Regionalized life cycle inventory data collection and calculation for perennial forage production in Canada: methodological best practices and limitations

Abstract

Purpose

In Canada, 95–99% of produced forages are consumed domestically each year, mainly by beef cattle. Despite their importance, their contribution to the Canadian livestock industry and associated ecosystem services has not been investigated. This study developed a life cycle inventory (LCI) of perennial forage production in Canada averaged from 2009 to 2018.

Methods

LCI data were sourced or calculated from up-to-date, regionally resolved sources and models. Inputs to perennial forage production included the following: concrete, steel, and plastic usage; machinery fuel consumption; electricity, natural gas, and water use for irrigation; and synthetic and organic fertilizer, lime, and herbicide use. Assessed emissions included ammonia and nitrous oxide (N₂O); carbon dioxide from energy use; herbicide, nitrate, and phosphate losses; and soil carbon accumulation. Results were expressed per metric tonne of harvested perennial forage dry matter at provincial and regional scales—Western Canada [British Columbia (BC), Alberta (AB), Saskatchewan (SK), Manitoba (MB)] and Eastern Canada [Ontario (ON), Québec (QC)].

Results and discussion

Rates of inputs varied, with generally lower nutrient but higher herbicide application in West vs. East. Irrigation was highest in BC, followed by AB and SK; energy consumption was highest in BC and lowest in QC. Higher N₂O losses and nutrient losses via leaching and runoff in the East were partially due to greater soil moisture. Although total harvested perennial forage area declined from 6.43 to 5.23 million hectares from 2009 to 2018, these lands continued to accumulate soil carbon. The time period used to calculate average yields affected LCI estimates, as prairie yields were lower 1994–2003/1999–2008 due to drought. Furthermore, soil carbon sequestration estimates were affected by the annual change coefficients employed, underscoring the need for careful interpretation of LCI outputs. Results were compared to other studies and highlighted the importance of the choice of data and methods in creating LCI, and the need for transparency.

Conclusions

This first national LCI of perennial forage production in Canada provides a baseline for LCI inputs and outputs associated with this sector, highlighting provincial and regional differences. Outputs can be used to conduct future life cycle assessments to assess the environmental impacts of forage production and generate recommendations to improve sustainability, and for education and marketing purposes. This study demonstrates methodological best practices for LCI data mining and calculations, within available data and model limitations, thereby identifying gaps and providing a roadmap for other countries or sectors to develop detailed forage LCI.

1 Introduction

The Canadian beef, dairy, sheep, and horse sectors are highly dependent on forage production, with 95–99% of the forage crop consumed domestically each year (Bonnefield Research 2016). The Canadian beef industry consisted of 3.78 million head of beef cows in 2021 and the cattle sector contributed $40 billion in production of goods and services and $15 billion to Canada’s GDP when considering direct and indirect impacts (Kulshreshtha and Nagy 2021; Statistics Canada 2022a). The beef cow herd is raised on improved and natural grasslands throughout the summer months, but during the winter perennial forage (hay) is the primary feed (Canfax Research Services 2022). Perennial forages also account for a proportion of the diet of steers and heifers during the backgrounding production stage (Legesse et al. 2016). Forage crops accounted for 15.2%¹ (5.72 million hectares, M ha) of the total crop area across the country in 2016 (Statistics Canada 2020a). Of the cropped area in SK and AB, 9.2% and 16.7%, respectively, are forages and account for more than half of Canada’s total forage production (Statistics Canada 2020a). Most forages are used by the beef cattle industry, with 92% of grass hay and 83% of grass-legume hay consumed by livestock in AB fed to beef cattle in 2016 (Cordeiro et al. 2022). At the national scale, approximately 80% of the total beef sector feed consists of forage (Legesse et al. 2016).

Compared with the production of annual feed crops, harvested perennial forages used in Canadian livestock production are typically low-intensity, low-input production systems that are often associated with more positive environmental outcomes. Synthetic fertilizers and herbicides are typically applied less frequently to perennial forage land, to a smaller proportion of these lands, and at lower rates per hectare (Fertilizer Canada 2017; Gagnon et al. 2016; Sheppard et al. 2010a). Perennial forage production also requires fewer inputs than annual cropping systems since it does not require replanting every year (Daneshfar and Huffman 2016). Well-managed rangelands and perennial crops can reduce surface runoff and soil erosion, improve water recharge and infiltration, and benefit soil biota and nutrient cycling (DuPont et al. 2010; Lobb et al. 2016; Weber and Cutlac 2017). Soil C stocks and nutrients are also enhanced in perennial forage production systems (Baron et al. 2001; Mapfumo et al. 2002), and the lower N fertilization rates typically result in lower ammonia (NH₃) and nitrous oxide (N₂O) emissions than annual cropping systems (Liang et al. 2020; Sheppard et al. 2010a). However, forage production can also have negative environmental effects, such as disposal of the plastic used to wrap bales or cover silage. In 2016, plastic used in the agriculture sector accounted for approximately 1.4% of all discarded plastic, of which 5% was recycled, 5% was incinerated (waste-to-energy), and the remainder landfilled (ECCC 2019). Incineration releases toxic gases and decomposition in landfills emits greenhouse gases (GHGs) (Azoulay et al. 2019; Verma et al. 2016). Plastics that do not immediately decompose can contribute to microplastic pollution of surface water and soils, reducing fertility and plant growth (Souza Machado et al. 2019; Stubenrauch and Ekardt 2020). Furthermore, the beef sector is scrutinized globally for its GHG emissions. Although the GHG intensity of Canadian beef production fell from 14.0 to 12.0 kg CO₂ equivalents per kg of liveweight produced between 1981 and 2011, total emissions rose by 28% over the same time period due to an expansion in beef production (Legesse et al. 2016).

The marketing of forages plays a significant role in the viability of the ruminant sector in terms of revenue generation and sustainability of the feed supply. Between 1981 and 2019, farm cash receipts from hay and forage seeds increased from ~ $42 to $739 million CAD (Statistics Canada 2020b). In Canada, due to the climate-driven uncertainty of forage productivity, beef producers are not always able to produce as much forage as required, and as a result, of the beef operations that harvested forages in 2011, 12% purchased annual forages and 31% purchased perennial forages (Sheppard et al. 2015). A portion of Canadian forage crops is also exported, with approx. 471,600 tonnes (t) exported in 2020, of which 66.1% was Timothy (Phleum pretense) hay, valued at $133.5 million CAD (GoA 2021).

The forage sector in Canada is likely to face multiple climate challenges as conditions become drier and hotter, increasing the incidence of drought across the southern prairies (Bush and Lemmen 2019). Competition from alternative land uses also presents a challenge, as the conversion of native grasslands and perennial forages to annual croplands, and their permanent loss to urbanization, reduces their availability to support grazing and forage-fed livestock and to provide ecosystem services. For instance, the conversion of ~ 50 M ha of Canada’s native grasslands to annual crops during the late nineteenth–early twentieth century reduced topsoil fertility and organic matter and enhanced soil erosion (Bailey et al. 2010). Despite its importance, the contribution of forage production to the Canadian livestock industry and associated ecosystem services has not been thoroughly investigated (Pogue et al. 2020).

In Canada, the production of harvested perennial forages for the livestock sector typically occurs in production systems distinct from those of annual crops. The Canadian Cow-Calf Cost of Production Network identified hay as the dominant winter feedstuff on cow-calf operations (Canfax Research Services 2021). The dependence on homegrown perennial forages means that these systems are frequently managed using a harvested (e.g., hay, haylage) and grazed rotation, with longer periods between rejuvenation in the west where stand establishment is more risky due to a lack of rainfall.

Previous life cycle assessments conducted elsewhere have examined the environmental effects of production of a single forage type (Bacenetti et al. 2018), while others have investigated multiple forages or mixtures (Hauggaard-Nielsen et al. 2016), or the effects of switching from one perennial forage type to another. For example, Bernas et al. (2021) reported multiple environmental benefits of using the cup plant (Silphium perfoliatum) as perennial forage compared with more conventional forages such as corn and lucerne for silage. When soil C changes were considered, the C footprint of a Canadian dairy farm was greater with corn than alfalfa (Medicago sativa L.) silage (Little et al. 2017). Samarappuli and Berti (2018) reported that although corn silage is a preferred forage and feedstock for biogas production, forage sorghum and forage sorghum-corn intercropping could replace at least a portion of corn silage, resulting in less serious environmental consequences.

The objectives of this study were to (i) conduct the first national life cycle inventory (LCI) of perennial forage production in Canada using the most recent data and models available at provincial and regional levels, averaged over a 10-year period (2009–2018); (ii) examine the effects of different yield scenarios on the LCI outputs; and (iii) examine the influence of alternative soil organic carbon (SOC) accumulation rates on the total amount of C stored in soils cropped to perennial forage.

2 Methodology

2.1 Scope of the life cycle inventory

The LCI assessed multiple environmental parameters associated with the production of perennial forages in Canada (Fig. 1). For the purposes of this study, all perennial forages grown and harvested for hay, haylage/silage, or seed were included, and hereafter are referred to as harvested perennial forages, as defined by Statistics Canada (2019a). According to the Farm Management Survey (Statistics Canada 2018), 82% of harvested forage is used to produce dry hay (stored in bales), with the remainder conserved as haylage/silage (stored in either vertical silos, horizontal silos, or plastic sealed, Fig. 2).

