Production Efficiency or Food Miles: Comparative Life Cycle Assessment of Local and Imported Peas and Lentils at Market in Western Europe

Abstract

A life cycle assessment was conducted to compare the impacts of peas and lentils produced in Canada, France, and Russia, transported to market in Western Europe, to assess the systems-level sustainability implications of changing production and consumption profiles of internationally traded commodity pulse crops. For all but 1–2 impact categories, imported Canadian peas and lentils outperformed those imported from Russia, due to the lower yields, higher levels of tillage, higher field-level emissions, and higher distances of truck transportation for Russian pulses. French peas had higher impacts of production than Canadian peas, for all categories but land use, due to higher levels of fertilizer inputs, irrigation, field activities, and field-level emissions. However, for 7 out of 12 impact categories, the impacts of the transportation of Canadian peas to Western Europe outweighed the higher impacts of the production of French peas. This demonstrates potential sustainability benefits of Canadian pulses, with some trade-offs from the additional impacts of transportation to market, adding nuance to the discussion around the importance of “food miles” in agricultural sustainability. Compared to previous studies, this demonstrates the importance of multi-criteria and regionalized assessments.

1. Introduction

Canadian pulse production is a growing industry, providing a high-quality plant-based protein source for food manufacturers, food service providers, and retailers [1]. A large proportion of Canadian pulses, such as peas and lentils, is exported, including to Europe, making Canada a leading exporter of pulses [1]. Pulses are high in protein and fibre, low in fat [2], and are cost-effective food products [3]. They have also shown significant environmental benefits when included in crop rotations, particularly with respect to biological nitrogen fixation (BNF) and reducing greenhouse gas (GHG) emissions. Pulse crops can increase sequestration of soil organic carbon (SOC) from carbon added to soils in their crop residues [4], and generally require lower inputs of pesticides and fertilizers than cereal crops—which lowers the field-level emissions from their application and the life cycle impacts associated with their production [5]. The BNF capabilities of pulse crops in rotation can also reduce fertilizer requirements for other crops, and improve yield quality and quantity [6].

Previous research has highlighted the relatively low environmental impacts of Canadian pulse production [6,7,8]. There can be substantial differences in the life cycle impacts of the same agricultural products produced in different regions due to differences in climate, soil, management practices, and supply chains. In order to assess these regional differences, recent studies [9,10] highlight the importance of building regional life cycle inventories (LCIs) and performing regionalized LCAs of agricultural products. It is therefore important to assess the sustainability of pulse production in different regions, for example Europe compared to Canada, with attention to the differences in soil types, climate, management practices, energy inputs, and yields, taking into account the relative contributions of transportation to market in Western Europe.

When assessing the life cycle impacts of imported products, transportation to market must be considered. It is often assumed that locally produced foods are more sustainable [11] due to the impacts of “food miles”. However, it has been demonstrated that transportation makes relatively small contributions to the life cycle impacts of food production—for example, only 11% of life cycle GHG emissions, with only 4% from delivery from producer to retail for the US food system [12]. On this basis, the authors [12] demonstrated that shifting diets towards foods with lower impacts of production is a much more effective strategy than eating locally produced foods, to lower the carbon footprint of the US diet. However, pulses have been identified as one of the least impactful food products available [13]; therefore, transportation impacts may be of relatively higher importance, given the low impacts of production. For this reason, the relative importance of transportation impacts, and the comparison of imported and locally produced pulses, including transportation, were the focus of this study.

In order to assess these differences in sustainability along the supply chains of pulses produced in either Canada or Europe and transported to market, life cycle assessment (LCA) can be used. LCA is a widely recognized methodology for quantifying resource inputs and emission outputs throughout a product’s life cycle for the purpose of multi-criteria environmental impact assessment and hotspot identification [14]. The objective of this study was to use LCA to compare the impacts of peas and lentils produced in Canada for export to Western Europe to those produced in both Western Europe (e.g., France) and Eastern Europe/Eurasia (e.g., Russia). These locations were selected based on market relevance (pers. comm. Denis Tremorin, Pulse Canada [15]).

In the literature, there are several LCA studies that compare pulse-based foods to meat products, including transportation impacts. Apaiah et al. [16] compared pea protein and pea soup to pork mincemeat, using an exergy analysis, including the contributions from transportation to market. However, they do not report the location of production for any of the supply chains; therefore, clearly, a regional comparison is not relevant to their study. Carlsson-Kanyama et al. [17] compared life cycle (including transportation to consumer) GHG emissions and energy consumption of a variety of food products consumed in Sweden, including peas, pork, and other plant-based staple foods. In this study, there were some foods produced locally in Sweden, and some imported (e.g., rice and tomatoes); however, this was to represent current average consumption patterns in Sweden, rather than potential scenarios to inform green procurement initiatives. Kekes et al. [18] compared different scenarios of producing a legume-based vegan burger, using LCA and life cycle costing (LCC). The system boundary included transportation to retail, but the focus was on eco-design, in other words, optimizing the production of the single vegan burger product. The only region- or transportation-based scenario was a scenario that varied the assumed transportation distances to retail, but this did not include a comparison of different production impacts in different regions.

