Livestock Ammonia Management and Particulate-Related Health

Livestock Ammonia Management and Particulate-Related Health Benefits ..... All of these measures trap ammonia to some degree, preventing its release t...
0 downloads 0 Views 60KB Size
Policy Analysis Livestock Ammonia Management and Particulate-Related Health Benefits D O N A L D R . M C C U B B I N , * ,† BENJAMIN J. APELBERG,† STEPHEN ROE,‡ AND FRANK DIVITA JR.‡ Abt Associates Inc., 4800 Montgomery Lane, Suite 600, Bethesda, Maryland 20814, and E. H. Pechan and Associates, 5528 Hempstead Way, Springfield, Virginia 22151

Agricultural operations are the largest source of ammonia emissions in the United States and contribute to the formation of ammonium nitrate and ammonium sulfate, two prevalent forms of fine particulate matter. Researchers have found an association between fine particulate matter and a variety of adverse healths effects, including premature mortality, chronic bronchitis, hospital admissions, and asthma attacks. Management practices that reduce ammonia emissions may decrease adverse health effects, resulting in significant economic benefits. We estimated the impact of a variety of emission controls, including diet optimization, alum, and incorporation of manure into the land. The results suggest that relatively modest management policies can have a significant impact on fine particulate formation in the atmosphere. Because of the heterogeneous nature of particulate matter, a key question is the importance of particulate matter size and composition. To the extent that ammonium nitrate and ammonium sulfate contribute to adverse health effects, ammonia management may have significant health implications. Our results suggest that a 10% reduction in livestock ammonia emissions can lead to over $4 billion annually in particulate-related health benefits.

Introduction It is well-known that animal wastes contribute to eutrophication and nitrate leaching, prompting the U.S. Environmental Protection Agency (U.S. EPA) to recently propose guidelines for feedlot operations (1). Livestock wastes can also cause significant odor problems, impair visibility, and contribute to lung disease in livestock workers (2). Less wellknown is that ammonia from livestock waste may contribute to significant health problems through secondary particulate formation. To our knowledge, cost-benefit analyses of livestock management controls have not considered this possibility. Our work suggests that the impacts may be substantial and should be considered in future analyses. Epidemiological work suggests that particulate matter causes a range of adverse effects. Recent work sponsored by the Health Effects Institute suggests that particulate matter leads to increased hospital admissions and premature * Corresponding author phone: (301)913-0521; fax: (301)652-7530; e-mail: [email protected]. † Abt Associates Inc. ‡ E. H. Pechan and Associates. 10.1021/es010705g CCC: $22.00 Published on Web 01/30/2002

 2002 American Chemical Society

mortality (3, 4). Although some studies question the role of particulate matter (5), a large body of evidence suggests that particulate matter is causally related to adverse health effects (6). Studies have shown significant associations between particulate matter and adverse health effects even after controlling for multiple pollutants and other potential confounders (3, 7). The question then focuses on the sizes and types of particulate matter that are harmful. Recently, attention has focused on fine, combustion particles. To better understand the potential biological mechanism for the adverse health effects of particulate matter, attention has focused on ultrafine particles and on the constituents of particulate matter (8-10). It appears that combustion sources are more toxic than soil-based particulate matter. However, the lack of widespread air quality monitoring data for fine particles and its constituents has hampered epidemiological research on this question. Livestock operations dominate ammonia emissions, with most ammonia coming from urea excreted from cattle and hogs and uric acid from poultry. In 1996, over 70% of anthropogenic ammonia emissions resulted from animal operations, with nitrogen fertilizer and motor vehicles contributing the bulk of the balance. Ammonia can react with gaseous nitric acid and particulate sulfate to form ammonium nitrate and ammonium sulfate. In some areas of the United States, ammonia is a limiting factor in the formation of both ammonium sulfate and ammonium nitrate, so a reduction in ammonia emissions can lead to reductions in these two components of PM2.5. Because of farms’ large contribution to ammonia emissions, changes in farm management practices may affect the formation of PM2.5 and in turn adversely affect people’s health.

Methods To test the impact of ammonia reductions, we developed a model for the United States based on the S-R matrix air quality model (11). We used the U.S. EPA’s national emission inventory for 1996 as a baseline (12). The control scenarios included percentage reductions in ammonia emissions from different livestock sectors and more detailed scenarios that examined specific ammonia control technologies. We used the S-R matrix to model the change in annual PM2.5 levels between the 1996 baseline and a given control scenario. We then estimated the impact on premature mortality of this change using recent epidemiological work by the Health Effects Institute (3). Finally, given a change in premature mortality, we estimated the value of this change using a value for a statistical life based on 26 studies (13). Baseline 1996 Emissions. The 1996 national emission inventory is a comprehensive national emissions inventory that the U.S. EPA (12) developed and has used in a variety of regulatory analyses (14). Livestock ammonia emissions dominate this inventory, with cattle contributing over 50% of emissions and hogs and poultry each contributing about 13%. To estimate livestock emissions, the U.S. EPA multiplied the number of cattle, hogs, poultry, and other livestock in each U.S. county by animal-specific ammonia emission factors developed by Battye et al. (15). Table 1 lists the emission factors for cattle, hogs, and poultry and the fraction of each emission factor that we attributed to different aspects of livestock production. Battye et al. (15) used emission factors generated for particular animal types in a report from The Netherlands (16). This included categories such as young cattle, dairy VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1141