Fig. 1

Geographical, temporal, and technical scope of the perennial forage life cycle inventory. BC, British Columbia; AB, Alberta; SK, Saskatchewan; MB, Manitoba; ON, Ontario; QC, Québec. ^aPlastics and paper utilized during field operations (prior to storage) include polypropylene (PP) used for totes; high-density polyethylene (HDPE) used for containers; and paper and cardboard. ^bPlastics used for the storage of hay bales include PP used for bale twine and HDPE used for bale net wrap. ^cPlastics used for the storage of haylage and silage include low-density polyethylene (LDPE) used for silage wrap and plastic covers for horizontal silos

Fig. 2

Overview of the distribution of harvested perennial forages in Canada, excluding the Atlantic provinces, among different conservation and storage types, and the storage materials considered in the LCI for each harvested forage and storage system

The system boundaries included the following inputs to the production process: seed; concrete, steel, and plastic used for storage; machinery fuel consumption, electricity, and natural gas for irrigation; water consumption for irrigation; and the use of synthetic or organic (manure) fertilizer, lime, and herbicides. Assessed outputs from the production process included direct N₂O emissions from land-applied manure and synthetic fertilizer, crop residues and mineralization; indirect N₂O emissions from N leaching and runoff and re-deposition of volatilized N; and NH₃ emissions from land-applied manure and fertilizer. Phosphate (PO₄) leaching and runoff to surface and groundwater and nitrate (NO₃) leaching to groundwater were also considered, as well as plastic and paper waste, land cover change, and the accumulation of SOC in perennial croplands.

Parameters were estimated at provincial and regional scales—Western Canada [British Columbia (BC), Alberta (AB), Saskatchewan (SK), and Manitoba (MB)], Prairies (AB, SK, and MB), and Eastern Canada [Ontario (ON) and Québec (QC)]—averaged over a 10-year period from 2009 to 2018. Regional LCI outputs were estimated as average values for the relevant provinces and weighted based on the proportion of total area or total production of harvested perennial forages in each province. The Atlantic Provinces were excluded as their forage production was minimal, representing only 3.6% of total national production from 2009 to 2018 (Statistics Canada 2019a). Inventory components were estimated per ha and per t of harvested forage corrected to 90% dry matter (DM, without storage losses). As inventory data were not available at the individual field scale, they were averaged across all perennial forage land/production, even when a particular LCI component was not relevant to the entire land base for a given province. For example, to estimate manure nutrient application rates per hectare, the total nutrient amount applied to perennial forages was divided by the total harvested forage land area within each province.

2.2 Harvested forage area and production

The functional unit used for LCI outputs was 1 t of harvested perennial forage corrected to 90% DM. Annual data regarding the area of harvested perennial forage (ha) and production (t ha⁻¹) between 2009 and 2018 were obtained from Statistics Canada (2019a) and a 10-year average was calculated for each province. Provincial yields were estimated by dividing total forage harvested by total area harvested for the 2009–2018 period. Field losses were accounted for in the estimates of total forage harvested and double counting of seed use was avoided by subtracting the quantity of seeds used to sow the crop from the harvested quantity of forage. Seeding rates were from provincial sources for AB (AARD 2010), SK (GoS 2019a), MB (GoM 2019), ON (OMAFRA 2019), and QC (Beauregard 2015). For BC, data from neighboring AB were used as a proxy. Annual seeding rates were adjusted based on the percentage of fields in the establishment phase in each province for each year (Statistics Canada 2018, Table 1).

Table 1 Seed input amounts and proportion of fields established each year for forage production by province

Dry matter yields were adjusted to reflect storage losses for different conservation methods (vertical silo, horizontal silo, and plastic-sealed haylage/silage; dried hay) (Rotz and Muck 1994). Provincial data regarding the percentage of total harvested forage stored using each method were obtained from Statistics Canada (2018, Table 2). The average percentage of harvested forage DM lost during storage for each province and storage type was estimated as the weighted average of the loss rate from each method of conservation used in each province. Forage yields with storage losses were calculated as the harvested DM yield before storage, minus the average storage loss for each province. Dry matter contents for the different conservation methods were estimated by province using storage distribution data and DM data for each method (Table 2).

Table 2 Percentage of total perennial forage production stored using different storage methods for each province and associated storage loss rates

2.3 Inventory inputs

2.3.1 External inputs—fertilizers and herbicides

Synthetic, organic, and lime fertilizers

Nitrogen (N), phosphorous (P), potassium (K), and sulfur (S) fertilizer were included in the LCI. For Western Canada, data regarding fertilizer application rates for perennial forage were available only for MB (Manitoba Agricultural Services Corporation 2019). These were average rates for 2009–2018 for alfalfa and alfalfa-grass hay for all municipalities and soil types, and were adjusted for the proportion of forage land in each province to derive fertilizer application rates for BC, AB, and SK (Eq. 1, Appendix 5.). For Eastern Canada, average annual fertilizer rates were determined using data from Stratus Ag Research (2018), which accounted for unfertilized fields. For ON rates were 39, 36, 76, and 9 kg ha⁻¹ and for QC rates were 68, 16, 34, and 2 kg ha⁻¹ for N, P₂O₅, K₂O, and S, respectively.

In Canada, livestock manure is typically applied to land within a very short distance of where it was produced (Nagy et al. 1999; Toma and Bouma Management Consultants 2006). Thus, it is reasonable to assume that livestock manure produced within a particular province was applied to agricultural land within that province. In the LCI, the amount of manure N excreted by all livestock types (beef cattle, dairy cattle, swine, poultry, and sheep) was estimated at the provincial scale based on the number of animals in each province (Statistics Canada 2019b) and the average manure N excretion rate for each animal group (ECCC 2022). Direct and indirect N losses from livestock manure (excluding manure deposited on pasture) during housing and storage were estimated based on the total N in each management system and total N loss fractions for each system (ECCC 2022). To estimate manure management system fractions for dairy manure, a relationship between farm size and time spent on pasture, exercise yards, and in barns was established for lactating and dry dairy cattle based on Sheppard et al. (2011a), and developed into a time series based on provincial farm size from the Census of Agriculture, and average provincial lactation days obtained from Lactanet. For dairy and swine manure deposited in barns, a manure storage time series was developed from the Farm Inputs Management Survey (1995), the Farm Environmental Management Surveys (2001, 2006, 2011) and the Livestock Farm Practices Survey (2005). The use of liquid systems was estimated by using a curve fitting process relating data on farm size to presence of a liquid management system for Eastern and Western Canada. The modelled liquid and solid fractions were further subdivided into livestock subtypes, manure subsystems and use of covers based on regional averages from multiple surveys. For liquid subsystems, crust formation was estimated based on data collected from the Livestock Farm Practices Survey (Sheppard et al. 2010b, 2011b). Provincial manure management system fractions for all other livestock were based on Marinier et al. (2005). This approach accounted for direct N₂O emissions and indirect N losses via NH₃ volatilization and leaching. Estimated losses were subtracted from the total N excreted to estimate the amount of N available in stored manure for application to agricultural land, by province, livestock type, and manure management system. To estimate the total amount of N in stored manure applied to harvested perennial forages, the amount available for application to land was multiplied by the fraction of manure applied to perennial cropland each month (vs. conventionally tilled or reduced till land) (Sheppard and Bittman 2012; Sheppard et al. 2009, 2010b, 2011b; Statistics Canada 2007) and the fraction of manure applied to harvested perennial forages (vs. tame/reseeded pasture), assuming an equal application rate across all perennials. The per ha application rate of manure N to harvested perennial forages was calculated as the total amount applied divided by the harvested perennial forage area, by province. Estimates of P and K applied to harvested perennial forage land in livestock manure were calculated based on the total amount of manure excreted by all livestock types (based on N excreted and manure N content), the P and K content of manure (as excreted), the proportion of stored manure applied to harvested perennial forage land, and the area of harvested perennial forages.

When used, application rates of lime to perennial forage were assumed to be constant and applied only in the year of establishment (Beauregard 2015). Rates were scaled within each province by the relative percentage of seeded forage area to which lime was applied. The annual lime fertilization rate (kg lime ha⁻¹) for forage land was calculated for each province (Eq. 2, Appendix 5.).

Herbicides

The active ingredients (a.i.) within herbicides considered were glyphosate, 2,4-D amine and MCPA (2-methyl-4-chlorophenoxyacetic acid). These were assumed to be applied only in the establishment year. Application rates were divided by the cropping duration of the stand (derived from Statistics Canada 2018) to provide yearly estimates. In ON and QC, the type of herbicides applied to Timothy-alfalfa silage and application rates were based on Beauregard (2015). For MB, application rates of glyphosate were derived from Wiens et al. (2014). Application rates were calculated based on the percentage of forage fields receiving herbicide during establishment (Statistics Canada 2018) and average annual provincial application rates. For BC, AB, and SK, no data were available; therefore, application rates for these provinces were derived from MB data. These data were adjusted proportionally to the percentage of lands on which herbicides were applied in each province.

2.3.2 Other inputs—irrigation and energy/fuel consumption

Irrigation water consumption

Blue water utilized for irrigation (water withdrawn minus water returned) in BC, AB, and SK was included in our analysis, but green water was not considered. Water utilized for irrigation was calculated primarily using data from the Agricultural Water Survey (Statistics Canada 2019c). Irrigated perennial forage was assumed to be used mainly for dairy cattle, horses, or export. For each province, the volume of irrigation water withdrawn (m³) per ha of harvested perennial forage was estimated based on the total area harvested (average 2010–2018, Statistics Canada 2019a) and total water used for irrigation of forage crops (hay + improved pasture, Statistics Canada 2019d). As no data were available for the amount of irrigation water used specifically for harvested perennial forages (2010–2018 average), it was estimated based on the total amount of water withdrawn for hay + improved pasture and the proportion of total irrigated ha of hay + improved pasture that were irrigated hay in 2010 (Statistics Canada 2019e).

Irrigation water used per ha of perennial forage was corrected for exported forage, based on the total area of exported forage (2010–2018 average; Statistics Canada 2019f) and multiplied by the volume of irrigation water used per ha. Finally, this value was adjusted to account for the irrigation efficiencies (the proportion of water withdrawn that is returned to ground or surface water) of three different irrigation systems: sprinkler, drip (micro), and flood irrigation (Nemecek et al. 2015). The volume of irrigation water withdrawn per ha of harvested perennial forage was multiplied by the percentage of farms in Canada using each irrigation technique (Appendix 6., Statistics Canada 2019g).

