Postcommercialization Testing and Monitoring for Environmental Effects of Transgenic Plants
As indicated in Chapters 2 and 5, short-term experiments and general characterization of plant traits may not pick up all environmental effects of transgenic crop plants. It is therefore important to conduct postcommercialization testing to determine if the precommercialization testing protocols adequately assessed risks (i.e., validation of precommercialization decisions). It also is important to set up long-term, postcommercialization monitoring programs to record trends in predicted effects, and to detect effects that were not predicted by precommercialization testing. Taken to an extreme, postcommercialization testing and monitoring could be prohibitively expensive, and if not carefully conceived, could lead to collection of uninterpretable data. This chapter explains the logic behind validation and monitoring programs, describes the status of current environmental monitoring programs, and makes suggestions for the general structuring of validation and monitoring programs. It is beyond the scope of the committee to offer suggestions for development of specific programs for postcommercialization testing and monitoring.
Postcommercialization testing or validation programs are an essential part of any quality control program. Whether it be an automobile part or a new hotel service, the producer does testing after commercialization to make sure that the precommercialization quality control program was effective. Precommercialization testing for environmental effects of new crop varieties is a new endeavor, so follow-up validation testing seems essential.
Monitoring of past and present status or trends in quantity and quality of a resource has often proven essential in making decisions regarding the management of that resource. As a very familiar example, a storeowner makes ordering decisions based on monitoring of inventory, and on factors such as past experience with consumer behavior, upcoming holidays, and time for the delivery of goods. Management abilities and predictive power for anticipating future needs improve with experience (or as data accumulate). Over time, changes in consumer preferences, and changes associated with technological advances in the delivery systems affect how storeowners manage their inventories. Models used to forecast climate, weather, or performance of stocks, are based on the sophisticated use of data on key parameters accumulated over time. Models improve as data series become longer, however the intrinsic degrees of uncertainty in these systems will never allow perfect predictions so monitoring will always be needed. In the economic world, productivity of workers, unemployment rates, stock indices, mortgage rates, and other variables are monitored and this information is used to predict and manage the local, regional, national and international economy. In contrast, we do not have reliable statistics or long-term monitoring indicators for most of the nations’ biological resources (NRC 2000a).
The charge to this committee as it pertains to environmental monitoring is to (1) evaluate the need for and approaches to environmental monitoring and validation processes and, if deemed necessary, to include recommendations for postcommercialization monitoring of transgenic plants and (2) provide guidance on the assessment of non-target effects, appropriate tests for environmental evaluation, and assessments of cumulative effects on agricultural and nonagricultural environments for transgenic plants.
THEORETICAL JUSTIFICATION FOR MONITORING AND VALIDATION AFTER COMMERCIALIZATION OF TRANSGENIC CROPS
Precommercial risk analysis has several inherent weaknesses. In general, small-scale precommercialization field experiments are not sensitive enough to detect anything but large effects. For any such experiment there will be some limit to what can be detected, and this limit will be rather high because the natural variability from one experimental replication to another is large. For example, in estimating the yield of a corn variety—a commercially important agronomic trait—it is necessary to run several hundred-yield trials to detect significant increases in yield. In the U.S. Corn Belt, corn yields may be greater than 150 bushels per acre. A variety that yields five additional bushels per acre is a significantly more
productive one, which at today’s depressed corn price of about $2 per bushel on a 500-acre corn farm would provide an extra $5,000 in gross income. This is only a 3% increase over previous varieties and yet requires hundreds of trials to demonstrate conclusively.
There is no environmental risk of any agricultural technology that will be tested in hundreds of experimental field trials prior to commercialization. Small-scale field trials will readily detect order-of-magnitude differences in an ecological effect, but smaller differences will be difficult to document. To illustrate this problem, a prospective power analysis (Oehlert 2000) can be conducted on published insect density data to model detection of potential non-target effects of transgenic crops. The population density of a pest insect—the European corn borer—was measured in three environments replicated four times each (Andow and Ostlie 1990). To detect statistically a 10% difference in population density among the experimental environments given the observed variation in density among replicates, it would be necessary to examine 134 replicates of the experiment. Thus, unless the effect of the transgenic crop were considerably larger (a twofold difference in population density could be detected with 10 replications), a small-scale field experiment is unlikely to detect 10% population reductions in non-target species. Yet ecological effects of 10% can be significant. For an endangered species, a 10% difference in survival (note that the committee is speaking of “survival”—not “density”) could mean the difference between recovery and extinction. Thus, a small experimental effect does not imply a small ecological effect, and the magnitude of a significant ecological effect may be no different from that of a significant commercial effect.
Postcommercialization testing is a typical component of most quality control programs and is appropriate for determining whether the precommercialization tests of transgenic plants were effective. Precommercialization testing of transgenic plants has involved the adoption of testing protocols used for other types of products such as synthetic pesticides. Because it is not clear that these protocols are completely appropriate for transgenic plants, postcommercialization validation testing is especially important. If evidence is collected over time that confirms the accuracy of precommercialization testing protocols, the need for validation testing would decrease, but due to the uniqueness of new products its utility is unlikely to disappear.
Even if the precommercialization risk analysis process is validated so it effectively identified and mitigated all order-of-magnitude ecological effects, low-probability events and low-magnitude effects would likely escape detection and management. As the transgenic crop is commercialized to larger spatial and temporal scales, it may become possible to ob-
serve smaller and less frequent ecological risks as long as there is a system of monitoring to look for them.
One type of potential adverse ecological effects of transgenic crops that may be difficult to detect is their potential for invasion of neighboring ecosystems. Short-term, spatially limited field trials are poor predictors of environmental impacts of invasions (Kareiva et al. 1996). Smallscale field trials conducted as part of a permit or petition for deregulation of a transgenic crop will ultimately have little predictive power regarding potential ecosystem effects or potential for invasiveness. In an exercise to explore the predictive power of a large dataset comparing potential invasiveness of transgenic and nontransgenic canola over three years in the United Kingdom (Crawley et al. 1993), Kareiva et al. (1996) found that results varied depending on how many sites and years were incorporated into the analyses. Years were more important than sites, but the magnitude of errors often exceeded 100%, making use of long-term data imperative to increase predictive power. Kareiva et al. (1996) conclude, “we have so little faith in models and short-term experiments regarding predictions about invasions, that we advocate extensive monitoring of any introduced [transgenic crop] with any ecologically relevant traits (such as disease resistance, herbivore tolerance, and so forth).”
