Modelling environmental impacts associated with the removal of productivity-enhancing technologies from Canadian feedlots: a case study

Abstract Greenhouse gas (GHG) and ammonia (NH3) emissions, land and water use associated with feedlot cattle (n = 40 hd treatment−1 trial−1) treated with or without productivity-enhancing technologies were modelled for a multiyear study (n = 4). Heifers (H) were assigned to the following treatments: (1) implanted (HTBA); (2) provided with melengestrol acetate (HMGA); (3) nonimplanted control, weight-adjusted (CON_Adj) to achieve the same final carcass weight (CW) as 1 (HCON_AdjTBA); or (4) CON_Adj to achieve the CW as 2 (HCON_AdjMGA). Steers (S) were assigned as follows: (1) implanted (STBA); (2) implanted and provided with ractopamine hydrochloride (SRAC; conducted in the last 2 years); (3) CON_Adj to achieve the same CW as 1 (SCON_AdjTBA); or (4) CON_Adj to achieve the same CW as 2 (SCON_AdjRAC). The GHG and NH3 emissions from HTBA, HMGA, STBA, and SRAC were 3.8%, 3.0%, 10.1%, and 8.5% lower and 4.3%, 2.9%, 7.4%, and 7.6% lower, respectively, than the respective control cattle. The land required to produce feed was also reduced by 6.6%, 4.8%, 9.9%, and 10.9%, while water use was reduced by 6.4%, 4.8%, 10.1%, and 11.1% for HTBA, HMGA, STBA, and SRAC, respectively. This modelling study clearly demonstrates that conventional beef production systems have a lower environmental footprint than nonconventional systems.


Introduction
The United Nation (UN) estimates that the global population will increase to 9.7 billion people by 2050 (UN 2019).Avenues by which food security can be achieved for the growing global population include the following: (i) investments in sustainable agricultural production and rural development, (ii) technology and productivity growth, and (iii) support to farmers, trade, and markets (Rockström et al. 2017).Therefore, evaluating the environmental footprint of existing and new technologies is a necessary step to assess impact and identify strategies to improve the environmental sustainability of Canadian agro-ecosystems.Of the total Canadian greenhouse gas (GHG) emissions in 2019, agriculture (livestock and crop production) accounted for approximately 8.1%, of which enteric methane (CH 4 ) emissions were responsible for 41% (Environment and Climate Change Canada 2021).Additional environmental implications associated with cattle production include but are not limited to natural resource use (i.e., land and water), biodiversity, odor, and volatilization and leaching of ammonia (NH 3 ) from cattle manure.Herein, the environmental metrics include GHG and NH 3 emissions, land use, and water use.
A number of modelling studies have estimated the environmental impacts associated with the cattle industry in the United States (Beckett and Oltjen 1993;Hristov et al. 2011;Stackhouse-Lawson et al. 2012;White and Hall 2017) and Canada (Ominski et al. 2007;Beauchemin et al. 2010;Alemu et al. 2017a;Legesse et al. 2016Legesse et al. , 2018aLegesse et al. , 2018b)).Potential strategies to reduce the environmental footprint associated with beef production, includes genetic improvement (Basarab et al. 2013), reducing days on feed (DOF) prior to slaughter (Capper 2012), increasing breeding stock longevity, improving weaning rates, conversion of crop land to pasture (Beauchemin et al. 2011), and diet modification, including feeding of higher quality forages (Guyader et al. 2017) or dried distillers grains (Hünerberg et al. 2014).
Improvements in animal production efficiency (i.e., average daily gain (ADG) and feed efficiency (FE; gain:feed)) have occurred via genetic advancements, improved management systems, and use of productivity-enhancing technolo-gies (PETs; Brameld and Parr 2016).PETs such as implants, melengestrol acetate (MGA), beta-adrenergic agonists (β-AA; e.g., ractopamine hydrochloride (RAC)), and ionophores (e.g., monensin) are conventional technologies frequently used in feedlots to improve the growth and FE of beef cattle (Aboagye et al. 2021).Despite improvements in production efficiency realized with the use of PETs, there has been a shift in demand toward "free-from" products (i.e., free from growth hormones or antibiotics) in domestic and global markets (Webb et al. 2017;Garmyn 2020;Yang et al. 2020).However, consumers are largely unaware of the implications of eliminating PETs from the production system on environmental sustainability and future food security.Efficiency losses from banning conventional technologies may make it increasingly challenging to feed a growing human population, with negative consequences for the environment (Aboagye et al. 2022).The effect of the removal of PETs on land use and the carbon footprint of beef production was examined in Canada by Basarab et al. (2012).In this study, the land required to produce a kilogram of beef without growth implants increased by 8.5%, with a 6% increase in the carbon footprint intensity.However, Basarab et al. (2012) did not include other environmental metrics such as water use and NH 3 emissions.More recently, a more extensive review of the environmental footprint, including ammonia emissions, as well as land and water use was conducted in the United States (Crawford et al. 2022) and Brazil (Capper et al. 2021).Thus, the objective of the current study was to model and compare the environmental footprint (i.e., GHG and NH 3 emissions, land use, and water use) of feedlot steers and heifers from a multiyear study using current conventional western Canadian production practices with or without PETs.We hypothesized that the environmental footprint of backgrounding and finishing beef cattle in a feedlot production system will be reduced with the use of PETs.