Energy and machinery fuel use

Machinery fuel consumption was estimated using the F4E2 model for 2011 in BC, AB, SK, MB, and ON (Dyer and Desjardins 2003, 2005). As no data were available for QC, machinery fuel use for field operations in this province was modeled using Holos (Little et al. 2013). The F4E2 and Holos models consider field operations associated with the production of perennial forages including tillage, sowing, fertilizer application, manure haulage and application, weed, pest, and pathogen control, harvesting (including cutting, raking, baling, or chopping), and transport of forage from the field (Dyer and Desjardins 2005; Little et al. 2013). Electricity and natural gas consumption for forage irrigation in each province (Table 3) was based on Dill (2010). Consumption estimates were based on the average of 11 irrigation systems (5 natural gas and 6 electric)² used in AB during the 2009 irrigation season. Holos provided the estimated energy requirement for herbicide application by Canadian region and soil type (Little et al. 2013). Irrigation-induced energy consumption for perennial forage production was compared to the Ecoinvent model v.3.5 life cycle inventory database, which contains estimates for average electricity use for the application of 1 m³ of irrigation water in the USA. This estimate was converted to per t harvested DM based on the irrigation volume consumed per ha in BC, AB, and SK.

Table 3 Energy use for perennial forage irrigation by province

2.4 Inventory emissions

2.4.1 Field emissions to air, water, and soil

Nitrous oxide

Annual N₂O emissions were estimated using Holos V4 (Pogue et al. in prep.). In Holos V4, the approach used to estimate direct and indirect N₂O emissions was based on the Canadian-specific method of Liang et al. (2020). Holos V4 operates on a daily timestep at the Soil Landscapes of Canada (SLC) polygon scale, and simulations were modeled for the 2009–2018 period for each SLC polygon across the six provinces, using input data for each province. Detailed data regarding soil characteristics, derived from the Soil Landscapes of Canada (SLC) database (Soil Landscapes of Canada Working Group 2010), climate, derived from the NASA POWER database (https://power.larc.nasa.gov//), and crop management were required. For crop management data, provincial per ha application rates for synthetic fertilizer and manure N were estimated as described in Sect. 2.3.1. Synthetic N fertilizer was assumed to be applied in the spring, as Sheppard et al. (2010a) reported that 82% of N fertilizer is applied to perennial forage and pasture crops between April and June in the East and between April and May in the West. Monthly manure applications were on the 15th day of each month, with solid manure being spread, and synthetic fertilizer and slurry being broadcast. Annual crop yields for tame hay were used at the Small Area Data Regions scale (Statistics Canada 2023) for each year of the simulation. Yearly above- and below-ground residue inputs were estimated using Holos based on yields and crop-specific defaults for moisture content and the ratio of above- and below-ground plant biomass.

Model outputs were per ha estimates for direct and indirect N₂O emissions from manure, fertilizer, mineralization, and crop residues (for direct emissions only). Indirect emissions included losses related to NH₃ volatilization and leaching/runoff. Additional outputs included per ha estimates for volatilized NH₃ and NO₃ lost via leaching/runoff. Outputs were provided at the SLC scale, and simple averages of these values were calculated for each ecodistrict. It was not possible to weight these averages based on the proportion of the ecodistrict perennial forage area in each SLC polygon, due to the large number of suppressed data points in the Interpolated Census of Agriculture at the SLC level (AAFC 2013a). Each SLC polygon and ecodistrict possesses a distinctive assemblage of topography, landforms, geology, soil, vegetation, water bodies, and fauna (Fig. 3, AAFC 2013a, b).

Fig. 3

Map of Canadian ecodistricts in the provinces of BC, British Columbia; AB, Alberta; SK, Saskatchewan; MB, Manitoba; ON, Ontario; and QC, Québec. Inset: Soil Landscapes of Canada (SLC) polygons for southern ON. Source: AAFC (2013a, b)

Each simulation was run twice: once without irrigation and once with irrigation (so that precipitation ≯ potential evapotranspiration). For those ecodistricts with irrigated perennial forage area, a weighted average of the estimates from the two scenarios was calculated based on the proportion of total forage area that was irrigated, as reported in the 2011 Interpolated Census of Agriculture (AAFC 2013b). Provincial estimates were then calculated as weighted averages based on the proportion of total provincial perennial forage area within each ecodistrict (AAFC 2013b). Provincial estimates were expressed as kg N₂O t⁻¹ of harvested forage (DM).

Carbon dioxide, nitrogen oxides, and ammonia

As for the approach used to estimate N₂O emissions, the methodologies used to calculate CO₂, NO_x, and NH₃ emissions are comparable to those recommended by the FAO LEAP Guidelines for the assessment of nutrient flows in livestock systems (FAO 2018). Carbon dioxide emissions from the application of lime fertilizer were calculated using IPCC (2006) methodology (Eq. 3; Appendix 5.), and CO₂ emissions associated with the production (i.e., upstream life cycle emissions) and application of different fertilizer blends were estimated using Holos V4. Emissions of NO_x were estimated based on EMEP/EEA (EEA 2016). Ammonia emissions from land-applied synthetic N fertilizer and manure were also estimated using Holos V4. For inorganic fertilizers, NH₃ emission factors (EFs) considered the type of inorganic N fertilizer, degree of soil incorporation, crop type, and soil chemical properties. For organic fertilizers, including manure, default EFs were used to generate province-specific EFs for dairy cattle and swine. A default EF of 0.21 kg (NH₃-N + NO_x-N) kg N⁻¹ was used for all other livestock types (ECCC 2022).

Herbicide emissions to air and soil

Estimates of herbicide (a.i.) application to perennial forages were obtained from Wiens et al. (2014) for MB and the other western provinces and Beauregard (2015) for ON and QC. Emissions of herbicide a.i. (glyphosate, 2,4-D amine and MCPA) to air and soil from field crops were estimated using dynamiCROP (Fantke et al. 2011; Fantke and Jolliet 2016), with the fraction of applied herbicide lost to the air fixed at 16.5% (10% from volatilization and 6.5% from wind drift). The remaining portion of herbicide was assumed to be deposited on the soil surface. In dynamiCROP, the fraction of land-applied pesticide emitted to the air during and after application is assumed to be pesticide-generic but crop-specific (Fantke and Jolliet 2016). As EFs for the application of herbicides to perennial forages were not available, those associated with wheat were used as default.

Phosphate, phosphorous, nitrate, and heavy metal emissions to ground and surface water

Default values for PO₄ flows to ground/surface water and P from the erosion of soil particles were derived from the Ecoinvent Report No. 15 (Nemecek and Kagi 2007), which used the SALCA-P model (Swiss Agricultural Life Cycle Assessment-P; Eqs. 4–7; Appendix 5.; Prasuhn 2006). Nitrate leaching to groundwater was estimated using Holos V4, which estimates leaching and runoff N losses from land-applied manure, as well as fertilizer and N mineralization, while considering a leaching fraction that is estimated based on precipitation and potential evapotranspiration (Liang et al. 2020). Heavy metal losses to the environment were also estimated based on Nemecek and Kagi (2007), which considered leaching to groundwater, flow to surface water, and deposition in agricultural soils.

2.4.2 Infrastructure and the production of agricultural waste

Field operations

Provincial estimates of the amount and types of waste generated in the production of perennial forages prior to harvest in each province were obtained from the CleanFarms report (ES Consulting Ltd. 2019), with consideration for the area of perennial forage production as a proportion of all agricultural land. Estimates were derived from a survey in AB that considered polypropylene (PP, polypropylene totes), high-density polyethylene (HDPE, pesticide containers), paper (paper and multi-walled bags), and cardboard. Per hectare estimates of paper and plastic waste associated with perennial forage production in AB were used for all provinces due to a lack of other data. These values were 55, 54, 80, and 79 g ha⁻¹ for PP, HDPE, paper, and cardboard, respectively.

Dried hay and haylage/silage storage

The median total tonnage of PP for twine and HDPE for net wrap in AB (ES Consulting Ltd. 2019) was multiplied by 60% to estimate the amount used for perennial forages (vs. annual straw) and divided by the total area of harvested perennial forages (2009–2018 average) to estimate usage per ha, adjusting for the proportion of total forage production that is conserved as hay bales. Due to a lack of suitable data for other provinces, the AB estimate was adjusted and applied to these provinces as follows: PP or HDPE per ha dried hay production in AB (g ha⁻¹) * ha harvested perennial forages used for dried hay production in province/ha harvested perennial forage in province.

Plastic wrap for hay bales and haylage/silage (low density polyethylene, LDPE) was modeled using two separate approaches for all provinces, firstly based on data from the CleanFarms report for AB and secondly using Groupe AGÉCO (2019) data, which is representative of agricultural plastic use in QC. Hay bales (dried), which account for 82% of harvested forage, were divided into square bales (25%), individual round bales (26%), and round bales stored in lines (49%) (Groupe AGÉCO 2019). For AB, SK, and MB, estimates of the percentage of bales that are wrapped were derived from the CleanFarms report at 14% of round bales stored in lines and 1% of round bales individually wrapped in plastic. Although these data relate to AB, they were also applied to BC, SK, and MB on the basis that the agricultural zones of these provinces employ similar plastic-wrapping practices based on comparable seasonal precipitation. For ON and QC, we assumed that 61% of all bales were plastic-wrapped, based on data for QC (C. Lajeunesse, Agrirécup, pers. comm., 2022). This percentage was applied to both provinces on the basis that broadly similar precipitation regimes exist across their agricultural regions.

Based on data from the CleanFarms report for AB, a median of 6033 g LDPE plastic were used per hectare of harvested perennial forages stored as plastic-wrapped hay bales. This value was estimated based on the reported quantity of LDPE plastic used for bale wrapping with the assumption that 60% of this amount was used for wrapping perennial hay bales, and the total area of harvested perennial forages, adjusting for the proportion of total forage production that is conserved using this method. To convert the AB estimate to an estimate of g LDPE bale wrap used per ha of harvested perennial forages for all provinces, the following calculation was used: g LDPE bale wrap per ha harvested perennial forages = g LDPE bale wrap per ha bale-wrapped hay (AB estimate) * ha harvested perennial forages used for hay production stored as plastic-wrapped bales in province / ha harvested perennial forages in province.

