Food system by-products upcycled in livestock and aquaculture feeds can increase global food supply

The current structure of the global food system results in suboptimal food availability, as a large proportion of the resources used in livestock and aquaculture feeds could be consumed by humans. Up to 40% of all arable land and more than 30% of cereal crop production is used for animal feeds1,2, and approximately 23% of all captured fish are destined for non-food uses, mainly for fish and livestock feeds3. This food–feed competition reduces the efficiency of the existing food system, as environmental and resource costs are higher when arable land is used for animal feed production instead of directly contributing to human consumption.

Increasing the feed use of food system by-products—that is, the secondary products created alongside the primary, human-consumable products—has been proposed as a solution to increase resource use efficiency, to reduce food–feed competition and to increase food system circularity. In addition, using food system by-products as feeds can reduce the environmental pressure on arable land and freshwater ecosystems, as well as reduce greenhouse gas emissions and fertilizer application. Increasing the use of by-products and crop residues as feed can also be cost-effective since many of them are widely available, low-cost materials. However, some non-food-competing feedstuffs are less suitable for feed use; for example, crop residues are fibrous and of low digestibility and poor protein quality, and others, such as some crop processing by-products, are protein dense but low on energy. Yet, some non-food-competing feedstuffs can be improved through processing or additives. Despite the challenges, part of the food-competing feed use could be replaced with non-food-competing feedstuffs without negatively impacting productivity (Supplementary Tables 7 and 8).

In this study, we assess the potential of improving circularity in the global food system by increasing the use of food system by-products and residues in animal feeds. This approach provides a much-needed systemic view of the highly interlinked global food system and advances the research field on three main fronts. First, global datasets including both feed material flows and the availability of food system by-products and residues at this level of detail do not exist. While different models and reports provide data on livestock or aquaculture feed use, these data are not harmonized throughout the global food system. Furthermore, while some studies have estimated feed use in both agriculture and aquaculture systems, they do not account for country-level differences in feed use or have only regional focus. Here we combined and harmonized data from various sources including crop, livestock and aquaculture production, as well as wild fisheries, and quantified the dynamics of global feed flows in remarkable detail (Fig. 1 and Methods).

Four main phases were considered in the analysis: (1) quantification of global food system material flows—we quantified global food system material flows including the potential production of by-products and the feed use in both terrestrial and aquatic food systems; (2) availability of by-products—we analysed the availability of the by-products by subtracting their current feed use from potential production, assuming that all non-used by-products would be available for feed; (3) nutritional limitations—using existing literature, we considered to what extent by-products can replace food-competing feedstuff in animal feeds, accounting for impacts on productivity; and (4) replacement potential—we estimated the amount of food-competing feedstuff that could be replaced with non-food-competing feedstuff. See Methods for the quantitative interpretation of the flows as well as the data used for each phase.

Second, material flows related to food system by-products have been estimated and reported only sporadically; thus, a comprehensive understanding of those flows is lacking. We overcame this by analysing the availability of different by-products and residues followed by quantifying their current feed use and the potential availability to further increase their feed use. Third, existing studies have assessed the feed use potential of individual by-products within specific production systems (Supplementary Tables 7 and 8) and have analysed scenarios of livestock production that could be sustained by restricting their feed use to non-food-competing feedstuffs. On the basis of our quantification of global food system material flows, we extend this knowledge by assessing the replacement potential of food-competing feedstuff with by-products and residues while simultaneously considering their regional availability and nutritional constraints in both the aquaculture and livestock production sectors (Methods). The nutritional constraints that dictate the replacement potential are based on an extensive literature review of feed experiment studies (Supplementary Tables 7 and 8). Combining the three advances, we are able to show that increased utilization of food system by-products and residues in animal feeds could theoretically lead to a considerable upsurge in the global food supply.

Food-competing feedstuff use

We first combined data from various sources, using production data for primary products for 2016–2018 (refs. 2,30) and feed composition data for livestock and aquaculture for 2010 (refs. 21,31) to analyse the current feed flows (Fig. 1 and Methods). We focused on the feed use of food-competing feedstuff including cereals, oilseed oils, pulses and whole fish used in fishmeal and fish oil production (Supplementary Table 3; see Methods for assumptions).

