Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

4.1 Adaptive Management

Adaptive management offers a way to address the pressing need to take steps to manage for factors affecting hypoxia in the NGOM in the face of uncertainties. The authors of a recent study undertaken by the National Research Council of the National Academy of Sciences identified six elements of adaptive management that are directly relevant to goal setting and research needs (National Research Council, 2004): (1) resources of concern are clearly defined; (2) conceptual models are developed during planning and assessment; (3) management questions are formulated as testable hypotheses to guide inquiry; (4) management actions are treated like experiments that test hypotheses to answer questions and provide future management guidance; (5) ongoing monitoring and evaluation is necessary to improve accuracy and completeness of knowledge; and (6) management actions are revised with new cycles of learning.

Perhaps the most important “take-home” lesson from their work is contained in the following statement:

Adaptive management does not postpone actions until “enough” is known about a managed ecosystem (Lee, 1999), but rather is designed to support action in the face of the limitations of scientific knowledge and the complexities and stochastic behavior of large ecosystems (Holling, 1978). Adaptive management aims to enhance scientific knowledge and thereby reduce uncertainties. Such uncertainties may stem from natural variability and stochastic behavior of ecosystems and the interpretation of incomplete data (Parma et al., 1998; Regan et al., 2002), as well as social and economic changes and events (e.g., demographic shifts, changes in prices and consumer demands) that affect natural resources systems.

Thus adaptive management provides an appropriate way for decision makers to deal with the uncertainties inherent in the environmental repercussions of prescribed actions and their influences on hypoxia.

Adaptive management can be conducted at the several management scales that occur in the NGOM and MARB. On the basin scale, adaptive management requires measurements of both nutrient loadings and hypoxia extent (area). Although it will not be possible to relate these changes to specific changes in the basin, these data will provide better understanding of the relationships between nutrients and hypoxia. On smaller scales, specific management actions can be treated as experiments that test hypotheses, answer questions, and thus provide future management guidance at that scale (for example, small watersheds).

The adaptive management approach requires that conceptual models are developed and used and that relevant data are collected and analyzed to improve understanding of the implications of alternative practices (e.g., Ogden et al., 2005). To help illustrate what is meant by a conceptual model, the Study Group has developed a diagram that shows major factors that affect hypoxia in the NGOM (Fig. 4.1). The corresponding conceptual model would estimate the relative contribution of each influence. Those estimates could serve as hypotheses of relative effects, and the diagram could illustrate hypothesized interactions and feedbacks. Such a conceptual model organizes how adaptive management research is conducted in a framework where the testing of hypotheses and the new knowledge gained is then used to drive management adaptations, new hypotheses, and the collection of new data on end points. Unlike the traditional model of hypothesis-driven research, adaptive management implies coordination with stakeholders and consideration of the economic and technological limitations on management. Unlike traditional demonstration projects, adaptive management implies an understanding that complex problems will require iterative solutions that will only be possible through generation of new knowledge as successive approximations to problem solving are attempted.

Fig. 4.1
figure 4_1_161148_1_Enfigure 4_1_161148_1_En

A conceptual framework for hypoxia in the northern Gulf of Mexico

Full size image

Successful implementation of the adaptive management process is occurring in the Grand Canyon (Meretsky et al., 2000) and the Everglades (Sklar et al., 2005). In addition, steps toward adaptive management are being examined in the Upper Mississippi River basin (O’Donnell and Galat, 2008). That work documents the need for greater collaboration between scientists and management agencies to plan, design, and monitor river enhancement programs. Problems exist in setting quantifiable success criteria, developing appropriate monitoring designs, and disseminating information. The Study Group expects similar difficulties in implementing adaptive management to occur in the MARB.

There needs to be a better understanding of the spatial and temporal aspects of basin-level responses to management practices and also a focus on other scales at which response can occur in a more timely fashion (Nassauer et al., 2007). Yet observations of a basin-level response to practices cannot be expected for some time, which calls for management and evaluation to be focused on a subbasin scale. Therefore it is important to obtain information at a scale where practices can be broadly and appropriately applied and where results are “meaningful and interpretable.” The relevant scale would likely be at smaller subwatershed scales, where local water quality and quantity benefits may become evident more quickly. Furthermore, the demonstration of adaptive management within a small sub-watershed may enhance practice adoption at other locations. Thus conceptual models need to be developed for this scale of resolution as well. Focus at the small watershed scale will also provide local water quality and quantity benefits. The results from small watershed studies must be able to be extrapolated to other small watersheds in the subbasin and, preferably, the entire MARB, if they are to be useful in reducing hypoxia in the NGOM.

Experiments that could be applied at small watersheds to help to improve understanding of the effects of different practices have the following characteristics:

  • Practices applied on the small watersheds should conform to accepted practice standards or make specific modifications of practices that can be implemented in new standards.

  • Monitoring should be at appropriate intensities (time and space) to determine effects of practices on water quality and quantity.

  • Monitoring should also measure co-benefits, including carbon sequestration, wildlife habitat, flood control, etc.

  • Practices should be applied in suites or systems, and components should be monitored to determine effects of component practices.

  • Changes in hydrology and crop productivity must be measured in addition to changes in water quality. Even at the small-scale, too many studies have focused just on nutrient concentrations in outflow water and neglected hydrologic or productivity changes.

  • All components of the cost of adopting and maintaining these practices should be measured and monitored. Such costs include direct equipment and structural costs, yield effects, changes in management time, changes in risk, and other costs.

  • These studies should be designed to improve our understanding at local, medium, and broad basin scale. Thus the experiments should be designed so that they can feed into conceptual models that operate at different scales.

  • Within practical limits, studies should be part of an adaptive management research strategy for the MARB to optimize the efficiency of research investments and to assure that results are coordinated, complimentary, and consistent.

Integrated modeling and monitoring play an important role in adaptive management. The cornerstone of adaptive management is the concept of learning about the impacts of actions and using that new understanding to guide future actions. Models can assist that learning by being used to evaluate impacts and uncertainties of proposed actions, such as targeted practices and locations or proposed policies, on both MARB and NGOM responses. In addition, monitoring must also be part of an adaptive management strategy in order to verify that the actions are addressing the stated goals or to test hypotheses. Monitoring is needed to improve the next generation of models and model assessments and to eventually verify that projected changes occur.

Adaptive management is also important to building infrastructure and to strategic planning and policy development of mechanisms of conservation practice implementation. For example, adaptive management can be used to evaluate if incentive-based programs are effective at bringing about changes in conservation practice acceptance and adoption at a local or small watershed level. At a basin level, other programs might be needed to facilitate adaptation of strategies and policies, and there must be constant feedback among all vested parties. As the scale of system increases (i.e., from a small watershed to the entire MARB), the complexity of adaptive management increases dramatically.

4.3 Setting Targets for Nitrogen and Phosphorus Reduction

To reduce hypoxia in the bottom waters of the NGOM, the Integrated Assessment set a target that N loading should be reduced by 30% in order to shrink the 5-year running average size of the hypoxic zone to below 5,000 km2 (1,930 mi2) by 2015. This reduction is significantly less than the three- to five-fold increase in N loading to the Gulf of Mexico due to human activity during the 20th century, and particularly in the past 30–50 years (Boyer and Howarth, 2008; Goolsby et al., 2001). Since the Integrated Assessment, a number of modeling efforts have provided a better depiction of how the area of hypoxia may respond to reduced N loading. The three available models were compared by Scavia et al. (2004), who concluded from these models that the 30% reduction in N is probably not sufficient to reach the goal of a hypoxia area of 5,000 km2 or less (Scavia et al., 2004). The consensus from these models is that N loads probably need to be reduced by 40–45% to reach the hypoxia reduction goal. In addition, a number of studies suggest that the consequences of climate change need to be considered, and this may require an N load reduction on the order of 50–60% to meet the original Integrated Assessment goal for hypoxic area (Donner and Scavia, 2007; Justić et al., 2003b). However, predicting the consequences of climate change on nutrient fluxes and hypoxia remains a very uncertain business (Howarth et al., 2006). The Study Group finds that the consensus of models reported by Scavia et al. (2004) and the new model of Scavia and Donnelly (2007), which uses the latest available load estimates from the USGS, supports a target of reducing the 5-year running average of N loadings by at least 45%. This target should be reassessed as more monitoring data are obtained, current models are refined, and new models are developed.

