day.

Estimated · wild finfish · 2019

2.7–5.2B / day

Converted from catch tonnage and estimated body weights. This is a range, not a census.

Sources: Mood et al., farmed fish estimates ↗ + wild fish estimates ↗ · Scope & caveats ↓

02 Climate footprint

Livestock is a material climate sector—not the whole climate problem.

FAO estimates livestock agrifood systems emitted 6.2 gigatonnes of CO₂-equivalent in 2015, about 12% of human-caused greenhouse-gas emissions.

Livestock-system emissions

6.2 Gt CO₂e2015 baseline

Share of all anthropogenic GHG

≈12%FAO scope

Share of agrifood emissions

≈40%FAO scope

Source: FAO, livestock agrifood emissions, 2015 baseline ↗ · Scope & caveats ↓

Global meta-analysis · full supply chain

Not all proteins carry the same footprint

kg CO₂e / kg food
Global mean life-cycle estimates from Poore & Nemecek’s dataset. Producer-to-producer variation can be large; product averages describe systems, not every farm.
View the values as a table

Source: Poore & Nemecek, global food footprints, Science 2018 ↗ · Scope & caveats ↓

What changes the total most?

Diet composition matters more than food miles for many products. Replacing ruminant meat usually produces the largest climate reduction; improving farms also matters, but even low-impact animal products often exceed plant substitutes in this global dataset.

03 Land & forests

The land equation is the clearest imbalance.

Animal-sourced foods use most agricultural land while delivering a minority of calories. Pasture expansion for beef is also the largest single driver in the cited tropical-deforestation attribution.

Poore–Nemecek global food-system estimate

Share of input versus output

World · around 2010
83%agricultural land
animal-sourced foods
37%protein
provided
18%calories
provided
“Animal-sourced foods” includes pasture plus cropland used for feed. Land quality differs; not all grazing land could grow crops. The ratio still captures the opportunity cost for habitat and carbon storage.

Source: Poore & Nemecek, land/input-output estimates ↗ · Scope & caveats ↓

Scenario, not forecast

If the world used a plant-based food system

The modelled agricultural footprint falls from 4.1B hectares to about 1.0B—a 75% reduction. It includes less pasture and less feed cropland, while direct human-food cropland grows.

4.1B hacurrent mix
1.0B haplant-based

Source: Poore & Nemecek, modelled plant-based land scenario ↗ · Scope & caveats ↓

41%

of tropical deforestation

was attributed to pasture expansion for beef in the 2005–2013 study average—about 2.1M hectares per year.

77%

of global soy

was fed to livestock in the cited 2018 end-use analysis; about 7% went directly to human foods such as tofu and soy milk.

Sources: OWID, tropical deforestation drivers ↗ + food-system soy end use ↗ · Scope & caveats ↓

Boundary: 41% is not “all global forest loss today.” It is a commodity-attribution result for tropical deforestation in a defined historical study period.

04 Freshwater & nutrient pollution

Agriculture dominates the pressure. Attribution within agriculture is harder.

Irrigation, livestock and aquaculture share an interconnected system. Two large global numbers reveal the scale—but neither is a livestock-only share.

Global freshwater withdrawals

69%agriculture, including irrigation, livestock and aquaculture

Ocean & freshwater eutrophication

78%caused by agriculture in the cited food-system assessment

Sources: FAO AQUASTAT, withdrawals ↗ + OWID food-system eutrophication synthesis ↗ · Scope & caveats ↓

✓

We can say: feed production, fertilizer and manure release nitrogen and phosphorus; agriculture is the dominant global system boundary for eutrophication.

×

We cannot say: livestock alone causes 69% of withdrawals or 78% of eutrophication. These sources do not make that allocation.

05 Oceans

More than a third of assessed marine stocks are overfished.

The state of fish populations and the number of fish killed are different questions. FAO measures the former by stocks and production tonnes; researchers estimate the latter in individuals.

37.7%unsustainable
Unweighted share of marine stocks monitored by FAO, 2021. The production-weighted sustainability measure is higher because large, well-managed fisheries count more.

Aquatic animal production · 2022

Capture fisheries92.3M tonnes
Aquaculture94.4M tonnes

Aquaculture surpassed capture fisheries as the main producer of aquatic animals for the first time. That shifts pressure; it does not remove welfare, feed, pollution or habitat questions.

Source: FAO, State of World Fisheries and Aquaculture 2024 ↗ · Scope & caveats ↓

06 Human health

The health case is about replacement and planning—not a vegan halo.

