Every kilogram of food on the planet carries a hidden water bill, and for the first time researchers have compiled an open database that breaks that bill into its three distinct parts: the rainwater consumed by growing crops, the freshwater drawn from rivers and aquifers, and the water notionally needed to dilute the pollution food production leaves behind. The new compilation, published in Frontiers in Sustainable Food Systems, synthesizes water footprint values from the published literature into 193 food items, 36 food subgroups and 14 food groups, and its analysis confirms a stark divide: animal-based products score significantly higher than plant-based foods on all three water components simultaneously.
Most openly accessible food footprint datasets report only a single aggregate value per commodity, which means that a tomato grown on rainfall and a tomato grown with irrigated water from a stressed aquifer can appear identical. The authors of the new database argue that these three quantities describe different pressures on water resources, are not interchangeable, and carry different implications for anyone trying to manage water sustainably.
To build the database, the team, led by Luigi Alessandrino of the University of Campania Luigi Vanvitelli together with colleagues at the CMCC Foundation, searched SCOPUS and Google Scholar and scrutinized 1,020 papers for eligibility, ultimately retaining 53 studies that reported the green, blue and gray components separately. These were combined with qualifying values from two earlier major reviews by Petersson and colleagues and by Poore and Nemecek. All values were converted to liters per kilogram, and where production involved more than one food item, mass allocation was used to split the footprint. The final database contains 790 green water records, 793 blue water records and 652 gray water records, and is released openly alongside descriptive statistics and the number of records underlying every aggregated value.
The statistical analysis was deliberately conservative. Because the Shapiro-Wilk test rejected normality for every distribution examined, the researchers used non-parametric methods throughout: a Mann-Whitney rank-sum test to compare animal-based with plant-based products, and a Kruskal-Wallis analysis of variance on ranks with Dunn’s post-hoc comparisons and Bonferroni correction to contrast the 14 food groups. The comparison confirmed that animal products exceed plant-based foods for green, blue and gray water alike, a pattern the authors attribute largely to the feed conversion ratio, since several kilograms of feed are needed to add one kilogram of animal body weight, on top of the water required for hydration, sanitation and farm operations.
At the subgroup level, the analysis showed that aquatic foods such as fish and crustaceans, along with unprocessed and processed legumes, carry lower water footprints than traditional animal protein sources including cheese, beef, pork and sheep meat. The authors note that this pattern is consistent with the planetary health diet, which proposes increasing plant-based consumption within environmentally sustainable dietary boundaries. They caution, however, that the sustainability of aquatic foods must also account for fish population trends and the overexploitation of the main consumed species’ stocks, pressures that a water footprint indicator alone cannot capture.
Within plant-based foods, a second significant distinction emerged between tree crops and herbaceous crops. Tree crops, with more extensive root systems and greater leaf area, tend to lose more water to the atmosphere and therefore show higher green water footprints, and where intensively managed they may also demand more irrigation and chemical inputs. But the composition of the footprint differs systematically: in the database, olive oil is dominated by the green component, reflecting a tree crop highly adapted to water-limited Mediterranean conditions, whereas seed oils from herbaceous crops show a larger blue water share, consistent with their shallower roots and reliance on irrigation during dry periods. The authors stress that this ordering is not an ecological ranking. Which profile is preferable depends on local water availability, management practices, and co-benefits such as soil conservation, biodiversity and carbon sequestration that lie outside the indicator entirely.
A parallel pattern appears in the nuts and seeds group, where nearly all tree crops are nuts and nearly all herbaceous crops are seeds, and the seeds show both lower blue and lower green water footprints. The researchers again urge caution in interpretation: almonds and pistachios are well suited to Mediterranean semiarid conditions, chestnuts are largely rainfed in mountain areas, and intensive nut production systems can exert substantial pressure on water resources. Yield per hectare, meanwhile, emerges as a strong modulator of the footprint per unit of product, which helps explain why fruit carries a much lower water footprint than nuts despite orchards typically being irrigated, and why many spices show extremely high values.
One result may surprise readers who assume rice is the thirstiest crop. In the database, rice shows a total water footprint of 1,281 liters per kilogram, built from a median green component of 693 liters, a median blue component of 401 liters and a median gray component of 187 liters, a total far from the highest values recorded. The authors point out that the global average water footprint of paddy rice between 2000 and 2004 was calculated at 1,325 liters per kilogram, a value close to their own, and that the figure reflects the substantial water percolation occurring in paddies. Percolating water is not considered to leave the system effectively, since it returns to the groundwater supply rather than being consumed.
The gray water component proved the most problematic. The team retained gray values only where they had been computed from nutrient loads, in practice nitrogen leaching from fertilization, and excluded estimates that also accounted for pesticides, fungicides and herbicides, because the two families of approaches differ by one to three orders of magnitude for the same commodity and cannot be pooled without artificial dispersion. As a consequence, none of the gray values in the database derive from pesticide-inclusive calculations. The authors identify several sources of uncertainty in this component, including the leaching fraction assumed for applied nutrients, which varies with soil, climate and irrigation regime and is rarely measured on site, and the water quality standard chosen as the reference concentration for dilution, which differs across national and international frameworks. They conclude that the gray component, though conceptually the most informative about the impact of agricultural practice on water quality, remains the least robust of the three.
The database also exposes how fragmented the evidence base remains. Of the 193 items, 123 rest on fewer than three records for green and blue water, and 135 for gray water, so medians could be computed for only a minority of items. Records originate from just 52 countries, covering 32 percent of the world’s nations, with sparse coverage of Africa in particular, and a substantial share of values refers to global averages rather than specific production contexts. The aggregated values should therefore be read as representative of the state of published knowledge, not of world agricultural production, and applying a global median to a specific region may be misleading, particularly for the blue component. The authors also flag that their total footprint values are synthetic indicators, since the sum of three independently computed central values falls below the median of record-level totals by 6 percent on average and by up to 51 percent for individual items. Ultimately, they emphasize, the water footprint measures volumes consumed or polluted but not the scarcity of the water involved or the ecological consequences of its abstraction, and it should be interpreted alongside carbon footprint, land use, biodiversity and nutritional quality rather than in isolation.
Subject of Research: Disaggregated water footprint of food commodities across green, blue and gray water components
Article Title: Green, blue, and gray water contribution to the water footprint of food products
Article References: Alessandrino, L., Masini, C., Palmieri, M., Mastrocicco, M., Chiriacò, M. V., & Castaldi, S. (2026). Green, blue, and gray water contribution to the water footprint of food products. Frontiers in Sustainable Food Systems, 10, Article 1917574. https://doi.org/10.3389/fsufs.2026.1917574
Image Credits: AI Generated
DOI: 10.3389/fsufs.2026.1917574
Keywords: water footprint, green water, blue water, gray water, food sustainability, animal products, plant-based foods, tree crops, rice, nitrogen leaching, database, sustainable diets