TABLE 1. Agricultural Livestock Emission Factors composite emission % of composite emission factor factor (kg of category NH3/head-year)a housingb storage spreading grazing cattle hogs poultry

22.9 9.21 0.179

21 34 40

9 14 0

51 52 60

19 0 0

a Composite emission factors are from Battye et al. (15). Percentages attributed to each subcategory are based on breakdowns from Asman et al. (16). Housing and storage was originally aggregated into one emission factor but was divided for use in this analysis. b Note that for cattle and hogs, we assumed a 70/30 split between housing and storage; for poultry, we assume that housing emissions dominate storage emissions because we only examine impacts for broilers, for which the housing litter also acts as storage for their waste.

cattle, fattening cattle, breeding bulls, etc. Battye et al. aligned the emission factors for each of these animal types as closely as possible to 1992 U.S. livestock populations (1991 populations for hogs and poultry) (17). After the relevant emission factor to each subset of the livestock population was applied, Battye et al. obtained a composite emission factor. In estimating the relative contribution of grazing, housing/ storage, and land application to the emission factor, we applied the subcategory-specific relative contributions presented by Asman (16) to the estimates from Battye et al. (15). An implicit assumption in this derivation is that the proportion of unconfined animals in each subcategory is the same in the United States and The Netherlands, which may be unlikely due to the greater amount of land available in the United States. Recent work by the U.S. EPA and others at farms in the United States will lead to revised emission factors. The distinction between emissions from confined and unconfined operations is most important for the cattle sector, where there is about a 5-fold difference between emissions from confined and unconfined cattle operations. Using a single emission factor results in an overestimate of emissions in areas dominated by grazing and an underestimate of emissions in areas dominated by confined operations. This is less of an issue for hogs and poultry, which typically live in confined operations. Control Options. For each livestock type, we developed control options that focused on ammonia volatilization controls in different aspects of the livestock production processsdiet, stabling of animals, storage of waste, and land application of waste. We focused exclusively on ammonia and assumed that there would be no change in primary particulate matter generated at farms. For each control option, we estimated a lower and upper bound for a technology’s potential to reduce emissions. This is the “control effectiveness”. Then we estimated what fraction of operators use the technology in the baseline scenario and the percentage that would use it in the control scenario. These are the “compliance rates”. Multiplying the control effectiveness and the difference in compliance rates, we obtained the percentage reduction in emissions. To estimate control effectiveness percentage, we took into account that the estimates reported in the literature are typically from controlled settings that may not be achieved on actual farms. In estimating compliance rates, we recognize that a certain percentage of farmers may currently be using a measure, at least to some extent, such as alum in broiler houses or optimization of feed nutrients. Control measures may not be in widespread use or strictly followed because farmers do not have the necessary information or because the measures do not improve profitability. When possible, we estimate this baseline compliance rate, and in other cases, we use a default compliance rate of zero. For the control scenario, we adopt a low estimate that reflects a voluntary, information-based approach with per1142

