Projected impact of future climate conditions on the agronomic and environmental performance of Canadian dairy farms

Agricultural Systems 157 (2017) 241–257

Contents lists available at ScienceDirect

Agricultural Systems journal homepage: www.elsevier.com/locate/agsy

Projected impact of future climate conditions on the agronomic and environmental performance of Canadian dairy farms

MARK

Marie-Noëlle Thiviergea, Guillaume Jégoa,⁎, Gilles Bélangera, Martin H. Chantignya, C. Alan Rotzb, Édith Charbonneauc, Vern S. Barond, Budong Qiane a

Agriculture and Agri-Food Canada, 2560 Hochelaga Blvd., Québec, QC G1V 2J3, Canada United States Department of Agriculture, Building 3702, Curtin Road, University Park, PA 16802, United States c Département des sciences animales, Université Laval, 2425 rue de l'Agriculture, Québec, QC G1V 0A6, Canada d Agriculture and Agri-Food Canada, 6000 C and E Trail, Lacombe, AB T4L 1W1, Canada e Agriculture and Agri-Food Canada, 960 Carling Ave., Ottawa, ON K1A 0C6, Canada b

A R T I C L E I N F O

A B S T R A C T

Keywords: IFSM Climate change Whole-farm model N footprint C footprint Environmental performance

Climate change is expected to increase agricultural productivity in Canada and in other northern countries but this increase will likely affect the environmental performance of dairy farms, one of the most important agricultural sectors in Canada. The objective of this study was to project the impact of climate change on the agronomic and environmental performance of a virtual dairy farm in each of three climatically contrasting areas of Canada through near future (2020–2049) and distant future (2050–2079) periods, using the Integrated Farm System Model (IFSM) and three climate models (CanESM2, CanRCM4, and HadGEM2). Under future climate conditions and relative to a reference period (1971–2000), projected yields of perennial forages and warmseason crops increased, whereas those of small-grain cereals decreased slightly. Projected ammonia emissions increased on virtual farms of the three areas and in all future scenarios (+ 18% to + 54%). Methane emissions from manure storage increased (+ 26% to +120%), whereas those from enteric fermentation and field manure application decreased. Projected farm N2O emissions changed only slightly relative to the reference period. Fossil fuel CO2 emissions related to field operations increased slightly, due to a larger number of forage cuts per year in future scenarios, but CO2 emissions related to grain drying decreased substantially. Projected losses of P increased on virtual farms of the three areas. The projected reactive N footprint of dairy farms in future scenarios varied more (− 15% to +46%) relative to the reference period than the C footprint (−5% to +9%). Although greenhouse gas mitigation should be a priority for dairy farms under future climate conditions, it should not overshadow the need for strategies to reduce reactive N losses.

1. Introduction Dairying in Canada and elsewhere is known to have significant environmental impacts. Total greenhouse gas (GHG) emissions from Canadian milk production, based on annual milk production of 81.8 million hL in 2015 (Canadian Dairy Information Centre, 2015), can be estimated at 8.4 Mt CO2 equivalent (CO2eq) (Quantis Canada et al., 2012). Nevertheless, the environmental impact of dairy farms is not limited to GHG emissions, as the dairy sector also generates NH3

emissions (Sheppard et al., 2011b) and contributes to water pollution through nitrate and P losses (Paul and Zebarth, 1997; Simard et al., 1995), as has been demonstrated in Canadian studies. Although a number of recent studies proposed mitigation measures for reducing the environmental impact of Canadian dairy farms (Hawkins et al., 2015; Chai et al., 2016; Jayasundara et al., 2016), little research has sought to project how climate change will affect the environmental performance of Canadian dairy farms in the future. Based on projections derived for many northern regions in the world (Tatsumi

Abbreviations: CAB, Central Alberta; CanESM2, Canadian Centre for Climate Modelling and Analysis Earth System Model; CanRCM4, Canadian Regional Climate Model; CHU, crop heat unit; CO2eq, CO2 equivalent units; DF, distant future; DM, dry matter; FPCM, fat- and protein-corrected milk; GDD, growing degree day; GHG, greenhouse gas; HadGEM2, Hadley Centre Global Environment Model version 2; IFSM, Integrated Farm System Model; NF, near future; QE, Quebec East; QSW, Quebec Southwest; RCP, representative concentration pathway; TAN, total ammoniacal nitrogen ⁎ Corresponding author. E-mail addresses: [email protected] (M.-N. Thivierge), [email protected] (G. Jégo), [email protected] (G. Bélanger), [email protected] (M.H. Chantigny), [email protected] (C.A. Rotz), [email protected] (É. Charbonneau), [email protected] (V.S. Baron), [email protected] (B. Qian). http://dx.doi.org/10.1016/j.agsy.2017.07.003 Received 3 August 2016; Received in revised form 29 June 2017; Accepted 3 July 2017 0308-521X/ Crown Copyright © 2017 Published by Elsevier Ltd. All rights reserved.

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

et al., 2011), climate change can be expected to have a positive impact on crop productivity in Canada given the expected increased CO2 concentration, warmer temperature, and longer growing season (Qian et al., 2016a, 2016b; Smith et al., 2013; Wang et al., 2012). However, Canadian studies using simulation models have projected an increase in N2O emissions from crop production systems as a consequence of higher N rates required to support expected greater crop yield (Smith et al., 2013), as well as an increase in annual NO3 losses from an agricultural watershed due to expected increase in precipitation (Dayyani et al., 2012). Climate change can reasonably be expected to affect other environmental emissions as well, since it is known, for example, that higher temperatures increase NH3 and CH4 emissions from manure (Sheppard et al., 2011b; Jayasundara et al., 2016) and that an increase in precipitation intensity leads to higher P losses (Messing et al., 2015). A better understanding of the overall agronomic and environmental effects of changes in temperature, precipitation, and atmospheric CO2 concentrations on dairy farming through modelling would enable the identification of the best suited mitigation measures and adaptation strategies for sustainable production in the future (Rotz et al., 2016). A dairy farm is a complex system, and comprehensive whole-farm simulations are required to describe the internal cycling of nutrients on the farm and the nutrient exchange that occurs between the farm and its environment (Schils et al., 2007). Several farm-scale models have been developed in recent years, such as DairyWise in the Netherlands (Schils et al., 2007), WFM (Whole-Farm Model) in New Zealand (Beukes et al., 2008; Wastney et al., 2002), GAMEDE (Global Activity Model for Evaluating the sustainability of Dairy Enterprises) in France (Vayssières et al., 2009a, 2009b), and the Integrated Farm System Model (IFSM) in the United States (Rotz et al., 2015). The IFSM is the only process-based farm-scale model that has been developed to represent dairy, beef, and cash-crop farms in the temperate regions of the northern United States and southern Canada. The model provides an assessment of the economic and environmental sustainability of dairy farms (Rotz et al., 2014). The model's components include crops and soils, harvest and storage, animal feeding, manure storage and handling, and economic analysis (Rotz et al., 2015). Jégo et al. (2015) previously showed that IFSM can be used to simulate the current yield and nutritive value of perennial forage crops and annual crops in eastern Canada. Thivierge et al. (2016) used IFSM to simulate the future yield and nutritive value of an alfalfa (Medicago sativa L.) and timothy (Phleum pratense L.) mixture in eastern Canada. Environmental losses simulated by IFSM (e.g. NH3 and GHG emissions; N and P losses to water) have been compared with reports in the literature and with farm measurements and have been found to be in the realistic range (Chianese et al., 2009a, 2009b, 2008; Rotz et al., 2014, 2011). The objective of this study was to examine the projected impact of climate conditions in the near (2020–2049) and distant (2050–2079) future on the agronomic and environmental performance of one virtual dairy farm in each of three climatically contrasting areas of Canada, by using IFSM with three climate models and under two representative concentration pathways (RCP 4.5 and 8.5). The main hypotheses were that under future climate conditions, (1) yield would increase for most crops except for small-grain cereals, (2) emissions of N2O, CH4, and NH3 in the atmosphere as well as losses of N and P in water through runoff and leaching would increase, and (3) N and C footprints would increase, particularly in the distant future.

(Fig. 1). For each virtual farm, daily minimum and maximum air temperatures, precipitation, and solar radiation were retrieved from the nearest weather stations for the 1971–2000 reference period (Fig. 1). The impact of climate change on dairy farms was studied by comparing IFSM predictions derived from synthetic climate data representative of the reference period (1971–2000) with predictions derived from synthetic climate data for the near future (NF; 2020–2049) and the distant future (DF; 2050–2079). For both of these future periods, two radiative forcing scenarios of atmospheric GHG concentration were applied: representative concentration pathways (RCP) 4.5 and 8.5. In RCP 4.5, GHG emissions increase only slightly until around 2040 and decline thereafter, while in RCP 8.5, GHG emissions keep increasing over time (IPCC, 2014). Both RCP 4.5 and 8.5 lead to an increase in atmospheric CO2 concentration in the future, but to a greater extent in RCP 8.5. The four future scenarios investigated in the present study are identified hereafter as NF4.5 and NF8.5 (near future with RCP 4.5 and RCP 8.5, respectively) and DF4.5 and DF8.5 (distant future with RCP 4.5 and RCP 8.5, respectively). Atmospheric CO2 concentrations averaged 346 μmole mol− 1 for the reference period; 447 and 469 μmole mol− 1 for scenarios NF4.5 and NF8.5, respectively; and 514 and 639 μmole mol− 1 for scenarios DF4.5 and DF8.5 (RCP Database version 2.0.5; Meinshausen et al., 2009). Climate scenarios used in this study were developed based on climate change simulations by three climate models: (1) the second-generation Canadian Centre for Climate Modelling and Analysis Earth System Model (CanESM2) (Arora et al., 2011) and (2) the Hadley Centre Global Environment Model version 2 (HadGEM2) (Johns et al., 2006; Martin et al., 2006; Ringer et al., 2006), which are global climate models, and (3) a newly developed Canadian Regional Climate Model (CanRCM4) (Scinocca et al., 2015; Qian et al., 2016b). To obtain acceptable estimates of climate risks, series of 300-yr synthetic weather data each representing a 30-yr period were generated for each of these models and RCPs, as well as for the reference period, using the stochastic weather generator AAFC-WG (Hayhoe, 2000; Qian et al., 2016b, 2004). The 300-yr synthetic weather data were then used to run the IFSM model. All three climate models used in the present study account for the greater likelihood of occurrence of extreme events in the future as determined by the dynamical processes in the models, but regional climate models like CanRCM4 are often more reliable than global climate models when it comes to simulating extremes at the regional scale. Changes in the likelihood of occurrence of extreme events are accounted for in the climate scenarios we used in this study since the stochastic weather generator AAFC-WG is able to reproduce historical climate extremes and to project changes in the future (Qian et al., 2008; Qian et al., 2010). Finally, because IFSM uses daily climate data as an input, it can account for some extreme events (e.g. very high temperatures, drought or soil water saturation). However, other extreme weather events such as hail or wind gusts are beyond the scope of this model.

2. Materials and methods

Average daily GDDs = Tmean − 5.0 (if Tmean < 5.0, GDDs=0.0)

2.1. Climate scenarios and weather data

Average daily CHUs = [1.8(Tmin − 4.4) + 3.33(Tmax − 10)

2.2. Projected climate conditions Table 1 describes the projected climate characteristics of the three virtual farms (CAB, QSW, and QE), derived from averaging the results from the three climate models. The average daily growing degree-days (GDD) or crop heat units (CHU) were calculated as follows:

− 0.084(Tmax − 10)2 ] 2

A virtual dairy farm was created for each of three climatically contrasting agricultural areas in Canada: Central Alberta (CAB) in the Prairies Ecozone, Quebec Southwest (QSW) in the Mixedwood Plains Ecozone, and Quebec East (QE) in the Atlantic Maritime Ecozone

where Tmean is the average daily temperature in degrees Celsius, Tmin is the daily minimum temperature set at 4.4 °C if < 4.4 °C, and Tmax is the daily maximum temperature set at 10 °C if < 10 °C, as per Brown and 242

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Fig. 1. Map of the terrestrial ecozones of Canada showing the locations of the three virtual farms representative of the Central Alberta (CAB), Quebec Southwest (QSW), and Quebec East (QE) areas along with the nearest weather stations.

crops on the farm minus the precipitation—increased for all farms in the distant future (+4 to +62 mm) (Table 1). A positive value for the precipitation deficit means that precipitation may not meet water demand for evapotranspiration (Bootsma et al., 2005b; Qian et al., 2013).

