Genetic Engineering of Trees to · · 2008-09-19Genetic Engineering of Trees to ... pest outbreak Kenneth F. Raffa ontrol of insect pests has of- ten been suggested as a prom- ising

Genetic Engineering of Trees to

Enhance Resistance to Insects

Evaluating the risks of biotype evolution and secondary pest outbreak

Kenneth F. Raffa

ontrol of insect pests has often been suggested as a prom- ising use of the emerging

gene-insertion technology. The most commonly proposed strategies would improve microbial pathogens of insects, incorporate defects into pest populations, and transfer genes that encode for insect-resistance properties into plants. Although some advantages and risks of these strategies have been discussed in general terms, there is little conceptual framework for applying theoretical principles to specific policy decisions (Simonsen and Levin 1988). In this article, I develop an approach to estimating risk by concentrating on one strategy-genetic transfer of resistance properties into otherwise susceptible hosts (Vaeck et al. 1987)-in one commodity frequently cited as a ben- eficiary of biotechnology-wood products (Dandekar et al. 1987, Far- num et al. 1983, Sederoff and Ledig 1985). Such an approach should improve our ability to evaluate and reduce the potential for inadvertent deleterious alteration of forest ecosystems. It may serve as a model for other transgenic applications.

I focus on one potential danger, biotype evolution, which is the selection for insect populations that can tolerate the new resistance property (Gould 1988). I do not consider here

Kenneth F. Raffa is an associate professor in the Entomology and Forestry Depart- ments at the University of Wisconsin, Madison, 53706. His research focuses on tree-insect interactions. ? 1989 American Institute of Biological Sciences.

In general, the risks are greatest in large

forested expanses

the impact of reduced insect populations on wildlife and nutrient cycling, problematic escaped plant material, and the great value of these technol- ogies as research tools.

Rationale for genetic enhancement of tree resistance

Much of the impetus for transferring foreign genes into trees arises from the unsuitability of traditional pest control tactics. For example, use of insecticides, a staple of insect control in many annual crops, is limited in forestry, where it is often infeasible, ineffective, or environmentally unacceptable. Application costs can be high because forests occur in im- mense, often inaccessible tracts; trees require a large amount of vertical coverage; and protection must extend over many growing seasons, rather than just one. Moreover, forests are often intended for multiple uses such as recreation, watershed management, and grazing, in addition to wood production.

Most forest insect management programs are based on a combination of silvicultural (stand density, rotation schedule, species-age composition, and site selection) and biological controls. Although often effective, these tactics also have limitations. In

many cases, environmentally sound and entomologically efficacious rec- ommendations cannot be adopted because they are not cost efficient.

Breeding for resistance against insects has not been pursued in forestry to the same extent as in agriculture. Although heritable resistance to some pests has been identified, operational problems have precluded implementation. For example, difficulties associated with breeding large plants, which require long periods before sexual maturity, can be prohibitive and pose formidable barriers to a scientist's career. Adding to this problem is the legacy of nineteenth- century high-grading, the cutting of only superior trees, which left defec- tive and pest-infected specimens as the genetic base of present forests (Barrett 1980).

Genetic improvement of trees, whether by traditional methods or biotechnology, can be highly compatible with silvicultural and biological insect control tactics. The attractive- ness of using gene alteration to accel- erate tree improvement has also been heightened by the conversion to more intensive tree management. When most forests were relatively unman- aged, insect feeding did not necessarily translate into commercial losses because companies simply purchased tracts as they matured rather than actively cultivating trees. However, intensively managed plantations, seed orchards, and energy farms comprise more sizable grower investments. These growing conditions also favor the survival and reproduction of some insects, just as in agriculture (Scho-

BioScience Vol. 39 No. 8

-

524

walter 1985). The Technical Associa- tion for Pulp and Paper Industries predicted that genetically engineered pest resistance would provide the greatest improvement to be gained by biotechnology over current practices, with significant impact within 3-5 years (Stomp 1987).

Genetic engineering, however, could also impose novel and unprecedented selective pressures on forest insect populations. The anticipated efficiency and efficacy, as opposed to the practical limitations of insecticides and traditional resistance breeding, pose major risks that have not yet been addressed in forest ecosystems. Resistant insect biotypes cause, at the least, a loss of efficacy and product failure. However, some resistant biotypes could cause greater problems than existed before the novel genes were deployed. Because trees serve as both commercial and natural re- sources, and because naturally regenerating forests located near planted stands comprise major terrestrial ecosystems, these risks must be addressed before large-scale outplantings can be deemed judicious.

General principles of biotype evolution

Generally accepted principles have emerged from several well-established disciplines, particularly the study of pesticide resistance, crop breeding, and plant-insect coevolution, that can be applied to tree gene manipulation. First, there is no physiological mode of insecticide action, if applied with sufficient intensity, that cannot be overcome by insect populations. Syn- thetic insecticides include a wide variety of exotic molecules that the target organisms did not previously encounter in their evolutionary histo- ries. Yet resistant races have emerged against all of them, often in manners that confer cross-resistance to related, and even unrelated, chemicals (Croft et al. 1982). So one cannot argue that genetically engineered resistance is immune to counteradapted biotypes because genes can be introduced from organisms unrelated to the host plant.

The pattern and intensity of selective pressures, rather than the actual mode of toxicity, most strongly af- fects the emergence of resistant races (Brattsten et al. 1987, Tabashnik and

Red pine defoliated by Neodiprion serti- fer. Photo by David Hall.

Croft 1982). Factors such as fre- quency of application, refugia among wild hosts, and persistence of additional selective pressures such as natural enemies are critical. Coevolu- tionary theory reinforces this lesson. In natural ecosystems, long-lived trees maintain defensive capacity against insects, despite the enormous differ- ences in their generation times, partly because of the conflicting and varying selective pressures imposed by the ag- gregate environment (Edmunds and Alstad 1978, Fritz et al. 1986, Raffa and Berryman 1987, Whitham 1983).

Second, there are numerous examples in which new or more severe pest problems than occurred before treatment have arisen (Forgash 1984). A common sequence is reduction of the target insect's natural enemies, fol- lowed by target insect resistance, fol- lowed by unchecked damage. Natural enemies usually evolve resistance more slowly than the target (Croft and Strickler 1983), and so new pesticides must be employed. The result is a classic pesticide treadmill. In- creased target-pest problems can also follow the introduction of resistant plant varieties. For example, adaptations to resistant cultivars of a preferred crop species can enhance a pest's ability to attack normally less- susceptible crops (Gould 1979).

Cross-resistance can take unpredictable forms, including immunity against currently effective methods. For example, Gould et al. (1982) demonstrated that mite adaptations to resistant plant cultivars can de- crease their susceptibility to organo- phosphates.

Previously innocuous nontarget insects can also be elevated to pest status through natural-enemy elimination, competitive release, and sometimes direct physiological benefit. Mite populations commonly rise after pyrethroid or DDT applications to which the target pest and mite natural enemies, but not the phytophagous mites, are susceptible. For example, the only major outbreak of the spruce spider mite, Oligoncyhus un- unguis Jacobi, in natural forests fol- lowed aerial DDT sprays against the western spruce budworm, Choristo- neura occidentalis Freeman (Furniss and Carolin 1977). Likewise, new plant cultivars have inadvertently created new pests by altering existing host-insect relationships (Oka and Bahagiawati 1984, Pathak 1975).

