Public Oversight of Ecologically Based Pest Management
The release of living predatory, parasitic, pathogenic, or antagonistic organisms (biological-control organisms), the deployment of biologically derived products such as toxins or semiochemicals (biological-control products), and the planting of resistant crop varieties (resistant plants) are fundamental components of EBPM. Wide-scale implementation of EBPM could include thousands of commercialized biological-control organisms and products, each of which could be quite specific with respect to target pest or disease as well as agronomic setting (area and crop) to which it would be applied. Any EBPM approach may entail risk to human health or to the environment. Appropriate oversight, therefore, is required to ensure that potential risks are properly assessed and managed, thereby promoting public acceptance of the use of biological-control organisms, biological-control products, or resistant plants. However, it is essential that the cost of meeting oversight requirements should not unduly constrain the development and implementation of promising methods of EBPM. In assessing risk, oversight by public agencies must use appropriate criteria and methods to avoid delays and unnecessary duplication.
An effective oversight system is essential for accelerating the development of EBPM. Such a system will quickly and adequately evaluate human and environmental health risk while considering the benefits of the new biological-control strategy. Construction of an effective oversight process presents a challenge for the future. Many specific biological-control organisms and products may be required to implement EBPM. To date, 43 microbial biological-control organisms have been registered by EPA.
Effective oversight requires risk assessment criteria suited to evaluating any
potentially harmful effects of ecologically based controls on human and environmental health. Generally speaking, the risks posed by biological-control organisms and resistant plants are not the same as those associated with broad-spectrum synthetic chemical controls. The specificity of the mode of action of the biological organism, or resistant plant with respect to target pest, the likely constrained geographic area of biologically based control measures, and the method of deployment all contribute to the likelihood of safety. Human and environmental health risks of biological controls are not only different in kind from those of most conventional chemicals but also lower in the degree of hazard potentially posed. In many cases, the risks of continued use of chemical controls may be greater than the risks of instituting ecologically based strategies. By the same token, for pest problems not currently addressed by conventional chemicals, the use of ecologically based controls may indeed present some risks, but to the best of this committee's knowledge there needs to be more data to make a more accurate risk assessment. Nevertheless, the history of biologically based resistance genes supports their continued use and their risks may well be acceptable compared to those attributable to lack of effective control.
Biological-control products include genes or gene products derived from living organisms that kill, disable, or otherwise regulate the behavior of living organisms. In this category of controls, the product of a living organism rather than the living organism itself is used to manage a pest or pathogen population. Assessment of the potential risks of biological-control products, as chemical pesticides, must include their effects on environmental and human health. The nature of these risks is in large part a function of the mode of action of biological molecules, which range widely from toxicity to attraction. There are many natural products, including Bt toxins, pheromones, floral attractants, insect growth regulators, and plant growth regulators that have been registered as biochemical pesticides (U.S. Environmental Protection Agency, 1994a). Many others are being proposed for exemption from regulation by EPA. Fermentation products and plant extracts with a toxic mode of action have promise, but their development may be hindered if they are automatically regulated as conventional synthetic chemicals.
Effective oversight must also account for the likelihood that the particular kinds of human or environmental risks posed will vary with the biological-control organism, product, or resistant plant. And, the risks may vary in degree if not in kind depending on the particular agronomic setting of the pest/crop combination. Consequently, human health considerations may be paramount in some cases, whereas environmental effects (e.g., adverse impacts on nontarget organisms) may be the only cause for concern in other cases. Accommodating such variation in risk profiles across the many types of ecologically based controls will be key to constructing effective but not burdensome oversight.
Significant markets for agrochemicals in many cropping systems justify investment in the regulatory costs involved in their development, and some biologi-
cal technologies similarly will have sufficient markets and profit potential to justify developers' covering the costs of complying with oversight requirements. However, pesticides used on minor crops or for limited purposes have less market potential. Some of the most efficacious and least risky products and approaches in EBPM will be modest in scale of use. Examples include release of classical self-perpetuating biological-control organisms that may produce great public good, but have virtually no commercial viability in the private sector. There is a need for a mechanism such as the IR-4 Program to encourage public support for development of biological-control organisms and products.
HUMAN HEALTH RISKS
Humans will come into contact with biological-control organisms, biological-control products, and resistant plants during their production and application and through exposure to organisms or residues that persist on crops and in the environment. The acute and chronic toxicities associated with conventional chemical pesticides are not found with most biological-control organisms and products. However, there may be other adverse health effects associated with their use. These include allergenic and other immune reactions and development of hypersensitivity through multiple exposure.
Health considerations are particularly relevant to workers involved in research on and production of biological controls. Workers applying or releasing the biological-control organisms, biological-control products, or resistant plants will also be exposed; therefore, packaging and application methods must address human exposure. EPA does require that hypersensitivity incidents be reported and may impose restrictions such as protective clothing to address potential problems associated with applicator exposure. It may be that adverse effects of biological-control organisms on human health could be restricted to certain individuals. For example, sensitivity to arthropod proteins is experienced only by some people who work extensively in arthropod rearing. Such risks can be managed by screening personnel for allergenicity or hypersensitivity and by providing protection against excessive exposure. When large-scale production of a biological-control organism is undertaken, concern about human exposure to proteins or other reactive materials is further warranted because sustained exposure to these materials can lead to acquired sensitivity and associated adverse reactions.