Bootsma (1993). Average temperature and the accumulation of GDD and CHU during the growing season increased in all future scenarios, with the largest increase found in scenario DF8.5 and the smallest increase in scenario NF4.5 (Table 1). The occurrence of days with high temperatures for cool-season crops was computed as the number of days when the daily maximum temperature reached at least 28 °C during the growing season, that is, the critical temperature at which yields are considered to decline in barley (Robertson et al., 2013) and wheat (Smith et al., 2013). This number of days with high temperatures increased in all future scenarios (+11 to +61 d, depending on the virtual farm and scenario), as reported from several studies (An and Carew, 2015; Qian et al., 2016b; Smith et al., 2013; Wang et al., 2012). The start of growth for perennial forage crops was defined as the date when the 5-d moving average of daily mean air temperature reached the base temperature of 5 °C, as per Qian et al. (2013); this date fell up to 27 d earlier in the future. The planting date for annual crops was simulated with IFSM, taking into account temperature, soil moisture conditions, and availability of machinery and labor. In future scenarios, it occurred earlier for corn (6 to 16 d), soybean (5 to 13 d), and coolseason annual crops (6 to 21 d). The first fall frost was delayed by 11 to 24 d. Cumulative precipitation during the growing season (April to October) increased for all virtual farms and future scenarios (+23 to + 73 mm) (Table 1). The increase was greater in eastern Canada (QSW and QE: + 39 to + 73 mm) than in western Canada (CAB: +23 to + 44 mm), which is consistent with the findings of Qian et al. (2013) and Smith et al. (2013). Despite the overall increase in precipitation, the projected increase in evapotranspiration during summer months may lead to more water stress (Thivierge et al., 2016; Qian et al., 2013). Potential evapotranspiration was calculated using the equation proposed by Priestley and Taylor (1972). From April to October, potential evapotranspiration increased more in the distant future (+38 to + 103 mm) than in the near future (+12 to +48 mm). Consequently, the precipitation deficit—the total potential evapotranspiration from all

2.3. Integrated Farm System Model The IFSM (version 4.2) is made up of nine main components that represent major processes on the farm: crop and soil, grazing, machinery, tillage and planting, crop harvest, crop storage, herd and feeding, manure storage and handling, and economic analysis (Rotz et al., 2015). Process-level simulation represents the interactions among components (e.g. crop harvest, animal feeding, and manure handling). The crop component is made up of five submodels for simulating purestand alfalfa, perennial grass (pure-stand and mixture), corn, small grains, and soybean. Daily potential total biomass accumulation is a function of solar radiation, day length, air temperature, atmospheric CO2 concentration, and crop leaf area. This daily accumulation is limited by four stress factors (temperature, soil moisture, soil N availability, and plant stored reserves) for determination of the actual daily biomass accumulation. Grazing can also be simulated but it was not done in this study because it is not a common practice on most Canadian dairy farms. A general soil model is used to predict the tractability of soil for field operations and the moisture and N availability for the growth and development of each crop. Precipitation, runoff, evapotranspiration, moisture migration, and drainage are tracked through time to predict the moisture content in multiple layers of the soil profile. Nitrogen movement and transformation within and among soil layers are modeled with functions mostly from the DAYCENT model (DAYCENT, 2007) along with some from the Nitrate Leaching and Economic Analysis Package (NLEAP) model (Shaffer et al., 1991). The machinery component of IFSM is used to determine the performance and resource-use rates for all machinery operations on the farm. Tillage, planting, and crop-harvest operations occur on days within a specified 243

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Table 1 Climate characteristics of the three virtual farms representing the Central Alberta (CAB), Quebec Southwest (QSW), and Quebec East (QE) areas for the reference period (1971–2000), and projected changes (increase [+] or decrease [−] relative to the reference period) in near future (NF; 2020–2049) and distant future (DF; 2050–2079) periods under representative concentration pathways 4.5 and 8.5 (averaged over 300 simulated years and the climate models CanESM2, CanRCM4, and HadGEM2). Climate characteristics

Temperature (average, annual) Temperature (average, Apr. to Oct.) - April - May - June - July - August - September - October High temp. daysa (Apr. to Oct.) First fall frost (< 0 °C) Precipitation (annual) Precipitation (Apr. to Oct.) Potential evapotranspirationc (Apr. to Oct.) Precipitation deficitd (Apr. to Oct.) Growing degree-day accumulation (GDDe base 5 °C, Apr. to Oct.) Crop heat unit (CHU) accumulationf Growth start for perennial foragesg Beginning of planting date – annual coolseason cropsh,i Beginning of planting date – corni Beginning of planting date – soybeani

Unit

CAB

QSW

Ref.

NF4.5

NF8.5

DF4.5

DF8.5

Value

Change from ref. (+/−)

°C °C °C °C °C °C °C °C °C No DOYb mm mm mm mm °C-d

2.3 10.5 4.1 10.4 14.0 15.9 14.8 10.3 4.0 7 252 486 398 577 179 1351

2.6 2.3 2.1 1.4 2.0 2.7 3.0 2.4 2.2 15 14 45 35 13 − 22 406

2.7 2.3 1.7 1.7 2.2 2.9 3.2 2.2 2.2 17 13 31 23 12 − 11 424

3.8 3.5 2.5 2.3 3.7 4.6 4.8 3.6 3.0 34 19 43 23 38 15 662

CHU DOY DOY

1884 111 131

623 − 10 −9

658 −8 −6

DOY DOY

133 N/Aj

−6 N/A

− 11 N/A

QE

Ref.

NF4.5

NF8.5

DF4.5

Value

Change from ref. (+/−)

5.1 4.8 3.6 3.4 4.5 6.1 6.6 5.1 4.4 47 24 74 44 48 4 918

6.0 14.0 5.5 13.1 17.9 20.6 19.1 14.2 7.8 22 277 1048 625 619 −6 2008

2.6 2.2 1.7 2.0 2.0 2.4 2.2 2.9 2.2 24 11 84 49 47 −2 438

2.6 2.4 2.0 1.8 2.2 2.6 2.3 3.1 2.6 25 12 81 55 48 −7 475

3.9 3.5 3.3 2.7 3.3 4.0 3.8 4.2 3.4 41 14 102 39 74 35 700

970 − 13 − 12

1319 − 17 − 18

3154 105 125

569 − 10 − 10

697 −9 − 11

−7 N/A

− 16 N/A

138 144

−6 −8

−6 −5

DF8.5

Ref.

NF4.5

NF8.5

DF4.5

DF8.5

Value

Change from ref. (+/−)

5.2 5.1 4.1 4.2 4.8 5.7 5.7 6.0 5.0 61 18 147 41 103 62 1030

3.1 10.7 1.6 8.5 14.3 17.5 16.1 11.3 5.3 7 272 922 551 544 −8 1393

2.6 2.2 2.0 2.0 2.1 2.5 2.3 2.4 1.8 11 12 97 61 43 −18 388

2.7 2.4 2.0 2.0 2.4 2.8 2.5 2.8 2.1 12 14 75 43 45 2 427

3.9 3.4 3.5 3.0 3.3 4.1 3.6 3.5 3.0 22 17 123 60 70 11 619

5.4 5.0 4.5 4.3 4.9 5.9 5.5 5.4 4.6 38 22 177 73 97 24 939

975 − 19 − 16

1375 − 25 − 19

2116 122 136

618 −10 −11

676 −9 −9

936 − 18 − 18

1375 − 27 − 21

− 10 − 10

− 12 − 13

146 N/A

−9 N/A

−7 N/A

− 11 N/A

− 18 N/A

a

Number of days when the maximum temperature reached at least 28 °C. DOY: day of the year. Calculated by the STICS model as per Priestley and Taylor (1972). d Precipitation deficit: potential evapotranspiration minus precipitation. e GDD: growing degree-days. f Crop heat units were accumulated from the last day of three consecutive days with mean daily air temperatures greater than or equal to 12.8 °C to the day of the first fall frost with a minimum temperature equal to or less than −2 °C, as per AAF (2015). g The start of the growing season for perennial forages was estimated as the last day of five consecutive days with a 5-d moving average daily mean air temperature greater than or equal to a base temperature of 5 °C, as per Qian et al. (2013). h Annual cool-season crops are barley in CAB and QE and wheat in QSW. i Beginning of planting dates were simulated with IFSM (Integrated Farm System Model), taking into account temperature, soil moisture conditions, and availability of machinery and labor. j Not applicable, because this crop was not grown on this farm. b c

time period when the soil and weather conditions are suitable for field work. The rate at which work is completed and the subsequent fuel use and labor requirements are all determined based on the size and type of equipment specified for each field operation. After harvest, a number of different storage options are provided for dry grain, high-moisture grain, dry hay, and ensiled forage. Feed allocation and animal response are related to the nutritive value of available feeds and the nutrient requirements of up to six animal groups making up either dairy or beef herds. The diet for each group is formulated using a cost-minimizing linear programming approach, which makes the best use of homegrown feeds and purchased supplements. The quantity and nutrient content of the manure produced are a function of the quantity and nutrient content of the feeds consumed. The manure composition, the type and size of storage, and the air temperature are the main drivers controlling gaseous losses that result from manure storage. Nutrient flows through the farm are modeled to predict potential nutrient accumulation in the soil and loss to the environment. The economic-analysis component uses a whole-farm budget that considers important fixed and variable production costs and the income from products sold, but that type of analysis is not discussed in this study. The system boundaries are set as the farm boundaries, i.e. cradle to farm gate (Rotz et al., 2015). Climate has many impacts on crop growth and development, machinery operations, crop losses, animal performance, and manure emissions. More details about simulated processes can be found in the appendix and in the IFSM reference manual (Rotz et al., 2015).

2.4. Virtual dairy farms The characteristics of virtual dairy farms representative of each agricultural area (Table 2) were determined in collaboration with a panel of experts on dairy systems and forage crops, and supplied to IFSM. Cows' annual milk production was averaged from 2009 to 2013 (QSW and QE, Robert Moore, Valacta, 2016, pers. comm.; CAB, Heikkila and Van Biert, 2014) and was reduced by 4% to account for on-farm milk losses (Édith Charbonneau, Université Laval, 2016, pers. comm.). Cows were housed in a tie-stall barn in QSW and QE and in a free-stall barn in CAB—the most common housing facilities for these respective agricultural areas (Sheppard et al., 2011a). Farm characteristics and daily weather data at a particular location were supplied to IFSM as input information. The direct effect of elevated CO2 on crops, called the fertilization effect, was included by modelling plant C fixation as a function of atmospheric CO2 concentration. The following soils were chosen because they represent a substantial proportion of the areas studied: a Penhold loam (Orthic Black Chernozem/Udic Boroll) in CAB, a St. Jude sandy loam (Gleyed HumoFerric Podzol/Mixed, Frigid Haplorthod) in QSW, and a St. André sandy loam (Orthic Humo-Ferric Podzol/Mixed, Frigid Haplorthod) in QE. The characteristics of the top soil layer (0 to 15 cm) (Table 2) were obtained from the Canadian Soil Information Service (CanSIS, 2015). The following cultivated crops were grown: pure alfalfa (3-yr stand life), silage corn (Zea mays L.), and grain and silage barley (Hordeum 244

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Table 2 Characteristics of the three virtual dairy farms representing the Central Alberta (CAB), Quebec Southwest (QSW), and Quebec East (QE) areas. Farm characteristics

Unit

CAB

QSW

QE

Animal production Number of lactating cows Milk production target (4% fat- and 3.3% protein-corrected milk, FPCM) Manure produced Manure nutrient concentrationa

Number L milk cow− 1 yr− 1 kg FPCM cow− 1 yr− 1 Mg DM %N–%P–%K

140 8610 8749 683 3.4–1.3–2.4

87 8491 8628 430 3.7–1.4–2.8

72 8282 8416 383 4.3–1.2–3.2

Cultivated area Perennial forages for silage Corn for silage Corn for grain Wheat for grain Barley for grain Barley for silage Soybean Total

ha ha ha ha ha ha ha ha

36 88 – – 65 35 – 224

28 36 36 7 – – 36 143

100 25 – – 25 – – 150

kg N ha− 1 kg N ha− 1 kg N ha− 1 kg N ha− 1 kg N ha− 1 kg N ha− 1 kg P2O5 ha− 1 kg P2O5 ha− 1 kg P2O5 ha− 1 kg P2O5 ha− 1 kg P2O5 ha− 1 kg P2O5 ha− 1

0 N/A 115 (17%)d N/A 75 (27%) N/A 20 (100%) N/A 60 (0%) N/A 35 (0%) N/A

N/Ac 106 (0%) 140 (36%) 70 (100%) N/A 0 N/A 65 (0%) 60 (0%) 40 (100%) N/A 10 (100%)

N/A 94 (0%) 120 (67%) N/A 70 (100%) N/A N/A 42 (0%) 39 (53%) N/A 40 (100%) N/A

Soil characteristics Predominant soil texture in IFSMe (surface layer)

%

Medium loam

Medium sandy loam

Silt–clay–sand concentration Organic C concentration Available water-holding capacity

% mm %

41–23–36 3.8 230

Medium sandy loam 35–12–53 1.8 90

Fertilization N fertilization from liquid dairy manureb or mineral fertilizer in the reference period

P fertilization from liquid dairy manure (M) or mineral fertilizer (F) in the reference period

Pure alfalfa Alfalfa–timothy mixture Corn Wheat Barley Soybean Pure alfalfa Alfalfa–timothy mixture Corn Wheat Barley Soybean