A second reason that ecological monitoring is needed after commercialization is that ecosystems are complex. As noted above, this complexity stems from interannual variations and indirect effects. Because laboratory and small-scale field experiments do not adequately replicate all the interactions that occur in an ecosystem, the only way to observe the full range of ecological effects of a transgenic crop is to observe it in actual ecosystems. Some of these effects cannot be predicted beforehand, so ecological monitoring will be necessary to detect any adverse ecological effects.
Social science provides an additional rationale for postcommercialization monitoring. From a social-psychological perspective, a compelling reason to monitor is that the public wants it. Rigorous monitoring assures the public that their concerns are being addressed seriously and reassures the public that scientists are being careful to keep the risks low. While it would be irresponsible to develop a large monitoring program for the sole purpose of reassuring the public, ignoring public concerns is also irresponsible.
Finding 6.1: There are several compelling theoretical arguments for the need for ecological monitoring and validation after commercialization of genetically modified organisms. These needs range from ecological to social and from the specific to the general. It is likely
that different kinds of monitoring will be needed to meet these various needs.
POSTCOMMERCIALIZATION VALIDATION TESTING
As indicated above, postcommercialization testing should be used to determine the effectiveness of the precommercialization risk assessments. Each of the four major categories of risk associated with transgenic plants that are examined before commercialization can be reexamined after commercialization (these are risks associated with movement of transgenes, impacts of the whole plant through escape and through impacts on agricultural practices, non-target effects, and resistance evolution). The specific approaches for validation will depend on the risk examined, and the intensity of testing may depend on the extent of the potential risk. Validation testing must always be based on testing specific hypotheses related to the accuracy and adequacy of precommercialization testing.
In some past cases, precommercialization tests have identified potential risks and restrictions that have been imposed on commercialization to limit these risks. One example is the risk of insect evolving resistance to Bt-producing transgenic crops. In this case precommercialization testing demonstrated that pest species had the genetic capacity to evolve resistance to Bt toxins (Tabashnik 1994), and in some cases the frequency of specific resistance genes was estimated (for example, Bt; Bentur et al. 2000). During the precommercialization period a number of biological attributes of target pests were estimated, and these estimates were used to develop simulation models that projected the dynamics of resistance evolution in the pest (Alstad and Andow 1995). Results from the simulation models and other sources were then used to determine conditions for commercialization that would limit the risks of resistance evolution (EPA 1995). At the time of commercialization there were still uncertainties regarding some of the key parameters used in the simulation models, and incorrect parameter estimates could result in inappropriately strict or lax restrictions on commercialization. Two types of postcommercialization testing could improve the match between risks and restrictions. First, further research to better estimate biological parameters of the system could increase the rigor of model predictions, and second, monitoring of changes in the frequency of resistance genes in the pest could be used to determine if evolution was proceeding at an acceptably slow rate. If information from parameter estimates or monitoring indicated that the precommercialization decisions about restrictions were inappropriate, adjustments to the conditions could be made.
In the case of risks due to non-target effects, there are limits to precommercialization tests because they are restricted to testing a small set of
organisms over a short time period, on a small spatial scale (Chapter 5). Once the transgenic crop is marketed there is an opportunity for largescale, multi-year testing to determine the adequacy of the precommercialization testing. Comparisons of the population dynamics of non-target organisms in large (about 100 acre) paired fields (transgenic versus near isolines) and surrounding natural habitats in a number of locations would be one approach to such validation testing. If such paired fields were repeatedly planted to the same crop variety for a number of years it would be possible to examine population trends in non-target organisms that would be impossible to examine at a smaller spatial and temporal scale. The committee recognizes that even the 100-acre scale would be insufficient for examining organisms that move long distances. For such organisms it would be necessary to utilize epidemiological approaches that examined relationships between the intensity of planting of a specific crop variety in a locality, and the population dynamics of these non-target organisms. At this level, a gray area develops between what might be considered a postcommercialization validation test and a monitoring program. While this committee treats validation testing and monitoring separately in this report because of the distinct and complex type of infrastructure needed for monitoring, it is important to recognize that these two approaches are used to accomplish broadly similar goals.
Recommendation 6.1: Postcommercialization validation testing should be used to assess the adequacy of precommercialization environmental testing. This validation testing must always involve testing specific hypotheses related to the accuracy and adequacy of precommercialization testing.
Recommendation 6.2: Postcommercialization validation testing should be conducted at spatial scales appropriate to evaluate environmental changes in both agricultural and more natural ecosystems.
A funding mechanism for postcommercialization testing will need to be established. It would be preferable to have such testing involve public sector scientists to help alleviate real or perceived conflicts of interest. Although validation testing is costly, the recent postcommercialization testing for impacts of Bt corn on monarch butterflies (Chapter 2) demonstrates that such testing is feasible.
The USDA Biotechnology Risk Assessment Research Grants Program is small ($1.9 million for fiscal year 2000) (USDA 2001a) and is directed primarily toward small-scale hazard identification. The objectives of this program expanded recently to include risk issues associated with commercialization. The USDA’s Initiative for Future Agriculture and Food
Systems (IFAFS) program has established a section concentrating on Biotechnology Risk Management ($3.4 million for fiscal year 2000, of which approximately $2.75 million is dedicated to Ecological Risk Management; USDA 2001b), but this may not be a recurring grant program within USDA. The budgets for these programs are small and the needs are so diverse, that many deserving issues are not funded. Validation testing grant proposals could fall between the main objectives of both of these programs.
Present public research programs, such as the Biotechnology Risk Assessment and Risk Management Programs will need to be expanded substantially to meet this need for postcommercialization testing. These grants may need to last longer than the average grant since many effects may not be seen until the crop has been grown for several years at the commercialized scale. In addition, these programs need to identify, clearly, program objectives that encourage research related to long-term monitoring. Such postcommercialization validation testing with well-crafted controls can provide solid experimental research data to form the scientific basis for decisions. The peer review process associated with such grants plays an important role in ensuring proper design of experiments. Some validation testing requires that the same location be examined for a long period of time and this may require specific targeting.
Recommendation 6.3: Present public research programs, such as in Biotechnology Risk Assessment and Risk Management, will need to be expanded substantially.