Materials and methods
The trials from which the data were garnered were conducted at the Agriculture and Agri-food Canada (AAFC) Research Development Centre located in Lethbridge, AB.The animal experimental protocols (ACC1527, ACC1635, ACC1715, and ACC1823) were reviewed and approved by the Lethbridge Research and Development Centre Animal Care Committee and the care of handling animals followed the Canadian Council of Animal Care Guidelines (CCAC 2009).

Description of farm location and climate
A partial lifecycle assessment (LCA) was conducted as reported by Boonstra (2022) to quantify whole-farm GHG emissions encompassing the feedlot (backgrounding and finishing) and associated inputs/outputs.The boundaries for the present LCA were set within the feedlot phase and did not include the processing and packaging stages of the final product, with the exception of the water footprint.This is because, generally, the latter stages represent less than 2% of beef chain emissions and packers and processors are reluctant to share proprietary information (Li et al. 2020).For the water footprint analysis, water used for feed production for the animals and at the processing plant was included and expressed per kilogram of boneless beef (Legesse et al. 2018a).The simulated model farm was located in Lethbridge, Alberta, Canada, within Ecodistrict 793.Additional site information, including soil characteristics, soil class, and subregion, was obtained (Alberta Soil File Information Center 2016) and characterized as dark brown soil zone of southwestern Alberta, 2A, and mixed-grass, respectively.Long-term average weather conditions were used in the present study to account for yearto-year variability in emissions and natural resource use.Average monthly temperatures (1991 to 2020) and growing season precipitation (May to October; 260 mm) were obtained from the Lethbridge Weather Station (Table S1).

Description of animal management scenarios
Growth performance data were obtained from four backgrounding and finishing trials conducted at the Lethbridge Research and Development Centre feedlot over 4 consecutive years from 2015 to 2018, with 120 heifers and 80 steers included each year for the first two trials and an additional 40 steers included in the final two trials (Ribeiro et al. 2021).The production cycle in each of the four trials began in the fall and ended the subsequent summer.Cattle were blocked by weight and sex and allocated into pens (n = 10 hd pen −1 ) bedded with barley straw.Cattle were backgrounded for the first 84 days, followed by a 28 day transition period, and finished during the remaining 148 ± 5 days of the production cycle.Both steers and heifers were fed a basal diet containing corn silage, barley grain, corn dried distillers grains (DDGS), and a mineral supplement to meet or exceed the nutrient requirements of growing and finishing beef cattle (Table 1; NASEM 2016).Slick bunk management was used to limit residual feed in the bunk while still meeting performance expectations.Diets were prepared daily as a total mixed ration and delivered to pens using a Beck 220 Mixer (Beck Implement Inc., Elgin, MN, USA).Feed samples were collected weekly, composited by weigh period, and analyzed as described by Ribeiro et al. (2021).Crude protein (CP; kg kg −1 diet dry matter (DM)) and total digestible nutrient (TDN; %) values were pooled by phase (Table 1).
Feeder cattle were managed using eight scenarios (n = 40 hd treatment −1 trial −1 ) as described by Ribeiro et al. (2021).However, as all groups were fed for 233 ± 8 days, the control treatments had lower final weights than the PET-treated cattle.Therefore, the number of DOF for the heifers (H) and steers (S) on the control (HCON and SCON, respectively) treatment was adjusted (Adj) by estimating the additional DOF and adding to the finishing period within each trial year (Tables 2  and 3) to achieve the same finished weight as the PET cattle.More specifically, HCON was adjusted to H implanted (HTBA) and H provided with MGA (HMGA) finishing weights, resulting in HCON_AdjTBA and HCON_AdjMGA, respectively, and SCON was adjusted to STBA and SRAC finishing weights, resulting in SCON_AdjTBA and SCON_AdjRAC, respectively.The resulting difference in DOF varied from −1 to 65 days (Tables 2 and 3).Therefore, treatments for the heifers were as follows: (1) HTBA; (2) HMGA; (3) HCON_AdjTBA; and (4) HCON_AdjMGA.Accordingly, treatments for steers were as  (Elanco Animal Health, Mississauga, ON).MGA (100 Premix; Zoetis Canada Inc., Kirkland, QC) was included in HMGA diets at a rate of 0.40 mg heifer −1 day −1 .Ractopamine (Optaflexx; Elanco Animal Health, Mississauga, ON) was included in addition to the implant protocol for SRAC at a rate of 30 mg RAC per kg of total diet during the last 42 days before slaughter.
Daily DM intake (DMI; kg day −1 ; Tables 2 and 3) was estimated by recording daily feed deliveries to each pen, divided by the number of cattle within the pen.Bunks in each pen were cleaned weekly, with orts subtracted from feed deliveries.Monensin (Rumensin; Elanco Animal Health, Mississauga, ON) was included in all diets at 33 ppm.Body weight (BW) was recorded on 2 consecutive days at the beginning and end of the backgrounding (n = 4) and finishing (n = 5) period, and ADG was estimated within each period (Tables 2 and 3).At the end of the finishing period, all cattle were processed at Cargill Ltd. (High River, AB).Carcass characteristics, including carcass weight (CW) and dressing percentage (DP), were summarized by treatment and year (Tables 2 and 3).Control cattle DP was applied to the control-adjusted cattle values to estimate the slaughter weight.Boneless beef was calculated as follows: Boneless beef = DP × slaughter weight × 0.73 where boneless beef is the lean meat associated with the carcass, kg; DP is the average dressing percentage, %; 0.73 is used to convert from CW to boneless beef (Agriculture and Agri-Food Canada 2021; Agriculture Marketing Guide 2021).

