This study used ISO-compliant life cycle assessment to compare the cradle-to-farm gate cumulative GHG emissions associated with four beef production strategies. The spring calving herd consisting of approximately 350 cows and related feedlot operations at the Agriculture and Agri-Food Canada, Lacombe Research Centre (Lacombe, AB, Canada) were used for the collection of diet ingredient and nutrient composition, feed intake, and cropping inputs and outputs from the four beef production strategies. Beef production practices in this herd are typical for western Canada and are described in detail by Basarab et al. [23,24]. The herd consists of Hereford-Aberdeen Angus and Charolais-Red Angus crossbred cows which are also common breeds in western Canada [25]. The cattle used for the GHG assessment included cows, breeding bulls, replacement heifers, replacement bulls, calves from birth to weaning, and feeders from weaning to slaughter. All animals were maintained and cared for according to the guidelines of the Canadian Council on Animal Care [26].

Sources of GHG included on-farm emissions of CH₄ from enteric fermentation and manure, on-farm emissions of N₂O from manure, off-farm emissions of N₂O from N leaching, run-off and volatilization, on-farm emissions of N₂O from cropping due to soils, fertilizer, roots and residue, and CO₂ emissions from energy use. Energy-use direct and indirect CO₂ emissions included: (1) embodied energy in equipment for field operations, baling, hauling, feed processing, feeding, bedding and manure removal; (2) crop inputs such as fertilizer, herbicide and seed; and (3) fuel and lubrication for field operations, baling, hauling, feed processing, feeding, bedding and manure removal. All feeds except for the protein supplement were grown on-farm, thus CO₂ emissions due to transporting feeds was negligible. All gases were expressed as CO₂ equivalents (CO₂e) to account for the global warming potential of each gas compared to CO₂, where CO₂ = 1, CH₄ = 23 and N₂O = 298 [21]. The carbon footprint or GHG intensity of beef was expressed as kg CO₂e kg⁻¹ live weight, CO₂e kg⁻¹ carcass weight, kg CO₂e kg⁻¹ live weight yr⁻¹, CO₂e kg⁻¹ carcass weight yr⁻¹.

The life cycle assessment for cows, bulls, replacement heifers and feeder cattle followed the production cycle of the cow herd and began in late May of each year when the cow-calf pairs were placed on pasture prior to the beginning of the breeding season and ended 365 d later since the objective of cow-calf management is to calve once every year. Two production cycles were followed for all cattle categories. Numbers of animals followed in each cattle category, feeding period and production cycle varied and are presented in Tables A1 and A2. Feed sampling, dry matter disappearance and quality analysis during the summer and fall grazing periods are described by López-Campos et al. [27]. For the remainder of the production cycle, diet ingredient composition and feed delivered to each cattle category were recorded daily by a feed truck (International DT4600 truck, Cattlelac 520 mixer, Weigh Tronic scale with onboard laptop) and later transferred to a daily time-step data base in Excel (Microsoft). Average diet ingredient composition, total digestible nutrients (TDN), crude protein (CP), body weight, ADG and DMI for cows, bulls and replacement heifers by feeding period are given in Table A1.

