Introduction

The Indian meal moth, Plodia interpunctella (Lepidoptera: Pyralidae) (Hübner), is a cosmopolitan pest of stored products found in bulk, processing, warehouse, and retail systems (Cox and Bell 1991). The range of products attacked by this pest includes: raw and finished cereals, nuts, dried fruit and vegetables, pulses, and garlic (Perez-Mendoza and Aguilera-Pena 2004). Moths cause damage by infesting both unprocessed, bulk products as well as finished, retail commodities. In bulk products, they are generally considered to be a nuisance pest. However, infestations in retail stores and warehouses result in significant damage, because the presence of a single larva or adult may result in the total loss of the infested product, as well as potential customer complaints or even lawsuits. The traditional management strategy for P. interpunctella has been the use of chemical insecticides, including: fog, fumigant, and grain protectants (Cox and Bell 1991). However, recent legislation has reduced the availability of chemical insecticides for use on stored products (Arthur and Rogers 2003).

Trichogramma (Hymenoptera: Trichogrammatidae) egg parasitoids have recently been studied as augmentative biological control agents for P. interpunctella in stored product systems, including: bulk peanut storage (Brower 1988), bulk wheat storage (Schöller et al. 1996), bakeries (Prozeil and Schöller 1998; Steidle et al. 2001), as well as in warehouses and retail stores (Schöller et al. 2006). Trichogramma spp. present some unique advantages that make them well-suited for use in the high visibility/low economic threshold systems characteristic of finished stored products. These include: causing mortality to the egg stage prior to the damaging larval stage, commercial availability, and perhaps most importantly, tiny body size (< 0.5 mm in length), making them unlikely to be noticed by consumers.

Selection of an appropriate species and/or strain of biological control agent is a critically important part of any augmentative biological control program (Smith 1996; Van Driesche and Bellows 1996), and has been discussed specifically in regards to choosing trichogrammatids to manage stored product moths (Schöller et al. 1994). The fact that P. interpunctella, unlike other pyralid moths such as Ostrinia nubilalis, lays its eggs both singly and in small clusters, and does so either directly on exposed products (Arbogast and Mullen 1978) or adjacent to compromised packages in response to food odors (Mullen 1994), results in a heterogeneous distribution of small (< 10) localized patches of host eggs. In general, variation in intra- and interpatch host density is known to affect parasitoid foraging behavior and success (Hassell 1982). In studies of Trichogramma spp., increased numbers of host eggs within patches have been shown to have a positive effect on host patch location (Morrison et al. 1980; Kfir 1983; Reznik and Umarova 1991; McCravy and Berisford 1998; Barnay et al. 1999) and foraging activity (Nordlund et al. 1981).

Therefore, the effects of four intra- and interpatch densities of P. interpunctella eggs were examined on the host-foraging success of individual naïve females of three Trichogramma species. The experimental objective was to select the most suitable species for use in an augmentative biological control program for this pest in retail stores and warehouses. Experimental assessments included measurement of overall parasitism, as well as a simple assessment of the relative aggregation of parasitism between two intrapatch host densities. The latter assessment was perceived as important relative to the management of P. interpunctella due to this pest's tendency to lay single or small clusters of eggs adjacent to compromised food packages or product spills (Arbograst and Mullen 1978; Mullen 1994), which may result in a concentration of eggs within the vicinity of such host resources. Thus, a more aggregated pattern of parasitism is likely to be desirable, as it would lead to a more thorough exploitation of areas likely to contain hosts. This experiment was performed to complement additional studies evaluating the foraging efficiency of the same three Trichogramma spp. in response to smaller scale habitat complexity in the form of small quantities of millet or flour (Grieshop et al. 2008), as well as their foraging response on retail shelving in the presence or absence of packaged goods (Grieshop et al. 2007b).

The three commercially available species selected for comparison were: T. deion (Pinto and Oatman), T. ostriniae Pang and Chen, and T. pretiosum (Riley). The commercial strains of T. deion and T. pretiosum used in this study were selected based on host preference tests performed by Schöller and Fields (2002). Their study demonstrated that these two species readily parasitized the eggs of P. interpunctella under both choice and no-choice conditions, while two strains of T. minutum Riley, and individual strains of T. sibericum Sorokina and T. brassicae Bezdenko did not. Furthermore, T. pretiosum has been collected from wild stored product moth populations in peanut warehouses (Brower 1984). T. ostriniae has been extensively explored as an augmentative biological control agent for O. nubilalis in field and sweet corn (Wang et al., 1997, 1999; Hoffman et al. 2002; Wright et al. 2002), was shown capable of parasitizing the eggs of P. interpunctella in bulk wheat (Jeffery Gardener, personal communication), and has a history of being reared on a similar stored product pyralid Ephestia kuhniella (Smith 1996). Likewise all three Trichogramma strains used in this study had been reared on E. kuhniella for more than 100 generations (Synthia Penn, personal communication; Jeffery Gardener, personal communication).

Materials and Methods

Data analysis

Egg patches and egg parasitization data were analyzed using the mixed procedure (SAS 2001). Replicates with zero parasitism were removed from the analysis to reduce variability, resulting in 15, 13, and 14 replicates for T. deion, T. pretiosum, and T. ostriniae, respectively. The numbers of egg patches and eggs parasitized were analyzed in a three by four mixed model ANOVA with fixed factors consisting of species and the four treatments, and with species nested within experimental runs as a random factor. In addition, four three-way mixed model ANOVAs were run to test for differences among species within treatments, with species used as the fixed factor and species nested within experimental runs as the random factor. Three four-way ANOVAs, one for each species, were run for both the number of egg patches and eggs parasitized. Degrees of freedom were adjusted using the Kenward-Rodgers correction for mixed models with nesting, and multiple comparisons were run for all models using a Tukey-Kramer adjusted LSD (p = 0.05). Additional paired t-tests were run for each species comparing the number of egg patches parasitized for the nine single-egg patches treatment with the nine four-egg patches treatment, as well as the 36 single-egg patches treatment with the 36 four-egg patches treatment. Similar, paired t-tests were run for the number of eggs parasitized for the 36 single-egg patches treatment with the nine four-egg patches treatment. To assess whether there were differences in the number of non-responsive females among the three species and four treatments a two-way χ² test was performed.

