Following the Boll Weevil Eradication Program the beet armyworm, Spodoptera exigua (Hübner) (Lepidoptera: Noctuidae), emerged as the most important threat to cotton production over large areas of the southeastern United States (Haney et al. 1996). Other important hosts of beet armyworms include corn, tomatoes, alfalfa, onions, asparagus, potatoes, and citrus as well as numerous non-economic species (Hendricks et al. 1995). Prior to 1991, repeated outbreaks of S. exigua occurred regularly in Georgia and elsewhere, e.g., 1977, 1980, 1981, 1988, and 1990 (Douce & McPherson 1991).

The threat of the fall armyworm, S. frugiperda (J. E. Smith) (Lepidoptera: Noctuidae), to cotton production also has increased (Riley et al. 1997). Historically, fall armyworms have not been an important problem to cotton production. Instead, they have been associated with corn, sorghum, and coastal Bermuda grass (Metcalf et al. 1951). In corn production alone, fall armyworms have been responsible for losses of $30- to 60-million annually to corn production (Sparks 1979,Sparks 1986). Additional environmental costs also have accrued due to management programs that emphasize insecticides (Riggin et al. 1994).

Often, less than 0.1% of pesticides applied to crops reach their target (Pimentel & Levitan 1986); the remainder can affect beneficial insects and biological control agents (Pimentel et al. 1980; Ripper 1956). For instance, Cotesia (= Apanteles) marginiventris (Cresson) (Hymenoptera: Braconidae) is an endoparasitoid native to the southeastern United States (Ashley 1979). It has a broad host range comprising a variety of taxa in the Lepidoptera (Tingle et al. 1978), and it is considered important to the management of armyworms (Loke et al. 1983; Lewis & Nordlund 1980; Ashley 1979). Tillman & Scott (1997) reported susceptibility of C. marginiventris adults to commonly applied rates of selected insecticides, including acephate, azinphosmethyl, bifenthrin, cyhalothrin, cypermethrin, endosulfan, esfenvalerate, fipronil, methomyl, methyl parathion, oxamyl, profenofos or thiodicarb. This vulnerability has inspired the continued pursuit of management tactics that either minimize or alternate insecticides in order to conserve natural enemies like C. marginiventris (Loke et al. 1983, Lewis & Nordlund 1980).

Several studies have demonstrated that parasitoids can use factors liberated from the food plant of their hosts as a method of locating a potential host habitat (Vinson 1975). Many investigators (Cortesero et al. 1997; Loughrin et al. 1994; Turlings & Tumlinson 1992;Turlings et al. 1991; Turlings et al. 1990; Loke et al. 1983) have reported that corn and cotton plants on which armyworms feed are attractive to C. marginiventris adult females in laboratory experiments. Organic volatiles from such plants emitted approximately 18-24 h after commencement of feeding by beet armyworms and released on a diurnal cycle have been implicated in this attraction (Turlings et al. 1990; Turlings et al. 1991; Turlings & Tumlinson 1992; Loughrin et al. 1994; Cortesero et al. 1997). An elicitor in the oral secretion from beet armyworms induces systematic production and emission of the attractants from plants (Alborn et al. 1997). Tumlinson et al. (1993) suggested that biological control may benefit from the use of plant breeding or genetic engineering to produce strains of plants that generate greater amounts of the herbivore-induced plant attractants.

Many aspects of this chemically-mediated tritrophic relationship have been studied in laboratory, wind tunnel experiments, including the influence of herbivore-induced plant volatiles on different host-foraging strategies of C. marginiventris and on another more host-specific parasitoid Microplitis croceipes (Cresson) (Cortesero et al. 1997). However, no data have been published concerning a fundamental aspect of this chemically-mediated tritrophic relationship, which is the attraction and increased performance of C. marginiventris females in the field by plants experiencing armyworm herbivory. Herein, we investigate the influence of herbivore-damaged corn and cotton plants on the recruitment of feral C. marginiventris adult females as measured by increased parasitism. Implications for management of armyworms are discussed.

Materials and Methods

Results and Discussion

Combining data from all trials, which included field plots comprised of a center corn plant that was undamaged (control) or damaged (treated) by herbivore feeding prior to artificial infestation with larvae and four adjacent (sentinel) plants, a total of 5,921 S. frugiperda (FAW) larvae were collected. These larvae represent 32.9% of all larvae (≈18,000) with which plants were artificially infested. Of the 5,921 larvae recovered, 2,637 were from field plots comprised of a center plant damaged by herbivore feeding prior to artificial infestation and 2,225 larvae were collected from field plots comprised of a center plant undamaged prior to artificial infestation. The mean (±S.D.) percentage of larvae collected from center plants in the treated plots (19.97 ± 10.6) was not significantly different from the mean percentage of larvae collected from center plants in the control plots (20.10 ± 10.5) (Table 1).

All of the parasitoids reared from S. frugiperda larvae collected from the corn plots were C. marginiventris. The mean (±S.D.) percent parasitism of all S. frugiperda larvae collected from treated plots (4.45 ± 11.3) and control plots (5.74 ± 12.9) was not significantly different. Likewise, there was no significant difference between the mean (±S.D.) percent parasitism of S. frugiperda larvae collected from sentinel plants in treated plots (5.55 ± 11.3) and control plots (4.43 ± 7.7). The mean (±S.D.) percent parasitism of larvae collected from the center plants of each plot was significantly (F = 11.58; df = 2, 35; P = 0.0009) greater for Trial 3 (10.80 ± 17.7) than for Trial 1 (1.56 ± 4.6) or Trial 2 (2.7 ± 7.2). However, differences between treated and control field plots with respect to the mean percent parasitism of larvae collected from the center plants were not significant (Table 1). Number of plants from which parasitized larvae were recovered, percent mortality (excluding parasitism) of recovered larvae, and percent mortality (including parasitism) of recovered larvae (Table 1) were not significantly influenced by herbivore feeding on the center corn plant in the treated plots prior to larval infestation.