Only recently has new evidence emerged for the need to control P inputs as well as N in the NGOM. Work by Sylvan et al. (2006) has shown P to be the limiting nutrient during periods of maximum primary production in the near-shore NGOM high-productivity zone. Because previous attention has focused on N, there has been limited effort to model the effects of P on hypoxic area. Scavia and Donnelly (2007) used the previously developed and calibrated model (Scavia et al., 2004) to evaluate both the effects of new USGS load estimates and to assess the potential for P to control hypoxia dynamics under current and historical conditions. Confirming the results of Sylvan et al. (2006), Scavia and Donnelly found that P could have become limiting in some areas and times because of the relative increase in N loads during the 1970 s and 1980s. While they concluded that P did frequently control hypoxia in near-field zone of NGOM, they noted that a P-only strategy would likely reduce production in the near-field but possibly increase production in downfield N-controlled areas of NGOM. Their work, using the new USGS load estimates, reinforced the need for a dual nutrient strategy combining a 45% reduction in N with a 40–50% reduction in the 5-year running average of P loading. While the far-field effects could possibly be reduced through an N-only strategy, they suggested that a prudent approach would be to reduce both N and P, simultaneously. They also noted that an N-and P-reduction strategy would not only reduce hypoxia in the NGOM but would also help to remove P-induced Clean Water Act impairments in the MARB. Based on this recent modeling work, the Study Group finds that a comparable P reduction is needed, again based on 5-year running average fluxes. As with the N target, this P target should be reassessed over time as more monitoring information is gained and new models are developed.

The CENR report and Scavia et al. (2004) made recommendations on an N reduction target with reference to average fluxes for 1980–1996. These fluxes were calculated using different methods (see Section 3.1) than in this book, but the N reduction target proposed recently by Scavia and Donnelly (2007) used a combination of the newer USGS 5-year LOADEST and composite estimates since 1980. In this book we only use the 5-year LOADEST results, since the composite estimates are incomplete; however, they are very similar to each other (again, see Section 3.1).

During the past 5 years of record, annual water flux to the NGOM has declined by 5.8%, whereas nitrate-N and TKN have declined even more, leading to a total annual N reduction of about 21% (Table 4.1). Considering the original reduction target of a 30% reduction in total N, it would seem that substantial progress was made beyond the reduction that would occur from less flow alone. However, the largest reduction was in TKN, with a large part of this decrease from the Missouri River (discussed in Section 3.1). For the important spring flux of N, there was little reduction in nitrate-N beyond the reduced water flow (–11 and –12.4% declines in water and nitrate-N flux, respectively). Again, TKN was greatly reduced (–31.5%) during spring flows, leading to most of the decline in total N (–19.2%), beyond the reduction in water flux. This suggests that during the important high-flow spring period (April, May, June), reductions in nitrate-N flux to the NGOM have not occurred under management systems and programs now in place since the most recent report. However, the annual nitrate-N reduction indicates that the tile-drained corn and soybean systems in the Upper Mississippi and Ohio River subbasins seem responsive on an annual basis to the recent reductions in net N inputs, as discussed in Section 3.2. Whether spring nitrate-N loads will respond to these changes in NANI is uncertain at this time.

Table 4.1 Annual and spring (sum of April, May, June) average flow and N and P fluxes for the MARB for the 1980–1996 reference period compared to the most recent 5-year period (2001–2005). Load reductions in mass of N or P also shown
Full size table

For total P flux, both annually and during the spring, there were increases of 12.2 and 9.5%, respectively. It is not clear why total P fluxes are increasing (with corresponding smaller water fluxes), and the result suggests that the reduction target of 45%, relative to the 1980–1996 period, is close to 50% for the 2001–2005 period. Likewise, the 45% N load reduction target, relative to the 1980–1996 period, is equivalent to a 30% reduction relative to the 2001–2005 period. Fertilizer P consumption in the MARB has been relatively constant since about 1984 and is similar to consumption during 1970–1975. Net P inputs to the MARB have declined since the 1970 s and have been predominantly negative since the mid-1990 s (see Section 3.2 and Fig. 3.25). Table 4.1 also indicates N and P reduction recommendations in units of mass with reduction targets of 45% N and 45% P, assuming that the reduction was spread across all forms of N and P, that occur both annually and during the spring.

While the Study Group finds that both N and P reductions are warranted, additional modeling and dose–response research are needed to refine the reduction targets, particularly for P loading. Scavia and Donnelly (2007) presented the only model results that relate P loads to hypoxia in the NGOM. Further, there are no experimental data relating phytoplankton responses there to different levels of P. Ideally, targets for reducing P based on water quality should have greater model support and should consider dose–response relationships for P responses by the in situ phytoplankton communities. In the meantime, the response of the Gulf system to a specific amount of P reduction remains uncertain and must await the formulation of new models and dose–response relationships for the receiving waters. Water quality models aimed at evaluating the effects of these reductions will also rely on this information. Dose–response relationships should be developed using in situ bioassays designed to “ask the phytoplankton” what the response relationships and bloom thresholds are. These bioassay experiments are a logical follow-up to the work of Sylvan et al. (2006), which has shown P to be the limiting nutrient during periods of maximum primary production in the near-shore NGOM high-productivity zone. Bioassays are needed on a seasonal basis, where the effects of hydrologic variability and changing N:P input (loading) ratios on primary production, phytoplankton community composition, and biogeochemical and trophic fate can be evaluated.

In Section 4.5.8 on Most Effective Actions for Industrial and Municipal Sources, the Study Group provides some ballpark estimates of possible N and P reductions from upgrading major municipal wastewater treatment plants. The Study Group’s example calculations demonstrate that sewage treatment plant upgrades to achieve total N concentration limits of 3 mg/L and total P concentrations of 0.3 mg/L could create reductions in total annual N flux to the Gulf by about 10% and the total spring N flux by about 6%. Upgrading to achieve P concentrations of 0.3 mg/L would create reductions in P fluxes from sewage treatment plants from 41,000 metric tons P/year (45,000 ton P/year) to 10,500 metric tons P/year (11,600 ton P/year) or about a 75% reduction in annual flux from sewage treatment plants to the MARB. These reductions, in turn, would translate into reductions of total annual P flux to the Gulf by about 20% and the total spring P flux by about 15%. If further investigation and data collection confirms the Study Group’s calculations, upgrades to major wastewater treatment plants in the MARB could accomplish nearly half of the Study Group’s recommended P reduction targets. This would represent very significant progress for both improving water quality in the MARB and reducing hypoxia in the NGOM.

Despite the need for additional model and bioassay work, the proposed target of a 45% reduction in annual P load should be used in an adaptive management framework to allow development of strategies that optimize both N and P reductions while more knowledge is acquired on P reduction impacts on near-field hypoxia. Unlike N, the P reduction strategy will help address water quality impairments in the MARB. Given the evidence that both N and P should be reduced in the NGOM, setting a goal for P reduction should not await the development of new models and availability of new experimental data. Enough information exists now to set a goal in an adaptive management context beginning with the P reductions that are already feasible given existing technologies and options.

In 2000, USEPA recommended nutrient criteria to states and tribes for use in establishing their water quality standards consistent with Section 303(c) of the Clean Water Act (CWA) (USEPA, 2000c). USEPA’s recommended criteria represent an estimated “reference condition,” and it is assumed that the reference condition concentration would protect all designated uses (including the most protected uses, such as high-quality fisheries, sensitive aquatic life). The Study Group asked USEPA for a comparison of the Study Group’s recommended 45% reductions for TN and TP flux to the reductions in nutrient levels that would correspond to USEPA’s ecoregional nutrient criteria for reference conditions (USEPA, 2006b). This comparison is provided in Appendix E. Although a number of assumptions were required to make this comparison (see the caveats in Appendix E), USEPA’s preliminary analysis suggests that the Study Group’s recommended targets for reducing TN and TP are, for most regions, not likely to be as stringent as would be obtained if states adopted USEPA’s recommended reference condition values into state water quality standards for all waters. This comparison should not be interpreted as the Study Group’s endorsement of USEPA’s recommended nutrient criteria but rather an emphasis on the need to consider both within-basin nutrient criteria and NGOM load reduction goals. Numeric nutrient standards being developed by the states of the MARB will almost certainly be concentration rather than load based and may be most stringent during warmer, lower flow periods when absolute loads can be relatively low but when local waters are most frequently impaired by excess nutrient levels. It will be important for USEPA and other agencies to evaluate and, if necessary, reconcile within-basin water quality standards with load reduction goals for the NGOM. Strategies are needed for integrating standards throughout the MARB to better manage hypoxia as well as local water quality.

A mechanism in the Clean Water Act for addressing water quality impairments is the development of Total Maximum Daily Loads (TMDLs), though it is important to note that the focus of TMDL development is identification of the source and causes of water quality impairment, rather than on implementation of change for improving water quality. Under Section 303(d) of the Clean Water Act, states, territories, and authorized tribes are required to develop lists of impaired waters (i.e., waters that have not met water quality standards). The law requires that the appropriate jurisdictions develop TMDLs for these impaired waters. The TMDLs specify the maximum amounts of pollutants that waterbodies can receive and still meet water quality standards. In addition, TMDLs allocate pollutant loadings among point and nonpoint sources.