There have been two studies in the literature that have examined the impacts of locally produced pulses compared to imported pulses, including the differences in production and transportation. One study assessed the life cycle impacts of Swedish pulses compared to imported pulses from China, the USA, Italy, Canada, and Turkey, including processing, packaging, and transportation [19]. They found that long transportation distances contributed considerably to energy use and climate change impacts, and the country of origin particularly affected biodiversity impacts. The other study assessed carbon footprints of Canadian rapeseed, wheat, and peas compared to French, German, Australian, and US counterparts, including transportation to market in Europe and/or Australia [20]. They found that, particularly for rapeseed and wheat, but also somewhat for peas, the impacts of producing Canadian crops were so much lower that they could be transported to market in Europe up to 17 additional times (in the case of rapeseed imported to Germany), before the impacts broke even with the higher production impacts of the locally produced crops. Importantly, this study was only a carbon footprint, and climate change was found in other studies to be one of the impact categories most influenced by transportation distance [19]. Also, there have not been any studies in the literature focusing on the particular combinations of pulse crops and competitor countries deemed most relevant by the Canadian industry experts.

Therefore, the goals of this study were as follows:

• To compare the life cycle environmental impacts of peas and lentils produced in Canada for export to Western Europe with peas produced in France and peas and lentils produced in Russia, using attributional LCA.
• To estimate the marginal impacts of transportation on the overall supply chain impacts of Canadian and Russian peas and lentils imported for consumption in Western Europe.

To date, there has been no such comparison made in the literature. This comparison was performed using methodologically consistent modeling practices for maximum comparability. The results of this comparison will inform production and consumption patterns for increased efficiency and sustainability in international markets for commodity pulse crops.

2. Methods

A cradle-to-market attributional LCA was carried out according to the ISO 14040/14044 standards for LCA [21,22].

2.1. Intended Applications

The results of this study can be used to make conclusions around the sustainability of Canadian, French, and Russian peas and lentils for market in Western Europe, as well as the relative contributions of transportation to the life cycle impacts of exported peas and lentils. These can form the basis of recommendations around modes of transportation and locations of production of peas and lentils to minimize environmental impacts. In turn, the results can also be used for sustainability and marketing purposes in the pulse industry.

2.2. Functional Unit

Both the functional unit and reference flow are defined as 1 kilogram (kg) of dried, marketable peas or lentils at market in Western Europe. In Canada, peas are dried to 16% and lentils to 14% moisture content [23,24].

2.3. System Boundaries

The overall scope of this attributional LCA was cradle-to-market, meaning it included all stages from the extraction of raw materials to the production of the peas or lentils on farms and their transportation to market in Western Europe. Subsequent “downstream” consumption and end-of-life stages were not included. The system boundary also excluded any processes that were associated with the establishment of the farms. The system boundaries of the study (represented in the flow diagram in Figure 1) are based on the Canadian pea and lentil LCI and LCA study from [8], where they are described in detail, and are briefly summarized here.

• Farm operations included tillage, harrowing, seeding, land rolling, fertilizer application, pesticide and other agri-chemical applications, swathing, and harvesting.
• Farm inputs/outputs included seed, fertilizer, plant protection products, desiccants, inoculants, irrigation (when relevant), energy carriers (i.e., fuels and electricity for farm operations), and harvested/dried pulses.
• Field-level emissions of carbon dioxide (CO₂), nitrous oxide (N₂O) (direct and indirect), ammonia (NH₃), nitrate (NO₃), nitrogen oxides (NOx), phosphate (P), and losses of crop protection products to air, soil, and water were included.
• N credits from biological N fixation and crop residues were considered as an additional function in the foreground system models. For the N fixed by the pulse crops, this was modelled as a negative input of ammonia fertilizer. This was chosen based on the logic that the N fixed by the pulse crop allows the farmer to apply that much less N fertilizer to the next crop in the rotation, and ammonia is the chemical building block for N fertilizers. Hence, system expansion was employed to deal with the multifunctionality of the system, which is detailed in the allocation section below.
• Drying (not including storage) of the pulse crops was included based on the required energy use.
• Transportation included the transportation of peas and lentils to market, as well as the transportation of farm inputs to the farm.

The following processes were excluded from the system boundary:

• Capital goods such as farm infrastructure and machinery used on the farm were not included in the inventory. Previous LCAs of peas and lentils in Western Canada have indicated that the production, distribution, storage, and disposal of on-farm capital goods (i.e., on-farm equipment) make only minor contributions to life cycle environmental impacts [6,25], and that impacts are highly sensitive to the assumed lifespan and use of the equipment.
• Storage (after drying) was excluded from the system boundary since a previous pea LCA carried out in the Canadian province of Alberta [7] indicated that grain storage was a minor contributor to the environmental impacts of Alberta pea production. Based on this and the advice of expert stakeholder groups (Canadian pulse growers’ associations and Pulse Canada), storage impacts were excluded.