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

haps modest economic incentives and a high estimate that reflects a stronger regulatory approach with larger costs for noncompliance and incentives for compliance. We assume that small to moderate farms will be less likely to comply than larger farms because of differences in information availability, costs, and enforcement. For most of the control measures in this analysis, we assume that compliance ranges from 25 to 50% for smaller farms and from 50 to 75% for larger farms. We assume no change in ammonia emissions from unconfined animals and focus the analysis on confined livestock operations. Confined operations have more opportunities to manage waste as compared to grazing animals, and the ammonia emissions per animal are greater in confined operations. The emissions are greater because manure and especially the urine from grazing animals often bind relatively quickly with the soil and do not volatilize as much as in confined operations. On the basis of personal communication with USDA staff (18), we estimated that 27% of cattle, 98% of hogs, and essentially all broilers and layers are in confined operations. For cattle, the emission factor is made up of three components: grazing, housing/storage, and spreading. Since control measures typically do not affect both housing and storage, we made an arbitrary 50% split between housing and storage to estimate the impact of a given measure. Similarly, we split the housing and storage components of hog production, assuming that one-half of the emissions come from housing and the other half from storage. For broilers, we assume that emissions come only from the housing since the waste stays in the poultry litter for the life of the bird. Ideally, we would have developed a complete farm model that included an estimate of the change in nitrogen utilization on farms under different control options and tracked nitrogen inputs and losses through ammonia volatilization, nitrate (NO3-) leaching, and nitrous oxide (N2O) emissions. A change in diet can reduce the amount of available nitrogen in subsequent stages and the potential for ammonia volatilization. On the other hand, while housing, storage, or land application control options decrease ammonia volatilization, they do not decrease the level of nitrogen in the system, so there may be an increase in nitrogen losses elsewhere in the system. A complete system approach, such as that by Rotz et al. (19), is beyond the scope of this analysis. We assume that diet optimization reduces ammonia emissions across the board so that a given percent reduction in excreted nitrogen results in the same percent reduction in overall ammonia emissions from confined animals. However, we treat housing, storage, and land application individually and assume no interacting effects. In addition to estimating percentage ammonia reductions from specific control options, we estimated the impact of 10, 20, and 40% reductions in farm-wide ammonia emissions. These span the realm of what is discussed for European regulatory goals and what researchers suggest is possible at a reasonable cost. Examining ammonia reductions from farms in Europe, Cowell and Apsimon (20) reported that relatively inexpensive measures could reduce overall emissions from about 10 to 30%, depending on the country and typical livestock management practices. In similar work, Bussink and Oenema (21) noted that up to 3-fold reductions in ammonia losses are possible at dairy farms in Europe, although this would come at a high cost. Gustavsson (22) reported that Sweden’s goal is to reduce ammonia emissions by 25-50% in parts of Sweden. In 1990, the Dutch government developed a goal of reducing emissions by 38% in 1996, 50% in 2000, and 70% in 2005 relative to 1980 levels. Erisman and Monteny (23) reported that the regulations, including slurry storage covers

and incorporation of waste into soil, reduced emissions in 1996 by about 10-20% relative to 1980 levelssabout half of the goal. Diet Optimization. Most of the nitrogen in feces is organically bound and tends to breakdown slowly, so the ammonia volatilization potential is low (21). Most ammonia volatilization in cattle and hogs comes from urine, particularly urea; in poultry, uric acid is the primary source. To model diet optimization, we focus on modifying feed protein levels. A variety of studies on cattle, hogs, and poultry suggest that a reduction in the protein content of feed and the addition of essential amino acids can significantly reduce ammonia volatilization. To estimate the control effectiveness of diet optimization, we develop low and high percentage reductions in excreted nitrogen for each livestock type. We assume that a given percentage reduction in excreted nitrogen affects all subsequent livestock production phases and reduces, by the same percentage, ammonia emissions from housing, storage, and land application. On the basis of a review of the literature, we assume a control effectiveness for cattle of 15-30% in our low and high estimates (24, 25). For hogs, we assume a rate of 1030% (26, 27), and for poultry we use a rate of 10-25% (28, 29). Housing. A wide range of animal housing practices can affect ammonia emissions, including the use of chemical additives, add-on controls to ventilation systems, and manure management practices, such as the quick removal of waste, scraping and flushing systems, and temperature control. In addition, changes in the construction of houses can affect ammonia emissions. For simplicity, we limit ourselves to estimating the impact of two chemical additives: urease inhibitors for cattle and hogs and alum for broilers. Urease inhibitors prevent the urease enzyme, produced from bacteria in the feces, from hydrolyzing urea to ammonia, which could then volatilize (30). However, an important question regarding the control effectiveness of urease inhibitors is the willingness of farm operators to apply them in an accurate and timely fashion. Varel et al. (31) reported that two types of urease inhibitors significantly reduced ammonia losses in feedlot experiments with cattle and hogs, and a manufacturer reported that the control effectiveness at cattle feedlots is 50%, and it should be similar at dairies (32). In our low estimate of control effectiveness, we assume a rate of 25%, taking into account that growers may not apply it effectively. In our high estimate, we assume a 50% rate. Alum, or aluminum sulfate (Al2(SO4)3‚18H2O), reduces ammonia emissions by lowering the pH of the litter. Acidforming compounds, such as alum, work by maintaining a low pH in the poultry litter and in turn reducing the conversion of ammonium to ammonia. The control effectiveness of alum can exceed 95% during the early part of the growing cycle when the young chickens are most susceptible to health problems from high ammonia levels. In personal communication, Moore (33) reported that the overall control effectiveness for alum is about 70%, for a treatment of about 250 lb/1000 ft2, with the effect of the treatment linearly related to the treatment rate. Moore estimated that under 10% of broilers are grown with alum and that most growers apply only 100 lb/1000 ft2. This relatively low treatment level controls ammonia concentrations for the first 10-14 days, the time when chicks are most sensitive. For our lower estimate of alum’s control effectiveness, we assume that growers use 100 lb/ft2, and this reduces ammonia emissions by 25%. In our high estimate, we assume an application rate of 250 lb/ft2 and a reduction of 70% in ammonia emissions. We only assume alum use in broiler production, as most of the information we have focuses on broilers.