STATUS OF LONG-TERM ENVIRONMENTAL MONITORING IN THE UNITED STATES
A multitude of individuals and private, local, state, and federal agencies manage the country’s abiotic (soil, water, air, nutrients, etc.) and biotic (organismal diversity) resources. Effective resource management relies on accurate and timely information on the extent, condition, and productivity of these resources and how they respond to management. Although historic calls for monitoring in natural areas such as in the national park system were made as early as the 1930s (Wright et al. 1933, Wright and Thompson 1935), it took until 1990 before the National Park Service initiated a natural resource inventory and monitoring program in select parks to aid in fulfilling its mission (Woodward et al. 1999). Many individuals and agencies are involved in long-term monitoring and research. Many formal censuses of animal species are regional in scale and short term in duration. Annual waterfowl surveys conducted to set hunting seasons and bag limits are a notable exception. Waterfowl popula-
tions were declining rapidly at the end of the nineteenth century and for their protection the Migratory Bird Treaty Act became law in 1918 (Anderson and Henny 1972). Over time the methods used to estimate annual production and harvest became more sophisticated and now include aerial surveys of breeding and wintering grounds, banding of individuals, reporting of harvests, and use of computer modeling (Shaeffer and Malecki 1996, Shaeffer 1998). Data collected in the summer are immediately used to develop harvest regulations for the fall, and this approach has proven extremely successful for managing waterfowl.
Most state natural resource, natural heritage, and fish and game departments conduct censuses of game species (fish, birds, and mammals). All officially endangered species are periodically censused. Other examples of long-term datasets include the Christmas Bird Count, the North American Breeding Bird Survey, and the Annual 4th of July Butterfly Count. However, a review of long-term programs (Woodward et al. 1999) found that vague objectives, a piecemeal selection of indicators, and poor linkages between monitoring projects and the decision-making process plague such programs. In addition, lack of consistent standards for data collection across agencies makes it difficult to increase statistical power by combining data.
Monitoring programs associated with agriculture have a more extensive history. One of the oldest programs is the National Agricultural Statistics Service (NASS), which has collected agricultural statistics for over 160 years. NASS’s charge has evolved over time, and today it collects a substantial and diverse set of agricultural data including crop acreage, farm expenditures, harvest yields, livestock inventories, and environmental data such as land usage and chemical use on those lands.
The Dust Bowl of the 1930s and loss of valuable land to erosion triggered the beginning of coordinated data-gathering efforts on the status of agricultural lands in the United States by the U.S. Department of Agriculture’s Soil Conservation Service (now the Natural Resources Conservation Service, NRCS). Over time the inventory has expanded from the original purpose of monitoring soil erosion in the 1930s to include monitoring of clean water, prime farmland, wetlands, wildlife habitat, and environmental effects of agriculture. At present the NRCS, (through the Rural Development Act of 1972, the Soil and Water Resources Conservation Act of 1977, and other supporting legislation) is charged with assessing the status, condition, and trends of soils, water, and related resources on nonfederal lands in intervals not to exceed five years (Nusser and Goebel 1997). These data now form part of the National Resources Inventory (NRI) program. A constant feature of the sample surveys is the use of numerous measures and the reliance on detection over time through repeated visits to the same points. The longitudinal survey encompasses a
network of 800,000 sampling points in primary sampling units (PSUs) across the United States. Each PSU is a 64.75 ha (160 acres) in area, and 0.8 km (0.5 miles) at each side (some PSU sizes vary; see Nusser and Goebel 1997 for details). Within each PSU, three sampling points are selected according to a restricted randomization procedure. For the comprehensive five-year NRI survey, sampling rates across the country generally range from 2 to 6% of the land area, but sampling rates increase for special studies. Statistical techniques and data collection protocols have evolved as inventory goals have become broader and more sophisticated. In recent years the NRI has conducted special studies to investigate such topics as changes in wetlands, erosion rates, and changes in field practices. Multidisciplinary teams using remote sensing techniques, and geographic information systems are now used for data collection.
Long-term monitoring activities of the NRI, however, have not been able to provide sufficient or relevant data for effective natural resource and environmental assessments or management (Goebel 1998). NRI data do not provide information about the presence or abundance of any plant or animal species (other than crop species) and thus fall short of a comprehensive assessment of natural resources (see BOX 6.1).
Despite the information available from the NRI and NASS, there are substantial gaps in current monitoring data from agricultural systems in the United States (The Heinz Center 1999). These gaps are particularly evident with regard to many of the ecological interactions that are important for assessing the long-term productivity and sustainability of agricul
BOX 6.1 Information Currently Contained in the NRI
tural practices. For example, there are no consistent or comprehensive data available on crop losses to pests and disease or on pesticide resistance. There are little if any consistent or comprehensive data on soil qualities such as salinity, organic matter, and compaction. The same can be said about the status and trends of unmanaged pollinator populations that are vital to many crops. These gaps and many others will need to be addressed if monitoring of new agricultural practices and their impacts, such as the introduction and use of genetically modified crop varieties, are to be seriously considered. To study associations between the intensity of use of a transgenic plant and changes in biological indicators, monitoring will need to be conducted at a county or township level in order to gather statistically useful data.
Finding 6.2: Our ability to assess the impacts of large-scale planting of transgenic crops is hampered in part by the lack of baseline or comparative data on the environmental impacts of agriculture.
Finding 6.3: The data provided through the NRI or the NASS are not sufficiently detailed to allow an independent assessment of the environmental effects of transgenic plants. Data need to be available on at least the county or township level to become meaningful. More detailed information is preferable to be able to make inferences about trends in areas with or without commercialization of transgenic plants.
Long-term monitoring programs of federal agencies were often established in response to legislative mandates (Endangered Species Act, Clean Water Act, Biomonitoring of Environmental Status and Trends Program, and the Environmental Monitoring and Assessment Program), yet information about biological diversity or individual species is patchy at best. For example, in its latest strategic plan, the U.S. Forest Service (USDA 2000) has identified species inventory and monitoring as key components to its decision-making processes for natural resource management plans. The improvement and integration of information systems, data structures, and information management as well as the development of indicators and monitoring protocols are expected to lead to improved stewardship. A milestone for fiscal year 2006 is the establishment of measurable objectives and monitoring programs for populations, habitats, and/or ecological conditions for threatened and endangered species, other species for which there are viability concerns, and other management indicator species/focal species. Of the 19 species explicitly mentioned to be included in monitoring programs are eight bird, five fish, three mammal, two plant, and one frog species or subspecies. The strategic plan recognizes that it may take many years before trends resulting from manage-
ment strategies can be separated from population fluctuations caused by other influences.
Two informative examples of the power and usefulness of monitoring come from the need for sustainable management of fisheries and waterfowl species, which has resulted in the design of standardized long-term monitoring programs. The National Marine Fisheries Service spent some $28.8 million on ship time to collect fisheries data (excluding personnel and analyses costs), $3.9 million for recreational monitoring, $9.2 million on observer programs (with another $10 million subsidy from private industry), and $2.8 million on a vessel monitoring system program in 1999 alone. The overall data collection expenditures approximate $45 million annually, while the total (commercial and recreational) value of fisheries’ harvests, including economic effects, are estimated at $45.7 billion annually (NRC 2000a).
