Agricultural Systems 108 (2012) 84–93
Contents lists available at SciVerse ScienceDirect
Agricultural Systems journal homepage: www.elsevier.com/locate/agsy
A case study of the potential environmental impacts of different dairy production systems in Georgia Jeff B. Belﬂower a, John K. Bernard b, David K. Gattie a, Dennis W. Hancock c, Lawrence M. Risse a, C. Alan Rotz d,⇑ a
University of Georgia, Driftmier Engineering Center, Athens, GA 30602, USA University of Georgia, Animal and Dairy Science, Tifton, GA 31793, USA University of Georgia, Miller Plant Sciences, Athens, GA 30602, USA d USDA-ARS, Pasture Systems and Watershed Management Research Unit, University Park, PA 16802, USA b c
a r t i c l e
i n f o
Article history: Received 20 May 2011 Received in revised form 13 January 2012 Accepted 23 January 2012 Available online 25 February 2012 Keywords: Farm simulation Dairy Pasture Carbon footprint Greenhouse gas Environment
a b s t r a c t The biological and physical processes of an intensively-managed rotational pasture-based dairy and a conﬁnement fed dairy in the southeastern United States were simulated with the Integrated Farm System Model (IFSM) to evaluate management effects on greenhouse gas emissions, soil carbon sequestration, carbon footprint, nitrate leaching, ammonia volatilization, erosion, phosphorus runoff, and phosphorus accumulation in the soil. Edge-of-ﬁeld erosion and phosphorus runoff were less for the pasture-based dairy per unit of land and per unit of milk produced, but nitrate leaching was greater. Ammonia emissions were greater from the conﬁnement dairy because of the greater handling of manure. Greenhouse gas emissions per cow were greater on the conﬁned dairy, but with greater milk production per cow, the carbon footprint of milk produced was similar to that of the pasture-based dairy. Considering the potential soil carbon sequestration following the conversion of crop land to perennial grassland, the carbon footprint of the milk produced by the pasture-based dairy was slightly less than that of the conﬁnement dairy. The results of this study were generally consistent with similar simulation studies done in the northeastern US with variations due to regional differences in climate, soil type, and agronomic practices. Simulated changes in production practices predicted that increasing milk production through improved animal management or feeding more corn decreased the carbon footprint of milk produced by the pasture-based dairy, while decreasing the inorganic nitrogen fertilizer application rate or raising replacement heifers on the farm had little effect. On the conﬁnement dairy, covering the manure storage and ﬂaring the biogas decreased the carbon footprint, using higher producing, pure-bred Holstein cows or producing less forage on the farm increased the footprint, and eliminating free-stall barns and placing all cattle on pasture had little effect on the footprint. The IFSM was capable of adapting to the climate and production practices of the southeastern US, but further improvements could be made to better represent the cropping practices used in this region. Published by Elsevier Ltd.
1. Introduction Dairy farming in the United States has evolved towards the conﬁnement of cattle and the feeding of imported feeds to achieve higher and more consistent milk production rates (Winsten and Petrucci, 2003). In contrast, pasture-based dairies feed cattle by growing grasses on the farm and rotating grazing cattle through carefully managed paddocks. The warm climate and soils in Georgia are conducive to pasture-based dairying. This favorable environment combined with the state’s milk deﬁcit, i.e., more milk is consumed than is produced, has led many to consider the viability ⇑ Corresponding author. Tel.: +1 814 865 2049; fax: +1 814 863 0935. E-mail address: [email protected]
(C. Alan Rotz). 0308-521X/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.agsy.2012.01.005
of expanding pasture-based dairying in this region (Hill et al., 2008). A concern in the expansion of this type of dairy production is the effect on the environment. Greenhouse gas emissions are a current societal focus, but this must be considered along with effects on air and water quality. Life cycle assessments have been performed to evaluate the greenhouse gas emissions that result from various types of dairy production in other regions of the United States (Rotz et al., 2009, 2010; FAO, 2010). However, to date, no published study has reported a life cycle assessment of representative farms in the southeastern region. Environmental beneﬁts of reduced erosion, phosphorus runoff, and ammonia emissions have been shown for well-managed pasture systems (Rotz et al., 2009), but this type of analysis has not been performed on
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
intensively managed, large-scale dairy production systems in the mild climate of the southeastern US. Reports such as ‘‘Livestock’s Long Shadow’’ (FAO, 2006) have shown that environmental impacts are a concern for public and regulatory approval of any large scale agricultural operation. Recent work in Ireland has evaluated greenhouse gas emissions from grass-based dairy systems and shown that a holistic LCA approach provides the most comprehensive evaluation of production systems (O’Brien et al., 2011). In a LCA comparison of extensive dairy production systems in New Zealand to intensive systems in Sweden, the extensive systems were found to have a slightly lower carbon footprint (Flysjo et al., 2011). Del Prado et al. (2010) used a whole farm modeling approach to evaluate GHG mitigation strategies in UK dairy farms that highlighted the interactions and tradeoffs among different forms of pollution and farm proﬁtability. Considering the importance and complexity of environmental evaluation, it is useful for the southeastern dairy industry to evaluate and use existing tools for determining the impacts of their farms on the environment. An available tool for evaluating the environmental impact of dairy production systems is the Integrated Farm System Model (IFSM; USDA-ARS, 2011a,b). This tool has been used to compare the performance and environmental impacts of pasture and conﬁnement based dairy production systems in the northeastern US. Our objectives were to evaluate the adaptability of the model to a milder southern climate, and to use the model to assess the environmental effects of two dairy farms in Georgia using pasture and conﬁnement feeding strategies. 2. Materials and methods Two actual dairy farms in Georgia were simulated using the IFSM. Comparing the model’s outputs to the actual farm performance provided an indication of the ability of the model to assess conditions in this region. Also, an examination of how management changes affected the predicted environmental impacts gives farm managers guidance on how management of their farm can positively or negatively affect the environment. 2.1. Model description The IFSM simulates the major biological and physical processes and interactions of a crop, beef, or dairy farm, providing a tool for comprehensive environmental assessment (Rotz et al., 2011). Crop