Microbes can produce numerous toxic and noxious metabolites, and mycotoxin contamination of food and feed from certain microbial control organisms is a legitimate concern (Betz et al., 1990). Appropriate screening for toxic components should therefore be included in risk assessment. In many cases, however, metabolites produced by biological-control agents are highly specific to the target pest and pose a minimal threat to human health. For example, Bt toxins are quickly degraded within the mammalian gut, reducing the potential for their accumulation within the mammalian system. Infrequently, the proteins cause
mild skin irritation in allergic individuals. If, as is typical, the Bt toxin is applied on crops until the day of harvest, there is a probability of residue present in or on edible portions of the treated crop, but these residues present little risk because of the low mammalian toxicity (Panetta, 1993). Acute toxicity data should be obtained about the toxins produced by macrobes or microbes used in ecologically based pest-management approaches in order to characterize the relative risks they pose. In the event that acute toxicity is observed in these initial tests, it may then be appropriate to conduct chronic testing on the toxin itself.
As with conventional pesticides, residues of biological-control organisms or products may persist; for example, low numbers of pests and their biological-control organisms may be found in the harvested crop. Alternatively, resistant plant cultivars may pose risks to consumers if the edible portion of the crop is altered in harmful ways. The acute and chronic toxicities associated with broad-spectrum conventional pesticides may be less likely with biological-control organisms and products and resistant plants. This lower level of toxicity in biological controls may be attributed to a nontoxic mode of action, a mode of action for which there is no counterpart in mammals, low application rate, low potential for persistence and bioaccumulation of natural molecules, and low potential for applicator exposure to high concentrations of a pesticidal substance (except in the case of purified, concentrated extracts or fermentation products).
Host plant resistance can be achieved by decreasing or deleting the plant's production of a chemical that attracts a pest or is a nutrient or vitamin required for pest growth; alternatively, resistance can be achieved by incorporating genes that produce substances that are toxic to, repel, or inhibit the growth of pests. In cases in which there is also a chemical change in the edible part of the crop plant, there is a potential for effects on human nutrition and health. There have been cases in which a resistant commercialized cultivar led to illness in people who consumed the product or harvested the crop, caused by elevated levels of specific toxic compounds. If new or higher levels of potentially toxic proteins or secondary metabolites are produced, then acute or chronic testing may be indicated.
The emphasis of genetic engineers is on altering the plant to express single proteins that are either toxic to pests or inhibit reproduction of the pest. But there is potential for engineering of more complex traits—for example, secondary biochemical pathways with altered levels of certain compounds or biosynthesis of novel compounds. If either scenario exists and the competitiveness of the recipient organism changes, then additional studies are required to examine impacts on human health and the ecosystem.
It is certainly possible that new compounds produced in plants could also be toxic to nontarget organisms at low concentrations. For example, the Bt genes currently in use by genetic engineers encode a toxin that is effective at 1 microgram per gram tissue or less. Most plant-produced toxins are less effective than the Bt toxin at low concentrations (Leemans et al., 1990).
It is not clear that toxic resistance factors produced through genetic engineer-
ing pose a greater risk than secondary plant compounds produced through traditional breeding. These secondary compounds often have broad-based toxicity; and the effects of this toxicity on human health are poorly understood. More important, the biochemical basis for the resistance achieved through traditional plant breeding is often unknown. Understanding the mechanism of the resistance in genetically engineered plants can provide more confidence that the resistant plant will not adversely affect ecosystems or human health.
Conventional chemical pesticides may adversely affect groups of nontarget, economically desirable organisms, such as pollinators, predaceous insects, fish, and other wildlife (Higley and Wintersteen, 1992). Broadcast chemical sprays contact nontargets as well as targeted organisms, both inside and outside of the treatment area as a result of drift and runoff of the pesticidal materials. Such broad impact has generated pressure to move to less disruptive management practices (Pimentel et al., 1980). The scale for potential harm is generally very much smaller when considering nontarget effects of biological-control organisms.
Adverse effects on nontarget (nonpest) organisms and creation of new pests are the risks to environmental health that may be presented by EBPM strategies. Use of biological-control organisms and products as well as of resistant plants potentially can harm nontarget organism populations directly or indirectly though altered interactions within the pest complex. Use of control organisms and resistant plants can result in a new pathogen or weed through genetic exchange with other microbes and plants (Betz et al., 1987; Higley and Wintersteen, 1992; Hollander, 1991; Howarth, 1983, 1991; Hoy, 1992; Hoy et al., 1991; Lockwood, 1993; Nafus, 1993; National Audubon Society, 1994; Pimentel et al., 1992; Tiedje et al., 1989; U.S. Department of Agriculture, Agricultural Biotechnology Research Advisory Committee, 1991). The kinds and severity of these risks vary with the organism, product, or plant and the environmental and agronomic setting of the pest-management problem. For example, the length of time an organism is expected to exert effects on the pest/host system will help determine its opportunities to harm nontarget organisms. Similarly, the host range of a biological-control organism will determine the set of nontarget organisms it might attack. Fortunately, there is a considerable knowledge base on which to determine these manifestations of risks (Alexander, 1990; Caltagirone and Huffaker, 1980; Charudattan, 1990a; Cook, 1993; Ehler, 1990; Howarth, 1991; Hoy, 1992; Lockwood, 1993; National Research Council, 1989a; Tiedje et al., 1989).
With biological-control macroorganisms, the most commonly cited risk is the potential for the organism to extirpate (eliminate through disruption) popula-
tions of nontarget organisms. Extirpation of nontarget species might result from predation, parasitism, pathogenicity, competition, or other attack on the nontarget species by the organism. This risk is most significant when a macroorganism is exotic (a nonnative species imported from outside the target area) and potentially not influenced by ecological controls to which an organism native to the area would be subjected. Still, the history of introductions of macroorganisms as biological-control organisms indicates a low incidence of problems (Clausen, 1978). The most notable examples of adverse nontarget effects from the use of macroorganisms as biological-control organisms involve the introduction of vertebrates such as the cane toad and mongoose (DeBach, 1974) and predatory snails (Howarth, 1991). However, many of these cases occurred in restricted or island populations, where habitat limitations appeared to have imposed constraints on the establishment of long-term balance between the introduced organism and its host/prey.
