A Landscape Approach to Agricultural Nonpoint Source Pollution
Most programs used to control agricultural nonpoint source pollution focus on in-field best-management practices, but there is a growing interest in the use of off-field control techniques (Clausen and Meals, 1989). The most commonly used off-field control practices are vegetative filter strips and riparian buffer zones. Vegetative filter strips are narrow strips of managed grassland situated directly adjacent to agricultural fields (Dillaha et al., 1989b). Riparian buffer zones are usually areas of natural forest vegetation situated between cropped areas and streams (Lowrance et al., 1984a). Interest in both of these practices has increased dramatically in recent years.
A focus on off-field controls changes the unit of analysis for physical and social science questions from the field to the landscape scale. Whereas nonpoint source pollution best-management practices and most agricultural policy instruments are directed toward activities that occur within agricultural fields, analysis of off-field nonpoint source pollution controls requires consideration of the interaction of crop fields with adjacent managed or unmanaged ecosystems and how those interactions affect water quality over an area larger than a specific crop field.
The emerging field of landscape ecology provides a conceptual basis for landscape analysis of agricultural nonpoint source pollutant problems (Forman and Godron, 1986). Landscape analysis considers the spatial juxtaposition and dynamic interactions between agricultural and adjacent ecosystems in the context of the water quality of the landscape
as a single unit, for example, a watershed or groundwater recharge zone. This chapter describes the conceptual and practical bases for landscape analysis of agricultural nonpoint source pollution and discusses options for and obstacles to implementing this approach.
NONPOINT SOURCE POLLUTANT ATTENUATION MECHANISMS
The basis for a landscape approach to agricultural nonpoint source pollutants is the use of particular areas as sinks for pollutants moving off agricultural fields. These sinks must be capable of intercepting the pollutants in either surface water runoffs and/or groundwater flows (Figure 12-1) and must support one or more of the processes that remove pollutants. These processes include plant and microbial uptake of nutrients and trace metals, microbial degradation of organic compounds, sediment trapping, microbial conversion of nitrate into nitrogen gas, and physical and chemical adsorption of metals and organic compounds.
Planning, implementation, and evaluation of the use of landscape sinks for nonpoint source pollutant control must consider two key factors: (1) the capability of a particular area to intercept surface water- and/or groundwater-borne pollutants and (2) the activities of different pollutant removal processes. Analysis of these factors is relevant in several contexts including field-scale development of specific off-field control practices such as grass vegetative filter strips, farm-scale analysis of where off-field controls should be established, and watershed-scale analysis of the effectiveness of existing sink areas such as riparian areas or wetlands.
Extensive research has demonstrated that grass vegetative filter strips have high sediment trapping efficiencies if the flow is shallow and the vegetative filter strips are not filled with sediment. Trapping efficiency has been found to decrease dramatically at high runoff rates (Barfield et al., 1979; Schwer and Clausen, 1989). Several short-term experimental studies have reported the effectiveness of grass filter strips in reducing the amounts of sediments in runoffs (Dillaha et al., 1988; Magette et al., 1987; Young et al., 1980). These short-term studies found that grass filter strips are effective at removing sediments and sediment-bound pollutants at trapping efficiencies that exceed 50 percent if the flow is shallow (shallow flow refers to water flowing in sheets across a field or grass filter strip). Grass filter strip
plots with concentrated flows (concentrated flow refers to water flowing in channels across fields or grass filter strips) similar to those expected under field conditions were reported to be much less effective than the experimental plots with shallow flows used in most vegetative filter strip research (Dillaha et al., 1989b).
Dillaha and colleagues (1989a) studied existing grass filter strips on 18 farms in Virginia and found them to be extremely variable in their effectiveness at removing sediments. Most grass filter strips in hilly areas were ineffective because runoff usually crossed the strip as a concentrated flow. In flatter regions, grass filter strips were more effective because slopes were more uniform and more runoff entered the strip as a shallow flow. Several 1- to 3-year-old vegetative filter strips were observed to have trapped so much sediment that they produced more sediments than adjacent upland fields. In these cases, runoff flowed parallel to the vegetative filter strip until it reached a low point, where it crossed the vegetative filter strip as a concentrated flow. These vegetative filter strips needed maintenance to regain their sediment-trapping abilities.