Also based on the CleanFarms data for AB, for plastic-wrapped haylage/silage, the amount of LDPE plastic used per ha of harvested perennial forages in AB was calculated as follows: g LDPE haylage/silage wrap per ha harvested perennial forage = g LDPE used for silage * 60% (used for perennials) / ha harvested perennial forage. To convert the AB estimate to estimates for the other provinces, the following calculation was used: g LDPE per ha harvested perennial forage = g LDPE per ha plastic-wrapped haylage/silage for AB * ha used for plastic-wrapped haylage/silage in province / ha harvested perennial forages in province. Based on Groupe AGÉCO (2019) data for QC, an estimated 0.00211 kg LDPE were used per kg of harvested forage in bales or plastic sealed haylage/silage. For each province, the amount of LDPE plastic used (for hay baling and haylage/silage) per hectare of harvested perennial forages was calculated based on: kg LDPE per ha baled hay/silage * (% harvested perennial forage in plastic-wrapped bales + % harvested perennial forage plastic-sealed as haylage/silage).

The concrete and steel used in the construction of horizontal and vertical silos in all provinces were estimated at 0.39 kg of concrete and 0.52 kg of steel t^–1 harvested forage year^–1 (over an assumed average silo lifespan of 35 years) using the Agri-footprint database (Durlinger et al. 2017). These values were converted to per hectare of harvested forage based on provincial forage yields and the percentage of perennial forage silage stored in horizontal and vertical silos. The amount of plastic utilized to cover horizontal silos was based on Groupe AGÉCO (2019) and Dubreuil et al. (2004) and assumed average silo dimensions of 2000 m³ and a forage density of 730 kg m^–3, with a capacity to store 1460 t of forage DM. The weight of this plastic was estimated at 86 kg with a lifespan of one year, representing 0.000059 kg LDPE kg^–1 forage.

2.4.3 Land occupation/transformation and effect on soil organic carbon change

The accumulation of SOC in land planted to perennial forage was estimated for each province using the National Inventory Report (NIR, ECCC 2022), which uses an annual linear coefficient over a 100-year duration to estimate SOC change based on the Century model (Parton et al. 1987, 1988). The accumulation of SOC was based on the assumption that the land was previously sown to perennial forage and that it continued to accumulate SOC throughout the 10-year period of this LCI.

The coefficients used to estimate SOC accumulation associated with perennial forages represented a balance between the rates of input from crop residues and losses through SOC decomposition. These coefficients consider environmental factors, such as soil type and characteristics derived from the Soil Landscapes of Canada database (Soil Landscapes of Canada Working Group 2010) and changes in perennial hectarage. Although the actual age of perennial forage stands was unknown, it was assumed that SOC levels had yet to reach equilibrium and continued to accumulate from 2009 to 2018.

2.5 Scenario analysis for perennial forage yield and soil organic carbon change

To examine the effect of variability in harvested perennial forage yield on LCI outputs, the results were re-calculated using average yield estimates from 1994 to 2003, 1999 to 2008, 2004 to 2013, and 2011 to 2015 (Statistics Canada 2019a). To examine the effect of using different annual SOC change factors, the inventory results were re-calculated using additional annual SOC change coefficients for conversion from cropland to permanent cover (reseeded grassland or perennial forage). These coefficients were derived from VandenBygaart et al. (2008, 2010), which reported field-based annual SOC change coefficients for multiple locations across Canada. For this LCI, provincial coefficients were calculated as the average of all reported values within a province. For provinces with no reported SOC change values, the average value for the relevant reporting zone was used.

3 Results and discussion

3.1 Forage area and production, with and without storage losses

Storage losses (% DM) ranged from an average of 5% in the West to 7% in the East (Table 4). The average stored conserved forage DM content for each province was higher in Western (83–85%) vs. Eastern (65–75%) Canada, likely due to the greater reliance on haylage/silage in the East.

Table 4 Dry matter losses (% DM) and average conserved forage dry matter content (%) by province and region

3.2 Inventory inputs

3.2.1 External inputs—fertilizers and herbicides

Synthetic/organic fertilizer and lime

Application rates of N, K, and S in synthetic fertilizers were lowest in AB and SK, ranging between 3.0 kg ha⁻¹ for S in SK and 22.3 kg ha⁻¹ for N in AB (data not shown). The per ha application rate of N in QC, at 68.5 kg N ha⁻¹, was more than double the national average of 27 kg N ha⁻¹. The application rate for P was also quite variable, ranging from 12.7 kg P₂O₅ ha⁻¹ in SK to 37.0 kg P₂O₅ ha⁻¹ in BC and MB (equivalent to 5.5 and 16.2 kg P ha⁻¹, respectively). The relatively low P application rate in QC of 16.2 kg P₂O₅ ha⁻¹ is likely a reflection of the stringent provincial regulations in this province (Québec Official Publisher 2021). Converting the per ha estimates to a per t harvested forage (DM) basis, application rates for N and K were also generally lower in the West vs. the East (Fig. 4). For example, N application rates ranged from 3.2 kg N t⁻¹ harvested forage (DM) in SK to 14.1 kg t⁻¹ in QC, while K application rates varied from 2.6 kg K₂O t⁻¹ harvested forage (DM) in SK to 14.8 kg K₂O t⁻¹ in ON. For P and S, the opposite was generally true, as per t estimates for P were lowest in QC and highest in MB (3.3 and 10.1 kg P₂O₅ t⁻¹, respectively), while for QC, the lowest value of 0.5 kg S t⁻¹ was reported, compared with 2.4 kg t⁻¹ in MB.

Fig. 4

Application rates of (a) synthetic fertilizer, (b) manure (nitrogen, phosphorous, and potassium), (c) lime, and (d) herbicides (glyphosate, 2,4-D amine, and 2-methyl-4-chlorophenoxyacetic acid) per t of harvested forage (DM)

Trends in nutrient application rates from manure were similar to synthetic fertilizer, with generally lower per ha rates for N application in the prairie provinces—6.2 kg ha⁻¹ in MB, 12.3 kg ha⁻¹ in SK, and 24.5 kg ha⁻¹ in AB, compared with 17.2 and 70.0 kg ha⁻¹ in ON and QC, respectively. On a per t harvested forage (DM) basis, N application rates ranged from 1.7 kg N t⁻¹ harvested forage (DM) in MB to a high of 14.4 kg t⁻¹ in QC (Fig. 4). For P and K, per t harvested forage (DM) application rates were lowest in ON (0.8 kg P t⁻¹ and 2.2 kg K t⁻¹) and MB (0.5 kg P t⁻¹ and 1.2 kg K t⁻¹), while rates were highest in QC (3.5 and 9.3 kg of P and S, respectively, t⁻¹ harvested forage (DM)). Of the total nutrients applied (synthetic fertilizer + manure) in the prairie provinces, most were as synthetic fertilizers, representing an average of 56.4% of N, 70.7% of P, and 54.2% of K applied t⁻¹ harvested forage (DM). In Eastern Canada, on average 59.4%, 54.1%, and 61.3% of N, P and K, respectively, were applied as synthetic fertilizers.

Sheppard et al. (2015) reported synthetic fertilizer N application rates to perennial forages in beef operations of 50 kg N ha⁻¹ and that, nationally, both synthetic N fertilizers and manure were applied to 19% of harvested perennial forages fed to beef cattle. Assuming that an average rate of 50 kg N ha⁻¹ was applied to perennial forages in the East and West, after considering those forage lands that received no fertilizer, the average synthetic fertilizer N application rate was 9.5 kg N ha⁻¹. This is significantly lower than the national average of 27 kg N ha⁻¹ estimated in the present LCI, but closer to the rates estimated for the West (between 8.6 kg N ha⁻¹ in SK and 25.2 kg N ha⁻¹ in BC and MB) compared with the East, where estimated rates for ON and QC were 38.8 kg N ha⁻¹ and 68.5 kg N ha⁻¹, respectively. The large discrepancy between the Sheppard et al. (2015) and the Eastern Canada estimates may be partly attributable to the low number of beef farms from ON and QC in their survey. For MB, the recommended rate of N application to grass stands and mixed grass-legume stands (with < 25% legume) is 44.8–67.3 kg N ha⁻¹ (Manitoba Government 2007), and for ON, OMAFRA (2017) reported a rate of 23 kg N t⁻¹ for grass stands (or 116.7 kg ha⁻¹, based on an average ON DM yield for 2009–2018 of 5.1 t ha⁻¹). When averaged across all fertilized (19% of perennial forage land) and unfertilized forage land, these values equate to approximately 8.5–12.9 kg N ha⁻¹ harvested forage (DM) for MB and 22.3 kg N ha⁻¹ for ON, estimates that are lower than ours. These differences may be partly explained by the fact that N application rates reported by Manitoba Government (2007) and OMAFRA (2017) are recommended rates, while actual application rates may be higher depending on several factors including the proportion of grass in the stand. Huffman et al. (2008) reported minimum and maximum N application rates (in synthetic fertilizer + manure) for hay and pasture of 1.3 and 24.9 kg N ha⁻¹ in Western Canada and 4.4 and 42.4 kg N ha⁻¹ for Eastern Canada (averaged across all fertilized and unfertilized land, assuming 19% of perennial forage land is fertilized). These estimates are generally lower than ours, possibly because they considered application rates for both hay and pasture, while our study only considered nutrient application rates per t of harvested perennial forages.

For manure N application, Legesse et al. (2016) used a value of 50 kg manure N ha⁻¹ applied to perennial forage land in Canada and assumed that 19% of grass and grass-legume hay land was fertilized with manure. If we apply this average rate to all perennial forages in the East and West, this results in an average N application rate of 9.5 kg N ha⁻¹, lower than the national average of 23.8 kg N ha⁻¹ estimated in our LCI, although Legesse et al. (2016) assumed manure N was applied at a rate to meet the N requirements of the crop, which may not always be the case. These Canadian rates can also be compared to N application rates reported outside of Canada. Hauggaard-Nielsen et al. (2016) modeled a forage legume-grass intercrop system in Denmark with low and high N application rates of 90 and 236 kg ha⁻¹ yr⁻¹. In contrast, Bacenetti et al. (2018) reported an application rate of 30 kg N ha⁻¹ for alfalfa in Italy in the first year, followed by no N application in following years to promote N-fixation. These studies illustrate the variability in N application rate depending on the crop, climate, and management conditions.