The status of nutrient criteria and TMDL development along the Mississippi River has been reviewed by the National Academy of Sciences (National Academy of Sciences, 2007). The National Academy of Sciences notes that none of the 10 Mississippi River mainstem states currently have numeric criteria for nitrogen or phosphorus applicable to the River and, that without such standards, there is little prospect of significantly reducing or eliminating hypoxia in the Gulf of Mexico. The National Academy of Sciences also describes how the process of developing numeric nutrient criteria and TMDLs for the Mississippi River could lead to water quality improvements in the Gulf of Mexico. NAS suggests that through such a process, USEPA could adopt the necessary numerical nutrient criteria for the terminus of the Mississippi River and waters of the northern Gulf of Mexico. Maximum nutrient loads could be assigned to each state and the loads could be translated into water quality criteria. Each state would then be required to develop a TMDL for waters that failed to meet the applicable criteria, and a coordinated effort could be undertaken to reduce point and nonpoint source loads to meet allocations established by the TMDLs. Thus, the NAS report identifies an approach through existing legislation (the Clean Water Act) that could be used to redress Gulf Hypoxia, but the SAB stresses that a great many steps exist between calling for “a coordinated effort” and implementing the full set of actions that must be undertaken for water quality to actually improve in the Gulf.

4.5 Protecting Water Quality and Social Welfare in the Basin

The Study Group has been asked whether social welfare can be protected while reducing hypoxia and improving water quality in the Basin. To thoroughly answer this question would require quantification of the full costs of all activities undertaken to reduce the necessary nutrient loading into the Gulf (from agricultural sources, point sources, air deposition, etc.) and the full benefits accruing from those activities. The benefits would include the direct benefits of reducing the size of the hypoxic zone (commercial fishery effects, recreational fishery gains, the value placed on preserving intact ecosystems, biodiversity, etc.) and the “co-benefits” (such as improved local water quality, increased wildlife habitat, flood control, aesthetic values).

Since the costs, benefits, and co-benefits will depend on the extent of coverage and specific locations of control options, a complete answer to the question would require knowing the details of how such nutrient reductions would occur. For example, if these reductions are to be achieved entirely through restoration of wetlands and tighter municipal source controls, it would be necessary to know where the wetlands would be located and where the point-source reductions would occur in order to estimate their costs and their co-benefits. In contrast, an entirely different set of co-benefits and costs would likely result from relying on a broader array of control options that also included nutrient management, increased perennials, riparian buffers, drainage management, and reductions in air deposition. Further, the exact policy approach (e.g., expanded EQIP funding, mandates, or taxes) would need to be specified if estimates of the incidence of the costs are to be estimated (i.e., whether the costs would ultimately be borne by taxpayers, by consumers, or by farmers and landowners).

To date, no set of models and/or studies have been undertaken that address all of the necessary components on a basin-wide scale to estimate the effects on social welfare. However, a number of studies, beginning with the research in the Integrated Assessment, have been done that address substantial components of this question. More complete efforts at quantifying the control costs than the benefits have been undertaken, though there remains a need for much more work on both sides of the equation. Integrated models at multiple levels and scales are needed to support this effort. The existing research focuses largely on agricultural nonpoint source control. This section summarizes findings from the limited set of large-scale economic-watershed models of agricultural nonpoint sources that have been applied to date.

4.5.1 Assessment and Review of the Cost Estimates from the CENR Integrated Assessment

Doering et al. (1999) in the Integrated Assessment undertook an ambitious cost-effectiveness analysis of several policy approaches to reach the N loss reduction goal of 20% established as part of the Integrated Assessment. The central modeling system they used was the US Mathematical Programming (USMP) model, which represents the agricultural sector in 45 production regions throughout the United States with 10 crops, 16 animal products, retail and processed products, and a range of domestic and international supply and demand relationships. Management practices include crop rotations, five tillage options, and varying fertilizer rates.

The environmental effects of various management practices and land uses in USMP are predicted by the EPIC model (the Environment Productivity Impact Calculator). USMP uses EPIC to predict changes in N loss, P loss, and sediment loss at the edge of the field from changes in land use and conservation practices. Donner et al. (2002) chose a 20% N loss reduction goal as “the best combination of sizable nitrogen loss reductions and acceptable economic costs” (Doering et al., 1999 p. 37). The remainder of their analyses focused on the evaluation of several policies that might achieve this environmental goal. Some key predictions from the modeling system include

  • A 20% reduction in fertilizer N application rates would result in the reduction of edge-of-field N loss by about 11%. In contrast, a 45% reduction mandate and fertilizer tax set to achieve a 45% reduction is predicted to result in the target goal of N loss reduction of about 20%. The less than proportional reduction in N loss coming from reduced fertilization in this modeling system is a result of predicted changes in acreage resulting from the feedback effect of price changes. Specifically, higher crop prices due to lower yields from the reduced fertilization rates induce more acreage planted to the fertilized crop, thereby partially offsetting the reduction in N. Whether the magnitude of the yield effects embedded in these models is accurate is an important question. For further discussion of this issue, see Section 4.5.6.

  • Some 7.29 million hectares (18 million acres) of wetland restoration would achieve the 20% reduction in N loss goal at a cost of over $30 billion.

  • Restoration of 10.9 million hectares (27 million acres) of riparian buffers was estimated to cost over $40 billion and generated relatively small reductions in N losses, suggesting that this strategy is not cost-effective for hypoxic zone control. In light of current evidence that phosphorous is also of concern, this result should be reconsidered as there is significant evidence that buffers can be quite effective in reducing sediment and phosphorous loss.

  • A “mixed policy” with a 2.02 million hectares (5 million acres) wetland restoration program in conjunction with a 20% fertilizer reduction is more cost-effective than most of the previous approaches, but the 45% reduction in fertilizer is more cost-effective yet.

  • The introduction of point–nonpoint source trading across the basin where the cap applies only to point sources will not achieve the 20% N loss reduction due to the relatively small magnitude of N contribution from point sources. Even with a stringent standard on point sources, only about 5% of the needed reductions occur.

  • These policies are likely to produce large “co-benefits” (i.e., other environmental benefits occurring within the basin and on-farm productivity benefits not immediately captured in the current profitability resulting from the policies). For example, the authors estimate that restoration of 405,000 hectares (1,000,000 acres) of wetlands would yield total benefits in the basin that exceed the costs, even without considering any benefits of hypoxia reduction.

Cost estimates used for the Integrated Assessment for a 20% reduction in N discharge coming from agricultural nonpoint sources range from $15 billion to $30 billion; however, these estimates suffer from a number of shortcomings including consideration of only a few options for reducing nutrient discharge and limited targeting. More inclusive assessments with better targeting of options to locations where they are most appropriate may reduce these costs.

In follow-up research, some of the same study coauthors (Ribaudo et al., 2001) compare nitrogen reduction methods with wetland restoration and low and high levels of N loss reduction. They find that nutrient management is more cost-effective at low levels of N loss reduction while wetlands restoration is more cost-effective at high levels. Tables 4.2 and 4.3 (listed at the end of this discussion) briefly summarize the key components of these studies and the other large-scale studies that are reviewed in the following discussion.

Table 4.2 Summary of study features of basin-wide integrated economic-biophysical models
Full size table
Table 4.3 Summary of policies and findings from integrated economic-biophysical models
Full size table

Due to limits on the understanding of the economics and natural science at the time, the work in the Integrated Assessment and its follow-up is based on assumptions that, in light of more recent research and availability of data, assessments could be improved upon in future work. The USMP model represents a wide variety of agricultural raw inputs and intermediate products at a relatively aggregate scale. However, it does not contain detailed description of land use, soil characteristics, yields, etc., at the individual field and/or subbasin scale. This inability to target finer scales could result in overstating the costs of meeting a particular reduction goal because significant cost savings can accrue from targeting land-management strategies.

The Integrated Assessment assumed a one-to-one relationship between the reduction in edge-of-field nitrogen loss and reduced loadings to waterways without incorporating the geographic differences in movement of N from the field of origination to the Gulf. Whether this shortcoming over- or understates the costs is an empirical question, but the results coming from a model that explicitly incorporates the fate and transport of nutrients and sediment might suggest very different results concerning the cost-effectiveness.