Description of cropping and land use
In the current study, silage and grains were considered to originate from the farm and were produced using lowtill management, while minerals and supplements were purchased.Total land required (ha treatment −1 ) to produce feed for cattle in the production trials was calculated using the ingredient inclusion rate (% DM), total DMI (kg hd −1 day −1 ), the average DM of the feed ingredient (%), as well as storage and feeding losses (Rotz and Muck 1994;Table 4).Estimates of crop yields were based on a 4 year average (2015 to 2018) and were expressed as kg ha −1 , as fed (Table 4).Yields and fertilizer rates for all crops were characteristic of those within the Lethbridge region (Agriculture Food and Rural Development 2004;Agriculture Financial Services Corporation 2019;Alberta Agriculture and Forestry 2021;Alberta Crop Reports 2021).Finally, DOF and number of cattle per pen (n = 10) were used to calculate the land requirements to produce the feed fed over the duration of the trial.
The GHGenius model (version 5.01a; http://www.ghgenius.ca)was used to calculate the quantity of DDGS produced from a tonne of corn grain processed in an ethanol plant (i.e., 300 kg tonne −1 ) based on the yield of corn within this Ecodistrict.It was also assumed that all crops were irrigated and treated with herbicides.Emissions associated with the manufacture of herbicides and fertilizers and those associated with fossil fuel use during their application were accounted for.
Estimating greenhouse gas emissions and system boundaries Holos (https://agriculture.canada.ca/en/agricultural-science-and-innovation/agricultural-research-results/holos-sof tw are-program), which employs IPPC Tier II algorithms modified to account for Canadian conditions and agricultural practices, was used to estimate on-farm GHG emissions (Little et al. 2008).The model has previously been used to assess the carbon footprint of Canadian agricultural practices (Beauchemin et al. 2010;Guyader et al. 2017;Little et al. 2017;Alemu et al. 2017b).Further, the model has been recently used to examine the impact of commercial Canadian beef production systems utilizing conventional and nonconventional technologies on GHG emissions (Aboagye et al. 2022).
The GHG emissions estimated from the Holos model included on-farm emissions of (i) CH 4 arising from enteric fermentation and manure decomposition, (ii) nitrous oxide (N 2 O; direct) from cropping, and (iii) carbon dioxide (CO 2 ) from energy use.Inputs for crop production, such as fertilizers and pesticides, and indirect emissions of N 2 O from N Table 2. Growth performance and carcass data of feedlot heifers backgrounded and finished with and without the use of productivity-enhancing technologies (n = 4 trials) a .leaching and volatilization were also included in the model.Emissions associated with transportation (i.e., animals and feed, feed additives, implants), other capital goods, and processing and manufacturing were not included.However, emissions from the production, transport, and processing of corn grain to ethanol and DDGS, and subsequent transport of DDGS from the ethanol plant, were estimated using GHGenius (mass allocation = 1032 g CO 2 e per kg of DDGS).Methane from enteric fermentation was calculated based on the CH 4 conversion factor of the diet (unadjusted), TDN, and CP content of the diet.Manure CH 4 emissions were cal-culated using the deep-bedding manure handling system and the associated conversion factor.
Direct N 2 O emissions from soil and cropping were estimated using total N inputs, including, crop residues, land applied manure, and fertilizer and the direct N 2 O emission factor (EF) from crops and soils for each specific Ecodistrict (Rochette et al. 2008).Adjustments within the Holos model accounted for climatic variables (i.e., growing season precipitation and potential evapotranspiration), soil variables (i.e., soil type and texture), tillage intensity, and topography.Monthly variables considered included (i) average temperature and (ii) soil N 2 O conversion--monthly soil N 2 O emis-Table 3. Growth performance and carcass data of feedlot steers backgrounded and finished with and without the use of productivity-enhancing technologies (n = 4 trials) a .sions (%).Direct N 2 O emissions from manure were estimated using the EF (0.01 kg N 2 O-N (kg N) −1 ; IPCC 2006) associated with a deep-bedding manure handling system.Indirect N 2 O emissions included N leaching and volatilization (i.e., amount of N lost to runoff, leaching, and volatilization).In direct N 2 O, emissions through leaching and volatilization were estimated using the EF 0.0075 and 0.01 kg N 2 O-N (kg N) −1 , respectively (IPCC 2006).Fossil fuels and energy required for crop production, including machinery use (2.39 GJ ha −1 , 70.00 kg CO 2 GJ −1 ) and irrigation (367.00 kg CO 2 ha −1 ), as well as animal feeding and housing per year (65.70 kWh beef cattle −1 , 0.20 kg CO 2 kW h −1 ) and manure handling (0.0248 GJ 1000 L −1 , 70.00 kg CO 2 GJ −1 ), were the primary sources of on-farm energy use that contributed to CO 2 emissions (Little et al. 2008).Secondary on-farm energy use included CO 2 emissions associated with crop production inputs such as herbicides (0.23 GJ ha −1 , 5.8 CO 2 GJ −1 ; Little et al. 2008) and N (3.59 kg CO 2 (kg N) −1 ) and phosphorus (0.5699 kg CO 2 (kg P 2 O 5 ) −1 ) fertilizers (Nagy 2000).Estimated GHGs, expressed in CO 2 equivalents (CO 2 e), were calculated as described in the 2006 Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories (IPCC 2006).In this approach, the global warming potential (GWP 100 ) of the gas is multi- e Sourced from Rotz and Muck (1994).
f Barley and corn grain, and corn silage (Agriculture Food and Rural Development 2004).g K c init , K c mid , and K c end , respective crop coefficients for initial, mid, and end growth stages to develop a K c curve during the growing season for each crop (ASCE 1996;Allen et al. 1998Allen et al. , 2007).h L init , L dev , L mid , L late , length of the growing season for initial, development, mid, and end growth stages, respectively (Allen et al. 1998).
i Blue (supplied through irrigation) and green water (precipitation that evaporates or transpired through the crops) were calculated using data from 2015 to 2018 (Lethbridge Weather Station, ID 3033875; Environment and Climate Change Canada 2020) to estimate the volume of water required by crops (L kg −1 DM).
plied by the total emissions of the gas.The GWP 100 for each of the GHG are as follows: CO 2 : 1, CH 4 : 28, and N 2 O: 298 (IPCC 2021).