In estimating compliance, we assume a rate of 5% in the baseline and a range of 50-75% in the control scenario. Compliance for alum may exceed rates achieved for other control measures because alum provides a number of benefits to farmers. Alum reduces the need for ventilation to maintain healthy growing conditions and working conditions. In addition, as it reduces ammonia volatilization, alum increases the nitrogen content of poultry waste and reduces the availability of phosphorus. The latter is a benefit or not depending on whether the local soils have excess phosphorus. Storage. Farmers frequently flush out animal waste in barns and housing operations into a lagoon or tank. In examining the impact of storage-related control measures, we focus on slurry tank and lagoon covers. The types of covers range from a tight tarpaulin covering or light construction roof to materials spread on the surface, such as plastic pellets or straw, or the natural formation of a crust on stored cattle waste (34, 35). All of these measures trap ammonia to some degree, preventing its release to the atmosphere and, in the process, producing a slurry richer in nutrients. On the basis of several estimates in the literature (20, 36, 37), we assumed a control effectiveness of 40-80%. Land Application. Land application of manure as fertilizer is an effective means to recycle animal waste; however, it can contribute to significant ammonia volatilization. Direct ground injection and tillage after surface application are two commonly examined measures to reduce volatilization and can significantly decrease ammonia losses from 50 to 90%, depending on soil conditions, climate, and application practice (38-45). To estimate the control effectiveness, we use a range of 20-60%, which reflects the published research on incorporating waste into the soil as well as the possibility that, in practice, farmers may have less success than the researchers in part because of delays in applying waste to fields. We do not include band-spreading, trailing-shoe, or other variations of surface application in our estimate. However, we recognize that these surface methods may be less expensive to adopt than incorporation. To estimate the compliance rate, we use a range of 2550% for operators of all sizes, recognizing that a number of factors may reduce the percentage of farmers that incorporate waste into the soil. Incorporation increases the nutrient availability of waste, and this tends to reduce the application rate and increase the acreage needed to dispose of the waste. This is particularly a problem in areas that already have a shortage of suitable land or have land with nitrogen levels that already exceed crop needs (23). The costs associated with incorporation of waste into soil may exceed the cost of purchasing chemical fertilizers, so farmers may have less incentive to incorporate the waste (40). Air Quality Modeling. We estimate particulate matter levels using the Phase II source-receptor (S-R) matrix, which E. H. Pechan & Associates (11) developed with the Climatological Regional Dispersion Model (CRDM) and then used to estimate the impact of new ambient air quality standards (46). For inputs, the S-R matrix uses emission sources through out the 48 states and borders areas and produces county-level annual average concentrations of various PMrelated species, including primary PM10 and PM2.5, nitrate, sulfate, and ammonia. The CRDM estimates county-level annual-average PM and precursor concentrations from county-level source areas across the United States, by parametrizing the processes of dispersion, chemistry, and deposition. It includes wet and dry deposition of gases and PM as well as linear chemical oxidation of SO2 to sulfate and NOx to nitrate. As in SLIM3 and ISC2LT, the model is based upon a sector-average approach for transport and dispersion (47) that is recommended for calculating long-term average concentrations. Using results from CRDM, E. H. Pechan & Associates Inc. VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1143

(48) developed four matrixes of S-R transfer coefficients that link emissions from every county and major elevated point source in the United States and border areas to air quality within every U.S. county. Each coefficient in a matrix represents the incremental ambient air quality impact of a given species at a given receptor from a particular area or point source. To estimate the change in PM associated with a change in ammonia emissions, we convert the transported NOx, SO2, and NH3 emissions to concentrations of ammonium sulfate and ammonium nitrate at the receptor. We then model the formation of secondary particles. We assume that ammonium reacts first with particulate sulfate to form ammonium sulfate or ammonium bisulfate, and any remaining ammonium may then react with gaseous nitrate to form ammonium nitrate. Concentration-Response Function. A number of researchers have found evidence suggesting that particulate matter causes a wide range of adverse health effects such as premature mortality, chronic bronchitis, hospital admission, and asthma attacks (49). To simplify the analysis and to focus on the single most important adverse health effect, we estimate the possible impact of ammonia reductions on premature mortality. To estimate mortality, we use a concentration-response function based on recent work by Krewski et al. (3):

