Background & Summary

Globally, food waste represents a significant loss of resources in terms of energy, labor, and economic value. On an annual basis, about a third of global food production, or 1.3 Gt, is thrown away across the value chain from the farm to consumers’ plates1. Food waste is composed of both avoidable and unavoidable food waste. Avoidable food waste (AFW) is the discarded edible (parts of) food products, intended for human food consumption while unavoidable food waste (UFW) is the expected, non-consumable waste stream2. This dataset follows Yussof et al.’s approach2, offering a classification of “waste avoidability” for each food product’s components based on the concepts of avoidable, potentially avoidable, and unavoidable food waste.

Avoidable food waste has been found to create needless environmental impacts across the world. AFW generated at the consumer stage has been found to emit 396 Mt CO2 eq (100 GWP), use 61 billion m3 (Gm3) of water and 198 million hectares of land (for food crops, feed crops, and grazing)3. Unavoidable food waste, by definition, cannot be readily reduced, yet its management as waste remains critical. The vast majority of these two waste streams currently ends up in landfills and open dumps, generating millions of tonnes of greenhouse gases every year4, especially due to their methane emissions5.

The most effective way to decrease the impacts of avoidable food waste is to reduce its generation at its source. As a result, in recent years researchers have conducted studies focused on reducing waste through consumer behavior and supply chain efficiencies. Studies have shown that plate size reduction6, promoting healthy food7, informational campaigns8, online food-sharing platforms9, and new legislations10 can help reduce AFW. Assessing the effectiveness of these solutions remains a challenge due to the difficulty of accurately measuring food waste flows11 and understanding their precise composition between avoidable food waste flows that can be reduced, and unavoidable food waste flows that are expected to remain stable, by definition.

While unavoidable food waste may not be reduced, it is critical to manage it to limit its environmental impacts at the waste treatment stage and explore valorization pathways to extract as much value as possible from these bio-resources12. As such, many studies have been conducted over the last decade to explore the valorization of food by-products - be it, for vegetable and fruit waste13, seafood14, nuts, cereals15,16, or meat waste17. By and large, these studies have explored both the nutritional content of many food by-products, as well as the generation of value-added products from these waste streams in the form of bio-fuels, bio-materials, value-added components, and bio-based adsorbents. A large majority of experimental studies operates at a lab scale, and therefore do not quantify or only superficially the available feedstock should these valorization solutions be scaled up to achieve industrial scale. Additionally, countries also elaborate their own food composition database (such as Japan18, France19, England20, Turkey21, and Norway22 however, similarly to food waste valorization studies, the vast majority focuses on the chemical and nutritional contents of food items. Some of these national tables may provide some indication of the mass share of the products that are edible or usually consumed, but they do not provide further details on the specific mass share of their different edible and inedible parts.

In this study, we present FOWCUS23, the first comprehensive analysis of the avoidable, potentially avoidable, and unavoidable components of 276 food commodities, post-harvest. The data is meticulously indexed according to FAOSTAT food commodities typology (incl. FAO FishStat typology for seafood products), enhancing its applicability for national-level quantification of food waste feedstock. This dataset not only fills a critical gap in existing literature but also empowers the development of strategies for the bio-chemical, bio-material, and bio-energy sectors. Thanks to its detailed accounting of the avoidable, potentially avoidable, and unavoidable parts of food commodities, this dataset also provides novel support in food waste quantification studies by helping predict the expected amounts of these different waste streams generated at the regional or national levels.

This dataset supports studies that use the ‘waste floor’ approach24,25 that is, those that aim to quantify the minimum amounts of food waste that may be generated in a specific geography, as well as national and global food waste monitoring studies26. This dataset can also be used as a tool to estimate the potential waste feedstock for bio-energy, bio-chemicals, bio-compounds, bio-absorbents, and nutrients, such as phosphorus, nitrogen, and potassium. Additionally, FOWCUS can be leveraged for more, stage-specific, supply chain analysis. For example, it can support the quantification of unavoidable food waste (e.g., by-products, residues) in specific food processing industries (e.g., slaughtering, milling, processed seafood) at an aggregated level, which could then be valorized by other industries, locally.

FOWCUS not only supports academic and industrial research but also assists policymakers in crafting targeted waste reduction strategies at a national level.