Ammonia emissions
Ammonia emissions were calculated as described by Legesse et al. (2018).More specifically, daily N excretion by cattle (N excretion rate ) was estimated as the difference between daily N intake and daily N retention (NRC 2000;Gavrilova et al. 2019).Daily N intake was estimated by dividing protein intake (the product of DMI and CP content, % of DM) by 6.25 (eq.1).The DMI was obtained from the energy requirements of feeder cattle in confinement for maintenance (depending on BW and climate) and weight gain (NRC 2000).
N intake = ((GE/18.45)× CP) /6.25 (1) where N intake is the daily N intake, kg N hd −1 day −1 ; GE is the gross energy intake, MJ hd −1 day −1 ; 18.45 is the conversion factor for GE kg −1 DM, MJ kg −1 ; CP is the crude protein of the feed, kg kg −1 ; 6.25 is the coefficient for the conversion from dietary protein to dietary N (Gavrilova et al. 2019).
The N excreted in urine (TAN excreted ) and the remaining N excreted in the feces as FecalN excreted (Dämmgen and Hutchings 2008) were also estimated as: where TAN excreted is the excreted N in urine, kg TAN hd −1 day −1 ; FecalN excreted is the N excreted in the fecal matter, kg fecal N hd −1 day −1 ; 0.57 is the fraction of excreted N in urine.As the CP of the diets was 13%, the fraction of urinary-N was assumed to be 0.57 for backgrounding and finishing cattle (Chai et al. 2014).
Daily NH 3 emissions from confined cattle housing (NH 3emissions, h ; Legesse et al. 2018b;Aboagye et al. 2022) were estimated as: NH 3emissions, h = TAN excreted × 0.9 × Feedlot temp adjustment × 17/14 (4) Feedlot temp adjustment = 1.041 average temp /1.041 17.7  (5) where NH 3emissions, h is the NH 3 emissions from confined animal housing, kg NH 3 hd −1 day −1 ; 0.9 is the emission factor associated with feedlot housing systems, kg NH 3 -N (kg TAN) −1 ; Feedlot temp adjustment is the adjusted temperature when cattle were housed in the feedlot, • C; average temp is the ambient temperature at the feedlot, • C; 17/14 is the coefficient for the conversion of NH 3 -N to NH 3 .
The periodic excreted TAN mass flow from feedlot cattle to pen manure (PTAN flow ; eq. 6) was assumed and estimated by subtracting periodic NH 3 volatilization from the periodic TAN excreted during confinement.Any contribution of NH 3 emissions from waste feeds or bedding was assumed to be negligible and was excluded from the analysis (Legesse et al. 2018b).
where PTAN flow is the periodic TAN mass flow from feedlot cattle to pen manure, Mg TAN hd −1 period −1 ; PTAN excreted, h is the periodic TAN in manure excreted by housed beef cattle (Mg TAN hd −1 feeding period −1 ); PNH 3emissions, h is the periodic emission rate of NH 3 , temperature corrected, from housed beef cattle (Mg NH 3 hd −1 feeding period −1 ); 14/17 is the conversion of NH 3 -N to NH 3 .
The NH 3 volatilization from stockpiled manure (eq.7) was also estimated from TAN in manure during stockpiling.In summary, the variation in TAN during stockpiling was estimated by combining TAN mass flow with TAN mineralized from organic matter while subtracting nitrified N from the TAN pool.
where PTAN stockpiled is the adjusted TAN in stockpiled manure used to calculate NH 3 emissions hd −1 during stockpiling, Mg TAN hd −1 period −1 ; F immob is the TAN fraction which is immobilized to organic N during stockpiling; F nitrify is the TAN fraction nitrified during stockpiling; PON storage is the periodic organic N (excreted fecal N) mass flow to the pen manure, mg organic N hd −1 period −1 ; F mineralize is the organic N fraction which is mineralized as TAN during stockpiling.It was assumed that no TAN was immobilized to organic N during stockpiling (F immob = 0), while the fraction of organic N mineralized (F mineralize ) as TAN was 0.28 and the TAN fraction nitrified from stockpiled manure (F nitrify ) was 0.14 (Chai et al. 2014).
In addition to commercial fertilizers, 57% and 43% of the manure was applied to tilled and untilled land, respectively, as per management practices on Canadian beef farms (Legesse et al. 2018b).Further, as the majority of stockpiled manure from beef farms is applied in the spring or fall (Sheppard et al. 2015), average temperatures from April to May and September to November were used to estimate NH 3 emissions.Volatized NH 3 during land application of manure was estimated based on the available TAN and the EF associated with the application practice, land, tillage, and month of application.The quantity of manure applied and the remain-ing NH 3 after stockpiling were used to estimate the monthly TAN transfer from stockpiled manure to land application as: PTAN land,tillage = F till/untill × PTAN stockpiled − PNH 3emissions,s × 14/17 (8) where PTAN land, tillage is the periodic TAN applied on tilled or untilled land during a specific period, Mg NH 3 hd −1 month −1 ; PNH 3emissions, s is the NH 3 emission rate during the stockpiling period, Mg NH 3 hd −1 month −1 ; F till/until is the fraction of manure applied on tilled or untilled land.