TABLE 2. Estimated Air Quality and Mortality Changes and Monetary Value for Confined Cattle Operations control scenario diet optimization urease inhibition storage management land incorporation emissions reduction

where y0 is the county-level annual all-cause death rate per person ages 30 and older; β is the PM2.5 coefficient ) 0.0046257; ∆PM2.5 is the change in annual mean PM2.5 concentration; and pop is the population of ages 30 and older. To estimate county-specific baseline mortality incidence among individuals ages 30 and over, this analysis used the average annual county mortality rate from 1994 through 1996 (50). The coefficient (β) is estimated by taking the natural log of the relative risk (1.12) and dividing by 24.5 µg/m3, the change in mean PM2.5 exposure associated with the relative risk (3). Valuing Premature Mortality. To value premature mortality, we used a damage function approach and multiplied the number of cases of premature mortality by an estimate of the value of a statistical life. This is an approach that is widely used in policy analyses (49). Perhaps the most controversial aspect is to estimate the value of a statistical life saved. To estimate the value of a statistical life, we start with an analysis by the U.S. EPA (49) that found that a Weibull distribution best described the distribution of 26 estimates of the value a statistical life. The mean of this Weibull distribution was $4.8 million (1990 dollars), or about $6.1 million, when updated to 1999 dollars. In addition, we assume that PM-related premature mortality occurs over a 5-yr period following exposure and then use a 5% discount rate to estimate the present economic value of the deaths that occur over this 5-yr period. On average, this results in an estimate of $5.6 million (1999 dollars) per statistical life.

mortality reduction

monetary value (millions 1999 $)

4.7 15.6 2.1 6.8 1.4 4.7 2.5 15.3 10 20 40

232 790 100 336 69 229 123 772 498 1 024 2 168

1 298 4 423 563 1 880 385 1 280 690 4 323 2 787 5 732 12 140

TABLE 3. Estimated Air Quality and Mortality Changes and Monetary Value for Confined Hog Operations control scenario diet optimization urease inhibition

∆mortality ) -[y0(e-β∆PM2.5 - 1)]pop

low high low high low high low high

emission reduction (%)

storage management land incorporation

low high low high low high low high

emission reduction

emission reduction (%)

mortality reduction

monetary value (millions 1999 $)

4.3 20.3 3.6 11.4 2.5 7.8 2.5 15.2 10 20 40

48 236 41 131 28 89 28 175 114 232 488

271 1 321 228 733 155 498 157 982 639 1 299 2 732

TABLE 4. Estimated Air Quality and Mortality Changes and Monetary Value for Poultry Operations emission monetary reduction mortality value (millions (%) reduction 1999 $)

control scenario diet optimization alum addition land incorporation emission reduction

low high low high low high

4.7 17.5 3.0 13.1 2.8 17.0 10 20 40

49 189 31 140 30 183 106 216 447

275 1 059 176 785 165 1 023 594 1 212 2 505

Results

has a significant effect. A 2% reduction in cattle ammonia emissions generates a half billion dollars in benefits. A 20% reduction in cattle emissions generates over 5 billion dollars in benefits. For a given percentage reduction, the impacts of hogs and poultry are roughly equivalent to each other and represent about 20% of the impact of cattle. Most of the technologies we examined reduced hog and poultry emissions by less than 10% with reductions from diet optimization and land application substantially higher.

Tables 2-4 present the impacts associated with nationwide ammonia reductions at cattle, hog, and poultry operations. Diet optimization and land application have the largest impact in each livestock sector. However, the percentage reduction in emissions varied substantially between cattle and the other two sectors because of the large fraction of unconfined cattle. For cattle, each of the technologies that we examined reduced ammonia emissions less than 20%, and most reduced emissions less than 10%. However, even a modest reduction

There is a large difference between the low and the high estimates in the percentage reduction in emissions because of the uncertainty in the underlying assumptions. In addition, the results are somewhat nonlinear with larger reductions generating more than a proportionate increase in benefits. This is due to the fact that an initial reduction in ambient ammonia may not reduce secondary particulate formation in areas with excess ammonia. If the available sulfate and nitric acid have already reacted with ammonia to form ammonium sulfate and ammonium nitrate and there is still

