FILE: <BC-73

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Introduction

In order to improve success rates in biological control, an understanding of events in past successful introduction programs is essential Luck et al. (1999). Successful cases can be used to test hypotheses about predator/prey interactions, and develop criteria for identifying effective natural enemies. Bellows & Van Driesche (1999) review the analytical bases developed in the late 1980's to estimate total losses from parasitism. Thay state that, "Because the population of an insect stage typically begins to lose members through death or development to the next stage in the life cycle before the entire recruitment to the stage is completed, at no time are all members of the generation present to be counted. This idea is analogous to a sink partly filled with water (i.e., the population), into which water is flowing (recruitment) and from which water is draining (death or advancement to the next stage)..." "To construct a life table, we need to know the total numbers that enter a stage (in this analogy, the total amount of water entering the sink). What biologists typically measure, however, is the number of animals present per sample unit at points in time (which is analogous to the amount of water in the sink at any given time). Although it is true that the volume of water present at any time is determined by the moment-to-moment balance of cumulative influx minus cumulative outflow, if these latter quantities are not known, it is not possible to determine total inflow from even the most detailed set of observations on the quantity of water in the tank at fixed moments in time. What is needed is a continuous record of recruitment for the whole period over which animals enter the stage of interest for the generation. This can be achieved by measuring recruitment for a series of contiguous intervals spanning the whole period when recruitment occurs (e.g., Van Driesche & Bellows 1988)."

Bellows & Van Driesche (1999) continue, "When the goal is to assess not only how many insects enter a given life stage over the course of a generation, but also to determine how many of that number subsequently become parasitized, the problem is compounded because the basic problem discussed above now applies to two quantities that must be measured; i.e., the total number of hosts recruited and the number that subsequently become parasitized. The linkages between these value are both dynamic and complex..." "Although there are some systems in which biology and life history characteristics are such as to produce nondynamic systems not subject to these problems (for example, cases where the sampled stage is in a diapause stage and accumulates without loss as, for example, is approximately the case for gypsy moth eggs, because dead or parasitized eggs remain countable) or systems such as some leafminers in which lost insects continue to be traceable in samples through their remains, the majority of insects do have overlapping recruitment and losses. For these cases, densities and percentage parasitism values seen in samples do not measure adequately the level of parasitoid effect."

Approaches in the evaluation process include (1) life table analysis, which is a descriptive method, (2) stage frequency analysis, (3) direct measurement of recruitment, (4) deat rate analysis, and (4) experimental manipulations in the field. The primary goal is to determine whether regulation of the host population exists and to identify the agents responsible for regulation. Luck et al. (1999) defined regulation as the biological processes involving natural enemies that suppress prey or host densities below levels that prevail in the absence of natural enemies. It must be determined whether the populations are regulated, measure the level of regulation and identify the forces involved in regulation. If the populations are not regulated, if the regulation is intermittent, or if the level of suppression is inadequate, then other options to consider are (1) introduction of additional natural enemies, (2) inoculative or inundative releases, (3) development of plant resistance, (4) change the cultural practices, etc. There are other key references pertaining to measurement of natural enemy impact (Thompson 1955, Richards et al. 1960, Hafez 1961, Kirtitani & Nakasuji 1967, Manly 1974, 1976, 1977, 1989; Ruesink 1975, Russell 1987, Kolodny-Hirsch 1988, Schneider et al. 1988, Van Driesche 1988, Bellows et al. 1989, Gould et al. 1989, Keating 1989, McGuire & Henry 1989, Van Driesche et al. 1989, 1991a,b, Buonaccorsi, J. P. & J. S. Elkinton 1990, Gould, J. R. 1990a,b, Hazzard et al. 1991).

There is probably no single method which can provide conclusive evidence that natural enemies are regulating a population. Natural enemies are not the only factor involved in many interactions, and the plant can significantly affect the natural enemies' ability to regulate (Flanders 1942, Starks et al. 1972, Price ta al. 1980). Luck et al. (1999) conclude that no research method if free of technical problems, and management decisions are made with insufficient knowledge. Therefore research aimed at developing an integrated pest management program is a continuous process in which hypotheses are continually being refined and tested (Way 1973). Classical biological control and augmentive biological control are important IPM tactics, but they must be pursued and expanded to include situations for which they have not ben emphasized (DeBach 1964, 1974, Ridgway & Vinson 1976, Carl 1982). Indigenous biological control forms the foundation for pest management and therefore must be utilized if IPM is to become more effective. Its presence in an agroecosystem can be demonstrated by disrupting it with insecticides (Folsom & Brondy 1930, Woglum et al. 1974, Brown 1951, Pickett & Patterson 1953, Ripper 1956, Bartlett 1968, Smith & van den Bosch 1967, Wood 1971, Ehler et al. 1973, Eveleens et al. 1973, Croft & Brown 1975, Luck & Dahlsten 1975, Luck et al. 1977, Reissig et al. 1982, Kenmore et al. 1984), or by comparing unsprayed, abandoned orchards with treated orchards. Insecticidal disruption provides one of the best experimental techniques for evaluating natural enemies. It can reveal the amount of control provided by indigenous entomophages (Stern et al. 1959, Smith & van den Bosch 1967, Falcon et al. 1968, MacPhee & MacLellan 1971, Wood 1971, Flint & van den Bosch 1981, Jones 1982, Metcalf & Luckmann 1982, Kenmore et al. 1984).

In the experimental evaluation of biological control, testing whether regulation exists and which natural enemies are responsible for the regulation, life tables and their analyses provide a quantitative framework in which to explore the consequences of a predator/prey interaction and to generate hypotheses. However, life tables cannot demonstrate the efficacy of natural enemies in suppressing a host or prey population in the field; only experimental methods can do this (Luck et al. 1999). Some populations cannot be manipulated with available technology because they are based on untested assumptions. Evidence is that natural enemies suppress host/prey populations and experimental results suggest that a host plant's nutritional quality, its physical structure and its chemical defenses play a role in pest suppression (Denno & McClure 1983, Futuyma & Peterson 1985, Whitham et al. 1984, Mattson 1980).

The development of an appropriate sampling routine is essential for the evaluation of natural enemies. The design is determined by the objectives of the experiment, the biology of the organisms involved and the cost of acquiring the information to meet the objectives. The sampling procedure used to acquire data and the statistical techniques used to analyze data must be decided before field evaluation begins. Appropriate experimental designs require preliminary studies to identify variation sources. Preliminary samples can save time and resources (Green 1979). For example Legner (1979 , , 1983, 1986) and Van Driesche (1983) described some of the problems associated with estimating and interpreting percent parasitism from field samples, while Van Driesche & Bellows (1988) discussed analytical procedures for dealing with some of the problems. Statistical randomness is important in population sampling and in the assignment of treatments. Randomness includes locating field plots and selecting sample plants and sample units. Each sample unit must have an equal chance of being selected. Nonrandom sampling makes analysis of the data questionable because of the uncertainty associated with the estimation of the values. Texts and articles on sampling and experimental design should be consulted before an evaluation of natural enemies or of biological control is begun (Morris 1955, 1960, Cochran 1963, Stuart 1976, Elliot 1977, Jessen 1978, Southwood 1978, Green 1979).

Evaluation in biological control must consider the following: Do natural enemies affect pest population densities; what natural enemies kill a pest; how quickly will an natural enemy kill a pest; how many pests will a natural enemy kill; how does an natural enemy respond to changes in pest densities in the field; and how do environmental changes affect the predator-prey/parasitoid-host interaction (Luck et al. 1999).

When evaluating indigenous natural enemy populations, it is necessary to determine whether biological control of the hosts exists. An effective means compares pest densities in an area not treated with pesticides to pest densities in an area subjected to traditional pesticide practices. Ceasing the use of pesticides in parts of a field does not constitute a previously unsprayed area, as prolonged pesticide use reduces natural enemies and alternate prey or hosts upon which the natural enemies depend. Time is required to reestablish interactions between natural enemy and prey/host populations. Also, the untreated area must be large enough to buffer the plots from pesticide drift and to insulate arthropod populations within from the dynamics and interactions of those in the adjacent areas. Estimating the degree of regulation exerted by the natural enemies residing in plots subjected to disruptive effects almost always underestimates the amount of potential biological control (Luck et al. 1999). Pesticide trials in which a small untreated block within a sprayed area is used to estimate the amount of control from factors other than the pesticide treatments are not adequate in that populations in the unsprayed area are overwhelmed by the dynamics of those in the surrounding treated blocks.

Techniques For Evaluation

Introduction / Augmentation of Natural Enemies.--In classical and indigenous biological control, a prey population is expected to be self sustaining. Control derived from augmentive releases is only temporary, lasting one season or less. The evaluation of each method poses different problems. In Classical biological control a natural enemy's impact can be demonstrated by comparing the change in a pest's density in the initial release sites with a control site of similar characteristics but lacking the natural enemy (Huffaker et al. 1962, Legner & Silveira-Guido 1983). A drop in the pest's abundance in the release site compared with the control site suggests that the natural enemy is responsible for the pest's decline. This conclusion is further supported if the pest's density in the control site also declines following the subsequent introduction or immigration of the natural enemy to that site. Replication of release and control sites adds confidence to the evaluation if the pattern of decrease is consistent across the experimental plot. A similar design can evaluate augmentive releases, but the results may be confounded if closely related or morphologically similar indigenous and released natural enemies attack the same pest (see Legner & Brydon 1966). However, Oatman & Platner (1971, 1978) showed that release and control plots are never identical ecologically. Exclusion, inclusion or interference methods are required to assess the difference between resident and released natural enemies. Introducing genetically marked individuals that differ from the resident population only in the genetic marker can also distinguish between resident and introduced populations (Legner et al. 1990a, 1990b; Luck et al. 1999).

The translocation of natural enemies to areas invaded by pest species and subsequent classical biological control gives additional proof that indigenous natural enemies can have a significant role in regulation of native populations (Wilson 1960, Dowden 1962, McGugan & Coppel 1962, McLeod 1962, DeBach 1964, CIBC 1971, Greathead 1971, Laing & Hamai 1971, Rao et al. 1971, Clausen 1978, Luck 1981, Kelleher & Hulme 1984, Cock 1985). Further proof is given when the introductions are repeated at several locations with similar results (DeBach 1964, Laing & Hamai 1976).

Exclusion / Inclusion of Natural Enemies.--Cages and other barriers have been used in exclusion and inclusion procedures to evaluate natural enemies (Smith & DeBach 1942, DeBach et al. 1949, DeBach 1955, Sparks et al. 1966, Lingren et al. 1968, Way & Banks 1968, van den Bosch et al. 1969, DeBach & Huffaker 1971, Ashby 1974, Campbell 1978, Richman et al. 1980. Aveling 1981, Faeth & Simberloff 1981, Frazer et al. 1981b, Jones 1982, Elvin et al. 1983, Chambers et al. 1983, Linit & Stephen 1983, Barry et al. 1984, Kring et al. 1985). Cages to exclude natural enemies were first deployed by Smith & DeBach (1942), using paired sleeve cages to test whether the introduced parasitoid Metaphycus helvolus (Compere) regulated the black scale, Saissetia oleae (Bern.). Comparison of the black scale in the open and closed cages showed that less black scale survived in the open cages. This technique was modified by using insecticide impregnated netting to kill natural enemies that emerged in the closed cages when the methods was used to evaluate other classical biological control projects (DeBach et al. 1949, DeBach 1955, DeBach & Huffaker 1971).

Cages with different sizes of mesh have been used to exclude natural enemies based on their size (Campbell 1978, Kring et al. 1985). Three types employed were (1) a complete exclusion cage with small mesh netting and sealed at both ends, (2) a control cage with similar netting and open at both ends and (3) a partial exclusion cage with large mesh netting and closed at both ends. The latter excluded large predators but allowed access of small predators and parasitoids.

