Uncharted links from mining to fed animal systems: a scoping review

Animal source foods are a growing component of human diets and the primary source of environmental pressure from global food production^1,2. For most farmed animal production (most pig, chicken, fish and crustaceans, and some ruminants), the use of human-made feeds is a necessity to support animal growth and health. Yet producing enough feed for farmed animals globally bears a huge environmental cost. Approximately 33–40% of agricultural land^2,3,4 and >9% of all aquatic animal production is destined for animal feed⁵. Moreover, the majority of environmental impacts embedded in the life cycle of farmed animals often comes from feed production^1,6,7. There has been considerable scrutiny of the use of numerous animal feed ingredients including soy products, palm oil, and wild-sourced fishmeal largely due to the ecological consequences of their production and a conflicting use of resources that could otherwise be diverted to direct human consumption. Most attention to the sustainability of feed resources has been directed towards the once living components of animal feed (i.e., agricultural and fisheries products). Yet very little focus has been devoted to the abiotic resources that are harvested for feed production.

Minerals are a critical component of animal feeds that are seldom discussed in sustainability assessments. Animals require a specific intake of minerals which differs between species, breeds, developmental stages and reproductive status⁸. To ensure they have the required amount, many fed animals are given mineral supplements as part of their diet to prevent nutritional deficiencies and to maximise productivity and product quality. These mineral supplements can be provided in many chemical forms and produced from numerous sources. For example, phosphorus supplements can be in dicalcium phosphate, monocalcium phosphate or other chemical forms, and can be sourced from phosphorus rock (via mining) or from bones that are byproducts of animal production or other industries⁹. Mineral supplements often make up a small percentage of the total feed; <5% for salmon¹⁰ and <3% for cattle¹¹. Though not all fed-animal systems globally use supplements, billions of production animals are being provided with these supplements, which even at low inclusion rates in feeds equates to millions of tonnes of minerals a year; using a feed conversion ratio of 1.35 and an mineral premix inclusion rate of 4.1%, the global salmon industry alone used ~111,785 t of mineral premix in 2024^10,12. Given this volume and the various ways that mineral supplements can be produced, understanding the environmental implications of their use is an important component of sustainable animal food production. One method to assess the environmental impacts of a product is to use Life Cycle Assessment (LCA), which is a tool used to assess the environmental impact of a product using a specific methodology, allowing different products to be compared.

To assess how thoroughly environmental impacts from mineral supplements are accounted for in sustainability assessments of animal feed, we conducted a scoping review of LCAs for fed farmed animals. We systematically identified 104 relevant peer-reviewed papers from Google Scholar, Web of Science, and Scopus that detailed LCAs of feed for either cattle, pigs, farmed fish and crustaceans, and chickens. The aims of this scoping review were to 1) identify whether mineral supplements were described in modelled feed and/or included in life cycle inventories, and 2) describe how their life cycle impacts are accounted for during assessment (e.g. via proxies). We then use examples to illustrate why exclusion or misrepresentation of mineral supplements in LCAs can cause the environmental pressures associated with animal feeds to be poorly estimated.

Treatment of mineral supplements in LCAs of farmed animal feeds

We identified a total of 104 peer-reviewed papers that met our search criteria, which contained 109 unique LCAs. This comprised 33 LCAs for cattle, 28 for chicken and 25 for fish/crustaceans and 23 for pigs (Fig. 1a). Pooled, most papers excluded mineral supplements from their LCAs (64%; n = 70). Across different taxa, the greatest proportion of studies where mineral supplements were omitted was for fish and crustaceans (84%), followed by cattle (73%), then chicken (50%) and pigs (48%). Of the 39 LCAs (36.5%) that did include mineral supplements, 25 of these LCAs (66%), did not model the mineral supplements detailed in feed descriptions but instead used a proxy (such as calcium carbonate) to represent the impacts of mineral use (Fig. 1b). Thus, only 14 LCAs (13%) modelled the impacts of mineral use that was described in the associated feed profile of the study.

Where proxies were used in place of mineral supplements, similar proxies were used across different LCAs (Fig. 2). Common proxies used were calcium carbonate/limestone (n = 15), elemental phosphorus (n = 3) and sodium chloride (n = 3; Fig. 2a, Supplementary Table S1). Most commonly, calcium carbonate or limestone was used to represent the impacts of the entire mineral premix (Fig. 2b), with some LCAs clarifying that the premix is assumed to be primarily composed of limestone¹³. Further, patterns of proxy use did not vary substantially across taxa, with calcium carbonate being the most commonly used for all farmed animal groups (Fig. 2b).

Though this is a scoping review with only 104 papers making the inclusion criteria, our results suggest that low resolution in the data and assessment afforded to the impacts from feed mineral supplementation appears to be common practice among LCAs across animal production sectors. Additionally, all the manuscripts that made the inclusion criteria were from the peer-reviewed academic literature. We recognise that these results therefore may not be representative of how LCAs are conducted in animal production industries and that confidential LCAs may routinely include the mineral components of feed. We therefore welcome more information from feed companies regarding their inclusion (or not) of the mineral component of feed in their LCAs.