Crop production
Water use associated with feed production was estimated based on water demand, green water (precipitation that evaporates or transpired through the crops), and blue water (irrigation) for each crop and the consumption (kg DM) of dietary ingredients (Legesse et al. 2018a;Aboagye et al. 2022).Gray water (freshwater pollution) was not included in the water footprint.The total volume of water evapotranspired and used to produce 1 kg of crop (L kg −1 DM) was estimated based on the respective yield of each crop.
Potential and actual evapotranspiration data, estimated using the National Drought Model and precipitation data from 2015 to 2018 (Lethbridge Weather Station, ID 3033875; Environment and Climate Change Canada 2020), were used to estimate crop water demand for green and blue water as follows: Crop water demand (CWD) = PT × K c ( 9) where PT is the potential evapotranspiration; AT is the actual evapotranspiration; K c is the the respective crop coefficient (American Society of Civil Engineers (ASCE) 1996).
Crop coefficients (K c ) were derived from the literature (ASCE 1996;Allen et al. 1998Allen et al. , 2007)), and a K c curve (Fig. S1) was developed for each crop, which considered the growing duration and crop development stage (K cinit , K cmid , and K cend ; Table 4).Crop-specific development stages (Allen et al. 1998) were based on growing conditions and management practices in the area, and each stage was associated with an appropriate K c .The length of the four stages of growth and the associated rate constant were as follows: (i) initial (L init ), with correspondence to K cinit , (ii) maturity (L mid ), corresponding to K cmid , (iii) development (L dev ), which was between L init and L mid , with a corresponding K c for the rate from K cinit to K cmid , and (iv) late-season period (L late ) with a corresponding K c for the rate from K cmid to K cend .An additional consideration in constructing the K c curve was the planting date, which was sourced from Huffman et al. (2015) and signified the onset of the growing season.The length of the growing season in days was calculated as the sum of the crop development stage (L init , L dev , L mid , L late ) lengths.