2.4. Geographical, Temporal, and Technological Boundaries

This study was representative of pea production in Canada, France, and Russia, and lentil production in Canada and Russia. The Canadian pea and lentil data were collected by Bamber et al. [8] at the provincial scale and were subsequently aggregated to a national average for this study.

The French pea production data were sourced from ecoinvent v.3.8, and are representative of the Barrois region in France [26]. The Russian pea and lentil production data were sourced from the report prepared by Lee [27] for Pulse Canada, and are reported at the Federal District scale, aggregated to the national level. The majority (85%) of Russian peas are produced in the Central, Southern, Volga, and Siberian Federal Districts, and 90% of Russian lentils are produced in the Volga and Siberian Federal Districts [27]. The datasets sourced from ecoinvent and [27] were remodelled for methodological consistency with [8], which also provides a detailed methodology for all modeling.

The Canadian pea and lentil data are representative of 2017–2019, the French pea data are from 2000 to 2004, extrapolated to the year 2021 (as indicated in the ecoinvent documentation, it was not indicated how this was carried out), and the Russian pea and lentil data are from 2018 to 2020. In all cases, the data are meant to represent the most technologically relevant inputs, outputs, and activities for peas and lentils produced in each region, and all deviations from this are accounted for in the data quality assessment (Section 2.6.3).

2.5. Allocation Procedure

In LCA, when a process produces multiple co-products, inventory flows and impacts associated with the process must be allocated between co-products. N credits from BNF were included as a co-product (or co-function) using system expansion, consistent with the ISO 14044 multi-functionality hierarchy [22], as described and justified in [8]. The pre-allocated processes in the ecoinvent APOS database used for background data were left as they are, and in the impact assessment calculations, physical allocation was performed in the openLCA software, to adhere to the ISO standard [22].

2.6. Life Cycle Inventory

LCI data for Canadian pea and lentil production were based on data collected from Canadian farmers (reported in [8]), and have been made available in ecoinvent versions 3.8 and later. The data for the production of peas and lentils in ecoinvent are available for each province; therefore, the national averages were calculated based on the tonnes of production represented by each provincial inventory [8]. The details on the ecoinvent processes used to model each input and output in the pea and lentil inventories are presented in [8], and the names and amounts of flows are included in the inventory results of this report (Table S4). An error was identified with the data available in ecoinvent v.3.8 (which was updated in a subsequent release), which was manually corrected. The inventory results presented in this manuscript represent the corrected values. See Table S1 for the processes modelled for Canadian peas and lentils, and the modifications made to them.

For Western European pea production, LCI data from ecoinvent were given priority since they are methodologically consistent with the Canadian data. Protein pea production in France (FR) was chosen as the most representative inventory from ecoinvent v.3.8 (Table S2). This dataset was modified to exclude the input of carbon dioxide, biogenic, from the air. This flow was included in the ecoinvent dataset to ensure a mass balance for the carbon that enters the crop from the air and then leaves the crop in a subsequent unit process. However, since the system boundary does not include the consumption and end-of-life of the crop, and for consistency with the Canadian inventories, this biogenic carbon input was removed. The packaging for fertilizer and pesticide containers was also removed, for consistency with the Canadian inventory, which only included the fertilizers and pesticides themselves. The names and amounts of the input and output flows are included in the inventory results of this report (Table S4).

There were no data available for pea and lentil production in Russia from ecoinvent, nor were there any published LCAs of lentil production in Russia at the time of writing. The primary source for Russian pea and lentil production data was the report prepared by [27] for Pulse Canada. This report provided some information on production inputs, outputs, and activities for peas and lentils produced in Russia.

In order to use consistent background datasets and modeling methodology, the ecoinvent processes for pea production in Alberta and lentil production in Saskatchewan were used as a template to model the Russian pea and lentil production systems. The input and output values were then modified based primarily on information provided by Lee (2022) [27]. For data points where no information was available from either of these sources, the values were assumed to be the same as in Canada (pers. comm. Denis Tremorin, Pulse Canada [15]). Based on Lee (2022) [27], the common field operations for Russian pulses include ploughing, cultivation, land rolling, planting, and harvesting. Inoculant, fungicide, and fertilizer are applied at the time of planting, and desiccant is applied at the time of harvesting Lee (2022) [27]. Lee (2022) [27] listed available fertilizers and pesticides, but did not indicate which ones were commonly used; therefore, it was assumed that these were the same as in Canada, since these were all also available in Russia. All values were scaled based on the differences in yield between Canadian and Russian peas and lentils, and providers were changed to be representative of Russian locations, when relevant. Pesticide emissions were calculated as in [8] for consistency with the Canadian datasets.

A detailed description of the processes modelled, and the modifications made to them, can be found in the Electronic Supplementary Material (ESM) Tables S1–S3.