In some cases, the desired plant property, such as high yield or resistance to another pest, bestows added benefit to another herbivore in the form of direct nutritive properties or toxicity to parasites (Campbell and Duffy 1979). For example, the green revolution introduced several high- yielding cultivars that resulted in new pest complexes (Harlan 1980, Pathak 1975). Naturally coevolved systems also provide numerous examples where host defensive chemicals favor adapted herbivores by repelling predators (Eisner et al. 1974), inhibiting gut pathogens (Andrews et al. 1980), or providing a food base (Bernays and Woodhead 1982). Therefore, identi- cal standards must be applied to trees genetically engineered for properties other than pest resistance, as any alteration of the host is likely to affect the selective pressures on closely adapted herbivores.

Third, biotype formation is not an occasional aberration, but rather an inevitable outcome of certain conditions. More than 428 arthropod species developed insecticide resistance by 1980 (Geourghiou and Mellon 1983), and 24 of the 25 major agricultural pests in California comprise either secondary pest outbreaks or

September 1989

I

525

Table 1. Mechanisms by which insect evolutionary responses to genetically altered trees could deplete efficacy, contribute to more severe problems with target pests, and create new pest problems. Each mechanism has been observed in response to insecticides and/or new plant cultivars. +: Outcome that could result from a particular mechanism. See text for explanation and

examples.

Outcome

Loss of Aggravated target Nontarget pest Mechanism efficacy pest problems emergence

Physiological adaptation No cross-resistance with current

tactics + Cross-resistance with current

tactics + + +

Cross-resistance between introduced and existing defenses in:

Primary host + +

Secondary hosts + + +

Natural enemy suppression by toxin + + +

Altered host availability patterns after

biotype evolution:

Primary pest threshold surpassed + +

Predisposition to secondary pests + - +

Competitor elimination -- +

Selection for host-preference shifts -+ +

pest resurgences facilitated by insecticides (Forgash 1984, Metcalf 1980). Immune biotypes have also emerged against such control practices as Ba- cillus thuringiensis (McGaughey 1985) application and alternate crop rotation (Krysan et al. 1986).

The potential mechanisms of insect response to and adverse consequences of genetic engineering in trees, and the mechanisms by which each could arise, are summarized in Table 1. Tree longevity intensifies these risks, because a single host generation may span several hundred insect generations, a full order of magnitude more than is normally required for insect biotypes to emerge against insecticides (Forgash 1984, Metcalf 1980) and resistant cultivars (Sosa 1981).

Given these possibilities, and the complexities of biological systems outside controlled laboratory conditions, one alternative is to conclude that the effects of tree gene manipula- tions simply cannot be predicted, and therefore implementation should be avoided. However, this carte blanche disapproval is itself fraught with risk: development and application will continue with or without the input of ecologists, and general reservations will simply not carry much weight as legal decisions are rendered, financial markets are explored, and ordinances from international to local levels are

modified to compete for sources of jobs and revenue. Also, molecular biologists have been subjected to such a cacophony of hypothetical worst-case scenarios that specific concerns must be detailed, lest real dangers be likewise summarily dismissed. By sug- gesting ways of using ecological factors to lessen the risks of biotype evolution, forest ecologists can provide some direction to future decisions and research needs. Without such input, single-strategy approaches become more likely, and these approaches are surely the most detrimen- tal.

The approach suggested here con- sists of developing the ability to char- acterize specific host-pest targets ac- cording to general levels of risk, devising tactics for reducing the chances of target and nontarget biotype evolution in systems deemed to have an acceptable level of risk, and initiating long-term strategies for fos- tering the environmentally safe use of plant genetic engineering in forestry.

Selection of target systems

Experiences with insecticides and resistant cultivars indicate that detailed knowledge of each target host/pest system is required before judicious decisions can be made. However, the enormous diversity of insect biolo-

gies, host physiologies, and tree- growing conditions makes a case- by-case appraisal unwieldly and undirected. Therefore, a framework for transferring general principles to specific evaluations of risk is pre- sented. Four factors are considered: the availability of local refugia for susceptible insect genotypes, the existing role of host defenses in the pest's population dynamics, the compatibility of novel genetic defenses with alternative management practices, and the ability of novel genes to be transferred to plant progeny.

Refugia consist of untreated plants, plant parts, or times in which the herbivore can successfully feed and develop to maturity without exposure to the novel trait. Because area-wide, consistent selection pressures acceler- ate biotype formation, whereas spa- tially disrupted, intermittent exposure favors preservation or restoration of previous gene frequencies, an abundance of local refugia is critical. Ap- plication of this concept under actual field conditions is complicated, however, because refugia for susceptible genes are determined by the target insect's biology, host physiology, plant community structure, application patterns, and various interactions thereof (Table 2).

Mere proximity between different host types is not sufficient to preclude race formation (Bush 1973). Precau- tions must be taken to avoid reproductive isolation, and susceptible genes can only be preserved in systems where there is a high likelihood that some insects will locate untreated suitable hosts within their lifetime and interbreed with exposed individuals.

Host plant physiology, distribution, and variability strongly affect the population dynamics, behavior, and gene frequencies of herbivorous insects (Alstad and Edmunds 1983, Berryman 1976, Raffa and Berryman 1983, 1987). The consequences of genetic engineering on these naturally occurring constraints must be considered with respect to survival and reproductive rates by new insect races or nontarget species released from competitors and natural enemies. For example, if a counteradaptation also conferred cross-resistance to existing plant defenses, the consequences would be most severe in systems

BioScience Vol. 39 No. 8 526

where these defenses strongly regulate current insect population behavior. The specific effects and modalities of novel, relative to existing, host properties and the interactions between insect population and plant defense thresholds must also be considered.

Compatibility with other pest management practices is essential because multiple, conflicting selective pressures imposed by the insect's overall environment reduce the chances of biotype evolution. An example of the stability conferred by conflicting selective processes in coevolving systems can be seen in wild potatoes that emit aphid alarm pheromones upon attack (Gibson and Pickett 1983). Presumably, tolerance to the plant defense would result in decreased es- cape from predators. Integrated pest management will be hindered in systems where the novel traits' mode of action is similar to other tactics. For example, consider widespread out- planting of trees with Bt delta- endotoxin. Once target pests became adapted, they could no longer be controlled by judicious spray applications that are currently employed only when populations are high enough to threaten tree health.

Altered gene expression must be limited to planted individuals through sterility or other means. Without this stipulation, the level of risk will be greatly raised. Expressive progeny would disrupt the required pattern of refugia. Also, the protection of multiple host generations by an introduced gene could reduce the selective pressures on trees for existing defensive traits by rendering metabolically expensive (Mooney et al. 1983) alle- lochemicals competitively unfit. Once biotypes evolved against the novel trait, forests could then become extremely susceptible to attack and require unprecedented levels of human protection. This process could dupli- cate the inadvertent removal of resistant genes from food crops during selection for desirable agronomic traits (Harlan 1980), with the result- ing heavy reliance on synthetic insecticides.

The availability of refugia for susceptible genes, importance of current plant defenses in insect population behavior, and compatibility with alternative management tactics are integrated into a common model in

Table 2. Specific properties of target insect-tree systems that may influence the likelihood and severity of heritable insect responses to genetically engineered trees. For each property, conditions deemed more likely to yield reduced levels of risk are proposed.