Aquatic and terrestrial farmed animals have diverse and specific mineral requirements in their feeds to support growth and health (Table 1). Furthermore, in some fed animal systems, it is recommended that many minerals are supplied in excess of requirements. For example, the National Research Council recommendations for dairy cattle nutrition suggest feeding animals 120% of the required amount of most minerals¹⁴.

Additionally, many species require similar elemental supplements but in vastly different quantities. For example, pigs require only 4600 mg/kg of calcium¹⁵, but due to their egg-laying abilities chickens may require almost four times that amount of calcium¹⁶. While iron is a critical supplement for fish¹⁷ it is not necessary for cattle as they receive more than they need from the plants in their feed¹⁸. Farmed species’ mineral requirements can also change throughout the production cycle. For example, freshwater fish will absorb minerals from the environment via the gills, saltwater fish will absorb minerals through the gills and the digestive tract (via drinking), and the mineral requirements of diadromous fish (e.g. Atlantic salmon) will change with their transfer from freshwater to seawater during the production cycle¹⁹. Thus, the diversity of mineral supplements used is therefore poorly represented by the relatively uniform use of proxies in the LCAs of different farmed animals.

The importance of improving the resolution of mineral supplement sustainability assessments

Poor representation of mineral supplements in animal feed LCAs undermines reliable environmental footprint assessments and understanding of their contribution to the impacts of global food production. Most minerals used in supplements are sourced from ore extraction, which yields considerable environmental pressures including land-use, greenhouse gas emissions, waste products, soil contamination, and water pollution²⁰. These contrast with agricultural or fishing activities which typically have a renewable resource base where producers have incentive to manage resources sustainably for long-term productivity. The magnitude of impacts per unit produced may also exceed those of other feed ingredients. For example, greenhouse gas estimates (kg CO₂ eq/t) range from 774 for mono-dicalcium phosphate to 8812 for elemental phosphorus^21,22, and from 2000 to 6120 for zinc mining and processing^23,24. By contrast, greenhouse gas estimates range from 423 to 735 kg CO₂ eq/t for wheat production^25,26.

Despite low inclusion rates of mineral supplements, their omission from LCAs can result in substantial impacts due to the large amount of feed used for animal production globally. For example, an estimated 83,767,500 t of feed were used in United States feedlots in 2025²⁷. Assuming a phosphorus inclusion of 0.48% (Table 1) and 8812 kg CO₂ eq/t for phosphorus mining and processing²¹, almost 72 million t CO₂ eq y⁻¹ would have been emitted from phosphorus supplementation of feedlot cattle. Fish and crustaceans had a particularly high proportion (95%) of studies where the mineral component was excluded from the LCA. Approximately 1,976,706 t of feed were used for Norwegian salmon in 2020¹⁰ and assuming a phosphorus inclusion of 0.94%¹⁰, an extra 163,700 t of CO₂ eq would have been emitted due to phosphorus supplementation of salmon feed. Omission of the mineral components of aquaculture feeds is particularly important as the contribution of mineral supplements in these feeds is increasing due to the transition away from fishmeal and oil and towards plant-based diets, which require a higher level of supplementation than established recommendations²⁸.

Using proxies as substitutes for mineral supplements compounds this problem by misrepresenting the true environmental impacts and failing to capture unique impact pathways. Mineral processing introduces entirely novel environmental pressures that have no equivalent in agricultural systems. While the use of proxies can capture some impact pathways, the extraction processes, sourcing locations, and operational contexts can vary considerably between actual minerals and proxies. For example, sulfuric acid production during zinc processing creates impacts that would be completely absent from feed LCAs that exclude or inadequately represent mineral supplements²⁹. Further, mining locations occur across a wide range of ecological contexts, including regions of relative high species richness (Fig. 3). Although spatial overlap between mine sites and species-rich areas does not by itself demonstrate biodiversity loss, it illustrates that mineral sourcing can create exposure to biodiversity-related risks that are not captured when premix is excluded or only represented by common proxies. These geographic and operational disconnects mean that LCAs of feed products may omit potentially important environmental pressures when mineral supplements are inadequately represented.

Toward more comprehensive feed assessments

While the exclusion of mineral supplements from LCAs is problematic from an environmental perspective, it is very understandable given several challenges. Their environmental impacts are very difficult to calculate, and the exact concentrations of each mineral in the feed, and the chemical species used (e.g. selenium as sodium selenite or selenomethionine) are often confidential¹⁰. There are also substantial data gaps around where mining is occurring and how ore is transported and processed, and thus, it is nearly impossible to know the process by which minerals for a specific feed were extracted or their geographical origin. These challenges are further complicated when minerals can be sourced from different systems. For example, phosphorus sources can be as diverse as animal bones or rock phosphorus, and the impacts embedded in both pathways will still depend on their source location and methods used to produce them, which can be allocated in different ways in LCA³⁰.