production, feed use, and the return of manure nutrients back to the land are simulated over 25 years of weather (Fig. 1). Growth and development of grass, alfalfa, corn, soybean, and small grain crops are predicted based upon daily soil and weather conditions. Tillage, planting, harvest, storage, feeding, manure handling, and irrigation operations are simulated to predict resource use, timeliness of operations, crop losses, and nutritive changes in feeds. Feed allocation and animal response are related to the nutritive value of available feeds and the nutrient requirements of the animal groups making up the herd. The quantity and nutrient content of the manure produced is a function of the feed consumed. Nutrient ﬂows through the farm are modeled to predict nutrient accumulation in the soil and loss to the environment (Rotz et al., 2011). Environmental impacts include N (ammonia) volatilization from manure sources, soil denitriﬁcation and nitrate leaching losses, sheet and rill erosion, and soluble and sediment-bound phosphorus runoff. Carbon dioxide, methane, and nitrous oxide emissions are tracked from crop, animal, and manure sinks and sources to predict the net greenhouse gas emission (Fig. 1). Secondary emissions occurring during the production of resources used on the farm, such as purchased feed, fuel, electricity, machinery, fertilizer, pesticides and purchased animals, are also included in the calculation of a carbon footprint of the production system. A portion of the net greenhouse gas emission is allocated to animals sold in proportion to their economic value relative to the milk sold. Whole-farm mass balances of nitrogen, phosphorus, potassium, and carbon are determined as the sum of all imports in feed, fertilizer, deposition, and crop ﬁxation minus the exports in milk, excess feed, animals, manure, and losses leaving the farm. Carbon sequestration can also be an important consideration in the evaluation of greenhouse gas emissions from pasture-based systems. Normally, the conversion of cropland to permanent perennial grassland stimulates the sequestering of carbon for up to 50 years until a new balance is established between carbon inputs and losses from the soil (Franzluebbers and Follett, 2005). To represent this process, the Comet-VR model (USDA-NRCS, 2009) was used to estimate carbon sequestration attained through changes in cropping practices. The predicted sequestration was then used to adjust the net greenhouse gas emission and carbon footprint following the conversion from tilled cropland to perennial grassland. The IFSM has been used to perform environmental assessments of various dairy farming methods in the northern and western US (Rotz et al., 2009, 2010). By setting model parameters to
Fig. 1. Processes on a dairy farm and their predicted environmental impacts as simulated by the Integrated Farm System Model (USDA-ARS, 2009).
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
characterize actual farms in Georgia and simulating those farms using historical weather data and soil characteristics for Georgia, the model was adapted to an environmental assessment of farms in this region. Our environmental assessment included the evaluation of greenhouse gas emissions and carbon footprints of two dairy production systems using a life cycle assessment (LCA). An LCA quantiﬁes the environmental impacts of a given product or process by accounting for all resources used in the process. An assessment over the full life cycle of a product is referred to as a cradle-tograve analysis for materials that are not recycled in the system and cradle-to-cradle analysis for materials that are reincorporated in the system’s overall process (Owens, 1997). Life cycle assessments of agricultural production typically end when products leave the farm. This forms a cradle-to-farm gate partial LCA (Rotz et al., 2010). By deﬁnition, the carbon footprint of a product is the total greenhouse gas emission associated with that product. The three greenhouse gases emitted by dairy farms in substantial amounts are carbon dioxide (CO2), nitrous oxide (NO2), and methane (CH4). In the IFSM, carbon dioxide emission and assimilation on a dairy farm result from carbon ﬁxation in plant growth, soil respiration, plant respiration, animal metabolism, respiration of microorganisms in manure on the barn ﬂoor and during storage, and fossil fuel combustion (Chianese et al., 2009a). Nitrous oxide emissions on a dairy farm result from nitriﬁcation and denitriﬁcation processes in cropland, the manure storage surface crust, and the manure in bedded packs or dry lots (Chianese et al., 2009c). Methane emissions result from enteric fermentation, manure on the barn ﬂoor, manure during storage, losses following manure application, and feces from grazing animals (Chianese et al., 2009b). Greenhouse gases have differing abilities for trapping heat in the atmosphere, known as global warming potential, which is expressed in carbon-dioxide-equivalent units (CO2e). In the IFSM, each unit of nitrous oxide is equivalent to 298 units of CO2, and each unit of methane is equivalent to 25 units of CO2 (IPCC, 2007). A carbon footprint of milk is determined as the net sum of all greenhouse gas emissions converted to CO2e units divided by the energy-corrected milk produced (Rotz et al., 2010). The term ‘impact’ encompasses both the positive and negative effects that agriculture has on the environment. Food production is integral to the survival of civilization, and thoughtfully performed agriculture can beautify the landscape and protect surrounding natural resources (Cederberg and Mattsson, 2000). The focus of our study is on emissions and nutrient losses to the environment, so by design, the more negative connotations are emphasized.
Calves were maintained on the farm in a barn for up to 4 weeks and then transported to another farm. These cattle were returned to the farm when they began lactation to replace culled cows. Cows had a 12-month calving interval with a 60-day dry period. Half of the cows were bred by artiﬁcial insemination on a fall cycle, calving around November 1. The other half was bred by artiﬁcial insemination on a spring cycle, calving around March 1. Lactating and dry cows were maintained on 100 ha of pasture throughout the year. The pasture consisted of two paddock systems irrigated with center-pivot units (Fig. 2). The two paddock systems were divided with high-tensile electric fence into 28 and 22 individual paddocks of about 2 ha in size. Tifton-85 hybrid bermudagrass was established on 77 ha of this pasture, which produced forage during late spring, summer, and early fall. The remaining 24 ha were planted with pearl millet which produced forage during the summer. All 100 ha were overseeded with annual ryegrass, oats, and arrowleaf clover in the fall using a no-till drill. These cool-season annuals produced forage during the late fall, winter, and spring. The pastures were fertilized with 336 kg N/ha of ammonium nitrate and 2.2 t/ha of chicken litter. Lactating cows were milked twice daily in a swing 48 herringbone parlor, which enabled rapid completion of the milking process. All of the manure, urine, and dirt deposited by cattle in the holding area and parlor were washed into grate inlets with pressurized water hoses after each milking. This efﬂuent was carried by gravity ﬂow into the waste management system consisting of a sand-trap, 114 m3 storage tank, and an overﬂow lagoon. Cattle were fed grain during the milking process. The grain consisted primarily of corn supplemented with soybean meal and cotton seed. The daily amount fed varied throughout the year from 2.5 to 9 kg/cow, with an average annual grain use of 3.6 kg/cow per day. The amount fed was inversely related to the energy provided by the available pasture forage. About 10% of the annual forage requirement was fed as needed in the pastures in the form of hay or silage to supplement available pasture forage. The conﬁned dairy maintained a herd size of 700 cows plus replacement heifers with 33% of the cows replaced each year.