Sleeve and field cages with more complex designs, such as those which enclosed whole plants, accompanied by samples of the prey and natural enemy populations, showed that the spring increase of predators eliminated black bean aphid, Aphis fabae Scop., colonies on its overwinter host, Euonymus europaenus L., after June (Way & Banks 1968). If spring aphid populations had been dense on the tree, the predators that remained after the aphids emigrated to their summer hosts prevented recolonization of spindle tree by late fundatrices during the summer, even though the spindle tree was capable of supporting an increasing aphid population. Closed field cages covered with dieldrin treated netting coupled with hand removal excluded natural enemies from some spindle trees whereas open field cages constructed with slatted walls allowed access of the natural enemies to the aphids on the uncaged trees but provided the same degree of shading as the closed cage (Way & Banks 1958, 1968). Such experiments and making census of populations on the sample twigs document the importance of predators in excluding aphids from the overwintering host plant during the summertime (Luck et al. 1999).

The evaluation of indigenous natural enemies of cereal aphids was done in large field cages and accompanying population samples. The experimental design combined field cages erected at several intervals after the aphids immigrated into a winter wheat field. The growth rates and peak densities of the aphid populations within the cages was compared with those in several open plots of similar size (Chambers et al. 1983). Samples showed that the abundance of Coccinella 7-punctata L. was negatively correlated with aphid abundance in the open plots but the incidence of parasitism and disease was not negatively correlated with aphid abundance. These latter two factors were more common in the caged plots. If the difference between the aphid densities in the cage and open plots was converted to per capita aphid consumption, based on the sampled coccinellid densities, the calculated values were within the range of known values. Coccinellids appeared to be the key agents limiting the growth rate and peak abundance of cereal aphids during mid season but they were unable to do so early in the season (Rabbinge et al. 1979, Carter et al. 1980).

Field cages with open field controls were used to determine whether the predator complex aggregated at dense patches of the pea aphid, Acrythosiphon pisum (Harris) (Frazer et al. 1981b). The cages excluded the predators and allowed the aphid population to increase to about 5X that of the open control plots. When the cages were removed the aphid populations declined to the densities that prevailed in the control plots and the decline was correlated with increased predator numbers aggregating at the denser aphid patches. Large field cages have also been used to evaluate the potential of predators in cotton to reduce egg and larval populations of the tobacco budworm, Heliothis virescens (F.) (Lingren et al. 1968). Evening releases of budworm moths initiated the prey populations within the cages. Fewer prey survived in the cages with predators than in cages excluding predators. Similar studies were conducted in California cotton to evaluate predation on the survival of larval populations of the cotton bollworm, Heliothis zea (Boddie) (van den Bosch et al. 1969). The cotton plants within the predator-free cages were treated with an insecticide to eliminate resident predators before bollworm larvae were introduced. Significantly fewer prey survived in the untreated cages and significantly more predators were collected from the untreated cages.

In order to determine whether indigenous natural enemies or microclimatic changes within a cage explained the increased survival of caged European corn borer, Ostrinia nubialis (Hübner) larvae, caged and uncaged plots and plots of similar size but enclosed with a cage within a cage were used (Sparks et al. 1966). The double cages was designed so that the screened panels on the inside cage were opposite that unscreened panels on the outside cage and vice versa. This arrangement allowed predators access to the plants inside while maintaining the same level of shading and air flow in both the complete cage and cage within a cage plots. Entomopathogenic fungi (Deuteromycotina) effects were also tested with cages for the black bug, Scotinophara coarctata F., in rice (Rombach et al. 1986a.). Adult bugs were introduced into screened cages and applications of fungi Beauveria bassiana (Bals.) Vuill, Metarhizium anisopliae (Metsch.) and Paecilomyces lilacinus Thom. were made with a backpack sprayer. The black bugs were significantly less abundant in all treatments when compared with untreated controls, with effects lasting to nine weeks. Similarly caged brown planthoppers, Nilaparvata lugens Stal, were treated with entomopathogenic hyphomycetes (Fombach et al. 1986b). Mortality from fungal infections ranged from 63-98% three weeks after application.

Ground predators, principally carabids, were excluded with trenches that contained insecticide soaked straw, from the cabbage root fly, Erioischia brassicae (Bouché) (Wright et al. 1960, Coaker 1965). Polythene barriers were used to exclude predators from two of three treatments in which the predator density was manipulated to determine its effect on the density of aphid populations (Winder 1990). Sticky bands around selected branches of a spindle tree were used to exclude the walking predators of Aphis fabae (Way & Banks 1968) and around the plant base to exclude walking predators of Trichoplusia ni (Hübner) (Jones 1982). Sticky circles around Trichoplusia ni eggs were used to exclude predators and parasitoids from attacking the eggs (Jones 1982).

Studies relating cage densities to the densities of resident field populations of predators outside the cages have been used for aphids and Lepidoptera (Frazer & Gilbert 1976, Campbell 1978, Aveling 1981, Frazer et al. 1981b, Chambers et al. 1983), providing useful hypotheses (Way & Banks 1968, van den Bosch et al. 1969, Campbell 1978, Carter et al. 1980, Aveling 1981, Faeth & Simberloff 1981, Frazer et al. 1981b, Chambers et al. 1983). Cages can provide quantitative information on predation rates (Elvin et al. 1983) but not without limitations. Small sleeve cages inhibit predator or prey movement and are good for experiments with sessile species or species with low vagility (smith & DeBach 1942). The abundance of citrus red mite, Panonychus (= Metatetranychus) citri (McG.), within sleeve cages was sometimes 12X greater than outside sleeve cages (Fleschner 1958) even though the mite population outside the cages was kept predator-free by continuous hand removal of predators. It was thought that the cage prevented the reproductive females from emigrating, that the microclimate within the cages favored rapid growth of the mite population, or both factors influenced population growth (Fleschner et al. 1955, Fleschner 1958).

It is not possible to identify which members of a predator/parasitoid complex are regulating a host population with exclusion cages unless the complex consists of one or a few species (Jones 1962). Partial exclusion cages may show whether small predators, pathogens or parasitoids regulate in the absence of large predators, but they cannot show whether large predators regulate prey in the absence of parasitoids or small predators (Luck et al. 1999). Cages may also inhibit predator or prey movement or interfere with natural enemy oviposition. Two leaf mining species on oak failed to reproduce within whole tree cages and a third species failed to reproduce in one cage (Faeth & Simberloff 1981). Aphid alates cannot emigrate from a cage, thus caged versus uncaged aphid populations may show differences in density because alate immigration reduces the uncaged aphid population. Some predator species aggregate at patches of high prey density in a numerical response (Readshaw 1973, Frazer et al. 1981a. Kareiva 1985). Such behavior may be inhibited by cage size because the spatial pattern in nature to which the predator species responds is larger than that present within the cage. Also, confining predators to a cage may causae them to search areas more frequently and thereby increases the likelihood that they will encounter prey. Under these conditions the predator may reduce prey densities to levels below normal, and in this way inclusion studies resemble laboratory experiments in which predators are confined with prey (van Lenteren & Bakker 1976, Luck et al. 1979).

Erroneous interpretations can result when prey are placed into a cage without consideration of their preferences for oviposition sites, their density and distribution patterns or their preferred feeding sites under field conditions. Some predators and parasitoids use kairomones to find their prey and hosts (Hassell 1980, Nordlund et al. 1981). Some kairomones are associated with feeding activity. Placing prey or hosts in new sites influences their risk of detection. Food quality may affect a phytophage's feeding time and increase its risk to predation because of the kairomones released while feeding (Nordlund et al. 1981). Detailed studies of a predator's searching behavior and capture rates and a prey's oviposition and feeding behavior are important (Fleschner 1950, Dixon 1959, Frazer & Gilbert 1976. Gilbert et al. 1976, Rabbinge et al. 1979, Carter et al. 1980. Baumbaertner et al. 1981, Frazer & Gill 1981, Sabelis 1981).

Whenever predator free controls are employed, it is difficult to exclude all predators, even when they have been treated with insecticides (van den Bosch et al. 1969, Irwin et al 1974, Elvin et al. 1983). Some predators may pass through excluding screens when in small developmental stages (Sailer 1966, Way & Banks 1968), or they are difficult to exclude because they become buried in the soil (Frazer et al. 1981a, Elvin et al. 1983). Cages also alter the microclimate through shading and inhibiting air flow. Exclusion and partial exclusion cages using terylene netting reduced the light intensity inside cages by 24-37% (Campbell 1978) and saran screen reduced solar radiation by 19% (Hand & Keaster 1967). Such shading required the use of a more shade tolerant cotton cultivar than was normally planted in the region (van den Bosch et al. 1969). Shading also affects plant physiology and thus may affect the plant's quality as a substrate for the host or prey population (Scriber & Slansky 1981). Temperatures within cages used in a corn borer study were 8-10°F lower than the temperature outside. The humidity fluctuated more moderately within and was 5-10% higher than that outside the exclusion cages (Sparks et al. 1966).

Solar radiation changes cause differences in leaf temperature by as much as 13°C (Hand & Keaster 1967). Leaf temperatures and moisture availability influence photosynthetic rates and evapotranspiration (Gates 1980). Leaf temperatures probably affect the behavior and feeding rates of phytophagous hosts and prey. Temperature related interactions between the growth rates of aphids and the searching rates of their predators are important (Frazer & Gilbert 1976, Frazer et al. 1981a). Screening also reduced wind speed within a cage by as much as 48% (Hand & Keaster 1967) which, depending on RH and wind velocity outside and inside a cage, influences the leaf's boundary layer within the cage (Gates 1980, Ferro & Southwick 1984). Instrumentation allows the monitoring of many of these effects but their influence on predator/prey interactions must be assessed (Luck et al. 1999).

Removal by Insecticide Treatment.--Natural enemy complex impact may be assessed through the application of insecticides. The method was first used to kill natural enemies of the long-tailed mealybug, Pseudococcus longispinus (Targ.), without affecting the mealybugs (DeBach 1946). Insecticides have been used to determine whether indigenous predator populations in cotton suppress populations of the beet armyworm, Spodoptera exigua (Hübner), and cabbage looper, Trichoplusia ni (Ehler et al. 1973, Eveleens et al. 1973). Early season insecticides applied to cotton were thought to interfere with natural controls (Ehler et al. 1973, Eveleens et al. 1973). Large blocks (3-4 square miles) were treated with an insecticide scheduled during early season, early and midseason and early, mid- and late season. A fourth plot served as an unsprayed control. Samples and observations showed that the absence of predators in the treated plots was correlated with the increased survival of beet armyworm eggs and first generation small larvae of the cabbage looper. The hemipteran predators, Geocorus pallens Stal, Orius tristicolor (White) and Nabis americoferus Carayon were implicated as the most important predators since they were the most affected by the treatments whereas Chrysoperla carnea Stephen was not so strongly affected. Insecticide treatment showed that the suppression of cabbage looper densities in celery arising from egg parasitism by Trichogramma spp. and predation of eggs and young larvae by Hypodamia convergens Guer. and O. tristicolor was sufficient to prevent economic damage before the production of the first marketable petiole in celery (Jones 1982).

Insecticides were also used to test whether the coccinellid, Stethorus sp. regulated the density of the two spotted spider mite, Tetranychus urticae (Koch), in a previously untreated apple orchard in Australia (Readshaw 1973). Two applications of malathion increased the density of the mite populations. Tetranychus urticae, unlike the predator fauna associated with it, was resistant to malathion. Stethorus regulated the mite population by numerically responding both aggregatively and reproductively to the denser mite patches. Even with insecticide disruption and stimulation of the mite reproduction (Chaboussou 1965, Bartlett 1968, van de Vrie et al. 1972, Dittrich et al. 1974), Stethorus was able to prevent the mite population from attaining an economic density of 100 mites/leaf on most trees.