Methods

Literature review and data compilation

The list of commodities included in this study was retrieved from the Food and Agriculture Organization’s (FAO) statistical database (FAOSTAT), detailing the global production of agricultural commodities in each country27. Additionally, the list of seafood commodities was retrieved from FAO’s FishStat database, which detailed the aquaculture production and wild capture of seafood commodities for each country28. This dataset is extensive and reports the landing or aquaculture production of more than 3500 seafood species. Due to the very large number, only the top 100 species by annual production volume were selected for this dataset. The top 100 species amounted to more than 80% of the global volume of seafood (2019 as base year) and therefore cover the majority of seafood items by volume. The remaining species were then grouped into different seafood categories. This process led to the inclusion of 276 commodities in this dataset, grouped into 12 food categories: vegetables, fruits, nuts, eggs, livestock, seafood, cereals, sugar, vegetable oils, stimulants, pulses, and root vegetables.

The methodology for developing the dataset began with an extensive literature review for each of the food commodities included in the final list. The goal was to collate exhaustive data on the mass of both edible and inedible components of each commodity, ensuring comprehensive coverage across all food categories. For this data collection process, the main key search terms used in combination with each of the names of FAO commodities were: ‘mass composition’, ‘mass share’, ‘mass balance’, ‘residues’, ‘valorization’, ‘yield’, ‘waste’, and ‘weight’. The main bibliographic sources were Google Scholar, Google Search Engine, and SCOPUS. More specific words were also utilized related to the specific morphologies of the commodities (e.g., endocarp, exocarp, etc.).

The data sources acquired and reviewed were peer-reviewed articles, guidelines, and reports from national agricultural ministries or departments (e.g., USDA, Australian Department of Agriculture, Italian Agricultural Ministry), international datasets from international organizations (e.g., FAO), and in some cases, industry reports. The data sources are compiled in the FOWCUS dataset for each food commodity and listed next to the relevant specific food commodities’ parts (see Data Records).

Setting mass balance boundaries

To accurately represent the mass composition of food products post-harvest, mass balance equations were employed for each commodity. These calculations were crucial in determining how the total mass of each harvested food commodity was distributed among its various components. This approach ensured that the summed mass percentages of all components, accounted for at the post-harvest stage, equaled 100%. This was essential to maintain the integrity and usability of the data, particularly for research applications in waste management and by-product valorization. The boundaries were set by first reviewing FAO commodity definitions and the Food Balance Sheet handbook prepared by the FAO Statistics Division detailing what is included in the production volumes. The description of the boundaries of the food products of each commodity group is compiled in Table 1.

Table 1 Mass balance boundaries.
Full size table

Each food commodity has a wide diversity of parts. We aimed to harmonize these parts to the greatest extent possible without losing too much granularity in the specificities of each food commodity. Figure 1 illustrates a general sketch of the main food commodities parts included within FOWCUS for each of the 12 food groups. A majority of food commodities in each food group are composed of these parts, although some specific commodities within the same food group may still differ significantly due to their anatomy (e.g., olive and palm fruits in the Vegetable Oils group, artichoke in the Vegetables group). We highlighted which parts are not usually included in FAO production statistics (dashed boxes in Fig. 1).

Fig. 1
figure 1

Illustrative breakdown of typical components across 12 food groups used in the FOWCUS dataset.

Full size image

Mass balance calculations

Once the boundaries and the inventory of parts composing each food commodity were set, the mass balance equations could be established. The mass balance equation was balanced based on the different data sources collected for each commodity, such that the total sum of the mass of each part equaled 100% of the mass of the food commodity (Eq. 1).

$${Food}\,{commodit}{{y}}_{{x}}\,{mass}={part}\,{1}_{{food}{commodity}{x}}+{part}\,{2}_{{food}{commodity}{x}}+\ldots +{part}\,{{n}}_{{food}{commodity}{x}}=100 \% $$
(1)

Data standardization and verification

Following the data collection and calculation phases, the resulting data were standardized to ensure consistency across all entries. This process involved 1) harmonizing to the greatest extent the parts' names; and 2) reviewing the mass composition of each food commodity within the same food group to ensure consistency in the boundaries set and in the definitions of the different parts (e.g., husk vs. hull). During this step, each part (e.g., hull, shell) was assessed against the FAOSTAT commodity definition to highlight whether or not it was accounted for in FAOSTAT production datasets (see Data records for description).