2.2. Farm characteristics Two actual dairy farms, one conﬁned and one pasture-based, were represented and simulated using the IFSM. Each farm was chosen based on the willingness of farm managers and owners to cooperate and how well these farms represented other dairies in the region. The farms were promised anonymity, so the farms are described to present the general characteristics of each system, but details that pinpoint the location or identity of the farms are omitted. The management intensive rotational pasture-based dairy maintained a herd size of 500 cows with an annual replacement rate of 10%. The breed composition of the herd was approximately 40% small-framed Holstein or Friesian and 60% Holstein and Jersey cross-breeds. The average mature cow weight of the herd was 500 kg, and annual milk production was 5000 kg/cow with a milk-fat concentration averaging 3.6%.
Fig. 2. Aerial view of the pasture-based dairy. The two irrigated circles are divided into paddocks for rotational grazing.
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
During this study, the dairy was transitioning to a cross-bred herd, and therefore the breed composition was rather complex. The goal of the breeding program was to obtain cattle that were 62.5% Holstein, 25% Swedish Red, and 12.5% Jersey. During this study, the average bodyweight of the herd was 590 kg. Annual milk production was 10,700 kg/cow with a milk fat concentration averaging 3.8%. Lactating cows were milked twice daily in a double-eight herringbone parlor. Calves on the farm were maintained in calf hutches for about 5 weeks and then placed on pasture. Cows were bred year-round using artiﬁcial insemination. Lactating cows were maintained in two free-stall barns with sand bedding (largest buildings in Fig. 3). Dry cows and replacement heifers were maintained on 100 ha of common bermudagrass pasture. These pastures were overseeded with annual ryegrass in the fall with a no-till drill. A ﬂush system was used to clean the concrete ﬂoors of the freestall barns. The waste management system consisted of a sandtrap, solid separator, and three lagoons in series. The water in the lagoons became progressively cleaner in each successive lagoon. Water from the bottom two lagoons was pumped back to the free-stall barns to provide ﬂushing solution. Lactating cows were fed a mixed ration composed of corn silage, ryegrass silage, hay, and grain. All of the silage and hay fed was produced on the farm. By weight, the rations for lactating cows consisted of 45% silage and 55% corn grain and other supplemental feeds (soybean meal, cotton seed, and fat). Crops grown on the farm supplied another dairy in addition to the one in the study, so the exact crop areas used to produce feed for the animals on this farm was not known. An allocation was made by multiplying the total land area of each crop grown by 0.7, which was the ratio of cows on the dairy studied to the total number of cows fed by the crops grown. From this allocation, 176 ha of corn double-cropped with annual ryegrass were grown for silage, and 100 ha of various grasses were grown for pasture and hay harvest. Also, 105 ha of soybeans were grown as a cash crop. Wastewater from the lagoons was pumped onto ﬁelds through center-pivot irrigation systems to fertilize the crops. The center-pivots also irrigated the ﬁelds with fresh water as needed. To account for the land used to feed cattle on the other dairy, 30% of the manure nutrients were exported from the farm. In addition to the recycled manure, approximately
Fig. 3. Aerial view of the conﬁned dairy including the pasture used for feeding replacement heifers and a portion of the cropland used to produce feed for cows.
56 kg N/ha of ammonium nitrate fertilizer was broadcast and incorporated prior to corn planting and 135 kg/ha of potassium was applied to the annual ryegrass. During the study, 6.7 t/ha of poultry litter was applied to 36 ha of corn prior to tillage incorporation. A one pass minimum tillage operation was used prior to corn and soybean planting and ryegrass was established through no-till seeding. A summary of the characteristics of each farm are listed in Table 1. 2.3. Farm simulations The characteristics listed in Table 1 were just a portion of the input parameters required to model farms with the IFSM. A complete list of all model parameters used to describe each farm are documented by Belﬂower (2010). The majority of the model inputs were determined during personal interviews with the managers of each farm. All other inputs were determined by on-farm measurements (e.g. lagoon dimensions, soil tests, silo dimensions) or other available media (e.g. Natural Resource Conservation Service web soil survey, GIS analysis to determine land areas, online databases of curve numbers and forage characteristics). For consistency, both farms were simulated using historical weather data (1980–2004) for Macon, Georgia. This major weather station was within 170 km and approximately equidistant from each farm. The boundaries of the production systems were deﬁned as the boundaries of the farm. Predicted emissions and nutrient losses were those leaving the farm across the physical surface boundaries, into the surrounding atmosphere, or into the groundwater below the root zone. All important amounts of carbon brought onto the farms, such as the imported poultry manure, were included as part of the production system. The carbon footprint LCA also included the emissions associated with the production of resources brought onto the farm. A model can never fully represent all management practices used on a farm. To assure that the most important aspects of the farm were properly represented, an evaluation was used to match certain model outputs with known conditions on the farm. This evaluation assured that the model accurately predicted the quantities and composition of feed produced and consumed and the amount of milk produced. When modeling the pasture-based dairy, the model predicted that nearly all forage was obtained from pasture. About 10% of the forage required was purchased as hay to supplement the pasture, similar to that found on the actual farm. The predicted average annual grain use was 3.6 kg/cow per day, the same as that used on the actual farm. The model also matched the annual milk production of 5000 kg/cow. Crops grown on the conﬁnement dairy were often doublecropped and sometimes triple-cropped during the year. To represent this farm, the farm area was set at 344 ha, the actual area producing crops to feed this herd. The individual crop areas were then set to represent the area of each crop harvested within an annual cycle. The farm parameters were set to represent 100 ha of pasture forage with 176 ha of double-cropped corn and annual ryegrass and 105 ha of soybeans. The farm had multiple corn-silage ﬁelds managed with different planting and harvest schedules. In the IFSM, the corn crop had to be managed as a single unit, so planting and harvest dates were set to represent the most typical strategy used to produce corn silage. An important aspect of representing farms with the IFSM is to properly predict the amount of feeds produced and used on the farm. Lactating cows on the conﬁnement farm were fed mixed rations of approximately 45% silage and 55% grain by weight. Diets formulated by the model for lactating cattle consisted of 40–50% forage depending upon forage quality and stage of lactation. Average annual grain and supplemental feed use including that for
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
Table 1 Major characteristics of the two simulated farms.