Feeder steers from the herd described above were used to create four beef production strategies: (1) hormone free cattle harvested at 11–14 mo of age, (2) growth implanted cattle harvested are 11–14 mo of age, (3) hormone free cattle harvested at 19–23 mo of age, and (4) growth implanted cattle harvested at 19–23 mo of age. These four production strategies were created because they represented the majority of youthful cattle, aged 10 to 24 mo of age, processed through Canadian packing plants [19]. Feeders harvested at 11–16 mo of age are referred to as “calf-fed” while those harvested at 17–23 mo of age are referred to as “yearling-fed”, which reflected the age at which the feeders are started on their finishing diet. In each of two years (2008 and 2009), 112 spring-born crossbred steer calves were equally assigned at weaning to two production systems (calf-fed; yearling-fed) and two implant groups (not implanted; implanted) based on breed cross, birth date, calf weight and dam age. In each year half the calf-fed steers (n = 28) were implanted with 200 mg progesterone and 20 mg estradiol benzoate (Component E-S, Elanco-Animal Health A Division of Eli Lilly Canada Inc., Toronto, ON, Canada) at weaning, and re-implanted with 120 mg trenbolone acetate and 24 mg estradiol (Component TE-S, Elanco-Animal Health A Division of Eli Lilly Canada Inc., Toronto, ON, Canada) approximately 90–100 d before slaughter. Similarly, half the yearling-fed steers each year (n = 28) were implanted at weaning and then four more times at 80–90 d intervals with 200 mg progesterone and 20 mg estradiol benzoate, and then with 120 mg trenbolone acetate and 24 mg estradiol 90–100 d before slaughter. A more detailed description of the animal management and experimental design can be obtained from López-Campos et al. [27]. The average diet ingredient and nutrient composition, body weight, ADG and DMI for the feeders by feeding period are given in Table A2. Since no feeder heifers were followed in the present study, it was assumed that the diet ingredient and nutrient composition for heifers was the same as for steers, and heifers had a 5% lower weaning weight, 10% slower post-weaning growth rate and 8% higher feed to gain ratio from weaning to slaughter compared with steers [23,28]. Both calf-fed and yearling-fed steers were targeted to be slaughtered at 8–10 mm of backfat in four groups of 14 per year. In 1–2 weeks intervals steers were trucked 3 km for processing at the Lacombe Research Centre abattoir. The values in Tables A1 and A2 were then used to calculate the daily enteric CH₄ emissions and emissions of CH₄ and N₂O from manure handling, storage and land application for each class of cattle and feeding period based on the IPCC [22] Tier 2 methodology and modified for nitrogen excretion according to NRC [20]. Tables A1 and A2 values were also used to calculate the total amount of feed ingredients delivered to each cattle category (kg DM hd⁻¹ d⁻¹).

The research was located near Lacombe, AB, Canada (52°27'23''N, 113°44'31''W). The actual whole-farm location is the Agriculture and Agri-Food Canada Research Centre farm, which is representative of Ecodistrict 737 in the Parkland ecoregion [29]. The soil is Orthic Black Chernozem with clay and sandy loam textures. The topography is moderately undulating. The long term average precipitation is 450 mm annually and 341 mm within the May to October growing season. The ecodistrict precipitation to potential evapotranspiration ratio (P/PE) is 0.65 for the growing season based on 1970 to 2000 annual weather data normal. Growing season precipitation during 2008, 2009 and 2010 was 311, 243, and 513 mm, respectively. Temperatures both during the growing season and during winter were generally slightly cooler than average (Table A3).

Crop production as required to feed all cattle classes was carried out on a farm scale. None of the land used could be classified as marginal. There was no summer-fallow. All pasture could be defined as cropland pasture in rotation with cereal and oilseed crops and not rangeland. Species varied between years, but field methods and crops used for feed were typical of the region. In 2008 barley (Hordeum vulgare L.) was used as silage and grain as feeds and straw for bedding; oat (Avena sativa L.) grain was used as feed. Summer pastures for cows consisted of perennial stands composed of largely meadow bromegrass (Bromus riparius Rehm.), but smooth bromegrass (Bromus inermis Leyss.) and Kentucky bluegrass (Poa pratensis L.) were also present depending on paddocks. Pastures had originally been sown as a mixture with alfalfa (Medicago sativa L.), but legume content was 20% or less. All of these pastures were over 10 yr-old. Pastures used for weaned calves and yearling stocker animals were planted in 2001 and consisted of meadow bromegrass and alfalfa with alfalfa less than 30% by 2008; Kentucky bluegrass existed in patches. In 2009 feed consisted of barley silage, grain and straw as well as straw for bedding. Pastures consisted of the same species mix, but areas used for grazing cows in summer were not identical.

