Management to reduce N losses in animal production requires a whole-farm approach. As discussed above, many changes can be made to reduce N losses in each step of manure management between animal excretion and crop uptake. However, the benefit for reducing the loss in any one component is low if steps are not taken to reduce losses occurring in other components. For example, reducing ammonia emission in the housing facility has little benefit if that retained N is simply lost due to poor management during manure storage and field application. Reducing ammonia emissions also may not provide overall benefit if that additional manure N leads to greater losses through denitrification and leaching. These losses of nitrous oxide emitted to the atmosphere and nitrate in groundwater may have a greater long-term cost to society than the ammonia emission. Only by providing similar levels of management to animal feeding, housing, manure storage, and field application can production systems be developed with reduced or optimum environmental impact.
Considerable effort is being given to farm level assessment and management of N. A comprehensive review is beyond the scope of this paper, but a brief review is included to demonstrate the type of work being conducted. Research efforts can generally be divided between actual farm studies and modeling studies. Although these two categories generally represent the type of work done, in practice these two approaches often join to provide the most comprehensive assessment and application of research on farming systems.
An awareness of the environmental problem in animal production began with the measurement of the potential nutrient imbalance in production systems (Bacon et al., 1990). Economic incentives during the last half of the last century drove farms toward greater animal densities per unit of land with less integration between crop and animal production. Numerous studies have illustrated the potential for excess N on the farm where feed and fertilizer inputs exceed the N contained in products sold off the farm. Over the long term, this excess N must be lost in some form to the environment. Case studies have taken an in-depth look at specific farms to predict where this N is lost (e.g., Hutson et al., 1998).
Experimental farms have been developed to closely monitor N flows through animal production systems. To evaluate grassland management systems for beef production, three farmlets were compared at two sites in South West England (Laws et al., 2000). Tactics including timely manure application, slurry injection, early housing of cattle, and the use of a mixed grass and white clover sward were successful in reducing surplus N, but animal and herbage production were reduced. The Karkendamm farm in Germany was established by the University of Kiel to measure N fluxes in the soil-plant-animal system (Taube and Wachendorf, 2000). Various management strategies in crop production and animal feeding were evaluated at the farm level with a goal of measuring and reducing all N flows and losses in dairy production.
A prototype farm in The Netherlands (De Marke) has been in operation for over 10 yr to research and demonstrate efficient nutrient management on Dutch dairy farms (van Keulen et al., 2000). The farm was designed to meet stringent environmental norms for N, P, and foreign substances, such as pesticides. Goals included reducing the annual surplus N on the farm to 128 kg/ha, with annual ammonia emission limited to 30 kg N/ha, nitrous oxide emission limited to 3 kg N/ha, and nitrate in upper groundwater below 50 mg/L. The farm was intentionally placed on soil that was among the driest and most leaching prone in The Netherlands.
Several techniques or strategies were used at De Marke to increase nutrient use efficiency and reduce losses to the environment. This began with efficient feeding of the animals. A unique floor system was used in the housing facility to separate urine from feces, which reduced the transformation of N to volatile ammonia and its subsequent loss. Urine and feces were combined and stored up to 6 mo in a covered storage tank. Manure was applied through shallow injection on grassland and deep injection on arable land. Minimum amounts of N fertilizer (120 kg N/ha) were used on grassland to maintain adequate yields and protein contents with no N or P fertilizer applied to other crops. A "catch crop" of annual ryegrass was seeded during the cultivation of corn land about 8 wk after the corn was established. In autumn, following corn harvest, this crop took up residual mineralized N to reduce potential leaching to groundwater. Implementing these N-conserving technologies reduced fertilizer N input by 74%. The farm has successfully demonstrated that the stringent environmental goals of The Netherlands can be met on a dairy farm (van Keulen et al., 2000).
A number of research efforts have focused on developing models to predict N loss during manure handling and following field application. Both empirical and mechanistic approaches have been used to develop models of ammonia emission during animal housing, manure storage, and field application (Hutchings et al., 1996; Ni, 1999; Sogaard et al., 2002). After the manure is incorporated into the soil, mechanistic models have been used to simulate N transformations through nitrification and denitrification and to predict nitrate leaching into groundwater and N emissions into the atmosphere (Shaffer et al., 1991; Eckersten et al., 1998; McGechan and Wu, 2001). These models have improved our understanding of the dynamic processes involved, and they provide ways of predicting or quantifying losses in each component of animal production.
Component models of N loss have been linked or combined to predict or assess N losses in whole-farm production systems. Bussink and Oenema (1998) compared dairy production systems in The Netherlands and found that up to threefold reductions in ammonia loss were possible along with marked reductions in mineral fertilizer use. In a comparison of dairy farming systems in the United Kingdom, Jarvis et al. (1996) found that using a tactical approach to fertilizer application, injecting slurry, or using 50% corn silage provided substantial reductions in ammonia, nitrous oxide, and nitrate losses. Dou et al. (1996) developed a computer worksheet to compare efficiencies of N utilization and nutrient balances among dairy production systems.
More sophisticated models have combined N loss with other production components to predict both environmental and economic implications of management changes. Schmit and Knoblauch (1995) used a linear programming approach to determine the economically optimal dairy herd intensities, manure application rates, and crop mix for unrestricted and restricted scenarios of N loss on New York dairy farms. They found that optimal cow numbers per hectare decreased by nearly 35% with a restriction on N loss. Stonehouse et al. (2002) used a mixed integer programming approach to generate optimal whole-farm plans for specialized swine finishing enterprises in Ontario. They found tradeoffs between economic and environmental goals, and environmental goals could only be reached at some loss in farm net return.
Computer simulation provides a powerful tool for evaluating the long-term impacts and interactions of management changes in animal production. McGechan and Wu (1998) developed a weather-driven simulation to compare slurry management options in dairy production in the United Kingdom. Environmental benefits were shown for long-term manure storage and injection, but neither practice could be economically justified. Kuipers et al. (1999) showed a similar result for The Netherlands, but manure storage and injection have become obligatory in that country.
A comprehensive farm simulation model was developed by Rotz et al. (1999b) to evaluate and compare the economic and environmental impacts of alternative dairy production systems. They demonstrated that including a low-RDP feed in ration formulation for a typical dairy herd reduced volatile N loss from the farm by 13 to 34 kg/ha of cropland with a small (1 kg/ha) reduction in leaching loss, and a $46 to $69/cow improvement in annual farm net return. Environmental and economic benefits for efficient protein feeding were generally greater with more animals per unit of land, higher milk production, more sandy soil, or daily manure hauling. The model was also used to evaluate and compare the long-term environmental and economic effects of various alfalfa, corn, soybean, and small grain cropping strategies on dairy farms (Borton et al., 1997; Rotz et al., 2001; Rotz et al., 2002). This model has been expanded to form the Integrated Farm System Model, which includes crop, beef, and dairy production options (Rotz, 2003).
Modeling has also been carried beyond the farm level. De Vries et al. (2001) modeled all of agriculture in The Netherlands to study the impacts of both structural measures and improved farming practices on major N fluxes, including ammonia and nitrous oxide emissions, leaching, and runoff. Improved farming practices were able to provide significant reductions in N loss, but these improvements were not enough to reach all targets set for those fluxes. The current ammonia emission target suggested for the year 2030 could not be met without eliminating all poultry and swine farming from The Netherlands and restricting all cattle to low-emission stables.