Number of cattle Breed Average body weight (kg) Housing
500 Holstein and Jersey Crosses 500 Young calves in bedded barn All cows on pasture
700 Holsteins and Holstein, Jersey, and Swedish Red Crosses 590 Lactating cows in two free-stall barns Dry cows and older heifers on pasture
Milk production Total (kg/cow/year) Milk fat concentration (%) ECM (kg/year)
5000 3.6 2538,000
10,700 3.8 7872,000
Feed production and use Harvested silage (t DM) Grazed forage (t DM) Purchased feed (t DM)
0 2045 595
3372 685a 2719
Dry cows and heifers were placed on pasture to graze on the conﬁned dairy.
replacement animals was 6.7, 2.7 and 1.2 kg/cow/day of corn grain, protein mix, and cotton seed, respectively. To include carbon sequestration effects, the Comet-VR model was used to predict potential sequestration in soils given the historical and current farming practices on each farm (USDA-NRCS, 2009). Carbon sequestration levels decrease with time over the 20- to 50-year period following a change in production practice. With time, the soil approaches a new level of carbon equilibrium and the amount of carbon sequestered diminishes (Rotz et al., 2010). The farmland on the pasture-based dairy was converted from row crops to perennial pasture about 3 years prior to this study. Therefore, for the next few decades carbon can be sequestered in the soil, reducing the carbon footprint of the farm. When comparing different farm production processes with LCA, it is important to scale the environmental impacts produced to the amount of commodity produced by the farm (Capper et al., 2009). Therefore, the environmental impacts predicted by the model were divided by the amount of milk produced on each farm. To account for differences in fat and protein contents, milk production was corrected to 3.5% milk fat and 3.1% milk protein, denoted as Energy Corrected Milk (ECM), as deﬁned in the IFSM (Rotz et al., 2011). 3. Results and discussion 3.1. Environmental impacts predicted by the model Simulated predictions provide insight into the interaction of management variables with the resulting greenhouse gas
emissions and carbon footprint of each farm (Tables 2 and 3). Enteric methane emissions per cow generally increase with the percentage of ﬁber fed in diets (Chianese et al., 2009b). Therefore, the higher forage diets on the pasture-based dairy increase this emission. The higher milk production of the conﬁned dairy herd and the greater feed intake required to produce that milk offsets the difference that lower forage content in the diet has on enteric methane emissions per cow. The long-term manure storage on the conﬁnement farm created considerable emission of methane. This emission source, along with other animal and manure handling differences, led to 70% greater methane emission per cow on the conﬁnement farm. Expressed per unit of ECM, the methane emission from the pasture-based farm was 30% greater than that of the conﬁnement farm (Table 2). Nitrous oxide emissions expressed per cow or per unit of ECM production were greater on the conﬁned dairy due primarily to the difference in soil type. The soil with a relatively high clay content on the conﬁnement dairy created greater nitrous oxide formation and emission compared to the sandy soil of the pasture-based dairy. Also, nitrous oxide emissions increase as the volume and residence time of efﬂuent in the waste management system increase (Chianese et al., 2009c), and both the volume and residence time of stored manure were greater on the conﬁned dairy. Engine emissions of CO2 increase with greater use of harvested crops because more ﬁeld operations are required for tillage, planting, harvesting, and feeding of these crops. Therefore, the emission per ECM from fuel combustion on the conﬁnement farm was 40% greater than that of the pasture-based dairy where little harvesting
Table 2 Simulated annual greenhouse gas emissions from the two dairy farms in Georgia. Greenhouse gas emission Energy Corrected Milk production (ECM) Methane
Carbon dioxide from fuel combustion
Total greenhouse gas Primary sources Secondary sources Net biogenic CO2 Not allocated to milk Total net Potential carbon sequestration
kg kg kg kg kg kg kg kg kg kg kg kg kg kg kg kg
CH4/cow CO2 equiv. CO2 equiv./t ECM N2O/cow CO2 equiv. CO2 equiv./t ECM CO2/cow CO2 equiv. CO2 equiv./t ECM CO2 CO2 CO2 CO2 CO2
equiv. equiv. equiv. equiv. equiv.
2548,000 119 1482,900 582 1.3 200,600 79 229 114,300 45.7
7872,000 202 3525,700 448 6.7 1402,700 178 684 478,500 63.9
1797,800 552,900 750,700 158,300 1441,700 230,000
5406,900 1492,500 2329,200 405,700 4164,500 0
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93 Table 3 Carbon footprints of the two dairies with and without including biogenic sources and sinks of carbon dioxide.
Including biogenic sources and sinks Carbon footprint (kg CO2 equiv./kg ECM) Carbon sequestrationa (kg CO2/kg ECM) Carbon footprint with sequestration (kg CO2 equiv./kg ECM)
0.58 0.09 0.49
0.56 0.00 0.56
Excluding biogenic sources and sinks Carbon footprint (kg CO2 equiv./kg ECM) Carbon sequestrationa (kg CO2/kg ECM) Carbon footprint with sequestration (kg CO2 equiv./kg ECM)
0.88 0.09 0.79
0.87 0.00 0.87
Estimated using the Comet-VR model (USDA-NRCS, 2009).