Barley and oat were managed identically from year to year, although fertilizer inputs varied between the years. Tillage could be described as minimum or reduced as opposed to zero-tillage or conventional tillage. Seeding occurred at recommended rates. In 2008 fertilizer was broadcast prior to seeding at 100 kg N as urea, 30 kg P₂O₅ and 30 kg K₂O ha⁻¹ and in 2009 at 80 kg N and 20 kg P₂O₅ ha⁻¹. All barley and oat crops received 1.236 L ha⁻¹glyphosate [N-(phosphonomethyl) glycine] as a pre-seeding burn-off. In addition crops used for grain production were applied with a mixture of dicamba, [3,6-dicloro-(2-methoxybenzoic acid)] mecoprop {2-[4-chloro-(2-methylphenoxypropoanoic acid)]} and MCPA amine [4-chloro-(2-methylphenoxy) acetic acid] at 1.24 kg ha⁻¹total active ingredient. In addition roller-packing after seeding, swathing, combining and grain hauling activities were taken into account. Straw removal for bedding and feed were assumed to occur on all cereals harvested for grain. Thus, baling and hauling of large round bales occurred after combining. Because barley and oat grain and straw were used for different livestock classes within the four production strategies and in some cases in different enterprises (e.g., could be sold off the farm), emissions (kg CO_{2 equiv} ha⁻¹) were divided between grain and straw in a 0.55 and 0.45 proportion according to crop dry matter distribution of the crop [30]. Barley silage activities included swathing, chopping, hauling and packing for a bunker silo.

Perennial hay, silage and pasture included inputs and operations for establishment. Establishment included seed at recommended rates for a 1:1 mixture of meadow bromegrass and alfalfa and no herbicide application or cover crop. Annual hay production included fertilizer broadcast, hay cutting with a haybine, baling and hauling. Hay and silage were included in the diets in 2008–09 only. Fertilizer inputs were 100 kg N as urea, 30 kg P₂O₅ and 30 kg K₂O ha⁻¹. Energy of all inputs and operations for hay and silage establishment were averaged over eight yrs, the expected life of the stand and 12.5% of energy or emission for establishment year was added to the total of annual hay and silage production. Pastures used for cows, calves and bulls during summer were assumed to have a 20-yr life span. Therefore establishment emissions of 5% of the establishment year total were added to those from annual input and operations. The cow-herd was grazed in two breeding groups of approximately equal size in different grazing cells. Cows were managed as in a controlled rotational grazing method, but rarely returned to any paddock for a second grazing. Carrying capacity averaged between the two herds within years was equivalent to 68 and 101 cow-calf pairs-d ha⁻¹ with bull size and grazing days accounted for. No fertilizer was applied to these pastures in 2008. In 2009 either 42 kg N or 58 kg N ha⁻¹ was applied to pastures for respective groups. Operations for fertilizer spreading, and moving and monitoring cows and calves while on pasture were recorded by time required for laborers and all terrain vehicle used.

Pasture used for weaned calves was similar to that used for yearling stocker animals and assumed to have an 8-yr life span. Therefore establishment emissions calculated as 12.5% of total emissions from the establishment year were added to annual emissions for both of these pasture-types. However, pastures used for weaned calves were used for a short period in each fall and had been used for hay production during the first cut. All calves weaned were co-mingled on the same pastures and managed identically. Therefore fertilizer added to these pastures was assumed to contribute only to the hay enterprise and not to the weaned calves. Carrying capacity for weaned calves was 560 and 550 animal-d ha⁻¹ over 50 and 26 pasture days during 2008 and 2009, respectively. Operations related to monitoring, moving and bedding were included in emissions attributed to these animals. Pastures used for yearlings were managed as controlled rotational grazing using an electrified wire that was moved at least twice per week. Carrying capacity of steers was 115 and 160 animal-d ha⁻¹ with 59 and 72 pasture days in 2009 and 2010, respectively. Fertilizer as 52 kg N ha⁻¹ and 75 kg N ha⁻¹ was broadcast in the spring of 2009 and 2010, respectively.