29–12–59 2.2 64

a

Manure nutrient concentration is expressed after losses in housing facilities and manure storage. Nitrogen fertilization from manure is expressed as available N. We assumed that available N from liquid dairy manure (C/N of 10, 31% N-NH4+) incorporated within 2 days in a medium loam or medium sandy loam soil was, respectively, 75% and 45% of total N for spring and fall applications to annual crops and 75% and 50% for spring and fall applications to perennial forage crops (CRAAQ, 2010). Manure was applied in the spring (50%) and fall (50%). c Not applicable, because this crop was not grown on this virtual farm. d Proportion (%) of N and P supplied by mineral fertilizer. e IFSM: Integrated Farm System Model. b

subsequent cut (Bélanger et al., 1999; Bootsma, 1984) in the QSW and QE areas. According to local experts working with farmers in the CAB area, nutritive value is generally prioritized over yield. Therefore, the minimum GDD accumulation between alfalfa cuts was set to 450 °C-d. For the reference period, there were two forage harvests per year in CAB and QE, and three in QSW. Excreta were removed daily from housing facilities and stored as liquid manure (8% to 10% dry matter [DM]) in a bottom-loaded tank with a 9-month storage capacity. Manure was incorporated within 2 d after application to the farm's cultivated area. There was no manure exporting off-farm. Mineral N and P fertilizers were applied to satisfy the N and P requirements in accordance with local recommendations [CRAAQ (2010) in QSW and QE; AAFRD (2004) in CAB], assuming soils had a P-to-Al ratio between 2.5 and 10. Amounts of N and P applied to each crop and farm are presented in Table 2. Conventional soil tillage was representative of the QSW and QE areas, with moldboard plowing in the fall, and one pass of a tandem disk and of a field cultivator in the spring prior to seeding. In CAB, reduced tillage was the most representative method of soil preparation and consisted of only one pass of a field cultivator before planting to deal with residues. Given the small area devoted to perennial forage crops and the resulting small proportion of perennial forages in the cow diet in CAB, 25% of the forage (from annual and perennial crops) in the ration was provided as hay, which was considered to be purchased, a common

vulgare L.) in CAB; alfalfa–timothy grass mixture (4-yr stand life), grain and silage corn, soybean (Glycine max [L.] Merrill), and grain wheat (Triticum aestivum L.) in QSW; and alfalfa–timothy grass mixture (4-yr stand life), silage corn, and grain barley in QE. The cultivated areas for each crop are specified in Table 2. The IFSM model has been shown to adequately simulate crop yield under current climate conditions in northern regions of North America, including eastern Canada (Jégo et al., 2015). Simulated yields of annual crops for the reference period were adjusted using the earliest reliable recorded yield data: 1995 to 2013 in the QSW and QE areas (FADQ, 2015), and 2011 to 2014 for grain barley (AFSC, 2015) and 2004 to 2014 for silage barley (AAF, 2014) in the CAB area. Silage corn yields were based on data taken from field experiments in the CAB area (Baron et al., 2006; Stanton et al., 2007). The forage species growth parameters used in IFSM were previously calibrated and evaluated by Jégo et al. (2015), and yields in the reference period were validated with a panel of experts as no official records were available for these species. In all three areas, forage crops were harvested as silage, all crops were rainfed, and no grazing was considered. As mentioned by Brassard and Singh (2007), this type of study emphasizes trends in the performance of dairy farms under a changing climate, and absolute yield values are therefore less important than the trends. Dates on which each harvest of perennial forages could begin were determined by considering a minimum accumulation of 450 °C-d (GDDs above + 5 °C) before the first cut and 520 °C-d between each 245

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

by animal genetics but negatively affected by heat stress on animals under a warmer climate. The choice of crops and the land area for each crop were kept constant in projected future scenarios. Planting dates of annual crops (corn, soybean, wheat, and barley) were adjusted in future scenarios based on the earlier start of the growing season, and corn relative maturity index and soybean maturity group were adjusted based on the increase in CHUs. Cultivars of wheat, barley, and forage crops remained unchanged in the future. Future genetic improvement of crops was not considered. Forages were harvested according to the previously stated GDD requirements, leading to a greater number of cuts per year in future scenarios (Thivierge et al., 2016). Finally, mineral N fertilization was adjusted in each future scenario to compensate for the additional exports of N associated with yield increases for each annual crop and for increased manure N losses during collection, storage, and land application. As an example, total N applied to corn increased by up to 67% in CAB, 45% in QSW, and 21% in QE. Since perennial forage crops included legumes at all three virtual farms, the projected increase in N exports associated with yield increases in the future was not offset by additional N fertilizer application. Phosphorus fertilization was not adjusted, because soils from those areas are relatively rich in P, and an increase in P export with crop harvest would not result in a P deficiency. All other parameters were kept constant between the reference period and future scenarios, and no technological development was taken into account. The effect of weeds, insects, and diseases on yield was not simulated in this study. Land use change was not considered: IFSM does not allow for long term sequestration or depletion of soil C. Accordingly, the potential increase in soil respiration leading to the loss of soil organic C under future scenarios was not taken into account, although some studies demonstrate that this will likely occur in northern countries such as Canada (Smith et al., 2009).

practice in Central Alberta. In QSW and QE, 10% of the forage in the ration was provided as hay (purchased). The herd was grouped for feeding into early-, mid-, and late-lactation cows, dry cows, and young and older heifers. The proportion of forage (consisting of silage from annual and perennial crops and hay) in a cow's daily feed DM intake was adjusted to 53% when averaged across groups of lactating cows and to 77% when averaged across all animal groups. Canola seed meal was selected as the crude protein supplement, and distillers' grain was selected as a less degradable protein and energy supplement. Wheat, barley, and corn harvested on the farm as grain or silage were primarily used as animal feed, and straw was used as animal bedding. Surplus crops and all soybeans were sold off farm. In QSW and QE, grains were fed to cows using a computerized automatic feeder, whereas silage and hay were hand-fed. In CAB, grain and silage were mixed and fed as a total mixed ration using a mixer wagon, whereas hay was hand-fed. 2.5. Projected performance parameters Projected parameters for agronomic and environmental performance are presented as the average of the three climate models. Agronomic performance was assessed by simulating crop yield and feed production. Environmental performance was assessed by simulating direct losses from the farm, including N and P losses through leaching and runoff as well as gaseous emissions (NH3, CH4, CO2, N2O, and other nitrogen oxides released during combustion of fossil fuels). Only anthropogenic CO2 emissions were included, that is, CO2 emissions from fossil fuel combustion. Environmental footprints, including C and reactive N losses, reveal the emission intensity of milk production and were expressed per kilogram of fat- and protein-corrected milk (FPCM) using a standard milk fat content of 4.0% and protein content of 3.3%. The reactive N loss footprint (g N kg− 1 FPCM) (hereafter called N footprint) was defined as the sum of all reactive N losses to the environment, and the C footprint (kg CO2eq kg− 1 FPCM) was defined as the sum of all GHG emissions, expressed in CO2 equivalent units, associated with the production of 1 kg of FPCM. In IFSM (version 4.2), the global warming potential values used to convert CH4 and N2O into CO2 equivalent units are 25 and 298 CO2eq kg− 1, respectively (IPCC, 2001). Dairy farms are regarded as a production system for milk and, therefore, only the environmental emissions associated with milk production are included in C and N footprints. This means that field-related environmental losses included in the footprints are calculated only for the portion of the cropped area used to feed cows. Indeed, when extra feed leaves the farm, the exported material becomes part of a different production system, as do the emissions associated with its production. On the other hand, all upstream emissions for the production of farm resource inputs (fuel, electricity, machinery, fertilizers, seeds, pesticides, and imported feed and animals) ultimately devoted to milk production are included in the environmental footprints of the dairy farm. On a dairy farm, animal products other than milk are produced, like calves and cull cows used for meat production. In IFSM, a biophysical allocation approach is used to determine which proportion of the resource inputs and environmental impact is allocated specifically to milk. This allocation approach is based on the physiological requirements for milk production and growth, including pregnancy (Rotz et al., 2015). In the present study, the allocation to milk ranged between 83.5% and 85.7% of the total environmental impact.

3. Results 3.1. Agronomic performance The yield of all silage crops (corn, barley, and perennial forage crops) increased in future scenarios (Table 3). Among perennial forage crops, pure alfalfa in CAB had the largest yield increase in future scenarios (+30% to + 62%), followed by the alfalfa–timothy mixture in QE (+ 23% to + 35%) and QSW (+ 3% to + 19%). The increase in silage corn yield was larger in CAB and QE (+ 46% to +84%) than in QSW (+ 15% to + 19%). Soybean and grain corn were grown only in QSW. Soybean yield increased in all future scenarios (+32% to + 45%), whereas grain corn yield increased in the near future only (+ 8% to +10%) and even decreased in DF8.5 (− 8%). The grain yield of cool-season cereals decreased slightly in all future scenarios in CAB (− 5% to − 11% for barley) and QSW (− 3% to −14% for wheat) but remained almost unchanged in QE (Table 3). Because surplus crops were considered to be sold off farm, the yield increase for silage crops (corn, barley, and perennial forage crops) resulted in a large increase in silage sales for the three farms under all future scenarios (Table 3). This increase in silage sales was larger in CAB (+ 505 to + 806 Mg DM farm− 1) than in QE (+308 to + 376 Mg DM farm− 1) and QSW (+ 15 to + 131 Mg DM farm− 1). Grain sales also increased in QSW (+27 to + 66 Mg DM farm− 1), because of grain corn and soybean yield increases. In CAB, lower barley grain yield caused an increase in purchases of grain in order to meet animal feeding requirements (Table 3). The amount of concentrates in the diet of cows (canola seed meal, distillers' grain, and mineral and vitamin mix) remained unchanged in all scenarios.

2.6. Assumptions Because this study was not aimed at assessing adaptation strategies, only minor changes were made to the farm parameters for future periods. Annual milk production per farm was kept constant in all future scenarios because of the existence of milk quotas in Canada. Milk yield per cow (Table 2) was also kept constant. It was not possible to predict its evolution over time, considering that it could be positively affected

3.2. Environmental performance Environmental emissions were projected for the whole farm area, including emissions related to crops that were projected to be sold off 246

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Table 3 Crop yield and feed production for the three virtual farms representing the Central Alberta (CAB), Quebec Southwest (QSW), and Quebec East (QE) areas for the reference period (1971–2000), and projected changes (increase [+] or decrease [−] relative to the reference) in near future (NF; 2020–2049) and distant future (DF; 2050–2079) periods under representative concentration pathways of 4.5 and 8.5 (averages over 300 simulated years and the climate models CanESM2, CanRCM4, and HadGEM2). Parameters

CAB (140 lactating cows, 224 ha) Ref.

NF4.5

Value

Change from ref. (+/−)

Yield (Mg DM ha ) Barley (dry grain) Barley (silage) Wheat (dry grain) Corn (silage) Corn (dry grain) Soybean (dry grain) Forage (pure alfalfa) Forage (alfalfa–timothy mixture)

3.3 6.6 N/A 11.5 N/A N/A 7.7 N/A

−0.4 0.1 N/A 6.1 N/A N/A 3.3 N/A

−0.2 0.6 N/A 8.2 N/A N/A 3.1 N/A

− 0.1 0.2 N/A 5.8 N/A N/A 2.3 N/A

Feed production (Mg DM farm− 1) Silage from perennial forages produced Silage from perennial forages sold Silage from annual grain crops produced Silage from annual grain crops sold Dry grain produced Dry grain purchased Dry grain sold (including soybean) Stover and straw produced Hay purchased

125 106 1038 426 195 10 0 196 154

79 80 423 459 −5 5 0 14 −4

72 73 399 432 −14 15 0 7 −5

50 53 457 491 − 26 28 0 6 −7

a

a b

NF8.5

QSW (87 lactating cows, 143 ha)

DF4.5

DF8.5

Ref.

NF4.5

NF8.5

DF4.5

Value

Change from ref. (+/−)

− 0.2 0.2 N/A 5.3 N/A N/A 4.8 N/A

N/Ab N/A 2.8 14.9 7.3 2.3 N/A 8.2

N/A N/A −0.1 2.5 0.6 0.8 N/A 1.4

N/A N/A −0.1 2.8 0.7 0.8 N/A 1.5

N/A N/A − 0.4 2.2 0.1 0.8 N/A 0.6

108 110 649 696 − 12 14 0 25 −6

132 9 400 36 349 0 270 25 38

31 17 35 48 65 0 65 1 0

33 34 94 97 48 0 48 1 0

12 3 9 12 54 0 55 1 1

QE (72 lactating cows, 150 ha) DF8.5

Ref.

NF4.5

NF8.5

DF4.5

DF8.5

Value

Change from ref. (+/−)

N/A N/A − 0.3 2.5 − 0.6 1.0 N/A 0.2

2.3 N/A N/A 10.3 N/A N/A N/A 6.6

− 0.1 N/A N/A 6.7 N/A N/A N/A 1.8

− 0.1 N/A N/A 6.4 N/A N/A N/A 1.7

0.0 N/A N/A 8.6 N/A N/A N/A 1.5

0.0 N/A N/A 7.9 N/A N/A N/A 2.3

−1 10 43 26 27 0 27 2 1

427 158 207 62 53 13 0 100 32

161 183 123 143 −2 0 0 5 −2

151 172 114 136 −2 0 0 5 −2

121 145 157 171 0 −1 0 13 −2

194 215 139 161 1 −1 0 17 −2

−1

DM: dry matter. Not applicable, because this crop was not grown on this farm.