was done (Table 2). A major portion of engine emissions was from irrigation, which was similar on each farm. The net biogenic assimilation of CO2 by crops produced on the conﬁnement dairy were much greater than that assimilated in the pasture-based dairy due to greater yields and the double-cropping systems used (Table 2). While carbon dioxide is emitted from other sources on a dairy, such as animal metabolism and manure respiration during storage, the emission and assimilation of CO2 that occurs during the production of feed heavily inﬂuences the total biogenic emissions of CO2 on each farm (Chianese et al., 2009a). Secondary emissions were greater on the conﬁned dairy because of greater use of resources produced off-farm including fuel, electricity, chemicals, and purchased grain. Another difference involved the production of heifers, where the conﬁned dairy raised replacement heifers on-site while the pasture-based dairy produced replacement heifers on a neighboring farm. Also, the conﬁned dairy raised more replacement heifers than needed to meet the replacement rate of lactating animals on the farm. The LCA in IFSM accounted for this exchange of heifers by including a secondary emission factor for the weight of heifers bought and sold from the farm (Rotz et al., 2010). Therefore, the emissions for the production of extra heifers sold were removed in the LCA, and the emissions occurring during the production of heifers produced off-farm were included. With all of these factors considered, the total secondary emission per unit of ECM was similar across farms. The conversion of land from annual tillage to perennial grassland potentially allows the soil to sequester carbon. The CometVR model (USDA-NRCS, 2009) predicted that the conversion of farmland from row crops to perennial grassland on the pasturebased dairy, would create an average annual sequestration of 230,000 kg CO2/ha during the transition to a new soil carbon equilibrium. Comet-VR was also used to analyze the carbon dynamics in the soils of the conﬁned dairy, where carbon was not sequestered in the soil because of the large amount of land that was intensively tilled each year. Thus, we assumed the soil in the conﬁned dairy was in carbon equilibrium. The IFSM calculates the carbon footprint of a farm with two methods. The ﬁrst includes biogenic CO2, which accounts for the carbon dioxide assimilated in the production of feed along with that emitted through plant, animal and microbial respiration (Chianese et al., 2009a). The second method is to ignore all biogenic sources and sinks of CO2 and include only that emitted through the burning of fossil fuels in farm operations. For a partial LCA that stops at the farm gate, the carbon footprint is best represented by including biogenic sources and sinks (Rotz et al., 2010). However, LCAs commonly exclude biogenic CO2 in the analysis of greenhouse gas emissions and carbon footprint, and this increases the carbon footprint of milk production (USDA-ARS, 2011b).
Similar differences in carbon footprint were found between the two production systems whether biogenic CO2 was included or excluded from the LCA. When included, the carbon footprint of milk from the pasture-based dairy was 4% greater than that of the conﬁnement dairy (Table 3). This difference was primarily due to the lower milk production on pasture, which required a greater maintenance of animals per unit of milk produced. When the potential sequestration of carbon in the soil of the perennial grassland was considered, the carbon footprint of milk from the pasture-based dairy was 12% less than that of the conﬁnement dairy. Carbon sequestration by soil under pasture would slow over time as the soil reaches a new level of equilibrium, and this beneﬁt would diminish. Differences in the model’s predictions of soil and water resource impacts between the two production systems (Table 4) result largely from differences in management practices and farm characteristics. For instance, erosion was greater on the conﬁned dairy because of soil type and tillage practices. The clay soils of the conﬁnement dairy site generated more runoff than the sandy soils of the pasture-based dairy, and this runoff resulted in a prediction of increased erosion and phosphorus loss. Also, minimum tillage for the establishment of row crops on the conﬁned dairy created substantially higher erosion rates than the perennial grass cover on the pasture-based dairy. Phosphorus accumulation in the soil was similar between the two dairies. Nitrate leaching was much greater on the pasture-based dairy because nitrate and water inﬁltrate and leach more readily through sandy soil than clay soil. Also on the pasture-based dairy, the greater than needed protein content and more degradable protein in pasture forage led to greater nitrogen excretion in urine (Rotz et al., 2011). Urine was deposited directly onto pasture in concentrated spots. The high concentration of nitrogen in these spots exceeded potential crop uptake allowing more nitrate leaching through the soil proﬁle. Ammonia volatilization per unit of ECM was about 50% greater on the conﬁned dairy compared to that emitted from the pasturebased dairy for a few reasons (Table 5). Urine deposits on pasture are absorbed into the soil, reducing ammonia emission. In contrast, manure lying on free-stall barn ﬂoors and in long-term storage facilities readily contributes to ammonia emission. Also, the conﬁned dairy applied more poultry litter and stored dairy manure onto ﬁelds, and ammonia volatilized following each application. Although the ammonia emission per cow was over 3 times greater on the conﬁnement farm, the greater milk production reduced the difference expressed per unit of ECM. Variations in nutrient ﬂows through each farm result largely from the differences in imported nutrients, exported milk, and crop uptake (Table 6). The amount of nitrogen imported to the pasturebased dairy, primarily as fertilizer, was relatively high compared to that exported in milk. The high use of nitrogen on this sandy soil
Table 4 Simulated annual erosion, phosphorus runoff, and nitrate leaching for the two types of dairy farms in Georgia. Environmental impact category
Erosion sediment loss (kg/ha) Erosion sediment loss (g/kg ECM) Sediment-bound P runoff (kg/ha) Sediment-bound P runoff (g/kg ECM) Soluble P runoff (kg/ha) Soluble P runoff (g/kg ECM) Soil P accumulation (kg/ha) Soil P accumulation (g/kg ECM) Nitrate N leaching (kg/ha) Nitrate N leaching (g/kg ECM)
40 1.57 0.028 0.0011
2443 107 2.49 0.109
0.057 0.0022 4.0 0.15 151 5.81
0.106 0.0046 1.0 0.0 7.0 0.31
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
Table 5 Predicted annual ammonia emissions for the two dairy farms in Georgia. Environmental impact category
Total ammonia volatilization (kg/cow) In housing facility During manure storage Following ﬁeld application During grazing Ammonia N volatilization (g/kg ECM)
26.4 3.3 0 4.1 18.9 4.3
88.5 40.6 29.2 10.8 7.9 6.5
led to high leaching and denitriﬁcation losses. The predicted nitrogen concentration in groundwater drained below the root zone was also high compared to a recommended drinking-water standard of 10 ppm (Table 6). On both dairies, the export of phosphorus was similar to that imported, essentially providing a whole farm balance of this nutrient (Table 6). Greater amounts of potassium were imported than that exported from the conﬁnement farm, which led to greater soil accumulation of that nutrient on this farm.