Data for P application to perennial forages in Canada are scarce. Heffer et al. (2017) reported that P and K fertilizer were not applied to grasslands in Canada (temporary and permanent grassland and pastures for hay, silage, and grazing) in 2014/15, which contradicts information provided in Canada’s Fertilizer Use Survey (Stratus Ag Research 2018). However, Heffer et al. (2017) did not account for the application of P to non-grass perennial forage crops, such as legumes. The Province of Manitoba (2006) suggested that alfalfa crops can typically benefit from P fertilization, as applying 33 kg P ha⁻¹ during seeding increased seedling size fourfold as compared to no P application. Hauggaard-Nielsen et al. (2016) reported P and K fertilizer application rates of 32 and 192 kg ha⁻¹ year⁻¹, respectively, for different perennial legume-grass intercrop systems in Denmark.

In Canada, livestock manure is typically applied within a short distance of where it is produced, primarily due to the economic restraints associated with transport. Nagy et al. (1999) reported that at two sites in SK, transport of cattle manure was economically viable up to 8 km, while from AB beef feedlots, it was up to 19 km (Toma and Bouma Management Consultants 2006). Assuming an equal application rate across all perennial forage lands ignores the fact that application rates are likely to be higher on both annual and perennial croplands that are in close proximity to intensive livestock operations. However, due to several factors, we could not determine the exact location of perennial forage lands in relation to intensive livestock operations. The Census of Agriculture failed to report the precise location of perennial forage lands over the timespan of the present LCI. Furthermore, there was limited information available regarding manure dispersal and application rates of manure to perennial forages. It is also likely that lands in the vicinity of intensive livestock operations are used to produce annual crops for use as feed. It is unlikely that these limitations had a significant impact on the results, as they were expressed on a per ha and per t of harvested perennial forage (DM) basis. Thus, it was most practicable to assume that manure within a province was evenly spread across all harvested perennial forage land.

The application of synthetic fertilizers to perennial forage land is comparatively small relative to annual cropland (Fertilizer Canada 2017). In the East in 2017, 69.9%, 53.9%, 66.7%, and 33.3% of perennial forage growers applied fertilizers containing N, P, K, and S, respectively, as compared to 96.6%, 86.4%, 88.1%, and 47.6%, respectively, for corn grain growers (Stratus Ag Research 2018). Sheppard et al. (2010a) reported average N application to tame hay of 1 kg ha⁻¹ in SK to 47 kg ha⁻¹ in BC and for alfalfa and alfalfa-grass mixtures of 1 kg ha⁻¹ in SK to 17 kg ha⁻¹ in QC. Annual crops receive far higher levels of N ranging from 50 to 81 kg ha⁻¹ in SK and QC, respectively (average for barley, canola, corn for fodder/grain, oats, and wheat) (Sheppard et al. 2010a). This is similar to N application rates for annual grain crops in the USA, ranging from ~ 45 to 80 kg ha⁻¹ (with no manure application) depending on soil, climate, and management factors (Cornell College of Agriculture and Life Sciences 2022). Similarly, P requirements are lower for perennial forages than for annual crops. In Canada, grasses required 112–224 kg P₂O₅ ha⁻¹ every 6 + years and alfalfa required 56 kg P₂O₅ ha⁻¹ per year, whereas annual crops required 32–61 kg P₂O₅ ha⁻¹ annually (AAFRD 2001, 2005; AARD 2013). In the USA, annual grain crops require 10–95 kg P₂O₅ ha⁻¹ annually (Cornell College of Agriculture and Life Sciences 2022), a wider but overlapping range compared to the Canadian values.

Application rates of ammonium-based fertilizers and urea were higher in the East compared to the West. This practice, coupled with the lower pH of soils in Eastern Canada, explains why this region is responsible for 80% of lime application in Canada (Dorff and Beaulieu 2014). The liming of acidic soils positively affects crop yield, including the yield and quality of grazed and conserved forages (Holland et al. 2018; Malhi et al. 1995). Application of S-based fertilizers can increase soil acidity, with application rates being similar between East and West. In addition to improving soil fertility, liming may also enhance SOC sequestration due to increased crop yields and return of residues to the soil (Paradelo et al. 2015). Furthermore, liming can also reduce GHG emissions through decreased N₂O emissions and increased CH₄ oxidation (Barton et al. 2013). However, land that is limed to grow perennial forage can act as either a net sink or source of CO₂ (Hénault et al. 2019; Kunhikrishnan et al. 2016). Liming has also been shown to decrease the uptake of toxic heavy metals by plants and to have positive effects on the biodiversity of soil invertebrate and microbial communities (Holland et al. 2018).

Herbicides

Glyphosate was the herbicide most commonly applied to perennial forage land in Western Canada, with annual application rates ranging from 0.260 kg a.i. ha⁻¹ in BC to 0.445 kg a.i. ha⁻¹ in MB, equivalent to 0.069 kg a.i. to 0.122 kg a.i. t⁻¹ DM, respectively. After considering forage yields, SK had the highest per t (DM) application rate of 0.151 kg a.i. (Fig. 4). In Eastern Canada, 2,4-D amine and MCPA were more frequently used in perennial forage production, with per t DM application rates higher in QC (0.082 kg t⁻¹ for 2,4-D amine and 0.004 kg t⁻¹ for MCPA) than in ON (0.065 kg t⁻¹ for 2,4-D amine and 0.003 kg t⁻¹ for MCPA). Glyphosate application reported by the Province of Manitoba (2019) was approx. 4 kg a.i. ha⁻¹ during forage establishment, resulting in approx. 0.52 kg a.i. ha⁻¹ or 0.14 kg t⁻¹ DM when averaged over a stand sustained for 8 years. Our reported herbicide application rates are lower than the rate of 2 kg ha⁻¹ year⁻¹ reported by Bacenetti et al. (2018) for alfalfa production in Italy, although this value is not averaged across all perennial forage land (i.e., including that proportion of forage that does not receive herbicide). Although in our study we only considered glyphosate for Western Canada and 2,4-D amine and MCPA for the East, these herbicides are also used in other regions (Health Canada 2022).

Herbicide application can have detrimental consequences for environmental and human health. For example, glyphosate can persist in soils, plant, and animal tissues and be transported via leaching and runoff to ground and surface waters (Kanissery et al. 2019; Van Bruggen et al. 2017). MCPA and 2,4-D amine are active ingredients in herbicides that control broadleaf weeds and can have negative effects on non-target plant species (Health Canada 2022; Morton et al. 2019; Tayeb et al. 2011). Increases in the area of land treated with herbicides from 1981 to 2011 are likely attributable to a shift from conventional to conservation tillage in the prairies and the expansion of glyphosate-tolerant crops (Gagnon et al. 2016), as well as a shift from pasture and forage production to annual cropping in Eastern Canada, also resulting in increased herbicide use (Clearwater et al. 2016). As herbicide use is primarily associated with annual cropland, environmental effects from perennial forages would be minimal as they account for < 1% of total herbicide use.

3.2.2 Other inputs—irrigation and energy/fuel consumption

Irrigation water consumption

The volume of irrigation water utilized per t of harvested forage (DM) was estimated as 90, 29, and 8.7 m³ t⁻¹ DM for BC, AB, and SK, respectively (Table 5). The much higher rate for BC is related to the fact that it accounted for 28% of total irrigated hay hectares, even though it possessed only 5.9% of total perennial forage area (2009–2018 average). When 28% of total irrigation water used for perennial forages was averaged across this relatively small area of harvested perennial forage, this high per ha (and thus per t) blue water utilization estimate was generated. In contrast, AB and SK generated a lower estimate as these regions have a smaller area of harvested irrigated forage relative to the total area of harvested forage. Furthermore, in BC, the percentage of perennial forages exported from the province was smaller (2%) than that of AB (32%) or SK (15%). Thus, a larger proportion of the total irrigation water withdrawn in BC “remained” in the province to be distributed among the total harvested perennial forage area. Although these estimates may obscure the amount of irrigation water utilized per hectare of irrigated forage, such estimates would still differ based on a number of factors including the proportion of forage-producing operations in each province using systems with different irrigation efficiencies. Furthermore, as irrigated forage in BC would typically yield six cuts a year vs. just two for AB and SK (S. Bittman, AAFC, pers. comm., 2022), average irrigation water use per t harvested forage would be significantly lower for BC, if total annual forage yield was considered.

Table 5 Irrigation water withdrawal per ha of harvested perennial forage and irrigation water utilized per ha and per t of harvested perennial forage, by province

In Canada, Legesse et al. (2018) estimated water use intensity (WUI, m³ of blue water used per t of forage) for grass or grass-legume hay in 2011 to range from 2 to 56 m³ t⁻¹, while Pfister et al. (2011) estimated blue WUI for alfalfa and mixed grass in Canada to range from 29 to 81 m³ t⁻¹. In comparison, Pfister et al. (2011) proposed Canadian WUI estimates of 328, 234, and 123 m³ t⁻¹ DM for wheat, barley, and corn, respectively. Across Canada, forage crops accounted for about 34% of total irrigation water in 2016, rising to 46% in 2018 (Statistics Canada 2017, 2019h). Although forage crops account for a significant proportion of irrigated land in Canada, Legesse et al. (2018) estimated that the blue WUI of grass hay and stockpiled perennial forages (2 m³ t⁻¹ DM) was lower in 2011 than many of the annual crops grown as feed for beef cattle. However, in their study, blue WUI estimates for grass-legume hay were similar to those of many annual feed crops, at 56 m³ t^-1 DM. Bacenetti et al. (2018) reported that irrigation is not typically used to produce Italian alfalfa. Neal et al. (2009) grew perennial forages in Australia under optimum and deficit irrigation; however, they did not report the WUI as irrigation was employed on the basis of soil moisture measurements.

Energy and machinery fuel use

On average across Canada, diesel consumption for agricultural machinery was 6.3 L t⁻¹ harvested forage (DM), electricity consumption for irrigation was 1.0 kWh t⁻¹ DM, and natural gas consumption for irrigation was 7.2 MJ t⁻¹ DM (Table 6). Differences in energy requirements between provinces reflect differences in local growing conditions and management practices. Perennial forage production in MB, ON, and QC did not use any electricity or natural gas, as irrigation is infrequently used in these provinces.