2.6.1. Soil Carbon, Nitrogen Credit, and Emissions Modeling

Emissions of pesticide active ingredients were modelled in the Canadian pea and lentil inventories as per Margni et al. (2002) [28]—assuming 76.5% to soil, 8.5% to water, and 10% to air (with 5% taken up by the plant and not emitted). In the French pea model, 100% of each active ingredient is modelled as an emission to soil, as per ecoinvent methods. The difference from 95% to 100% is negligible, and the differences in emissions to soil, water, and air are also negligible since the characterization factors were the same for each. Therefore, these methods of pesticide emissions modeling can be directly compared. The Russian pesticide emissions were modelled the same as the Canadian emissions, since the inputs were also assumed to be the same (but scaled relative to the difference in yield between the two countries). Heavy metal emissions for all countries were modelled according to the SALCA model [29].

The Canadian direct N₂O emissions were calculated using the IPCC Tier 2 model, with country-specific factors from [30]. Direct N₂O emissions for France were taken directly from the ecoinvent dataset since it was also calculated using the same IPCC methodology. Russian direct N₂O emissions were calculated using the emission factors from the Russian National Inventory [31]. This report gives emission factors for both chernozem and sod–podzolic soils (as well as a generic factor for “other” soils), which are the two main soil types in the pea and lentil growing regions in Russia [32]. The emission factors for the three different soils were averaged based on their proportions in Russian agricultural soils (chernozems 64.1%, sod-podzolic 14.7%, other 21.2%) [31]. The Russian Federation (2021) [31] also provided factors for N inputs from above and below-ground residues; however, the document is in Russian, and the units are not clear in translation. Therefore, the Canadian values for above and below-ground residues from [33] were used. For all countries, indirect N₂O, NO₃, NO_x, NH₃, and CO₂ emissions were calculated using IPCC Tier 1 [34], and P emissions using the SALCA model [35].

Soil carbon sequestration is not included in the ecoinvent inventories; therefore, it was calculated independently for all three countries. Bamber et al. [8] estimated soil carbon sequestration for Canadian peas and lentils using the DNDC model. However, there was insufficient access to location, climate, soil, and management data to use this methodology for France and Russia. Therefore, the most commonly applicable model is the IPCC Tier 1 methodology for the estimation of soil carbon. According to the IPCC (2006) methodology [34], changes in SOC come from changes in land management, and if there are no changes in land management, then it is assumed to be no change in SOC. Since historical land management data for each crop type are not available for all countries under study, the estimates of total carbon sequestration/emissions from land use and management changes on all cropland from each country’s National Inventory Report (NIR) were consulted [31,36,37]. The total estimate of soil carbon change from cropland was divided by the total cropland in each country (from [38]) to give an estimate of soil carbon change per hectare of cropland. This was then scaled by the yield of each crop to give an estimate per kg of crop. Therefore, the SOC estimates are not crop or specific, and are simply allocated to peas and lentils based on their share of the total cropland per functional unit (based on yield).

The credit for fixed N from peas and lentils that is left behind to be used for the next crop was calculated using the methods specified by [39] for Top Crop Manager, for each country. This was modelled as a negative input of ammonia fertilizer, to approximate the benefit to the next crop of requiring less N fertilizer due to the N left by the pulse crops.

2.6.2. Transportation

Transportation methods and distances from Russia and Canada to Western Europe were obtained from Lee (2022) [27] and personal communications with Pulse Canada.

Table 1. Types of transportation modelled for each pulse transported to Western Europe.

Crop/Region	Method of Transport	Process Modelled (Ecoinvent v.3.8)	Distance (km)
Lentils CA	Trucking to Regina (roundtrip)	market for transport, freight, lorry, unspecified|transport, freight, lorry, unspecified|APOS, S-RoW	150
	Rail from Regina to Montreal	market for transport, freight train|transport, freight train|APOS, S-US	3072
	Container vessel from Port of Montreal to Port of Rotterdam	market for transport, freight, sea, container ship|transport, freight, sea, container ship|APOS, S-GLO	6280
Peas CA	Trucking to Saskatoon (roundtrip)	market for transport, freight, lorry, unspecified|transport, freight, lorry, unspecified|APOS, S-RoW	150
	Rail from Saskatoon to Thunder Bay	market for transport, freight train|transport, freight train|APOS, S-US	1438
	Bulk vessel from Thunder Bay to Antwerp, Belgium	market for transport, freight, sea, bulk carrier for dry goods|transport, freight, sea, bulk carrier for dry goods|APOS, S-GLO	8991
Lentils RU	Road (70%) from Volga FD to Volga (inland) port	market for transport, freight, lorry, unspecified|transport, freight, lorry, unspecified|APOS, S-RER	743
	Rail (30%) from Volga FD to Volga (inland) port	market for transport, freight train|transport, freight train|APOS, S-Europe without Switzerland	319
	Container vessel to Marseille	market for transport, freight, sea, container ship|transport, freight, sea, container ship|APOS, S-GLO	4786
Peas RU	Road (70%) from Central FD and Southern FD to Novorossiysk port	market for transport, freight, lorry, unspecified|transport, freight, lorry, unspecified|APOS, S-RER	523
	Rail (30%) from Central FD and Southern FD to Novorossiysk port	market for transport, freight train|transport, freight train|APOS, S-Europe without Switzerland	224
	Bulk vessel to Marseille	market for transport, freight, sea, bulk carrier for dry goods|transport, freight, sea, bulk carrier for dry goods|APOS, S-GLO	4786