The other example is that of the U.S. Fish and Wildlife Service’s Office of Migratory Bird Management, which has developed a continental monitoring program to provide information for sustainable harvest of waterfowl. As indicated earlier, each year management decisions such as season dates and harvest limits are set based on models incorporating information on breeding pairs, weather, and the age structure of populations (Shaeffer and Malecki 1996, Shaeffer 1998). Hunters (via their purchase of duck stamps) pay for many of these monitoring programs, and spend about $1 billion for hunting and related activities. These two programs illustrate that monitoring with specific goals to (1) provide information to set harvest limits and (2) provide information about the response of the resource to management actions can be a powerful tool to manage resources. In particular, the waterfowl monitoring efforts illustrate the use of conceptual models for evaluating factors regulating waterfowl populations (such as snowfall for Canadian geese on breeding grounds or population structure and precipitation for mallards; Shaeffer and Malecki 1996, Shaeffer 1998).
Long-term monitoring provides the necessary data to develop indicators and to model annual recruitment that are then easily measured and used in management decisions. Long-term monitoring data also provide baseline information necessary to evaluate whether changes in management or other environmental variables result in population changes. However, for most biological resources such long-term data do not exist.
Finding 6.4: The United States does not have in place an adequate environmental monitoring program for agricultural and natural ecosystems that would permit assessment of the status and trends of the nation’s biological resources.
Selection of Appropriate Variables to Monitor
The real challenge in developing a long-term monitoring program is in determining what are the most important variables to monitor and how to monitor them in a cost-efficient manner. The natural world is itself variable and populations of organisms fluctuate widely in abundance and distribution even in relatively short time spans. Examples of such fluctuations are the cycles of many forest pests such as tent caterpillars or the lemming cycle. The ability to identify and separate ecological effects attributable to chemical, physical, or biological perturbations from inherent variability is central to characterizing effects and estimating risks in ecological systems. The frequency and amplitude of such oscillations make the development of sampling designs to detect population trends due to other factors challenging. “Choosing from the myriad of potential parameters to monitor is truly a daunting task for managers and biologists. Literally, any biotic or abiotic feature of an ecosystem can be monitored. Moreover, changes in natural resources can manifest at spatial scales from the individual to the landscape, biologic scales from genetic to community, or temporal scales from a few milliseconds to millennia” (Woodward et al. 1999).
Arguing with confidence that environmental conditions are better, worse, or just different owing to changes in crop varieties or cultivation practices is impossible unless natural oscillations (due, for example, to fluctuations in climate) can be separated from changes in the abundance of species and ecosystem function caused by chemical, physical, or biological stressors. The identification of natural fluctuations is complicated because information is lacking about causes and consequences of population oscillations for most organisms. Recent evidence on large population fluctuations caused by the North Atlantic Oscillation and the El Niño Southern Oscillation in birds, insects, and mammals (Sillett et al. 2000, Mysterud et al. 2001) demonstrate the importance of climate variability and long-term data collection. Historical information about diversity or disturbance can also be valuable in explaining the response of different communities to the same perturbation—that is, the invasion of a species, a drought or flood, and so forth. The sequence in which disturbances occur may lead systems to very different succession trajectories (Fukami 2001), again highlighting the need for long-term data collections.
Another challenge in determining what to monitor comes from the potential for strong indirect effects in ecosystems. Ecosystems can be viewed as aggregates of populations functioning together—however, ecosystems also have unique features (spatial structure, diversity, etc.) and function (nutrient cycling, food web relationships, etc.) that must be considered when assessing changes associated with the commercialization of
transgenic plants. Outcomes of perturbations are difficult to predict due to the high degree of indeterminacy in the strength and direction of ecological interactions (Attayde and Hansson 2001). A broad message from community ecology in recent decades is that indirect interactions among species are pervasive in natural communities (Holt and Hochberg 2001). Such interactions are often quantitatively as important as direct trophic or interference interactions. It is therefore possible that a transgenic plant that is toxic to a single soil insect species could affect the diversity of soil microbe species. Addition or removal of individual species have both been found to cause dramatic changes in communities, and recent evidence suggests that even species with “weak” interactions (as opposed to keystone species) may play a significant role in stabilizing communities (Berlow 1999). Indirect interactions arise because most species live in a complex web of interactions, and in principle this makes it difficult to predict the response of even well-understood systems to environmental change (Yodzis 1988, Polis and Strong 1996). The prevalence of indirect effects in natural communities limits the accuracy of even the most basic predictions (e.g., the addition of a predator will reduce the population of prey; Attayde and Hansson 2001).
A central goal for postcommercialization monitoring is to understand (and potentially predict) the outcome of species interactions in agricultural and natural communities. However, understanding and predicting the outcomes of species interactions require knowledge about which species are dynamically coupled, either indirectly or through intermediate species (Attayde and Hansson 2001).
DEVELOPMENT OF MONITORING PROGRAMS FOR TRANSGENIC CROPS
The effects of commercialization of transgenic crops can range from a change in the state or dynamics of an organism, a population, or an ecological system. These effects can be characterized as either direct or indirect. Direct effects involve interactions of the transgenic plant with another organism (e.g., direct killing of an insect pest feeding on a Bt crop). Indirect effects are interactions that result from direct effects. For example, herbicide treatment of resistant crop varieties is intended to reduce weed infestation (a direct effect), but reductions in weeds may reduce populations of species feeding on the weeds (an indirect effect).
Many different types of metrics can be used to assess the impacts of transgenic plants or invasive species on populations and ecosystems (Parker et al. 1999). This committee recommends that a two-part approach be used to assess potential environmental impacts of transgenic crops.
The first element of this approach involves trained-observer monitoring involving technical personnel in agricultural and natural areas management. Trained observers are frequently in the field and could be important in detecting environmental effects. Their observations would enter into a nonmonitoring process of validation. Laboratory and field testing would be conducted to validate scientifically the observations made by the observers. If the effect is validated and shown to be possible to occur in the field, it will be possible to generate scientific hypotheses for how the effect would be observed in the field. The second leg of the monitoring system would be long-term monitoring programs using bioindicators.
Recommendation 6.4: The committee recommends a two-part approach to postcommercialization monitoring of transgenic crops: (1) trained-observer monitoring involving technical personnel in agricultural and natural areas management, and (2) long-term moni toring programs using appropriate indicators (see below).