Consumption and processing
Daily water consumption was estimated based on animal category, average BW (Ribeiro et al. 2021), and ambient temperature over the feeding period (Environment and Climate Change Canada 2020), as described by Legesse et al. (2018a).
Total drinking water use = WU coeff × n d × n hd where WU coeff is the water use coefficient (L hd −1 day −1 ; NASEM 2016); n d is the number of days in the feeding period; n hd is the number of animals (hd treatment −1 year −1 ).Water intake was assumed to remain constant at temperatures ≤4.4 • C but to increase as temperature increased beyond this point.Water used for cleaning cattle and facilities was considered negligible (Beaulieu 2007;Legesse et al. 2018a) and was excluded.Water use for processing beef was based on the average water use efficiency among processing plants in western Canada at 16.5 L kg −1 boneless beef (Legesse et al. 2018a).

Water use
Similarly, water use requirement over the 4 year trial was lower for PET cattle than control cattle (Table 5).Water use decreased for implanted heifers by 6.4% compared with HCON_AdjTBA (5.89 vs. 6.29 m 3 per kg boneless beef) and by 4.8% for HMGA relative to HCON_HMGA (5.76 vs. 6.05 m 3 per kg boneless beef).Use of implants also decreased water use intensity in steers by 10.1% compared with SCON_AdjTBA (5.74 vs. 6.38 m 3 per kg boneless beef), and by 11.1% for SRAC compared with SCON_AdjRAC (5.78 vs. 6.50 m 3 per kg boneless beef; Fig. 1).

Greenhouse gas emissions
The observed 3.0% to 10.2% reduction in total GHG emissions (kg CO 2 e kg CW −1 ) associated with PET use can be attributed to several factors, including sex, DOF, and type of PET administered.Specifically, GHG emission intensities were lower for steers than heifers and for cattle that received PET vs. those that did not.In addition, a greater percentage reduction in GHG emission intensities was also observed in implanted cattle than MGA-and RAC-administered heifers and steers, respectively.The increased magnitude of GHG reduction for steers compared with heifers may be attributed to the observed increase in performance as TBA implants increased weight gain by 11.3% and 7.4%, respectively, an outcome attributed to the higher DMI of steers.Further, increased DMI may account for the decreased GHG emission intensity response in implanted steers compared with implanted heifers (Aboagye et al. 2022).Similarly, an increased magnitude of reduction in the GHG emissions (kg CW −1 ) for implanted steers compared with heifers (13.7% vs. 9.6% reduction) was observed in a commercial feedlot for finishing beef cattle in western Canada (Aboagye et al. 2022).
A recent review by Aboagye et al. (2021) summarized the environmental impacts, including GHG emissions, associated with the use of PET in several studies conducted in North America.In Canada, Basarab et al. (2012) estimated a 4.9% to 5.1% reduction in GHG emission intensity for cattle receiving PET, but the type and timing of the administration differed from the present study.Basarab et al. (2012) estimated GHG emissions from heifers and steers that were either (i) implanted with 200 mg progesterone and 20 mg estradiol benzoate at weaning and reimplanted with 120 mg TBA and 24 mg estradiol 90 to 100 days before slaughter, or (ii) implanted with 200 mg progesterone and 20 mg estradiol benzoate at weaning and then implanted four more times at 80 to 90 days intervals, followed by a 120 mg TBA and 24 mg estradiol implanted at 90 to 100 days before slaughter.In the current study, heifers received three TBA/estradiol implants or MGA, and steers received the same implant regime as the heifers, with or without RAC.Our study also estimated emissions during the backgrounding and finishing periods, whereas Basarab et al. (2012) estimated GHG emissions from weaning to slaughter.As far as we are aware, our study is the first to include multiyear experiments specifically to assess the impact of PET on the carbon footprint of feedlot cattle.
In the United States, Stackhouse-Lawson et al. (2013) reported a considerably greater reduction in GHG emissions (kg CW −1 ; 22%) from PET cattle compared with control cattle.Although the PETs used in the Stackhouse-Lawson et al. (2013) study and the current study were similar, i.e., (i) monensin only, (ii) monensin plus implant, and (iii) monensin plus implant plus RAC, the duration of the feeding period differed between the two studies (365 vs. 245 ± 18 days).Further, the inclusion of monensin as a treatment independent of the control allowed for a closer investigation of their impact on GHG emissions than the current study, with monensin being associated with an 11% reduction in total CH 4 and a 6% reduction in CO 2 (Stackhouse-Lawson et al. 2013).In the current study, all cattle received monensin, so it was not possible to determine whether reductions in GHG emissions as a result of its use were additive to other PETs.If they were additive, it might account for the lower reduction in GHG emission intensity with PET vs. controls as compared with some other studies.