Table 6 Energy use per t (DM) of harvested perennial forage, by province

Manitoba (Province of Manitoba 2019) reported 21 L diesel ha⁻¹ year⁻¹ were used for the production of perennial forages, similar to our value of 22 L ha⁻¹ year⁻¹. Wiens et al. (2014) estimated diesel consumption in MB of 36 L ha⁻¹ forage land year⁻¹. For ON, OMAFRA (2019) reported 19 L diesel ha⁻¹ forage land year⁻¹, compared with our estimate of 34 L ha⁻¹ forage land year⁻¹. Beauregard (2015) estimated 27 L ha⁻¹ forage land year⁻¹ for QC and Little et al. (2013) estimated 21 L ha⁻¹ forage land year⁻¹ for Canada. These estimates are close to our national average of 23 L ha⁻¹ year⁻¹, but lower than the 37.5 L fuel ha⁻¹ reported by Dalgaard et al. (2001) for forage production in Denmark.

Fuel use reported by OMAFRA (2019) omitted bale wrapping and fertilizer application, which may explain the lower estimated fuel use in ON. Furthermore, soil type can affect the degree of soil compaction and thus fuel consumption (Zabrodskyi et al. 2021), a factor that was not considered in our estimates or in the F4E2 model. Compared to annual cropping systems, machinery fuel consumption for harvested perennial forages is typically lower as perennial crops do not require annual tillage or seeding. Irrigation-induced energy consumption for perennial forage production in this study (based on an AB irrigation system energy trial reported by Dill (2010)) was compared to the Ecoinvent model v.3.5 life cycle inventory database, which contains estimates for the average energy use for the application of irrigation water in the USA (0.00315 kg diesel and 0.239 kWh electricity per m³ irrigation water). These estimates were converted to t⁻¹ harvested forage (DM) based on the irrigation volume consumed per hectare for BC, AB, and SK based on the 10-year average perennial forage yields for these provinces. This yielded estimates of 21.6, 7.0, and 2.1 kWh t⁻¹ DM for BC, AB, and SK, respectively, which are significantly higher than the 5.8, 1.9, and 0.6 kWh t⁻¹ DM estimated in the present study. Differences in these estimates could be explained by variation in the efficiency of the different irrigation systems and differences in the use of irrigation per unit area or per unit weight of forage.

Although direct energy CO₂ (i.e., CO₂ emissions arising from energy use for on-farm heating, ventilation, etc.) accounts for only 1.6% of total GHG emissions from North American cattle production systems (FAO 2017), any reduction in these emissions through perennial vs. annual cropping systems could be seen as a climate change mitigation measure.

3.3 Inventory emissions

3.3.1 Field emissions to air, water, and soil

On average across Canada, 0.31 kg, 57.95 kg, 0.41 kg, and 1.85 kg of N₂O, CO₂, NO_x, and NH₃, respectively, were emitted per t (DM) of harvested forage. Emissions were significantly higher in Eastern Canada (particularly QC) with estimates of 0.56 kg, 68.67 kg, 0.63 kg, and 2.47 kg of N₂O, CO₂, NO_x, and NH₃, respectively, compared to 0.19 kg, 52.42 kg, 0.29 kg, and 1.54 kg t⁻¹, respectively, in Western Canada (Fig. 5). In comparison, Fathollahi et al. (2018) reported emissions of 0.19 kg N₂O, 1.75 kg CO₂, 0.04 kg NO_x, and 1.50 kg NH₃ per t of alfalfa hay produced in Iran. Differences in estimates from the two sources could be attributable to variation in climate, management practices, or the parameters considered.

Fig. 5

Nitrous oxide (N₂O), carbon dioxide (CO₂), nitrous oxides (NO_x), and ammonia (NH₃) emissions and losses of phosphate, phosphorous, and nitrate per t of harvested forage (kg t⁻¹ DM), by province

Perennial vegetation can reduce surface runoff and erosion, increase water infiltration, and reduce the flow of sediment and nutrients that can negatively impact surface water quality. At the national scale, 28.22 kg NO₃ and 0.058 kg PO₄ were lost to groundwater, and 0.125 kg of PO₄ t⁻¹ harvested forage (DM) flowed into rivers (Fig. 5). This can be compared to 9.80 kg NO₃ t⁻¹ and 0.13 kg P t⁻¹ lost to water for alfalfa hay produced in Iran (Fathollahi et al. 2018). Nitrate losses were higher in Eastern Canada with 41.7 kg t⁻¹ compared to 21.26 kg t⁻¹ in Western Canada (Fig. 5). Although our estimates are higher than those of Fathollahi et al. (2018), they considered NO₃ losses to water only from organic and mineral fertilizers, while ours also included leaching losses from soil N mineralization. From our analysis, NO₃ losses from just manure and fertilizer equate to average loss rates of 10.3 kg NO₃ t⁻¹ harvested forage (DM) for Western Canada and 24.2 kg NO₃ t⁻¹ for Eastern Canada, with our West estimate being similar to that of Fathollahi et al. (2018). In comparison, Drury et al. (2016) reported average N loss rates in leaching and runoff across all agricultural land in 2011 of between 0.4 (in AB) and 31.7 kg N ha⁻¹ (in QC). Our estimates are much higher when expressed on a kg NO₃-N ha⁻¹ basis, ranging from 10.7 (SK) to 62.5 kg NO₃-N ha⁻¹ (QC). However, our estimates included NO₃-N losses from soil N mineralization as well as from manure and synthetic fertilizer application and atmospheric N deposition. In contrast, Drury et al. (2016) considered only additions of fertilizer and manure N, fixation of N by leguminous plants, and atmospheric dry and wet deposition of N. If we consider only manure and fertilizer N and atmospheric N deposition, our estimates are between 4.5 (SK) and 41.0 kg NO₃-N ha⁻¹ (QC).

Phosphate losses to ground and surface water per t DM were higher in the West (0.066 kg t⁻¹ to groundwater and 0.141 kg t⁻¹ to surface water) than in the East (0.043 kg t⁻¹ to groundwater and 0.093 kg t⁻¹ to surface water). These higher loss rates per t in the West can be attributed to the generally lower yields in these provinces, leading to higher leaching and runoff rates on a per t harvested DM basis.

The generally higher emission rates in Eastern Canada can be largely explained by climatic differences and higher fertilizer application rates. In the semi-arid prairies, N₂O emissions are typically lower than those in higher moisture soils in the East (Worth et al. 2016), although losses can also be driven by temperature (Butterbach-Bahl et al. 2013; Griffis et al. 2017). Higher ambient temperatures also result in increased NH₃ losses from manure and fertilizers (Robertson et al. 2013; Skjøth and Geels 2013), with average spring temperatures being higher in Eastern (4–6 °C) than Western (2–6 °C) Canada (ECCC 2021a). As most synthetic fertilizer and manure is land-applied in the spring, these temperature differences significantly impact overall emissions (Sheppard et al. 2010a, 2015).

Compared to Eastern Canada and BC, the prairie provinces are generally less susceptible to nutrient leaching and runoff from agricultural fields due to lower precipitation (Cade-Menun et al. 2013; Roussin and Binyamin 2018; Zhang et al. 2019). Thus, < 1% of farmland in the prairies is thought to pose a high or very high risk of contaminating ground or surface water with N, as compared to 41% and 75% in ON and QC, respectively (Drury et al. 2016). Likewise, in 2011, approx. 2% of farmland in AB, SK, and MB was proposed to be at high or very high risk of contaminating surface or ground water with P as compared to 11% in ON and 53% in QC (Reid et al. 2016). Although the amount of P in surface water was found to be positively correlated with precipitation across Canada (Liu et al. 2021), the effect of perennial cropping as compared to annual cropping on nutrients in surface runoff from agricultural fields is strongly driven by nutrient and crop residue management (Liu et al. 2014). For example, Liu et al. (2014) reported that, for a southern MB watershed, the release of nutrients from perennial plant residues during freezing may result in greater P losses in surface snowmelt as compared to annuals.

3.3.2 Infrastructure and the production of agricultural waste

Field operations

On average across Canada, per t harvested perennial forage DM, 15 g of PP and HDPE plastic, 21 g of cardboard, and 22 g of waste paper were produced (Table 7). On a per t basis, the amount of waste produced was lower for the East due to higher yields of harvested forage than in the West. As it was not possible to completely delineate these waste products from annual crops vs. perennial forages, it is likely that the production of waste products from conserved perennial forage is overestimated as pesticides are more commonly used in annual cropping systems.

Table 7 Quantity of agricultural waste produced and materials used per t (DM) of harvested forage by province during the field and storage stages

Dried hay and haylage/silage storage

The amount of plastic waste generated from bale-wrapping and plastic wrapping for haylage/silage ranged from 0.33 (AB) to 1.45 kg (QC) LDPE t⁻¹ harvested forage (DM) based on CleanFarms data, and from 0.23 (AB) to 1.34 kg (ON) LDPE t⁻¹ based on Groupe AGÉCO data (Table 7). Using the average estimates from the two data sources, LDPE usage for plastic-wrapping was highest in the East (1.20–1.37 kg LDPE t⁻¹ harvested forage) compared with the West (0.28–0.33 kg LDPE t⁻¹), as more bales are plastic-wrapped and a higher proportion of harvested forages are ensiled as opposed to sun-dried as is the case in the West. In addition to LDPE used for wrapping, silo covers contributed a further 0.001–0.0047 kg LDPE t⁻¹, with the highest usage rates in the East. Plastic waste can have a large negative impact on the environment due to its longevity and slow decomposition, as well as the energy and petroleum resources used for its production. Most agricultural plastics do not readily biodegrade, and therefore if not recycled, they are incinerated or landfilled, producing GHG emissions. The emissions throughout the life cycle of LDPE contribute to climate change, acidification, eutrophication, ozone depletion, and the formation of photochemical oxidants.