Average yields (kg DM ha⁻¹) for all crops and feeds were determined as harvested yields of grain, straw or whole-plant forage on a field basis by recording truck-loads with representative fresh and dry weights. Where this was not possible, within field representative quadrat sampling (kg DM m⁻²) was extrapolated. All pastures were monitored using a cage method for available forage and residue yields on a dry matter basis as animals moved from paddock to paddock as was the case for cows and calves, or as electrified wire was moved within paddocks as was the case for weaned calves and yearlings. Duration of measurement period in the case of cow-pasture was no longer than 2 wks or less as cows moved from paddock to paddock more frequently. Concentrations of nitrogen (N) and carbon (C) were determined using a Leco C and N Analyser [31] on sub-sampled dried and ground subsamples.

Sources of CO₂ emission derived from on-farm energy use are summarized in Table A4. All energy used in manufacture and transportation of equipment (embodied), operation and maintenance (fuel and lubrication) was accounted for in all cropping activities for each crop species and cropping or pasture enterprise. Energy (MJ ha⁻¹) was converted to diesel fuel equivalent (L ha⁻¹) and then to CO₂e (kg CO₂e ha⁻¹) [32]. Equipment used for each crop or animal management activity was referenced to Nagy [33] and Saskatchewan Agriculture Farm Machinery Custom Rental Guide 2008–09 [34] to determine a work rate (ha h⁻¹) for the actual equipment combination used or its’ equivalent size and type. Then, for cropping activities, the equipment combination was matched to embodied fuel and lubrication energy required hourly [33] and a total determined for the energy required (MJ ha⁻¹) for annual crop production from pre-seeding operations to harvest. For feeding (included processing, delivery and hauling), bedding and manure removal, the actual daily or periodic events were timed, the amounts of energy expended calculated using the hourly energy utilization rate (MJ h⁻¹) for the equipment combination used [33] for embodied, fuel and lubrication energy, then totaled and expressed on an animal unit basis for each specific classification of animal units (MJ hd-d⁻¹).

Energy used in the manufacture and transportation of seed, fertilizer and herbicides used were determined (Table A4) for each crop and feed combination in each year. The embodied energy coefficients for seed and herbicide were from Nagy [32,33] and Zentner et al. [35], and fertilizer from Snyder et al. [36]. Energy (MJ kg⁻¹) was converted to a diesel fuel equivalent (MJ L⁻¹) and then to CO₂ equivalent (kg CO₂e kg⁻¹) for the specific input [32]. For each crop or feed all energy sources (inputs, equipment and production activities) were summed on per ha basis (kg CO₂e ha⁻¹).

Sources of direct and indirect nitrous oxide emission from cropping systems and pastures are summarized in Table A5. Methods for calculation of N₂O generally follow the outline for defining nitrogen fractions and emission factors for crops on an ecodistrict basis as described by Rochette et al. [37]. The base emission factor for Ecodistrict 737 incorporates the factors for tillage, topography, irrigation and soil texture typical of the Bowden to Wetaskiwin farming areas of Alberta. Irrigation was not used and was not a consideration. No manure was applied to farm lands. Manure from grazing animals is not considered as part of the crop input component [37], and was included in the animal component for CH₄ and N₂O emissions from manure. N mineralization was assumed be at steady-state, therefore net mineralization was equal to 0.0 and soil mineralized-N not added to the sum of crop N-inputs. Relevant information for the ecodistrict as supplied by Worth and Desjardins [38] were P/PE = 0.65, Emission factor _soil = 0.0095 kg N₂O-N kg^{− 1} N.