The action of two parasitoids of the olive scale, Parlatoria oleae (Colvee), was evaluated using insecticides (Huffaker & Kennett 1966). This scale is bivoltine on olive in the San Joaquin Valley of Calviornia. One generation occurs during the autumn and spring and the second generation during summer. Aphytis paramaculicornus DeBach & Rosen and Coccophagoides utilis Doutt was introduced for biological control (Rosen & DeBach 1978). Aphytis dominated during the autumn and spring scale generation whereas Coccophagoides dominated during summer. Three DDT treatments were used to exclude the parasitoids: (1) a spring treatment to exclude Aphytis, (2) as summer treatment to exclude Coccophagoides and (3) a spring and summer treatment to exclude both parasitoids. Untreated trees were left as controls. It was thought that DDT residues on the foliage and twigs inhibited the parasitoids but did not affect the scale's reproduction and survival. Treatments which excluded Coccophagoides had higher scale densities than the untreated controls but lower densities than the treatments which excluded Aphytis. Treatments that excluded only one of the parasitoids had lower scale densities than treatments that excluded both parasitoids. Treatments also indicated that together the parasitoids provided better biological control than either did alone even though the mortality contributed by Coccophagoides was only about 5%.

Inoculation of fumigated (12 hrs with methyl bromide) and unfumigated poultry manure with Musca domestica L. eggs demonstrated 53.4 to 99.4% mortality in the presence of predatory and scavenger arthropods (Legner 1971). Significant negative correlations of parasitization with increasing host densities were explained by parasitoid behavior. Inherently, single female parasitoids without interference from other individuals of the same or different species respond positively with increases in host density; parasitization rates increase, which appears to be correlated with increases in the production of progeny (Legner 1967). However, when groups of parasitoids concentrate their search among several host pupae, as is common in nature, their efficiency per female is decreased through mutual interference, that apparently involves combinations of physical interruption and chemical effects. There was some evidence that female parasitoids were strongly attracted to denser concentrations of their hosts in their habitat (e.g., Legner 1969), which evidence further tends toward increases in the interference factor at natural high host densities. Furthermore, any interference that would deter some female parasitoids from oviposition during the first few days of adult life would lower fecundity and longevity (Legner & Gerling 1967). Operating collectively, these several forces would tend to produce the observed apparent negative correlation between parasitization and host density.

Several problems are associated with interpreting results from an insecticide treatment, however. The pesticide may stimulate reproduction of the prey population. There may be a pesticide induced sex ratio bias, and pesticide induced physiological effects on the plant may arise. Mites that are exposed to sublethal doses of some pesticides are stimulated reproductively and occasionally even increase female biased sex ratios (Charboussou 1965, Bartlett 1968, van de Vrie et al. 1972, Dittrich et al. 1974, Maggi & Leigh 1983, Jones & Parrella 1984). Such effects may also extend to aphids (Bartlett 1968, Mueke et al. 1978), and delphacids (Chelliah et al. 1980, Reissig et al. 1982). Differential mortality resulting from pesticide treatments has also been reported. Male black pineleaf scale, Nucalaspis californica (Coleman) (Edmunds & Alstad 1985), and California red scale, Aonidiella aurantii (Maskell) (Shaw et al. 1973) are more susceptible to pesticides than females. Plant physiology is also affected by insecticide applications (Kinzer et al. 1977, Jones et al. 1983). Row crops treated with certain insecticides become attractive oviposition sites for Lepidoptera (Kinzer et al. 1977). Interactions between aphid reproduction, insecticides and cultivars have been reported on alfalfa (Mueke et al. 1978). Knowledge of the biology and interactions is required to properly time an insecticide application to disrupt the natural enemy populations while minimizing their effects on prey or host. Because insecticides potentially stimulate arthropod reproduction and effect plant physiology, estimates of predation rates with this exclusion method should be done cautiously. Although insecticide treatments stimulated the brown planthopper, Nilaparvata lugens Stal, reproduction, the amount of stimulation could not account for the high levels of resurgence. Only the reduction of natural enemies could. Insecticides can be used to determine the relative importance of natural enemies when the complex is composed of a few species showing temporal separation of their effects, in seasonal occurrence or in the generations they attack (Luck et al. 1999).

Removal of Natural Enemies by Hand.--Although laborious, hand removal has been used to evaluate the predators of tetranychid mites on citrus and avocado and to compare results obtained with other exclusion methods (Fleschner et al. 1955, Fleschner 1958). It has also been used to evaluate the mirid, Crytorhinus fulvus Knight, introduced to control the taro leafhopper, Tarophagus proserpina (Kirkaldy) (Matsumoto & Nishida 1966). Predation of Aphis fabae was also assessed in part by removing adult predators by hand when they flew onto predator free branches (Way & Banks 1968). A sticky band at the base excluded walking predators from feeding on A. fabae individuals placed on the branch.

Luck et al. (1999) believe that the hand removal method deserves more attention, especially as a method of checking for bias in other exclusion methods. However, it seems to be limited to studies of predator/prey interactions with species of low vagility, those that occur at reasonable densities and are diurnally active or are undisturbed by night lights (Luck et al. 1999).

Prey Enhancement.--Prey may be placed directly on plants in the field to stimulate predator attraction. This procedure involves tethering prey to a substrate (Weseloh 1974, 1982) or placing them on leaves or other plant parts where they would normally occur (Ryan & Medley 1970, Elvin et al. 1973, van Sickle & Weseloh 1974, Weseloh 1974, 1978, 1982; Torgensen & Ryan 1981). Some studies marked the prey with dyes before placing them in the field (Hawkes 1972, Elvin et al. 1973). The prey were visited frequently to measure predation, and if predation was observed, the predator's identity was noted. Predators such as spiders can be observed in the field with their prey *Kiritani et al. 1972), and web spinning spiders leave cadavers in or beneath their webs (Turnbull 1964).

It is sometimes more practical to use greenhouse grown plants of the same age, size and variety as plants used in field studies. Plants can be caged in the greenhouse or field for pest oviposition. Then the infested plants are transferred to the field and monitored for parasitism and predation. Van der Berg et al. (1988) used eggs of several foliage-feeding rice pests to determine predation. The egg chorion showed that eggs were attacked by predators with chewing or sucking mouthparts.

Predation and parasitism was thought to alternate as principal mortality factors during the year in studies that followed the seasonal incidence of predation and parasitism of eggs of the yellow stemborer of rice, Scirpophaga incertulas (Walker) (Shepard & Arida 1986). The technique of prey enhancement may be used to advantage with cages and or insecticides. However, a major limitation is that prey must be limited to sessile forms such as eggs, pupae or some scale insects, although there are possibilities with tethered hosts (Weseloh 1974). Kairomones and other chemical cues may be important to establishing the appropriate interaction (Nordlund et al. 1981).

Methods For Detecting Predation/Parasitism

Serology.--Predators have been associated with their prey with serological methods (Dempster et al. 1959, Dempster 1960, 1964, 1967; Rothshild 1966, 1970, 1971; Frank 1967, Ashby 1974, Vickermann & Sunderland 1975, Boreham & Ohiagu 1978, Sunderland & Sutton 1980, Gardner et al. 1981, Greenstone 1983). Predations rates have also been estimated with serology (Dempster et al. 1959, Dempster 1960, 1964, 1967). A precipitin assay has been also used (Boreham & Ohiagu 1978, Ohiagu & Boreham 1978, Southwood 1978). Other methods are the enzyme-linked immunosorbent assay (ELISA) (Vickermann & Sunderland 1975, Fichter & Stephen 1979, 1981, 1984; Ragsdale et al. 1981, Crook & Sunderland 1984, Sunderland et al. 1987, Sopp & Sunderland 1989), and an assay based on passive hemagglutination inhibition (PHI) (Greenstone 1977, 1979). Agglutination assay employs polystyrene latex particles coated with antibody (Boreham & Ohiagu 1978, Ohiagu & Boreham 1978). Such methods detect prey particles in the gut of predators by its reaction with antibodies obtained from a vertebrate, such as a rabbit, that has been sensitized to the prey. The reaction is a visible precipitate. (Also see Boreham & Ohiagu 1978, Miller 1978 and Sunderland 1988).

Detection of prey in a predator's gut is influenced by the size of prey, size of meal, time since the meal was taken, the rate of digestion, whether the natural enemy is a sucking or chewing predator, the abundance of taxonomically closely related prey and the sensitivity of the test. Sensitivity of the assay can be increased if the antibody is linked to an enzyme (ELISA). When the antibody reacts with prey, the enzyme carried with the antibody allows amplification of the reaction because one enzyme molecule can convert many molecules of substrate. This assay may detect hemolymph dilutions of more than 260,000 (Fichter & Stephen 1981) and is often sufficient to differentiate among prey stages (Ragsdale et al. 1981).

Both precipitation and ELISA techniques are useful for identifying the prey in a predator's diet and estimating predation rates (Sunderland 1988). ELISA is more sensitive to the presence of small amounts of antigen (prey protein or carbohydrate), is suitable for large scale testing and can be used with a minimum of equipment. Material necessary for the tests may be prepared and stored under refrigeration for six months (Sunderland 1988).

The passive haemagglutination assay (PHA) is a method for increasing sensitivity of the precipitin test. Sheep red blood cells (rbc) are chemically coated with the antigen of the suspected prey. Antigen coated rbc's are added to a solution of specific antibody and combine with the antigen molecules on the rbc to form a mat (agglutination). Small amounts of antibody cause agglutination. In antibody-free controls the rbc's do not agglutinate and this inhibition forms the basis of the assay. The amount of antibody required to cause agglutination is determined and added to an extract of a predator's gut contents. If prey protein or carbohydrate (antigen) is present it binds with the antibody. When antigen coated rbc's are added, they will not agglutinate because the antigen from the predator's gut has been bound by the antibodies (Luck et al. 1999). A small amount of antigen produces inhibition which explains the assay's greater sensitivity than that of a comparable precipitin assay (Greenstone 1979). Freshly sensitized erythrocytes have to be prepared each time an assay is conducted (Boreham & Ohiagu 1978), and this requires skilled operators.

The precipitin test was originally used to document arthropod predation of mosquito larvae (Bull & King 1923, Hall et al. 1953, Downe & West 1954) and latter was applied to terrestrial predator/prey interactions (Downe & West 1954). The first prey for which estimates were attempted from field samples was a chrysomelid beetle Gonioctena (= Phytodecta) olivacea (Forster) feeding on broom (Dempster 1960). Tests revealed six mirids, two anthocorids, a nabid, a dermaptern and red mites feeding on the beetle in the field. Laboratory tests showed that only the older mirid and anthocorid stages fed exclusively on younger stages of G. olivacea. A single laboratory feeding by the mirids and anthocorids could be detected 24 hrs after they had ingested a meal, and feeding by a dermapteran could be detected 60 hrs after it had fed (see Luck et al. 1999).

The degree of overlap between older stages of the predator and younger stages of the beetle influenced the number of beetles preyed upon. Densities of prey and predators were estimated from field samples. The fraction of positive responses in predator samples estimated the fraction of the predator population that had fed on G. olivacea. Because G. olivacea were scarce in the field while alternative prey were abundant, encounters between G. olivacea and the predators were infrequent. Therefore, if a predator tested positive to G. olivacea antibody, it was interpreted as a single predation event. Then the number of beetles preyed upon by each predator could be estimated suing the equation:

P_a = (N_piF_piT_pi) / R_pi

where P_a is the number of prey killed; N_pi the density of the predator (or stage of predator) i; F_pi the fraction of positive tests of the ith predator in a sample; T_pi the duration in days that the appropriate prey and predator stages are coincident in the field; and R_pi the retention time of a single prey feeding by ith predator (or stage of predator). Estimates from the precipitin test of egg and larval mortality due to predator for two beetle generations were found to agree closely with the independent estimates of "unknown" losses of eggs and young larvae during the same two beetle generations (Richards and Waloff 1961).