This distinction is critical for the utilization of this dataset to quantify the amounts of unavoidable food waste generated during the production of these commodities. Indeed, for tree nuts commodities, such as almonds, the hull is not included in FAO production statistics. In the case of almonds, this dataset describes the composition of almonds, such as the hull, the shell, the skin, and the kernel amounting to 100% of the volume of this commodity. In the almond example, to quantify the volume of these four parts using FAOSTAT production data for a given country, the production amount was adjusted to include the almonds’ hulls. The adjustment is described in Eq. 2.

$${Food\; commodit}{y}_{{Adjusted\; production\; volume}}=\frac{{Food\; commodity}\,{X}_{{FAO\; production\; volume}}}{(1-\sum {mass\; share\; part}\,{x}_{{not\; accounted\; in\; FAO}})}$$
(2)

Where \({Food\; commodity}\,{X}_{{Adjusted\; production\; volume}}\) is the adjusted production volume (tonnes) of food commodity X, \({Food\; commodity}\,{X}_{{FAO\; production\; volume}}\) is the volume (in tonnes) retrieved from FAOSTAT production datasets, and \(\sum {mass\; share\; part}\,{x}_{{not\; accounted\; in\; FAO}}\) is the sum of the mass share of the parts not included in the FAOSTAT production datasets.

Once the production volume is adjusted to account for parts (e.g. almonds’ hull) not included in FAO production, the mass of each part can be readily computed by using the mass share estimate from the FOWCUS dataset.

$${Part}\,{{x}}_{{Commodity}{X}}={Foodcommodity}\,{{X}}_{{Adjusted}{produciton}{volume}}\ast {massshar}{{e}}_{{part}{x}}$$
(3)

Where \({Part}\) \({x}_{{Commodity\; X}}\) is the mass (tonnes) of Part x of food commodity X and \({mass\; shar}{e}_{{part\; x}}\) is the percentage mass share of this part.

These calculation steps were performed for each food commodity to ensure that the dataset and its mass share estimates were in full alignment with the production data from FAOSTAT and to avoid any quantitative distortion in the underlying FAO production data. Using the example of almonds, these verification steps ensured that the quantities of almond shells, skins, and kernels, summed together, amounted to the FAO production data for Almonds, in shell, in addition to providing the mass of almond hulls produced.

Finally, the estimates gathered in this dataset were cross-referenced with data points from multiple sources where available and iteratively validated. This validation is further described in the Technical Validation section.

Unavoidable and avoidable food waste typology

Every part of each food commodity was classified as avoidable, potentially avoidable, and unavoidable. This classification has been used extensively by many researchers in the field of household food waste assessment29,30,31. This was done following the approach developed by Yusoff et al. (2024) where the main edible components of food products, considered safe to eat, are assigned as avoidable. Potentially Avoidable Food Waste (PAFW) is another category often added for ambiguous food components that depend on individuals’ preferences32 and is shaped by cultural norms33. Certain foods, like chicken feet, may be commonly eaten in some regions but are not widely consumed in others. However, in most instances, local experts can determine what is generally considered avoidable food waste and potentially avoidable food waste in their country. As such, food components, whose consumption depends on individual and cultural preferences are assigned potentially avoidable food waste, similar to Nicholes et al.34 and Okayama et al.31approaches. Finally, food components that have never been and will never be consumed by humans under normal circumstances were classified as unavoidable.

Data Records

This dataset is available on Figshare23 (https://doi.org/10.6084/m9.figshare.27203688). The detailed mass compositions of food commodities are archived as a structured spreadsheet, and as such, are presented in a format that supports wide-ranging research activities from waste management to by-product valorization studies. FOWCUS is systematically arranged across several sheets within a single spreadsheet, one for each of the food group, enabling a straightforward navigation for researchers and practitioners.

  • General Information (1 sheet)

    This section provides metadata on the dataset, including the scope and a concise description of the approach employed for FOWCUS data assembly as well as an index briefly describing the content of each sheet of the dataset.

  • Commodity Composition Data (12 sheets)

    Each sheet presents a detailed tabulation of each food commodity with the mass percentages of all its components, as well as their waste qualification as avoidable, potentially avoidable, or unavoidable food waste. There is a sheet for each of the 12 food groups. All the relevant data sources and references are directly compiled alongside the mass share estimates.

  • Classification Notes (1 sheet)

    This sheet details the classification and any grouping (e.g. seafood species) made during the data compilation.

    To better illustrate the organization of the FOWCUS dataset, Table 2 provides an overview of the fields included in the Commodity Composition sheets, along with their descriptions.

    Table 2 Food Commodity Mass Composition Data Fields.
    Full size table

Technical Validation

In order to validate our dataset, we collected data from additional sources to verify that the order of magnitude of the different parts composing a food commodity was similar to those reported in other sources.

The edibility percentage of food commodities was of special importance and was meticulously reviewed, since a key purpose of this dataset is to understand the share of the food that should (avoidable) or may (potentially avoidable) be consumed by humans, and the share that may be discarded as unavoidable food waste. Since our dataset describes in detail the different components of food commodities and whether they are usually consumed by humans, we compared the sum of the edible parts estimated in FOWCUS for each food item to the edibility or yield share estimates provided by various sources.