System properties

Refugia: Opportunity for nonadapted insect survival Plant community structure

Number of untreated trees Insect life history

Mobility Mating site Number of trees oviposited by female Fecundity

Plant physiology Distribution and expression of trait

Insect life history *Plant community structure Available acceptable species

Insect life history *Plant physiology Number of insect generations/tree growing cycle

Insect population regulation Effect of existing host defenses on insect reproduction Effect of novel host plant defense on insect Action of existing and novel defense mechanisms

Compatibility of novel trait with other pest management methods

Biological control Host range of major natural enemies Exposure of natural enemies to novel trait Tolerance of natural enemies to novel trait Dispersal capability of natural enemies

Silvicultural control Optimal insect strategy for coping with novel plant

trait and patterns of suitable host availability Feasibility of insect removal by sanitation

Insecticidal control (synthetic and microbial) Availability of cost-effective sprays Mode of action of directly applied and plant-

incorporated toxins *Interaction between plant and insect properties.

Figure 1. By applying the specific traits for each system as outlined in Table 1, potential targets can be situ- ated within general zones of risk. Po- tential targets will be addressed in a descending hierarchy, in which the cropping system is considered first. Within each cropping system, some major pests will be evaluated with emphasis on the parameters shown in Figure 1. The population behavior and management considerations per- taining to each pest are discussed in turn. The objective is not to address every important pest, but rather to provide examples of how specific cases can be evaluated.

Initially, biotype formation is considered for the target pest alone. This constraint is subsequently relaxed in the discussion on nontarget pest emergence. In practice, this distinc- tion is arbitrary, because the tactics in

Attributes reducing risk

High

High Distant from host

Many Low

Varied

High

Low

Low Nonlethal? Dissimilar

Broad Low High High

Conflicting High

High

Dissimilar

the latter discussion must also be applied to each target pest.

Growing conditions Commercial forests. Large commercial forests, consisting of either naturally regenerating stands or plantations, comprise a major portion of wood and fiber production. Planta- tions are usually even-aged monocultures derived from seed orchards, whereas the more expansive self- regenerating tracts vary in species, age, and genetic diversity, depending on their ecological status and logging practices. Ownership is distributed among government agencies, large corporations, and regional timber companies. Even though the total value of this resource is enormous, per-acre profits are low, and so traditional methods of insect control are

September 1989 527

GROWING CONDITIONS, INSECT POPULATION BEHAVIOR, AND POPULATION MANAGEMENT FACTORS AFFECTING LIKELIHOOD AND IMPACT OF INSECT BIOT

EVOLUTION TO GENETICALLY ENGINEERED TREES

Scarce . AVAILABILITY OF LOCAL REFUGIA ? Abu

\ EDUCED ._

MODERATE

SEVERE

o intense weak

Population Regulation by Host Defenses

V EDUCED

MODERATE

SEVERE

intense weak

Population Regulation by Host Defenses

intense

Population Reguli by Host Defens

Figure 1. Conceptual framework for evaluating the relative risks of insect evolution against genetically engineered trees. Each growing system is characte a relative availability of local refugia in the form of untreated plants, plant pa times. Availability of local refugia is determined by growing conditions, stand st and insect properties such as vagility, mating behavior, and host range. The lik of biotype evolution is increased in systems where the insect population is , regulated by host tree properties and where transgenic resistance can dimii effects of other management tools and selective pressures. The manifold sel zones of severe, moderate, and reduced risk varies along the refuge ava continuum. Note that favorable attributes of any two parameters do not super unacceptable level of the third.

not cost effective in these systems. Likewise, insecticides pose greater en- vironmental hazards in large forests than in other tree-growing systems, because uses such as watershed management and recreation may exceed the importance of wood products. Although these factors lend support to genetic engineering in large forests, widespread, uninterrupted deployment of novel genes would post a high likelihood of biotype evolution. The availability of untreated refugia could be extremely small, because there would be little gene flow between exposed and unexposed insects, and harvest intervals are length- ier than for all other conditions except landscape ornament. There- fore, special precautions to preserve susceptible insect genes must be employed in these high-acreage commercial systems.

Short-rotation intensive cultivation. Recent advances in the development of fast-growing species, such as Pop- ulus and Salix, may lead to revolu- tionary changes in the tree-growing industries. In addition to providing traditional wood products, agrifor-

estry promises to be a major so energy. Selected cultivars have tremendous capacity for biom< duction (Ek et al. 1983), and r cycles of only 1-10 years hav proposed. Currently, these tr typically planted on low acrea der intensive management, ar may yield high cash values.

Short rotation times and ir cultivation methods are well su the deployment of spatial and te genetic mosaics. By alterating ant foreign genotypes, mixing varieties, and integrating plan tance with biological, cultur chemical controls in areas wh treated hosts are abundant, the < of biotype formation can be red managed wisely, genetic engi may even help preserve existin that confer protection from insel current intense selection and plasm manipulation of fast-g hardwoods poses the challe achieving superior production ties without sacrificing pest re: (Harrell et al. 1981), as happl food plants. Gene transfer n may prove most efficient at cor such multiple traits.

Seed orchards and high-value planta- YPE tions. Seed orchards are generally low- acreage, high-value, intensively managed sites. They are often surrounded by

indant large forests which could provide a substantial reservoir of susceptible, intermingling insect genotypes. Because even low levels of insect feeding cause severe

EDUCED economic losses, biological controls are not always satisfactory. Likewise, cul-

TE tural remedies may conflict with growing practices, just as they often do in agriculture. Because seed orchards must produce a consistently high yield, they

weak provide a plentiful, predictable resource a to specialized herbivores that have

,es?o evolved high reproductive capacity as an adaptation to a naturally scarce, unpredictable food supply. High-value planta-

biotype tion trees (e.g., Christmas trees and ve- rized by neer) pose approximately the same level lrts, and of risk as seed orchards. :ructure, :elihood Ornamental trees. Trees grown for strongly ornament and shade are among the rush the ,arating most valuable on a per-individual ba- ilability sis, with large lawn trees contributing rcede an greatly to real estate values (Payne et

al. 1973). Extensive monocultures are not common: usually an age and species mosaic is strewn across the sub-

urce of urban and urban landscape, through- shown out which untended woodlots are

iss pro- scattered and around which larger otation forests often occur. re been Several features of this system ees are could provide for the critically needed ges un- susceptible-insect gene refugia. First, id they the outlying regions from which ur-

ban pests often arise could be left itensive unaltered, with alternative methods ited for being used in the forests. Second, ur- .mporal ban parks and woodlots could be left

resist- untreated and municipal foresters clonal could continue current practices.

it resis- Third, not all homeowners would al, and choose any one variety of any one ere un- species. Introduction of foreign genes chances is already well underway in this set- luced. If ting, as some homeowners plant Jap- neering anese white birch to avoid bronze g genes birch borer and Chinese elm to avoid cts. The Dutch elm disease. No biotype evolu-

germ- tion has been reported. Genetic engi- ;rowing neering in this context would simply nge of be a more efficient and directed ver- proper- sion of the same process. sistance ened in Insect groups nethods nbining Bark beetles. Bark beetles (Co-

leoptera: Scolytidae) are the most

BioScience Vol. 39 No. 8

E = . _

e c 0i-0

a ,

o_o o m o 0

528

damaging forest pests to mature trees (Coulson and Witter 1984). Silvicul- tural tactics can reduce losses, but no method is totally satisfactory. There- fore, these insects might seem a logi- cal group to suppress by gene trans- ferral.

