Abstract
Understanding and communicating the environmental impacts of food products is key to enabling transitions to environmentally sustainable food systems [El Bilali and Allahyari, Inf. Process. Agric. 5, 456–464 (2018)]. While previous analyses compared the impacts of food commodities such as fruits, wheat, and beef [Poore and Nemecek, Science 360, 987–992 (2018)], most food products contain numerous ingredients. However, because the amount of each ingredient in a product is often known only by the manufacturer, it has been difficult to assess their environmental impacts. Here, we develop an approach to overcome this limitation. It uses prior knowledge from ingredient lists to infer the composition of each ingredient, and then pairs this with environmental databases [Poore and Nemecek Science 360, 987–992 (2018); Gephart et al., Nature 597, 360–365 (2021)] to derive estimates of a food product’s environmental impact across four indicators: greenhouse gas emissions, land use, water stress, and eutrophication potential. Using the approach on 57,000 products in the United Kingdom and Ireland shows food types have low (e.g., sugary beverages, fruits, breads), to intermediate (e.g., many desserts, pastries), to high environmental impacts (e.g., meat, fish, cheese). Incorporating NutriScore reveals more nutritious products are often more environmentally sustainable but there are exceptions to this trend, and foods consumers may view as substitutable can have markedly different impacts. Sensitivity analyses indicate the approach is robust to uncertainty in ingredient composition and in most cases sourcing. This approach provides a step toward enabling consumers, retailers, and policy makers to make informed decisions on the environmental impacts of food products.
Generated Summary
This research article presents a method for estimating the environmental impacts of over 57,000 food products. The study utilizes publicly available information, including ingredient lists and environmental databases, to derive initial environmental impact estimates across four indicators: greenhouse gas emissions, land use, water stress, and eutrophication potential. The approach involves inferring the composition of ingredients using prior knowledge from similar products, pairing this with environmental databases, and then estimating the environmental impact of each food product. The study investigates the correlation between environmental and nutritional impacts, as well as the variation in environmental and nutritional impacts of similar or potentially substitutable foods. The research aims to provide consumers, retailers, and policymakers with the information needed to make informed decisions regarding the environmental sustainability of food products. The methodology was tested and validated, and the results indicate the algorithm’s capacity to reasonably estimate the environmental impacts of food products.
Key Findings & Statistics
- The study analyzed the environmental impact of 57,185 food products.
- Only ~3% of the food products in the dataset had full quantitative composition information publicly available.
- The estimated environmental impact score was, on average, across all products tested, 1.6% lower (95% CI = 3.3% lower to 0.2% higher) than the known environmental impact.
- The estimated environmental impact score was within 10% of the known score for 65.7% of products and within 25% for 84.6% of products.
- The average difference between the known and estimated environmental impacts was 0.05 kg CO2e and 0.08 m² of land.
- The median average estimated environmental impact score is 1.6, the 75th percentile score is 4.1, and the 95th percentile score is 14.1.
- The algorithm’s average accuracy across Shelves was within 10% of the known score for 70.8% of Shelves (450 of 635) and within 25% of the known score for 87.9% of Shelves (558 of 635).
- The mean estimated environmental impact score for products in Aisles with the lowest estimated environmental impacts were often sugary drinks and other beverages.
- The estimated environmental impact score for products in Aisles with the highest estimated environmental impacts primarily contained beef and lamb products.
- Sensitivity analyses indicated that the estimated median impact of a product is on average 87% and 33% higher than the 5th and 25th observed percentile impacts, while the 75th and 95th observed percentile impacts are 72% and 300% higher than the median impact, respectively.
- For sausages, products primarily containing beef or lamb had a 240% higher impact (95% CI = 155 to 320%) than chicken and turkey sausages.
- For sausages, the nutritional impact was 20% higher (95% CI = 12.0 to 28.3%) for beef, lamb, or pork sausages compared to chicken and turkey sausages.
Other Important Findings
- The algorithm accurately estimated a product’s environmental impact (Fig. 2).
- Foods with a low environmental impact for one indicator tend to have low impacts for other indicators (P < 0.05 for all Spearman’s correlations).