Specifically, we collected data from national (Japan19, United Kingdom20) and regional (West Africa35) food composition tables for fruits, vegetables, root vegetables, and nuts. For seafood, we collected additional data on fish processing from FAO’s own seafood food composition table36. For some food groups, the edibility percentage was not the most apt metric to validate our estimate, as it was not the way the literature assessed these food groups. For livestock, we collected additional data sources on dressing yields for key animals (i.e., cattle37,38, pig39, chicken40, horse41, sheep42) as it is one of the key metrics (closely linked to edibility) to determine yield from livestock slaughtering. For nuts43,44 and stimulants45, additional data on kernel/nut ratio were used, which is another ratio closely linked to the edibility percentage of nuts. For sugar crops (sugar cane46 and sugar beet47,48, the sugar content, the edible portion of the plant, was the metric used to validate the mass share estimates (Table 3). For cereals49,50,51, the kernel share relative to the seed grain was the key metric analyzed.

Table 3 Key validation metrics and validation source types.
Full size table

Where available, we listed the minimum and maximum edibility values (or another associated metric) found in the literature and compared the mid-point value of the range to the values present in our dataset (Eqs. 4 and 5).

$${Valu}{e}_{{midpoint}}=\frac{\left({Valu}{e}_{\max }+{Valu}{e}_{\min }\right)}{2}$$
(4)
$${Percentag}{e}_{{difference}}=\frac{{Valu}{e}_{{FOWCUS}}-{Valu}{e}_{{midpoint}}}{{Valu}{e}_{{FOWCUS}}}\ast 100$$
(5)

This process was done iteratively, using a flag system to highlight any food commodity in the dataset with an edibility value above or below the ±20% range from the mid-point value. Whenever a food commodity in our dataset was flagged, the sources were reviewed and amended if necessary to refine the estimate in the dataset. This process was performed for all food commodities for which additional data could be collected.

Figure 2 describes the workflow followed to validate the data compiled in FOWCUS.

Fig. 2
figure 2

Workflow of Validation Process.

Full size image

The validation process and the subsequent iterations in refining and adjusting the mass composition estimates led to the vast majority of the estimate points being within ±20% of the midpoint value. A few outliers remain, mostly due to the novel boundaries set for certain food items by this dataset to offer a more comprehensive view of the mass compositions of food items than most studies. Validation results are illustrated in Fig. 3. Figure 3 visualizes the validation results by showing the deviation of each food commodity’s estimate from the midpoint of the corresponding validation range. Commodities falling within ±20% of the midpoint value were considered validated, while those outside this range are flagged for further review.

Fig. 3
figure 3

Validation results for each food category.

Full size image

We acknowledge that even with this validation process, inherent variations, and therefore uncertainties, remain due to the variety of species within a specific food item and the lack of repeated measurements for certain food commodities, which may make it difficult to fully validate the estimate of the mass share of their parts. It is the intention of this dataset to be updated as new data become available that can further validate or refine the data compiled.

Usage Notes

This dataset, accessible on FigShare, provides a comprehensive spreadsheet format suited for researchers and practitioners in food science, waste management, and bio-waste valorization. For optimal use, it is advised to familiarize oneself with its structure via the initial metadata sheet. Users can employ analysis tools like R, Python, or any spreadsheet-based software for data manipulation and are encouraged to integrate this dataset with other data sources for enriched insights.

To utilize the dataset at the national, regional, or global levels, the users may readily use FAOSTAT production datasets since our dataset is aligned with FAO food commodities typologies. To quantify animal parts, we advise the use of the Domestic Animal Diversity Information System (DAD-IS)52 to extract live-weight estimates of specific livestock raised within the national boundaries. This data should then be combined with the respective FAOSTAT estimate on the number of livestock slaughtered in this country to yield the total national live-weight production of livestock.

For food commodities belonging to the Cereals, Nuts, Legumes/Pulses, and Stimulants, this dataset includes parts (e.g., husk/hull/pod) that are not included in FAOSTAT production datasets. This is highlighted in the dataset (see Data Record section). To quantify the correct amounts of avoidable, potentially avoidable, and unavoidable food waste quantities for these commodities based on FAOSTAT production data, users must follow the quantification protocol described in the Methods section, which first adjusts the total production volumes from FAOSTAT to account for parts not included in its estimate (e.g., husk) and then apply the mass share percentage from this dataset.

While the data is robust and careful attention has been spent on validating the estimate whenever data allowed, we acknowledge inherent variability and measurement uncertainties in interpretations due to the variety of products covered in this dataset. As such, contributions to enhance dataset accuracy are welcome, adhering to our transparency and reproducibility standards to continuously improve FOWCUS and its accuracy.