Approximately 15% (940 million tons in dry matter) of the total feedstuffs (6,100 million tons) used in livestock and aquaculture production consisted of food-competing feedstuff that could be directly used as human food (Fig. 2). This is in line with existing estimates (Supplementary Information). However, animal production systems differ substantially in their food-competing feedstuff use. At the global level, up to 49% of feed use in aquaculture (total feed use, 67 million tons), 68% in poultry (total 421 million tons) and 38% in pork meat production (total 1,200 million tons) consisted of food-competing feedstuff, while for cattle meat (total 5,200 million tons) and dairy (total 1,920 million tons), the share was only 3–4% (in quantities of feed in dry matter). The low share of food-competing feedstuff in cattle feed is mainly due to the large share of global cattle production being extensive grazing systems, which have high feed conversion ratios (kg feed per kg output), consuming high amounts of feed consisting mainly of roughages such as grass and hay. Diets in industrial feedlot cattle systems often include a higher share of food-competing feedstuff. For example, in some North American and European industrial beef cattle systems, the diets in the finishing phase can consist of more than 70% (ref. 21) food-competing feedstuff. However, these systems are highly optimized, having lower feed conversion ratios and consequently lower total feed consumption. Furthermore, we found that regional variation in food-competing feedstuff use is notable, the minimum being less than 4% in Africa and values ranging through 7% in Oceania, 8% in Latin America, 15% in Europe and 16% in Asia to a maximum of almost 20% in North America. These differences reflect regional variations both in animal species farmed (that is, the relative proportions of different animals) and in production mode (that is, intensive versus extensive systems).

a, All feed flows. b, Flows for only food-competing feedstuff. The percentages refer to the shares of the feed use categories (on the left) and the shares of the feed use in specific animal production groups (on the right) of the total global feed use. The feedstuffs included in each category are described in Supplementary Tables 3 and 4. The data sources used are given in Methods. Material flows in terms of protein content are shown in Extended Data Fig. 1.

At the global level, however, most of the feed use consists of materials not suitable for human consumption, mainly roughages such as grass foraged by cattle (Fig. 2a). Cereals form by far the largest group of food-competing feedstuff use, in which maize is the most important feed cereal, followed by wheat, rice and barley (Figs. 2b and 3a,d). Notably, fish products (fishmeal and fish oil) are important sources of protein (Fig. 3b and Extended Data Fig. 1) and fat (Fig. 3c) when examining the global feed protein and fat flows. They can also be important for the supply of healthy fatty acids (EPA and DHA). Similarly, the importance of oilseed oils increases when looking at the fat content (Fig. 3c).

a–d, The values represent averages over 2016–2018 for energy content (kcal) (a), protein content (b), fat content (c) and quantities (dry matter weight) (d). The error bars represent the total uncertainty range for different feedstuffs (Methods). The use of the feed is colour-coded to each feed group.

Availability of food system by-products

In this study, we identified four main categories of food system by-products: (1) crop residues (that is, the plant material remaining after harvesting, including parts such as straw, leaves, stalks, roots and stover); (2) crop processing by-products, including cereal bran and distiller’s grains (including distiller’s grains from biofuel and brewer’s grains from barley beer industries), sugar by-products (including sugarcane and sugar beet molasses and sugar beet pulp), oilseed meals (including rapeseed, soybean, sunflower seed, palm kernel, sesame seed, cottonseed, groundnut and other oilseed meals) and citrus pulp; (3) livestock by-products from non-ruminant origins (that is, processed animal protein from pig and poultry production, here blood meal, hydrolysed feather meal, meat and bone meal, poultry by-product meal, and poultry oil); and (4) fisheries and aquaculture processing by-products, hereafter referred to as fish by-products, processed into fishmeal and fish oil. The total availability of food system by-products was estimated by multiplying the production quantities of primary products by conversion factors for different by-products and residues, or applying global statistics2 or other literature (Methods). The feed use of these by-products and residues was then subtracted from their total availability to estimate the potential availability of materials not already used as feed.

Only relatively small shares of crop residues and livestock by-products are currently used as feed, while nearly all oilseed meals (that is, oilseed meals and cakes, here grouped together under meals) and more than half of the total potential of crop processing by-products, such as sugar beet pulp and cereal bran, are already used as feeds (Fig. 4). It should be noted that while only a small share of crop residues are used for feed, these resources also have other uses such as biofuels or bedding material for livestock (Discussion). Here the use of crop residues in maintaining soil quality was taken into account in sustainable harvest ratios (Methods), but other uses were excluded, assuming that all by-products not already used for feed would be available to replace the human-edible feeds in animal diets. Despite their multiple uses, the highest theoretical potential in utilizing by-products as feed still lies in crop residues, as their theoretical availability exceeds the availability of other by-products (Fig. 5).

The increased feed use of by-products when applying the replacement potential analysed in this study. Uncertainty range for current feed use (left), additional feed use when replacement is applied (middle) and other uses/theoretical availability (right) is shown. Negative values refer to the feed use exceeding potential availability. For the individual products included in the combined categories, see Supplementary Table 4.