Main

International commodity crop markets face evolving expectations of sustainability, reflecting increasing awareness of environmental issues, in particular, climate change1,2. The carbon footprints of major commodity field crops such as non-durum wheat (herein also referred to, interchangeably, as wheat), rapeseed and peas are of increasing interest to diverse stakeholders—wheat as a major staple crop3, rapeseed as an important first-generation biofuel feedstock (https://www.prnewswire.com/news-releases/world-biofuels-report-190062341.html) and peas as a leading source of plant-based protein (https://www.prnewswire.com/news-releases/increased-inclination-toward-plant-based-diet-generating-promising-sales-opportunities-for-dried-peas-market-players-tmr-301167046.html). Canada is a key producer and exporter of these crops, with production concentrated in the prairie provinces of Alberta, Manitoba and particularly Saskatchewan4.

The carbon footprints of these crops exported from Canada were compared with those from other major producers (France, Germany, Australia and the United States), based on the International Standards Organization (ISO) 14067 (ref. 5) standard. The break-even point for transportation of Canadian field crops was calculated as the distance that they can be shipped before emissions from production plus transportation equal those of crops produced in the destination countries. These results enable meaningful comparisons between carbon footprints of commodity field crops in international markets. They can be used to identify priority targets within their supply chains for future mitigation efforts, inform marketing and procurement policy decisions and contribute necessary nuance to ‘food miles’ discourse.

Results

Generally, field-level nitrous oxide (N2O) emissions and soil organic carbon (SOC) changes were the largest determinants of observed differences in carbon footprints. In most cases, even when SOC change was excluded, Canadian crops had smaller footprints compared to the same crops produced in other countries. Only Australian rapeseed was lower because of low field-level N2O emissions due to aridity and lack of irrigation, despite higher tillage rates (resulting in higher impacts from field-level activities) (Fig. 1). With SOC change included, Canadian crops always had the lowest emissions because their soils currently sequester carbon due to low- or no-till practices, reduced summer fallow and carbon inputs from manure and crop residues, despite SOC losses from conversion of forest and grassland to cropland6. All other countries considered had net SOC losses7,8,9,10. Australia had the lowest rates of SOC loss due to a decreasing trend in the amount of forest land converted to cropland and an increase in adoption of conservation and no-till practices8. The next lowest country was the United States, which had net sequestration from cropland remaining cropland due to conservation tillage and C inputs (although with a declining trend), but with larger SOC losses, mainly from conversion of forest and grassland10. France had higher SOC losses due to land conversion from grassland, although increased adoption of conservation tillage practices has slowed this trend over recent years7. Germany had the highest SOC losses, also driven primarily by conversion of grasslands to agricultural land, although similar increases in adoption of conservation tillage practices has decreased losses recently9.

Fig. 1: Carbon footprint of rapeseed, non-durum wheat and peas produced in Canada, Australia, France, Germany and the United States.
figure 1

Mean net values (including soil carbon) are indicated by the red dot with associated error bars representing standard deviation (selected for visibility of error bars), as well as by the numerical values for each crop–country combination. The contributions to the carbon footprint (kilograms CO2 equivalent per kilogram crop) are indicated for field-level N2O emissions, fertility management (including fertilizer and manure application and biological nitrogen fixation), fuel use for field activities, SOC change and other inputs and activities. All differences between countries are statistically significant (P < 0.05), using a single-factor analysis of variance. Error bars and significance tests were calculated using Monte Carlo simulation (1,000 iterations). CA, Canada; AU, Australia; FR, France; DE, Germany; US, United States.

Source data

Full size image

For rapeseed (Fig. 1), production of fertilizers (35%) and field-level N2O emissions (47%) were the main contributors to the carbon footprint of Canadian crop production. Carbon dioxide from combustion of diesel for field activities contributed 7% of the total footprint of Canadian rapeseed. Australian and Canadian rapeseed crops had similar footprints when SOC changes were excluded (0.48 and 0.53 kgCO2e kg−1 (kg CO2 equivalent per kg crop)), while Australian emissions were higher when they were included. Although Australian soils were found to have net emissions of CO2 compared to net sequestration in Canadian soils, these losses only increased the carbon footprint of Australian rapeseed by 10% when they were included. The highest contributor to the carbon footprint for Australian rapeseed was fuel use for field activities (30%), followed by fertilizer production (26%). Field-level N2O emissions were low compared to all other regions due to differences in climate, contributing only 11% of the total carbon footprint. German and French rapeseed had the largest footprints. Field-level N2O emissions were the highest contributors to the carbon footprint of German and French rapeseed (55–58%). When SOC was included, it added an additional 24–39% to the carbon footprint. Production of fertilizers was also an important contributor (25–27%).