The precipitin test also was used to identify the predator species and to determine the fraction of Pieris rapae (L.) eggs and young larvae that died due to predation (Dempster 1967). Because of the relative scarcity of P. rapae a positive precipitin test was interpreted as one predation event. Studies of the delphacid Conomelus anceps (Germar) employed precipitin tests to identify ten of 91 potential predators (Rothchild 1966). The precipitin test could not be used to estimate predation rates because multiple predation events were possible.

For estimating predation rates with the precipitin test it is necessary to have information about predator and prey densities, densities of alternate prey, the period during which a meal can be detected in each predator and prey and predator stages involved. Precipitin tests estimate predation rates of prey which form a small fraction of the available prey or infrequent predation events. A slight bias may arise in such estimates if predators have fed on other predators that have fed on the prey, if a suspected predator is phytophagous but ingests sessile prey stages while feeding on the plant or if a suspected predator feeds on prey carrion (Boreham & Ohiagu 1978). The precipitin test may also yield biased estimates of predation rates from cross reactions between the antibodies of closely related species. Therefore, a knowledge of the local fauna which might serve as prey and the predator's propensity for local movement is essential to the successful application of this test (Luck et al. 1999). Also the serum developed from one prey stage may not react with the antigen of another (Boreham & Ohiagu 1978). Sufficient resources must be committed in order to use this technique: prey must be collected in sufficient numbers to elicit an immunological response when injected into the vertebrate. As such the procedure is not ideal when applied to small prey such as mites (Murray & Solomon 1978).

When used in conjunction with other population studies, precipitin assays may be very helpful. Few other methods can provide quantitative estimates of predation rates under natural field conditions. Although they cannot be used to estimate predation rates under all situations, they are valuable for identifying predator species or stages that feed on a prey. This method deserves more attention especially as more sensitive tests such as ELISA are available (Vickermann & Sunderland 1975, Fichter & Stephen 1981, 1984; Ragsdale et al. 1981, Crook & Sunderland 1984, Sunderland et al. 1987, Soop & Sunderland 1989). A great advantage is that predation is allowed to occur naturally.

Electrophoresis & Isoelectric Focusing.--Predators may be associated with the prey with electrophoretic techniques. Electrophoresis separates proteins based on charge and size differences in an electrical field. Differences in charge and size commonly occur among isoenzymes (proteins catalyzing the same reaction) from different taxa. If the prey and predator have isoenzymes with different electrophoretic mobilities, the analysis of homogenates prepared from predators fed on prey should exhibit protein bands corresponding to the predator and the prey. Also if there are several potential prey of a predator, and if the prey have electrophoretically distinct isoenzymes, analysis of predator homogenates can reveal the prey species inside the predator.

Electrophoresis can be successful if the prey isoenzymes are detectable after predator feeding, and electrophoretic variation occurs among the prey and predator isoenzymes. Isoenzyme detection depends on prey size, in vitro activity of the isoenzyme, presence and volume of the predator foregut, and the type of electrophoresis employed (Murray & Solomon 1978, Giller 1984, Lister et al. 1987, Soop & Sunderland 1989). Electrophoretic variation depends on the suite of isoenzymes available for comparison and the type of electrophoresis. Standard electrophoretic procedures (starch gel and polyacrylamide gel electrophoresis) can detect prey isoenzyme activity for several isoenzyme types involving relatively large prey (>2-3 mm body length). Under this size, the number of detectable prey isoenzymes is diminished and hence the chance of distinguishing closely related prey is decreased.

Enhanced sensitivity of electrophoretic methods include conventional electrophoresis in cellulose acetate membranes (Easteal & Boussy 1987, Höller & Braune 1988) and isoelectric focusing (IEF) (see Luck et al. 1999). IEF has advantages over other techniques involving small and large prey. In IEF, proteins are "focused" into narrow bands along relatively broad pH gradients. Focusing enhances the detection of enzymes compared to other techniques which gradually spread the proteins into diffuse bands. In addition, because relatively broad pH gradients are used in IEF, enzymes with different charges, such as may occur between unrelated prey taxa, will remain sharply focused on the gel. The fine resolution of IEF does not affect the ability to distinguish enzymes with very similar charges. With standard techniques, these contrasting problems are difficult to solve simultaneously as one set of conditions (buffer type and pH, gel type) may be optimal for one prey type but not others.

The prey of several arthropod predators have identified with electrophoresis. Polyacrylamide gradient gel electrophoresis was used to detect prey protein (esterases) in the gut of predators after they had fed on known prey (Murray & Solomon 1978). The technique detected esterases of Panonychus ulmi (Koch) in the predaceous mite Typhlodromus pyri (Scheuten), and in two anthocorids, Anthocoris nemoralis (F.) and Orius minutus (L.) that had fed on the mite in the laboratory. Dicke & DeJong (1986) used methods to determine whether T. pyri and Amblyseius finlandicus (Oudemans) also fed on the apple rust mite, Aculus schlechtendali (Nalepa) as an alternate host in the field. Electrophoresis was also used to identify the prey species exploited by A. nemoralis on alders in the field (an aphid Pterocallis alni [DeGeer]) (Murray & Solomon 1978). Electrophoresis with polyacrylamide disc gels detected esterases of several prey species in the gut of the waterboatman Notonecta glauca L. (Giller 1982, 1984, 1986). A meal was detectable from 17-48 hrs depending on temperature and meal size, and was strongly correlated with the length of time the meal spent in the foregut (Giller 1984). Giller (1986) used electrophoresis to identify the prey of N. glauca and N. viridis Delcourt in the field. Lister et al. (1987) used polyacrylamide gel electrophoresis and electrophoresis and esterase allozymes to determine the diet of some microarthropods and the predation rate by the acarine predator Gamasellus racovitzai (Trousessart).

Predation rate estimates with serological methods and electrophoresis requires substantial resources. The techniques call for the development of antibodies or methods for identifying the isozymes of the prey species or stage, the development of methods to estimate the predator and prey densities, including those needed to estimate the densities of alternate prey, and the identification of the predator and prey stages involved. Initially the use of these techniques to estimate predation rates appeared limited to prey populations which form a small fraction of the available prey or in which predation events by a predator are frequent. Frequent predation confounds interpretation of a positive test because a single large meal cannot be distinguished from several small meals.

Immature parasitoids within aphids have been detected with electrophoresis (Wool et al. 1978, Castanera et al. 1983), and in whiteflies (Wool et al. 1984). The parasitoid Aphidius matricariae Hal. was detected in the green peach aphid Myzus persicae (Sulz) and parasitism of the white fly, Bermesia tabaci (Gennadius) by the endoparasitoids Encarsia lutea (Masi) and Eretmocerus mundus (Mercet), was detected with electrophoresis and histochemical staining for esterases. But the whitefly parasitoids could not be identified to species (Wool et al. 1984). Electrophoresis allows the processing of large numbers of hosts to estimate the fraction that are parasitized and sometimes the parasitoid species involved. This contrasts with the traditional methods in which field samples are dissected while fresh or reared. Electrophoresis can detect within a host immature parasitoids without dissection and parasitoid enzyme activity within a prey cannot be confused with host's enzyme activity.

Marking Prey.--Predator species and/or predation rates have been identified with marking techniques. Markers have included radioactive isotopes -¹⁵¹europium (Ito et al. 1972), ³²phosphorus (Jenkins & Hassett 1950, Pendleton & Grundmann 1954, Jenking 1963, McDaniel & Sterling 1979, McCarty et al. 1980, Elvin et al. 1983) and ¹³⁷cesium (Moulder & Reichle 1972), ¹⁴carbon (Frank 1967), rare elements (Stimmann 1974, Shepard & Waddill 1976), and dyes (Hawks 1972, Elvin et al. 1983). Prey are fed (Elvin et al. 1983, Frank 1967, Room 1987) or injected (McDaniel & Sterling 1979, McCarty et al. 1980) with the radioactive isotope and the radioactivity is detected in a predator with scintillation, a Geiger counter, or autoradiography. For autoradiography suspected predators are collected after exposure to labelled prey and are glued to paper, which is placed against X-ray film (McDaniel & Sterling 1979). The film is developed, and dark spots on the film produced by the rays from ³²phosphorus indicate labelled predators. Methods involving isotopes require training and necessary equipment to perform the assays. Safety regulations and environmental considerations may limit the use of the method in some situations. Other disadvantages, as with electrophoresis and serological techniques, include the inability to detect whether a predator had fed on other predators that had consumed labelled prey or whether a prey was scavenged (Luck et al. 1999). Experiments using isotopes, especially those using autoradiography, are simpler to conduct than serological and related techniques. Methods using labelled elements require several manipulations, but they provide more information per unit effort than other kinds of marker tests.

Such rare elements as rubidium and strontium also have application as labels. They can be sprayed on foliage or placed in the diet of the prey, incorporated into the prey's tissues and then transferred to the predators or parasitoids who feed on labelled hosts (Stimmann 1974, Shepard & Waddill 1976). The mark should be retained for life, and self-marking is possible via a labelled plant. However, the technique requires an atomic absorption spectrophotometer, which is expensive, and placement of the labelled prey on plants may expose them to abnormal predation rates. Phytophages seldom choose feeding or oviposition sites on their plant hosts at random (Ives 1978, Wolfson 1980, Denno & McClure 1983, Guerin & Stadler 1984, Whitham et al. 1984, Myers 1985, Papaj & Rausher 1987). Parasitoids and predators do not search their habitats uniformly (Weseloh 1974, 1982; Fleschner 1950). Therefore, without the proper behavior studies, the degree of bias in determining the natural enemy complex or in estimating predation rates is unknown.

Genetic markers have been used to track parasitoids and assess their impact against hosts, such as common muscoid flies. Legner & Brydon (1966) liberated a thelytokous race of parasitoid on poultry farms which they were able to tract and derive host mortality data from. Legner et al. ( 1990a, 1990b ) derived similar information by releasing gregarious strains of Muscidifurax raptorellus Kogan & Legner, and a temporary interference of several weeks with resident parasitism during the establishment phase was detected. However, this was later overcome when the released strain had a chance to multiply naturally at the site.

Visual Counts.--There are several advantages of using visual counts over many of the exclusion techniques. There is no manipulation of the environment required. Prey can be added or predators removed to determine the response of the predator to changes in prey density. A visual record reveals the predator's diet in the field. Perhaps serology and electrophoresis share these three advantages, but the latter require considerable technology. visual counts require a substantial commitment of time to observe the predation and to determine the feeding rates for different combinations of predator/prey stages. Vision cannot be used if the predator is cryptic, easily disturbed or escapes from the observer. Also, the time a predators spends consuming a prey may vary depending on the range of prey stages attacked, the hunger level of the predator, interference or stimulation by other predators or prey that are active in the area, and differences among individual predators due to genera, reproductive stage or molting (Luck et al. 1999). These in turn can determine the probability that a predator will be observed in the field with a prey. Laboratory data on the time spent by four predators consuming prey was highly variable, leaving the investigators pessimistic about the visual method's utility for estimating predation rates (Kiritani et al. 1972). But the approach may still be valid for some predators, and has been used to determine the fraction of diurnal predation for each predator species in a complex (Elvin et al. 1983).