Several models have been developed for vegetative filter strip design
and evaluation. GRASSF is an event-based model developed for designing vegetative filter strips with respect to sediment removal (Barfield et al., 1979; Hayes et al., 1979). The model was evaluated by using plot data from multiple events; and predicted values were in good agreement with observed values, even though the model does not consider deposition in the water ponded upslope of the grass strip, which is where most deposition occurs in grass filter strips (Hayes and Hairston, 1983). By neglecting the deposition of sediment in the upslope ponded water, a model tends to underpredict the trapping capability of the strip. GRAPH (Lee et al., 1989), a derivative of the GRASSF model, simulates nutrient transport in vegetative filter strips. GRAPH considers the effects of advection (transport by water flows) as well as adsorption and desorption processes. The model also considers the effects of changes in sediment size distribution and the chemical transport processes in vegetative filter strips.
The chemicals, runoff, and erosion from agricultural management systems (CREAMS) model can also be used to evaluate the trapping of sediment by grass filter strips from overland and concentrated flow (Williams and Nicks, 1988) and from deposition where the upper edge of a vegetative filter strip has redirected runoff from overland to concentrated flow. Flanagan et al. (1989) derived simplified equations to design grass filter strips and compared estimates from their equations and from the CREAMS model with observed data. In both cases, they found good agreement. An important advantage of the CREAMS model is the ability to compute the trapping of sediment in the ponded water created by a filter strip placed across an area where concentrated flow is occurring. RUSLE, the revised universal soil loss equation, also computes the effect of grass strips on erosion and sediment yield.
If grass filter strips are so narrow that the strips completely fill with deposited sediment, CREAMS overestimates the trapping of sediment because the model does not account for sediment deposited in the grass strip. However, most filter strips used where overflow is occurring are usually wide enough that the width of the grass strip is not a factor in the amount of sediment that is trapped. The critical factor is how well the model reflects the reduced transport capacity that is created when the grass strip reduces the velocity of the runoff water.
Riparian forest buffer zones have the ability to absorb as many or more sediments than grass filter strips (Cooper et al., 1987; Lowrance et al., 1988). As with grass filter strips, concentrated flow, sediment accumulation, and buffer zone disturbances can reduce the sediment-trapping ability or cause the accumulated sediments to be released from riparian forest buffer zones.
Plants can effectively take up nutrient elements such as nitrate and phosphate and can also absorb many heavy metals such as lead, cadmium, copper, and zinc. The importance of plants as a nonpoint source pollutant sink depends on their ability to absorb the nutrients moving in either surface or subsurface water flows. Pollutants moving in groundwater are accessible to plants only when the water table is high in the soil profile, such as in wetlands. In these situations, plants can be an important sink for groundwater-borne pollutants (Ehrenfield, 1987). Plant roots may not be able to take up nutrients if water flow is too rapid. The nutrients in water moving across the soil surface in a concentrated flow or percolating rapidly through soil macropores may not be susceptible to plant uptake. Large rainfall events—which often transport a very high percentage of nonpoint source pollutants—readily produce concentrated surface flows and macropore-dominated percolation.
Plant species differ in their abilities to take up different pollutants and in the rate at which uptake occurs. There is ample opportunity to manage the plant community in vegetative filter strips, such as through selection of appropriate grass species and harvesting, which removes the accumulated pollutants (Brown and Thomas, 1978). In riparian forests, management of the plant community is more difficult, but it can be accomplished through selective cutting and replanting. Actively managing riparian forests could substantially increase their effectiveness in preventing water pollution. Selection of the optimal grass species for vegetative filter strips and development of management plans for riparian forests are major topics of research at several locations in the United States.
Although some plants are able to take up pollutants, they may not be reliable long-term sinks for pollutants in the landscape. Uptake of pollutants by plants necessarily declines or stops during the winter, which is often when most movement of surface water and groundwater from upland areas toward water bodies occurs. The use of a mixture of plant materials (for example, cool and warm season grasses) can extend the period of plant activity, but in many areas, a significant dormant season is inevitable.
A major concern with uptake of pollutants by plants is that the nutrients trapped in plant tissues can later be released back into the soil
solution as these tissues decompose. Storage of pollutants in the structural tissues of trees represents a relatively long-term attenuation, but it still does not result in removal of pollutants from the ecosystem. The nutrients released from decomposing plant tissues may be attenuated by microbial, physical, or chemical pollutant mechanisms in surface soils. Release of pollutants by decomposition may be beneficial if the vegetation removed the nutrients from groundwater, where the potential for attenuation is often quite low.
In addition to seasonal dynamics, longer-term temporal dynamics affect the ability of plants to act as pollutant sinks. Over time, plants in a riparian buffer zone or vegetative filter strip can become ''saturated" in their ability to absorb nutrients, resulting in a decline in their absorption capacity (Aber et al., 1989). Although most plants show marked growth responses to nitrogen and phosphorus, after a period of high input, other nutrients become limiting and nitrogen and phosphorus are no longer absorbed. Unless there is some type of nutrient removal through harvesting, saturation will likely occur at some time.