4.5.2 Other Large-Scale Integrated Economic and Biophysical Models for Agricultural Nonpoint Sources

Since completion of the Integrated Assessment, several basin-wide studies have evaluated policies that might reduce Gulf hypoxia and/or have effects on other environmental amenities that could be considered co-benefits (including carbon sequestration and upstream, local water quality indicators). The models can be divided into those that use the USMP modeling framework and those based on econometric estimates of behavioral response to economic drivers.

Booth and Campbell (2007) used a regression model to estimate the cost of reducing N losses when targeting conservation dollars to those areas with the highest proportion of fertilizer use. They modeled a hypothetical case in which conservation enrollment rises in direct proportion to the nonlinear rise in nitrate flux that occurs as fertilization intensity increases. The result was an increase in the amount of land in the high fertilizer watersheds enrolled in the Conservation Reserve Program by 2.7 million hectares (6.67 million acres) (a 29% increase over 2003 CRP levels) at a cost of $448 million. Booth and Campbell (2007) describe this as a 6.2% increase over the combined cost of commodity support and conservation programs. They account for the drop in commodity support spending that would accompany the enrollment of commodity-farmed land in the CRP. Booth and Campbell (2007) do not specify the percentage reduction in nitrate loading that would result from this scenario.

Wu et al. (2004) and Wu and Tanaka (2005) developed an econometric model of crop choice and tillage choice using the National Resources Inventory for the upper Mississippi River basin. They estimated the probability of adopting conservation tillage and crop choice based on a variety of physical and economic variables including land quality, slope, climate conditions, and profits. They used over 40,000 crop land points observed for 16 years, although only a subset of the observations were used for model fitting. These adoption models then simulate adoption profiles under alternative policies. Finally, the environmental effects of the policies are predicted with a biophysical model. Wu et al. (2004) used a set of environmental production functions estimated via a meta-modeling approach (Wu and Babcock, 1999), based on data generated from the EPIC model. They found that crop rotations are not a cost-effective strategy to N reduction.

Wu and Tanaka (2005) used the SWAT model to predict water quality changes from the policies. They considered the same two policies as Wu et al. (2004), as well as a policy that would increase the amount of land set-aside in a Conservation Reserve-type program and a fertilizer tax at various rates. They found a fertilizer tax to be the most cost-effective of policies they considered.

Kling et al. (2006) employed a similar econometric modeling approach. Like Wu et al. (2004), they used the National Resource Inventory data to link the cost data with the SWAT model. They estimated the costs and water quality benefits of implementing a set of conservation practices associated with implementation rules based on distances to a waterway, slope, and erodibility indices. The conservation practices assessed include grassed waterways, nitrogen management, terraces, buffers, land retirement, and conservation tillage. They estimated that this placement of conservation practices on the landscape would cost over $800 million annually (or roughly $16 billion if viewed as a lump sum cost assuming a 5% rate of discount) and would achieve a 22% reduction in N loadings into the upper Mississippi River basin at Grafton, IL. Within the UMRB, they estimated a 40–66% reduction in sediment loads, a 6–47% reduction in P loads, and a 9–29% reduction in N loads. These estimates (like those from all of the studies reviewed here) are likely to be very sensitive to the set of conservation practices included and the specific scenarios studied.

Greenhalgh and Sauer (2003) used the USMP augmented in two important ways: (1) they configured the model by watersheds and added information on municipal waste water treatment plants and (2) they included “attenuation” coefficients derived from the SPARROW model to reflect the transport component of N flows between watersheds. The focus of their work was on policy options for hypoxia that also contribute to greenhouse gas reductions. The policies they considered include N trading between point and nonpoint sources, GHG trading assuming external carbon prices of $5/ton and $14/ton, N trading with additional payments for GHG emission reductions, an N fertilizer tax, a subsidy to farmers willing to shift from conventional to conservation tillage, and an expansion of the CRP program to 16 million hectares (40 million acres) nationwide. Of the policies evaluated, none achieved the 20% reduction goal of the Doering et al. (1999) analysis. The largest reductions were achieved in their simulation of point/nonpoint source trading with a stringent N standard. The most cost-effective policies were also the trading programs.

Ribaudo et al. (2005) also considered the possibility of N trading between point and nonpoint sources using the USMP model. They found that trading has significant potential to reduce costs relative to a requirement that wastewater treatment plants be required to install stringent nutrient removal technology. Footnote 1

These studies shed light on the costs of addressing the hypoxia problem from conservation practices in the agricultural sector and the way these costs may vary depending on the policy instrument chosen (trading program, conservation payment, tax, etc.). These studies also directly bear on the question of how much it will cost to address local water quality in the MARB. However, as noted above, shortcomings of the integrated models have prevented assessment of many policies as well as conservation practices and sinks. None of the models include point source and nonpoint source control options. With the exception of Booth and Campbell (2007), most models have not adequately addressed the cost savings associated with targeting. Nonetheless, results to date suggest that there is large variability in the costs of alternative policies. The issue of who pays these costs may also be important to consider since the incidence (who must pay the costs) may differ dramatically across policies. A notable example is a fertilizer tax, which has the same social costs as a restriction but which may have a much higher incidence on farmers.

Improved estimates of the costs of installing and maintaining conservation practices could be generated with the current suite of models by considering alternative sets of conservation practices. This can be accomplished using the following steps: (1) identifying conservation practices that are most likely to be effective in reducing nutrients important for hypoxia and (2) identifying scenarios that place these conservation practices on the landscape. These scenarios could be based on rules of thumb (identifying, for example, a particular conservation practice to be used on cropland with specific climate and soil characteristics); algorithms for optimal placement to minimize costs; multiple goals, such as maximizing in basin co-benefits or income support; or policy-relevant methods, such as the use of an environmental benefits index; or computing cost estimates from economic models and water quality changes from watershed models.

4.5.3 Research Assessing the Basin-Wide Co-benefits

As noted above, many of the same practices that could contribute to reductions in the hypoxic zone could also have significant effects on local water quality, carbon sequestration, wildlife habitat, flood control, and other ecosystem services. The physical co-benefits of many conservation practices and sinks are described in Section 4.5.10. On the basin-wide scale, there are a few studies that provide physical measures of one or more co-benefits that are associated with implementation of conservation practices that would address hypoxia, particularly related to carbon sequestration and water quality (see, for example, Feng et al. (2005), Lewandrowski et al. (2004), Greenhalgh and Sauer (2003)). These studies consistently indicated that significant co-benefits are present, but these estimates are not monetized and are reported in physical units. Further, the policies analyzed are not focused on hypoxia reduction.

Thus, the work reported in the Integrated Assessment remains the most complete coverage to date of the potential value to MARB residents of the water quality and other co-benefits. The estimates provided there suggested that the monetized value of the benefits to the basin were larger than the costs based primarily on benefit estimates of the value of erosion control and wetlands restoration. A more complete accounting of these benefits could be developed using benefits transfer techniques, although there are many ecosystem services for which currently accepted methods are not likely to adequately fully capture the value of the benefits. But, in any case, because the Integrated Assessment was not able to quantify all co-benefits, total co-benefits within the basin would almost certainly be larger than those estimated.

Due to the incredible complexity in this system, as well as limits in data, modeling, and research, definitive statements on social welfare are not possible. For example, there is incomplete information on the costs of farm-level actions to reduce edge-of-field nutrient losses. There is even greater uncertainty in quantifying the effectiveness of farm-level nutrient control actions in reducing watershed-level nutrient flux and about the relationship between watershed-level nutrient flux and the spatial and temporal dimensions of the hypoxic zone. These uncertainties are further exacerbated by the possibility of regime shift in the Gulf of Mexico, whereby the system could become more susceptible to hypoxia following the initial occurrences. If regime shift is a factor, then historic data on the relationship between nutrient flux and the size of the hypoxic zone does not provide guidance on the decrease in nutrients required to achieve a given reduction in the size of the hypoxic zone. Hence, a return to historic lower levels of nutrient fluxes might not be adequate to return to a corresponding size of the hypoxic zone.

There are many sources of uncertainty in the economic, hydrologic, and Gulf systems that make it difficult to render definitive conclusions about social welfare. Indeed, it is precisely because of these many uncertainties and need for additional research that we recommend an approach based on an adaptive management strategy that aims to move in a “directionally correct” fashion, rather focusing on achieving a precise outcome.