1144

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

TABLE 5. Urease Inhibitors and Alum: Comparison of Estimated Management Costs and Monetary Value for Mortality Reduction management practice

baseline emissions (ton)

alum: poultry

310 000

emission reduction (%) low high

2.9 12.6

emission reduction (ton)

cost ($ per ton)

cost ($ million)

monetary value of mortalityreduction ($ million)

ratio: monetary value to cost

8 990 39 060

6 100 5 400

55 211

170 754

3.1 3.6

excess ammonia, then a reduction in ammonia will not affect particulate matter levels. Comparison of Management Costs and Monetary Value of Mortality Reduction for Alum. This analysis focused on the benefits of ammonia reduction, and did not systematically explore the costs of implementation. However, taking advantage of easily obtainable data, we considered the costeffectiveness of alum by comparing its implementation costs and the estimated monetary value of the associated mortality reduction. To estimate the costs of alum we used data from Moore (33), who estimated that the cost of alum is $200300/ton. On the basis of a cost of $250/ton and an application rate of 2 ton/flock of 20 000 birds to achieve 70% control effectiveness, the cost of treatment is $0.025/head. With six grow-outs per year, the cost of reducing ammonia emissions works out to $5400/ton of NH3 reduced. Similar calculations for our low estimate yield a cost of about $6100/ton of HN3 reduced. Table 5 shows that the benefit cost ratio ranges from 3.1 to 3.6, suggesting that alum would be cost-effective. However, we caution that our cost-effectiveness analysis is limited in a number of respects. In estimating benefits, we estimate only the reduction in premature mortality and ignore other air quality health benefits as well as benefits related to water quality and visibility improvement. On the cost side, we took a fairly simplistic approach by focusing just on the cost of the alum and do not attempt to estimate potential negative impacts of alum, such as the possibility of higher aluminum levels in the soil and water in areas with highly acid soils. Nevertheless, we consider the analysis illustrative of some of the important categories to include in a more complete analysis of social costs and benefits.

Discussion Our work is best viewed as a scoping analysis to identify the potential health impacts of changes in livestock ammonia emissions. We used readily available information to examine a previously ignored facet of livestock management, and in the process we identified a number of areas that future analyses could develop more fully. Perhaps the biggest question revolves around the relative toxicity of different PM constituents. Assuming that different constituents of fine particulate matter are equally harmful, we find that controlling ammonia emissions from livestock operations results in significant economic benefits. In estimating the health impact, we used the latest epidemiological research linking fine particulate matter (PM2.5) to premature mortality (3). However, as we noted, it is unclear whether all of the constituents of PM2.5 are equally harmful. It could be that specific portions of PM2.5, such as ammonium sulfate or, say, combustion-related organic compounds, are most harmful (10). The current ambient air quality standard only distinguishes particulate matter by size and not by composition. Future work may identify particular constituents of concern that could affect the benefits estimates that we presented. Another area of concern is the ammonia emission inventory. To develop better estimates of the impact on particulate formation, future work could develop more precise, less-aggregated estimates. Indeed, ongoing work in the United States is in the process of developing more refined emission factors. Currently the U.S. EPA emission inventory

uses national composite emission factors developed for the most part in Europe in the early 1990s. Using a single national emission factor for each livestock type can lead to significant biases in estimated emissions. Confined animals have significantly higher emissions than unconfined animals, and different areas of the United States have different proportions of these two livestock types. Necessarily, we have simplified the extremely complicated process of nutrient cycling on farms. We considered a relatively limited number of control options individually, in a partial equilibrium approach, basing our estimates of the control effectiveness for each individual control on the range of values observed in the literature. This ignores the fact that reducing ammonia volatilization at one stage will likely create a more nutrient-rich waste, leading to a greater availability for ammonia loss at a later stage, such as land application. We also ignore that control measures taken individually tend to minimize the estimated change in ambient particulate matter. In areas with excess ammonia, an initial ammonia reduction has little or no impact on particulate matter levels, so practices taken in concert generally have a greater impact than the sum of individual practices. We might also add that as policies to control NOx and SO2 take effect over the coming years, this will likely increase the amount of excess ammonia and decrease the impact of ammonia reductions on PM formation. The scenarios that we considered clearly have impacts on other environmental concerns, such as nitrate leaching, phosphorus runoff, and nitrous oxide formation. For example, the higher nutrient level of alum-treated poultry waste may increase nitrate leaching (51, 52) or nitrous oxide emissions (53, 54) without careful monitoring of crop nutrient needs perhaps in combination with nitrification inhibitors (55). On the other hand, alum may reduce phosphorus runoff. Moore and Miller (56) reported that alum-treated poultry waste reduces phosphorus solubility, making poultry waste less susceptible to phosphorus leaching and runoff. This suggests the need for a complete farm model, such as the DAFOSYM model by Rotz et al. (19), to track nutrient inputs and losses throughout the stages of livestock production. Finally, we note that to estimate the impact on ambient particulate matter, we used a relatively simple air quality model. For a scoping analysis, our approach is perhaps sufficient, but more detailed estimates would require a model better suited to examining secondary particulate formation.

Acknowledgments Portions of this work were completed as part of a contract with the U.S. Environmental Protection Agency, Office of Policy, Economics and Innovation, National Center for Environmental Economics. However, all views expressed in this paper are those of the authors and do not necessarily reflect those of any named institutions. The authors thank Drs. Brian Heninger and Cynthia Morgan of the National Center for Environmental Economics for their support.