CH4 emissions from enteric fermentation, and manure (27% to 36%), through N2O and CH4 emissions during manure storage and field application. Methane (CH4) emissions in the reference period, averaged across the three virtual farms, were mostly from housing facilities (79.9%, mostly from enteric fermentation) and manure storage facilities (19.9%), with negligible emissions (0.2%) from field application of manure (Table 4). In future scenarios, CH4 emissions from manure storage increased, whereas those from housing facilities and field manure application decreased, resulting in an overall increase,

farm in future scenarios. Ammonia (NH3) emissions in the reference period were mainly from N applied to fields (59%) and, to a lesser extent, from housing (26%) and manure storage (15%) facilities (Table 4). Total farm NH3 emissions increased on all farms and in all future scenarios (+ 18% to +54%). The main GHGs emitted by the three virtual farms during the reference period were CH4 (470 to 757 Mg CO2eq farm− 1, depending on the farm), followed by N2O (238 to 376 Mg CO2eq farm− 1), and CO2 (23 to 83 Mg CO2 farm− 1). The main sources for these GHG emissions were animals (48% to 52%, depending on the farm), mostly through

Table 4 Selected environmental emissions of the three virtual farms representing the Central Alberta (CAB), Quebec Southwest (QSW), and Quebec East (QE) areas for the reference period (1971–2000), and projected changes (increase [+] or decrease [−] relative to the reference) in near future (NF; 2020–2049) and distant future (DF; 2050–2079) periods under representative concentration pathways of 4.5 and 8.5 (averages over 300 simulated years and the climate models CanESM2, CanRCM4, and HadGEM2). Emissions by farm

CAB (140 lactating cows, 224 ha) Ref.

NF4.5

NF8.5

DF4.5

Value

Change from ref. (+/−)

QSW (87 lactating cows, 143 ha) DF8.5

Ref.

NF4.5

NF8.5

DF4.5

Value

Change from ref. (+/−)

QE (72 lactating cows, 150 ha) DF8.5

Ref.

NF4.5

NF8.5

DF4.5

Value

Change from ref. (+/−)

DF8.5

−1

Ammonia (kg NH3 farm ) Housing facilities 2070 Manure storage 1503 Field application 4461

314 430 856

313 437 699

503 761 1147

636 1079 1702

738 1935 4248

225 492 927

228 421 859

365 748 1417

549 997 1561

416 1597 3149

122 490 343

136 553 463

215 749 799

355 1172 1273

Methane (kg CH4 farm− 1) Housing facilitiesa 25,157 Manure storage 5044 Field application 66

− 1609 2276 −4

−1524 2325 −4

− 1371 4073 −5

− 1504 6050 −6

16,774 5759 51

− 180 2215 0

− 314 2319 −3

− 111 3782 −2

−76 5889 −5

15,471 3286 51

− 909 859 −5

− 886 988 −4

− 952 1747 −5

−887 3115 −5

Nitrous oxide (kg N2O farm− 1) Manure storage 676 Farmland 584

− 38 87

−40 71

− 46 85

− 56 146

541 257

−8 16

− 32 6

− 25 17

−55 2

656 261

22 − 57

28 − 49

10 − 74

11 −90

Carbon dioxide (Mg CO2 farm− 1) Fossil fuel combustion 82.9 3.5

2.6

2.0

5.5

65.4

− 29.0

− 26.9

− 31.3

−30.9

23.4

5.7

5.3

5.5

8.5

Losses by leaching and runoff N loss (kg N ha− 1) 1.92 P loss (kg P ha− 1) 0.07

0.58 0.01

0.81 0.02

5.46 0.03

52.43 0.51

− 4.02 0.13

− 2.47 0.19

− 1.23 0.23

−0.26 0.35

21.33 0.13

− 13.54 0.07

− 13.15 0.07

− 12.49 0.09

−14.36 0.17

a

1.47 0.02

Methane in housing facilities mostly comes from enteric fermentation.

247

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Fig. 2. Reactive N losses (g N kg− 1 of 4% fat- and 3.3% protein-corrected milk [FPCM]) from the different contributing sources and the reactive N loss footprint (emissions allocated to milk, g N kg− 1 FPCM) for the three virtual farms representative of the Central Alberta, Quebec Southwest, and Quebec East areas in the reference period (1971–2000) and for the near future (NF; 2020–2049) and distant future (DF; 2050–2079) periods under the representative greenhouse gas concentration pathways 4.5 and 8.5, averaged across the following climate models: Canadian Centre for Climate Modelling and Analysis Earth System Model (CanESM2), Canadian Regional Climate Model (CanRCM4), and Hadley Centre Global Environment Model version 2 (HadGEM2).

winter damage to alfalfa associated with climate change (Bélanger et al., 2002). Moreover, the projected increase in the number of cuts per year in future scenarios could affect regrowth in the following year, as demonstrated for timothy (Jing et al., 2014). In IFSM, this effect is only taken into account for pure alfalfa. Projected increases in silage corn yield in the future were much smaller in QSW than in the colder areas (CAB and QE). Brassard and Singh (2007) also projected greater corn yield increases in areas where climate conditions in the reference period were not the most suitable for corn growth. Indeed, in the reference period, CHU accumulation was not optimal for silage corn in CAB and QE. In the future, the greater CHU accumulation projected for CAB and QE should favor silage corn growth. The large yields projected in future scenarios (up to 19 and 20 Mg ha− 1 in QE and CAB, respectively) are commonly reported at present in the north central United States (Sanford et al., 2016) and therefore seem realistic for Canada in the future. Assuming constant milk production, the large yield increase projected for silage crops (from annual and perennial crops) means that less land area would be needed to produce the same amount of feed for cows in future scenarios. The land made available for other purposes would represent approximately 28% to 36% of the total cropped area in CAB and QE, and 5% to 8% in QSW, the area that would have a smaller crop yield increase. Possible options for the extra land include growing grain crops or bioenergy crops for sale, or using the extra silage to support an increased herd size and milk production. Although these options are not addressed in the present study, they would likely result in an increase in net farm income. The use of an optimization model would be the next step to determine which option would provide the best agronomic, environmental, and economic benefits in the future. Under current climate conditions, grain corn is grown only in the QSW area, because climate conditions in CAB and QE are not suitable for this crop. In the present simulations, despite the use of longer-season hybrids that have more time to accumulate DM (Bootsma et al., 2005b), the projected yield of grain corn in QSW increased only slightly in the near future (+9% on average) and decreased in the distant future (− 3%), unlike the projected yield of silage corn that increased in all scenarios in the same area (+17% on average). This suggests a lower harvest index (ratio of grain to biomass yield) for corn in future scenarios, which could likely be explained by greater temperatures and precipitation deficits projected in the distant future and their effect on the physiological development of corn. Indeed, it has been demonstrated that higher temperatures increase cornstalk biomass without

especially in the distant future (+ 4% to + 26%). Due to the manure handling system, a larger proportion of nitrous oxide (N2O) emissions was from manure storage (64%) than from farmland (36%, including manure application and synthetic fertilizer application). In future scenarios, manure storage N2O emissions changed only slightly relative to the reference period (− 10% to +4%). Farmland N2O emissions increased in CAB (+ 12% to +25%), remained almost unchanged in QSW (+ 1% to + 6%), and decreased in QE (−19% to −35%) relative to the reference period (Table 4). Fossil fuel CO2 emissions increased in CAB (+ 23% to + 36%) and in QE (+2% to + 7%), but decreased in QSW (− 41% to − 48%). Losses of N in water through runoff and leaching increased in CAB, remained relatively constant in QSW, and decreased in QE, and losses of P increased on all three virtual farms (Table 4). Footprints reveal the emission intensity of milk production, and results are expressed per kg of FPCM. Field-related environmental losses are therefore calculated only for the portion of the cropped area used to feed cows. The N footprint in the reference period ranged from 6.6 g N kg− 1 FPCM in CAB to 16.6 g N kg− 1 FPCM in QSW (Fig. 2). In future scenarios, it increased in CAB (+ 15% to + 46%) and QSW (+ 5% to + 16%) and generally decreased in QE (− 15% to 0%). The C footprint in the reference period ranged from 0.95 kg CO2eq kg− 1 FPCM in CAB to 1.20 kg CO2eq kg− 1 FPCM in QE (Fig. 3). The C footprint varied only slightly compared to the reference period in all future scenarios (+ 1% to − 5%) except for the distant future in QSW (+ 5% to + 9%). 4. Discussion 4.1. Crop yield The projected yield increase for the alfalfa–timothy forage mixture in QSW and QE under future climate conditions is consistent with the findings of Thivierge et al. (2016). Indeed, the projected longer growing season and greater GDD accumulation resulted in an increased number of forage cuts per year. The smaller yield increase projected under the most extreme scenario (DF8.5) in QSW was due to greater water and temperature stresses (Thivierge et al., 2016). Thivierge et al. (2016) also reported that the proportion of alfalfa in the mixture would increase in all future scenarios, given that N-fixing species are more responsive to elevated atmospheric CO2 than non-fixing species. For this reason, a large yield increase for pure alfalfa stands was expected in CAB in the future. The IFSM model does not consider the greater risk of 248

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Fig. 3. Greenhouse gas (GHG) emissions (kg CO2 equivalent units [CO2eq] kg− 1 of 4% fat- and 3.3% protein-corrected milk [FPCM]) from the different contributing sources and the C footprint (emissions allocated to milk, kg CO2eq kg− 1 FPCM) for the three virtual farms representative of the Central Alberta, Quebec Southwest, and Quebec East areas in the reference period (1971–2000) and for the near future (NF; 2020–2049) and distant future (DF; 2050–2079) periods under the representative GHG concentration pathways 4.5 and 8.5, averaged across the following climate models: Canadian Centre for Climate Modelling and Analysis Earth System Model (CanESM2), Canadian Regional Climate Model (CanRCM4), and Hadley Centre Global Environment Model version 2 (HadGEM2). The conversion of GHG to CO2 equivalent units was done in the Integrated Farm System Model (IFSM) using a global warming potential of 25 for CH4 and 298 for N2O, as specified in IPCC (2001).

conditions in the present study. Indeed, the projected period to reach maturity under future climate scenarios was shortened by up to 3 d for barley in CAB and QE and by up to 5 d for wheat in QSW, which agrees with the findings of Singh et al. (1998). With the projected earlier planting dates and shorter period to reach maturity, the overall exposure of these crops to heat units should not be greater in future scenarios than in the reference period (Bootsma et al., 2005a) and they should have even less time to accumulate DM (Bootsma et al., 2005b). Moreover, the present study indicated that days with high temperatures should occur mainly in June and July, which is when the flowering and grain-filling stages of cool-season cereals are likely to occur. High temperatures during the flowering and grain-filling stages can cause a severe decline in grain yield in cool-season cereal crops (An and Carew, 2015; Wang et al., 2012). Several previous studies also projected a decrease in yield for wheat in the southwestern area of Quebec (Singh et al., 1998; Tatsumi et al., 2011) and for barley in the eastern area of Quebec (Singh et al., 1998), the Atlantic region of Canada (Bootsma et al., 2005b), and Manitoba (An and Carew, 2015), although those reports did not account for the CO2 fertilization effect. In contrast, a small increase in wheat yield (5% to 9%) was projected for the southwestern area of Quebec in 2040–2069 with the DSSAT (Decision Support System for Agrotechnology Transfer) model (Brassard and Singh, 2007). The DSSAT takes into consideration the enhanced water-use efficiency of crops under elevated atmospheric CO2 concentration, which could result in greater yield due to lower plant water requirements and water stress in future scenarios. Projections generated by IFSM could be improved by taking into account the enhanced crop water-use efficiency under future climate conditions. Projecting the agronomic performance of dairy farms makes possible the identification of adaptation strategies that deserve investigation in future studies. For instance, legume crops like alfalfa and soybean could be exploited to a greater extent, since they stand to benefit the most from the increase in atmospheric CO2 concentrations. With higher temperatures, grain corn and soybean could become suitable crops for current colder areas, with potential benefits for farm income. In warmer areas like QSW, an alternative to timothy in the alfalfa-timothy mixture might be needed. Some forage grasses are known to be less sensitive to high summer temperatures than timothy. Because the yield of cool-season cereals will likely be limited in the future by the occurrence of high temperature days during the flowering and grain-

affecting grain number or weight (Yang et al., 2014) and that the corn harvest index decreases under water deficit (Amer, 2010). For the QSW area, Brassard and Singh (2007) predicted a slight increase (+2% to + 8%) in grain corn yield for the 2040–2069 period relative to the 1960–1990 period, assuming no changes in corn hybrids. For a more distant future (2090–2099), Tatsumi et al. (2011) projected a 20% decrease in corn yield for southwestern Quebec, although they did not account for the CO2 fertilization effect. Given that potential genetic improvement of corn hybrids (except the change in hybrid maturity) is not taken into account in the present study or in the cited literature, the increase in grain corn yield with climate change was likely under-estimated in future scenarios. The projected increase in soybean yield in the present study (up to + 45% for the DF8.5 scenario in QSW) is consistent with the results of Brassard and Singh (2007) who projected increases of 27% to 111% for the same area. Soybean yields have been found to be highly correlated with available CHUs in the Atlantic region of Canada (Bootsma et al., 2005b). According to Brassard and Singh (2007), 36% to 76% of the projected increase in soybean yield in the QSW area could be attributed to the CO2 fertilization effect. This could partly explain why Singh et al. (1998) and Tatsumi et al. (2011), who did not take CO2 fertilization into account, projected a decline in soybean yields in southwestern Quebec between 1961–1990 and 2090–2099. The CAB and QE areas are unsuitable for grain corn and soybean production in the reference period, with 1900 to 2100 CHUs, but will likely become suitable in the near future (Singh et al., 1998). Bootsma et al. (2005b) stated that it was conceivable that grain corn could replace 30% of small-grain cereals and silage corn in Atlantic Canada by 2040–2069. Soybean is also expected to expand its production and to compete with corn, given soybean's lower production cost, high protein content, and higher selling price (Bootsma et al., 2005b). Therefore, growing grain corn and soybean on extra land made available due to silage yield increases should be investigated as a climate change adaptation strategy. Unlike the situation for warm-season and over-wintering crops, the number of days to reach maturity for cool-season cereal crops is expected to decrease under future climate conditions owing to the acceleration of the maturity process caused by a greater occurrence of high temperatures (An and Carew, 2015; Qian et al., 2013; Robertson et al., 2013). This is consistent with the projected decrease in yields of cool-season cereal crops (barley and wheat) under future climate 249