Biological-control organisms can attack less favored organisms if the target organism becomes scarce (Howarth, 1983, 1991). Local populations of organisms closely related taxonomically and ecologically to the target of the biological-control organism are most at risk to nontarget effects. Although an introduced organism may parasitize or prey on a species closely related to the target (Clausen, 1978; DeLoach, 1978; Krombein et al., 1979; Wapshere, 1982), this is not of serious concern unless extinction of the nontarget population is possible (Howarth, 1991). Host-switching exhibited by a biological-control organism often involves attack on a nontarget host or prey that falls within the organism's natural range of suitable hosts and thus does not necessarily involve a change in behavior or genetic adaptation by the organism. Even with some elasticity in host range or habitat range, there are limits to host/prey switching that confer a level of specificity to the use of macroorganisms as controls and lessen the risk of nontarget effects (Harris, 1988; Watson, 1985). Even after thorough investigation of the physiological host range of a new biological-control organism, and with data on ecological host ranges of closely related organisms, additional assessments of affected hosts must be made after the organism is introduced into the field. Although there is virtually no empirical evidence of harmful prey switching by exotic biological-control organisms, few thorough studies have been reported (National Audubon Society, 1994).
Experimentation can provide useful insights into the impact of changes in host-range specificity of pathogens over time. The potential for changes in host range is an important influence in assessing the risk of using pathogenic biological-control organisms. Typically with an exotic pathogen, a specific pathotype is identified for introduction and studied in its native environment to delineate its host range and pathogenic potential. With an indigenous pathogen, a specific isolate or pathotype may be used initially, but additional selection may be necessary to identify a more effective strain. In this situation, the host-range specificity of the original pathotype may not represent that of the selected strain. Hence, it is advisable to assess the host range of each improved pathotype or strain.
Hypotheses regarding the innate ability of natural enemies to parasitize or prey on nontarget species can be tested under microcosmic conditions in a laboratory or greenhouse. But the physiological host range revealed by such experiments may not represent the true (ecological) host range of the organism (Olckers et al., 1995; Watson, 1985). Single-choice experiments (one organism with one host) are poorly discriminatory and are likely to give false-positive results. Nonetheless, they are useful in eliminating concern about potential nontarget hosts. Multiple-choice experiments (one organism with two or more different hosts) reduce the chances of false-positives; nevertheless, the ecological host range expressed under field conditions is likely to be much more narrow than the physiological host range characterized by laboratory choice tests (Ridings et al., 1978; TeBeest, 1988; Watson, 1985). Any prediction of the ecological host range must consider both the experimentally determined physiological host range and the biological, ecological, and taxonomic host-range of any closely related organisms.
The prediction of ecological host range is particularly difficult with self-perpetuating organisms that function at the tertiary trophic level, such as predatory and parasitic arthropods. Although it can be determined in laboratory conditions whether a predatory or parasitic arthropod will utilize a range of hosts, such data cannot be freely extrapolated to field conditions. Complex spatial, temporal, and behavioral interactions commonly displayed by self-perpetuating macroorganisms ultimately determine whether an organism has even the opportunity to encounter a given host.
After release of the organisms into the field, appropriate monitoring will be necessary to check the reliability of the physiological host-range data. Monitoring over several years will identify direct and indirect effects on natural communities. Every introduction provides the opportunity to validate and/or modify protocols for estimating ecological specificity from microcosm studies. Climate, geography, floral phenology, and trophic interactions are key aspects that combine to define the relationship of an organism with coexisting organisms. Thus, ecological host range can be estimated with some accuracy, combining physiological host-range data with ecological evidence from other geographical areas where a similar system (ecological analogy) is in place, particularly if closely related biological-control organisms have previously been monitored there after release (National Audubon Society, 1994).
In general terms, a threshold level of a pest is required to sustain reproduction of a predatory or parasitic biological-control organism; and if high numbers of the biological-control organism suppress the capacity of the pest to reproduce, then negative feedback adversely affects the survival of the biological-control organism (National Audubon Society, 1994). As pest population levels decline, there is a decreased likelihood of contact between the biological-control organism and the pest, resulting in reduced reproduction of the biological-control organism. Although this traditional theory of predator-prey relationship remains
valid, other components such as foraging behavior also contribute to these ecological interactions (National Research Council, 1986b). Current knowledge of host-prey relationships is useful as a starting point in assessing risk of nontarget effects on macroorganisms. However, the continuing contributions of ecologists, biologists, entomologists, weed scientists, botanists, zoologists, and other scientists will be necessary for a comprehensive evaluation of these complex ecological systems.
Because of the enormous diversity of microbial pathogens that might be used as pest controls, there is incomplete understanding of their potential adverse effects on nonpathogenic microorganisms in natural systems. Because microorganisms occupy numerous soil or plant habitats, laboratory experiments can provide data only on selected populations, and evaluation of nontarget effects from these studies may not be applicable to agricultural ecosystems (Hollander, 1991; Tiedje et al., 1989). Though the species composition of any particular microbial ecosystem may not be known, there is considerable knowledge about some of the functional roles and mechanisms of microbes found there; this information can be used to identify potential effects on nontarget species (Cook, 1991; National Research Council, 1989a; Tiedje et al., 1989).