shows the methods of transportation for each crop/region, and the processes used to model them. For Canadian transportation, data on methods and distances of land transportation, as well as the methods of transportation to ports in Western Europe, came from Pulse Canada (pers. comm. Denis Tremorin [15]). To calculate the distance from each Canadian port to Western European ports, an online sea route calculation tool was used (available at ports.com/sea-route). Russian transportation data came from the report on Russian pea and lentil production from Lee (2022) [27], which provided a rough estimate that 70% of land transportation is by truck, and 30% by train. They also provided distances from the most common regions of production to the most commonly used Black Sea ports. The online sea route calculation tool was again used to estimate the distances from the Black Sea ports to Western European ports. The port of Marseille, France, was used for this calculation since it is the closest Western European port to the Black Sea.

2.6.3. Data Quality and Uncertainty

Data quality scores were assigned to all LCI data points using the standard ecoinvent pedigree matrix available in openLCA (ESM Figure S1). Uncertainty was estimated using the parameter uncertainty scores associated with the data quality scores (ESM Figure S2), plus the base uncertainty associated with the variability of each value calculated from the raw data as a lognormal uncertainty distribution, when there were multiple replications of the same measurement, or estimated using the generic factors from Frischknecht et al. (2005) [40] (ESM Table S5). Uncertainty was propagated using 1000 runs of a Monte Carlo simulation to assess the uncertainty of the LCIA results, following the same methods as [8].

2.7. Life Cycle Impact Assessment

Impact assessment was performed in the openLCA software version 1.10.3, with the IMPACT World+ LCIA methods suite chosen for this study for consistency with previous LCAs performed on Canadian pulses for Pulse Canada [8,41]. Based on the ISO requirements, and previous LCA studies of pulses, the impact categories chosen were climate change, mineral resources use, fossil and nuclear energy use, terrestrial acidification, freshwater eutrophication, freshwater ecotoxicity, particulate matter formation, water scarcity, land occupation—biodiversity, photochemical oxidant formation, ionizing radiation, and ozone layer depletion (ESM Tables S6–S17). Impact assessment results were reported at the mid-point for all categories as well as the endpoint for biodiversity. As per the ISO standard [22] for LCAs, making comparative assertions, normalization, grouping, and weighting were not performed.

2.8. Sensitivity Analysis

Sensitivity analyses were performed to assess the sensitivity of the results to any parameter data, methodological choices, or assumptions that were deemed particularly uncertain or of low data quality, and/or that contributed significantly to the LCIA results. These included the method of transportation of Russian pulses to port, the tillage practices for Russian pulses, and the emissions modeling for field-level GHGs.

Lee (2022) [27] indicated that the exact proportion methods of transportation of Russian pulses to port was unknown, and they assumed a 70%/30% split of rail and truck transport. For the sensitivity analysis, scenarios were included where the transportation was either 100% rail or 100% truck. Lee (2022) [27] also indicated an example of typical field operations, which included ploughing, cultivation, land rolling, planting, and harvesting. As a sensitivity analysis, Russian pea and lentil production processes were instead modelled with the same field operations as Canadian production conditions. The main difference between the two models in the baseline scenario is the lower incidence of tillage on Canadian pulse farms.

For the baseline LCI, the field-level GHG emissions for Canadian and French pulses were taken directly from the ecoinvent databases. For the Canadian pulses, these values were previously calculated by [8] according to the IPCC methodology, which is the reported methodology for the French pea dataset as well. For Russian pulses, the emissions were calculated according to the Russian NIR [31], based on IPCC methods, as described in Section 2.6.1. For methodological consistency, a sensitivity analysis was included where the Canadian and French values were recalculated according to the newest French NIR methodology for France (at the time of writing) [37], and, for Canada, the Canadian Roundtable for Sustainable Crops (CRSC) carbon footprint reports [42]. These differ from the Canadian NIR [36], particularly with respect to their treatment of emissions from crop residue. Researchers with Agriculture and Agri-Food Canada recently published new guidelines to update the Canadian Tier 2 emission factors (EFs) for N₂O emissions from agricultural soils—in particular regarding the differentiation between sources of N from synthetic fertilizer, manure, or crop residues [43]. Specifically, they propose scaling factors to be applied to the N₂O EFs for sources of N of 1.0, 0.84, and 0.28 for synthetic N, animal manure N, and crop residue N, respectively. In practice, this means that estimates of N₂O from crop residue N are now only 28% of previous estimates using the same EF for all sources of N. However, the CRSC report uses a factor of 0.84 for crop residue N, which is the same factor used for organic manure, rather than the value of 0.28 suggested by [43]. This choice is justified by stating that there is insufficient evidence for the lower emission factor.