While we cannot definitely say that we can achieve the 5,000 km2 (1,930 mi2) goal while maintaining social welfare, there is evidence that suggests it is feasible to do so. First, and perhaps most importantly, welfare losses in the Basin will be at least partially or even totally offset by co-benefits of nutrient reduction actions. For example, if wetlands restoration is used to control nutrient flux, it will result in improvements in wildlife habitat and local water quality, both of which will improve welfare in the Basin. Findings from the Doering et al. (1999) assessment point out that the benefits accruing locally from wetlands restoration might well exceed the costs, even without any Gulf hypoxia reductions. Similar estimates are reported in Hey et al. (2004) for substantial restoration of wetlands in flood plains (see Section 4.4.2). Management actions that reduce farm-level nutrient losses may lead to better local water quality, thereby improving welfare for affected residents within the Basin. If management actions are undertaken to control air emissions, thereby reducing atmospheric deposition of nitrogen, it will result in improvements in air quality, reduction in acid precipitation, lower emissions of greenhouse gasses, etc. Thus, co-benefits within the Basin will at least partially and perhaps fully offset welfare losses associated with the costs of implementing management actions. And in the longer term, a transition from corn to perennial crops could benefit farmers and other Basin residents. Thus, there may be larger scale transitions in the agronomic system that provides opportunities to reduce nutrient flux while maintaining welfare in the Basin.

A second reason for optimism is that cost-effective approaches, such as targeting low cost sources and using emissions trading, have not yet been applied. These approaches have the potential to reduce the costs of nutrient control, possibly considerably, thereby reducing the burden of complying with the goal. Thus, there may be opportunities to control the cost of nutrient reduction.

4.5.4 Principles of Landscape Design

Another perspective for protecting social welfare can be drawn from the principles of landscape design. A landscape perspective involves broad-scale consideration of how decisions affect resources, particularly in the long run. Guidelines have been proposed as a way to facilitate land managers considering the ecological ramifications of land-use decisions (Dale et al., 2000). These guidelines are meant to be flexible and to apply to diverse land-use situations, yet require that decisions be made within an appropriate spatial and temporal context. These landscape design guidelines can serve as a checklist of factors to be considered in making decisions that relate to implications for hypoxia in the Gulf.

  • Examine the impacts of local decisions in a regional context. The spatial array of habitats and ecosystems shapes local conditions and responses (e.g., Patterson, 1987; Risser, 1985) and local changes can have broad-scale impacts over the landscape. Hypoxia is a classic example of such impacts (Russo et al., 2008), for fertilizer applications in the Midwestern states can affect oxygen conditions in the Gulf of Mexico. This guideline notes that it is critical to examine both the constraints placed on a location by the regional conditions and the implications of decisions for the larger area. Therefore, it is critical to identify the surrounding region that is likely to affect and be affected by the decision and examine how adjoining jurisdictions are using and managing their lands. Forman (1995) suggests that land-use planning should first determine nature’s arrangement of landscape elements and land cover and then consider optimal spatial arrangements and existing human uses. Following this initial step, he suggests that the desired landscape mosaic be planned first for water and biodiversity; then for cultivation, grazing, and wood products; then for sewage and other wastes; and finally for homes and industry. Of course, planning under pristine conditions is typically not possible. Rather, the extant state of development of the region generally constrains opportunities for land management.

  • Plan for long-term change and unexpected events. Impacts of decisions can, and often do, vary over time as a result of delayed and cumulative effects. Future options are often constrained by the decisions made today as well as by those made in the past. For example, areas that are urbanized are unlikely to be available for any other land uses because urbanization locks in a pattern on the landscape that is hard to reverse. Thus, management actions should be implemented with some consideration as to the physical, biological, esthetic, or economic constraints that are placed on future uses of resources. External effects can extend beyond the boundaries of individual ownership and thus have the potential to affect surrounding owners. Planning for the long term also requires consideration of the potential for unexpected events, such as variations in temperature or precipitation patterns or disturbances. Long-term planning must also recognize that one cannot simply extrapolate historical land-use impacts forward to predict future consequences of land use. The transitions of land from one use or cover type to another often are not stable over time because of changes in demographics, public policy, market economies, and technological and ecological factors.

  • Preserve rare landscape elements, critical habitats, and associated species. This guideline implies a hierarchy of flexibility, and it implicitly recognizes ecological constraints as the primary determinants in this hierarchy. For example, a viable housing site is much more flexible in placement than an agricultural area or a wetland dedicated to improving water quality and sustaining wildlife. Optimizing concurrently for several objectives requires that planners recognize lower site flexibility of some uses than others. However, given that most situations involve existing land uses and built structures, this guideline calls for examining local decisions within the regional context of ecological concerns as well as in relation to the social, economic, and political perspectives that are typically considered.

  • Avoid land uses that deplete natural resources over a broad area. Depletion of natural resources disrupts natural processes in ways that often are irreversible over long periods of time. The loss of soil via erosion that can occur during agriculture and the loss of wetlands and their associated ecological processes and species are two examples. This guideline requires the determination of resources at risk, which is an ongoing process as the abundance and distribution of resources change. This guideline also calls for the deliberation of ways to avoid actions that would jeopardize natural resources and recognition that some land actions are inappropriate in a particular setting or time, and they should be avoided.

  • Avoid or compensate for effects of land use on ecological processes. Negative impacts of land-use practices might be avoided or mitigated by some forethought. To do so, potential impacts need to be examined at the appropriate scale. At a fine scale, farm practices may interrupt ecoregional processes. At a broad scale, patterns of watershed processes may be altered, for example, by changing drainage patterns as part of the land use. Therefore, how proposed actions might affect other systems (or lands) should be examined. For example, human uses of the land should avoid uses that might have a negative impact on other systems; at the very least, ways to compensate for those anticipated effects should be determined. It is useful to look for opportunities to design land use to benefit or enhance the ecological attributes of a region.

  • Implement land-use and -management practices that are compatible with the natural potential of the area. Local physical and biotic conditions affect ecological processes. Therefore, the natural potential for productivity and for nutrient and water cycling partially determine the appropriate land-use and management practices for a site. Land-use practices that fall within these limits are usually cost-effective in terms of human resources and future costs caused by unwarranted changes on the land. Nevertheless, supplementing the natural resources of an area by adding nutrients through fertilization or water via irrigation is common. Even with such supplements, however, cost-effective management recognizes natural limitations of a site. Implementing land-use and -management practices that are compatible with the natural potential of the area requires that land managers understand a site’s potential. For example, land-management practices such as no-till farming reduce soil erosion or mitigate other resource losses. Often, however, land uses ignore site limitations or externalize site potential. For example, building shopping malls on prime agriculture land does not make the best use of the site potential. Nevertheless, land products are limited by the natural potential of the site.

Together these guidelines form the basis of a landscape design perspective that should improve the ability to understand and manage the complex system that is affecting hypoxia in the Gulf of Mexico.

4.7 Cost-Effective Approaches for Nonpoint Source Control

While the Action Plan and this Study Group urge the reliance on adaptive management principles, a variety of tools can be used as the vehicle for implementation within adaptive management. The current Action Plan indicates a principle of encouraging “actions that are voluntary, practical, and cost-effective” (page 9). Additionally, the plan will “utilize existing programs, including existing State and Federal regulatory mechanisms,” as well as identify needs for additional funding. These statements include a variety of tools ranging from purely voluntary programs (those with no associated financial incentives) to current conservation programs funded by state and federal agencies (such as the Conservation Reserve Program [CRP] and the Environmental Quality Incentive Program [EQIP]) to water quality trading. Research assessing the costs and effectiveness of these approaches is addressed in this section.

Complicating the design of cost-effective approaches is the geographic distance between the sources of nutrients and the receiving waters downstream. Two identical farm fields in different locations (with resulting differences in the hydrology of the local watershed) will send differing amounts of nutrients to the Gulf. Hence, the effectiveness of a practice or sink in a particular location depends on what sources and sinks are present elsewhere in the watershed. Whether it is cost-effective to install a buffer at a particular location may depend upon whether there is a wetland at the base of the watershed, whether conservation tillage is being practiced elsewhere, etc. Thus, rather than focus on individual practices, policy options that can simultaneously encourage the adoption of practices and sinks that are jointly cost-effective will best protect social welfare in the Basin.

It is important to clarify the concept of “costs.” Here, “costs” refers to the least amount of compensation needed to effect change, e.g., the compensation that would be necessary for a landowner or farmer to adopt a conservation practice. This is the standard concept of economic cost, relevant to any good or service. This cost includes “direct” costs, such as the cost of new equipment, building of structures, and labor to manage a practice, as well as a myriad of potential “indirect” costs, such as lost profits from adopting the practice and compensation for added risk from the practice. Components of these costs can be negative; that is, it may actually increase profitability to adopt some practices (conservation tillage in certain circumstances is a notable example).