Several microbial processes can serve to attenuate nonpoint source pollutants in different components of the landscape. Like plants, microbes can take up or "immobilize" nutrients and metals in their tissues to support growth. As in plants, this immobilization is reversible, and the accumulated pollutants can be released upon microbial death and decomposition. Since microbial turnover is quite rapid (Paul and Clark, 1989), immobilization in microbes is likely not a significant long-term sink for pollutants in the landscape.
Microorganisms have the ability to degrade organic compounds such as pesticides and many pesticides are highly susceptible to microbial degradation. Landscape areas that support high levels of organic matter and microbial populations (for example, vegetative filter strips or riparian forests and wetlands) may be important sites for degradation of the compounds that leave agricultural fields.
Denitrification is a microbial process that converts nitrate, the most common form of nitrogen leaving agricultural fields, into nitrogen gas. This process occurs under anaerobic conditions and has been found to be an important nitrate attenuation mechanism in the surface soils of wet riparian forests (Ambus and Lowrance, 1991; Jacobs and Gilliam,
1985; Lowrance et al., 1984b; Peterjohn and Correll, 1984) and, to a lesser extent, in vegetative filter strips (Groffman et al., 1991).
Investigators have expressed great interest in assessing the potential for denitrification in groundwater. Although some studies have found significant potential for denitrification in groundwater (Slater and Capone, 1987; Smith and Duff, 1988; Trudell et al., 1986), others have found little or no denitrification activity (Parkin and Meisinger, 1989). A key question relates to the availability of carbon to support microbial activity in the subsurface (Obenhuber and Lowrance, 1991). Groundwater-borne nitrate beneath wet riparian zones may be subject to attenuation more than the nitrate in groundwater beneath more upland areas is, since water tables are high in wetlands, allowing groundwater-borne nitrates to interact with the biologically active zone of the soil with high levels of carbon.
A major mechanism of nitrogen loss in the surface soils of riparian wetlands may be denitrification of the nitrate removed from groundwater by plants (Groffman et al., 1992; Lowrance, 1992b). Uptake of nitrate
from groundwater by plants can lead to increases in the amount of nitrogen in plant litter (Lowrance et al., 1984a). High nitrogen levels in plant litter can lead to increases in nitrogen mineralization and the amount of available nitrate in surface soils. This nitrate is then subject to denitrification.
Several physical and chemical adsorption processes in soil attenuate pollutants in vegetative filter strips, riparian buffer zones, and other landscape sink areas. These processes include the attraction of cations to negatively charged sites on clay and organic matter, chemical binding of organic compounds on clay and organic matter, and physical fixation of the ions within clay minerals. Adsorption processes are controlled by the amount of clay and organic matter in the soil.
Two key factors control the effectiveness of landscape-scale nonpoint source pollutant sinks: (1) the capability of a particular area to intercept surface water- or groundwater-borne pollutants and (2) the ability of a particular area to support different pollutant removal processes. By considering the discussion of attenuation mechanisms presented above, it becomes possible to identify those key landscape components that may be effective as nutrient sinks.
Riparian forests, especially those dominated by wetlands, are major potential pollutant sinks. Because of their physical position in the landscape, they can intercept a high percentage of the surface runoff and groundwater flow that moves from upland areas before it reaches streams. Wetland areas have a unique ability to interact with groundwater because the water tables in these areas are close to the soil surface, allowing for the interaction of roots and microorganisms with groundwater-borne pollutants. In addition, riparian areas have the potential to attenuate many pollutants. In most naturally vegetated areas, plant uptake is vigorous and soil organic matter levels are high, which increases the potential for microbial attenuation and chemical adsorption processes. For these reasons, riparian areas have justifiably been the focus of much research on landscape-scale sinks of agricultural nonpoint source pollutants.
A considerable body of evidence confirms that riparian forests can be effective sinks for agriculturally derived nonpoint source pollutants. Several studies have found that strips of riparian forest vegetation are
important for maintaining stream water quality in areas where uplands are formed intensively (Groffman et al., 1992; Jacobs and Gilliam, 1985; Karr and Schlosser, 1978; Lowrance et al., 1984b; Peterjohn and Correll, 1984; Simmons et al., 1992). These studies have included comparisons of watersheds with and without riparian vegetation as well as process-level studies of pollutant removal from riparian soils and vegetation.