3. Results and Discussion

3.1. Life Cycle Inventory

The inventory includes inputs of seed, crop protection products (including herbicides, fungicides, and insecticides), inoculants, fertilizers, irrigation, and field operations (Table S4). It also includes outputs of yield, N credit, and field-level emissions. The complete openLCA model for Canadian pea and lentil production from [8] is publicly available on the Open Science Framework and in the Canadian Agri-Food Life Cycle Data Centre. The Canadian provincial inventories are also available in ecoinvent v.3.8, along with the French pea production inventory used in this study. A summary of the inventory values used for pea and lentil production in Canada, France, and Russia is presented in ESM Table S4.

Key differences between regions include higher yields in France and lower yields in Russia, leading to lower and higher, respectively, land use than in Canadian crops per functional unit. Canadian pulses generally had lower levels of field activities and chemical inputs than French or Russian pulses. In addition, Canadian soil, climate, and management conditions resulted in lower field-level emissions and higher levels of soil carbon sequestration.

3.2. Life Cycle Impact Assessment

In general, at market in Western Europe, imported Canadian peas and lentils outperformed peas and lentils imported from Russia, as well as locally produced French peas for some impact categories (Figure 2 and Figure 3). Russian pulses had 58–1414% higher production impacts (cradle-to-farm gate) than Canadian pulses in all impact categories, due to their lower yields (leading to higher impacts per kg crop), higher levels of tillage, and higher field-level emissions (e.g., GHGs, phosphate, nitrogen oxides, ammonia)—due to higher levels of fertilizer inputs, as well as soil and climate factors.

Canadian pulses generally had 9–60% higher transportation impacts than Russian pulses due to the higher distances of rail and sea transportation, despite the higher truck transportation for Russian pulses. However, at market in Western Europe, Canadian peas still had 19–148% lower impacts than Russian peas in all impact categories except photochemical oxidant formation (30% higher) and terrestrial acidification (7% higher), and Canadian lentils outperformed Russian lentils in all but photochemical oxidation (24% higher).

French peas generally had 44–2302% higher impacts of production (cradle-to-farm gate) than Canadian peas due to higher levels of fertilizer inputs, irrigation, field activities, and field-level emissions. However, for fossil and nuclear energy use, mineral resource use, ozone layer depletion, particulate matter formation, photochemical oxidant formation, and terrestrial acidification, the impacts of the transportation of Canadian peas to Western Europe outweighed the higher impacts of the production of French peas, leading to overall 10–1620% higher impacts of Canadian peas, at market in Western Europe. French peas also had 22–57% lower impacts at market than Russian peas for mineral resource use, fossil and nuclear energy use, terrestrial acidification, land occupation, photochemical oxidant formation, ionizing radiation, ozone layer depletion, and freshwater ecotoxicity.

Overall, transportation contributed 9–80% of the cradle-to-market impacts of peas, and 8–83% of lentils imported from Canada to Western Europe (Figure 2 and Figure 3). For peas and lentils imported from Russia, transportation instead contributed 4–55% and 3–51%, respectively. These regional differences were due to the generally higher total transportation impacts when importing from Canada (due to the longer distances travelled), as well as the lower production impacts of Canadian crops, leading to a higher relative importance of transportation. Acidification and photochemical oxidant formation were the impact categories with the highest cradle-to-market contributions from transportation, thus leading to the higher relative importance of transportation distance in determining the most sustainable location of production among imported pulses. Conversely, transportation contributed the least to land occupation and eutrophication impacts, and therefore, for those impacts, differences in production impacts were the determining factor in the relative sustainability of production locations.

Most impact categories followed the general trends presented above, with some notable exceptions. Notably, French peas had the lowest farm-gate (22% lower than Canada) and at-market (23% lower than Canada) impacts for land occupation due to their higher yields, leading to lower production-related land use, as well as lower transportation. French peas were the only irrigated crop; therefore, the water scarcity impacts of French peas were much higher (979–1753%) than Canadian or Russian peas. Irrigation contributed 93% of the production-related impacts of French peas. In addition to water scarcity, the inclusion of irrigation also significantly impacted the production impacts for ionizing radiation and freshwater ecotoxicity, contributing 47% and 15%, respectively.

Soil carbon sequestration had an important effect on the results for the climate change impact category. Without soil carbon, Canadian peas had the lowest overall carbon footprint at market in Western Europe (70% lower than France), and Russian peas had the highest (347% higher than Canada), including the impacts of transportation (Figure 2A). When soil carbon was included, French peas had the highest impacts (199% higher than Canada), followed by Russian peas (148% higher than Canada), and Canadian peas had the lowest, since French agricultural soils have net carbon emissions, and Canadian and Russian soils have net sequestration (

Table 2. Summary of climate change impacts (kg CO₂e/kg crop) for peas and lentils in each country.