Although some environmental effects of specific transgenic crops might be predicted, many effects (intermittent, low-magnitude, or cumulative effects) may remain undetected during precommercial field trials, and it might be possible to detect them only after scale-up to commercial plantings. A system of monitoring based on trained observers can be used to help detect such effects. Monitoring for novel or acute ecological effects, such as detection of a new nonindigenous invasive species, has relied on such a system for several decades (Kim 1983). Detecting new effects of transgenic plants, such as a pathogen of herbicide-tolerant soybean or a decline in bird populations in areas planted with certain transgenic crops, would also require a network of trained observers who have an incentive for detecting such unexpected effects.
One concern that emerges from lessons learned from introduction of the ornamental plant known as kudzu (Pueraria montana) into the United States is that widespread planting may increase the likelihood of invasiveness. Ironically the initial spread of this species was linked to the commercial seed trade—plants were grown as ornamentals as well as forage crop (Winberry and Jones 1974, Mack 1996) in addition to its use for erosion control. Good surveillance and monitoring are essential for detecting early invasive tendencies and for allowing early eradication or control measures should they become necessary.
Such a network of trained people would be prohibitively expensive to maintain for the sole purpose of detecting the effects of commercialization of transgenic plants. However, trained professional and volunteer networks already exist in agricultural and natural areas, and it would be useful to determine how to integrate monitoring with preexisting networks.
In the United States, the Agricultural Extension Service and the Animal and Plant Health Inspection Service (APHIS) have provided a detection network for invasive species. In addition, crop consultants and farmers themselves may notice environmental changes associated with transgenic crops. Building and maintaining the capacity to detect new acute ecological effects is paramount to an effective monitoring system. In natural areas, resource professionals in federal and state agencies (e.g., Fish and Wildlife Service, state departments of natural resources, natural heritage programs, National Park Service, Bureau of Land Management, Department of Defense) already assess and inventory or observe native species and ecosystems. The same is done in civil society organizations (such as the Nature Conservancy stewards, Ducks Unlimited, Trout Unlimited, land trusts) and by professional and volunteer naturalists (hikers, birders, entomologists, botanists organized to form exotic pest plant councils, for example, native plant societies, etc.). Any deviations or unusual occurrences can be reported and then verified.
A critical need for assessing whether such reports are associated with the commercialization of transgenic crops will be access to detailed information (at the township level at least) on the spatial and temporal patterns of planting of specific transgenic varieties (see below). Systematic monitoring in areas planted with the same variety is needed to gather data used to build an understanding of processes and changes in the field and surrounding natural areas. Ecological effects with a low frequency of occurrence will probably occur in a spatially heterogeneous pattern. Monitoring for these effects can be improved by taking this spatial heterogeneity into account. For example, because the probability of a change being detected is likely to correlate spatially with the amount of transgenic crop planted in an area, spatial maps of planting density can lead monitoring personnel to the optimal monitoring locations.
Many of these networks of people are already stretched thin by their present responsibilities, so it is important that any additional activities associated with detecting potential environmental effects of transgenic
crops be integrated into their ongoing activities. For example, the Agricultural Extension Service maintains offices in many counties. These extension service personnel have many responsibilities and spend considerable time observing agriculture. They have little spare time, and it would be ineffective to mandate that they spend additional time specifically looking for environmental effects of transgenic crops. But because they are trained observers of agriculture, it is possible that while they are conducting their normal activities they could be on the watch for potential environmental effects of transgenic crops. This may be facilitated by a short half-day or one-day workshop on the potential environmental hazards of transgenic crops, so that these extension service personnel can understand better the types of effects that might occur. Crop consultants are another group of trained observers in agriculture. It may be possible to add a training module to the short courses they routinely take that would prepare them to observe environmental effects of transgenic crops.
LONG-TERM MONITORING AND THE USE OF BIOINDICATORS
The release of transgenic plants is often compared to the introduction of nonindigenous species (Kareiva et al. 1996, Marvier and Kareiva 1999). Lessons from invasion biology and management of nonindigenous plants suggest that the reasons for differences in invasiveness among species are poorly understood (Williamson 1996), and that invasiveness may evolve or be delayed (Blossey and Nötzold 1995, Ellstrand and Schierenbeck 2000), often in the form of a “lag” phase between initial introduction and explosive spread (Williamson 1996). Ultimately, we have little power to predict which species will be a successful invader, which ecosystem may be particularly vulnerable to invasion (Williamson 1996, Williamson and Fitter 1996, Lonsdale 1999), or what the impact of invasions will be (Hengeveld 1999). In fact, for some of the most well-known invasive bird species, such as European starlings and the House sparrow, multiple introduction events were necessary before populations became established (Williamson 1996). Lag times in invasive species may be explained by (1) inherent factors associated with population growth and range expansion, (2) changes in environmental conditions favoring the introduced species, and (3) genetic factors (Crooks and Soulé 1999). Moreover, time-lagged effects are particularly evident after establishment of exotic perennial plants compared to annuals, and in association with range expansion of the invaders (Williamson 1996). Overall, short-term experiments (even if conducted over multiple years or a decade) cannot substitute for long-term time series observations because of the potential for lag times and
the general unpredictable outcomes of altered ecosystem interactions. Long-term monitoring to assess potential changes associated with commercialization of transgenic crops is essential. The inherent difficulty in predicting indirect interactions and cumulative or synergistic effects makes the use of long-term monitoring essential in assessments of the environmental effects of new technologies.
Monitoring Transgenic Crops
Systematic monitoring of the spatial and temporal patterns of an area planted with different transgenic varieties will provide a basis for all other monitoring efforts. Ecological effects with a low frequency of occurrence will probably occur in a spatially heterogeneous pattern, and the probability of occurrence will be predicted to be proportional to the area of the transgenic crop planted. Information on the planting pattern will permit use of epidemiological methods (Waggoner and Aylor 2000) to evaluate reports received from the trained-observer monitoring system. Without this systematic monitoring data, it will not be possible to separate coincidental anecdotes from real ecological trends. The information should also be used to plan long-term monitoring sampling plans to optimize effort allocation. Moreover, in any analyses of long-term monitoring, indicator data need to be linked to the spatiotemporal planting pattern.
Annual reporting of this planting pattern is essential. The spatial pattern should be reported at as fine a spatial scale as possible because environmental effects are likely to be localized and aggregation of spatial data may eliminate the correlation between the occurrence of the crop variety and the environmental effect or may induce artificial correlations. The committee suggests that spatial resolution of planting patterns to a 6 × 6 mile grid may be sufficient, but spatial statisticians should develop the appropriate sampling scenario. Because different transformation events associated with the same trait can have different environmental effects (e.g., transgenic corn varieties Event 176 vs. Bt-11/Mon 810), planting patterns should be reported separately for the various transformation events, instead of aggregated across the type of trait within a crop.