For wheat (Fig. 1), fertilizer inputs (38%) and associated field-level N2O emissions (37%) were the main contributors to emissions for Canadian production. Field activities contributed 8% of the Canadian wheat carbon footprint. When included, SOC change reduced the overall carbon footprint by 62%. All other regions had much higher emissions than Canada due to the higher life cycle emissions of production and net carbon emissions from soils. Among the other regions, Australia (0.52 kgCO2e kg−1) had the lowest emissions, followed by France (0.59 kgCO2e kg−1), the United States (0.60 kgCO2e kg−1) and Germany (0.65 kgCO2e kg−1). For German and French wheat, field-level N2O emissions made the highest contribution to the carbon footprint (47–48%). This was followed by SOC change (contributing an additional 20–36% when included) and production of fertilizer inputs (19–20%). For the United States, field-level N2O emissions contributed 27% to the carbon footprint. Fertilizer and manure production (collectively, fertility) contributed 24% of the carbon footprint, and fuel use for field activities contributed 18%. SOC change added an additional 10% to the carbon footprint of US wheat. For Australia, fuel use for field activities contributed 19% to the carbon footprint, fertilizer and manure production contributed 19%, and field-level N2O emissions 17%. When included, SOC loss contributed an additional 6% to the carbon footprint. The rest of the emissions for Australian wheat (45%) came from other inputs and activities, including seed, CO2 emissions from lime and urea, and post-harvest energy use.

The highest contributor to the carbon footprint of Canadian pea production (Fig. 1) was field-level N2O emissions (62%). Field activities contributed 19% of the emissions of Canadian peas. Overall, fertility management contributed 10%, including emissions from fertilizer and manure production, as well as a credit for biological N fixation. When included, SOC change decreased the carbon footprint of Canadian peas by 87%. As all other countries already had higher production emissions and much higher emissions from soil carbon changes, their combined footprint compared to Canada (0.03 kgCO2e kg−1) ranged from 21 times higher in France (0.64 kgCO2e kg−1) to 34.5 times higher in Germany (1.05 kgCO2e kg−1). The highest contributor to the carbon footprint of French peas was field-level N2O emissions (48%) followed by field activities (18%). Overall, fertility management contributed 12%, and SOC change increased the carbon footprint by an additional 51%. Field-level N2O emissions contributed 56% of the carbon footprint of German peas. Fertility management contributed 15% of the carbon footprint, field activities 13% and SOC change an additional 64%. Field-level N2O emissions were the highest contributors to the carbon footprint of US peas (43%). Fertility management contributed 17%, field activities 16%, and SOC change an additional 17%.

See Supplementary Discussions 1–11, Supplementary Tables 1–29 and Supplementary Figs. 1–6 for detailed life cycle inventory and life cycle impact assessment results.

When SOC was included, crops produced in Canada then shipped overseas to Australia, France and Germany still had lower carbon footprints than crops produced in each destination country. In the most extreme cases, differences in production emissions were sufficient to offset shipping the crops from Canada to Europe an additional 17 times before breaking even (for German peas and canola), equivalent to circumnavigating the globe more than three times (Fig. 2 and Extended Data Table 1). With SOC included, transportation accounted for 21–88% of the footprint of Canadian crops shipped overseas. Canadian peas had the lowest carbon footprint of all the crop types, and therefore transportation made the highest proportional contribution, ranging from 67% of the emissions for peas at market in Germany to 88% at market in Australia. By contrast, rapeseed had the highest production emissions, and thus transportation contributed only 21–50% of the total carbon footprint at market in Europe/Australia.

Fig. 2: Break-even distances for wheat, peas and rapeseed produced in Canada and transported to Australia, France and Germany, compared to the travel distance to each destination country.
figure 2

The break-even distances (blue, gray and orange bars) (mean ± standard error) represent the amount of ocean transport possible before the emissions from transportation break even with the difference in production plus land transportation emissions between regions. A negative break-even distance means that production emissions in the destination country are lower than in Canada. a,b, Results are presented with SOC change included in (a) and excluded from (b) the production emissions. Error bars (standard error) and significance tests were calculated using Monte Carlo simulation (1,000 iterations).