Fertilizer-N applied and above ground residue and root contributions were determined and summed as direct emissions. Residue and root-N were quantified using actual data for residue dry matter, root mass and N concentration or using methods and ratios supplied by Janzen et al. [39] for product: above ground residue: root and appropriate N concentrations. All residue-N and root-N was assumed to be returned to the soil each year for annual crops, but only 10% of perennial pasture and hay residue was included in crop N-inputs [39]. Root mass of all perennial pasture and hay stands was assumed to be 10,000 kg ha⁻¹ as approximated in smaller paddock and plot sites at Lacombe [40]. Indirect N₂O-N emission consisted of leaching of fertilizer-N and root and residue-N that had mineralized within the year. In this case a leaching fraction (FRAC) _Leach = 0.19 kg N ha⁻¹ of N-inputs and volatilization of applied fertilizer-N of = 0.10 kg N ha⁻¹ of fertilizer-N inputs. Emission factor for leaching (EF _Leach) was 0.0075 kg N₂O kg⁻¹ N ha⁻¹ and the emission factor for volatilization = 0.01 kg N₂O kg⁻¹ N ha⁻¹.

Enteric CH₄ and manure CH₄ and N₂O were expressed in kg head⁻¹ d⁻¹ and multiplied by 23 for CH₄ and 298 for N₂O and then by the number of days in each feeding period for each cattle class to give kg CO₂e head⁻¹ period⁻¹. Carbon dioxide equivalents from enteric fermentation and manure were then summed across feeding periods for each cattle class to give kg CO₂e head⁻¹ yr⁻¹ or kg CO₂e head⁻¹ period⁻¹ for calf- and yearling-fed steers and heifers where period refers to the days from birth to harvest. These values were then scaled to a herd size of 160 cows assuming 85% calves weaned of cows exposed to breeding (n = 136), 10% culled cows in the fall due to reproductive failure (n = 16), 5% cows culled in the spring due to failure to deliver a live calf, calving difficulty or temperament (n = 8), 15% replacement heifers to replace the culled cows and keep the herd growth at zero (n = 24), a 25:1 cow:bull ratio (n = 6 bulls), six young replacement bulls, six cows to produce the six replacement bulls and 68 slaughter steers and 44 slaughter heifers that is based on 136 calves with 24 females being kept as herd replacements. These proportions are based on the Alberta Cow-calf Audit [41]. The actual live slaughter and carcass weight for the calf- and yearling-fed steers were taken from López-Campos et al. [27] and multiplied by 68 to obtain total beef sold for steers. Live slaughter was 10% lower and dressing percentage 0.5 percentile points lower in heifers than steers [23,28] and these values were used to calculate the live slaughter and carcass weight of heifers that were then multiplied by 44 to obtain total weight of beef sold from slaughter heifers. The diet ingredient composition (DM basis) and DMI within each feeding period and cattle category (Tables A1 and A2) were used to calculate total feed or crop ingredients (kg DM) used. These values were then multiplied by the appropriate energy CO₂ and cropping N₂O emission coefficient within year (Table A6) to given total kg CO₂e from energy use and cropping.

Yearling systems required approximately 85.8 ha more land to complete the production cycle than calf-fed systems. However, the proportion of land composed of annual and perennial crops was identical and carcass productivity per unit land area was similar between yearling and calf fed systems (2.0 to 2.5% mean difference). Carcass productivity per unit land area was similar because carcass size of yearling fed systems averaged 31% heavier than the calf-fed systems (Table A7). Implanted compared to non-implanted carcass output per unit area was 6.9% and 7.7% larger within calf-fed and yearling-fed systems, respectively, because implanted carcasses were about 9% heavier (Table A7). The more intensive Parkland system produced 7.8 times the carcass weight per unit area compared to a more extensive Southern Alberta system described by Beauchemin et al. [11], which was based on native range and dryland crop production. The difference was the 2040 ha of native range [11] required for the cow herd compared to 237 ha pasture in the current study (Tables A9 and A10). In total the Parkland systems in this study had carcass outputs averaging 140.2 kg ha⁻¹ compared to 18 kg ha⁻¹ in the Southern Alberta study.