Crop-Country	Crop Production Without Soil Carbon	Soil Carbon	Crop Production Including Soil Carbon	Crop Production, Soil Carbon, and Transportation
Peas CA	0.088	−0.065	0.023	0.181
Peas FR	0.288	0.254	0.542	0.543
Peas RU	0.392	−0.050	0.342	0.450
Lentils CA	0.178	−0.120	0.058	0.308
Lentils RU	0.737	−0.113	0.624	0.778

). These values were calculated based on the estimates of total carbon sequestration/emissions from croplands from the National Inventory Reports for each country [31,36,37]; therefore, they do not represent crop-specific data. A summary of the climate change impacts of peas and lentils from each country is presented for crop production only (without soil carbon), crop production including soil carbon, and crop production (with soil carbon) plus transportation to Western Europe (Table 2). Full LCIA tables are presented in ESM Tables S18 and S19.

3.3. Life Cycle Interpretation

3.3.1. Sensitivity Analysis

Transportation

For all impact categories, the train scenario had lower impacts than the original (up to 17% lower for ozone layer impacts of peas), and the truck scenario had higher impacts (up to 8% for particulate matter formation impacts of peas) (ESM Figures S3 and S4). However, these changes were small in comparison to the magnitude of the impacts and the differences between regions. For most impact categories, the ranking of the relative sustainability of pulses produced in Canada, France, or Russia was not changed by altering the modelled transportation methods. Transportation, therefore, represents a potential area for improvement for Russian pulse producers, but would not be sufficient to outweigh the sustainability benefits of Canadian pulses, due to their higher yields, less intensive field operations, and lower field-level emissions.

Field Operations

Overall, the Canadian field operations scenario reduced the impacts of Russian pulse production, up to 47% for ozone layer depletion impacts of Russian lentils (ESM Figures S5 and S6). Again, most of these changes were still relatively small compared to the overall magnitude of the impacts and the differences between regions of production, and, therefore, the relative ranking of the sustainability of pulses produced in each region did not change for the majority of impact categories. For peas, for ionizing radiation, Russian peas had higher impacts than both Canadian and French peas in the original scenario, but had lower impacts than French peas in the alternate scenario. Thus, reducing tillage practices also represents a potential area of improvement for Russian pulse production, but, again, this would not outweigh the sustainability benefits of Canadian pulses for the majority of impact categories.

Field-Level Greenhouse Gas Emissions

Recalculating the Canadian field-level GHG emissions according to the CRSC carbon footprint methodology [42] and the French GHGs according to the French NIR [37] increased both Canadian and French impacts, but the relative rankings of the countries remained the same (ESM Figure S7). Russian peas and lentils still had higher climate change impacts than Canadian peas and lentils. French peas had higher impacts on production than Canadian peas (both with and without soil carbon). With soil carbon and transportation-to-market included, Canadian peas still had the lowest impacts.

3.3.2. Comparison to Previous Studies

Bamber et al. [20] showed that Canadian crops (more so rapeseed and wheat, but also peas somewhat) had much lower carbon footprints than the same crops produced in Europe, even including the transportation impacts. The current study included additional pulse crops and different production regions, but, most importantly, included a comprehensive suite of impact categories to determine any trade-offs between different types of impacts. The results from the current study regarding the relative carbon footprints of Canadian and European crops were similar. However, when looking at other impact categories and other countries, the results became more nuanced. The production impacts of Russian pulses were, similarly, always higher than Canadian pulses, and for most impact categories (including climate change), impacts were still lower for Canadian pulses after transportation. However, this was not the case for photochemical oxidation and terrestrial acidification, which were higher for Canadian pulses than Russian pulses, at market in Western Europe. French pulses had lower production impacts than Canadian pulses for all impacts other than land use, due to their higher yields, thus requiring less agricultural land occupation. However, when transportation to market was included, Canadian pulses had lower impacts for only 5 of 12 impact categories assessed, including climate change. For the other seven impact categories, the high transportation impacts offset the lower impacts of production for Canadian pulses. This demonstrates the importance of performing a full LCA instead of a single-criterion assessment like a carbon footprint.

Tidaker et al. [19] also assessed pulses grown in Europe (Sweden) compared to imported pulses from a variety of countries, including Canada. They found that transportation to market contributed considerably to energy use and climate change impacts. The results of the current study differ somewhat from this, since the climate change impacts of imported Canadian pulses to Europe were always lower than local European (French) production, meaning that although transportation was a significant contributor, it did not affect the relative sustainability of production locations. However, of all the pulse products imported to Sweden [19], the Canadian pulses had the lowest climate change impacts. The same large influence was seen on energy consumption in both studies, given that fossil and nuclear energy use, and mineral resource use impacts were higher for imported Canadian pulses when transportation impacts were included. One potential explanation for the differences in conclusions between this study and [19] is the differences in production conditions between Sweden and France. Despite both being in Europe, they have quite different climates and management practices. This further underscores the importance of regionalized LCA modeling to accurately assess the impacts of products produced under diverse conditions.