Trials conducted in cotton plots yielded results similar to those conducted in corn plots except the number of larvae recovered from cotton was less than the number of larvae recovered from corn. A total of 730 S. frugiperda larvae were collected from artificially-infested cotton plants in field plots comprised of a center plant that was undamaged (control) or damaged (treated) by herbivore feeding prior to larval infestation, and four adjacent (sentinel) plants. These 730 larvae represent a recovery of 6.1% of the total number (≈12,000) of larvae used to infest the plants in the two trials. 343 larvae were collected from field plots containing a center plant that was damaged by herbivore feeding prior to larval infestation, and 387 larvae were collected from field plots containing a center plant that was undamaged prior to larval infestation. Mean (±S.D.) percent larvae collected from center plants in the treated plots (21.96 ± 15.9) was not significantly different from mean percent larvae collected from center plants in the control plots (25.53 ± 17.3) (Table 2).

Similar to the trials conducted in the corn plots, all of the parasitoids reared from S. frugiperda larvae collected from the cotton plots were C. marginiventris. There was no significant difference between mean (±S.D.) percent parasitism of all S. frugiperda larvae collected from treated cotton plots (3.47 ± 5.0) and control cotton plots (5.92 ± 7.5). Likewise, there was no significant difference between mean (±S.D.) percent parasitism of S. frugiperda larvae collected from sentinel plants in treated plots (3.69 ± 5.7) and control plots (6.06 ± 8.8). Mean (±S.D.) percent parasitism of larvae collected from the center plants of each plot was significantly (F = 5.14; df = 1, 23; P = 0.0468) greater for Trial 1 (6.40 ± 10.9) than for Trial 2 (2.12 ± 9.8). However, differences between treated and control field plots with respect to mean percent parasitism of larvae collected from the center plants were not significant (Table 2). Herbivore feeding on the center cotton plant in the treated plots prior to larval infestation did not significantly affect the number of plants from which parasitized larvae were collected, the percent mortality (excluding parasitism) of collected larvae, nor the percent mortality (including parasitism) of collected larvae (Table 2).

This study represents the first field experiments testing recruitment of C. marginiventris females from feral populations by herbivore-damaged corn and cotton plants. The approaches used in these field experiments are congruent with those described previously for laboratory bioassays, accommodating such important factors as time during which volatiles are released from plants (Loughrin et al. 1994; Turlings et al. 1990), time of day during which C. marginiventris females forage most actively and during which they have been used previously in bioassays (Turlings et al. 1991; Loughrin et al. 1994), and stage at which S. frugiperda larvae are preferred hosts (Loke et al. 1983; Riggin et al. 1994). Conclusions drawn previously from the results of laboratory bioassays concerning recruitment of C. marginiventris females by herbivore-damaged corn and cotton plants are neither confirmed nor contradicted by results of the present field study. We tested for significant difference in the level of parasitism, or larval mortality between field plots containing herbivore-damaged plants and plants that were undamaged. Under the conditions of this study, we found no evidence that systemic host plant volatiles induced by herbivore feeding were used by feral C. marginiventris to improve foraging and parasitism at specific sites within a field of corn or cotton.

In view of our findings, it is interesting to consider the observations made by Ruberson & Whitfield (1996) in a study of the facultative parasitization of S. exigua eggs by C. marginiventris, conducted in cotton fields where larval populations of S. exigua were very low. Because herbivore-induced plant attractants are lacking on plants with only egg masses present and are limited or absent in fields with low or no larval populations, they surmised that C. marginiventris was quite successful at locating egg masses even when foraging cues induced by larval feeding were rare. They suggested that female wasps may have been attracted to the field by the feeding of the few larvae present and then attacked the egg masses. Also, they concluded that larval damage is not the only source of cues to which C. marginiventris is capable of responding in close-range host location. Our results are congruent with the observations and conclusions of Ruberson & Whitfield (1996). Considering the substantial body of published work on the attractiveness of herbivore-damaged plant volatiles to foraging parasitoids and our findings that female C. marginiventris did not parasitize more larvae at herbivore-damaged sites than undamaged sites, we suggest that C. marginiventris may be attracted to the field by the herbivore-induced plant volatiles, but then rely on additional cues to locate the specific site of the host larvae.

Identifying alternative or supplemental tactics to insecticides for managing armyworms in cotton and corn continues to be a valuable pursuit. Among these tactics are biological control and host plant resistance. Tumlinson et al. (1993) suggested that biological control and host plant resistance may benefit from the use of plant breeding or genetic engineering to produce strains of plants that generate greater amounts of the herbivore-induced plant attractants. Although this tritrophic system is an interesting one with considerable support from laboratory experiments, data collected from our field study reveal that knowledge of how this tritrophic system is influenced by factors in the field is incomplete. Additional field studies are necessary before conclusions may be drawn about the potential of exploiting these herbivore-induced plant attractants to serve industry and the agricultural community.