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

filling stages, the use of winter cereals, which are sown in the fall and harvested earlier in the following year, could be considered. Lastly, irrigation could become an interesting avenue and deserves further investigation.

the storage structure, which favors the formation of a crust at the manure surface. Although this crust helps reduce NH3 and CH4 emissions from the storage structure, it also creates an environment favorable to nitrification and denitrification, which substantially enhances N2O emissions (Rotz et al., 2015). Overall, projected changes in farmland N2O emissions in the future were the result of a trade-off between two opposing trends. On one hand, the additional amount of N lost as NH3 in housing facilities and during manure storage reduced the actual amount of N available for in-field N2O losses, as previously observed by Rotz et al. (2014). On the other hand, the increasing rates of mineral N fertilizers applied to counterbalance the increased N uptake associated with higher yields led to higher soil nitrate and ammonium concentrations, thereby contributing to increased farmland N2O emissions. The highest increase in N2O emissions occurred in CAB, which had the largest projected increases in annual crop yields, resulting in larger fertilizer N inputs. Results for this area are consistent with the findings of Smith et al. (2013), who projected an increase in N2O emissions in Canada because of the increase in mineral N applied in anticipation of larger crop yields. Larger fossil fuel CO2 emissions were expected in CAB and QSW compared with QE in the reference period because these virtual farms in CAB and QSW have a larger proportion of annual crops that require more fuel use per hectare for field operations than perennial crops, and because virtual farm in QSW has a larger proportion of grain crops that require drying, an operation that also requires fossil fuel. In future scenarios, the projected CO2 emissions increased slightly in CAB and QE because more fuel was required for field operations due to an increase in the number of forage cuts per year. The large decrease in CO2 emissions in QSW is specifically related to the reduced need to dry grains under future climate conditions. The QSW virtual farm has a large area dedicated to grain corn, a crop that requires more drying than small-grain cereal crops like wheat and barley. Leaching N losses are predicted on a daily basis in IFSM, based on the amount of nitrate in the soil solution and the simulated proportion of solution that drains through the soil profile. Larger leaching N losses on the QSW virtual farm than on the virtual farms in CAB and QSW in the reference period can be attributed to a greater concentration of N in the soil solution (13.9, 6.1, and 2.8 g kg− 1 in QSW, QE, and CAB, respectively) and greater precipitation in the growing season. Because the QSW virtual farm grows mainly annual crops, N was frequently applied to bare soils in spring or fall, a practice that may explain the greater N concentration in the soil solution. Lower precipitation during the growing season in CAB (only 64% to 72% of the amounts in QSW and QE in the reference period) explains the very low leaching N losses projected for this area. The projected increase in N losses from leaching in CAB in the future is related to the higher rates of mineral N fertilizers applied in future scenarios, mostly on silage corn. In QE, most of the manure was applied to perennial forage crops, which represented 67% of the cropped area in the reference period. Despite the projected yield increase in future scenarios, no additional N fertilizer was applied to perennial forage crops due to the presence of alfalfa in the mixture. Therefore, the increase in N exports associated with the higher yields of perennial forage crops in future scenarios resulted in lower N losses from leaching in QE. In IFSM, the calculation of P lost through surface runoff and leaching includes losses from applied fertilizers and from soil inorganic and organic P pools (Rotz et al., 2015). The driving variables for P losses that varied the most among farms were daily precipitation, crop canopy, and residue cover. Indeed, the greatest P losses were found in QSW, where precipitation levels were the highest and where 50% of the crop area was devoted to corn, a crop with wide inter-row spacings and minimal soil coverage. Using IFSM, Ghebremichael et al. (2009) found that increasing the corn area by 3% on small and medium-sized dairy farms in the northeastern United States increased sediment-bound P losses by 18%. The projected increase in P losses in future scenarios is related to the general increase in precipitation for all farms.

4.2. Environmental emissions In the present study, most of the NH3 emissions occurred during field applications of manure, a finding that is consistent with results reported by Sheppard et al. (2011b) for dairy farms across Canada. The proportion of the total farm NH3 emissions associated with housing facilities was greater in CAB than in QE and QSW. Indeed, IFSM considers a larger soiled area per cow in a free-stall barn (3.5 m2 per cow in CAB) than in a tie-stall barn (1.2 m2 per cow in QSW and QE), and emissions of NH3 per animal inside housing facilities are higher when manure is spread over a larger area (Rotz et al., 2015). Lastly, NH3 emissions during manure storage depend on the exposed surface area, the air temperature, and the total ammoniacal N (TAN, includes ammonium and ammonia) concentration of the manure. The lower proportion of farm NH3 emissions from manure storage in CAB than in QE and QSW, despite the larger exposed storage surface area, can be explained by the greater NH3 losses in the housing facilities prior to storage; those losses reduced the TAN concentration in the stored manure, thereby reducing the source intensity for NH3 emissions. Projected field-related NH3 emissions increased in all future scenarios, primarily due to higher temperatures. Higher temperatures are known to increase the rate of urea hydrolysis forming TAN, the ammonia fraction of TAN in the manure solution, and the ratio of gas to aqueous ammonia—all factors that increase the potential NH3 emission rate (Rotz et al., 2015). Wind speed was not monitored by the selected weather stations and, therefore, IFSM used a default value of 3 m s− 1 for all farms. Wind speed over open manure storage facilities and air velocity inside open barns affect NH3 emission rates in IFSM (Rotz et al., 2015). The proportions of GHG emissions that are attributable to animals and manure (average of 44% and 29%, respectively, on a CO2eq basis) are in line with a recent life-cycle assessment of Canadian dairy farms that attributed 46% of farm GHG emissions to livestock management and 27% to manure management (Quantis Canada et al., 2012). The partitioning of on-farm methane emissions is consistent with the findings of Vergé et al. (2007), who also estimated the respective contributions of enteric fermentation and manure storage to CH4 emissions on Canadian dairy farms at 80% and 20% with minor emissions from field-applied manure. To simulate CH4 emissions from housing facilities, which mostly comes from enteric fermentation, IFSM takes into account cow type and weight as well as the amount and composition of the cow's diet, including starch, neutral and acid detergent fiber concentrations, and metabolizable energy intake. Because the number of cows and their milk production was assumed to remain unchanged in future scenarios, the change in feed quality likely explains the small projected decrease in CH4 emissions from housing facilities in CAB and QE. Hatew et al. (2016) demonstrated that the increased maturity of silage corn at harvest results in higher starch concentration and lower fiber concentration, which reduce CH4 emissions from dairy cows. In IFSM, emissions of CH4 from manure storage depend mostly on temperature and storage time (Rotz et al., 2015). Projected storage times did not change significantly, given that spring and fall manure applications were both advanced by a similar number of days. Therefore, temperature change is likely responsible for the large projected increase in CH4 emissions from manure storage in future scenarios (+62% on average). Nitrous oxide emissions on dairy farms originate mainly from nitrification and denitrification processes in soils (Chianese et al., 2009b), and manure storage is a secondary source. In the present simulations, however, N2O emissions from manure storage were greater than those from farmland. Manure was assumed to be loaded into the bottom of 250

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

farm systems, Bell et al. (2011) found that low-forage dairy systems (higher intensification level) had a lower global warming potential, which is equivalent to our C footprint, when expressed per unit of energy-corrected milk. In our study, the smallest C footprint was obtained for the farm with the lowest proportion of perennial forage crops in the cows' diet (CAB, 0.95 kg CO2eq kg− 1 FPCM) and the largest C footprint for the farm with the highest proportion (QE, 1.20 kg CO2eq kg− 1 FPCM). Interestingly, Bell et al. (2011) found that low-forage dairy systems had a larger global warming potential (C footprint) when expressed per unit of land area instead of per unit of milk production. This finding suggests that the relative ranking of the farms in terms of C footprint could have differed if a land area functional unit (ha) had been used instead of a milk production functional unit (kg− 1 FPCM).

Through simulations of the environmental performance of dairy farms in future scenarios, some emission sources were identified as deserving special attention with regard to the development of mitigation strategies: NH3 emissions from manure application, CH4 emissions from manure storage, and field-related P losses through soil erosion. The CAB virtual farm, which had the largest projected increase in annual crop yields, clearly illustrates that field-related N losses through denitrification and N leaching may need to be addressed in the future given the projected increase in N fertilizer rates. 4.3. Environmental footprints The main difference in reactive N losses among the virtual farms of the three regions pertained to N leaching, which was much lower in CAB than in QSW and QE because of lower precipitation. The smaller N losses and greater farm milk production in CAB explain its smaller N footprint compared to that of QE and QSW. Similarly, differences in N leaching losses mostly explain the larger N footprint in QSW than in QE, given that a larger proportion of land is devoted to annual crops in QSW. In addition, cows in CAB consumed more annual silage crops (mostly corn) and less perennial forages than cows in QSW and QE, resulting in lower N concentrations in manure (33.8, 36.6, and 43.4 g N kg− 1 DM in CAB, QSW, and QE, respectively) and, therefore, lower N emissions from manure application. Despite the projected smaller cropped area needed to grow feed for cows in the future, the reactive N footprint increased in all future scenarios in CAB and QSW, mostly because of the increase in NH3 emissions (Fig. 2). A larger amount of N lost per unit of milk produced is an indication of inefficient management of this expensive-to-replace crop nutrient (Sheppard et al., 2011b). In QE, the reduction in leaching N losses was larger than was the increase in NH3 emissions, except in scenario DF8.5, where they were almost equal, resulting in an N footprint identical to that of the reference period. The C footprint estimates for the three virtual farms in the reference period (0.95 to 1.20 kg CO2eq kg− 1 FPCM; Fig. 3) were slightly greater than the values previously reported (0.93 to 1.12 kg CO2eq kg− 1 FPCM) for Canadian dairy farms (Mc Geough et al., 2012; Quantis Canada et al., 2012; Vergé et al., 2007). However, methodologies differed among studies, particularly with respect to manure management, including off-farm exports. Manure management can have a particularly strong influence on C footprint, given that it is the second largest contributor to GHG emissions from dairy farms. The C footprint varied less among scenarios (−5% to + 9% relative to the reference period) than did the N footprint (− 15% to +46%). Under future climate conditions, projected farm GHG emissions from animals, feed production, fuel combustion, and the production of resource inputs declined, whereas projected emissions from manure increased. In QE, this trade-off resulted in a small projected decline in the C footprint in the near future and a value close to that of the reference period in the DF8.5 scenario (+1%). In CAB and QSW, this trade-off resulted in only a slight change in the C footprint in the near future and a small increase in the distant future (+1% to + 4% in CAB, and + 5% to + 9% in QSW). The N and C footprints differed greatly among farms. In addition to having a different climate, each virtual farm was designed to be a realistic system for the targeted agricultural area, including farm size and crop selection. The CAB virtual farm is associated with a higher level of intensification than the QSW and QE virtual farms, because of the greater number of cows and greater milk production per cow. The QSW farm has a higher level of intensification than QE. The lower C footprint in CAB is likely related to this greater intensification. An increase in milk production per cow is generally accompanied by a decrease in forage grass intake (Salou et al., 2017). This is true in the present study, where only 25% of daily forage DM intake consisted of perennial forage crops in CAB, in comparison with 31% in QSW and 68% in QE. When comparing contrasting intensification levels of dairy