4. Conclusions

The cradle-to-market life cycle impacts of peas and lentils produced in Canada, France, and Russia, for market in Western Europe, were assessed and compared. Overall, Canadian peas and lentils had lower impacts of production (cradle-to-farm gate) than French and Russian pulses. Including the impacts of transportation to market in Western Europe, local French peas had lower impacts than imported Canadian peas for 7 out of 12 impact categories. Imported Russian peas had lower impacts than imported Canadian peas for two impact categories, and Russian lentils for one. This was due to the higher impacts of the transportation of Canadian pulses to Western Europe. These transportation impacts were generally larger than the impacts of the production of Canadian peas and lentils. This shows that Canadian peas and lentils have high production efficiency, since they have lower production impacts, and sometimes lower to-market impacts, even including the large transportation impacts to Europe.

It is often assumed that locally produced foods are more sustainable [11] due to the impacts of “food miles”. However, in general, transportation makes relatively small contributions to the life cycle impacts of food production—for example, only 11% of life cycle GHG emissions, with only 4% due to delivery from producer to retail [12] for the US food system. On this basis, [12] demonstrated that shifting diets towards foods with lower impacts of production is a much more effective strategy than eating locally produced foods to lower the carbon footprint of the US diet. However, the current study focused on the same foods (peas and lentils) produced in different regions, and pulses have been identified as one of the least impactful food products available [13]. We found that, in this case, transportation impacts had relatively higher importance, given the low impacts of production, particularly of Canadian peas and lentils. However, this relative importance also varied between impact categories, with transportation being the main driver behind differences in some categories (e.g., photochemical oxidation and acidification), and not making significant contributions to others (e.g., eutrophication and land occupation).

The main drivers of the differences in production efficiency between Canadian and French pulses were the inputs of fertilizers, field activities, and field-level emissions of N₂O and P. These higher emissions were due to the higher application of fertilizers, as well as differences in soil and climate factors. French peas are also irrigated, and Canadian peas are not; therefore, the water use impacts of French peas were much higher. Russian pulses consistently had higher production and generally higher transportation impacts than Canadian peas. The production-stage differences were due to the lower yields for Russian pulses, the higher prevalence of tillage, and higher field-level N₂O emissions due to differences in soil and climate. The transportation-stage differences were due to higher distances of truck transportation, which generally has the highest impact among the transportation methods considered in this study. These results indicate the potential sustainability benefits of Canadian pulses in comparison to their competitors in the European market, with some trade-offs due to the required transportation to market.

This study represents the first LCA comparison of these important low-impact, high-nutrition crops from these various international competitors with high market relevance. The inclusion of the impacts of transportation to market in Western Europe is novel and adds nuance to discussions on both production efficiency and food miles. The results seen in this study serve to reinforce some results from the previous literature, such as the relative importance of transportation impacts to life cycle energy use impacts. On the other hand, the relative climate change impacts of crops transported long distances vary in the literature, depending on the regions being assessed. This demonstrates the need for regionally specific assessments, since relying on published studies from other regions may lead to erroneous conclusions. While inconclusive in terms of relative sustainability at market in Western Europe, these results may be used to inform evidence-based green procurement policies to choose between locally produced European pulses or imported Canadian pulses. This study does, however, provide fairly strong evidence against sourcing Russian pulses.

Limitations and Future Directions

Comprehensive LCI data for pea and lentil production in Russia were not available; therefore, it was assumed that Russian production conditions were similar to Canadian production wherever specific data were lacking, unless otherwise indicated. Similarly, the exact modes of transportation for Russian peas and lentils were not available; therefore, it was assumed that they were transported to port by 70% road and 30% rail, based on estimates in Lee (2022) [27]. It is assumed that they are then transported by bulk carrier (peas) and container ship (lentils) to ports in Western Europe based on expert opinion (pers. comm. Denis Tremorin, Pulse Canada [15]).

Historical land use change and management data for each crop were not available for all countries under study; therefore, the changes in soil carbon were estimated based on the National Inventory Report data for cropland as a whole [31,36,37], which means that the soil carbon estimates are not crop specific, and are instead estimates of the soil carbon fluxes for the entire area of cropland in each country, scaled by the yield of each crop to reflect the functional unit.

The French data sourced from the ecoinvent database were representative of an earlier time frame than the primary data collected for Canada and Russia. Although they were all part of the same version of ecoinvent and, therefore, followed a consistent methodology, the inventory data may no longer be representative of current conditions in France. In the future, we recommend a dedicated data collection effort in all countries of interest, conducted at the same time and in a methodologically consistent manner, in order to provide the most accurate basis for comparison.