Recommendation 6.5: The committee recommends that any postcommercialization monitoring program that is adopted should include monitoring of the spatial distribution of transgenic crops.
Monitoring Using Biological Indicators
The committee recognizes the difficulty in providing sufficient financial and taxonomic expertise to monitor more than a fraction of the biota
occurring in agroecosystems and natural areas. Indeed, for transgenic crops, the role of monitoring in relation to the specific indicators and our capacity to monitor requires additional deliberation (BOX 6.3). The committee endorses the development of ecological indicators as proposed by the National Research Council (NRC 2000a) for both agricultural and nonagricultural environments. The development of indicators is based on the assumption that monitoring an indicator is more cost effective and accurate than monitoring individual processes or species. Using indicators will simplify what is communicated from such programs.
The NRC (2000a) has developed several criteria for evaluating ecological indicators. The criteria recognize that some ecological indicators have been less useful than anticipated because they have not been clearly linked to underlying ecological processes or because the data requirements are overly complex and extensive. The criteria provide a framework for evaluating indicators to assess the potential importance of a proposed indicator, its properties, its domain of applicability, and its limitations:
General importance. Does the indicator provide information about changes in important ecological processes?
Conceptual basis. Is the indicator based on a well-understood and generally accepted conceptual model of the ecosystem to which it is applied?
Reliability. What is the evidence that the indicator is reliable?
Temporal and spatial scales. Does the indicator inform us about national, regional, or local ecological changes? Are the changes measured likely to be short-term or long-term? Can the indicator detect changes at the appropriate spatial and temporal scales without being overwhelmed by variability?
Statistical properties. Are the statistical properties of the indicator (accuracy, sensitivity, precision, robustness) sufficiently understood that changes in its values will have clear and unambiguous meaning?
Data requirements. How many and what kinds of data are needed to estimate the indicator and to detect trends in the indicator?
Skills required. What technical and conceptual skills must the collectors of data for an indicator possess?
Data quality. Will sampling and analytical methods be documented sufficiently that future researchers will be able to understand how the indicator was estimated?
Data archiving. Data, physical samples necessary to recalibrate the entire dataset, and analytical models should be archived so that interested parties have access to the information.
Robustness. In an ecological sense, is the indicator relatively insen-
sitive to expected sources of interference, such as external ecological perturbations, and technological change in monitoring capacity?
Internal compatibility. Is the indicator compatible with any other indicator being developed or used by other nations or international groups?
Costs, benefits, and cost effectiveness. What is the cost to develop, implement, and refine the indicator? Benefits are more difficult to estimate, but what are the expected ones? Is there another method that can provide similar information at lower cost?
Details for the justification and rationale for each indicator are contained in the earlier NRC report (2000a) and the arguments will not be repeated here. For some of the recommended indicators, data are already being collected through the NRI; others, particularly the monitoring of biological diversity, will need further elaboration. This committee is not charged with and will not provide detailed guidance on organismal groups/taxa for monitoring. However, the committee encourages further research into the potential use of indicator or umbrella species to assess ecosystem health or changes associated with the commercialization of transgenic plants (see BOX 6.2). The USDA and many other federal and state agencies are committed to increasing the monitoring of biological and natural resources. A scientifically rigorous design modeled after the NRI and with cost sharing among agencies should reduce overall costs for such a program. More research is needed to identify organisms and bio-
BOX 6.2 Indicators of the Nation’s Ecological Capital
The Committee to Evaluate Indicators for Monitoring Aquatic and Terrestrial Environments (NRC 2000a) chose the following as indicators of the nation’s ecological capital:
The committee chose as indicators of ecological functioning or performance:
The committee chose as indicators for agricultural ecosystems in particular:
logical processes that are especially sensitive to stresses and perturbations. Future research may suggest better indicators, but their performance and reliability need to be carefully evaluated before implementing them nationally.
The committee suggests that the development of indicators used as “common currency” should be an open democratic process involving agencies, industry, and other stakeholders. Only if there is agreement of what to monitor will the results be accepted.
Recommendation 6.6: There should be an open and deliberative process involving stakeholders to establish criteria for environmental monitoring programs.
The committee follows the arguments for a common currency developed by the indicators committee (NRC 2000a):
Indicators are more likely to be useful if they are understandable, quantifiable, and broadly applicable. They are likely to command attention if they capture changes of significance to many people in many places. Although indicators of local effects are not without value, they must be aggregated into some composite indicator if they are to serve broad policy purposes. Indicators are most policy-relevant if they are easily interpreted in terms of environmental trends or progress toward clearly articulated policy goals, and if their relevance is made clear (Landres 1992). In other words, indicators that convey information meaningful to decision makers and in a form these decision makers and the public can understand are more likely to be observed and acted on. Indicators are also more likely to be influential if they are few in numbers and capture key features of environmental systems in a highly condensed but understandable way. The manner in which data are aggregated to yield a small number of general indicators should be clear, especially to those who wish to understand how the indicators were developed. The reason for choosing indicators, and the selection criteria, should also be clear (Landres 1992).
Any objective ecological indicators should be expressed numerically, so that results can be compared with those of indicators in other places and times. For the indicators to command attention and be used, the data and calculations they are based on must be credible. The choice of what indicators to use and how to define them is necessarily somewhat subjective, but the procedures for measurement and calculations associated with a particular indicator, once defined, must be clearly specified, repeatable, and as free of subjective judgments as possible. Where they are unavoidable, the sources of subjectivity should be defined and identified (Landres 1992, Susskind and Dunlap 1981). For example, the Consumer Price Index and the percent of people unemployed are calculated by well-defined rules that have been agreed on, regardless of a person’s view about the value of full employment or low inflation or even the
validity of these indices. Debates about these numbers do not involve who calculated them. Similarly, ecological indicators need to be based on calculations that are well defined and agreed on.
In addition to being based on credible measurements and calculations, the choice, motivation, and interpretation of indicators should be publicly trusted for them to be of greatest use. That means that the people and organizations who produce the indicators should be generally trusted (Greenwalt 1992). The committee cannot specify the best methods for achieving this goal, but notes that in at least some cases separating the responsibility for preparing indicators from responsibility for carrying out policies based on them seems to enhance trust in the indicators. For example, the Bureau of Census has no policy-making responsibility; so, despite recent political arguments about the validity of sampling as opposed to counting everyone, the population estimates produced by the Bureau are usually trusted. Similarly, the National Weather Service has no responsibility for environmental policies, and so, beyond some scientific questions about the nature and placement of its instruments, its statistics are generally widely respected and trusted. The importance of public trust in the indicators is even more critical if ecological indicators are to be used as input for a national assessment of the state of the nation’s ecosystems.