Statistical Sampling.--Obtaining field samples during parasitoid/predator liberation periods can provide useful information about the ability of a species to effect its host/prey density. Parasitoid impact was thus measured on the pink bollworm, Pectinophora gossypiella (Saunders). A significant positive relationship was found between the total number of parasitoids released and the host density, which was most pronounced during a mid autumn period (Legner & Medved 1979). Releases of egg-larval and larval-parasitoids produced small measurable reductions in P. gossypiella moth emergence from mature cotton bolls, but did not significantly reduce % boll infestation. By releasing parasitoids at three densities, it was possible to show significant differences between controls and a low and medium release rate, but excessive parasitoid dispersal out of the release areas into the cotton field at large explained a leveled slope after the medium release rate (Legner & Medved 1979).

The potential of Goniozus spp. and Pentalitomastix plethorica Caltagirone to regulate navel orangeworm, Amyelois transitella (Walker) was judged from seasonal positive functional responses to host density and with k-value analyses (Varley et al. 1974, Legner & Silveira-Guido 1983 ). Application of this technique requires that the host show minimal overlapping of generations, however. Goniozus emigratus (Rohwer) and Goniozus legneri Gordh demonstrated a significant capacity to recognize and respond in a regulative fashion in mid summer by increasing attack rates on higher host densities. However, no such tendency was indicated during cooler periods of late autumn.

Indigenous parasitism of Rhagoletis completa Cresson in its native range of western Texas and southeastern New Mexico was also assessed with k-value analysis,which showed a significant impact of combined natural mortality on host reduction (Legner & Goeden 1987). Biosteres sublaevis Wharton demonstrated the greatest measurable activity as a cause of natural mortality.

Legner & Brydon (1966 ) were able to show an increased parasitism and house fly host mortality closer to liberation sites of parasitoids. Legner et al. (1990a, 1990b ) also charted increases and spread of muscoid fly parasitism from release sites. The importance of proper field sampling, measurement of host destruction and unpredicted upsets to organisms in different guilds in these and similar studies was emphasized (Legner & Bay 1964, Legner 1979, Legner 1983a, 1986 ).

Long term sampling of the width of aquatic weed masses following the introduction of phytophagous cichlid fish established a regulatory capacity for Tilapia zillii Gervais in California irrigation canals (Legner & Fisher 1980 , Legner & Murray 1981). The data clearly showed seasons of high, medium and low regulatory capacity.

MOLECULAR TECHNIQUES

Unruh (1999) discussed various molecular techniques and procedures as they may be applied to classical biological control work. Diehl & Bush (1984) pointed out that emphasis is often to seeking new races or biotypes of beneficial species which had demonstrated partial control of a pest. Usually this effort is directed at finding races which are better suited for climates in specific parts of the pest's range. Collections of morphologically identical populations of a natural enemy from geographically separate and ecologically distinct habitats often result. In such cases scientists are left with the problems of evaluating the subsequent collections for their inherent variability for a variety of biological attributes, such as climatic, reproductive, sexual or host preference. Populations of the various races are usually released in a random way followed by biological studies of those which become established. Such studies have revealed cryptic species or races with distinct biological attributes, as exemplified by host specific races of Comperiella bifasciata (Howard) which attack the Chinese race of the California red scale, and the Japanese race of the yellow scale (Diehl & Bush 1984). Nature screens the adaptive diversity of natural enemies, which is believed more accurate and cost effective than any regimen of prerelease screening (van Lenteren 1980). This procedure nevertheless precludes the development of new procedures and theory for evaluating and using races that are morphologically indistinguishable (Unruh 1999).

Three frequent errors of omission in many biological control programs are (1) the low priority allotted to pre- and post-release evaluation of natural enemy effectiveness. Traits such as developmental rates, diapause, host range, sex ratio and resistance to encapsulation are of special importance; (2) the tools for identifying and studying infraspecific categories of natural enemies both prior and subsequent to their release are not used; and (3) descriptions of the ecological and genetic relations of natural enemies in their native distribution are rare.

In spite of increasing sophistication in the theories of natural enemy/pest interactions, it is still not possible to choose the best natural enemies even when give sound ecological studies in a laboratory or in the native range (van Lenteren 1980). In order to improve the understanding of biological control introductions, particularly those employing biotypes, it is necessary to have better methods for discriminating populations. Traits to identify populations must reside in the organism itself. although short term field or laboratory studies may often employ dyes, heavy metals or radioactive tracers. Where morphology cannot distinguish, the next level of identification is biological or molecular markers. In some instances where biological traits under study change through time, molecular differences must be sought among species, races, strains or even individuals. Molecular traits reflect a natural hierarchical organization that begins with the DNA sequence and ends with the protein composition of the organism (as well as other phenotypic expressions). DNA sequences are the most difficult and expensive to obtain and estimates of the electrophoretic mobility of particular proteins, such as enzymes, are the least expensive and most straightforward.

Hutchinson (1965) compared evolution to a play and the ecological environment to a theater. Unlike most researchers of natural variation in organisms, biological control workers remove populations from their home town theaters and attempt to run many plays simultaneously at an out of town stage (Unruh 1999). The problems encountered may partially explain the seeming preoccupation with the concepts of biotype, race, geographic, sibling, semi-, and cryptic species (DeBAch 1969, Rosen & DeBach 1973, Rosen 1978, Gonzalez et al. 1979). In biological control it must be considered how previous evolutionary history affects adaptation to a novel ecological milieu. The ecological and genetic dynamics which arise when two or more species are released into the same area against the same host need to be understood. When questioning the likelihood of pre- and postzygotic isolating mechanisms of geographic species, the query is more than academic. The idea that biological control programs represent evolutionary experiments on a grand scale (Myers 1978) is important, but very little effort has been expended on postrelease study of natural enemy adaptations to their new home. Only concerted effort in these aspects of a biological control program can transform an empirical practice into an evolutionary experiment.

Molecular methods are important for biological control research because they represent a set of tools to employ in dissecting the biology of natural enemies (Unruh 1999). Sometimes these tools may be irreplaceable but in many cases the questions posed may be addressed by a different route. Using molecular methods to the exclusion of more traditional methods, or at the expense of developing navel methods, is not logical. Avoiding molecular methods because of a perception that they are too complicated is also foolish. With natural enemies systematic investments in molecular approaches should yield rapid returns. However, studies of the biological control practice of importing few specimens, rearing them in the laboratory and releasing their progeny into a hostile environment may uncover radical effects associated with colonization (Unruh 1999).

Isozymes

Hunter & Markert (1957) and Unruh (1999) discussed gel electrophoresis and histochemical localization of enzymes within the supporting gel matrix, which is a powerful method with which to observe genetic variation within and between animal populations (Hubby & Lewontin 1966, Lewontin 1974). Differences in mobility among allelic variants of specific enzyme gene products is the most accessible and practical form of biochemical trait variation available.

Protein electrophoresis does not reveal all amino acid sequence differences that may exist in a protein nor all nucleic acid differences in the genes which code for proteins. About 29% of the nucleotide substitutions which occur are synonymous, arising from redundancy in the genetic code. Of the remaining 71% which cause amino acid substitutions, only about 25-30% change the net charge of a protein (Nei & Chakraborty 1973). A few substitutions that do not change charge may affect conformation (tertiary structure) and be detectable by electrophoresis (Lewontin 1974). Even variants differing in charge may not be detectable with a single buffer system of those commonly employed. This hidden genetic variation (Johnson 1977) may be exposed by use of sequential electrophoresis (multiple electrophoretic separations using a sequence of buffer pHs to detect heterogeneity in a given mobility class) with starch and acrylamide gel supports or by isoelectric focusing (Coyne 1982, Aquadro & Avise 1982), or with cellulose acetate supports (Easteal & Boussy 1987). No single set of conditions has proven able to detect all differences, therefore the careful worker usually tries several buffer systems and potentially two or more supporting matrices. Isoelectric focusing is not significantly superior to zone electrophoresis in detecting hidden variation (Aquadro & Avise 1982, Coyne 1982). But isoelectric focusing is much superior in its ability to concentrate proteins while it separates them (Righetti 1983) and is potentially superior for studies with minute insects where enzyme concentrations in homogenates are very dilute.

Isozyme data are typically of the form of allele (mobility class) frequencies at one to several enzyme loci. Phenotypes of individuals as seen on the stained gels should be interpretable under a genetic model that corresponds to the insect's ploidy and mode of inheritance of the gene in question. This may require genetic analysis in order to eliminate nongenetic variation (Richardson et al. 1986).

Enzyme electrophoresis is in such wide use as a tool in systematics and genetic studies that a comprehensive enumeration for the insects alone would be impractical. But, numerous reviews in relation to systematics of insects and other animals exist (Avise 1974, 1983; Berlocher 1984, Bush & Kitto 1978, Buth 1984, Thorpe 1983), as well as many useful references on technique (Harris & Hopkinson 1976, Eastel & Boussy 1987, Richardsson et al. 1986, Selander et al. 1971, Shaw & Prassad 1970). As of 1991 few studies have been conducted on natural enemies of insects and there have been limited in scope (Unruh et al 1986).

These data can be used in two basically distinct ways in systematics: (1) presence or absence of alleles can be analyzed directly as character states; (2) allele frequency data may be summarized into a matrix of distances among taxa which is then analyzed. The distance, phenetic, or so called quantitative approach (Avise 1983) has been most popular, but persuasive arguments for the superiority of a character state approach exist. This topic has been variously reviewed by Avise (1983), Berlocher (1984), Buth (1984), Felsenstein (1982) and Hillis (1987). This dichotomy in analytical approaches exists for all discrete types of biochemical characteristics including restriction fragment polymorphisms and sequence data, but for analyses involving DNA the character states approach is preferred because restriction fragment or sequence differences can be unambiguously polarized (Hillis 1987). The discrete nature of isozyme, restriction fragment, and sequence data contrast with immunological distance and DNA-DNA hybridization which yield only distance data (Unruh 1999).

DNA Restriction Fragment Analyses

A critical step in the growth of molecular genetics came with the discovery and utilization of restriction enzymes in the 1970's (Roberts 1982). Their use in characterizing polymorphisms in natural populations followed within 10 years (Avise et al. 1979). Restriction enzymes are a group of endonucleases that cleave DNA at characteristic 4-6 base sequences. For example the restriction enzyme known as EcoRI cleaves 5'-GAATTC-3' between G and A. Differences between individuals in the presence or absence of these specific recognition sequences throughout a DNA sequence are detectable as different length pieces of cleaved DNA upon separation in an electrophoretic assay. The variants are called restriction fragment length polymorphisms (RFLP) or more briefly restriction polymorphisms. Some 350 different restriction enzymes are known (Roberts 1982) and >50 are available commercially. Typically, purified DNA is digested and the fragments, which are separated by electrophoresis based on size, are visualized by a direct DNA stain. Alternatively, in a method which is ca. 100X more sensitive, the DNA is transferred from the gel to a nitrocellulose membrane and visualized there by autoradiography, or by hybridization with a preselected radioactive or biotinilated (Dykes et al. 1986) DNA probe (Lansman et al. 1981). probes are pieces of DNA (or RNA) which, by virtue of their complementary sequence, bind to the DNA (or RNA) of interest. The combined procedures of transferring nucleic acid to a membrane and detecting complementary sequences bound there is called blotting (see Maniatis et al. 1982 and Unruh 1999).

mtDNA

RFLP studies of natural populations have emphasized the mitochondrial genome because of several attributes of this extrachromosomal DNA (Brown 1983, Moritz et al. 1987): (1) mtDNA is a relatively small molecule of 15-20 thousand base pairs which are arranged in a covalently closed circle; its small size allows the molecule, or a few fragments from it, to be analyzed with a single electrophoretic procedure. (2) mtDNA is easily isolated because mitochondria, themselves, are discretely sized organelles, easily segregated by centrifugation, and numerous in almost all cells. Also, mtDNA often has a buoyant density significantly different from nuclear DNA further facilitating its isolation. (3) The molecule is quite simple compared to nuclear DNA (ca. 4 order of magnitude shorter; has the same 37 genes in all taxa; each gene is represented by only a single copy of the mtDNA molecule). (4) It is haploid precluding the high possibility of heterozygotic individuals for RFLP as might be seen in a region of genomic DNA. This fact allows for straightforward analysis of the resulting fragments on the gels. (5) mtDNA is maternally inherited simplifying genetic analysis. (6) Usually all mitochondria within an individual have identical DNA sequences (= homoplasmy). (see Moritz et al. 1987, and Unruh 1999).