4.4. Differences in projections from the three climate models A multi-model ensemble is known to be more robust than results from individual climate models. Nonetheless, a brief discussion on the impact of climate models on the elements of agronomic and environmental performance is warranted. In the present study, differences among the three climate models (CanESM2, CanRCM4, and HadGEM2) for a given scenario had non negligible effects on some elements of the agronomic and environmental performance for each farm simulated by IFSM. Differences [(highest value − lowest value) / lowest value × 100] in yield for perennial forage crops ranged from 5% to 14% in QE and QSW for the forage mixture, which is near the values obtained by Thivierge et al. (2016) for the same areas, and from 10% to 32% for pure alfalfa in CAB. Differences in yield for annual crops ranged from 4% to 26% for barley and wheat, 4% to 29% for silage corn, 8% to 20% for grain corn, and 6% to 12% for soybean. In most areas and future scenarios, the highest yields were obtained from the projections with the CanRCM4 climate model. This climate model projected more growing-season precipitation in future scenarios than the other climate models did, especially in May and June in CAB and April, July and August in QE. Moreover, CanRCM4 projected lower temperatures during the growing season in QSW, an area where the prevalence of days with high temperature (days when the daily maximum temperature reached at least 28 °C) was found to be an issue. The impact of climate models on yield simulated by IFSM was not increased in the distant future when compared to the near future. However, differences in yield were generally larger in the CAB area than in the other two areas. This is likely related to the fact that CAB was the area with the largest differences in precipitation among the three climate models. Indeed, cumulative precipitation during the growing season (April to October) increased for all virtual farms and future scenarios (+23 to + 73 mm), except in CAB with the HadGEM2 climate model, where it decreased (− 7 to − 21 mm). The impact of climate models on variations in emissions simulated by IFSM never exceeded 11% for anthropogenic CO2, 12% for NH3, 6% for CH4, and 4% for N2O. Losses of N and P through leaching and runoff showed the largest differences among climate models. These differences were much larger in CAB (77% to 210% for N and 0% to133% for P) than in QSW and QE (4% to 29% for N and 3% to 33% for P). The large differences found in CAB were due to the particularly low N and P loss values obtained with the HadGEM2 climate model, likely related to the decrease in precipitation projected in this area with this climate model while the other two climate models projected an increase. Although the differences in projections among the three climate models appear large when expressed in relative values, they never exceeded 6.9 kg N ha− 1 and 0.2 kg P ha− 1 in absolute terms. Differences in projected C and N footprints simulated by IFSM due to climate models never exceeded 6.0% and 9.4%, respectively. In most areas and future scenarios, the lowest footprints were obtained from the CanRCM4 climate model: the larger projected yields with this model resulted in a smaller portion of the cropped area being used to feed cows and being accounted for in the environmental footprints. 251

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

(Hatfield et al., 2011; Morgan et al., 2004) and reduced freezing tolerance in alfalfa (Bertrand et al., 2007), but neither phenomenon is currently included in IFSM.

4.5. Limitations of this study The outcomes of the present study are predictions, and should be interpreted with caution. Modelling involves the use of certain assumptions, which can affect the results significantly. A major assumption made in the present study is that milk yield per cow would remain constant in the future. A large part of the expected increase in milk yield in the future would come through cow genetic improvement. However, higher temperatures in the future are also expected to increase the heat stress for cows, which is known to affect negatively milk yield (West, 2003). Therefore, the decision was made to keep milk yield per cow constant. A different choice would have affected the C and N footprints, which are calculated by dividing total emissions by the farm's milk production. Another assumption was that crop areas would remain unchanged with the same area devoted to each crop in the future. Because silage crops were projected to have the largest increase in yield in the future, and because herd size and milk yield did not change, large amounts of forage and corn silage were projected to be sold off farm in future scenarios. Other options are likely to be available to dairy farmers in the future, such as using perennial crops to produce biofuel or switching to more lucrative grain crops (e.g. corn or soybean). The potential of these options warrants further investigation. The present study neglected some potential effects of climate change on Canadian dairy farms. For instance, the impacts of future climate change on the prevalence of certain crop diseases and pests (Smith et al., 2013) and on crop lodging (Bootsma et al., 2005a) are not well known and were not considered. Moreover, the expected reduction in the winter survival of sensitive species such as alfalfa in the future due to less efficient hardening in the fall and less snow cover during the winter (Bélanger et al., 2006, 2002; Castonguay et al., 2006) was not addressed. The development of improved cultivars, such as heat-resistant spring cereals or freezing-tolerant alfalfa, would certainly facilitate adaptation to the projected warmer climate conditions. Also, the potential effect of climate change-induced heat stress on cattle was only partially addressed, without considering the long-term negative effect on milk yield and on cows' health and reproduction (Rotz et al., 2016). The IFSM model could be improved to better account for the effects of climate change on crop physiology. Under elevated CO2 concentrations, for instance, studies have reported increased water-use efficiency

5. Conclusions This study simulated the effects of climate change on the agronomic and environmental performance of Canadian dairy farms in three climatically contrasting agricultural areas in Canada. Under future climate conditions, yields of perennial forages and warm-season crops were projected to increase, especially in the two colder areas (CAB and QE), whereas yields of small-grain cereals were projected to decrease. The environmental emissions that were projected to increase the most included NH3 emissions from manure storage and field application, CH4 emissions from manure storage, field-related P losses and, for most of the farms, field-related N losses through denitrification. These results provide avenues for selecting strategies for mitigation of emissions and for dairy farm adaptation to climate change. With regard to the environmental impact directly associated with milk production and expressed as C and N footprints, emissions from manure were projected to increase the most in the future, whereas the share of field-related emissions was projected to decrease because less area would be needed to produce the same amount of feed for cows. Overall, the projected changes in future scenarios were more pronounced for the N footprint (−15% to +46%) than for the C footprint (−5% to +9%) relative to the reference period. These findings suggest that, in the context of climate change, future management strategies on dairy farms should be aimed not only at reducing GHG emissions but also at improving overall farm N-use efficiency. Acknowledgments This study was financially supported by the Dairy Research Cluster as part of the Canadian Agri-Science Clusters Initiative of Agriculture and Agri-Food Canada (AAFC). The senior author is grateful to AAFC for the financial support it provided through the Visiting Fellowships in Canadian Government Laboratories Program. The authors warmly thank Sheilah Nolan from Alberta Agriculture and Forestry for reviewing the manuscript, René Morissette from AAFC for his assistance with data processing, and François Thibodeau and Marianne Crépeau from AAFC for their thorough work on IFSM simulations.

Appendix A The key outcomes identified in this study showed that under future climate conditions:

• Yields of perennial forages and warm-season crops were projected to increase, especially in the two colder areas (CAB and QE), whereas yields of small-grain cereals were projected to decrease slightly. • The environmental emissions that were projected to increase the most included NH emissions from manure storage and field application, CH 3

4

emissions from manure storage, field-related P losses and, for most of the farms, field-related N losses through denitrification. More details about calculations of these processes are provided in the following sections.

A.1. Crop growth, establishment and harvest processes The crop component is made up of five submodels for simulating pure-stand alfalfa, perennial grass (mixture), corn, small grains, and soybean. Table 1 identifies the original crop model adapted for use in IFSM with references for more detail on those models.

Crop

Model

Reference

Alfalfa Perennial grass Corn Small grains Soybean

ALSIM1 Level 2 GRASIM CERES-maize CERES- small grain SOYGRO

Fick (1977) Mohtar et al. (1997) Jones and Kiniry (1986) Tsuji et al. (1994) Jones et al. (1991)

252

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

All of these models simulate growth and physiological development on a daily time step as influence by weather and soil conditions. They use a radiation use efficiency approach and the equation used to calculate biomass growth can be summarized as:

AGB = IntPAR × RUE × min (Ws, Ns, Ftemp) × f (CO2) Where: AGB above ground biomass, g m− 2 IntPAR intercepted potentially active solar radiation, MJ m− 2 RUE radiation use efficiency, gAGB MJ− 1 Ws water stress index, dimensionless (between 0 and 1) Ns nitrogen stress index, dimensionless (between 0 and 1) Ftemp temperature function, dimensionless (between 0 and 1) f(CO2) CO2 concentration factor (= 1 when [CO2] = 330 ppm) Warmer temperatures due to climate change affect crop growth and physiological stage of development through a number of interactions specific to the crop. Projected changes in precipitation patterns and increased evapotranspiration can both positively and negatively affect growth. Potential photosynthetic fixation of carbon is increased through projected increases in atmospheric CO2 concentration using nonlinear relationships specific to each crop (Tsuji et al., 1994). Tillage, planting and harvest processes are simulated through time based upon the capacity of the equipment used and days suitable for field operations. Days suitable for field work are primarily controlled by soil moisture content, where the moisture level must be below a critical level to allow tractability of the equipment. During harvest, the moisture content of the crop is controlled by the drying conditions, i.e. ambient temperature, solar radiation, precipitation and soil moisture. Through these interactions with daily weather patterns, project changes in climate affect the timing of field operations and the crop losses and quality changes occurring during harvest. A.2. Ammonia emissions from manure storage and field application The hourly rate of ammonia emission is a function of the overall mass transfer rate and the difference in ammonia concentration between the manure and surrounding atmosphere (Datta, 2002):

J = 3600 K ( Cm –H ( Ca ) ) Where: J Cm Ca H K

ammonia flux, kg m− 2 s− 1 concentration of ammonia in manure, kg m− 3 concentration of ammonia in ambient air, kg m− 3 Henry's Law constant, g liquid g gas− 1 overall mass transfer coefficient, m s− 1

Cm = F × CTAN Where: CTAN = concentration of total ammoniacal nitrogen (TAN) in the manure solution, kg m− 3 F = ammonia fraction of TAN

F = 1 (1 + 10−pH K a ) Where: 10(0.05–2788/T) surface pH of manure or urine temperature, °K Thus, ammonia emission rate is very sensitive to manure and ambient air temperatures. Projected warmer temperatures in the future will increase ammonia emission rates with all other aspects of manure management held constant.

Ka pH T

A.3. Methane emissions from manure storage Emission of CH4 from slurry or liquid manure storages is predicted using equations derived from the model proposed by Sommer et al. (2004):

ECH 4, man = (( 24·Vs, d·b1) 1000)·exp [ln (A) − (E RT )] + (( 24·Vs, nd·b2) 1000)·exp [ln (A) − (E RT )] Where: ECH4 , man emission of CH4 from the storage, kg CH4 day− 1 Vs , d and Vs , nd degradable and non-degradable VS in the manure, g b1 and b2 rate correcting factors, dimensionless A Arrhenius parameter, g CH4 kg VS− 1 h− 1 E apparent activation energy, J mol− 1 R gas constant, J °K− 1 mol− 1 T temperature, °K 253

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Therefore, CH4 emissions increase with projected increases in ambient temperature. A.4. Field-related P losses The sediment organic P loss is simulated using enrichment ratios to predict bioavailable and labile P losses as functions of erosion sediment loss (Sharpley, 1985).

Porgl = Pbiol + Plabl Where: Porgltotal sediment P loss (kg P ha− 1) Pbiol bioavailable P loss (kg P ha− 1) Plabl labile P loss (kg P ha− 1)

Pbiol = sed × Pbio × CER Where: sed Pbio CER

erosion sediment loss (kg sediment ha− 1) bioavailable P (kg P ha− 1) enrichment ratio

sed = 11.8 ( Qsurf qpeak Ahru )0.56 (K )(L)(S )(C )(P ) Where: Qsurf qpeak Ahru K L S C P

daily runoff depth, mm peak runoff, m3 s− 1 field area analyzed, m2 soil erodibility factor slope length factor slope steepness factor cover management factor support practice factor

CER = exp(1.63 − 0.25·Ysed ) Where: Ysed = amount of daily erosion occurring from the given cropland, kg erosion day− 1 An important process simulated for the inorganic pools is the loss of P from the upper soil layer through runoff. Using the theory of an extraction coefficient, labile P is withdrawn from the soil reservoir and enters runoff. The mass of soluble P lost in runoff, Psol [kg P ha− 1], is a function of the runoff depth Q [m], the extraction coefficient, soil depth, and soil bulk density.