Bark beetles, however, might cause even greater problems after biotype evolution than they do currently, because of their particular relationship with host plants. Individual trees within a host species vary greatly in their levels of resistance. All trees can resist low numbers of beetles by se- creting repellent and toxic resins, but high attack densities mediated by ag- gregation pheromones can exhaust these defenses (Raffa and Berryman 1983, Rudinsky 1962). Thus the interaction is quantitative in that the outcome is dictated by the interaction between tree defense capabilities and insect numbers, but it is also discrete in that colonization attempts result in either tree death or insect failure to reproduce. Moreover, a tree's threshold of resistance-the number of beetles required to overcome its defenses-is greatly diminished by envi- ronmental stresses, such as water def- icit, disease, lightning, and other insects.

Under most conditions, even the most damaging bark beetle species usually attack stressed trees (Rudin- sky 1962), because the ability to de- tect weakened hosts increases their chances of reproduction. If there is a sudden increase in susceptible trees due to an area-wide stress such as drought, however, the population can surpass a critical density above which beetle behavior changes (Berryman 1976, Raffa and Berryman 1987). Beetle densities can become high enough to overcome the resistance of almost all trees in the stand, even after the stress has been removed. Under these conditions, beetles func- tionally expand their own food supply by successfully attacking healthy trees, and devastating outbreaks occur. Thus bark beetle population levels relate closely to the vigor of their host plants (Berryman 1976). Natural enemies and temperature play important roles, but most forest managers associate beetle outbreaks with old trees, dense stands, and physical stress.

Effective bark beetle management

Neodiprion lecontei mating pair. Photo: G. Lintereur.

maintains conditions that confine populations to a scavenging existence by promoting stand vigor through early rotation, thinning, sanitation, proper site selection, and age mosaics. Because beetles compete with other beetles (not trees), we need to provide conditions under which the best beetle strategy is to orient away from healthy trees, and individuals that are not repelled by healthy trees are at a competitive disadvantage (Raffa 1988).

This stability could be reduced by deployment of a foreign gene conferring resistance. If all trees, stressed and healthy alike, expressed the resistance factor, the selective pressures on beetle populations would be enormous and unidirectional. Behavioral attributes that orient beetles to weakened trees would confer little advan- tage. The best chance of beetle reproductive success would be offered by combined physiological immunity to both the novel and current tree defensive traits. With the role of host plant resistance removed, no effective control strategies would remain. Al- though dissimilarities between natural and exotic plant defense mechanisms can reduce this possibility, they do not provide sufficient assurance.

Even if beetles only overcame the novel gene but not natural resistance mechanisms, damaging conditions could still result. An accumulated pool of previously protected, stressed trees would become available, allowing populations to rise above out-

break thresholds. A widespread, synchronous appearance of suitable hosts, as opposed to a chronic, scattered reservoir of stricken trees, greatly increases the likelihood of damaging bark beetle outbreaks (Raffa and Berryman 1987). The pattern would be similar to what has already proven to be a failed, now abandoned, fire control strategy: by suppressing small fires, managers al- lowed tinder to accumulate to unnat- ural volumes, and uncontrollable, devastating fires resulted.

Transgenic trees might also suppress natural enemy populations, because most arthropod predators and parasites of scolytids complete development within trees killed by the beetles. Once beetles evolved resistance to the novel property, not only would there be an enormous backlog of available hosts, but there could also be a scarcity of natural enemies. Al- though not the principal factors regulating scolytid abundance, these arthropods are important mortality agents, in whose absence tree losses would increase (Amman and Cole 1983). Moreover, adverse effects would probably spread beyond treated stands, as emigrants from outbreaks often damage neighboring forests. There are no effective measures against such high populations (Berry- man 1976).

Mosaic strategies cannot provide sufficient assurance that this threat would be alleviated. Leaving some trees untreated would mostly provide a reservoir of vigorous plants, and so would not maintain current selective advantages to beetles that orient to unhealthy trees. Biotype formation could theoretically be delayed by de- liberately stressing untreated trees, but this approach demands a level of precision that far exceeds both current scientific capabilities and anticipated operational fidelity. In an unpredictable environment of drought, lightning, and additional biotic agents, the optimal proportion of untreated stressed trees cannot be determined.

Defoliators. Defoliators comprise the second-most-damaging insect group. The spruce budworm, Choristoneura fumiferana Clemens, is generally considered the most important forest defoliator in North America (Coulson

September 1989 529

and Witter 1984, Morris 1963). Bal- sam fir, Abies balsamea L. Mill., is the most preferred and susceptible species (Mattson et al. 1983), although there may be geographic variation in these relationships. Populations normally remain low for several decades, as physical and biotic mortality agents offset reproductive gains. Outbreaks occur when there is an abundance of mature host trees combined with several consecutive years of hot, dry summers (Blais 1973, MacLean 1980). Once populations rise, outbreaks rapidly expand to most spruce species.

Intense, continuous selection by resistant trees would probably select for immune C. fumiferana biotypes, just as widespread DDT applications did previously (Randall 1965). However, behavioral and ecological attributes could possibly be used to lessen the chances of biotype formation against this pest. If factors encoding for resistance were incorporated into Picea but not Abies, then a large proportion of the population would complete development on untreated plants. Peri- odic outbreaks would continue to occur on the Abies refuge, but natural enemies would eventually cause population collapse. Protection of the more desirable Picea at the expense of the less desirable Abies would probably be acceptable to forest managers, based on the twofold difference in their pulpwood values (Peterson 1988). Because young trees seem less vulnerable to attack (Mattson 1985), the possibility of limiting gene expression to older trees should also be considered.

With such a scheme, genetically engineered resistance could possibly be integrated into current silvicultural and biological controls. Because most natural enemies of budworms do not directly interact with the plant, for example, there may be less exposure to the novel trait than with bark beetles. However, indirect effects require further investigation.

If efficacy were lost due to biotype evolution, C. fumiferana populations would not necessarily exceed their current virulence. Host defensive mechanisms mostly cause nonlethal effects, such as smaller size and reduced fecundity, rather than direct toxicity (Mattson et al. 1983). This relationship could possibly reduce the

selective pressures for cross-resistance, especially where silvicultural practices provided species, age, and variety mosaics.

One danger is that resistant trees, although not directly affecting natural enemies, would reduce their numbers by depleting the supply of C. fumiferana, and thereby allow a severe pest resurgence once biotypes evolved. Several factors, however, might reduce this danger. First, most of the major predators and parasites that maintain C. fumiferana populations at low levels between outbreaks are generalists (Morris 1963) that could subsist on alternate insect hosts. Sec- ond, the specialists that help termi- nate outbreaks are adapted to long periods of low C. fumiferana population densities. Third, unlike traditional pesticide treatments that con- tinually suppress natural enemies while the resistant herbivore population rises, a plant trait that does not directly harm beneficial insects could allow their populations to respond immediately to increased prey densities. Effective emergency measures, including a broad array of synthetic and microbial insecticides, are available to prevent major losses if outbreaks occurred despite the above factors (Schmitt et al. 1984).