Crucially important in the data collection process is that the data satisfy standards of timeliness, level of detail, accuracy, accessibility to users, coverage or completeness, and credibility of the data collection and management processes using the data. Because data collections and use will be at various spatial scales, data system designers must allow for demands among users at these various scales. Any system must be able to deal with the needs of users to work at different spatial resolutions and different degrees of timeliness. For the National Marine Fisheries Service, credibility is a major concern. Many stakeholders mistrust data that are collected and analyzed by the same agency that makes policy recommendations, conducts stock assessments, and enforces fishery regulations. These multiple responsibilities create mistrust about the collection and use of fisheries data (NRC 2000d).
Recommendation 6.7: The committee recommends that an independent body separate from APHIS be charged with the development of a monitoring program. This monitoring program/database should allow participation by agencies, independent scientists, industry, and public-interest groups. The database depository should be available to researchers and the interested public.
Establishment of a database depository would build stronger confidence and allow access to more data if collection techniques were standardized. Major advances in monitoring could be achieved by establish-
ing new computing and communications capabilities and by increasing integration and standardization of data management on local, regional and national bases.
Expertise for developing statistically reliable designs is available within the federal government or at universities. For example, the Resources Inventory Division of NRCS develops policy, procedures, and standards and provides guidance for natural resource data collection efforts by NRCS. The division also ensures that NRCS’s data collection efforts are coordinated with other federal, state, and local, government agencies. The Federal Geographic Data Committee (FDGC) coordinates the development of the National Spatial Data Infrastructure (NSDI). The NSDI encompasses policies, standards, and procedures for organizations to cooperatively produce and share geographic data. The 17 federal agencies that make up the FGDC are developing the NSDI in cooperation with organizations from state, local, and tribal governments; the academic community; and the private sector in an effort to improve the quality of the collected information while minimizing the burden for individual agencies through the use of information technologies.
Recommendation 6.8: The establishment of long-term monitoring efforts is recommended to permit assessment of potential environmental changes associated with the commercialization of transgenic plants.
RESPONSES TO MONITORING
The purpose of monitoring (using bioindicators, general surveys by trained personnel, and long-term assessments) is to allow timely and informed responses to changes in agricultural and natural ecosystems (positive or negative) associated with the commercialization of transgenic plants. Without long-term monitoring, informed management decisions are impossible to make because of uncertainties about the causes of shifts in abundance of individual species or of ecosystem function. The longer the data series from a monitoring program, the easier it is to distinguish true ecological relationships from fluctuations due to natural variability. A tradeoff between accuracy and utility develops because the longer we wait for accurate assessments, the less likely it is that responses can be developed to “restore” ecosystem function or rebuild species populations. When there is a considerable lag time before impacts become detectable, the problem of timeliness of response is exacerbated (Byers and Goldwasser 2001). Examples of such delayed impacts are extinction
BOX 6.3 Role of Monitoring
The question of what to monitor depends on the larger purposes that monitoring is intended to serve. With respect to the environmental effects of commercialization of transgenic crops, it is important to bear in mind certain points discussed in previous chapters. First, as discussed in Chapter 2, the focus should be on the environmental effects of transgenic crops rather than the transgenes themselves. These effects depend on the phenotypic characteristics of the crop rather than the fact that transgenes have been moved into the crop through rDNA techniques. Furthermore, even if transgenes move into other crops and wild relatives, such movement may or may not have further effects on such factors as crop performance, food quality, and non-target organisms. As such, the key questions that must be addressed in discussing the need and potential for monitoring have to do with our ability to detect such effects should they occur.
Social science research suggests that some members of the public may regard an unintended movement of the transgene itself as an environmental effect (Gaskell et al. in press). This is especially the case for focus groups conducted outside the United States, where participants are likely to interpret effects on farming practices in environmental terms. Thus, events that complicate production of crops that meet organic marketing standards are likely to be interpreted by some as environmental effects, even if they do not involve detectable effects on non-target organisms or ecosystem functioning. Others will regard the unintended presence of transgenes in a commodity as a direct economic or social effect, rather than an environmental effect, in that they affect the value of farmers’ conventional and organic crops and possibly consumers’ ability to find products that satisfy their preferences.
A system to test crops for the presence of transgenes, either in the field or at various points in shipment, is technically feasible. One could thus monitor nontransgenic crops for the movement of transgenes. The question is whether this is an objective that is consistent with the goals of environmental monitoring. There are both scientific and social value judgments that bear on this question. Given a clear set of criteria for the ecological health of conventional or organic agroecosystems, it might be possible to evaluate whether the presence of specific transgenes in particular farming systems is likely to have detrimental effects. There also may be specific cases, such as organic growers’ use of Bt toxin, where monitoring for specific outcomes, such as increased resistance to Bt among crop pests, might prove beneficial to producers. In general, however, there are no scientifically agreed upon parameters for evaluating the ecological health of agroecosystems (Peck et al. 1998). Thus, the argument for undertaking such a system of monitoring involves an appeal to economic and social values or an interpretation of environmental effects that is broad enough to encompass many impacts that most Americans would characterize in social or economic terms.
While technically feasible, detecting low levels of gene flow into crops or noncrops may be challenging. Gene flow by pollen varies with a tremendous number of parameters, including pollen vector behavior, compatibility of the populations involved, and relative sizes of the source and sink populations (e.g., Levin 1981; Ellstrand et al. 1999). Likewise, seed dispersal can depend on many factors. Typically, the dispersal of pollen and seed follows a leptokurtic curve (Levin and Kerster 1974), such that the great majority of the propagules disperse a very short distance (on the order of meters), but the curve has a long tail (on the order of hundreds or thousands of meters). Of course, the sites to be sampled will be in that tail region of dispersal. The events in the tail are rare enough that they will not be uniformly distributed and are especially subject to idiosyncratic factors (e.g., wind direction) that will increase their patchiness in space. For monitoring to be effective, samples must be large enough to detect rare and patchy events (Marvier et al. 1999). Although
flow into noncrop species may be reasonably interpreted as a necessary (but not sufficient) threshold for certain environmental effects to occur, monitoring gene flow may be of considerably less value for detecting environmental effects than direct monitoring of nontarget species or ecosystems themselves.