Psol = [ Pil ( Q ) (Cextr ) Dlayer ( rBD )] Where: Cextr extraction coefficient, Mg m− 3 Dlayer depth of the soil layer, m rBD bulk density of the soil, Mg m− 3 A simple relationship is used to predict leaching loss of soil inorganic P loss based upon the work of Vadas (2001):

Plch = 0.01 (Clp )( Ls ) [ Dlayer Dsoil ] Where: Plch soil P leached from layer, kg ha− 1 Clp concentration of P in leachate from soil layer, mg kg− 1 Ls amount of leachate flowing from soil layer, mm Dsoil depth of root zone in soil, m Soil P loss is affected by the amount and intensity of precipitation along with soil management practices. Projected increases in precipitation in the northeast region with more intense storms can increases P runoff in both soluble and sediment forms. A.5. Field-related N2O losses through denitrification and nitrification Emission of N2O from soils is predicted as the sum of nitrification and denitrification losses:

EN 2O, soil = EN 2O, soil, N + EN 2O, soil, D

254

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Where: EN2O , soil total emission of N2O from soils, kg N2O ha− 1 day− 1 EN2O , soil , N emission from soils due to nitrification, kg N2O ha− 1 day− 1 EN2O , soil , D emission from soils due to denitrification, kg N2O ha− 1 day− 1 Emission from nitrification is predicted as:

EN 2O, soil, N = K2·RNO3·FN , conv Where: K2 fraction of nitrified N lost as N2O flux, g N g N− 1 RNO3 soil nitrification rate, g N m− 2 day− 1 FN , conv conversion factor, 15.7 (kg N2O ha− 1 day− 1) / (g N m− 2 day− 1). Emission of N2O due to denitrification is predicted as:

EN 2O, soil, D = [( min (Fd, NO3·Fd, CO2 )·Fd, WFPS ) (1 + RNratio)]·ρsoil ·dsoil ·FN , mass Where: EN2O , soil , D emission of N2O from soil, kg N2O ha− 1 day− 1 Fd , NO3 factor for the effect of soil nitrate concentration, μg N g soil− 1 day− 1 Fd , CO2 factor for the effect of soil respiration, μg N g soil− 1 day− 1 Fd , WFPS factor for the effect of soil moisture, dimensionless RNratio ratio of N2 to N2O emission, μg N μg N− 1 ρsoil bulk density of the soil, g cm− 3 dsoil active soil depth of layer simulated (upper, lower), cm FN , mass unit conversion factor, 0.157 (kg N2O ha− 1 day− 1) / (μg N cm− 2 day− 1) These processes are indirectly influenced by projected changes in climate. Denitrification is influenced by soil moisture and nitrate contents. Projected changes in temperature and precipitation control soil moisture. As indicated above, increased temperature increases ammonia emissions, which affects the amount of N applied and thus the amount available for both nitrification and denitrification processes. A.6. Other processes Details about other processes can be found in the reference manual of IFSM (Rotz et al., 2015), available at: https://www.ars.usda.gov/northeast-area/up-pa/pswmru/docs/integrated-farm-system-model/ More specifically, information about ammonia emission from animal housing can be found at pages 131 to133, carbon dioxide emission at pages 155 to161, methane emission from enteric fermentation and field-applied manure at pages 162 to169, and nitrous oxide emission from barn and from manure storage at pages 173 to177. Finally, information about the calculation of environmental footprints can be found at pages 203 to 209.

forage crops in eastern Canada: causes, mitigation, and prediction. Can. J. Plant Sci. 86, 33–47. http://dx.doi.org/10.4141/P04-171. Bell, M.J., Wall, E., Russell, G., Simm, G., Stott, A.W., 2011. The effect of improving cow productivity, fertility, and longevity on the global warming potential of dairy systems. J. Dairy Sci. 94, 3662–3678. http://dx.doi.org/10.3168/jds.2010-4023. Bertrand, A., Prévost, D., Bigras, F.J., Castonguay, Y., 2007. Elevated atmospheric CO2 and strain of rhizobium alter freezing tolerance and cold-induced molecular changes in alfalfa (Medicago sativa). Ann. Bot. (Lond.) 99, 275–284. http://dx.doi.org/10. 1093/aob/mcl254. Beukes, P.C., Palliser, C.C., Macdonald, K.A., Lancaster, J.A.S., Levy, G., Thorrold, B.S., Wastney, M.E., 2008. Evaluation of a whole-farm model for pasture-based dairy systems. J. Dairy Sci. 91, 2353–2360. Bootsma, A., 1984. Forage crop maturity zonation in the Atlantic region using growing degree-days. Can. J. Plant Sci. 64, 329–338. http://dx.doi.org/10.4141/cjps84-047. Bootsma, A., Gameda, S., McKenney, D.W., 2005a. Impacts of potential climate change on selected agroclimatic indices in Atlantic Canada. Can. J. Soil Sci. 85, 329–343. Bootsma, A., Gameda, S., McKenney, D.W., 2005b. Potential impacts of climate change on corn, soybeans and barley yields in Atlantic Canada. Can. J. Soil Sci. 85, 345–357. Brassard, J.P., Singh, B., 2007. Effects of climate change and CO2 increase on potential agricultural production in Southern Québec, Canada. Clim. Res. 34, 105–117. Brown, D.M., Bootsma, A., 1993. Crop Heat Units for Corn and Other Warm Season Crops in Ontario. Factsheet, Agdex 111/31. Ontario Ministry of Agriculture, Food and Rural Affairs, Toronto, ON, Canada. Canadian Dairy Information Centre, 2015. Canada's dairy industry at a glance: 2015 highlights. http://www.dairyinfo.gc.ca/index_e.php?s1=cdi-ilc&s2=aag-ail (accessed 14 Dec. 2016). CanSIS, 2015. Detailed Soil Surveys. Canadian Soil Information Service. http://sis.agr.gc. ca/cansis/nsdb/dss/v3/index.html (accessed 2 Aug. 2015). Castonguay, Y., Laberge, S., Brummer, E.C., Volenec, J.J., 2006. Alfalfa winter hardiness: a research retrospective and integrated perspective. Adv. Agron. 90, 203–265. Chai, L., Kröbel, R., MacDonald, D., Bittman, S., Beauchemin, K.A., Janzen, H.H., McGinn, S.M., Vanderzaag, A., 2016. An ecoregion-specific ammonia emissions inventory of Ontario dairy farming: mitigation potential of diet and manure management practices. Atmos. Environ. 126, 1–14. http://dx.doi.org/10.1016/j.atmosenv.2015.11. 030.

References AAF, 2014. Alberta Greenfeed and Silage Production Survey Results 2004–2014. Alberta Agriculture and Forestry. http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/ all/sdd12583 (accessed 15 Dec. 2015). AAF, 2015. Agroclimatic Atlas of Alberta: Agricultural Climate Elements. Alberta Agriculture and Forestry. http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/ all/sag6301 (accessed 27 March 2016). AAFRD, 2004. Alberta Fertilizer Guide. Alberta Agriculture, Food and Rural Development. Agdex 541-1. AFSC, 2015. Insurance Statistics. Agriculture Financial Services Corporation. https:// www.afsc.ca/Default.aspx?cid=1889&lang=1 (accessed 15 Feb. 2016). Amer, K.H., 2010. Corn crop response under managing different irrigation and salinity levels. Agric. Water Manag. 97, 1553–1563. http://dx.doi.org/10.1016/j.agwat. 2010.05.010. An, H., Carew, R., 2015. Effect of climate change and use of improved varieties on barley and canola yield in Manitoba. Can. J. Plant Sci. 95, 127–139. http://dx.doi.org/10. 4141/cjps-2014-221. Arora, V.K., Scinocca, J.F., Boer, G.J., Christian, J.R., Denman, K.L., Flato, G.M., Kharin, V.V., Lee, W.G., Merryfield, W.J., 2011. Carbon emission limits required to satisfy future representative concentration pathways of greenhouse gases. Geophys. Res. Lett. 38, L05805. http://dx.doi.org/10.1029/2010GL046270. Baron, V.S., Najda, H.G., Stevenson, F.C., 2006. Influence of population density, row spacing and hybrid on forage corn yield and nutritive value in a cool-season environment. Can. J. Plant Sci. 86, 1131–1138. Bélanger, G., Kunelius, T., McKenzie, D., Papadopoulos, Y., Thomas, B., McRae, K., Fillmore, S., Christie, B., 1999. Fall cutting management affects yield and persistence of alfalfa in Atlantic Canada. Can. J. Plant Sci. 79, 57–63. http://dx.doi.org/10.4141/ P98-035. Bélanger, G., Rochette, P., Castonguay, Y., Bootsma, A., Mongrain, D., Ryan, D.A.J., 2002. Climate change and winter survival of perennial forage crops in eastern Canada. Agron. J. 94, 1120–1130. http://dx.doi.org/10.2134/agronj2002.1120. Bélanger, G., Castonguay, Y., Bertrand, A., Dhont, C., Rochette, P., Couture, L., Drapeau, R., Mongrain, D., Chalifour, F.-P., Michaud, R., 2006. Winter damage to perennial

255

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

A.K., Mosier, A.R., Newton, P.C.D., Niklaus, P.A., Nippert, J.B., Nowak, R.S., Parton, W.J., Polley, H.W., Shaw, M.R., 2004. Water relations in grassland and desert ecosystems exposed to elevated atmospheric CO2. Oecologia 140, 11–25. http://dx.doi. org/10.1007/s00442-004-1550-2. Paul, J.W., Zebarth, B.J., 1997. Denitrification and nitrate leaching during the fall and winter following dairy cattle slurry application. Can. J. Soil Sci. 77, 231–240. http:// dx.doi.org/10.4141/S96-052. Priestley, C.H.B., Taylor, R.J., 1972. On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Weather Rev. 100, 81–92. Qian, B., Gameda, S., Hayhoe, H., De Jong, R., Bootsma, A., 2004. Comparison of LARSWG and AAFC-WG stochastic weather generators for diverse Canadian climates. Clim. Res. 26, 175–191. http://dx.doi.org/10.3354/cr026175. Qian, B., Gameda, S., Hayhoe, H., 2008. Performance of stochastic weather generators LARS-WG and AAFC-WG for reproducing daily extremes of diverse Canadian climates. Clim. Res. 37, 17–33. http://dx.doi.org/10.3354/cr00755. Qian, B., Gameda, S., de Jong, R., Falloon, P., Gornall, J., 2010. Comparing scenarios of Canadian daily climate extremes derived using a weather generator. Clim. Res. 41, 131–149. http://dx.doi.org/10.3354/cr00845. Qian, B., De Jong, R., Gameda, S., Huffman, T., Neilsen, D., Desjardins, R., Wang, H., McConkey, B., 2013. Impact of climate change scenarios on Canadian agroclimatic indices. Can. J. Soil Sci. 93, 243–259. Qian, B., De Jong, R., Huffman, T., Wang, H., Yang, J., 2016a. Projecting yield changes of spring wheat under future climate scenarios on the Canadian Prairies. Theor. Appl. Climatol. 123, 651–669. http://dx.doi.org/10.1007/s00704-015-1378-1. Qian, B., Wang, H., He, Y., Liu, J., De Jong, R., 2016b. Projecting spring wheat yield changes on the Canadian Prairies: effects of resolutions of a regional climate model and statistical processing. Int. J. Climatol. 36, 3492–3506. http://dx.doi.org/10. 1002/joc.4571. Quantis Canada, Groupe Agéco, CIRAIG, 2012. Environmental and socioeconomic life cycle assessment of Canadian milk. Prepared for Dairy Farmers of Canada. http://lait. org/en/brochure/environmental-and-socioeconomic-life-cycle-assessment-ofcanadian-milk/ (accessed 20 July 2016). Ringer, M.A., Martin, G.M., Greeves, C.Z., Hinton, T.J., James, P.M., Pope, V.D., Scaife, A.A., Stratton, R.A., Inness, P.M., Slingo, J.M., Yang, G.-Y., 2006. The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (Had-GEM1). Part II: aspects of variability and regional climate. J. Clim. 19, 1302–1326. http://dx.doi.org/10.1175/JCLI3713.1. Robertson, S.M., Jeffrey, S.R., Unterschultz, J.R., Boxall, P.C., 2013. Estimating yield response to temperature and identifying critical temperatures for annual crops in the Canadian prairie region. Can. J. Plant Sci. 93, 1237–1247. Rotz, C.A., Kleinman, P.J.A., Dell, C.J., Veith, T.L., Beegle, D.B., 2011. Environmental and economic comparisons of manure application methods in farming systems. J. Environ. Qual. 40, 438–448. http://dx.doi.org/10.2134/jeq2010.0063. Rotz, C.A., Montes, F., Hafner, S.D., Heber, A.J., Grant, R.H., 2014. Ammonia emission model for whole farm evaluation of dairy production systems. J. Environ. Qual. 43, 1143–1158. http://dx.doi.org/10.2134/jeq2013.04.0121. Rotz, C.A., Corson, M.S., Chianese, D.S., Montes, F., Hafner, S.D., Bonifacio, H.F., Coiner, C.U., 2015. The Integrated Farm System Model, Reference Manual Version 4.2. Agricultural Research Service, USDA. https://www.ars.usda.gov/northeast-area/uppa/pswmru/docs/integrated-farm-system-model/#Reference (accessed 17 Feb. 2016). Rotz, C.A., Skinner, R.H., Stoner, A.M.K., Hayhoe, K., 2016. Evaluating greenhouse gas mitigation and climate change adaptation in dairy production using farm simulation. Trans. ASABE 59, 1771–1781. http://dx.doi.org/10.13031/trans.59.11594. Salou, T., Le Mouël, C., van der Werf, H.M.G., 2017. Environmental impacts of dairy system intensification: the functional unit matters!. J. Clean. Prod. 140, 445–454. http://dx.doi.org/10.1016/j.jclepro.2016.05.019. Sanford, G.R., Oates, L.G., Jasrotia, P., Thelen, K.D., Robertson, G.P., Jackson, R.D., 2016. Comparative productivity of alternative cellulosic bioenergy cropping systems in the North Central USA. Agric. Ecosyst. Environ. 216, 344–355. http://dx.doi.org/10. 1016/j.agee.2015.10.018. Schils, R.L.M., de Haan, M.H.A., Hemmer, J.G.A., van den Pol-van Dasselaar, A., de Boer, J.A., Evers, A.G., Holshof, G., van Middelkoop, J.C., Zom, R.L.G., 2007. DairyWise, a whole-farm dairy model. J. Dairy Sci. 90, 5334–5346. http://dx.doi.org/10.3168/jds. 2006-842. Scinocca, J.F., Kharin, V.V., Jiao, Y., Qian, M.W., Lazare, M., Solheim, L., Flato, G.M., Biner, S., Desgagne, M., Dugas, B., 2015. Coordinated global and regional climate modeling. J. Clim. 29, 17–35. http://dx.doi.org/10.1175/JCLI-D-15-0161.1. Shaffer, M.J., Halvorson, A.D., Pierce, F.J., 1991. Nitrate leaching and economic analysis package (NLEAP): model description and application. In: Follett, R.F., Deeney, D.R., Cruse, R.M. (Eds.), Managing Nitrogen for Groundwater Quality and Farm Profitability. Soil Science Society of America, Inc., Madison, WI, pp. 285–298. Sharpley, A.N., 1985. The selective erosion of plant nutrients in runoff. Soil Sci. Soc. Am. J. 49, 1527–1534. Sheppard, S.C., Bittman, S., Swift, M.L., Beaulieu, M., Sheppard, M.I., 2011a. Ecoregion and farm size differences in dairy feed and manure nitrogen management: a survey. Can. J. Anim. Sci. 91, 459–473. http://dx.doi.org/10.4141/CJAS2010-004. Sheppard, S.C., Bittman, S., Swift, M.L., Tait, J., 2011b. Modelling monthly NH3 emissions from dairy in 12 ecoregions of Canada. Can. J. Anim. Sci. 91, 649–661. http://dx.doi. org/10.4141/CJAS2010-005. Simard, R.R., Cluis, D., Gangbazo, G., Beauchemin, S., 1995. Phosphorus status of forest and agricultural soils from a watershed of high animal density. J. Environ. Qual. 24, 1010–1017. Singh, B., El Maayar, M., André, P., Bryant, C.R., Thouez, J.-P., 1998. Impacts of a GHGinduced climate change on crop yields: effects of acceleration in maturation, moisture stress and optimal temperature. Clim. Chang. 38, 51–86.