Despite the challenges outlined, the lack of information on specific sourcing locations should not completely preclude inclusion of mineral premix components in feed LCA. The FAO Livestock Environmental Assessment and Performance (LEAP) Partnership published a guide in 2020 advising that vitamins and mineral pre-mixes be included in livestock feed LCAs, and advocated for the use of proxies only when no other sources are available³¹. LCA information is available for the extraction, processing and transport of many minerals. For example, zinc is a nutritional requirement of all farmed species listed in Table 1, but it is not represented by any commonly used proxy (Fig. 2) and even animal feed LCAs, which include zinc in their supplement inventory do not quantify its impacts³². However, LCA information is available for zinc mining and extraction via different methods²³, and recent efforts mapping the global mining footprint provide opportunities to quantify regional land-use, water-use, and biodiversity impacts³³. At the very least, life cycle inventory databases (e.g. EcoInvent) could include proxies that are more similar to the actual mineral inputs to enable more accurate LCAs. Ideally, to enable representative sustainability assessments for animal feeds, industry and researchers need to have better access to more information about the mineral supplement ingredients and their provenance. A comprehensive, regularly updated database of mineral supplements and their environmental impacts that could be used for LCAs would help to resolve this. If this is not possible, mineral supplement suppliers need to provide data on the environmental pressures associated with their mineral supplements, which could be achieved without breaching confidentiality associated with exact formulation of the supplement. Better collaboration between feed manufacturers, mineral suppliers and researchers would aid resolution of feed LCAs.

Despite the prevalence of mineral supplements in farmed animal feeds, their impacts are poorly represented in Life Cycle Assessment- mostly (64% of the time) they were excluded from assessments. When they were included, proxies were often used in the LCA. LCAs for farmed fish and crustaceans were especially poorly resolved; 95% of studies omitted the mineral supplement component of the feed entirely from the LCA. Although understandable due to difficulty in obtaining data on the source of minerals for supplements, use of proxies such as calcium carbonate will misestimate the impacts of mineral supplements. Therefore, we strongly argue that mineral supplements need to be specifically and directly included in LCAs for the nature and location of the impacts of farmed animal feeds to be accurately assessed. We argue that mineral supplement and/or animal feed companies have a responsibility to provide this information to those wishing to conduct LCAs. These companies can give this information for pre-mixed mineral supplements, and thus can do so without compromising their intellectual property by revealing their specific formulations.

Scope and databases

We conducted a scoping literature review using a PRISMA- ScR framework³⁴ (Supplementary Table S2) to quantify the extent to which mineral supplements/premix were resolved in LCAs for the most common farmed animals (chickens, pigs, cattle, carp, tilapia, salmon and shrimp). We used three databases for our search; Scopus, Web of Science (WOS) and Google Scholar. These represent the three most widely used literature repositories for environmental sciences and align with published systematic frameworks (e.g., Pickering and Byrne³⁵). Search terms for each database can be found in Supplementary Table S3.

Search process

For Scopus, we entered the search terms with the TITLE-ABS-KEY, ordered the results by relevance and refined it to include English language and article format only. For WOS, we searched using ‘All Fields’, ordered the search results by relevance and refined the results to include English language and article format only. For Google Scholar, we entered the search terms and retained only peer-reviewed journal articles sorted by relevance. We excluded all other article types (e.g. book chapters, theses, pre-prints). We collected the first 100 results from each of the three databases. We conducted all searches on 21/02/2024.

Exclusion and data extraction process

Of the 300 papers resulting from the searches, we removed 64 duplicates.

We split the remaining 236 papers between three reviewers, and the title and abstract were reviewed for each paper, with 10% of the papers reviewed by all three reviewers to check consistency of review. We excluded papers from further review based on the workflow in Fig. 4. Rationale behind the exclusion criteria can be found in Supplementary Table S4.

Before excluding papers where it was not clear as to whether they did or did not include minerals in the inventory, every effort was made to contact the lead author/s via email and ask them for more details. We contacted authors of 25 papers and received a response rate of 36%. Once two emails were sent and no response was received, we assumed that no response was forthcoming and therefore these papers were rejected due to ‘Not possible to understand if minerals were in the inventory’ (Fig. 4).

We shared the 104 papers between the three reviewers and extracted the following data (Supplementary Table S1):

1. 1.

   Details of the paper (title, DOI, author, year of publication)

2. 2.

   Animal species and animal type (e.g. salmon, fish/crustacean) in each LCA in the paper

3. 3.

   Farming country

4. 4.

   Whether or not mineral supplements were included in the LCA inventory

5. 5.

   Whether a proxy for the minerals was used in the LCA and if so, what the proxy was

6. 6.

   Other details about the paper regarding how the minerals were included in the LCA

Two of the authors extracted the data from ~10% of the papers to cross-check data extraction. No discrepancies between the two authors were found. As some papers included LCAs from multiple species, a total of 109 LCAs were used in the data extraction (Supplementary Table S1).

Data

The raw data used in this study and the associated metadata can be found in Supplementary Table S1 in Supplementary Information. All analyses were conducted and Figs. 1–3 were made using RStudio (Version 2026.05.0). Figure 4 was made in Microsoft Powerpoint. All data and figures are available to view on the project’s GitHub repository³⁶.