Investigators have several uncertainties about the performance of riparian forests as nonpoint source pollutant filters. Regional variations in their effectiveness may be important. Although a relatively large body of research suggests that these areas are effective in the Southeast (Jacobs and Gilliam, 1985; Lowrance et al., 1984b; Peterjohn and Correll, 1984), fewer data have been collected for other parts of the United States. Several studies have suggested that riparian zones are effective in the Corn Belt (Huggins et al., 1990; Kovacic et al., 1991; Schlosser and Karr, 1978) and Northeast (Simmons et al., 1992), but more data for these regions need to be collected.
A major concern with riparian zones is their long-term effectiveness. Over time, the ability of these areas to absorb sediments and nutrients may decline. Sites for sediment trapping may become filled, and the capacity of plants and microorganisms to take up nutrients may become saturated. The accumulated pollutants may be released if riparian areas are disturbed by logging, fire, windstroms, or flooding. Investigators now need to research riparian zones that have been absorbing pollutants from uplands for many years.
Given what investigators know about pollutant attenuation processes, predicting the value of upland grass filter strips as pollutant sinks is more problematic than predicting that of riparian forests. The plant communities in filter strips and riparian areas potentially could be managed intensively, but plants may be the least reliable pollutant attenuation mechanism. Upland areas generally have relatively low organic matter levels, so microbial sinks in upland areas are usually less vigorous than they are in riparian zones. There is little potential for upland grass filter strips to interact with groundwater. The main advantage of vegetative filter strips is their proximity to agricultural fields and the potential for aggressive management of the plant community and the physical condition of the vegetative filter strip.
IMPLEMENTING A LANDSCAPE APPROACH
Making full use of off-field nonpoint source pollutant control mechanisms presents distinct challenges to both physical and social scientists. For example, determining which areas must be placed into vegetative
filter strips or riparian buffer zones and how these areas should be managed is a challenge to physical scientists. Development of policy instruments that can be used to implement these changes is a challenge to social scientists. Evaluation of the performance of new policies and practices is a challenge to both groups.
For field-scale applications, the U.S. Forest Service of the U.S. Department of Agriculture (USDA) has produced specific guidelines for riparian buffer zone planning, design, and maintenance (Welsch, 1991). The guidelines call for three zones on which different management practices are used. Zone 1, nearest the stream, consists of undisturbed forest, mostly for stream habitat protection. Zone 2 consists of managed-forest and is the site of most pollutant removal activity. Zone 3 is a grassed area, with management that differs with upland land use. If the upland land use is row crops, zone 3 may contain water bars (small dams placed to slow and redirect runoff), spreaders, or other devices to convert concentrated runoff into a dispersed flow. If the upland land use is pasture, management of zone 3 is less intense, with controlled grazing permitted under certain conditions.
At the landscape scale, determination of areas to be used for vegetative filter strips or riparian buffer zones must be based on scientific evaluations of how effective these areas are likely to be for pollutant control. Moreover, unless off-field control practices are instituted in a systematic way across a given landscape unit, overall landscape water quality improvements will be minimal (Phillips, 1989). The challenge, then, is to identify key areas that need to be managed for their pollutant control value and to develop policy instruments to facilitate land use changes.
Approaches such as USDA's Conservation Reserve Program will have to be substantially restructured to increase the production of riparian areas. Although land suitable for buffer zones is eligible for enrollment in the Conservation Reserve Program, most producers prefer to enroll whole fields. Figure 12-2 points out both the problems with land retirement on a field-by-field basis and the potential for more effective use of programs such as the Conservation Reserve Program. Programs like the Conservation Reserve Program will not produce the necessary land use changes unless water quality problems and riparian areas are clearly stated priorities (Dillaha et al., 1989a).
Conversion of different types of land into different types of pollutant sinks will require a diverse set of policy instruments. Implementation of field-scale vegetative filter strips is the most conceptually straight forward approach. For example, if investigators determine that all fields that exceed a certain length or slope criterion must have a vegetative
filter strip at the field's edge, then voluntary or mandatory instruments can be devised to motivate the land manager to install the vegetative filter strip. This type of implementation may require the loss of productive land.
Implementation of riparian buffer zones on a landscape scale will be more challenging. Since these areas are necessarily linked to certain water bodies, designation of the areas to be maintained as buffer zones will be based on watershed delineations that likely are not consistent with land ownership patterns or governmental jurisdictions. A major problem will arise from the fact that not all lands in a watershed will have riparian forests. It is important to consider situations in which one land user owns lands with riparian forest that filters pollutants that run off the lands of many upland land users. A system could be worked out whereby producers in areas without riparian compensate producers that establish or protect riparian areas for the loss of land necessary to meet watershed-wide water quality goals.