Verge et al., 2007 and Masuda, 2007). In previous work, the IFSM was veriﬁed to predict similar carbon footprints as those determined in these studies when similar system boundaries and assumptions were used (Rotz et al., 2010). Greenhouse gas emissions and other losses determined for the two dairies were compared to those predicted by the IFSM for similar dairy production systems in the northeastern US (Rotz et al., 2009; Fig. 4). The consistency and differences in the results of the two studies support the ability of the IFSM to adapt to the Southeast. The emission categories with large variations between the studies highlight the regional climate, soil, and agricultural differences between Georgia and Pennsylvania. The sandier soil in the southeast grazing dairy led to less erosion and nutrient runoff and greater nitrate leaching compared to the similar production system in the northeast (Fig. 4). The double
3.2. Evaluation of predicted environmental impacts The best evaluation of this type of simulation study is to compare predicted losses to measured values, but measuring emissions throughout the life cycle of milk production is difﬁcult, expensive, and likely impossible. Some evaluation was done though, by comparing predicted emissions and nutrient losses to limited available data. Two published studies have assessed the greenhouse gas emissions from dairy production systems using the IFSM (Rotz et al., 2009) or a closely related but simpler tool, DairyGHG (Rotz et al., 2010). Several other studies have conducted life cycle assessments to determine net emissions from milk production systems (Cederberg and Mattsson, 2000; Basset-Mens et al., 2009; FAO, 2010;
Fig. 4. Comparison of selected environmental impacts to those predicted in a similar study conducted in the northeastern US by Rotz et al. (2009).
Table 6 Average annual nutrient balances for the two dairy farms simulated by the Integrated Farm System Model over 25 years of climate data for Macon, GA. Nutrients available, used, and lost
Nitrogen imported to farm Nitrogen exported from farm Nitrogen import/export Nitrogen available on farm Nitrogen lost by volatilization Nitrogen lost by leaching Nitrogen lost by denitriﬁcation Nitrogen concentration in leachate Crop removal over that available on farm Phosphorous imported to farm Phosphorous exported from farm Phosphorus import/export Phosphorous available on farm Phosphorous loss in runoff and leachate Soil phosphorous build up Crop removal over that available on farm Potassium imported to farm Potassium exported from farm Potassium import/export Potassium available on farm Potassium loss through runoff Soil potassium build up Crop removal over that available on farm Carbon imported to farm Carbon exported from farm Carbon loss as carbon dioxide Carbon loss as methane Carbon loss through runoff
kg/ha kg/ha kg/ha kg/ha kg/ha kg/ha ppm % kg/ha kg/ha kg/ha kg/ha kg/ha % kg/ha kg/ha kg/ha kg/ha kg/ha % kg/ha kg/ha kg/ha kg/ha kg/ha
563 148 3.8 964 110 151 124 91 55 29.8 25.6 1.2 76.5 0.2 4.0 100 34.8 38.8 0.9 620 31.0 0.0 109 49,900 1700 47,700 452 0.0
47.6 0.0 NA 47.3 4.2 84.6 63.2 78.7 3.0 0.9 0.0 NA 10.2 0.2 0.9 36 5.7 0.0 NA 91.7 4.6 6.9 53 8400 0.0 8400 0.5 0.0
484 261 1.9 518 148 7.0 93.3 3.7 56 45.5 41.9 1.1 33.1 2.6 1.0 88 168 103 1.6 212 10.6 54.7 69 19,600 2300 16,900 308 8.1
16.9 9.7 NA 21.3 7.3 6.6 30.1 3.5 2.0 1.2 1.3 NA 22.8 2.0 3.4 19 8.4 4.6 NA 128 6.4 13.2 15 1900 290 1700 9.1 4.6
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
cropping strategy used on the southeastern conﬁnement dairy provided year around ground cover and crop uptake of nitrogen, which reduced erosion, nutrient runoff, and nitrate leaching compared to the similar production strategy in the northeast. Ammonia volatilization was not affected much by location in these studies. Nitrous oxide emissions were less in the southeastern farms. For the pasture-based farm, the sandier soil reduced nitriﬁcation and denitriﬁcation processes. With the conﬁnement strategy, doublecropping increased nitrogen uptake, reducing the soil nitrogen available for nitriﬁcation and denitriﬁcation processes during the winter and early spring seasons. Methane emissions were not affected much by location, but differences in manure handling practices gave a little lower emission for the southeastern conﬁnement farm. Differences in methane and nitrous oxide emissions together gave lower net greenhouse gas emissions for the southeastern farms. This difference along with differences in milk production gave similar carbon footprints for the grazing dairies when potential carbon sequestration was not considered. The sandy soil in the southeast had a much lower potential for sequestering carbon, so the carbon footprint of milk produced on the southeast grazing dairy was not reduced nearly as much through potential sequestration as found for the northeast grazing dairy. While measured values of greenhouse gas emissions from dairy farms were not available, other environmental impacts resulting from dairy production have been measured. Therefore, effort was made to ﬁnd measured values of environmental impacts from conditions similar to those of the pasture-based and conﬁned dairies in this study. Predicted environmental impacts were compared to measured values from a variety of sources (Table 7). Compared to the limited data available, the model generally predicted values within the range of values obtained from the other sources. One exception was the erosion sediment loss from the conﬁnement dairy. The large sediment loss predicted by the USLE was representative of a conventional tillage system with bare soil throughout much of the year. The lower value predicted by the IFSM better represented the minimum tillage and double cropping system used on this actual farm. The IFSM predicted leaching of nitrate on the pasture-based dairy was high due to the relatively high application of nitrogen fertilizer on this sandy soil. Potential carbon sequestration on this farm was also low compared to that reported, likely because of the lower sequestration potential for the sandy soil on this farm.