Soils under crop management (e.g., cereals and oilseeds) that has occurred for 100 years after breaking may be at a soil organic carbon (SOC) equilibrium level typical of that management, but with lower SOC stores. Conversion of cropland to grassland may result in increasing SOC stores that occur rapidly over the first 20 years, but may continue at a much slower rate for up to 50 or 100 years [47,48]. The majority of SOC accumulation may occur within the first decade of grassland conversion unless additional practices such as application of fertilizer-N occur in concert with the conversion [49]. Soil organic matter gains on conversion of cropland to grassland will be composed of relatively high proportions of light fraction SOC, which may be highly degradable and subject to loss under some conditions [50,51,52]. The reverse process of SOC loss after grassland conversion to cropland follows a trajectory of high annual SOC loss-rates immediately after conversion, then decreasing rates with time [47,48]. Land used in the current study has been under cultivation for over 100 years, but has been in rotation between forage, cereal and oilseed crops frequently over that period. Thus, the trends in SOC may be increasing for the forage and pasture stands and at equilibrium or decreasing for lands used for cereal grain and silage [53,54]. The amounts of land and short period of the study prevented a detailed assessment of changes in soil carbon and effectiveness as a greenhouse gas sink. Therefore gain and loss of SOC was attributed to each land type based on literature, available measurement, management and time from conversion from or to cropland and grassland (i.e., stand age).

Using the Century model, the Canadian National Inventory Report [48] estimates for conversion of cropland to grassland in the Prairie Parkland a linear sequestration rate of 0.55 Mg ha⁻¹ yr⁻¹ SOC over the first 20 yr of conversion and a mean rate of 0.2 Mg ha⁻¹ yr⁻¹ SOC over 100 years; a land management change from grassland to cropland or in reverse requires SOC loss of the same magnitude. Net Ecosystem CO₂ flux (NEE) was determined annually and continuously with the Bowen Ratio Energy Balance (BREB) method on cropland converted to fertilized, grassland [55] from 2002 until 2008, including establishment, for areas used in feed production of this study. Including the year of establishment a BREB annual NEE rate averaged 0.6 Mg C (2.1 Mg CO₂) ha⁻¹ yr⁻¹ on this converted perennial stand. A rotation of cereals and oilseeds (2005 to 2009) in an adjacent field, broken from timothy, averaged a NEE loss of 0.45 Mg C (1.65 Mg CO₂) ha⁻¹ yr⁻¹. These local sequestration and emission rates are in general agreement with the current literature [47,48,56,57] for conversion into or out of grassland. Lower sequestration rates were attributed to hay and haylage stands than new pasture (Table A9), because pasture had higher residue-C inputs. Hay, silage and grain lands had residues removed from fields for feeding and therefore had lower direct C inputs to soil than pasture [58]. However, no C grown within the farm or study boundaries left the farm. No C was transported off the farm in the form of feed, fiber or manure.

System net emissions per unit area were reduced by 11% and 16% for calf-fed and yearling fed systems, respectively, when SOC loss and sequestration were included in the determination of greenhouse gas intensity (Tables A9 and A10). The difference between the calf- and yearling-fed systems was the use of recently converted, highly productive pasture in the yearling-fed system (Table A9), to which was attributed the highest SOC-sequestration rate as well as more silage and grain to which slightly lower loss-rates occurred. The native pasture used in the study of Beauchemin et al. [11] may have been at equilibrium as it had not been cultivated, although the carbon stores may have been very high. Soil carbon stores [59] to a 30-cm depth under long-term stands of unfertilized tame grass-legume, native grass and old grass mixed species were 107.4, 90 and 76.2 Mg C ha⁻¹ for a Dark Brown soil at Scott, SK. Comparable side by side cropped sequences to each grassland treatment at Scott, SK were 65.0, 51.4 and 52.6 Mg C ha⁻¹, respectively. In addition the GHG intensities, expressed as kg CO₂e per kg carcass weight or kg CO₂e per kg carcass weight per yr were reduced by 10.9%, 10.9%, 16.1% and 15.6% for the calf-fed non implanted, calf-fed implanted, yearling-fed non implanted and yearling-fed implanted beef production strategies, respectively. Again the difference in GHG emission reduction was due to more tame pasture used (62 ha) and thus more C-sequestration in the yearling-fed production system compared to the smaller loss-rates incurred by the extra silage (13.5 ha) and grain (7 to 10 ha).