Although the concept of conferring genetic resistance against C. fumiferana merits further consideration, a greater understanding of this insect's ecology and behavior is essential before deployment can be deemed relatively safe. It is critical that budworms feeding on Abies and Picea do not become reproductively isolated. The high dispersal ability and multiple-oviposition behavior of the spruce budworm lends some confidence to this strategy, but additional research should focus on gene flow and geographic variation (Hardy et al. 1983).

The specifics of the spruce budworm-balsam fir-spruce system cannot be generalized to all defoliators. However, refugia may be maintained for some other species by restricting genetic alterations to certain growing conditions. For example, the gypsy moth, Lymantria dispar L., causes widespread forest defoliation, but much of its economic damage occurs in urban and suburban settings. Therefore, a policy of deploying resistant genes for ornamental or short-

rotation production purposes, while practicing traditional and developing integrated pest management strategies in forests, could protect high- value trees yet exert only moderate selective pressures on the insect. Al- though large acreages would still be periodically defoliated, the actual economic and aesthetic impact would be greatly diminished. Again, the specific biologies of each insect must be considered, and the inability of female gypsy moths to fly dictates cau- tion (Table 2).

A similar strategy with novel genes limited to short-rotation intensive cultivations could be used against insects such as the cottonwood leaf beetle, Chrysomela scripta F., and the forest tent caterpillar, Malacosoma disstria Hubner, which can cause more severe economic losses in intensive cultivations than in extensive forests.

Root insects. The impact of root- feeding insects has been greatly increased by modern forest plantation practices (Schowalter 1985). Losses to these species, primarily weevils and white grubs, are generally low in mature stands. However, when new seedlings are established after harvest or reclamation, root injury due to larval feeding and/or adult stem gir- dling can devastate plantings. Natural enemies are valuable but inadequate, and applied biological controls have been unsuccessful. Likewise, there are no totally acceptable silvicultural remedies against species that feed on living roots. Detection is difficult, so soil-permeating, persistent insecticides such as lindane are sometimes applied. These chemicals have been banned for most other uses. Thus, if transgenic resistance could be employed against root insects without adverse effects, both tree production and environmen- tal safety would benefit.

Root-feeding insects may provide targets against which time-specific expression of resistance genes could reduce the chances of biotype evolution (Gould 1988, Raffa 1987). A better understanding of the relationship between host age and susceptibility is necessary, however, to devise appropriate tactics. It is not known, for example, whether losses are most severe in young stands because larger trees are more tolerant and can better


withstand feeding, if younger trees are more attractive or less able to resist attack, or if physical attributes of the soil created by a closed canopy simply reduce insect replacement rates.

If large trees are commonly exploited without suffering severe damage, then limiting resistance expression to young trees could provide refugia analogous to the fir trees left untreated for spruce budworms. If no such reservoir exists, biotype evolution would be likely. Moreover, more serious losses than currently occur could result. A synchronous large-scale surge of root insect populations rather than the current steady mortality to young trees would provide a vast sub- strate for species such as the pales wee- vil, Hylobius pales Herbst, that breed in dead tissue and then as adults girdle nearby live seedlings. If a substantial reservoir exists on mature trees but temporal relaxation of gene expression is not provided, the consequences could likewise be devastating. Increased root feeding on mature trees by adapted insects would increase susceptibility to bark beetles (Raffa 1988), which in turn could catapult scolytid populations across their threshold density and result in the massive outbreaks de- scribed previously.

Seed and cone predators. These species, primarily Lepidoptera, Co- leoptera, and Hemiptera, pose major limitations on seed orchard productivity because their feeding translates directly into yield loss. They are only vulnerable to insecticides for brief periods of their life cycles, so timing is critical for effective control. Because multiple pest complexes are the norm, seed orchard managers must sample with an array of pheromones, consider numerous action thresholds, and spray for each population peak.

The use of resistant trees to suppress seed and cone pests could prove more compatible with biological control than current methods. With fewer insecticide applications, natural enemy populations of both target and currently sprayed nontarget species may rise, and multiple selection pressures could be enhanced. Regions outside seed orchards should be left unmanipulated. Usually seed orchards contain a large array of host lineages, thus providing an underly- ing genetic diversity that would fur-

ther reduce unidirectional selective pressures.

The presence of neighboring untreated refugia is not sufficient to prevent biotype formation, however, as evidenced by numerous fruit orchard pests with similar biologies that evolved insecticide resistance (Met- calf 1980). Untreated conebearing trees must be included within each planting. The greatest risks would probably be with species such as Conopthorus that enter cones as adults and oviposit in only one or several trees, rather than most Lepi- doptera, which oviposit externally on the cones of many trees (Table 2). If biotypes arose in a seed orchard, there is no obvious mechanism by which neighboring forests would be threatened, as food scarcity would probably remain a major selective force in these areas.

Wood borers. These Coleoptera (Bu- prestidae, Cerambycidae), Lepi- doptera (Cossidae, Aegeriidae), and Hymenoptera (Siricidae) primarily colonize weakened trees. Forests can be protected by early stand rotation, thinning, judicious site selection, and sanitation. In urban environments, however, numerous stresses such as root compaction, pollution, and me- chanical damage render trees susceptible to wood borers. Thus, insects such as the bronze birch borer, Agrilus anxius Gory, are major pests of ornamental trees. Forest management strategies are not applicable to homeowners, and so insecticides are often required. These applications are expensive and only marginally effective, however, and insecticide drift poses major problems when large trees are sprayed in urban areas.

Widescale outplantings in commercial forests would pose some of the same risks as with bark beetles and are unnecessary. As with the gypsy moth, however, wood borers may provide targets where economic impact can be greatly reduced, with relatively low risk of biotype evolution, by limiting gene alterations to high- value trees.

Nontarget pest emergence and biotype-delaying tactics

In nature, all tissues of all tree species are exploited by a variety of insects.

Evaluating the influence of transgenic resistance on nontarget insects is extremely difficult. Because of their current nonpest status, we know the least about these insects. For most commercial tree species, the complete guild of insect herbivores has not even been cataloged (Niemela and Neu- vonen 1983).

Although we have insufficient knowledge to predict nontarget pest emergence, experience with pesticides and resistant cultivars and observa- tions from natural coevolved systems suggest some useful tactics. First, a mode of action that is relatively specific to the target organism and as distinct as possible from existing plant defense mechanisms should be selected. Although unpredictable forms of cross-resistance can occur, this approach would at least reduce the selection for coadaptive insect genes that provide immunity from both introduced and existing plant traits.

Second, expression of novel properties should be limited to the tissues and times at which the target feeds, and preferably expression should be induced by the target herbivore (Gould 1988, Raffa 1987). Naturally occurring tree-insect systems provide examples of how uneven phytochem- ical distribution can favor stability. Pines allocate resin acids to new but not old foliage, thereby protecting their photosynthetically most productive tissues while allowing herbivores to graze the less-valuable needles (Ike- da et al. 1977). Larvae that prefer the less-protected tissue presumably out- compete those not repelled by young needles, so physiological tolerance to resin acids has not evolved in most sawflies. Likewise, many plants limit the expression of defensive traits to critical periods of the growing season, certain age categories, or induction by herbivore activities.