to reduce the carbon footprint of dairies (Chianese et al., 2009a, 2009b, 2009c, 2009d and Rotz et al., 2010). Suggested improvements in management have included increased production per animal, more grain and higher quality forage in rations, reduced manure storage time, covering the manure storage and burning the biogas produced, incorporating managed rotational grazing into conﬁnement operations, and reducing the resource inputs to the farm. The IFSM was used to analyze four potential changes in management on both the pasture-based and conﬁned dairies to determine how these changes affected their carbon footprints. On the pasture-based dairy, the following four management changes were modeled: Annual milk production was increased from 5000 to 6100 kg/ cow assuming that changes in cattle genetics and management could provide this increase in production. The inorganic nitrogen fertilizer application rate on grassland was reduced from 336 kg/ha to 168 kg/ha. In the model, nitrogen availability is reduced, potentially reducing grass yield and protein content (Rotz et al., 2011). All replacement heifers were grown on the farm instead of being raised on a nearby farm. This change affected the carbon footprint, but the LCA included a carbon footprint of heifer production regardless of where they were grown. More corn silage and grain were incorporated into cattle diets with an increase in annual milk production to 6100 kg/cow. Of the four management options, only those that increased production provided a substantial reduction in the carbon footprint of the production system. A 22% increase in milk production reduced the footprint by about 15% whether the increased production was obtained through animal management or feeding more corn. Reduced fertilizer use and raising replacement heifers on the farm had little effect on the carbon footprint of the full production system (Fig. 5). Through further simulations of the conﬁned dairy, the following four management changes were studied:
3.3. Management changes and the resulting impact on the environment Previous studies have conducted sensitivity analyses of farm input parameters to emissions and used the results to suggest ways
The cross-breeding program currently undertaken on the farm was removed by increasing the average body weight, decreasing the milk fat content, and increasing the milk production to levels representative of pure-bred, large frame Holstein cattle. The land area used to produce silage was reduced by 50%, thereby causing the farm to import more forage and grain. The manure storage was covered and a ﬂare was used to burn the captured biogas converting the methane to carbon dioxide. This was modeled as documented by Chianese et al., 2009b and Rotz et al., 2010.
Table 7 Comparison of model predicted annual environmental impacts to measured or empirical values.
a b c d e f g
Environmental impact category
Pasture-based dairy model
Conﬁnement dairy model
Erosion sediment loss (kg/ha) Sediment-bound P runoff (kg/ha) Soluble P runoff (kg/ha) Total P runoff (kg/ha) Soil P accumulation (kg/ha) Nitrate N leaching (kg/ha) Ammonia N volatilization (kg/ha) Potential sequestered carbon (kg CO2/ha)
40 0.028 0.056 0.084 4.0 151 110 2178
54b NMFa NMF 0.007–13c NMF 1.7–38e 7–186e 11,980f
2443 2.49 0.106 2.60 1.0 7.0 148 0.00
15,700b NMF NMF NMF 4.5d NMF NMF Negativeg
NMF = No measurement found. The Universal Soil Loss Equation (USLE). Romeis, 2008. Conﬁned Dairy Farm Manager. Personal Communication. September 1, 2010. Eason, 2010. Dr. Nicholas Hill, University of Georgia. Personal Communication. September 8, 2010. Franzluebbers and Follett, 2005.
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93
modiﬁcation throughout its 40 years in operation to improve efﬁciency and decrease negative environmental impacts. In contrast, the pasture-based dairy has been operating for less than 5 years, and the farm managers are still learning how best to operate the farm. It is likely that the pasture-based dairy will become more efﬁcient as the farm managers reﬁne their system.
Fig. 5. Effects of various changes in management on the carbon footprint of the pasture-based and conﬁned dairies.
Free stall barns were removed and all cattle were placed on pasture throughout the year. In addition to the 100 ha currently used for grazing, an additional 100 ha of land currently used for annual ryegrass and corn silage production were converted to perennial pastures. The existing common bermudagrass base was maintained, although it was assumed that clover and annual ryegrass were no-tilled into the existing forage mix each fall. Milk production was decreased to 8200 kg/cow per year. Of the four strategies tested on the conﬁnement dairy, only the use of the manure storage cover and the burning of biogas provided a substantial reduction in the carbon footprint (Fig. 5). Both the use of pure-bred Holsteins and the importation of more feed caused substantial increases in the carbon footprint of the production system. The use of grazing for the lactating herd along with a reduction in milk production had a relatively small impact on the carbon footprint. This further illustrates that the use of grazing may not have much impact on the carbon footprint of milk production when compared to conﬁnement feeding (Rotz et al., 2009). When comparing these very different production strategies, many assumptions must be made and these assumptions can favor either of the two systems. Therefore, it is difﬁcult to conclude that either system provides a consistent advantage in lower carbon footprint. The changes in the modeled carbon footprint resulting from alternative management scenarios demonstrate the potential of the IFSM to positively impact the dairy industry. While this study did not address economics, utilization of the model’s economic prediction capabilities would further illustrate the potential economic impact of each of these changes in management, thereby providing dairy producers with guidance on decreasing their carbon footprint while maintaining or improving proﬁtability. While emissions and energy consumption are an important consideration for any dairy producer concerned about the environment, when a change in management is considered for the purpose of decreasing negative environmental impacts, a holistic approach that considers all factors must be taken. For instance, a particular change might decrease greenhouse gas emissions but increase erosion and nutrient runoff. These consequences must be weighed against each other to establish which management decision is better for the environment. The environmental impacts of changes in management have been explored in more detail by Belﬂower (2010). It is worth noting that the conﬁned dairy in the study has a noted record of environmental stewardship, with continual
The IFSM provided a valuable tool for evaluating the environmental impact of dairy farms in the southeastern United States, even though it could not fully represent the triple cropping practices sometimes used in this region. The carbon footprints of the two dairy production systems were similar with the footprint of the pasture-based system about 4% greater than that of the conﬁnement system. When the potential for carbon sequestration was considered, the carbon footprint of the pasture-based dairy was 12% less than that of the conﬁnement dairy when biogenic sources and sinks of CO2 were included and 9% less with the exclusion of biogenic CO2. Other environmental impacts varied widely between the farms. Erosion and phosphorus runoff were much greater on the conﬁned dairy because of the high clay content in the soil and the large area of land tilled each year to produce annual crops. Nitrate leaching was greater on the pasture-based dairy because of the coarser soil texture and because urine and feces deposits on pasture created high nitrogen concentrations on small land areas. Ammonia volatilization was greater from the manure on barn ﬂoors, during storage, and following land application on the conﬁned dairy compared to emissions from grazing animals on the pasture-based dairy. Management changes were explored that provided reductions in the carbon footprint of milk production using either production strategy. Because different land bases were used for the two dairy farms, this analysis was not intended to provide a comprehensive comparison of pasture-based and conﬁnement feeding dairy production systems. This analysis does support that environmental beneﬁts for well-managed grazing-based dairy systems primarily come through reduced erosion and phosphorous runoff and reduced gaseous emissions from manure, such as ammonia. The carbon footprint of milk production systems is dependent upon many characteristics of the production systems compared, so a general conclusion cannot be made in comparing the carbon footprint of milk produced by pasture-based and conﬁnement feeding strategies.