Processed meat has an established causal link to colorectal cancer. A balanced vegan diet can provide needed nutrients, but “vegan” alone does not mean nutritious.

WHO / IARC

Group 1

Processed meat

Carcinogenic to humans. An analysis of ten studies estimated that eating 50g daily was associated with an 18% higher relative risk of colorectal cancer.

Same evidence class does not mean same danger as smoking. Classification measures confidence that it can cause cancer, not risk magnitude.

WHO / IARC

Group 2A

Red meat

Probably carcinogenic to humans, based on limited epidemiological evidence plus strong mechanistic evidence.

“Limited” means chance, bias or confounding could not be ruled out in the human evidence.

Source: WHO/IARC, red and processed meat evidence ↗ · Scope & caveats ↓

NHS guidance

A well-planned vegan diet can meet nutritional needs.

Variety, fortified foods and appropriate supplements matter. Special life stages deserve professional guidance.

  • B12
  • Vitamin D
  • Iodine
  • Iron
  • Calcium
  • Selenium
  • Omega-3

Source: NHS, vegan diet guidance ↗ · Scope & caveats ↓

07 The answer, for now

Does moving away from animal farming materially change the picture?

Yes—especially for animals, land, beef-driven deforestation and food-system emissions.

The strongest global evidence supports a shift toward plant-rich diets. It does not support collapsing every environmental or health problem into one livestock statistic, nor pretending all animal foods, farms, places or people are interchangeable.

Strong

Direct evidence

Animal slaughter counts; fish production; livestock emissions; animal-food land use; beef-linked tropical deforestation; processed-meat cancer evidence.

Shared

System-level evidence

Freshwater withdrawals, eutrophication and antimicrobial resistance involve animal farming, crop farming and other pathways. Attribution must stay explicit.

Missing

What data still fails us

Comparable global welfare measures, a true individual fish census, consistent intensive-farm definitions, and timely farm-level pollution reporting.

If the goal is impact rather than identity

  1. 01
    Reduce ruminant meat first.

    Beef and lamb dominate many land and climate comparisons.

  2. 02
    Replace—do not merely remove.

    Use varied legumes, grains, vegetables, nuts, seeds and fortified foods; plan B12.

  3. 03
    Count welfare as well as carbon.

    Switching from cattle to chicken can lower emissions while increasing the number of animals killed.

  4. 04
    Keep producer variation visible.

    Better farming practices matter even when global product categories remain far apart.

08 Evidence trail

Every headline number has a boundary.

Primary international statistics and peer-reviewed research take priority. Synthesis pages are used where they expose the underlying study and definition clearly.

01

Observe

Use FAO counts and assessments for what is directly measured.

02

Estimate

Label modelled fish counts as ranges, never official totals.

03

Attribute

Separate livestock-specific evidence from agriculture-wide shares.

04

Compare

Keep periods, units, denominators and study scopes visible.

Drivers of Deforestation ↗

Our World in Data, synthesizing Pendrill et al. · Updated 2024

Tropical deforestation attribution and global soy end use

Research synthesis

AQUASTAT methodology: Water use ↗

Food and Agriculture Organization of the United Nations · Living methodology

Global freshwater-withdrawal shares and category definitions

UN methodology and statistics

The vegan diet ↗

UK National Health Service · Living guidance

Nutritional adequacy and nutrients requiring planning

Public-health guidance

Antimicrobial resistance ↗

World Health Organization · 2026-07-16

Health burden and cross-sector drivers of antimicrobial resistance

Public-health fact sheet
Data and editorial method

Updateable observations. The land-animal count is the sum of seven major species in the latest World row of the FAO-derived OWID series. The update script also rebuilds the product-footprint comparison from the published Poore–Nemecek dataset.

Fish. FAO reports aquatic animals in tonnes. Individual totals are peer-reviewed estimates that divide production or catch by estimated species/country body weights. They are presented separately because uncertainty is large.

Scenarios. A 75% land reduction under a global plant-based diet is a modelled counterfactual. It is not a prediction of consumer behavior or a date-bound forecast.

Health. Relative risk is not absolute risk. Diet adequacy depends on foods, quantities, fortification, supplements and life stage—not on a label alone.

Refresh. Run python3 signals/scripts/update_food_data.py from ctw.studio, then review the diff, periods and source definitions before publishing.

Next Expand the lens

Housing: are homes becoming less attainable?

Prices, income, rents, mortgage costs, supply and household formation—without turning one national market into “the world.”

Read brief →