Source data

Full size image

With SOC excluded, transportation accounted for 20–48%, 17–42% and 10–29% of the emissions of Canadian peas, wheat and rapeseed, respectively, at market in Europe/Australia. For Canadian crops, the exclusion of SOC meant higher overall production emissions, as Canadian soils had net carbon sequestration. Therefore, transportation overseas made a relatively smaller contribution. However, when calculating the break-even distances without SOC change, production-related emission differences between countries were lower, and hence transportation of Canadian crops was proportionately more important (Fig. 2). Without SOC change, Canadian crops shipped to Europe still had lower emissions, but the opposite was true for Australia. For German rapeseed and peas, the offset was still sufficient to ship the crops produced in Canada to Germany an additional 4–5 times before breaking even, or ~1 circumnavigation of the globe (Extended Data Table 1).

Discussion

This study revealed key drivers of, and differences in, greenhouse gas (GHG) emissions for field crop production and transport to market between countries—in particular, with respect to N2O and SOC. Both are priority foci for emissions abatement efforts. Carbon sequestration in cropland was an important determinant of the relatively low carbon footprint of Canadian crops compared to other countries. This is, in part, attributable to the widespread adoption of low- and no-till practices11, particularly in the Canadian Prairies where up to 89% of agricultural land is managed accordingly12, along with conducive soil and climatic conditions, and reductions in summer fallow. Adoption of these practices in Canada was historically motivated by their positive impact on soil moisture retention/management in arid regions13. It has since been recognized that reduced tillage is critical for restoring SOC14, as well as reducing energy use from tillage, making it a key climate change mitigation strategy for all countries included in this analysis15,16,17,18. However, achieving net SOC sequestration through changes in tillage may not be universally successful, evidenced by the lower levels of SOC sequestration observed in Australia despite widespread use of conservation and no-till practices8. Moreover, following a change in tillage regime, SOC levels will eventually reach a new equilibrium, halting further sequestration19. Increasing SOC sequestration cannot, therefore, be considered a universal nor a long-term emissions abatement strategy and should be accompanied by other policy initiatives and technological innovations targeting other key emission sources in the agri-food system20.

Field-level N2O emissions, influenced by amounts and types of N fertilizer applied21 and regional climate and management factors (particularly related to crop residue management)22,23, were another key driver of regional differences in carbon footprints. For example, Australian N2O emissions were found to be lower than any other country in this analysis due to arid climate conditions and low fertilizer application rates, while France and Germany had relatively high field-level N2O emissions due to their higher rates of fertilizer application and precipitation7,8,9. Reduction of field-level N2O emissions is another important emissions mitigation strategy that may be accomplished by increasing fertilizer-use efficiency following the Nutrient Stewardship 4R framework (that is, applying the right fertilizer at the right rate, time and location) to match crop requirements and minimize nutrient losses24. Precision agriculture technologies, such as variable rate fertilizer application based on spatially resolved assessment of plant nutrient requirements, are particularly salient in this context. Additional technological innovations, such as alternative methods for ammonia synthesis25 and P fertilizer production26, may also greatly reduce the contributions of fertilizer production to crop carbon footprints.

It is often assumed that local products have lower carbon footprints than those produced elsewhere in the world and shipped to market27. It is shown here that this is not necessarily the case, particularly when there are large regional differences in resource efficiencies, production conditions and climate factors—even if transportation distances are large. Although the crops considered rank among the lowest carbon intensity agricultural products, differences in production efficiency typically far outweigh the contribution of transportation in determining their relative carbon footprints in global markets. However, the relative proportion that transportation contributes to the overall carbon footprint increases as production-related emissions decrease. The ultimate impact of these commodities will also depend on the processing and distribution of the final product, as well as its end of life. Consideration of regional differences in production emissions should therefore play a key role in marketing and procurement policies developed with the intention of reducing the carbon footprints of food systems.

Methods

Data selection and quality control

Data were sourced for 13 crop/country combinations (Supplementary Note Section 1.1), including 3 Canadian crops for export and key producers of these crops internationally. The Canadian average was calculated as a production-weighted average of the prairie provinces of Alberta, Saskatchewan and Manitoba, as these represent 98–99.9% of the production volume of Canadian rapeseed, non-durum wheat and field peas and are representative of Canadian national exports28. These data included crop yields, inputs of seeds, fertilizers and pesticides, energy use (irrigation, field activities, post-harvest, transportation) and field-level emissions of GHGs. Data sources included publicly available national and commercial life cycle inventory databases, national and international statistics databases, publications from relevant governmental agencies for each country, crop growers’ associations, national and international sustainability consortia and peer-reviewed literature (see Supplementary Data 1–3 for complete list of sources screened). The ecoinvent database (v3.8 APOS) was chosen as a single background data source to ensure methodological consistency for all background data such as fertilizer, fuel and pesticide production.