The potential adverse effects of genetically engineered, whole-plant expression can be anticipated by a reexamination of the spruce budworm example, depicted earlier as a possibly safe target. If the defensive property inadvertently protected cor- tical tissue and thereby resisted spruce bark beetles, Dendroctonus rufipennis Kirby, then the processes de- scribed earlier as favoring bark beetle adaptations would operate. Although

September 1989 531

this species is usually only a moderate problem, D. rufipennis can undergo outbreaks under favorable conditions, and it has caused some of the most severe losses ever recorded for any forest insect (Furniss and Carolin 1977).

Third, genotype mosaics should be employed within each planting, providing mixtures of transgenic traits superimposed on mixtures of various seed sources. Untreated trees should be intermingled within each planting. This strategy could reduce the likelihood of oligophagous and polypha- gous herbivores emerging as new pests of formerly less-preferred trees by undergoing genetic shifts in behavior to avoid the novel defense (Gould 1984). Such genetic variation stabi- lizes potential outbreak species like the black pineleaf scale, Nuculaspis californica Coleman, in nature. Al- though this insect can develop within- tree biotypes that correspond to host genotypes (Edmunds and Alstad 1978), diversity within the host population provides a complex array of selective pressures that renders the adaptive value of various insect genes, such as those regulating movement and resource breadth, in opposition (Alstad and Edmunds 1983).

Fourth, we should adopt a major lesson from pesticide usage and devise comprehensive, preconceived biotype management programs (Brattsten et al. 1987). Insect monitoring is a critical component of this scheme. Both previous experience and theoretical models emphasize that tactics for suppressing biotype evolution are most effective before the newly favored al- lele becomes common (Brattsten et al. 1987, Tabashnik and Croft 1982). Both target and nontarget populations should be periodically appraised.

The population genetics of emerging biotypes should be thoroughly stud- ied, as gene frequencies and inheritance patterns are critical in determin- ing optimal suppression methods (Tabashnik and Croft 1982). Like- wise, the mechanisms of biotype im- munities should be characterized to help develop appropriate countermea- sures (Brattsten et al. 1987).

Developing strategies to minimize risks

The preceding analysis suggests specific tactics for estimating and reducing the risks of deleterious insect responses to transgenic trees. A scheme for integrating these actions in a co- hesive fashion is proposed in Table 3. However, implementation of appropriate tactics will require a broad, interdisciplinary approach as an inte- gral component of plant biotechnology. Several long-term strategies must be initiated.

First, specific guidelines on transgenic release that consider the chances of biotype evolution and nontarget pest outbreaks need to be established. Detailed, rigorous regulations already apply to laboratory and field practices with regard to human safety and/or accidental release (Brill 1985). Equivalent standards should be developed governing the impact of deliberate releases at the population and ecosystem levels. Replacing ad hoc decisions with established guidelines could also safeguard against the danger that each deployment result- ing in no apparent and/or immediate adverse effects will lead to acceptance of other uses that are less judicious.

Policies must be based on research specifically directed at insect evolutionary responses to transgenic plants

Table 3. Principal ecological and management strategies for reducing the risks and impact of biotype evolution associated with deployment of genetically engineered tree resistance.

Restriction of novel gene deployment to systems where multiple, opposing, and ephermeral selective pressures can be maintained

Implementation of biotype-delaying tactics at planting Mode of action of introduced trait distinct from existing defense mechanisms Temporal and spatial limitations on and herbivore induction of resistance trait expression Host genotype mosaics at multiple layers Intermingling with untreated trees

Biotype management Monitoring of field populations Genetic characterization: mode of inheritance, dominance

Physiological characterization of biotypes: mode of detoxification, development of biotype modality inhibitors

Integration with other forest protection practices and multiple forest resource uses

(Gould 1988). Research in this area is critically lacking. Laboratory models should be devised to test specific hy- potheses, such as those emerging from Figure 1 and Table 2. Comple- mentary studies on tree-insect interactions, insect population behavior and genetics, and community ecology should focus on nontarget pest emergence. Development of methods for restricting gene transfer (Bej et al. 1988) and limiting expression to specific tissues, times, and herbivore levels is a critical need requiring the skills of molecular biologists.

Heightened regulations always in- cur the risk of being counterproduc- tive. Stipulations could become so restrictive as to render genetic alterations impractical and/or unattrac- tive, thereby reducing the benefits this tool can bring to forestry. However, some of the systems where genetic engineering appears to have the high- est margin of safety comprise large, well-defined, and easily accessible markets. Therefore, ecological and commercial considerations are often, or can be made, compatible.

The recommendation to limit expressed resistance to planted stock, for example, is of obvious benefit to biotechnology companies. Likewise, species such as Salix and Populus, which are most suitable for intensive short-rotation systems and gene mosaics, have also proven to be particularly amenable to protoplast manipulation and genetic engineering. Extension of patent life should also be considered as an incentive for accept- ing such guidelines.

The attributes of some potential target systems are likely to demand such expensive safeguards that deployment is not practical. This outcome is not justification for applying less-restrictive criteria, however, because the potential consequences of an erroneous decision are too severe. This philosophy does not discrimi- nate solely against genetic engineering, but rather it is currently applied to such traditional tactics as importa- tion of biological control agents.

Second, integrated risk management programs that involve all af- fected disciplines must be developed. Guidelines regulating gene transfers must be compatible with overall forest resource management. For example, wildlife biologists may oppose a


particular approach because of its effects on insectivorous birds or other components of the food web. Like- wise, trees genetically engineered to resist foliar fungal pathogens could possibly inhibit endophytes that repel insects (Carroll 1988), thereby caus- ing nontarget insect outbreaks.

Finally, ecologists and plant protection specialists should become more involved in the training of molecular biologists. Although a proposed plant protection tactic may comprise a major scientific advance at the molecular level, its implementation could entail a quite primitive approach from a population perspective. Recent advances from all levels of biological organization must be integrated to enhance the efficacy and environmen- tal safety of pest management tools. More exposure to population genetics, population dynamics, and crop protection should be provided in the core curriculum of students intending to conduct plant genetic engineering. A historical context of previous tech- nological capabilities that outpaced ecological understanding, such as high-grading, calendar pesticide application, and total fire suppression, would help better prepare molecular biology students for the contributions they can make.

Conclusions

An approach has been developed for using the general principles of biotype evolution to generate specific esti- mates of risk with regard to genetic engineering in trees and to devise possible preventive tactics. The major criteria include the tree-cropping system, compatibility with other pest management techniques, and specific attributes of the target insect's biology. This approach may apply to other forms of plant genetic engineering, as well as to other biotechnolog- ical approaches to controlling insects. Based on these analyses:

* In some systems, biotype evolution poses a severe threat. Possible adverse effects include both decreased efficacy and alteration of existing plant-insect relationships so as to worsen current conditions. In other cases, genetic engineering could be more compatible with biological control than are current insecticide treatments or it could be used to provide

greater genetic diversity than traditional breeding methods.

* In general, the risks are greater in large forested expanses than in seed orchards, rapid rotation systems, and ornamental plantings.

* Expression of resistance should be nontransferable to host progeny.