Second, the focus of most economic studies is on total costs with little or no consideration paid to what subset of society actually bears the costs (incidence) of the policy. This focus on efficiency (seeking the lowest cost approach) is based on the premise that compensation could always be paid to those bearing the cost in some form so that society will be best off if the lowest cost option is pursued. However, since such compensations are rarely paid, the issue of who pays is likely to enter the policy decision. Complete information on the incidence of alternative tools in this context is not available, but where appropriate, we note the likely incidence considerations.

4.7.1 Voluntary Programs – Without Economic Incentives

There is a small and growing literature concerning the effectiveness and optimal design of voluntary agreements that do not have positive or negative financial incentives associated with them (Morgenstern and Pizer, 2007; National Research Council, 2002). Key insights were presented in a game-theoretic model by Segerson and Miceli (1998), who identified the conditions under which voluntary agreements are likely to yield efficient pollution levels without significant economic incentives. They studied voluntary agreements that are based on threats of harsher outcomes if the goals are not met, using the example of mandatory abatement requirements if the voluntary agreement does not succeed in meeting the pollution goal. The premise is that firms will voluntarily agree to reduce pollution if they can avoid the costs that future mandatory controls would otherwise bring. In the absence of financial compensation, the presence of a positive probability of a penalty (or cost in the form of mandatory control) is required to support Segerson and Miceli’s findings that there are situations in which efficient levels of pollution control can be achieved with voluntary agreements (without economic incentives). They found that pollution reduction is likely to be small when the background threat is weak.

Empirical work also sheds light on the efficacy of voluntary agreements that do not have financial incentives. Mazurek (2002) identified 42 voluntary environmental initiatives sponsored by the federal government since 1988. Although the programs she identified are largely outside the realm of agriculture, her conclusions are relevant. Mazurek concluded that a variety of implementation problems have led to “lower-than-expected” environmental results for voluntary (without financial incentive) agreements, a result consistent with findings of a 1997 USGAO (1997) report concerning four voluntary agreements related to climate change.

In the same National Research Council report (2002), Randall identified three essential functions for government if voluntary agreements (without financial incentives) are to be effective. These key functions are meaningful monitoring to back up a threat of government inspection, “credible threat of regulation” if the goals are not met, and a clear liability system to punish “blatant polluters and repeat offenders.” Randall concluded that “voluntary (or negotiated) agreements, industry codes, and green marketing should be viewed as promising additions to the environmental toolkit, but they should supplement, not supplant, the regulatory framework. They make a nice frosting on the regulatory cake. But the cake itself must be there.”

Finally, Morgenstern and Pizer (2007) presented seven case studies on voluntary agreements (without economic incentives) in the United States and elsewhere. Point estimates of environmental improvements attributable to the voluntary programs ranged from negative values (actual declines in environmental performance) to a maximum of 28% improvement in environmental performance. Morgenstern and Pizer concluded “that voluntary programs have a real but limited quantitative effect….”

Given the historical aversion to imposing mandatory requirements in agriculture, the collective weight of these studies suggest that voluntary agreements that do not have incentives associated with them are not likely to be adequate on their own to achieve significant reductions in nutrient runoff. In short, voluntary programs without incentives can have small effects but cannot be relied upon to induce major environmental improvements.

4.7.2 Existing Agricultural Conservation Programs

Currently, the largest incentive-based conservation programs related to agriculture are the EQIP and CRP. A potentially significant program introduced in the 2002 Farm Bill was the Conservation Security Program (CSP), which has been funded only partially and implemented incrementally. The CRP pays farmers to retire land, and the other two pay farmers to implement conservation practices on their farms (EQIP is a cost-share program; CSP was intended to cover the full costs of adoption). Numerous studies undertaken by USDA’s Economic Research Service and others have estimated the magnitude of environmental benefits from these programs in physical terms (e.g., tons of erosion reduction, acres of habitat preserved, acres of wetlands restored) and some efforts have been made to monetize these benefits [see Claassen et al. (2004) for a summary of CRP studies as well as Haufler (2005)]. The Conservation Effects Assessment Program (CEAP) was initiated in an attempt to provide nationwide estimates of the benefits provided by the full suite of conservation programs; a national assessment of the water quality benefits is being developed currently (Bob Kellogg, presentation to SAB Hypoxia Advisory Panel, December 6, 2006).

The CRP pays landowners to take their land out of crop production and place it in perennial vegetation or trees, depending on the region of the country, with a goal of creating wildlife habitat and reducing erosion (and originally to reduce crop production). The CRP enrolls about 10% of total US cropland, nearly all in 10-year contracts although there is significant concern that high corn prices due to ethanol expansion may rapidly reduce this amount. A number of studies have identified large environmental benefits associated with the CRP [Smith and Alexander (2000), Feather et al. (1999)]. The program has used an Environmental Benefits Index (EBI) since 1990 to prioritize parcels for inclusion in the program that gives points to land based on particular environmental attributes and cost. The movement from targeting erodible lands (prior to 1990) to the use of the EBI for targeting has been estimated to have doubled the benefits from the program (Feather et al., 1999). Ribaudo (1989) estimated that a CRP enrollment that targets lands based on environmental damages (benefits) would have significantly greater benefits still. By redesigning the weights in this index, the program could target land that is predicted to contribute high nutrient loadings to the Gulf.

Many other studies have addressed the cost-effectiveness of land retirement to achieve environmental benefits within the context of the CRP. In a series of papers assessing the efficiency of the Conservation Reserve Enhancement Program (CREP) in Illinois, Khanna et al. (2003) linked the AGNPS model with site-specific characteristics of parcels to examine the relative efficiency of alternative targeting mechanisms (Yang et al., 2003, 2004, 2005). Extremely large gains from targeting were reported; for example, Yang et al. (2004) estimated that with targeting, 30% less cropland could have been retired (at almost 40% less total cost), while achieving 20% reductions in erosion instead of the actual 12% reduction.

The EQIP program is a cost-share program for conservation practices in livestock facilities and on land that remains in agricultural production. A prospective benefit cost analysis (as required by Executive Order 12866) predicted over $5 billion in net benefits from the EQIP program as implemented under the 2002 Farm Bill, even though not all of the benefits could be monetized (US Department of Agriculture, 2003).

The Wetland Reserve Program (WRP), Grassland Reserve Program (GRP), and Wildlife Habitat Incentive Program (WHIP) are all smaller land retirement programs that also could potentially benefit efforts to reduce Gulf hypoxia. Additional information on the large-scale potential for wetlands is provided by Hey et al. (2004), who addressed the question of whether the social benefits from restoring up to 2.83 million hectares (7 million acres) of cropland in the 100 years floodplain of the upper Mississippi River basin to wetlands exceed the costs. The benefits include reduced flood-related crop damages; reduced crop subsidies; and non-flood-related recreation benefits of wetland conversion, including fishing, hunting, and general recreation usage. These benefits were compared to estimates of the costs of cropland conversion comprised of farm rental rates (representing the present value of farmland income) and the costs of wetland construction and maintenance. Hey et al. (2004) estimated that the benefits exceed the costs in all locations considered except one county in Missouri. In the context of NGOM hypoxia, this difference is especially striking because the benefits exceed the costs for this conversion even without considering any benefits from reduction of the hypoxic zone. As the authors carefully pointed out, the social efficiency of converting 2.83 million hectares (7 million acres) does not mean that private benefits will exceed the private costs for all parties. Individual landowners would stand to lose while recreationists accrue benefits.

These findings represent an important addition to the assessment of wetlands in the Integrated Assessment. While Doering et al. (1999) concluded that wetland restoration was less cost-effective than fertilizer reductions, their analysis did not include cost savings from crop subsidy reductions nor flood-related crop damages. In addition, the Hey et al. (2004) work focused on wetlands targeted in flood plains. The study suggests two points of key importance for NGOM hypoxia: (1) there is a large amount of acreage that is situated in locations that potentially could serve as nutrient sinks in the upper Mississippi River basin, and (2) the co-benefits of this action are large enough, in and of themselves, to justify the social efficiency of converting this land to nutrient sinks even without considering the benefits associated with reducing Gulf hypoxia.

The programs mentioned above can be categorized into one of two groups: land retirement programs and “working” land programs. Both the CRP and WRP are examples of land retirement programs, since landowners receive payments in exchange for taking land out of active agricultural production and putting the land into perennial grasses, trees, or wetlands restoration. In contrast, EQIP and the CSP are examples of working land programs whereby landowners or producers receive payments to cover part or all of the costs of making changes in conservation practices or management decisions on their land that remains in agricultural production. Some research has addressed the cost-effectiveness of working land programs versus land retirement programs. For example, Feng et al. (2006) found that a cost-effective allocation of resources to sequester carbon in agricultural soils favors working land (via conservation tillage subsidies) over land retirement (via payments to retire land and plant it in perennial grasses). It is important to note, however, that this study focused on stylized working land and land retirement programs rather than attempting to address the cost-effectiveness of existing conservation programs as actually implemented.