Supporting Information Available Appendix A discusses assumptions underlying the estimation of 1996 baseline ammonia emissions. Appendix B reviews the literature for the control options used in this analysis VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1145

and discusses their effectiveness in reducing ammonia emissions. Appendix C briefly reviews the development of the S-R matrix and then presents maps of the impacts of different control scenarios on PM2.5 formation. Appendix D presents the dollar value of reductions in cattle, hogs, and poultry ammonia ranging from 5% to 20% as well as a sensitivity analysis where we consider the impact of varying the assumed split between emissions from housing and storage. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) U.S. EPA. Code of Federal Regulations, 40 CFR 122 and 412; Fed. Regist. 2001, 66, 2960. (2) Kirkhorn, S. R.; Garry, V. F. Environ. Health Perspect. 2000, 108, 705. (3) Krewski, D.; Burnett, R.; Goldberg, M.; Hoover, K.; Siemiatycki, J.; Jerrett, M.; Abrahamowicz, M.; White, M. Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of Particulate Air Pollution and Mortality; Health Effects Institute: Cambridge, MA, 2000. (4) Samet, J. M.; Dominici, F.; Curriero, F. C.; Coursac, I.; Zeger, S. L. N. Engl. J. Med. 2000, 343, 1742. (5) Moolgavkar, S. H. J. Air Waste. Manage. Assoc. 2000, 50, 1199. (6) Bates, D. V. Environ. Health Perspect. 2000, 108, 91. (7) Samet, J.; Zeger, S.; Dominici, F.; Curriero, F.; Coursac, I.; Dockery, D.; Schwartz, J.; Zanobetti, A. The National Morbidity, Mortality, and Air Pollution Study; Health Effects Institute: Cambridge, MA, 2000. (8) Wichmann, H. E.; Spix, C.; Tuch, T.; Wolke, G.; Peters, A.; Heinrich, J.; Kreyling, W. G.; Heyder, J. Daily Mortality and Fine and Ultrafine Particles in Erfurt, Germany. Part I: Role of Particle Number and Particle Mass; Health Effects Institute: Cambridge, MA, 2000. (9) Schwartz, J.; Neas, L. M. Epidemiology 2000, 11, 6. (10) Laden, F.; Neas, L. M.; Dockery, D. W.; Schwartz, J. Environ. Health Perspect. 2000, 108, 941. (11) E. H. Pechan & Associates Inc. Control Measure Analysis of Ozone and PM AlternativessMethodologies and Results, Final Technical Support Document; Prepared for U.S. EPA, Office of Air Quality Planning and Standards, Innovative Strategies and Economics Group: Research Triangle Park, NC, 1997. (12) U.S. EPA. National Air Pollutant Emission Trends, Procedures Document, 1990-1996; Office of Air Quality Planning and Standards: Research Triangle Park, NC, 1998. (13) U.S. EPA. The Benefits and Costs of the Clean Air Act: 1990 to 2010: EPA Report to Congress; U.S. EPA, Office of Air and Radiation, Office of Policy: Washington, DC, 1999. (14) U.S. EPA. Technical Support Document for the Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements- Air Quality Modeling Analyses; U.S. EPA, Office of Air Quality Planning and Standards: Research Triangle Park, NC, 2000. (15) Battye, R.; Battye, W.; Overcash, C.; Fudge, S. Development and Selection of Ammonia Emission Factors: Final Report; Prepared by EC/R Incorporated, Durham, NC; Prepared for U.S. Environmental Protection Agency, Office of Research and Development: Washington, DC, 1994. (16) Asman, W. Ammonia Emissions in Europe: Updated Emissions and Emission Variations; National Institute of Public Health and Environmental Protection: Bilthoven, The Netherlands., 1992. (17) U.S. Department of Agriculture. Agricultural Statistics 1992; U.S. Government Printing Office: Washington, DC, 1992. (18) Miller, J. USDA, personal communication with S. Roe, 2000. (19) Rotz, C. A.; Satter, L. D.; Mertens, D. R.; Muck, R. E. J. Dairy Sci. 1999, 82, 2841. (20) Cowell, D.; Apsimon, H. Atmos. Environ. 1998, 32, 573. (21) Bussink, D. W.; Oenema, O. Nutr. Cycling Agroecosyst. 1998, 15, 19.