* Genetic mosaics involving spatial, temporal, and herbivore-induced within-tree variation, multiple sources of resistance, treatment mixtures, and refugia of untreated trees can reduce risk. Natural systems provide valuable examples of stable tissue-protection strategies, and these systems should be emulated as models for ecologically sound transgenic tactics.

* Integrated biotype management practices that provide multiple and conflicting selective pressures, cou- pled with well-planned monitoring of and response to insect biotype emergence, could reduce risk. Introduced traits should be based on narrow modes of action that are as distinct as possible from existing host defense mechanisms.

* Biotype evolution needs to be considered for all gene transfers, regardless of their intended function. Likewise, intended resistance against insects must accommodate equivalent concerns from other ecological disciplines. Comprehensive, multi- disciplinary ecological criteria governing the release of genetically altered trees should be researched and instituted.

* Greater emphasis should be placed on the scope of biological variation, the history of pesticide use, ecological feedback, and crop-protection principles during the training of molecular biology students intending to develop transgenic plants.

Acknowledgments The critical reviews and helpful

suggestions of L. Brattsten, Depart- ment of Entomology, Rutgers Univer- sity; F. Gould, Department of Ento- mology, North Carolina State University; W. J. Mattson, USDA Forest Service; R. T. Roush, Depart- ment of Entomology, Cornell Univer- sity; M. Wagner, Department of For- estry, Northern Arizona University; B. McCown and D. Ellis, Department of Horticulture, University of Wis- consin; R. L. Giese, Department of

Forestry, University of Wisconsin; and S. Codella, K. Klepzig, and D. Robison, Department of Entomology, University of Wisconsin are greatly appreciated. This work was sup- ported in part by the University of Wisconsin-Madison College of Agri- cultural and Life Sciences.

References cited Alstad, D. N., and G. F. Edmunds Jr. 1983.

Adaptation, host specificity, and gene flow in black pineleaf scale. Pages 413-426 in R. F. Denno and M. S. McClure, eds. Variable Plants and Herbivores in Natural and Man- aged Systems. Academic Press, New York.

Amman, G. D., and W. E. Cole. 1983. Moun- tain pine beetle dynamics in lodgepole pine forests. Part II: Population dynamics. US For. Serv. Gen. Tech. Rep. INT-145.

Andrews, R. E., L. W. Parks, and K. D. Spence. 1980. Some effects of Douglas fir terpenes on certain microorganisms. Appl. Environ. Mi- crobiol. 40: 301-304.

Barrett, J. W. 1980. The northeastern region. Pages 25-66 in J. W. Barrett, ed. Regional Silviculture of the United States. John Wiley & Sons, New York.

Bej, A.K., M.H. Perlin, and R. M. Atlas. 1988. Model suicide vector for containment of genetically engineered microorganisms. Appl. Environ. Microbiol. 54: 2472-2477.

Bernays, E. A., and S. Woodhead. 1982. Plant phenols utilized as nutrients by a phytophagous insect. Science 216: 201-203.

Berryman, A. A. 1976. Theoretical explanation of mountain pine beetle dynamics in lodgepole pine forests. Environ. Entomol. 5: 1225-1233.

Blais, J. R. 1973. Control of spruce budworm: current and future strategies. Bull. Entomol. Soc. Amer. 19: 208-213.

Brattsten, L. B., C. W. Holyoke Jr., J. R. Leeper, and K. F. Raffa. 1987. Insecticide resistance: challenge to pest management and basic research. Science 231: 1255-1260.

Brill, W. J. 1985. Safety concerns and genetic engineering. Science 227: 381-384.

Bush, G. L. 1973. Sympatric race formation in Tephritidae. Pages 3-23 in M. J. D. White, ed. Genetic Mechanisms of Speciation in Insects. D. Reidel, Dordrecht, The Nether- lands.

Campbell, B. C., and S. S. Duffey. 1979. Toma- tine and parasitic wasps: potential incompat- ibility of plant antibiosis with biological control. Science 205: 700-702.

Carroll, G. 1988. Fungal endophytes in stems and leaves: from latent pathogen to mutual- istic symbiont. Ecology 69: 2-9.

Coulson, R. N., and J. A. Witter. 1984. Forest Entomology: Ecology and Management. John Wiley & Sons, New York.

Croft, B. A., and K. Strickler. 1983. Natural enemy resistance to pesticides: documenta- tion, characterization, theory, and application. Pages 669-702 in G. P. Georghiou and T. Saito, eds. Pest Resistance to Pesticides. Plenum, New York.

Croft, B. A., S. W. Wagner, and J. G. Scott. 1982. Multiple- and cross-resistance to insecticides in pyrethroid-resistant strains of

September 1989 533

the predatory mite, Amblyseius fallacis. En- viron. Entomol. 11: 161-164.

Dandekar, A. M., P. K. Gupta, D. J. Durzan, and V. Knauf. 1987. Transformation and foreign gene expression in micropropagated Douglas-fir (Pseudotsuga menziesii). Bio- technology 5: 587-590.

Edmunds, G. R. Jr., and D. N. Alstad. 1978. Coevolution in insect herbivores and coni- fers. Science 199: 941-945.

Eisner, T., Y. S. Johnessee, and Y. Carrel. 1974. Defensive use by an insect of a plant resin. Science 184: 996-999.

Ek, A. R., J. E. Lenarz, and A. Dudek. 1983. Growth and yield of Populus coppice stands grown under intensive culture. Pages 64-71 in E. A. Hansen, ed. Intensive Plantation Culture: 12 Years of Research. US For. Serv. Gen. Tech. Rep. NC-91.

Farnum, P., R. Timmis, and J. L. Kulp. 1983. Biotechnology of forest yield. Science 219: 694-702.

Forgash, A. J. 1984. History, evolution, and consequences of insecticide resistance. Pestic. Biochem. Physiol. 22: 178-186.

Fritz, R. S., C. F. Sackli, and P. W. Price. 1986. Competition versus host plant phenotype in species composition: willow sawflies. Ecol- ogy 67: 1608-1618.

Furniss, R. L., and V. M. Carolin. 1977. West- ern Forest Insects. USDA Forest Service Mis- cellaneous Publication 1339. Washington, DC.

Geourghiou, G. P., and R. B. Mellon. 1983. Pesticide resistance in time and space. Pages 1-46 in G. P. Geourghiou and T. Saito, eds. Pest Resistance to Pesticides. Plenum, New York.

Gibson, R. W., and J. A. Pickett. 1983. Wild potato repels aphids by release of aphid alarm pheromone. Nature 302: 608-609.

Gould, F. 1979. Rapid host range evolution in a population of the phytophagous mite, Tet- ranychus urticae Koch. Evolution 33: 791- 802.

1984. Role of behavior in the evolution of insect adaptation to insecticides and resistant host plants. Bull. Entomol. Soc. Amer. 30: 33-41.

.1988. Evolutionary biology and genetically engineered crops. BioScience 38: 26- 33.

Gould, F., C. R. Carroll, and D. J. Futuyma. 1982. Cross-resistance to pesticides and plant defenses: a study of the two-spotted spider mite. Entomol. Exp. Appl. 31: 175- 180.

Hardy, Y. A., A. Lafond, and C. Hamell. 1983. The epidemiology of the current spruce budworm outbreak in Canada. For. Sci. 29: 715-725.