Scientific considerations alone do not provide a sufficient basis for general monitoring programs that would attempt to keep track of transgenes that may move either to conventional and organic crops or to noncrop plants. One should not rule out the possibility of future transgenic crops that would warrant such an approach on scientific grounds. Some of the scientific, social, and economic factors that could be considered in evaluating whether the movement of transgenes beyond the cultivars in which they are commercialized should be monitored are summarized below.
SHOULD FEDERAL AGENCIES MONITOR FOR THE MOVEMENT OF TRANSGENES?
“debts” due to habitat fragmentation or invasion of nonindigenous species, global climate change, and groundwater contamination with nitrate and pesticides (Holden et al. 1992, Phillips et al. 2000). Although monitoring may detect some unexpected effects so that action may be taken to prevent or ameliorate those effects, in other cases monitoring may detect those effects so late that environmental damage may be irreversible (e.g., extinction).
Finding 6.5: Monitoring cannot substitute for rigorous precommercialization regulatory risk assessments.
Long-term monitoring programs are expensive and labor intensive, and require standardization of monitoring units and verification of data collected. Society will benefit only if timely responses become part of the framework for evaluating the environmental impacts of transgenic plants. One category of responses to a finding of environmental hazard from a monitoring program would be the option to reassess the environmental risks associated with a specific transgenic plant implicated in the finding. Another category of responses could be the adaptive integration of knowledge gained from monitoring one transgenic cultivar in developing more appropriate evaluation protocols prior to commercialization and postcommercialization monitoring systems for future cultivars.
In general, risk analysis can improve safety using a combination of prerelease assessment and management and postrelease monitoring and mitigation. At present, APHIS has no formal process for responding to data gathered through environmental monitoring. However, some of the responses described below have already occurred to some extent (e.g., research to assess the potential risk of Bt corn pollen on monarch butterfly larvae) and have influenced the activities and regulatory processes of APHIS. A summary follows of the types of responses to monitoring that are essential to the objectives underlying the different types of monitoring approaches described and recommended above.
Currently there are no provisions for an APHIS regulatory response to the detection of environmental hazards after deregulation of a crop (but see Chapter 7). One of the underlying assumptions in recommending postcommercialization monitoring is that a feedback system in the regulatory process will be implemented, to respond to data accumulated during long-term monitoring. These responses may be informational or regulatory.
Examples of Responses
Validation testing could have a direct effect on regulation if, for example, testing revealed a substantial increase in the frequency of a Bt
toxin resistance gene and a federal agency implemented a change in the recommended resistance management plan. Validation testing could be used more broadly to test for causal relationships between the distribution of transgenic crops and their hypothesized environmental effects after commercialization. For verification of precommercialization risk assessment testing, this type of testing would provide feedback on effects at larger spatial scales that would either confirm the validity of current risk assessment procedures or suggest that they should be modified. In the latter case, a formal response to the results of validation testing would be a change in the type or extent of precommercialization testing that an applicant would need to conduct before a permit would be issued or a petition approved.
Long-term monitoring of as many appropriate indicators as is reasonable (including the exact location and acreage of transgenic crops planted each year) might provide critical information on the spatial association between the intensity of use of a transgenic plant and specific environmental effects. Indicator monitoring, coupled with the potential detection of an unexpected effect through a network of observers, will allow the detection and interpretation of spatial relationships between the distribution of specific events and environmental change. Accounting for event location then allows “epidemiological”-style research methods to detect and associate specific environmental effects with specific transgenic crops. Another type of feedback or response to data resulting from indicator monitoring would simply be the provision of spatial distribution data to design additional, more focused monitoring efforts. By using spatial data we can detect quickly which changes are occurring near the use of transgenic crops and so may be resulting from these introductions. Finally, the historical placement of the various events will provide baseline data and trends against which change is assessed as part of the follow-up from potential detection of unexpected effects by observers. Of course, any regulatory response, such as disallowing the continued use of a specific transgenic crop associated with detrimental environmental consequences, would be aided by data about the status of bioindicators at the location of those events.
Any adverse effects detected through monitoring by trained observers would likely be correlational or anecdotal; it must be verified with experiments, tested in field environments to demonstrate that it could happen in such environments, and/or stimulate additional, more focused monitoring to establish the cause of these effects and their potential connection to planting of transgenic crops. Several examples are provided above in the discussion of the trained-observer monitoring system.
Thus, the typical appropriate response to effects reported by trained observers would be a formal verification process to ensure that the results
of observer monitoring activities are acted upon. Of course, in some cases the observed effect may be so dramatic that it would justify immediate action, which would later be followed by a verification process. In addition to the various feedback functions described above, there should be serious consideration given for a process that allows clear regulatory responses to findings from environmental monitoring. For example, one of the possible responses should be to disallow continued planting of the transgenic crop. Such a response could consist of a two-step process. First, the problem should be identified, described, and a probationary warning issued that specifies an observation period during which, if the problem continues, planting will be disallowed. Second, during that observation period, mitigation efforts could be attempted and evaluated. At this time, the burden of proof should shift, so that one must prove that mitigation is successful or planting will be disallowed. Such a response to environmental disruption would require regulatory decision making to match different levels of risk or types of hazards to the timing of the response, mitigation and evaluation periods, the mitigation goal required, and the degree of evidence required. Under the present USDA regulatory system, such matching of a measured regulatory response to the degree of identified risk is not possible.
Recommendation 6.9: The committee recommends that a process be developed that allows clear regulatory responses to findings from environmental monitoring.
Sustaining the quality of the nation’s ecological and natural resources requires effective management of those resources (Grumbine 1994, Olsen et al. 1999). Effective resource management relies on accurate, timely, and complete information on the extent, condition, and productivity of those resources. To obtain this information, federal, state, and local agencies have established ecological and natural resource monitoring programs (Olsen et al. 1999). Monitoring is also used in the identification and definition of environmental problems yet to be recognized or that may emerge in the future.
Major monitoring programs such as the NRI and the NASS provide valuable long-term datasets relevant to agriculture, but they are not sufficiently detailed or focused to allow an independent assessment of the environmental effects of transgenic plants. Currently the environmental monitoring of agricultural and natural ecosystems in place in the United States is inadequate for assessing the potential impacts of commercialized transgenic crops.
To be able to do so, the committee recommends that a two-part approach be used to assess the potential environmental impacts of transgenic crops after commercialization: (1) Trained-observer monitoring would involve technical personnel in agricultural and natural areas management making and recording observations; and (2) long-term monitoring would help identify and distinguish patterns of biotic and abiotic variation in natural and agricultural ecosystems from impact events due to transgenic crops. In order for the findings of such monitoring efforts to be useful, a clear and coordinated regulatory response must be in place.