Certain microorganisms, such as B. thuringiensis, suppress plant pests by producing toxins or antibiotics, and risk evaluation of these biological-control organisms is focused on the persistence of the toxin and its effects on nontarget species. Experience has demonstrated a basis for concern about potential effects of -endotoxin (Bt) on nontarget arthropods within the class against which the biological-control product is active, particularly in the case of certain endangered species of butterflies. In these cases, the habitats of concern have been identified and product use has been prohibited there (Hutton, 1992). This approach is especially appropriate when a gene encoding a toxin is introduced into an organism that occupies a habitat where the toxin would not otherwise be found. The encapsulated Bt protein toxin is nontoxic to vertebrates and has been demonstrated to be of minimal risk to other nontarget organisms (Panetta, 1993); it has established an excellent precedent, but similar benign properties must be established for other microbial toxins that may be useful in EBPM. Molecular techniques can provide scientists with the tools to make precise alterations in toxin-encoding genes, thereby improving product performance and increasing the ability to focus the effect of the control on the target organism.
The long history of plant breeding suggests that resistant plant cultivars rarely cause significant effects on nontarget organisms. Plant resistance factors range from chemical to physical and can be specific to a single pest species or have broad effects on an array of organisms as different as arthropods and pathogens.
The potential risks to nontarget organisms resulting from new or elevated levels of toxins in resistant plant cultivars are not well known. If a cultivar resistance is toxicity based, then there is potential for native herbivores such as caterpillars or honey bees to be harmed, though these occurrences are not well
Collego®: A Mycoherbicide Approach for Weed Management
COLLEGO® is one of two commercial bioherbicide products commercially available since the early 1980s. COLLEGO® is a formulation of the spores of a fungal plant pathogen Colletotrichum gloeosporioides [Penz.] Sacc. f. spp. aeschynomene (Coelomycetes). The product was developed by scientists of the Arkansas Agricultural Experiment Station at the University of Arkansas; the U.S. Department of Agriculture's Agricultural Research Service's Rice Research Station at Stuttgart, Arkansas; and The Upjohn Company. COLLEGO® is used to manage northern jointvetch, Aeschynomene virginica (L.) B.S.P., a native leguminous weed in rice and soybean crops in Arkansas, Mississippi, and Louisiana.
The fungus produces an anthracnose disease that can kill both seedling and mature northern jointvetch plants. The natural concentrations of the fungus found in fields are generally inadequate to control northern jointvetch. Effective weed control can be achieved, however, by augmenting natural populations of the fungus by spraying fields with a suspension of fungal spores. The fungus can be easily grown in culture and produces spores abundantly, making commercial formulation easy. COLLEGO® is a wettable powder of dried spores sold-in three components: (1) the spore powder, (2) a hydrating liquid, and (3) activated charcoal to clean spray tanks. It is applied using aerial or ground sprayers after the crop has emerged, preferably soon after rain or irrigation.
COLLEGO® has consistently provided more than 90 percent control of northern jointvetch. Although the fungus has a much broader host range than originally thought (it can infect several nontarget legumes including English peas [Pisum sativum]), it has not posed any danger to nontarget plants under field conditions during nearly 2 decades of experimental and commercial use.
COLLEGO® has served as a model for bioherbicide science, technology, and regulation. Since its commercial introduction in 1982, it has been used repeatedly and successfully, performed consistently, been integrated with other pest management and cultural practices, proved highly stable in its virulence, and produced excellent scientific information garnered from a decade of follow-up research.
SOURCE: TeBeest, D. O., and G. E. Templeton. 1985. Mycoherbicides: Progress in the biological control of weeds. Plant Dis. 69:6-10.
known. Additional field and laboratory studies can improve an assessment of such risks.
Some resistance factors are expressed only at one stage of the crop's growth—for example, resistance to second-generation corn borer (Brindley et al., 1975). Some are expressed in only one organ of the plant—for example, in the silks of corn—and some are expressed only when the plant is under attack by the pest or pathogen (Ryan, 1983). Other resistance factors are present in all stages of growth and in most organs of the plant. The range of species affected by a plant resistance factor is an indicator of potential effects on nontarget organisms. This
is analogous to host range and specificity determining the potential effects of a biological-control organism on nontarget organisms. Superior resistance factors are those that decrease multiple pest or pathogen populations without adversely affecting nontarget organisms (Kennedy and Barbour, 1992). However, there are cases in which improved resistance to one pest decreases resistance to others (Dacosta and Jones, 1971), which could be a severe limitation to product efficacy in some situations.
In some cases, host plant resistance changes the plant in a way that benefits natural enemies of a pest (Price et al., 1980). For example, the production of extrafloral nectaries or changes in leaf architecture may enhance the ability of a biological-control organism to find its target pest organism (Schuster and Calderon, 1986). Certain cultivars of wheat are thought to be resistant to a soilborne disease as a result of enhanced populations of antagonists. However, host-plant resistance can adversely affect biological-control organisms as well as pests (Robb and Bradley, 1968) by reducing populations of arthropod hosts parasitized or preyed on by control organisms. There are no clear examples where negative effects on natural enemies have overridden the positive effects of pest reduction. Applied entomology has a history of assessing potential negative effects on beneficial arthropods before resistant cultivars are released.
Exacerbation of Plant Pests
Assessment of the safety of a biological-control organism introduced into the environment involves several factors, including familiarity with the organism and its function, prior history of use, and characteristics of the target environment. Much has been learned from prior releases of biological-control organisms that can provide a basis for assessment. If the organism is unfamiliar or there is uncertainty about the environment into which it is introduced, a careful evaluation must be conducted prior to introduction into the environment (National Research Council, 1989a).