The existing working land and land retirement programs are implemented with features that likely affect the cost-effectiveness of the programs for achieving environmental gains in different ways. For example, the CRP uses an EBI that favors admitting land into the program that achieves environmental benefits at relatively low costs. All else equal, this component of the program will improve its cost-effectiveness. In contrast, the CSP provides payments for ongoing stewardship of farmers so that program expenditures are used to reward past behavior rather than to change existing behavior. This, all else equal, will reduce the program cost-effectiveness for achieving environmental gains. The lack of competitive bidding and clear targeting also reduces the cost-effectiveness of this program. Finally, it is worth noting that targeting and competitive bidding were explicitly disallowed in the EQIP program during its most recent reauthorization. Again, this will reduce its cost-effectiveness.

4.7.3 Emissions and Water Quality Trading Programs

Emission trading is a regulatory approach that sets a maximum allowable level of overall emissions and then allows sources to exchange pollution allowances. A properly structured trading program can reduce the costs of achieving emission standards by allowing the flexibility necessary to focus pollution reductions on sources that are less expensive to control. In theory, a broad-based emissions trading program could help to reduce the air and water contributions of nutrients to the NGOM. Water quality trading is simply the name given to the extension of emissions trading to achieving water quality objectives.

In a recent survey of the programs to support water quality trading in the United States, Breetz et al. (2004) identified 40 water trading initiatives and an additional six state policies with specific programs related to water quality trading. USEPA has supported these programs (USEPA, 2004a) and has produced explicit policies related to their implementation. Many states and regions also have explicit policy guidance. However, the effectiveness of these programs appears to have been quite limited as very few trades are actually occurring. Further, little evidence of environmental improvement associated with these programs exists (Breetz et al., 2004).

A key problem with these programs is the lack of a required water quality improvement necessary to generate adequate demand for credits (King, 2005). To achieve “cap and trade,” an effective cap is necessary. A cap could come from a tight enough cap on point sources such that they would find it cost-effective to purchase credits from agricultural nonpoint sources. Alternatively, the cap could be extended to agricultural sources. While some have conjectured that the Total Maximum Daily Load (TMDL) program may eventually play this role, there is no current mandate for agricultural sources to restrict nutrient runoff. Also problematic are a range of restrictions on allowable trading, such as requirements that a particular baseline set of conservation practices be in place with credits accruing only for additional conservation activity.

While trading could be a significant contributor to cost-effective nutrient control, the necessary institutions for water and/or air emissions trading to be an effective policy instrument are not broadly in place. In addition to clear and enforceable limits on emissions or water quality contributions (from point and/or nonpoint sources), enforceable rules concerning trading ratios, liability when standards are not met, monitoring, etc. must be established before these markets can flourish. Ideally, a trading program to address NGOM hypoxia would be broad based and include highly diverse sources (such as air deposition and many agricultural nonpoint sources) to maximize the potential for cost savings.

4.7.4 Agricultural Subsidies and Conservation Compliance Provisions

US farmers have been the recipients of farm payments for decades. These payments support prices and/or income, especially of farmers growing bulk commodities such as corn and soybeans. Economic theory suggests that, all else equal, such payments will increase the intensity and acreage of farming, possibly resulting in increased water quality problems. Research by Reichelderfer (1985) provided empirical evidence that these payments encourage crop production on highly erosive land. Likewise, a recent study from USDA’s Economic Research Service (Lubowski et al., 2006) quantified the effect of one major program, subsidized crop insurance, on the location and acreage of cropland and its environmental effects. Lubowski et al. (2006) estimated that about a million hectares (2.5 million acres) were brought into production as a result of the program and that these lands are more vulnerable to erosion, are more likely to include wetlands, and have higher levels of nutrient losses than average.

To some extent, USDA’s conservation programs (see Section 4.4.2) exist to counteract the “perverse effects” or unintended consequences of its crop subsidies inasmuch as government financial support has encouraged farmers to choose commodity crops that require more fertilizer, maximize yield without regard to soil and water quality consequences, and cultivate marginal land. Restructuring or eliminating existing subsidies could serve to mitigate some of these perverse effects (e.g., by shifting subsidies to reward less fertilizer-intensive crops as well as by requiring, as a condition of receiving subsidies, certain conservation practices).

Taheripour et al. (2007) provided additional evidence on this point. First, their model suggests that removal of all crop subsidies would reduce nitrogen pollution by 8.5% and that the reduced need for distortionary income taxes to support these subsidies could increase social welfare by $1.2 billion. Further, they found that tax-neutral policies to achieve nitrogen reduction can generate significant double dividends (a double dividend refers to a situation where a policy not only internalizes an externality but also reduces the deadweight losses associated with distortionary taxation, such as an income tax). They provide an estimate of the magnitude of the double dividend for a range of nitrogen-reduction goals and policy approaches including a nitrogen tax, a nitrogen reduction subsidy, a tax on output, and a combined output tax and nitrogen reduction subsidy and find that a double dividend from these instruments can be significant.

While environmental improvements associated with agriculture have largely been pursued via cost-share or subsidy programs, one significant regulatory approach has been the implementation of environmental compliance provisions that require farmers who receive farm program payments (including price support and income support) to undertake some environmental performance practices. Specifically, in the 1985 Food Security Act, conservation compliance provisions required owners of highly erodible land (a categorization of land based on its slope and soil type) to implement soil conservation plans, and a “swampbuster” provision disallowed payments to go to farmers who converted wetlands to crop land. Claassen et al. (2004) estimated that up to 25% of the reduction in soil erosion that occurred between 1982 and 1997 was attributable to conservation compliance. Many believe these gains could have been higher if there had been stronger enforcement of the mechanism. While no direct estimates are available of the increased benefits that could come from more enforcement, there is evidence of very limited reporting and penalizing of violations (Claassen, 2000).

Claassen et al. (2004) assessed the prospect for reducing nutrient losses from the Mississippi River basin by extending compliance requirements to nutrient management. They used “nutrient management” to refer to the range of activities related to the timing and level of fertilization decisions that best minimizes soil nutrients in excess of crop needs at any point in time. They noted that the ideal set of nutrient-management practices will vary considerably across farms and regions and that the costs of these activities will also vary notably across this space. Using data from the EQIP program, they summarized the distribution of incentive payments needed to induce willing adoption of nutrient management practices as defined under EQIP. For the Heartland region (ERS Farm Resource Region), the average annual incentive payment is about $7/ac, and 95% of the payments are $12/ac or less.

While these data provide an excellent starting point for assessing the cost-effectiveness of nutrient management methods addressing local water quality and NGOM hypoxia, several additional pieces of information would be needed for a full assessment. First, these costs represent the compensation needed for those farmers who have already adopted practices under the EQIP program; those who have not adopted are likely to have at least as high costs, possibly substantially higher. In this regard, these costs could be viewed as a lower bound. Second, these costs are specific to the EQIP requirements for nutrient management. Whether these requirements are effective enough to yield substantial off-site benefits is not addressed. Nonetheless, based on this cost assessment and a comparison with the annual commodity program payments farmers typically receive, Claassen et al. (2004) concluded that substantial nutrient management could occur with extension of conservation compliance provisions to nutrients.

Claassen et al. (2004) also considered whether buffer practices could be induced under conservation compliance provisions. They included riparian buffers, filter strips, grassed waterways, and contour grass strips in their discussion of buffer practices. To assess the costs of these practices and how they vary across locations, they looked at information on producers’ willingness to accept compensation for adoption of the practices observed for continuous CRP priority areas. Owners of these lands received an average payment of about $90/year in addition to 50% cost share for installation of the buffer practice. Based on this analysis, as an example, Claassen et al. (2004) computed the annual costs per area for a filter strip and concluded that, in many cases, this payment would be below the average subsidy received by producers, thereby suggesting that buffer practices might also be successfully adopted under nutrient compliance provisions.

Finally, Claassen et al. (2004) noted that conservation compliance provisions are likely to have few transaction costs relative to other policies (although enforcement costs would need to be considered) and require very low budgetary outlays beyond the payments that are already provided for commodity or insurance programs. Claassen et al. (2004) also argued that conservation compliance requirements have been relatively cost-effective due to the flexibility with which they can be implemented. Producers in different regions of the country, with differing soil and weather conditions, can meet their compliance obligations with different practices. This flexibility means that the most appropriate technologies can be used for the location of the practice.