Data quality assessment

All data points were screened using a fit-to-purpose, modified pedigree matrix to assess quality in terms of reliability, completeness and correlation29. The best fit data for each data point were chosen by converting pedigree matrix quality scores into base uncertainty factors, in line with current best practices in life cycle assessment29,30. Adjustments were made to pedigree matrix descriptors to account for factors such as inter-annual variability in yields, the differences between verified and estimated data values, number of replicate measurements and the number of sites covered in a particular region (Supplementary Note Section 1.3). Once uncertainty values were calculated for each data point from each data source, the calculated uncertainty values for data points representing the same inputs for each crop/country combination were compared to identify the data point/source with the highest quality (that is, that introduced the least amount of uncertainty into the final results). The choice of best fit data for modelling each data point for each crop/country combination therefore took into account these overall data quality scores. See Supplementary Note Section 1.5.6 for a complete list of data sources chosen.

Carbon footprint methods

A carbon footprint analysis was conducted following the ISO 14067 standard5. Results for each crop/country combination were based on a functional unit of 1 kg of product (that is, rapeseed, wheat grain or field peas) at the farm gate. The system boundaries included farm-level inputs of fertilizers, plant protection products and seed, energy for irrigation, field activities, post-harvest activities such as product drying, outputs of grain (and straw, when relevant) and field-level GHG fluxes. Intergovernmental Panel on Climate Change (IPCC) Tier 2 methods31 were used to model direct and indirect N2O emissions and SOC changes, using data and emission factors from each country’s 2022 National Inventory Report6,7,8,9,10, with a key limitation that these SOC change values are regionalized but not crop specific. IPCC Tier 1 methods31 were used for modelling CO2 emissions from lime and urea. A detailed description of the carbon footprint estimation method is provided in Supplementary Note Section 1.5, including soil carbon and emissions modelling in Supplementary Note Section 1.5.8.

Allocation methods

Specific allocation methods were implemented where necessary to distinguish between the agricultural practices used in each crop or crop/country combination, including (a) whether or not manure was used as fertilizer, (b) nitrogen credits from pea crops fixing their own nitrogen and (c) the effect of crop residues.

Manure inputs sourced from animal production systems contain nutrients that largely originated from synthetic fertilizers (that is, those used to grow the livestock feed). Manure nutrients were hence modelled based on synthetic fertilizer production processes, taking into account their recycling by applying a 50/50 allocation of production emissions between their first and second uses32.

Nitrogen credits were modelled using system expansion and substitution. The nitrogen credit provided by peas for the next crop in rotation was modelled as an avoided input of ammonia fertilizer, reflecting the fact that the next crop in rotation would require a smaller nitrogen input due to the nitrogen fixed by the peas. This was modelled as ammonia because this is the simplest nitrogen fertilizer, and it is used as the building block for all other nitrogen fertilizer types.

A portion of wheat straw is generally harvested and removed from fields to be used in other processes. Therefore, wheat grain and straw were considered to be co-products of wheat production systems. In the absence of high-quality, consistent data regarding the total amount of residues removed per region, a standardized rate was applied to all regions based on data for Canada. For all wheat production models, we therefore assumed that 8.3% of wheat straw was removed from fields, reflecting 34.5% of the straw removed from 24.1% of the non-durum wheat area33,34,35. This was then used to calculate the mass-based allocation factors for wheat grain and straw.

Calculation of break-even transportation distances

To determine the break-even points for transportation of exported Canadian field crops to each overseas importing country, first the differences were calculated between the emissions of production and transportation to port in Canada and in each other country producing each crop domestically, both including and excluding soil carbon changes. These differences in total footprint were divided by the emissions of shipping to get the break-even transportation distances. Then, the distances to ship to each country were subtracted from the break-even distances. If the break-even distance was higher than the shipping distance, this means that the emissions of producing the crop in Canada and shipping it to the other country were still lower than local production in that country. The leftover transportation distances (break-even minus actual shipping distance) were then expressed as the number of trips from Canada to the country and as circumnavigations of the globe (as an accessible example of distance, rather than a practical shipping route). See Supplementary Data 4 for detailed calculations.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.