If there is a risk associated with introducing biological-control organisms, products, and cultivars into the environment, it is the low probability of producing a pest that previously was not a pest. Microorganisms have been determined to exchange genetic information in the environment, but with low frequency. Transfer of genetic information may result in (a) acquisition of virulence or (b) enhancement of host range or virulence. Although such enhancement may contribute to enhanced pathogenicity or compromise the durability of biological tools introduced into the environment, the pathogenicity of a microorganism is known to result from a complex interaction among a number of genes and gene products of the pathogen and host. Hence the pathogenic potential of a microbe can be anticipated.
The probability of a nonpathogenic species becoming a virulent pathogen is extremely low. To confer pathogenicity, a pathogen must first attach itself to a
suitable host, compete for nutrients, and also, in the case of saprophytes, resist the defense systems of the host. It is highly unlikely that moving one or a few genes from a pathogen to an unrelated nonpathogen will cause the recipient to become pathogenic. Species most at risk are those that are most closely related to the pathogen, while those less closely related must overcome considerable genetic barriers (National Research Council, 1989a).
Some documentation does exist to confirm gene transfer between microorganisms. Resistance genes of some introduced microorganisms are transferred to indigenous microorganisms such as A. tumefaciens. Cisar and colleagues demonstrated an enhancement of host range or virulence of the mycoherbicide Collego® through transfer of genetic material to a related indigenous pathogen (Cisar et al., 1994). Pathogens with enhanced host range or virulence may be generated through genetic instability; for example, if rearrangements occur in the genome of disabled viroids placed into transgenic plants, there is potential for generation of virulent forms (Hammond, 1994). Some other unknown consequences of genetic exchange may exist, but comprehensive field evaluation can demonstrate occurrence of these events in a natural setting.
Weeds may result from crosses between introduced and native plants. Wild plants have been brought into the United States from other countries for breeding purposes. These wild plants have been crossed with cultivated crop varieties in open-field tests often with few precautions against the development of new weeds. Origins of weeds can be traced to (a) wild colonizers, (b) hybridization between wild and cultivated races of domestic species, and (c) abandoned domesticated varieties (Oka and Morishima, 1982). Experience with intensive crop breeding during this century indicates that hybridization between wild relatives and germplasm used for crop improvement is relatively rare.
The experience with sorghum is quite instructive and can be used as a general model for understanding the potential for development of weed-crop complexes and new pernicious weeds. According to De Wet (1966), the weediness of johnsongrass (Sorghum halepense) was enhanced coincidentally with its introgression with cultivated sorghums (Sorghum bicolor) in the United States. When these johnsongrass populations extend their already major ecological role outside agricultural fields, they represent the most extreme category of known risk associated with gene flow from a crop to a weedy relative.
Other examples of gene exchange (gene flow or introgression) between a domesticated crop and its wild relative has been reported for cultivated rice (Oryza sativa) and wild rice (Oryza spp.) (Parker and Dean, 1976). Gene exchange between corn and its wild relative teosinte, between Eastern carrots and wild carrots, and between durum wheat and wild emmer wheat have been documented. The genus Amaranthus is prone to natural hybridization between the few cultivated species and several weedy relatives (Simmonds, 1979). In some cases, the hybrids between the cultivated and wild amaranths could out-compete their weed parents (Tucker and Sauer, 1958).
Biological Control of Crown Gall
Agrobacterium radiobacter strain K84 is a naturally occurring bacterium that controls crown gall, a plant tumor caused by the related soil bacterium A. tumefaciens. Crop losses caused by crown gall occur worldwide and can be extensive, particularly in nurseries growing rosaceous plants, grapevines, and stone fruit trees. A. tumefaciens enters the plants through wounds and causes tumors, which can weaken, reduce the aesthetic quality, and eventually kill the host plant. There are no effective chemical controls currently available for crown gall.
The biological control agent A. radiobacter strain K84 has been used commercially for more than a decade in many regions of the world, including Australia, Greece, Israel, Italy, Japan, New Zealand, South Africa, Spain, and the United States. K84 can be applied to wounds on cuttings, bare-rooted seedlings, grafts, and on field-grown plants. K84 protects wounds from infection by A. tumefaciens in part because of the production of agrocin 84, an antibiotic with specific toxicity against sensitive strains of A. tumefaciens.
Genes determining the production of agrocin 84 and immunity of the host bacterium to agrocin 84 are present on pAgK84, an indigenous, conjugative plasmid of strain K84. If plasmid pAgK84 is transferred to A. tumefaciens through the natural process of bacterial conjugation, the pathogen becomes immune to agrocin 84 and less sensitive to biological control by strain K84. In response to concerns that the predominance of A. tumefaciens harboring pAgK84 may reduce the efficacy of biological control, a derivative strain of K84, lacking a region (tra) required for conjugal transfer of pAgK84, has been constructed (Jones et al., 1988). This strain cannot transfer pAgK84 to other bacteria and its use is expected to minimize the risk that biological control will break down due to the presence of agrocin 84-resistant strains of A. tumefaciens in nursery soils. The strain containing the tra deletion is used commercially in Australia.
Gene flow has apparently occurred from cultivated rye (Secale cereale spp.) to wild relatives in California, where a weedy rye probably derived from a cross between S. cereale and S. montanum (wild relative) has become increasingly crop-like. This introgression has proceeded to such an extent that farmers are said to have abandoned efforts to grow cultivated rye for human consumption; instead they deliberately sow hybrids for forage (Jain, 1977; Suneson et al., 1969).