Silos in Canada required between 0.01 and 0.11 kg concrete and between 0.01 and 0.12 kg of steel t⁻¹ harvested forage (DM), with values in the East generally higher than in the West (Table 7). The production of cement for forage storage structures leads to additional energy consumption and CO₂ emissions (Müller et al. 2014).

3.3.3 Land occupation/transformation and effect on soil organic carbon change

The area of harvested forages declined in all provinces from 2009 to 2018. Across the six provinces included in the LCI, the total area declined from 6.43 M ha in 2009 to 5.23 M ha in 2018, a decrease of 18.7%. Across the individual provinces, rates of decline varied from 10.7% in AB to 29.2% in ON (Fig. 6). Between 1980 and 2004, the area of harvested perennial forage in AB increased from 1.42 to 2.39 M ha, and in SK from 0.77 to 1.77 M ha (Statistics Canada 2019a). An increase in the cattle population between 1981 and 2006 in AB from 4.2 to 6.4 million head and from 2.4 to 3.4 million head in SK likely increased forage demand (Statistics Canada 2022b). Financial challenges in the annual crop sector also caused a shift to lower cost perennial crops in many provinces during this time (Statistics Canada 2014). Harvested forage area in BC, MB, ON, and QC either showed smaller increases or remained relatively stable from 1980 to the mid-2000s, after which all provinces experienced a decline. Between 2006 and 2016, the price index of grain crops in Canada rose by 66% and that of oilseed crops by almost 80% (Statistics Canada 2020c). Over this same period, the production (t) of perennial forages fell in most provinces, by between 11.1 (AB) and 39.2% (ON), although it rose by 21.7% in BC (Statistics Canada 2019a). In ON, the largest conversion from perennial to annual crops occurred in the Lake Huron watershed between 2011 and 2015, where short-term perennial vegetative cover (pasture/forages, fruits, etc.) declined by 22% (Corry 2017).

Fig. 6

a) Annual area of harvested perennial forage (ha) in BC, AB, SK, MB, ON, and QC from 2009 to 2018 (Statistics Canada 2019a); b) Average annual C sequestered by soils (t C t⁻¹ DM) under forage land by province for the 2009–2018 period

Perennial vegetation systems can play an important role in climate mitigation via the sequestration of SOC. Forage lands that did not undergo conversion to annual crops continue to accumulate SOC (Fig. 6). Others have shown that the conversion of annual cropland to perennial forages increases the accumulation of soil C (Asgedom and Kebreab 2011; Hutchinson et al. 2007; VandenBygaart et al. 2010). Although the SOC change coefficients used in our study accounted for conversion of annual to perennial crops, they did not consider the effects of different management practices on the rate of SOC accumulation. These practices could include changes in fertilization or irrigation practices, or the potential effects of climate change on forage yield or inputs or losses of C from the SOC pool. The conversion of perennial to annual crops since 2006 would have contributed to a decline in net C sequestration and a decline in SOC stocks (ECCC 2022). However, while perennial forage production systems can be beneficial in terms of sequestering C in their soils and hay frequently provides a complete ration for Canadian cows, there are trade-offs as cattle fed a diet rich in forages have higher enteric methane emission rates compared with cattle fed a grain-rich ration (Beauchemin et al. 2010, 2011; Legesse et al. 2016).

Annual rates of C sequestered per ha of harvested perennial forage varied from 0.17 t C in BC to 0.25 t C in ON and QC, with intermediate rates of 0.205 t and 0.2 t ha⁻¹ for AB/SK and MB, respectively. Expressed per t of harvested forage (DM), the prairie provinces had higher sequestration rates, with the highest in SK (0.075 t C t⁻¹ harvested forage (DM)) and lower rates of approx. 0.05 t C t⁻¹ in ON/QC and 0.045 t C t⁻¹ in BC. These trends can be attributed to the fact that although the per hectare C sequestration rates are lowest in the prairies, the rates per t of harvested forage are higher due to the relatively low yields in these provinces. In comparison, Su et al. (2009) reported a C sequestration rate of 0.4 t C ha⁻¹ year⁻¹ for perennial forage in China. Enhanced plant growth as a result of adequate precipitation increases the accumulation of SOC (Badgery et al. 2013; Dlamini et al. 2016; Follett et al. 2012). In contrast, higher temperatures can result in increased soil microbial decomposition and a loss of SOC (Dlamini et al. 2016; Zhao et al. 2017). Soil texture and pH also affect SOC gains and losses, as soils with a higher clay content are generally reported to have greater SOC content (Follett et al. 2012; Jobbágy and Jackson 2000). Furthermore, soils with a higher pH can have higher SOC accumulation rates, as a result of reduced microbial decomposition of soil organic matter (Zhao et al. 2017). There are a wide variety of soil types throughout Canada, which contributes to variability in SOC accumulation. Although growing season average temperatures in southern ON (1981–2010) were generally higher than in the prairies, the higher yields facilitated by greater precipitation (ECCC 2021a, b) and a subsequent higher input of aboveground C in residues and belowground C in roots, could outweigh any increase in C losses from the soil via decomposition. This could result in overall higher rates of SOC accumulation per ha in Eastern vs. Western Canada.

3.4 Scenario analysis

3.4.1 Perennial forage yield

At the national scale, the estimated average forage yields for 2004–2013 and 2011–2015 were 1.7% and 0.2% lower, respectively, than 2009–2018 (Table 8). For 1994–2003 and 1999–2008, yields were significantly lower than for 2009–2018, particularly in the prairie provinces where yields were 22.4–29.0% and 14.0–20.6% lower, respectively. The lower yields for these two 10-year periods were attributed to severe drought in many parts of the prairies in 1999–2001. In Eastern Canada, yields were only 1.7 to 6.7% lower for 1999–2008, 2004–2013, and 2011–2015, compared to 2009–2018.

Table 8 Average provincial perennial forage yields (kg ha⁻¹) for the 1994–2003, 1999–2008, 2004–2013, and 2011–2015 periods, compared with 2009–2018 10-year average yields

As a result of lower yields, particularly in the prairies, more land was required to produce each t of harvested perennial forage DM, which led to increases in the amount of C sequestered in these soils as a result of this additional land use. For example, SOC sequestered per t harvested DM was 121% and 134% that of the 2009–2018 estimate for the West in 1999–2008 and 1994–2003, respectively. In Eastern Canada, small changes in yield resulted in only a slight increase in the amount of land required per t forage and consequently increases in SOC were modest. During low-yield periods, more land area in the West was required to produce 1 t of harvested forage, resulting in higher emissions to air and water and losses of nutrients and pollutants to soil (Figs. 7, 8, 9). This analysis demonstrates that the flows into and out of the perennial forage production system as measured by the LCI can differ depending on the timeframe under examination, with changes in average forage yield for these different timeframes being the major driver of these differences. This highlights the care needed when interpreting the LCI outputs on the basis of yield, as this could be significantly higher or lower for the study period compared with a longer timeframe where growing conditions would be more variable (e.g., due to severe drought).

Fig. 7

Life cycle inventory parameters per t (DM) of harvested perennial forage for Canada calculated based on average yield (DM) values for 2009–2018, 1994–2003, 1999–2008, 2004–2013, and 2011–2015. LDPE, low-density polyethylene; HDPE, high-density polyethylene; PP, polypropylene. These charts depict, for each LCI parameter, the outputs of the LCI based on average yields for the 1994–2003, 1999–2008, 2004–2013, and 2011–2015 timeframes as a percentage of the LCI outputs based on average yields for the 2009–2018 timeframe

Fig. 8

Life cycle inventory parameters per t (DM) of harvested perennial forage for Western Canada calculated based on average yield (DM) values for 2009–2018, 1994–2003, 1999–2008, 2004–2013, and 2011–2015. LDPE, low-density polyethylene; HDPE, high-density polyethylene; PP, polypropylene. These charts depict, for each LCI parameter, the outputs of the LCI based on average yields for the 1994–2003, 1999–2008, 2004–2013, and 2011–2015 timeframes as a percentage of the LCI outputs based on average yields for the 2009–2018 timeframe

Fig. 9

Life cycle inventory parameters per t (DM) of harvested perennial forage for Eastern Canada calculated based on average yield (DM) values for 2009–2018, 1994–2003, 1999–2008, 2004–2013, and 2011–2015. LDPE, low-density polyethylene; HDPE, high-density polyethylene; PP, polypropylene. These charts depict, for each LCI parameter, the outputs of the LCI based on average yields for the 1994–2003, 1999–2008, 2004–2013, and 2011–2015 timeframes as a percentage of the LCI outputs based on average yields for the 2009–2018 timeframe

3.4.2 Soil organic carbon change

Re-calculating the amount of SOC sequestered per t of harvested forage DM using different annual SOC change coefficients significantly affected the LCI results. The coefficients were derived from the Century model-based estimates of SOC accumulation over the first 20 years after land was converted from annual crops to perennial forage (Table 9). These coefficients assume a higher annual rate of SOC gain over a shorter time period. In other work, VandenBygaart et al. (2008) extracted field-based estimates from the literature of SOC change in systems shifting from low C input associated with frequent tillage to managed grassland. Applying these estimates, SOC sequestration ranged from 0.56 to 0.72 t SOC ha⁻¹ year⁻¹ for the prairie provinces and 0.6 to 1.07 t SOC ha⁻¹ year⁻¹ for the Eastern provinces (Table 9). VandenBygaart et al. (2010) used a single SOC change coefficient derived from Elora, ON (0.32 t ha⁻¹ year⁻¹) for Eastern Canada, which was lower than the 20-year coefficients from ECCC (2022) and those employed earlier by their laboratory (VandenBygaart et al. 2008). For the prairies, data were derived from four locations in AB ranging from 0.23 to 0.93 t SOC ha⁻¹ year⁻¹, with an average value of 0.6 t SOC ha⁻¹ year⁻¹ applied to AB and SK, and an estimate of 0.76 t SOC ha⁻¹ year⁻¹ derived from the Parkland region applied to MB (Table 9). Despite the variation in these values, there is a general consensus that SOC accumulation increases in perennial forage systems as compared with annual cropping systems. Overall, VandenBygaart et al. (2010) estimated that the effect of replacing annual crops with perennial cover would result in an average SOC accumulation rate of 0.53 t C ha⁻¹ year⁻¹ (9.0 ± 1.5 t C ha⁻¹ over 16.9 years) at the 0–30 cm soil depth.