Harlan, J. R. 1980. Origins of agriculture and crop evolution. Pages 1-8 in M. K. Harris, ed. Biology and Breeding for Resistance to Arthropods and Pathogens in Agricultural Plants. Texas A&M University Press, Col- lege Station.

Harrell, M. O., D. M. Benjamin, and J. G. Burkot. 1981. Evaluation of adult cottonwood leaf beetle, Chrysomela scripta (Co- leoptera: Chrysomelidae), feeding preference for hybrid polars. Great Lakes Entomologist 14: 181-184.

Ikeda, T., F. Matsumura, and D. M. Benjamin.

1977. Chemical basis for feeding adaptation of pine sawflies Neodiprion rugifrons and Neodiprion swainei. Science 197: 497-499.

Krysan, J. L., D. E. Foster, T. F. Branson, K. R. Ostlie, and W. S. Cranshaw. 1986. Two years before the hatch: rootworms adapt to crop rotation. Bull. Entomol. Soc. Amer. 32: 250-253.

MacLean, D.A. 1980. Vulnerability of fir- spruce stands during uncontrolled spruce budworm outbreaks: a review and discussion. For. Chron. 56: 213-221.

Mattson, W. J. 1985. Role of host plants in the population dynamics of the spruce budworm. Pages 124-125 in C. J. Sanders, R. W. Stark, E. J. Mullins, and J. Murphy, eds. Recent Advances in Spruce Budworms Research. Proceedings of the CANUSA Spruce Budworm Research Symposium, Bangor, Maine, 16-20 September 1984. USDS Forest Service, Washington, DC.

Mattson, W. J., S. S. Slocum, and C. N. Koller. 1983. Spruce budworm (Choristoneura fumiferana) performance in relation to foliar chemistry of its host plants. Pages 55-60 in R. L. Talerico and M. Montgomery, eds. Forest Defoliator-Host Interactions: A Comparison Between Gypsy Moth and Spruce Budworms. US For. Serv. Gen. Tech. Rep. NE-85.

McGaughey, W. H. 1985. Insect resistance to the biological insecticide Bacillus thuringiensis. Science 229: 193-195.

Metcalf, R. L. 1980. Changing role of insecticides in crop protection. Annu. Rev. Ento- mol. 25: 219.

Mooney, H. A., S. L. Gulmon, and N. D. Johnson. 1983. Physiological constraints on plant chemical defenses. Pages 21-36 in P. A. Hedin, ed. Plant Resistance to Insects. American Chemical Society Symposium Se- ries 208. American Chemical Society, Wash- ington, DC.

Morris, R. F., ed. 1963. The Dynamics of Epidemic Spruce Budworm Populations. Mem. Entomol. Soc. Can. 31.

Niemela, P., and S. Neuvonen. 1983. Species richness of herbivores on hosts: how robust are patterns revealed by analyzing published host plant lists? Ann. Entomol. Fenn. 49: 95-99.

Oka, F. N., and A. H. Bahagiawati. 1984. De- velopment and management of a new brown planthopper (Nilaparvata leguns Stal) biotype in North Sumatra. Indonesia Contribu- tions from the Central Research Institute for Food Crops Bogor 71: 1-33.

Pathak, M. D. 1975. Utilization of insect-plant interactions in pest control. Pages 121-148 in D. Pimental, ed. Insects, Science and Soci- ety. Academic Press, New York.

Payne, B. R., W. B. White, R. E. McCay, and R. R. McNichols. 1973. Economic analysis of the gypsy moth problem in the northeast. II. Applied to residential property. US For. Serv. Res. Pap. NE-285.

Peterson, T. A. 1988. Wisconsin Forest Prod- ucts Price Review. University of Wisconsin- Extension, Boltwood Edition, April 1988. Madison, WI.

Raffa, K. F. 1987. Devising pest management tactics based on plant defense mechanisms- theoretical and practical considerations. Pages 303-329 in S. Ahmad and L. B.

Brattsten, eds. Molecular Mechanisms in In- sect-Plant Interactions. Plenum Press, New York.

. 1988. Host orientation behavior of Dentroctonus ponderosae: integration of to- ken stimuli and host defenses. Pages 369- 390 in W.J. Mattson, J. Levieux, and C. Bernard-Dagan, eds. Mechanisms of Woody Plant Resistance to Insects and Pathogens. Springer-Verlag, New York.

Raffa, K. F., and A. A. Berryman. 1983. The role of host plant resistance in the colonization behavior and ecology of bark beetles. Ecol. Monogr. 53: 27-49.

.1987. Interacting selective pressures in conifer-bark beetle systems: a basis for re- ciprocal adaptations? Amer. Nat. 129: 234- 262.

Randall, A. P. 1965. Evidence of DDT resistance in populations of spruce budworm, Choristoneura fumiferana (Clem.) from DDT-sprayed areas of New Brunswick. Can. Entomol. 97: 1281-1293.

Rudinsky, J. A. 1962. Ecology of Scolytidae. Annu. Rev. Entomol. 7: 327-348.

Schmitt, D. M., D. G. Grimble, and J. L. Searcy, eds. 1984. Managing the Spruce Bud- worm in Eastern North America. USDA For- est Service Agricultural Handbook 620. Washington, DC.

Schowalter, T. D. 1985. Insect adaptations to disturbance. Pages 135-152 in S. T. A. Pick- ett and P. S. White, eds. The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, New York.

Sederoff, R. R., and F. T. Ledig. 1985. Increas- ing forest productivity and value through biotechnology. Pages 253-276 in R. Ballard, ed. Weyerhaeuser Scientific Symposium on Forest Potentials, Productivity, and Value. vol. 4. The Weyerhaeuser Co., Tacoma, WA.

Simonsen, L., and B. R. Levin. 1988. Evaluat- ing risks of releasing genetically engineered organisms. Pages S27-S30 in J. Hodgson and A. M. Sugden, eds. Planned Release of Ge- netically Engineered Organisms. Trends in Biotechnology/Trends in Ecology and Evolu- tion Special Publication. Elsevier, Cam- bridge, UK.

Sosa, I. 1981. Biotypes J and L of the Hessian fly discovered in an Indiana wheat field. J. Econ. Entomol. 74: 180-182.

Stomp, A. M. 1987. Views of forest biotechnology: report from a conference on tree improvement by genetic engineering. Pro- ceedings of the Technical Association of the Pulp and Paper Industry Pulping Confer- ence. TAPPI Press Publ., Washington, DC.

Tabashnik, B. E., and B. A. Croft. 1982. Man- aging pesticide resistance in crop-arthropod complexes: interactions between biological and operational factors. Environ. Entomol. 11: 1137-1144.

Vaeck, M., A. Raynaerts, H. H6fte, S. Jansens, M. DeBeuckeler, C. Dean, M. Zabeau, M. Van Montagu, and J. Leemans. 1987. Trans- genic plants protected from insect attack. Nature 328: 33-37.

Whitham, T. G. 1983. Host manipulation of parasites: within-plant variation as a defense against rapidly evolving pests. Pages 15-41 in R. F. Denno and M. S. McClure, eds. Variable Plants and Herbivores in Natural and Managed Systems. Academic Press, New York.


Documents

Genetic Engineering of Trees to · · 2008-09-19Genetic Engineering of Trees to ... pest outbreak Kenneth F. Raffa ontrol of insect pests has of- ten been suggested as a prom- ising