Although hybridization between a crop and its wild relative may not be preventable, there is little likelihood that desirable domesticated traits will be retained in the wild relative. Much of the emphasis in plant breeding has been on traits that would reduce adaptation to the wild. Important commercial traits, such as pest resistance, that have the potential to alter the ecology of wild relatives have not been a problem, with the possible exception of gene transfer from cultivated sorghum to johnsongrass (National Research Council, 1989a). In gen-
eral, experience with traditionally bred resistant plants suggests that although the potential for genetic transfer between resistant varieties and wild relatives exists, it occurs infrequently. Where it has occurred, it caused a problem only rarely because the weed became less weedy and more domesticated. Still, there is need for vigilance in the search for evidence of negative effects of resistance-gene transfer (National Research Council, 1989a).
RISK ASSESSMENT AND MANAGEMENT
Evaluation of risks associated with deployment of biological-control organisms and products and resistant plants should be based on evidence relative to both the type of organism or product and its method of deployment.
EBPM involves the deployment of a range of tactics from application of synthetic chemical pesticides to liberation and establishment of live organisms. This range of tactics might be viewed as a continuum, with application of synthetic chemical pesticides on one end and relocation of living self-perpetuating arthropods or microbes on the other. Along the continuum one might encounter ''natural" pesticides derived from plants (sabadilla, rotenone) or microbes (Bt -endotoxin, others) that are deployed using methods and expectations similar to those used for synthetic pesticides. Further yet along the continuum one could encounter living, nonperpetuating and self-perpetuating organisms whose effects may be derived by the in situ production of toxins or other deleterious factors. Among these are entomopathogenic viruses and plants containing pest or disease resistance capabilities. It should not be difficult to envision, then, that the processes, fundamental principles, and expectations of deployment of tactics along this "continuum" would vary widely from one portion of the continuum to another. Therefore, basing the conceptual or empirical framework for oversight of all organisms or products on characteristics, risk factors, and history of one could lead to inappropriate risk evaluation and failure to meet the goal of ensuring human and environmental safety.
Drawing on Experience and Experimentation
All experience including that gained from the natural occurrence of the biological materials and their toxicity and field testing should be fully considered in regulatory review of new organisms or products.
There are innumerable combinations of controls (organisms, products, resistant plants), pests (pathogens, weeds, arthropods), and agroecological settings (geographic location, crop). If attempted on a case-by-case basis, requiring experimentation unique to each, oversight requirements and costs could well become prohibitive, particularly given the small market size of many biological controls. Fortunately, the task of risk assessment for one situation can be greatly
expedited by drawing on relevant experience with similar controls, pest, and agroecologies. Knowledge gained from past releases or uses of related organisms or products is the best guide for evaluating the potential risks and benefits of new releases of organisms in the same category. As experience grows with related taxa or functionally similar organisms, adjustments can be made in the categorization and oversight required for new organisms. For example, USDA's Animal and Plant Health Inspection Service (APHIS) proposed that introduction of certain categories of genetically modified plants would require notification rather than a permit (Federal Register, 1992). The EPA proposed exemption from the notification requirement for certain subgroups of microbial pesticides, as information warranting such action becomes available (U.S. Environmental Protection Agency, 1994b).
For any particular biological-control organism or product or resistant plant, expeditious risk assessment will rely on information generated by field experience with closely related organisms or substances and/or by appropriate experimentation in the laboratory or greenhouse. Consequently, the protocol for human and environmental risk assessment will vary accordingly. Relevant risks must be assessed, the information or additional inquiries needed to support risk assessment must be identified and evaluated, and a synthesis of the findings must be made: but by whom? For EBPM, the range of expertise and experience required to make such judgments transcends that which public agency professional scientific staff can reasonably be expected to cover. Consequently, public oversight must employ expert review committees consisting of scientists from relevant disciplines in agriculture, ecology, and human health as well as practitioners with experience with closely related organisms or products. Although public officials remain the ultimate arbiters of the acceptability of any risks posed by a new pest control, the diverse nature of biological-controls dictates involvement of scientists with relevant experience or expertise simply on the grounds of efficiency, avoiding imposition of unnecessary or duplicate information requirements.
Experience, experimentation, and expert opinion should direct oversight attention to broad-spectrum organisms and products or resistant plants and their use on major acreage crops where risk factors are greatest or most difficult to assess. At the same time, effective review will exempt or remove from oversight those organisms, products, or resistant plants for which accumulated experience indicates low risk. Generally speaking, higher human and/or environmental risks will be associated with larger scale in geographic use and/or with the duration of effect of the control organism, product, or resistant plant. Conventional broad-spectrum pest control chemicals often pose risks because of persistence in the environment or as residues in food or because of wide use.
Scale of Use
The significance of scale of use of a pest control is already recognized, for example, in the differential oversight treatment by EPA of microbial control organisms depending on whether the proposed release is for small-scale testing (less than 10 acres), large-scale testing, or full commercial product registration (widespread use). Many biological-control organisms, even if fully commercialized would be limited to relatively small geographical areas because of their environmental or host specificity. Limitation in scale of use also limits human health risk because fewer individuals may be exposed in production or application of the control and because any dietary exposure would be restricted. Environmental risk is similarly restricted because the geographic specificity is related to the fact that the control is not useful or cannot survive in other agroecological settings.
The persistence of a control organism may also affect the possibilities for adverse effects on the environment or increases in human exposure. Biological-control organisms can be
self-perpetuating organisms that become permanently established;
self-perpetuating organisms that can reproduce through one or multiple generations, but will ultimately expire (for reasons such as climatic extremes—e.g., mealybug destroyer in southern California not able to overwinter); or
organisms that are incapable of self-perpetuation in the environments into which they are introduced.