Ammonia emissions
The observed reduction in total NH 3 emissions intensity with PET in the current study (2.9% to 7.6%) was lower than that reported in the United States by Stackhouse et al. (2012;6.0% to 13%).This greater reduction in NH 3 emissions may reflect the lower dietary CP (11.8% vs. 13.4%),longer backgrounding period (182 vs. 84 days), and different system boundaries examined by Stackhouse et al. (2012).In addition to examining the NH 3 emissions from the feedlot, stockpiled manure, and field-applied manure, the latter study also examined emissions from cattle on grazed pasture and rangeland Fig. 1.Average percentage differences in the intensities (i.e., greenhouse gas (GHG) emissions (CO 2 e), ammonia emissions (g), and land use (m 2 )) are expressed in terms of kg of carcass weight, while water use (m 2 ) is expressed in terms of kg of boneless beef of feedlot heifers or steers administered productivity-enhancing technologies as compared to respective controls (n = 4 trials).The number of days on feed for control cattle to reach their finished weight was adjusted using the same final weight as PET cattle.
-12.00% -10.00% -8.00% -6.00% -4.00% -2.00% 0.00% GHG Ammonia Land Water HTBA HMGA STBA SRAC (Stackhouse et al. 2012).Including emissions from grazing animals during the backgrounding phase decreased the NH 3 emissions of the entire beef production system (Stackhouse et al. 2012).This reduction in NH 3 emissions may be partly due to lower NH 3 volatilization associated with manure deposited on pasture (Hristov et al. 2011).Furthermore, in the present study, a greater percentage decrease in NH 3 emissions for implanted heifers relative to MGA-treated heifers, or RAC-treated steers than implanted steers, reflects differences in study design, including the use of heifers vs. steers, and the different technologies used.For example, the RAC-treated steers also received implants, potentially resulting in synergistic activity to decrease NH 3 emissions.In another US study, NH 3 emissions directly measured in steers decreased by 43% with the combination of implants and the β-AA, zilpaterol chloride (ZC), compared with implant only (0.19 vs. 0.33 g of NH 3 kg CW −1 day −1 ; Stackhouse-Lawson et al. 2013).Further, the use of another β-AA, lubabegron, has recently been approved by the US Food and Drug Administration with the claim that it decreased estimated NH 3 emissions by 3.8% to 14.6% (15.2, 14.3, and, 13.5 g NH 3 kg CW −1 ) when fed at different dosages (1.5, 3.5, or 5.5 mg kg −1 DM) with implants, monensin, and tylosin compared with steers that received only the combination of monensin, implants, and tylosin (15.8 g NH 3 kg CW −1 ; Kube et al. 2021).Although these traditional (RAC and ZC) and the novel (lubabegron) β-AA feed technologies administered to cattle have different modes of actions with regard to their impacts on N metabolism (Dilger et al. 2021), it is likely that both β-AA type products impacted rumen degradation of dietary protein, increased protein deposition, reduced urinary urea excretion, and thereby decreased the intensity of NH 3 emissions.The greater magnitude of response with β-AA technologies compared with other PETs aligns with the results of our study.

Land use
Average crop yield and fertilizer rates (2015 to 2018), temperature (1991 to 2020), and precipitation (2015 to 2018) were consistent between treatments and trial years, eliminating any potential annual variation that these parameters may have had on the estimated land required to produce the feed required for the experiments.Therefore, in the current study, the observed improvement in land-use efficiency (m 2 kg CW −1 ) was attributed to the increased CW of PET heifers and steers compared with control cattle (Ribeiro et al. 2021; Tables 3 and 4).Similarly, in Canada, improved land-use efficiency due to an increase in the CW of implanted cattle ranging from 4.0% to 9.0% has been observed (Basarab et al. 2012;Aboagye et al. 2022).Furthermore, a recent study in Brazil also showed that, compared with control cattle (no implants), the quantities of feed and land required to produce a unit of beef were reduced in implanted cattle over the entire production system, with reductions proportional to the level of animal performance by the implant (high > medium > low; Capper et al. 2021).Consequently, due to the lower ADG of cattle raised without these additives, Capper and Hayes (2012) reported a 9.1% increase in the amount of land required to produce feed when PETs were eliminated in the US production system.