1146

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

(22) Gustavsson, J. Nutr. Cycling Agroecosyst. 1998, 51, 81. (23) Erisman, J.; Monteny, G. Environ. Pollut. 1998, 102, 275. (24) Paul, J.; Dinn, N.; Kannangara, T.; Fisher, L. J. Environ. Qual. 1998, 27, 528. (25) Smits, M.; Valk, H.; Monteny, G.; van Vuuren, A. In Gaseous Nitrogen Emissions from Grasslands; Jarvis, S. C., Pain, B. F., Eds.; Cab International: Oxon, U.K., 1997; p 101. (26) Sutton, A.; Kephart, K.; Verstegen, M.; Canh, T.; Hobbs, P. J. Anim. Sci. 1999, 77, 430. (27) Hobbs, P. J.; Pain, B. F.; Kay, R. M.; Lee, P. A. J. Sci. Food Agric. 1996, 71, 508. (28) Elwinger, K.; Svensson, L. J. Agric. Eng. Res. 1996, 64, 197. (29) Summers, J. D. Poult. Sci. 1993, 72, 1473. (30) Svennerstedt, B. J. Agric. Eng. Res. 1999, 72, 19. (31) Varel, V.; Nienaber, J.; Freetly, H. J. Anim. Sci. 1999, 77, 1162. (32) Axe, D. IMC Agrico Feed Ingredients, personal communication with S. Roe, 1999. (33) Moore, P. University of Arkansas, Fayetteville, personal communication via e-mail with Don McCubbin, January 11, 2001. (34) Mannebeck, H. In Odor Prevention and Control of Organic Sludge and Livestock Farming; Nielsen, V. C., Voorburg, J. H., L’Hermite, P., Eds.; Elsevier Applied Science Publishers: New York, 1985; p 188. (35) Minnesota Pollution Control Agency. Feedlot Air Qualtiy Summary: Data Collection, Enforcement, and Program Development; Minnesota Pollution Control Agency: Minneapolis, MN, 1999. (36) Hornig, G.; Turk, M.; Wanka, U. J. Agric. Eng. Res. 1999, 73, 151. (37) Xue, S. K.; Chen, S.; Hermanson, R. E. Trans. ASAE 1999, 42, 1095. (38) Frost, J. P. Grass Forage Sci. 1994, 49, 391. (39) Kirchman, H.; Lundvall, A. Nutr. Cycling Agroecosyst. 1998, 51, 65. (40) Malgeryd, J. Nutr. Cycling Agroecosyst. 1998, 51, 51. (41) Mattila, P. Nutr. Cycling Agroecosyst. 1998, 51, 47. (42) Morken, J.; Sakshaug, S. Nutr. Cycling Agroecosyst. 1998, 51, 59. (43) Smith, K. A.; Jackson, D. R.; Misselbrook, T. H.; Pain, B. F.; Johnson, R. A. J. Agric. Eng. Res. 2000, 77, 277. (44) Sommer, S.; Ersbøll, A. J. Environ. Qual. 1994, 23, 493. (45) Weslien, P.; Klemedtsson, L.; Svensson, L.; Galle, B.; KasimirKlemedtsson, Å.; Gustafsson, A. Soil Use Manage. 1998, 14, 200. (46) U.S. EPA. Regulatory Impact Analyses for the Particulate Matter and Ozone National Ambient Air Quality Standards and Proposed Regional Haze Rule; U.S. EPA, Office of Air Quality Planning and Standards: Research Triangle Park, NC, 1997. (47) Turner, D. B. Workbook of Atmospheric Dispersion Estimates, AP-26; U.S. EPA, Office of Air Programs: Research Triangle Park, NC, 1979. (48) E. H. Pechan & Associates Inc. Regional Particulate Control Strategies Phase II; Prepared for the U.S. Environmental Protection Agency, Office of Policy, Planning, and Evaluation: Washington, DC, 1996. (49) U.S. EPA. The Clean Air Act Amendments (CAAA) Section 812 Prospective Study of Costs and Benefits 1999: Advisory by the Advisory Council on Clean Air Compliance Analysis: Costs and Benefits of the CAAA; Prepared by the Advisory Council on the Clean Air Compliance Analysis, Science Advisory Board, U.S. Environmental Protection Agency: Washington, DC, 1999. (50) U.S. Centers for Disease Control. CDC Wonder, 1999. (51) Shreve, B. R.; Moore, P. A.; Daniel, T. C.; Edwards, D. R.; Miller, D. M. J. Environ. Qual. 1995, 24, 106. (52) Kingery, W. L.; Wood, C. W.; Delaney, D. P.; Williams, J. C.; Mullins, G. L. J. Environ. Qual. 1994, 23, 139. (53) Bremner, J. M. Nutr. Cycling Agroecosyst. 1997, 49, 7. (54) Jarvis, S. C.; Pain, B. F. Clim. Change 1994, 27, 27. (55) Freney, J. R. Nutr. Cycling Agroecosyst. 1997, 49, 1. (56) Moore, P. A.; Miller, D. M. J. Environ. Qual. 1994, 23, 325.

Received for review March 5, 2001. Revised manuscript received November 29, 2001. Accepted November 30, 2001. ES010705G