An established self-perpetuating organism exerts permanent effects on the local native populations, the pest-pathogen complex, and on the gene pool. If any of these effects are potentially negative, the environmental risk of release is magnified by self-perpetuation.
The characteristics that confer a biological-control organism with the ability to self-perpetuate are
tolerance of the environment and habitat,
ability to compete with established populations of indigenous organisms,
appropriate life-cycle synchrony with a target host (particularly important for macroorganisms), and
ability to reproduce when host density is high.
Self-perpetuation requires the ability to survive during periods of low host abundance, to tolerate winter and other adverse conditions, or to survive by assuming a dormant form. In some cases, the likelihood of self-perpetuation then can be deduced from the taxon to which the organism belongs; in addition, the method
and timing of deployment can influence establishment of the organism and its longevity.
With a nonperpetuating biological-control organism, site-confined application is possible. The fact that a nonperpetuating organism will not become a permanent component of an ecosystem may substantially reduce risk where there is a concern about human health and environmental effects (Howarth, 1991).
Microbial pathogens of plants can persist in nature. If susceptible hosts are present at adequate densities and environmental conditions are conducive, microbial pathogens will spread from the initial focus of establishment. When populations of the target host are reduced, then dispersal of the microbial pathogen slows and its numbers decline. Over time, the system assumes equilibrium, or homeostasis, when the control agent becomes endemic. Fitting this model are the postrelease histories of the following weed-control agents:
Puccinia chondrollina, a rust fungus from the Mediterranean region used to control skeleton weed (Chondrilla juncea) in Australia (Cullen, 1985) and California (Supkoff et al., 1988);
Entyloma compositarum, a smut fungus from Jamaica used to control hamakua pamakani (Ageratina riparia) in Hawaii (Trujillo et al., 1988);
Phragmidium violaceum, a rust fungus from Europe used to control wild blackberries (Rubus constrictus and R. constrictus) in Chile (Oehrens, 1977); and
Puccinia carduorum, a rust fungus from Turkey used to control musk thistle (Carduus thoermeri) in Virginia (Baudoin et al., 1993).
Long duration of effect, by itself, is not a risk factor; the risk depends on the nature of the control organism. If an organism is host specific, for example, longevity will pose little risk to nontarget organisms. Indeed, the primary effect of longevity will be to increase the benefits of the organism as a pest or disease-control system. If the organism is not host specific, its longevity will increase the potential for effects on nontarget organisms. A persistent organism can become a pest or a nuisance in its new environment (Caltagirone, 1981; Harris, 1988; Howarth, 1983), but it is possible to minimize such risk. Phylogenetic, ecological, and biological relationships are indicative of the host ranges of related groups of biological-control organisms. When empirical evidence indicates a restricted host specificity, then risk from introduction of an organism from that group can be estimated (National Audubon Society, 1994).
Conventionally, risk assessment is taken to be a separate exercise from risk management (National Research Council, 1983). As applied to human health concerns, risk assessment involves hazard identification (Does the agent cause the adverse effect?), does-response assessment (What is the relationship between dose and effect in humans?), and exposure assessment (What exposures are cur-
rently experienced or anticipated under different conditions?). The resulting risk characterization answers the question, What is the estimated incidence of the adverse effect in a given population? Risk assessment is performed by public-or private-sector analysts in support of public agency decisions about risk management and acceptable levels of risk. In the case of human health effects, especially with respect to carcinogens, legislative mandate dictates the tolerable level of risk.
Although levels of acceptable risk are established in the public arena, not by the scientific community, guidance should be sought from experience in managing risks similar to those posed by biological-control organisms and products or resistant plants. For example, the Food and Drug Administration's current view of the health risks of genetically engineered plants (that the fact of their modification does not a priori raise concerns) is instructive and demonstrates that the requirements of public oversight need not be burdensome. Environmental risks posed to nontarget organisms, particularly when extinction might be a possibility, have been managed through the registration process for conventional chemicals. Determination of the acceptable risk should be made without regard to the process that produced the modified organism (i.e., conventional breeding versus genetic engineering), recognizing the relevance of experience already gained and the precision of newer techniques (National Research Council, 1989a).
Gaps and Inconsistencies in Current Oversight
The current oversight responsibility for biological-control organisms, products and resistant plants is shared primarily between EPA and USDA/APHIS and is characterized by application of different oversight requirements for microorganisms versus macroorganisms, for exotic versus indigenous organisms, and for organisms considered to be plant pests and those not considered plant pests. The complexities and anomalies of the current system may be attributed to the overlapping jurisdiction of several agencies, the diversity of organisms to be regulated, and the attempt to make the decision making "template" developed for registration of conventional chemical pesticides applicable to biologically based controls.
Current federal oversight generally evaluates the risks of biological-control macroorganisms using the procedures developed for exotic plant pests under the Federal Plant Pest Act (FPPA); microbial biological-control agents are subject to oversight procedures developed for chemical pesticides under the Federal Insecticide, Fungicide, Rodenticide Act (FIFRA). Neither set of procedures is tailored for registration of biological-control agents; therefore, both pose unnecessary barriers to registration of biological-control organisms.
The goal of FPPA is to prevent the introduction or interstate movement of exotic plant pests. Although the statute does not specifically provide for comprehensive regulation of all biological-control organisms, all such organisms (even
those that are only capable of attacking pest arthropods) are assumed by APHIS to be potential plant pests and therefore subject to regulation under FPPA.
FIFRA was intended for regulation of chemical pesticides and poses severe constraints to the efficient testing of microbial biological-control organisms. Once separated from exotic or endemic plant pests under quarantine or similar containment conditions, biological-control organisms should be subject to risk assessment protocols commensurate with their biological and ecological attributes.