Chianese, D.S., Rotz, C.A., Richard, T.L., 2008. Simulating methane emissions from dairy farms. In: 2008 ASABE Annual International Meeting, Providence, RI, pp. 3462–3482. Chianese, D.S., Rotz, C.A., Richard, T.L., 2009a. Simulation of carbon dioxide emissions from dairy farms to assess greenhouse gas reduction strategies. Trans. ASABE 52, 1301–1312. Chianese, D.S., Rotz, C.A., Richard, T.L., 2009b. Simulation of nitrous oxide emissions from dairy farms to assess greenhouse gas reduction strategies. Trans. ASABE 52, 1325–1335. CRAAQ, 2010. Guide de référence en fertilisation, 2e édition. Centre de référence en agriculture et agroalimentaire du Québec, Quebec City, QC, Canada. Datta, A.K., 2002. Biological and Bioenvironmental Heat and mass Transfer. In: CRC Press. Marcel Dekker, Inc., New York, NY. DAYCENT, 2007. DAYCENT. Ver. 4.5. Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO. Dayyani, S., Prasher, S.O., Madani, A., Madramootoo, C.A., 2012. Impact of climate change on the hydrology and nitrogen pollution in a tile-drained agricultural watershed in eastern Canada. Trans. ASABE 55, 389–401. FADQ, 2015. Assurance récolte, rendements réels. La Financière agricole du Québec. http://www.fadq.qc.ca/statistiques/assurance-recolte/rendements-reels (accessed 20 July 2016). Fick, G.W., 1977. The mechanisms of alfalfa regrowth: A computer simulation approach. Search Agric. 7 (3), 1–28. Ghebremichael, L.T., Veith, T.L., Cerosaletti, P.E., Dewing, D.E., Rotz, C.A., 2009. Exploring economically and environmentally viable northeastern US dairy farm strategies for coping with rising corn grain prices. J. Dairy Sci. 92, 4086–4099. Hatew, B., Bannink, A., van Laar, H., de Jonge, L.H., Dijkstra, J., 2016. Increasing harvest maturity of whole-plant corn silage reduces methane emission of lactating dairy cows. J. Dairy Sci. 99, 354–368. http://dx.doi.org/10.3168/jds.2015-10047. Hatfield, J.L., Boote, K.J., Kimball, B.A., Ziska, L.H., Izaurralde, R.C., Ort, D., Thomson, A.M., Wolfe, D., 2011. Climate impacts on agriculture: implications for crop production. Agron. J. 103, 351–370. http://dx.doi.org/10.2134/agronj2010.0303. Hawkins, J., Weersink, A., Wagner-Riddle, C., Fox, G., 2015. Optimizing ration formulation as a strategy for greenhouse gas mitigation in intensive dairy production systems. Agric. Syst. 137, 1–11. http://dx.doi.org/10.1016/j.agsy.2015.03.007. Hayhoe, H.N., 2000. Improvements of stochastic weather data generators for diverse climates. Clim. Res. 14, 75–87. http://dx.doi.org/10.3354/cr014075. Heikkila, R., Van Biert, P., 2014. Dairy Cost Study: The Economics of Milk Production in Alberta 2013. vol. 73 Alberta Agriculture and Rural Development. Agdex 821-1. IPCC, 2001. Climate change 2001: the scientific basis. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A. (Eds.), Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK. IPCC, 2014. Climate change 2014: synthesis report. In: Core Writing Team, Pachauri, R.K., Meyer, L.A. (Eds.), Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, . https:// www.ipcc.ch/report/ar5/ (accessed 10 Apr. 2015). Jayasundara, S., Appuhamy, J.A.D.R.N., Kebreab, E., Wagner-Riddle, C., 2016. Methane and nitrous oxide emissions from Canadian dairy farms and mitigation options: an updated review. Can. J. Anim. Sci. 96, 306–331. http://dx.doi.org/10.1139/cjas2015-0111. Jégo, G., Rotz, C.A., Bélanger, G., Tremblay, G.F., Charbonneau, É., Pellerin, D., 2015. Simulating forage crop production in a northern climate with the Integrated Farm System Model. Can. J. Plant Sci. 95, 745–757. http://dx.doi.org/10.4141/cjps-2014375. Jing, Q., Bélanger, G., Qian, B., Baron, V., 2014. Timothy yield and nutritive value with a three-harvest system under the projected future climate in Canada. Can. J. Plant Sci. 94, 213–222. http://dx.doi.org/10.4141/cjps2013-279. Johns, T.C., Durman, C.F., Banks, H.T., Roberts, M.J., McLaren, A.J., Ridley, J.K., Senior, C.A., Williams, K.D., Jones, A., Rickard, G.J., Cusack, S., Ingram, W.J., Crucifix, M., Sexton, D.M.H., Joshi, M.M., Dong, B.-W., Spencer, H., Hill, R.S.R., Gregory, J.M., Keen, A.B., Pardaens, A.K., Lowe, J.A., Bodas-Salcedo, A., Stark, S., Searl, Y., 2006. The new Hadley Centre Climate Model (HadGEM1): evaluation of coupled simulations. J. Clim. 19, 1327–1353. http://dx.doi.org/10.1175/JCLI3712.1. Jones, C.A., Kiniry, J.R. (Eds.), 1986. CERES-Maize: a simulation model of maize growth and development. Texas A & M Univ. Press, College Station Texas. Jones, J.W., Boote, K.J., Jagtap, S.S., Mishoe, J.W., 1991. Soybean development. In: Hanks, J., Ritchie, J.T. (Eds.), Modeling Plant and Soil Systems. Agronomy Monograph 31 ASA-CSSA-SSSA, Madison, WI. Martin, G.M., Ringer, M.A., Pope, V.D., Jones, A., Dearden, C., Hinton, T.J., 2006. The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (HadGEM1). Part I: model description and global climatology. J. Clim. 19, 1274–1301. http://dx.doi.org/10.1175/JCLI3636.1. Mc Geough, E.J., Little, S.M., Janzen, H.H., McAllister, T.A., McGinn, S.M., Beauchemin, K.A., 2012. Life-cycle assessment of greenhouse gas emissions from dairy production in Eastern Canada: a case study. J. Dairy Sci. 95, 5164–5175. Meinshausen, M., Riahi, K., Smith, S.J., Thomson, A., van Vuuren, D.P.P., 2009. RCP Database. International Institute for Applied Systems Analysis. http://tntcat.iiasa.ac. at/RcpDb (accessed 18 Feb. 2016). Messing, I., Joel, A., Wesström, I., Strock, J., 2015. Influence of higher rain intensities on phosphorus movements in the upper half meter of macroporous clay soil. Acta Agric. Scand. Sect. B 65, 93–99. http://dx.doi.org/10.1080/09064710.2014.996588. Mohtar, R.H., Buckmaster, D.R., Fales, S.L., 1997. A grazing simulation model: GRASIM, A: Model Development. Trans. ASAE 40 (5), 1483–1493. Morgan, J.A., Pataki, D.E., Körner, C., Clark, H., Del Grosso, S.J., Grünzweig, J.M., Knapp,

256

Agricultural Systems 157 (2017) 241–257

M.-N. Thivierge et al.

Vayssières, J., Bocquier, F., Lecomte, P., 2009a. GAMEDE: a global activity model for evaluating the sustainability of dairy enterprises. Part II – interactive simulation of various management strategies with diverse stakeholders. Agric. Syst. 101, 139–151. Vayssières, J., Guerrin, F., Paillat, J.-M., Lecomte, P., 2009b. GAMEDE: a global activity model for evaluating the sustainability of dairy enterprises part I – whole-farm dynamic model. Agric. Syst. 101, 128–138. Vergé, X.P.C., Dyer, J.A., Desjardins, R.L., Worth, D., 2007. Greenhouse gas emissions from the Canadian dairy industry in 2001. Agric. Syst. 94, 683–693. http://dx.doi. org/10.1016/j.agsy.2007.02.008. Wang, H., He, Y., Qian, B., McConkey, B., Cutforth, H., McCaig, T., McLeod, G., Zentner, R., DePauw, R., Lemke, R., Brandt, K., Liu, T., Qin, X., White, J., Hunt, T., Hoogenboom, G., 2012. Short communication: climate change and biofuel wheat: a case study of southern Saskatchewan. Can. J. Plant Sci. 92, 421–425. http://dx.doi. org/10.4141/cjps2011-192. Wastney, M.E., Palliser, C.C., Lile, J.A., Macdonald, K.A., Penno, J.W., Bright, K.P., 2002. A whole-farm model applied to a dairy system. Proc. N. Z. Soc. Anim. Prod. 62, 120–123. West, J.W., 2003. Effects of heat-stress on production in dairy cattle. J. Dairy Sci. 86, 2131–2144. Yang, Y., Massa, G.D., Mitchell, C.A., 2014. Temperature DIP at the beginning of the photoperiod reduces plant height but not seed yield of maize grown in controlled environments. Ind. Crop. Prod. 53, 120–127. http://dx.doi.org/10.1016/j.indcrop. 2013.12.002.

Smith, W.N., Grant, B.B., Desjardins, R.L., Qian, B., Hutchinson, J., Gameda, S., 2009. Potential impact of climate change on carbon in agricultural soils in Canada 2000–2099. Clim. Chang. 93, 319–333. http://dx.doi.org/10.1007/s10584-0089493-y. Smith, W.N., Grant, B.B., Desjardins, R.L., Kroebel, R., Li, C., Qian, B., Worth, D.E., McConkey, B.G., Drury, C.F., 2013. Assessing the effects of climate change on crop production and GHG emissions in Canada. Agric. Ecosyst. Environ. 179, 139–150. Sommer, S.G., Petersen, S.O., Møller, H.B., 2004. Algorithms for calculating methane and nitrous oxide emissions from manure management. Nutr. Cycl. Agroecosyst. 69, 143–154. Stanton, D., Grombacher, A.W., Pinnisch, R., Mason, H., Spaner, D., 2007. Hybrid and population density affect yield and quality of silage maize in central Alberta. Can. J. Plant Sci. 87, 867–871. Tatsumi, K., Yamashiki, Y., Valmir da Silva, R., Takara, K., Matsuoka, Y., Takahashi, K., Maruyama, K., Kawahara, N., 2011. Estimation of potential changes in cereals production under climate change scenarios. Hydrol. Process. 25, 2715–2725. http://dx. doi.org/10.1002/hyp.8012. Thivierge, M.-N., Jégo, G., Bélanger, G., Bertrand, A., Tremblay, G.F., Rotz, C.A., Qian, B., 2016. Predicted yield and nutritive value of an alfalfa–timothy mixture under climate change and elevated atmospheric carbon dioxide. Agron. J. 108, 585–603. http://dx. doi.org/10.2134/agronj2015.0484. Tsuji, G.Y., Jones, J.W., Uehara, G., Balas, S. (Eds.), 1994. DSSAT version 3. University of Hawaii, Honolulu, HI.

257