References Basset-Mens, C., Ledgard, S., Boyes, M., 2009. Eco-efﬁciency of intensiﬁcation scenarios for milk production in New Zealand. Ecol. Econ. 68 (6), 1615–1625. Belﬂower, J., 2010. Environmental Assessment of Pasture-based and Conﬁned Dairy Farms in Georgia. University of Georgia, Athens, GA, Master’s Thesis. Capper, J.L., Cady, R.A., Bauman, D.E., 2009. The environmental impact of dairy production: 1944 compared with 2007. J. Anim. Sci. 87 (6), 2160–2167. Cederberg, C., Mattsson, B., 2000. Life cycle assessment of milk production – a comparison of conventional and organic farming. J. Cleaner Prod. 8 (1), 49–60. 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 (4), 1301–1312. Chianese, D.S., Rotz, C.A., Richard, T.L., 2009b. Simulation of methane emissions from dairy farms to assess greenhouse gas reduction strategies. Trans. ASABE 52 (4), 1313–1323. Chianese, D.S., Rotz, C.A., Richard, T.L., 2009c. Simulation of nitrous oxide emissions from dairy farms to assess greenhouse gas reduction strategies. Trans. ASABE 52 (4), 1325–1335. Chianese, D.S., Rotz, C.A., Richard, T.L., 2009d. Whole-farm greenhouse gas emissions: a review with application to a Pennsylvania dairy farm. Trans. ASABE 25 (3), 431–442. Del Prado, A., Chadwick, D., Cardenas, L., Misselbrook, T., Scholeﬁeld, D., Merino, P., 2010. Exploring systems responses to mitigation of GHG in UK dairy farms. Agric. Ecosys, Environ. 136, 318–332.
J.B. Belﬂower et al. / Agricultural Systems 108 (2012) 84–93 Eason, N., 2010. Nitrogen Dynamics on Pasture-based Dairy Farms in Georgia. Master’s Thesis. University of Georgia, Athens, GA. FAO, 2006. Livestock’s Log Shadow. Food and Agriculture Organization, Rome, Italy. FAO, 2010. Greenhouse Gas Emissions from the Dairy Sector: A Life Cycle Assessment. Food and Agriculture Organization, Rome, Italy. Flysjo, A., Henriksson, M., Cederberg, C., Ledgard, S., Englund, J., 2011. The impact of various parameters on the carbon footprint of milk production in New Zealand and Sweden. Agric. Syst. 104, 459–469. Franzluebbers, A., Follett, R., 2005. Greenhouse gas contributions and mitigation potential in agricultural regions of North America: introduction. Soil Tillage Res. 83 (1), 1–8. Hill, N., Hancock, D., Cabrera, M., Blount, A., 2008. Improved efﬁciency of pasturebased dairies using complementary pasture species and irrigation scheduling. , October 2009. Intergovernmental Panel on Climate Change (IPCC), 2007. Climate change 2007: The physical science basis: contribution of working group I to the fourth assessment report of the Intergovernmental Panel on Climate Change’’. (Accessed 21.09.10). O’Brien, D., Shalloo, L., Buckley, F., Horan, B., Grainger, C., Wallace, M., 2011. The effect of methodology on estimates of greenhouse gas emissions from grassbased dairy systems. Agric., Ecosys. Environ. 141, 39–48. Masuda, K., 2007. Environmental Impact Assessment of Dairy Farms Using the Life Cycle Assessment Method (LCA). Memories of the Research Faculty of Agriculture-Hokkaido, Hokkaido University, Japan. Owens, J., 1997. Life cycle assessment: constraints on moving from inventory to impact assessment. J. Indust. Ecol. 1 (1), 37–49.
Romeis, J., 2008. Phosphorus Loading in Agricultural and Forested Headwater Streams in the Upper Etowah River Basin, Georgia. Doctoral Dissertation. University of Georgia, Athens, GA. Rotz, C.A., Corson, M.S., Chianese, D.S., Montes, F., Hafner, S.D., Jarvis, R., Coiner, C.U., 2011. Integrated Farm System Model: Reference Manual. USDA Agricultural Research Service. Available at: . (accessed 05.05.11). Rotz, C.A., Soder, K.J., Skinner, R.H., Dell, C.J., Kleinman, P.J., Schmidt, J.P., Bryant, R.B., 2009. Grazing can Reduce the Environmental Impact of Dairy Production Systems. Forage and Grazinglands. doi:10.1094/FG-2009-0916-01-RS. Rotz, C., Montes, F., Chianese, D., 2010. The carbon footprint of dairy production systems through partial life cycle assessment. J. Dairy Sci. 93 (3), 1266–1282. USDA-ARS, 2011a. The Integrated Farm System Model (IFSM). Pasture Systems and Watershed Mgt. Res. Unit, University Park, PA. Online. . USDA-ARS, 2011b. The Dairy Greenhouse Gas Model (DairyGHG). Pasture Systems and Watershed Mgt. Res. Unit, University Park, PA. . USDA-NRCS, 2009. The Voluntary Reporting of Greenhouse Gases-Carbon Management Evaluation Tool (COMET-VR)’’. Natural Resource Ecology Lab, Colorado State Univ., Fort Collins, CO. Verge, 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. Winsten, J.R., Petrucci, B.T., 2003. Seasonal Dairy Grazing: A Viable Alternative for the 21st Century. American Farmland Trust, Washington, DC. 26p.