Although there are cases in which the pesticide "template" functions effectively for large-scale commercial registration of biological-control products, they do not function efficiently for registration of biological-control organisms. Neither the intent of the biological-control organism application nor the process by which it operates parallels the chemical pesticide model.
A case in point is the convoluted process required for the registration of encapsulated Bt-endotoxins (a killed microbial pest control agent that was engineered to produce the Bt toxin). Killed microbial products produced through the fermentation of a live intermediate are further regulated by EPA under the Toxic Substances Control Act (TSCA) as pesticidal intermediates, despite the fact that such agents are contained within a fermentor until the final killed microbial pesticide is produced. Such evaluation includes a review of the potential health and environmental effects of the inadvertent release of the live intermediate. This situation results in a dual review process within EPA, with the manufacturing portion of the process reviewed separately from the field release of the killed microbial pesticide. This is apparently the result of the dual jurisdictions of FIFRA and TSCA for review of "chemical intermediates," a distinction that is irrelevant to microbial pesticides.
The existing fragmented oversight process employs different assessment criteria to biological-control organisms that pose similar human or environmental risks, resulting in unnecessary scrutiny of some low-risk organisms and almost complete lack of regulatory review of others. The most important inconsistency is the differential treatment of macroorganisms by USDA/APHIS under FPPA and of microorganisms by EPA under FIFRA. For example, neither EPA nor APHIS regulates nonpathogenic nematodes, whereas both agencies regulate microorganisms. In addition, microbes are exempt from oversight if vectored by a nonpathogenic, indigenous macroorganism (such as a nematode). There is no clearly defined process for risk assessment of biological-control macroorganisms that are not plant pests, resulting in indecisiveness in granting permits for field release. The current regulatory oversight of macroorganisms is thus reduced to case-by-case evaluations through environmental assessments or environmental impact statements under the National Environmental Protection Act (NEPA). To provide appropriate oversight of biological-control organisms and products, USDA, EPA, and other federal and state agencies with oversight authority should jointly develop criteria and protocols specifically for the testing, registration, and
use of living control organisms, biological-control products, or plants resistant to pests or diseases.
No formal oversight structure exists for traditionally bred resistant plants, whereas genetically engineered resistance will, it appears, be subject to close scrutiny (U.S. Environmental Protection Agency, 1994a). A question confronting policy makers is whether host plant resistance accomplished through genetic engineering should receive a higher degree of scrutiny than resistance achieved through traditional plant breeding. Such a distinction imbedded in regulatory practice would not necessarily reflect differences between actual human or environmental health risks posed by the two technologies.
Options for Improvement
The barriers to effective, efficient oversight for biological-control organisms and products and resistant plants are inconsistencies in the existing framework of laws and regulations and the resultant overlap or lack of coordination among agencies with jurisdiction. In fashioning a more coherent approach to assessing the human and environmental risks of EBPM, there is a need to depart from current practice. Most important, the criteria for risk assessment must vary with the possibilities presented by each organism, product, or resistant plant.
To accommodate the demand for information based on experience as well as experimentation, oversight must draw on the resources of the scientific community and of field practitioners. Technical Advisory Groups (TAG), a consortium of representatives from government and research established to provide advice to the government on the safety of biological organisms for weed control, may be useful for oversight of supplements of EBPM.
The U.S. Environmental Protection Agency and the U.S. Department of Agriculture, currently responsible for oversight of pesticide regulations, should develop and publish a guide to risk criteria, data requirements, and oversight procedures that apply to importation, movement, introduction, testing, and release or registration of biological-control organisms or products.
A regulatory road map would be a first and important step toward reducing the uncertainties and delays caused by gaps and inconsistencies in current regulatory treatment of biological-control organisms. Such a road map would be particularly valuable now in the early stages of development of a broad-based industry producing biological-control organisms and products and resistant plants. Potential developers of new controls especially those in academic settings may be unfamiliar with oversight requirements and procedures, and an available road map would reduce the costs of acquiring that knowledge. As discussed previously, the costs of complying with oversight requirements can be an important determinant of commercial viability of biological pest control with small poten-
tial markets. Even today, the slow pace of registration may be at least partially attributable to the significant costs of gaining regulatory approval.
Streamlining the oversight process would enable those involved in research, development, and use of biological-control organisms and products to anticipate the requirements of public evaluation of product efficacy and human and environmental health risks. Those involved should be able to identify, through an integrated road map or similar mechanism, the agencies with authority for the specific organism or product to be evaluated. The road map should also provide detailed guidance through the procedures and requirements. Where procedures have associated timetables, these should be spelled out. Through the development of a logistical or procedural road map, an individual or group would be able to identify the full process at the outset, and would also be able to assess the data requirements, criteria for evaluation, and costs. In addition, an estimate of the time period necessary for completion of the evaluation process would be available to assist in planning and prioritization of projects.
This committee visualizes EBPM as the foundation for an approach to not only managing but also to assuring the durability of biological-control organisms, biological-control products, and resistant cultivars. Principles of ecology that lay the foundation of EBPM must be incorporated into implementation and oversight. An important goal of EBPM is to restore and preserve balance to the managed ecosystem by duplicating natural processes to the maximum extent possible. Risk assessment also should reflect that principle. Biological-control products must be developed and implemented in ways that complement managed ecosystems and facilitate the biological and natural controls already existing to suppress pests. Monitoring of new products and processes is key. The resultant information will lead to early identification of durability problems. The knowledge gained from this monitoring of the dynamic interactions among organisms will increase the understanding needed to manage old and new pests in a safe, profitable, and durable way.