Restriction fragment length polymorphism and mapping in mtDNA is most useful at or below the species level (Wilson et al. 1985, Avise et al. 1987). For higher taxonomic categories its utility is limited by the rapid accumulation of transitional nucleotide substitutions (versus base transversions). Transitional substitutions apparently become saturated within about 10 million years in mammals, after which substitution rates fall 5-10 fold to , or below, that characteristic of mammalian nuclear DNA (Brown 1983). The situation in invertebrates is less clear (Powell et al. 1986).

Genomic DNA

Restriction polymorphism studies of nuclear DNA are technologically more difficult than that for mtDNA because of the great size of the nuclear genome. In mtDNA there are 37 genes in about 18,000 base pairs; in eukaryotic nuclear genomes there are 40,000 or more genes in 200 million base pairs (Spradling & Rubin 1981, Lewin 1985). In most mtDNA (except yeast mtDNA) there is almost no spacer DNA, no intervening sequences, no satellite DNA and each gene is unique (in some cases sequences overlap). By contrast, eukaryotic nuclear genes (cistrons) are often repeated, interrupted and surrounded by non-coding sequences (introns and exons) which are often composed of characteristic repeating sequences. In addition there are large blocks of DNA called satellite DNA (the heterochromatin visible in cytological studies) that consist solely of highly repetitive sequences. About half of eukaryotic genomic DNA consists of moderately to highly repetitive DNA (Unruh 1999).

To study RFLP in nuclear DNA, the sequence of interest must first be isolated. This requires cloning the sequence. For example, if restriction fragment variation in alcohol dehydrogenase (ADH) is to be examined, the major tasks would be (1) to develop a probe specific for the ADH sequence and (2) to clone the region of DNA containing the gene (for which the probe is complementary). The second step may be necessary for each individual or population to be assayed. Since the ADH sequence represents only about 1 X 10 ^-5% of the total genomic sequence, developing a probe is a major effort demanding the facilities and expertise of a fully operational molecular biochemistry laboratory.

Sequencing

Although quite a few genomic and mitochondrial DNA regions have been sequenced from a variety of animals, no complex eukaryotic genome has been completely sequenced. In 1983 the complete mtDNA sequences was available for only three species: the mouse, humans and the cow (Avise & Lansman 1983). There are numerous articles treating of the merits of a unified effort to completely sequence the human genome. Comparative studies of sequences of genomic DNA have been conducted only where a history of protein study precedes the work (ADH in Drosophila species) (Dreitman 1983, Coyne & Kreitman 1986), or, more often, in humans and diseases of humans and animals (MacIntyre 1985). Generally, the same difficulties encountered in restriction fragment analysis is also encountered in DNA sequencing. This is because sequencing must be preceded by restriction analysis and cloning of each of the restriction fragments. Selected cloned fragments are finally sequenced. About 1,000 bases may be sequenced in a day, and with automated gel scanning sequencers can attain a rate of 600 bases per hour (Prober et al. 1987). However, the techniques to isolate a specific sequence of interest from the billion base4 pairs comprising the genome remains a formidable obstacle. This topic is discussed in Beckendorf & Hoy (1985), Legin (1985), and the mathematics of analyzing sequence data has been reviewed by Nei (1987).

In biological control, the cost of DNA sequencing is excessive. The technical difficulty and expense of DNA sequencing presently exceeds that of protein sequencing (see Kimura 1983).

Ribosomal DNA

There is only one exception toe the technological constraints posed by the large size of eukaryotic genomes. Ribosomal RNA (rRNA) is an important functional component of the ribosomes, occurring in multiple, in tandem repeating copies which are often localized in a part of the genome (one or a few chromosomes). In insects as much as 1% of the total genomic DNA is rDNA (ribosomal DNA). Probably because of the structural role rRNA plays in the ribosome, its sequence is highly conserved among all life (excepting viruses). Across the eukaryotes rRNA sequence homology is above 70%. Sequence divergence of rRNA among bacteria represents the most compelling evidence supporting the recent recognition of a third kingdom of organic life, the archaebacteria (Woese 1981). In addition to the highly conserved rRNA, the genes for rRNA spacer regions show considerably less sequence conservation than mature rRNA (the RNA remaining after all excision of spacers). Creating genomic libraries for rDNA is facilitated by its high copy number and probes developed for completely unrelated taxa retaining sufficient homology to be useful for restriction studies as well as sequencing.

Phylogenetic studies based on restriction site variation of rDNA have begun to appear in the literature (e.g., for Rana spp. by Hillis & Davis 1986). The organization of rDNA in insects was reviewed by Beckingham (1982). With restriction analysis, we may concentrate on space DNA (both transcribed and nontranscribed) where sequence homology can be quite low and genetic difference among taxa are well known for insects (Beckingham 1982). Heterogeneity within the spacer regions of Drosophila populations has also been observed by restriction analysis (Coen et al. 1982).

Probably more significant for systematic application is a rapid sequencing technique developed by Lane et al. (1985). It was observed that the highly conserved nature of rRNA allowed side stepping the development of a probe; instead it is possible to synthesize or purchase a probe based on published sequences for other taxa in the same phylum or kingdom. Then, this short oligonucleotide (ca. 20 bases) "probe" is used as a primer in more routine procedures to sequence from a highly conserved domain into one that is variable and taxonomically informative. Because rRNA exists in enormous proportions in the cell (10% by dry weight in E. coli; Lewin 1985), the rRNA from a single, or few individuals, may be sufficient to sequence >200 without a need for cloning or probe development. The highly conserved nature of rRNA, although permitting this method, may also restrict its value to the systematics of higher categories, such as families and genera.

Direct Estimation of Genetic Distance

Two methods are used to measure genetic distance directly, as compared to averaging of many qualitative differences: Immunological distance and DNA-DNA hybridization. The application of both methods is limited to phylogenetic questions because they provide only distances.

Immunological methods to measure genetic distance among related organisms was reviewed in Bush & Kitto (1978) and Beverly & Wilson (1985). The basis of the technique is to isolate a protein, produce antibodies to it and use the purified antibody in microcomplement reactions within and between taxa. The resulting distance measurements are phenotypic, and subject to the peculiarities of antigen-antibody interactions.

Problems with the method include significant stochasticity in the rate at which amino acids are substituted in antigenic proteins and immunological distance is not always proportional to the number of amino acid substitutions (Nei 1987). Both sources of error combine to make the molecular clock from immunological distance slightly more inaccurate than that from amino acid sequence data (Wilson et al. 1977). However, the correlation between immunological distance and nucleotide substitution usually is high (r = 0.9). This association holds for proteins differing by as many as 30% of their amino acids, or, on another scale, comparisons remain possible where divergence times approach 150 million years (Wilson et al. 1977). Thus, the technique has an important role in estimating phylogeny of higher taxonomic ranks and remains useful down to specific differences.

Microcompliment fixation is highly sensitive and repeatable and the amount of antibody harvested from each immunized rabbit is adequate for thousands of comparisons. The methods greatest fault is the technical difficulty and expense of isolating sufficient quantities of pure protein for immunization from each test population. The protein selected as antigen must be homologous and relatively abundant. Larval hemolymph protein was used by Beverly & Wilson (1985) to examine drosophilids. It is not certain what an ideal protein would be for parasitic Hymenoptera. The difficulty in obtaining pure, homologous protein from the taxa of interest has limited the use of this method.

DNA-DNA hybridization provides a direct physical estimate of sequence homology over the whole coding genome making its use for quantifying genetic distance among taxa theoretically attractive. it is based on the proportionality of the thermal stability of a DNA duplex and sequence homology between reannealing strands. The procedures consist of purifying the DNA, shearing or digesting it into standardized size pieces, disassociating the duplexes, and estimating the amount of reassociation of mixtures between and within taxa at various temperatures.

The occurrence of repetitive sequences throughout the genome represents a nuisance in DNA hybridization studies. About 20-40% of the DNA in insects is repetitive (Spradling & Rubin 1981). When a sequence exists as a single copy in a mixture of segments of DNA from the organism, its probability of encountering (= rate of hybridizing with) its complementary sequence is the length of the sequence divided by the total genome size. When a sequence exists in multiple copies, the probability of one copy encountering a complement is this ratio multiplied by the copy number of the sequence. Thus, repetitive sequence hybridize much more rapidly than unique sequences. The genome has been classified into fast, intermediate, and slow components which corresponds to their rate of hybridization and copy number (Lewin 1985). In actuality, the nonrepetitive, or slow, DNA component is isolated from the fast and intermediate components by thermal chromatography and then used for hybridization (Britten et al. 1974). A small amount (0.1%) of a tracer composed of radio-labelled single copy DNA from one test organism is mixed with a large mount of unlabelled single copy DNA (driver DNA) from another organism and reassociation of the strands is estimated from the measured radioactivity of different aliquots from a second thermal chromatographic separation.

Based on DNA "melting" experiments with thousands of bird species and primates, Sibley & Ahlquist (1983, 1984), estimated that a one degree C. difference in melting temperature between homo- and heteroduplex comparisons (T5OH) corresponds to about 4-5 million years since divergence of genomes. In contrast, for cage crickets (Coccone & Powell 1987), Drosophila (Powell et al. 1986) and sea urchins (Britten 1986), one degree C. corresponds to ca. 1 million years since divergence. Therefore, invertebrate DNA seems to be evolving at 4-5 times the rate of mammalian DNA.

But DNA-DNA hybridization requires several grams of tissue in order to extract sufficient DNA to do all the reciprocal hybridization reactions required to compare a modest number of populations. As the number of populations increase, the amount of material required increases geometrically. Therefore, completely orthogonal comparisons among a large number of populations or species is impossible. The use of single copy DNA hybridization for insects dates to 1975 (Sohn et al. 1975). The large part of the genome examined and the physico-chemical precision of hybridization of single copy DNA represents a strong argument for its use as an evolutionary clock in phylogenetic analyses (Gould 1985).

Therefore, immunological distance offers a highly repeatable estimate of genetic (phenotypic) distance of a single homologous gene product, which may be compared across many taxa because of the large amount of antibody produced, and can be used to compare clearly differentiated species and up to genera, families and probably superfamilies depending on the rate of divergence of the protein selected. DNA hybridization offers an estimate of total (single copy) genome divergence for a smaller group of populations, and in insects is probably restricted to comparisons ranging from among well differentiated populations (or cryptic species) to comparisons among genera.

Uses of Molecular Techniques in Biological Control

There have been no insect bio-control agents studied with molecular involving nucleic acids, and only a few studies have employed the isozyme method (Unruh et al. 1986). However, there has been considerable use of molecular methods with insect viruses (Crook et al. 1985), entomogenous bacteria (Oeda et al. 1987) and with viruses found in calyx fluid of some parasitic wasps (Theilman & Summers 1986). The problem in analyzing the genome of these forms of life (viruses and bacteria) is much less than for comparable studies of higher eukaryotes.