4.7.5 Taxes

The use of a per unit tax to internalize the costs of externalities of production is well known to be highly cost-effective when the tax is placed directly on the externality generating activity; these “Pigouvian” taxes are the equivalent of placing the appropriate price on the pollutant (Baumol and Oates, 1988). Taxes can be a powerful market signal, communicating the need to change behavior, Baumol and Oates (1988) demonstrated that subsidies (essentially just negative taxes) can also be designed that provide the equivalent market signals for changes in behavior. This argument is often used to support the design of environmental programs that pay participants for the provision of environmentally friendly practices rather than using taxes to change behavior. A potentially important exception to this equivalence can occur when the provision of a positive payment induces entry into the farming sector generating production on otherwise unprofitable lands. This possibility was addressed in Section 4.4.4 in the context of general agricultural subsidies and conservation compliance.

A tax directly on an input into production that is highly correlated with the pollutant can be an efficient second-best policy. The possible use of a nitrogen fertilizer tax was considered in Doering et al. (1999) and found to be as cost-effective as any of the policies they considered (they note that the initial incidence falls on farmers). Fertilizer taxes already exist in some states but are set at much smaller levels than those studied by Doering et al. (1999). The inelastic demand for fertilizer (Denbaly and Vrooman, 1993) means that the magnitude of taxes needed to induce behavioral change would likely be large.

The incidence of a tax (and thus determination of who pays the costs) is likely to fall on farmers and consumers of food products made from crops that use fertilizer. In contrast, the incidence of conservation program payments is largely on taxpayers. Finally, it is important to note that tax instruments will be more efficient the more broadly they are applied to the various nutrient sources identified as pollutant contributors; so ideally a tax would be applied to all nutrient sources rather than singly to fertilizer.

4.7.6 Eco-labeling and Consumer Driven Demand

The idea that environmentally friendly producer behavior can be induced by consumer demand is one basis for eco-labeling and certification programs. Dolphin safe tuna (Teisl et al., 2002) and organic fruits and vegetables (Loureiro et al., 2001) are two successful examples. Research analyzing the effectiveness of eco-labeling suggests some promise.

Thogersen (2002) summarized three schemes, all implemented in Europe, that have been credited with significant reductions in emissions from heating appliances and paint solvents (the German “Blue Angel” brand) and reductions in pollutants from paper production and household chemical and laundry emissions (the Swedish “Good Environmental Choice” label and the Nordic “Swan” label). Although not specific to a particular product, Clark and Russell (2005) noted that several studies of the Toxic Release Inventory have shown that information can affect firms’ choices.

Could consumer-driven demand affect the changes in land-use and agricultural management necessary to contribute notably to nutrient flows into the Gulf? This approach would require the labeling of food and fiber products made from agricultural outputs in the MARB to indicate that they were produced in such a way as to reduce or eliminate nutrient contributions to hypoxia. Consumers would then need to respond to this labeling by purchasing products, presumably at a higher cost, in adequate quantity to change the market behavior. Given that much of the grain produced in the Corn Belt is used for livestock feed and not directly traceable to its field of origin, it will be difficult to distinguish products that were produced with “hypoxia -friendly” production practices from those that were not. It is not clear that labeling can credibly be produced without significant government involvement and expense (Crespi and Marette, 2005). Nor is it clear that consumer response would be adequate to drive changes in production practices, even if the labeling challenges could be overcome. One area in which labeling may prove effective is in animal agriculture, where the tracking of an individual unit from producer to final consumer is more straightforward.

4.9 Options for Managing Nutrients, Co-benefits, and Consequences

4.9.1 Agricultural Drainage

The Integrated Assessment reports identified several research needs related to agricultural drainage. Brezonik et al. (1999) emphasized the importance of agricultural drainage in nutrient transport from cropland and identified increased spacing of subsurface drainage tile and controlling water table levels (controlled drainage) among those practices that could potentially reduce nitrate losses from cropland. Mitsch et al. (1999) noted that controlled drainage was not widely practiced in US Corn Belt and that most of the research on controlled drainage had been conducted in more southern climates.

4.9.1.1 Alternative Drainage System Design and Management

Relatively few field studies have addressed the effects of subsurface drain depth and spacing on N losses from cropland. Overall, results suggest a trend of decreased subsurface flow and decreased N loss at wider tile spacing or decreased tile depth. Reported reductions in nitrate export are primarily due to reductions in the volume of flow rather than reductions in nitrate concentration. Drain flows and N loss can be affected by both drain spacing and depth (Hoffman et al., 2004; Kladivko et al., 2004; Skaggs et al., 2003, 2005), and use of drainage intensity (Skaggs et al., 2005) normalizes some of the variability in results of drainage spacing studies. Drainage intensity increases with deeper tile depths and closer tile spacing. Research suggests that reducing drainage intensity by either shallower tile depth or wider tile spacing will reduce subsurface flow and nitrate loss. However, adjustments in tile spacing and depth are only possible when drainage systems are being installed, and the Corn Belt is already extensively drained. As these systems are replaced, repaired, and upgraded over the next few decades, there will be opportunities to consider alternative drainage designs to minimize nutrient losses. In the meantime, there may be opportunities to achieve similar benefits by retrofitting existing drainage systems with control structures that allow some management of subsurface drainage.

Drainage management (controlled drainage) is currently an area of active research and development (http://extension.osu.edu/~usdasdru/ADMS/ADMSindex.htm). Research suggests that drainage management could reduce nitrate transport from drained fields by 30% for regions where appreciable drainage occurs in the fall and winter (Cooke et. al., 2008). Although water table management could potentially alter nitrification and denitrification reactions, reported reductions in nitrate export with controlled drainage are primarily due to reductions in the volume of flow rather than reductions in nitrate concentration. Some uncertainty arises from difficulties in closing water balances (and therefore N balances) in field studies, and an unknown amount of subsurface flow reduction could be due to lateral seepage and/or increased surface runoff (Cooke et al., 2008). Simulation studies predict increased surface runoff when higher water tables are maintained using controlled drainage (Singh and Helmers, 2006; Skaggs et al., 1995), suggesting a potential tradeoff between reduced subsurface drainage and increased surface runoff. Although raising the water table can decrease the volume of infiltrating water entering drainage tile, higher water tables can also increase surface runoff resulting in increased erosion and loss of particulate contaminants such as soil bound phosphorous.

Controlled drainage requires relatively flat and uniform topography, and slopes of less than 0.5 or 1% are recommended (Cooke et al., 2008; Frankenberger et al., 2006). Concerns for erosion and surface runoff increase with increasing slope, and slopes greater than 0.5–1% can require an impractical number of control structures. There has been speculation that new technologies could make the practice economically feasible at slopes of 2% or more, but this would raise even greater concerns over surface runoff. Although tile drainage is widespread throughout the Corn Belt, it is not clear what portion of this tile drainage can be retrofitted with structures for controlled drainage. A first approximation might be an estimate of the fraction of tile-drained lands with slopes less than 0.5–1%, but this approach requires higher resolution topography than is generally available in the Corn Belt. These estimates are available for a few large drainage districts in north central Iowa for which very high resolution topography were developed. Although 50–75% of the cropland in these drainage districts is tile drained, only about 10% has a slope less than 1% and only about 3% has a slope less than 0.5% (Matt Helmers, Iowa State University, Ag Drainage Website, http://www3.abe.iastate.edu/agdrainage). These results suggest that controlled drainage may be applicable to a relatively small fraction of tile-drained land in Iowa, but this may not be representative of other regions of the Corn Belt. Based on STATSGO soils data, Illinois, Indiana, and Ohio may have twice as much cropland suitable for controlled drainage as Iowa (Dan Jaynes, National Soil Tilth Lab, Ames, IA). High-resolution topography could provide a much better basis for this assessment.

4.9.1.2 Bioreactors

Denitrification bioreactors have been installed in the field as treatment systems for tile drain effluent (Van Driel et al., 2006) and as denitrification walls (a trench filled with carbonaceous material to intercept subsurface flow) (Robertson et al., 2000; Schipper et al., 2004, 2005; Schipper and Vojvodic-Vukovic, 1998, 2001). Bioreactors on tile drains are typically bypassed during high flows and "are most usefully applied in the treatment of baseflows rather than peak flows." Current knowledge indicates that denitrification walls are effective for at least 5–7 years with little or no loss of nitrate removal capacity (Robertson et al., 2000; Schipper and Vojvodic-Vukovic, 2001). A variety of materials such as corn stalks, wood chips, and sawdust are potential organic amendments to enhance denitrification in bioreactors. Continued research is needed to determine whether denitrification bioreactors could be installed around lateral tile drain lines and whether this would be technically and economically feasible. Future redesign of tile drain systems may include integrated denitrification enhancements around tile lines and at the outlets of smaller tile lines.