Water use
As water is the first limiting resource for many agricultural products, the efficient use of water in livestock production through any feed technology is very critical to sustainable animal production.The reduction in water use intensity observed in the present study was similar to that reported by Aboagye et al. (2022) who reported an average of 14.6% reduction in water requirements when PETs were removed in a commercial feedlot for finishing cattle in western Canada.Similarly, Capper (2012) who estimated a 15.2% reduction in water requirements over the entire US production system as the "natural" production system was shifted to PET-raised beef cattle (0.573 vs. 0.486 m 3 per kg of beef).Aboagye et al. (2022) and Capper (2012) attributed the reduction in water use with PET to an increase in growth rate and slaughter weight.In our study, the reduced water use for PET cattle was attributed to increased ADG, resulting in fewer DOF.In the United States, Webb (2018) also estimated that the use of PETs in a cow-calf to finish production system reduced water use by 1.0%, 5.8%, and 4.4%, with ionophores, implants plus ionophore, and ionophores plus implants plus β-AA, respectively.Differences between our results and those reported by Webb (2018) may be attributed to the type of model used to estimate water use and the location of the trial (Alberta vs. South Dakota and Nebraska) as soil type, climate, precipitation level, and crop yields differ regionally and/or seasonally, impacting water use.Further, Capper and Hayes (2012) evaluated the effects of PET (β-AA plus implant plus ionophore plus MGA) in producing an equivalent yearly amount (454 million kg) of beef as a "natural" production system using backgrounding and finishing cattle and estimated a 4.5% (1.05 vs. 1.10 m 3 per kg of beef) reduction in the amount of water required.As with GHG and NH 3 emissions and land use, the magnitude of water use reductions was also greater for steers than heifers.

Conclusion
Results from this modelling study suggest that the use of PET will lower GHG and NH 3 emissions, as well as land and water use, thereby decreasing the environmental footprint associated with feedlot cattle during backgrounding and finishing under the weather conditions and management prevailing in western Canada.Improvements in ADG, FE, and slaughter weight as a result of PET were the main factors that contributed to the decrease in each of the environmental indices.This study adds to the growing body of evidence that the use of PET results in a significant reduction in the environmental footprint of beef within North American production systems.Although recent trade agreements and consumer demand have signalled opportunities regarding the adoption of PET-free production, the economic benefit to producers remains to be elucidated.It is clear from the research presented here and elsewhere that the environmental cost of eliminating PET in the Canadian cattle production system would be significant.

Table 1 .
Inclusion rate of ingredients and nutrient composition (mean ± standard deviation) of experimental diets offered to heifers and steers during backgrounding and finishing (n = 4 trials) a .

Diet ingredient (% of DM)
Ribeiro et al. 2021ded in addition to the implant protocol for SRAC at a rate of 30 mg RAC per kg of total diet during the last 42 days before slaughter); SCON_AdjTBA and SCON_AdjRAC, control steers with an adjusted final body weight (BW) to achieve the same finished weight as STBA and SRAC; monensin was included in all diets at 33 ppm.The average final BW of unadjusted control steers was 625.8 kg (10.7% and 13.0% less than STBA and SRAC, respectively;Ribeiro et al. 2021).

Table 4 .
Cropping system input parameters associated with production of barley and corn grain and corn silage in the diets of feedlot heifers and steers a .

Table 5 .
Environmental intensity associated with feedlot heifers and steers backgrounded and finished with and without the use of productivity-enhancing technologies (n = 4 trials) a .Environmental metrics are expressed on an intensity basis; GHG and ammonia emissions and land use are expressed in terms of kg of carcass weight, while water use is expressed in terms of kg of boneless beef.b HTBA, heifers implanted with Component TE-100 (containing 100 mg trenbolone acetate + 10 mg estradiol USP + 29 mg of tylosin tartrate and a single implant of Component TE-200 containing 200 mg trenbolone acetate + 20 mg estradiol USP + 29 mg of tylosin tartrate); HMGA, heifers administered melengestrol acetate (MGA 100 Premix was included in HMGA diets at a rate of 0.40 mg heifer −1 day −1 ); HCON_AdjTBA and HCON_AdjMGA, control heifers with an adjusted final BW to achieve the same finished weight as HTBA and HMGA; monensin was included in all diets at 33 ppm.c STBA, steers implanted with Component TE-100 (containing 100 mg trenbolone acetate + 10 mg estradiol USP + 29 mg of tylosin tartrate and a single implant of Component TE-200 containing 200 mg trenbolone acetate + 20 mg estradiol USP + 29 mg of tylosin tartrate); SRAC, steers administered ractopamine (Optaflexx was included in addition to the implant protocol for SRAC at a rate of 30 mg RAC per kg of total diet during the last 42 days before slaughter); SCON_AdjTBA and SCON_AdjRAC, control steers with an adjusted final BW to achieve the same finished weight as STBA and SRAC; Monensin was included in all diets at 33 ppm.Mean, average for each dietary treatment across trial and within environmental parameter; Std., standard deviation.
a d GHG, greenhouse gas emissions.e