Systematics.--Unruh (1999) pointed out that systematists often reiterate the maxim that from phylogenetically sound classifications emerges the ability to predict characteristics of yet unseen species. For example, all known species in the aphelinid genus Marietta are hyperparasitoids of diaspidid scale insects (Clausen 1972). Therefore, it would be safely predicted that any newly encountered Marietta also had this habit. It is obvious that when higher taxonomy, especially at the generic and subgeneric levels, corresponds to such biological attributes it is very useful to the biological control researcher. In many important parasitoid genera, molecular techniques would be valuable in both organizing species within species groups and these within genera and by helping taxonomists find natural gaps on which to split genera into more workable entities.

Assumptions in the use of isozymes for phylogenetic analysis are that amino acid substitutions accrue at an about equitable rate among lineages within a taxon or, at least, that novel amino acid substitutions are much more likely than reversions to a previous sequence. Procedures and theory for phylogeny estimation using isozymes (Berlocher 1984b) and other molecular data (Felsenstein 1982, Nei 1987) and for consensus between molecular and morphological data (Hillis 1987) are very extensive topics. It is possible to compare the association between genetic distance from isozyme surveys to proposed taxonomic rank of taxa. Judging taxonomic rank by genetic distance measures is both overtly phenetic and typological, but genetic distances can alert us to problems when groups are observed which display considerable morphological variation without associated genetic differences, or conversely, high genetic differentiation without associated morphological differences. Genetic distances observed at various taxonomic ranks for several allozyme studies of insects are summarized in Unruh (1999). The mean values of Nei's standard genetic distance for studies employing 15 or more genetic loci are shown. Nei's distance is defined as D = -1n(I) where I = Jxy/(JxJy)^1/2 and J represent estimates of sums of squares for the proportion of shared alleles. Nei's distance, as opposed to many other distance measures commonly employed in isozyme studies, has well characterized theoretical properties in relation to rate of amino acid substitutions and evolutionary time (Nei 1987). However, the measure cannot be used for some methods of phylogeny estimation because it is not metric (Farris 1972, Hillis 1987). Isozyme studies sometimes prompt authors to relabel infraspecific ranks or to suggest revision of current taxonomic rank.

The association of genetic distance to presumed taxonomic rank is very inconsistent, particularly at levels below morphologically distinct species. Comparisons among species groups or genera result in D values of 1 or >1. Interspecific comparisons are less consistent, ranging between ca. 0.1-1.4 (see Unruh 1999). Most interesting are those cases where morphology, behavior or cytology has indicated specific rank while isozyme data reveals few differences. An example is the relationship between 10 species of Speyeria which includes one cryptic species pair: S. mormonia and S. egleis. These morphologically identical species are among the most genetically distinct of the 10 species (D = 0.224). In contrast, S. callipe, another species nearly indistinguishable morphologically from S. egleis but differing from it in karyotype, is very similar in isozymes (D = 0.015). Generally, the Speyeria species studied (Brittanacher et al. 1978) are a very homogenous group which does not include some of the more distinctive species in the genus (Brussard et al. 1985).

Another example is found in the study of Gryllus by Harrison (1979). A cryptic species pair, G. veletis and G. pennsylvanicus, previously thought to be products of allochronic speciation, and which display broadly overlapping distributions, were found to be quite distinct (D = 0.165). In contrast, three morphologically distinct and either macro- or microgeographically allopatric species differed very little (G. pennsylvanicus, G. firmus, and G. ovisopis; D<0.03).

The parasitoid genus Muscidifurax consists of five distinct species and various biotypes, based on both morphological and ethological (mating) behavior data (Kogan & Legner 1970, Legner 1987, van den Assem & Povel 1973). Isozyme analyses have borne out very closely the morphological and behavioral data, and strain differences are clearly shown (Kawooya 1983). Unique extranuclear phases to inheritance in one species, M. raptorellus were mentioned previously (Legner 1989, 1991).

Examples of very low differentiation of species include Magicicada spp. and Drosophila silvestris and D. heteroneura. Although each may have speciated recently, they still represent cases where biological or morphological differentiation has outstripped genetic differentiation detected by isozyme variation. Each displays aspects of their biology, and probably history, which are consistent with rapid development of reproductive isolation. In Magicicada reproductive isolation is imposed by 13 and 17 year life cycles; isolation may have been reinforced by a history of glaciations which fragmented the species' ranges (Simon 1979). In the Hawaiian drosophilids, D. silvestris and D. heteroneura, highly ritualized courtship behavior (Speith 1981) may have mediated rapid isolation after colonization-founder events triggered strong genetic shifts (Templeton 1980). More examples of lower than expected genetic distances exist in many of the other studies but are hidden in averaging (e.g., Berlocher & Bush 1982, Bentz & Stock 1986, Pashley et al. 1985, Unruh 1999) or by taxonomic confusion (e.g., Stock & Castrovillo 1981).

Low genetic distances among some species contrasts markedly with high distances among geographic populations of single species. This is true of cave crickets where some species dwell in forests and populations are relatively free to interbreed compared to species which inhabit caves and populations are strongly subdivided. In such comparisons D ranged from 0.02 for forest inhabiting species versus values exceeding 0.2 for cave inhabiting "cave" crickets (Caccone & Sbordoni 1987). Genetic differences exceeding 0.2 are probably associated with postreproductive isolating barriers in cave inhabiting invertebrates (see Caccone & Sbordoni 1987). High D is also seen for intraspecific populations of butterflies in the genus Euphydryas.

The genetic distance among entities variously defined as biotypes, subspecies or semispecies are highly variable (Unruh 1999). These must be interpreted in light of the dispersal abilities of the organisms (level of gene exchange among populations), rate of amino acid substitutions (the molecular clock), the effective size of populations (probability of fixation of genes through drift) and any behavioral or physiological attributes of the insects life system which may reinforce isolation among subpopulations (host mediated mate finding, allochrony) or which are associated with divergent selection. Historical phenomena, such as founder events and hybridization, may also produce unexpected patterns in genetic distances.

Genetic distances calculated from isozymes, particularly those that are higher than expected may prove useful in judging the rank of taxa. Low genetic distances among taxa known to be distinct based on other characteristics is a common but often inexplicable observation. Such studies can reveal new species and provide evidence for the existence of species complexes.

Bush & Kitto (1978) suggested that DNA-Hybridization was valuable for comparisons at taxonomic levels of species to class. Recent studies of parthenogenetic populations of Drosophila mercatorum (Caccone et al. 1987) and cave crickets (Caccone & Powell 1987) show that DNA melting provides resolution below the species level. Phenograms for cave crickets from DNA hybridization (Caccone & Powell 1987) are similar but not identical to those derived from isozymes (Caccone & Sbordoni 1987) and were also consistent with those based on morphology save a few exceptions. Caccone & Powell (1987) argued that the DNA hybridization data are more compelling evidence of genetic divergence than other techniques because so much more of the genome is analyzed and because amino acid substitutions (measured with isozyme, protein sequencing and immunological distance methods) may be subject to greater selective constraint than total single copy DNA. This resolution derives partly from the rate of differentiation of invertebrate DNA (high compared to mammalian DNA) and in part from the highly isolated (and genetically differentiated) nature of cavernicolous camel crickets and parthenogenetic lineages. The high rate of evolution of invertebrate DNA (Caccone & Powell 1987, Powell et al. 1986, Britten 1986) also suggests that DNA melting will prove unreliable much beyond the generic level in these taxa.

Mitochondrial DNA restriction patterns may prove to be even more sensitive in these kinds of studies. At the species level, mtDNA RFLP patterns for the homosequential (uniform inversion patterns on chromosomes) Hawaiian Drosophila have been used to clarify phylogenies already addressed by morphology, behavior, allozymes and DNA hybridization (DeSalle & Giddings 1986). Harrison (Harrison et al. 1987) used the method to analyze the structure of a hybrid zone between two Gryllus species and reviewed other similar studies.

Therefore, Unruh (1999) concluded that at specific and generic levels, isozymes, single copy DNA-DNA hybridization and rDNA restriction analysis are all potentially valuable. mtDNA analysis offers the greatest resolution for infraspecific groupings (Avise et al. 1987), but there are still too few data from insect groups to form a clear picture of its upper range of resolving power (moritz et al. 1987). At the range of species and genus, DNA-DNA hybridization and ribosomal DNA restriction analysis are most powerful. Only the latter method can be subjected to cladistic analyses. At higher taxonomic ranks, genera to family, rRNA sequencing appears to hold most promise. Immunological distance may also be considered for interspecific through family comparisons.

Biotypes.--Populations that fall into the region between species and clearly undifferentiated populations have been classified many ways. Diehl & Bush (1984) classified insect biotypes on a combination of genetic and geographic relationships. They began with the premise that some biological trait is variable and suggested that there are five categories into which biotypic variation falls: nongenetic polyphenisms, genetic polymorphisms, geographic variation, host races and species. In contrast, Gonzalez et al. (1979) provided a nomenclatoral classification of all infraspecific categories. They suggested biotypes are reproductively compatible populations which display differences in some biological attributes. This definition of biotype is more narrow and perhaps more functional, than popular usage. Neither classification excludes the other, nor collectively do they completely treat the biological patterns seen in infraspecific categories.

Molecular methods have disclosed that sympatric biotypes associated with different hosts (or habitat, etc.) are probably species if each is also characterized by striking isozyme differences (fixed allelic differences). These same isozyme markers would only allow a definition of biotypes as "geographic races," semispecies or subspecies if populations were allopatric. Most morphologically indistinguishable entities dealt with in biological control were once allopatric. Whether more genetic variation in adaptive traits exists between such allopatric populations compared to within sympatric populations is questionable (Gould 1983, Diehl & Bush 1984, Fox & Morrow 1981)

When allozymes differ only in frequency among geographic populations they can no longer provide unambiguous markers of races. Here mtDNA restriction mapping could be extremely valuable. mtDNA restriction differences can be ordered into a series from which infraspecific phylogenies (clonal lineages of the mtDNA molecule itself) can be constructed (Avise et al. 1987). Differences in the presence or absence of restriction sites among populations can be ordered into networks connecting haplotypes by the minimum number of steps. The ability to overlay these clonal phylogenies onto the geographic distribution of sexually breeding populations may allow us to separate historical and adaptive processes responsible for producing racial or biological differences in animal populations. If two biotypes are sympatric but show an mtDNA phylogeny that segregates them, species status is indicated.

One potential of mtDNA lies with the ability to find additional variation in restriction studies beyond that found with allozymes, a case made in several vertebrate taxa (Avise et al. 1987), but less well documented for arthropod species. More important is that carefully mapped restriction fragment differences can be ordered into a series (or an unrooted phylogeny) whereas frequency variation in allozyme loci cannot (Felsenstein 1982).

Establishment, Introgression and Spread of Biotypes--An important use of molecular methods for biological control is to document the establishment of new natural enemy races or species and monitoring their spread subsequent to colonization. Studies of postcolonization adaptation and introgression of races of natural enemies is substantially dependent on these molecular classifications. There are a few examples of this application of the allozyme method.

Aphidius ervi Halliday is one of several species that have been released into North America to control the pea and blue alfalfa aphids (Gonzalez et al. 1978). After several releases, comprising thousands of specimens from populations of A. ervi complex, A. eadyi, A. smithi, Praon barbatum and Ephydrus sp. at various sites in southern California, A. ervi has become the dominant parasitoid of these aphids (Unruh 1999). Isozyme analysis of established populations of A. ervi from southern California dna of early generation samples of populations from throughout its wide Palearctic distribution have clarified their genetic relationship. Genetic differentiation among four A. ervi populations from western Europe and north Africa was low (D = 0.016-0.029). This contrasts with the relatively high differentiation found in populations from Pakistan and Japan (Unruh 1999).