- The environmental impact estimates remain available for situations that are better suited to the disaggregated estimates (e.g., companies with targets focusing on a single environmental outcome, such as net zero greenhouse gas emissions policies).
- More nutritious foods tend to be more environmentally sustainable, although there are exceptions to this trend.
- In general, many of the highest impact products were dried beef products.
- The study showed that replacing meat, dairy, and eggs with plant-based alternatives could have significant environmental and health benefits.
- The study showed that replacing meat, dairy, and eggs with plant-based alternatives could have significant environmental and health benefits.
- For pesto sauces, lasagna, and cookies, the environmental impacts of direct substitutes were found to vary.
- The accuracy of the algorithm increased when more composition information about a product was known.
Limitations Noted in the Document
- The study is limited by the lack of detailed environmental impact information, especially the exact amounts and supply chains of ingredients, which are often trade secrets.
- The algorithm’s accuracy is affected by the reliance on ingredient lists, where the composition of some ingredients is not fully known.
- The study’s environmental database contains limitations and biases, including a bias toward commodities and production systems in high-income regions.
- The Nutri-Score, used for assessing nutrition quality, has limitations; for instance, it doesn’t account for the effects of processing or home preparation.
- The findings are based on products available in the United Kingdom and Ireland, and the generalizability to other regions might be limited.
- The uncertainty in sourcing may not have a large influence on the estimated environmental impact of most products, but it does indicate that more transparent ingredient sourcing is needed to derive more accurate environmental impact estimates.
Conclusion
The study provides a significant step towards understanding and communicating the environmental impacts of food products. By developing and testing an algorithm to estimate these impacts across a wide range of products, the authors have created a framework that can inform decision-making by consumers, retailers, and policymakers. The core finding is that the algorithm can reasonably estimate the environmental impacts of food products, even when complete composition data is unavailable. The algorithm’s capacity to estimate these impacts, even with incomplete data, is a substantial advancement, setting a baseline for future research. The findings highlight that more nutritious foods often have lower environmental impacts, supporting the potential for win-win scenarios in food choices. The research also reveals substantial variability in the environmental impacts of similar food products, pointing to opportunities for targeted interventions, such as ingredient sourcing. The work provides a tool for assessing the environmental impact of food products and identifies areas for improvement, such as the need for more transparent ingredient sourcing and the importance of considering the context in which foods are consumed. This research underscores the potential for food systems transformation by informing choices that can lead to more sustainable and healthy diets. As stated in the article, “By estimating the environmental impacts of food products in a standardized way, our approach provides a step to enable informed decision making by end users such as consumers and policy makers.”
IFFS Team Summary
- Meta-analysis estimating the environmental impact of 57,000 food products in the U.K. and Ireland
- Builds on the work from Poore & Nemecek’s (2018) and the foodDB—a Big Data research platform at the University of Oxford that collects and processes data daily on all food and drink products available in 12 online supermarkets in the U.K. and Ireland, and a comprehensive review of 570 studies of the environmental impact of food production, which includes data from 38,000 farms in 119 countries. Also pairs with data from Gephart et al., Nature 597, 360–365 (2021)
- Study was specific to UK and Ireland because government regulations there require ingredients to be listed on each product in decreasing order of their abundance and for the percent composition of characterizing ingredients (e.g., the beef in beef lasagna)
- Looked at greenhouse gas emissions, land use, water stress, and eutrophication potential and then combined these four scores into a single estimated composite environmental impact score per 100 g of product.
- Finds many meat alternatives had a fifth to less than a tenth of the environmental impact of meat-based equivalents.
- This is apparently the first time a transparent and reproducible method has been developed to assess the environmental impacts of multi-ingredient products.
- Limitations:
- Each ingredient and their supply chain in each food product are often considered a trade secret and lack considerable transparency, especially on country on origin of (ie. fish)
- Some nutrition method concerns about NutriScore were communicated
- Some food combinations are odd i.e. ”roasted potatoes, chips, onion rings, and rice” are combined, and considered healthy as a group, despite some items being a fried junk food.
- Some prepared foods could vary greatly, i.e. “pizza”
- Packaging is not included as an impact, which we understand is not within scope of the paper, but bears mentioning
- Legumes (beans, lentils, soy beans) are not included in charts despite being key plant foods