Table 9 Soil organic carbon change coefficients used in the SOC scenario analysis

Additional field-based estimates of SOC change over time for annual to perennial conversion or for longer term perennial forage stands were scarce, particularly for the 0–30 cm soil depth or greater. Su et al. (2009) reported a 20.2% increase in SOC 6 years after cropland was re-established to perennial alfalfa in China. They estimated a C sequestration rate of 0.4 t C ha⁻¹ year⁻¹, a value similar to that derived from other 20-year SOC change coefficients for Western Canada (Table 9). Ellert and Bettany (1995) also reported that pasture systems stored more SOC (52.4 ± 2.5 t C ha⁻¹) than annual cropland (46.2 ± 0.9 t C ha⁻¹). The greater capacity of perennial systems to accumulate SOC can be attributed largely to the greater root biomass and distribution in perennial vs. annual crops (DuPont et al. 2010).

3.5 Sources of uncertainty

Life cycle inventory data has associated uncertainty that must be considered (Bamber et al. 2020). The main data sources used in this LCI and associated estimates of uncertainty are provided in Table 10. For example, the fertilizer application data sourced from Manitoba Agricultural Services Corporation (2019) and Stratus Ag Research (2018) were based on data from over 300 and 200 farms, respectively, with a broad range of fertilizer application rates. All data sourced from Statistics Canada reported ranges of coefficients of variance rather than individual values. The coefficients of variance for the Liang et al. (2020) methodology for calculating N₂O ranged from 32 to 128%, and the NO_x emissions from EEA (2016) had a 95% confidence interval of ± 100 to 300%, although this was a generic estimate based on data quality. Overall, although the most up-to-date input data and models were employed in this LCI, using Canadian data and approaches where possible, there was considerable uncertainty associated with both input and emissions data, based on variability in farm practices and modeling emissions under different conditions. This inherent variation demonstrates the value of regionalized LCI and LCA results conducted by province and region, recognizing that aggregating over different regions can obscure some of these variables and increase the uncertainty of the overall model.

Table 10 Sources of data used in the LCI and their associated uncertainty estimates

4 Summary and conclusions

Canada’s forage sector plays a significant role in sustaining Canada’s livestock population and in contributing to the Canadian economy. While the production of harvested perennial forages leads to the consumption of natural resources, these rates of consumption were generally lower than those required by annual cropping systems, given the reduced use of irrigation, synthetic fertilizers, and energy. Consequently, and also as a result of other properties of perennial crops (such as their greater root biomass and distribution and the N-fixing capacity of legumes), these systems can lead to lower levels of pollutant emissions to the atmosphere and water. For instance, the inclusion of N-fixing legumes in crop rotations reduces energy-related emissions and the need for commercial N fertilizers, limiting NO₃-N leaching and N₂O emissions. Although the total area of these lands declined in all provinces between 2009 and 2018, those that were not converted to other land uses continued to retain and sequester SOC at a higher rate than annual cropping systems. However, estimates of the total amount of SOC accumulated over this period in each province depended on the annual SOC change coefficients employed, emphasizing the need to improve the spatial accuracy of these estimates.

Inventory outputs varied depending on the province or region, with some inputs higher in the West than in the East. For example, irrigation water utilization and associated energy use were only relevant to BC, AB, and SK, while losses of nutrients and pollutants to the air and water were typically higher in the East due to greater fertilizer use and higher soil moisture levels. However, LCI outputs per unit of harvested forage depended on the yield estimates employed, and this study found that using yields associated with time periods that experience drought led to an overall increase in the LCI outputs associated with forage production.

This study also highlighted data gaps and limitations that need to be addressed, particularly in relation to the paucity of relevant data for specific provinces. For example, waste generation at the field stage of forage production was only available for AB, and for AB and QC at the storage stage. Fertilizer use data for forages in MB had to be adapted to AB and SK due to the lack of data for these provinces, and information on the use of herbicides across all provinces was limited. Addressing these data gaps and updating the LCI models would generate a more accurate assessment of the resource use and emissions associated with these production systems.

Life cycle inventory and assessment offer insights regarding strategies that may be considered by stakeholders and policy makers to improve the environmental and economic performance of a product or service. The outcomes of this LCI provide an opportunity to examine the environmental impacts of forage production systems and generate recommendations for improvements in resource use, as well as education and marketing associated with the sustainability of Canadian forage products.

Appendices

Appendix 1

Manitoba nutrient application rates for synthetic fertilizer were applied to BC, AB, and SK as follows:

$$\begin{aligned}{NutrientApp\_rate}_{province}= & \;{NutrientApp\_rate}_{MB}\\&*{ForageAreaFert}_{province}\\&/{ForageAreaFert}_{MB}\end{aligned}$$

(1)

where:

NutrientApp_rate_province:

Provincial nutrient application rate of synthetic fertilizer (kg ha⁻¹)

NutrientApp_rate_MB:

Nutrient application rate for MB (kg ha⁻¹) 25, 37, 21, and 9 kg ha⁻¹ for N, P₂O₅, K₂O, and S, respectively

ForageAreaFert_province:

Forage area in province receiving commercial fertilizer (%): BC (35%), AB (31%), SK (12%) (Statistics Canada 2018)

ForageAreaFert_MB:

Forage area in MB receiving commercial fertilizer (35%) (Statistics Canada 2018)

$$\begin{aligned}{Lime\_rate}_{province}= & \;{Lime\_rate}_{perennials}\\&*{Area\_limed}_{province}*{Perennials}_{established}\end{aligned}$$

(2)

where:

Lime_rate_province:

Annual lime fertilization rate (kg lime ha⁻¹)

Lime_rate_perennials:

Lime fertilization rate (2500 kg ha⁻¹, from Beauregard (2015))

Area_limed_province:

Average % of cropland and tame/seeded pasture area in the province receiving lime = 1.3, 0.1, 0.3, 0.5, 3.2, and 7.7% for BC, AB, SK, MB, ON, and QC, respectively, estimated as the total area of land receiving lime (average of 2011 and 2016 data) (Statistics Canada 2019i)/total area of land used for crops (excl. Christmas tree area) and tame/seeded pasture (average of 2011 and 2016 data (Statistics Canada 2019j))

Perennials_established:

% perennial forage fields in the establishment phase each year. An adjustment factor was applied based on crop duration data (Statistics Canada 2018) to take into account the fact that lime is applied during establishment of perennial forages; % hay fields in the establishment phase each year = 14, 16, 13, 16, 26, and 26% for BC, AB, SK, MB, ON, and QC, respectively

$${CO}_{2Liming}=\left({M}_{Limestone}\times {EF}_{Limestone}\right)\times \frac{44}{12}$$

(3)

where:

CO_2Liming:

Annual lime fertilization rate (kg lime ha⁻¹)

M_Limestone:

Annual amount of calcic limestone (kg CaCO₃ year⁻¹)

EF_Limestone:

Emission factor (kg of C kg⁻¹ limestone); the default emission factor of 0.12 for limestone from IPCC (2006) was used

44/12:

C to CO₂ conversion factor

Phosphate leaching was estimated as the average quantity of P leached to groundwater for permanent pasture (which was assumed to be representative of perennial forage) according to the equation:

$${P}_{gw}={P}_{gwl}\times {F}_{gw}$$

(4)

where:

P_gw:

Quantity of P leached to groundwater (kg ha⁻¹)

P_gwl:

Average quantity of P leached to groundwater (kg ha⁻¹), which was 0.06 kg P ha⁻¹ for permanent pastures and meadows

F_gw:

Correction factor for fertilization by slurry, this was equal to 1 in the present analysis as it was assumed that no sewage sludge was used (Nemecek and Kagi 2007)

As a proxy for harvested perennial forage, PO₄ runoff was estimated as the average quantity of P lost via runoff for extensive permanent pastures and meadows multiplied by a correction factor that accounted for differences between mineral fertilizers and manure according to the equation:

$${P}_{ro}={P}_{rol}\times {F}_{ro}$$

(5)

$${F}_{ro}=1+\frac{0.2}{80}\times {P}_{2}{O}_{5min}\times \frac{0.7}{80}\times {P}_{2}{O}_{5man}$$

(6)

where:

P_ro:

Quantity of P lost through runoff to rivers (kg ha⁻¹)

P_rol:

Average quantity of P lost through runoff (kg ha⁻¹), which is 0.25 kg P ha⁻¹ for intensive permanent pastures and meadows and 0.15 kg ha⁻¹ for extensive permanent pastures and meadows

F_ro:

Correction factor for fertilization with P

P₂O_5min:

Quantity of P₂O₅ contained in mineral fertilizers (kg ha⁻¹)

P₂O_5man:

Quantity of P₂O₅ contained in solid manure (kg ha⁻¹) (Nemecek and Kagi 2007)

Emissions of P to rivers due to the erosion of soil particles were estimated according to the equation:

$${P}_{er}=10000\times {S}_{er}\times {P}_{es}\times {F}_{r}\times {F}_{erw}$$

(7)

where:

P_er:

The quantity of P emitted through erosion to rivers (kg P ha⁻¹)

S_er:

Quantity of soil eroded (kg ha⁻¹)

P_er:

P content in the top soil (kg P kg⁻¹ soil), the average value of 0.00095 kg kg⁻¹ was used for all provinces included in the analysis

F_r:

Enrichment factor for P, the average value of 1.86 was used—this factor takes into account the fact that eroded soil particles contain more P than average soil particles

F_erw:

Fraction of the eroded soil that reaches the river, the average value of 0.2 from the Ecoinvent report was used (Nemecek and Kagi 2007)

Appendix 2

Table 11 provides a breakdown of the steps to estimate irrigation water utilization for harvested perennial forage production in BC, AB, and SK.

Table 11 Breakdown of the steps taken to estimate irrigation water utilization per ha of harvested perennial forage (excluding exports) for British Columbia (BC), Alberta (AB), and Saskatchewan (SK)