At one liberation site in Riverside, California, established A. ervi populations were identical to the African type (Unruh et al. 1986). A recent analysis of A. ervi from a second site in Escondido showed that these were of the Japan type. Two diagnostic allelic differences exist between the African and Japan types and not intermixing has yet been found in the field. Sexual compatibility between the Japan and African races is only slightly lower than between intrapopulation comparisons and there is no evidence of postreproductive barriers (Unruh 1999).

Isozymes were also used to clarify the status of a natural enemy of weeds. Rhinocyllus conicus, a weevil specializing on the flower heads of thistles in the Palearctic has been introduced into North America to control milk, musk and Italian thistles. Several studies suggested that at least two distinct races of R. conicus exist, one adapted to milk thistle (Silybum marianum) and another adapted to Italian thistle (Carduus nutans), and probably a third adapted to musk thistle (Carduus nutans). Studies of weevil and thistle species associations throughout Europe showed that R. conicus feeds on a small subset of potential hosts in a given region (some plants are hosts in one region but not in another, and vice versa) (Zwölfer & Preiss 1983). Early attempts to establish this weevil in California on milk thistle failed, probably because the weevils belonged to a biotype adapted to musk thistle. Eventually, all these races were established, each on its own thistle, presumably because they were matched corresponding to their original host in Europe (Goeden & Kok 1986). Subsequent to establishment of these biotypes, an electrophoretic analysis showed that the milk thistle biotype was genetically distinct from the other two biotypes at one enzyme locus and was different in allele frequencies at two others (Goeden et al. 1985).

In 1985 about a decade after the races attacking milk and Italian thistle were established on their respective hosts in southern California, R. conicus was discovered feeding and reproducing on two species of native north American thistles, Cirsium californicum and C. proteanum. Electrophoretic analysis of these populations showed that only the Italian thistle race had shifted to the native, nonpest, thistles despite an apparently equal opportunity for each race to do so (Unruh & Goeden 1987). The results were not surprising, however, since the Carduus attacking biotypes are known to accept European Cirsium spp. in their range (Zwölfer & Preiss 1983). These observations suggested that isozyme analysis, and perhaps mtDNA restriction studies, could effectively supplement biological screening (host range studies) of candidates for biological control of weeds (Unruh 1999).

Testing of Cloned or Selected Lines.--Allozymes have been useful in verifying the establishment of two synthetic pyrethroid resistant mite predators in apple orchards (Whalon et al. 1982). Unique to this study was that the isozyme loci assayed also identify the biological distinctiveness of the strains. That is, susceptible and resistant mite populations were characterized by differences in general esterases. Resistant populations displayed distinctive mobility differences in their major esterase bands and these stained much more intensely. This intensity of staining is probably reflective of higher enzyme titer (as opposed to greater activity per molecule) in the resistant forms. The esterases themselves are important in detoxifying both organophosphates and pyrethroids (Chang & Whalon 1986). Esterase of green peach aphid has been used to follow the seasonal flux of resistant and susceptible pest populations in the field (Baker 1979). Immunological procedures which detect the titer of a specific esterase have proven even more sensitive and specific than esterase assays (Devonshire et al. 1986). The future use of allozyme markers to follow the progress of genetically engineered strains is likely (Unruh 1999).

Population Biology

Only discrete characters can be effectively used as markers in studies of population phenomena. Among the methods that yield discrete characters only isozymes and to a lesser extent mtDNA are practical for large surveys of many individuals. Unruh (1999) discussed several potential applications of these methods in biological control.

Effects of Colonization.--Baker & Moeed (1987) emphasized that populations of animals or plants introduced by humans to ecologically and climatically disparate regions of the world are of great interest to evolutionary biologists because they provide developing case histories of differentiation under the constraint of reduced population size in the founding population. The effects of these genetic bottlenecks are of special interest in light of theories suggesting that rapid genetic shifts may occur in founders and may mediate speciation (Templeton 1980). The genetic systems of populations may respond to drift and a novel selective environment by a rapid shift to a new adaptive peak, which Templeton (1980) called genetic transilience. These genetic events are more apt to occur when the originating population is large and panmictic and when there are few founders. Such a shift has been postulated as an important force in the speciation of Drosophila silvestris - D. heteroneura (Templeton 1980). There are several changes likely to be evident in the genetic makeup of a newly colonized population following a bottleneck. First, most alleles which were rare in the endemic source population will be lost (Nei et al. 1975). Second, inbreeding is likely to increase as is gametic phase and linkage disequilibrium (Templeton 1980). Third, when several isolated populations are founded they are likely to display more interpopulation variation in gene frequencies (greater Fst) than would population from the native range of the introduced species.

Each of these phenomena can be estimated using polymorphic enzyme loci. In contrast, only the loss of alleles is measurable with restriction analysis of mtDNA. But, since there is only one mtDNA haplotype inherited from a mating, opposed to four of genomic DNA, the effective population size of mtDNA is one-fourth that of chromosomal genes and should be much more sensitive indicators of population bottlenecks (DeSalle & Giddings 1986). With mtDNA this would be measured as the number of restriction polymorphisms in the source population versus that in the colonized population.

Molecular studies of colonized populations are rare but they are consistent with the predicted effects. Berlocher (1984) noted higher Fst and linkage disequilibrium for eight colonized populations of the walnut husk fly. Baker & Moeed (1987) saw significantly fewer rare alleles in colonized Myna birds and St. Louis & Barlow (1988) saw the same in the Eurasian tree sparrow. The adaptive nature of these changes were not discussed, however. Latorre et al. (1986) used mtDNA restriction maps to show that Drosophila subobscura, populations which recently colonized the Nearctic were derived from European (as opposed to African) progenitors. Although not especially sensitive in the last case, mtDNA studies may be very sensitive in discovering the source of pest populations and thereby facilitate a search for adapted natural enemies.

Genetic bottlenecks occurring during and after colonization may not result in reproductive isolation and speciation but may still induce genetic shifts in natural enemies that may improve or reduce their efficacy. Also, other genetic changes, especially in highly heritable quantitative traits, may be driven by selection alone, without invoking reduced population size. However, as noted previously, the nature of polygenic inheritance is still not clearly understood in many species, and certainly the parasitic Hymenoptera are known to show some bizarre inheritance patterns (Legner 1987, 1989, 1991). Phenomena associated with colonization of biological control agents have not been examined using molecular methods, but some work with quantitative genetic changes has been reported. Myers & Sabath (1980) detected changes in polygenic traits in colonized populations of the cinnabar moth, a phytophage of tansy ragwort. Stearns (1983) observed changes in life history traits of Gambusia fish introduced to water reservoirs in Hawaii that corresponded with the stability of water levels in the reservoirs through time. Genetic bottlenecks were not implicated in either of these cases. In most biological control projects, bottlenecks are common because of the few founder individuals introduced.

Population Structure and Mating Systems.--The distribution of genetic variation among individuals within a population may be a consequence of habitat structure or phenomena intrinsic to the ecology and behavior of a species. Studies at this level can illuminate other, more far reaching aspects of natural enemy behavior and evolution. For example, Templeton's theory of genetic resilience may be particularly inapplicable to highly inbred parasitic Hymenoptera. Unfortunately, the degree of inbreeding in most parasitoids is unknown with the exception of those whose biology allows direct enumeration of the number of founders to a patch of hosts and the interactions of their progeny which issue from it.

The level of inbreeding characteristic of a particular parasitic life style has implications beyond the tendency to form species. For example, inbreeding may influence the optimal sex ratio of a population (Hamilton 1967, Charnov 1982, Waage 1986). The levels of inbreeding in a population can be estimated using isozyme variation if the discrete patches utilized by the progeny of founding female parasitoids can be adequately defined. This estimate can be used to the predicted sex ratio based on local mate competition (LMC) theory (Hamilton 1967) and compared to the observed sex ratio. Deviations from LMC predictions may arise from fitness differences between females and males developing in different size hosts and from various deviations from the assumptions of LMC (Charnov 1982, Waage 1986). Methods of estimating relatedness of individuals occurring in groups such as nests of eusocial Hymenoptera (but directly applicable to discrete patches of parasitized hosts) using isozyme or related characters has been reviewed by Pamilo (1984) and Weir & Cockerham (1984). Because mtDNA is clonally inherited it would not be useful unless sequence variation was extremely high in local populations.

Population structure at a more crude level can provide other valuable insights but caution is needed in interpreting the data. Allozyme frequency differences among populations may reflect three forces: selective differences, isolation and effective population size. Also, historical phenomena which may include the accumulated effects of one or more of these is very important but difficult, if not impossible, to separate from ongoing, dynamic (equilibrium) processes. These phenomena can seldom be isolated with certainty, and often they are hopelessly confounded. For example, allozyme based methods to estimate gene flow (Slatkin 1987), would be questionable when the assumptions of the estimation methods are violated by recent colonization events or other perturbations. Significant genetic differentiation among geographic populations could mean that populations have experienced one or several bottlenecks and gene exchange has been or is too low to erase the evidence of genetic drift, or, that there are disruptive selective forces which have shaped and maintain these differences. Although selection may be suspected if populations displayed different host use patterns or some other characteristic biological attribute, in general allozyme variation is treated as if selectively neutral (Lewontin 1974). In fact, populations may appear undifferentiated electrophoretically but still display distinctive biological attributes, such as host races (Unruh 1999).

Other Uses for Molecular Methods

Molecular methods may be used to detect internal parasitoids; both immunological methods and isozymes have been valuable (Luck et al. 1987). Isozymes have been used to detect parasitoids of Lepidoptera (Menken 1982), aphids (Wool et al. 1978, castanera et al. 1983) and of whiteflies (Wool et al. 1984). New technology, such as cDNA probes (complementary to parasitoid DNA) in "dot-blot" assays, could also be used to detect internal parasitoids. These should prove both more sensitive and more selective than immunological techniques. Probes of this sort are having wide application in plant virus detection (Jayasena et al. 1984).

Comparative studies of the level of a polymorphism in animal populations have raised the question of why do some taxa, such as Hymenoptera or aphids, display significantly genetic variation than other insect groups? The subject has been addressed for Hymenoptera (Sheppard & Heydon 1986, Unruh et al. 1986) and for aphids (Wohrmann et al. 1986, Suomalainen et al. 1980).

Isozymes have been valuable in quality control assessment of insect laboratory populations (Bush & Neck 1976, Berlocher & Friedman 1981, Pashley & Proverbs 1981, Stock & Robertson 1982, Unruh et al. 1983). as should other molecular markers if they become less expensive. However, changes in allele frequencies and decline in isozyme variation do not necessarily mean reduced performance of laboratory stocks. In mass cultures of screw worm an allele frequency changes was shown to be responsible for a loss in competitiveness (fitness) (Bush & Neck 1976), but in other studies such an association has not been made. If it is accepted that most allozymes detected by electrophoresis are selectively neutral, then loss of variation in these genes under laboratory culture provides an estimate of inbreeding and drift but not of selective changes. For some species, highly inbred lines may be as fit as their outbred natural counterparts (Unruh 1999).

Exercises:

Exercise 73.1: What do biologists typically measure when they sample arthropod populations?

Exercise 73.2: How may it be determined that natural enemies are responsible for regulating a population?

Exercise 73.3: What must be considered in the evaluation of a biological control agent?

Exercise 73.4: List and explain the five principal techniques used for evaluating biological control organisms.

Exercise 73.5: What are five common methods used for detection of predation or parasitism?

GENERAL REFERENCES: <bc-73.ref.htm> [ Additional references may be found at MELVYL Library ]