In addition to patch use and parasitism, the landscape class metric “normalized aggregation index” (nAI) was calculated for egg patches parasitized by each species in the 36 single-egg patches treatment and the 36 four-egg patches treatment. The nAI equals the total length of edge (or perimeter) involving the corresponding class, given in number of cell surfaces, minus the minimum length of class edge (or perimeter) possible for a maximally aggregated class, also given in number of cell surfaces, which is achieved when the class is maximally clumped into a single, compact patch, divided by the maximum minus the minimum length of class edge (McGarigal et al. 2002). The nAI has a range from zero to one, with a value of zero indicating a maximally aggregated class and a value of one indicating a maximally dispersed class.

For the purpose of nAI calculation, patches were treated as 7.5 x 7.5 cm cells in a 6 x 6 landscape array with two class values denoting either a parasitized patch or a patch lacking parasitism. nAI was calculated using the Fragstats software package (McGarigal et al. 2002) using the equation:

Where e_i equals the actual total length of edge (or perimeter) of class i, min e_i equals the minimum possible total length of edge of class i based on the number of cells in class i, and max e_i equals the maximum possible total length of edge of class i based on the number of cells in class i. nAI values for the two treatments were analyzed for each of the three species using a paired t-test. Values were transformed using an angular transformation (arcsin(y^0.5)) to stabilize variance.

Results

Number of non-responding females

A similar pattern of non-responsive females was found among the three species with the highest number of non-responsive females found in the nine single-egg patches treatment and the lowest number found in the 36 four-egg patches treatment (X²= 1.80, df = 3, p = 0.62). The exception was T. ostriniae, where the lowest number of non-responsive females was found in the 36 single-egg patches treatment (Table 3).

Figure 1.

Mean (+SEM) patches and eggs parasitized by species and treatment. Bars with different lowercase letters over them were found to be significantly different among species within the same treatment, as were whole treatments with different uppercase letters (Tukey-Kramer adjusted LSD with p < 0.05). High quality figures are available online.

Discussion

Although patterns of parasitism varied considerably among species, no one species performed consistently better than the other two. The fact that all three species parasitized the most eggs where egg density was highest (the 36 four-egg patches treatment), and the fewest where host density was lowest (the nine single-egg patches treatment) suggests that host-finding occurred in a density-dependent manner (Figure 1). The pattern of density-dependent parasitism is consistent with observations reported for Trichogramma under laboratory (Kfir 1983) and field conditions (Wang and Ferro 1998). For T. deion and T. ostriniae, the number of host patches parasitized was highest when interpatch distance was lowest (36 patch treatments) (Table 1). Other authors have attributed increased host patch discovery to increased intrapatch density, both in the laboratory (Barnay et al. 1999), and in the field (Morrison et al. 1980; McCravey and Berisford 1998).

Table 1.

Mean ± SEM number of patches parasitized and number of eggs parasitized by Trichogramma species and treatment (eggs/patch).*

Table 2.

Paired t-test results for number of patches parasitized and number of eggs parasitized by trichogrammid species in test arenas (differences between means tested).

Table 3.

Number of replicates by trichogrammid species and patch size treatment with non-responding females (out of 15).

The three species responded differently in intra- and inter-patch comparisons. For T. deion, the number of patches parasitized was about two times greater in the nine four-egg patches treatment compared to the nine single-egg patches treatment. However, patches parasitized were similar in the 36 four-egg and single-egg patches treatments (Table 1). In contrast, T. ostriniae parasitized significantly more four-egg patches in both nine and 36 patch arrangements, and there were no intrapatch differences for T. pretiosum (Table 2). One possible explanation for the differences in response to host densities among species is that the three species perceive “patch structure” differently. The greater spatial aggregation of patches parasitized in the 36 four-egg patches treatment compared with the nine single-egg patches treatment for both T. deion and T. ostriniae could be due to a foraging response to “meta” or super patches consisting of more than a single disk when four-eggs were present (Figure 2). In contrast, the lack of difference for T. pretiosum suggests that T. pretiosum may have a tendency to respond to the egg patches as individual patches irrespective of intrapatch density. The difference between T. pretiosum and the other two Trichogramma species may be the result of the latter two species exhibiting an area-restricted search in response to the discovery of patches containing more hosts. For T. ostriniae this may be due to an evolutionary association with a host that lays dense egg masses versus sparsely distributed eggs.

Figure 2.

Mean (+SEM) normalized aggregation index for one-egg-36 patch and four-egg-36 patch arenas. Values closer to zero indicate a higher level of aggregation. Pairs of bars with stars over them are significantly different (Paired t-test ± < 0.05, ** < 0.01). High quality figures are available online.

An alternative hypothesis that may explain increased patch discovery as interpatch density increases is that host patch discovery depends on the detection of a minimum level of kairomones associated with the host. Although untested in this study, the host eggs may have produced volatile compounds identifiable by Trichogramma and used as host-foraging cues since other studies have shown that Trichogramma responds to kairomones produced by or associated with host eggs, including scales from adult moths, sex pheromones, and compounds associated with egg metabolic processes (Jones et al. 1973; Morrison and Nordlund 1980; Nordlund 1981; Boo and Yang 2000). In addition, it has been hypothesized that Trichogramma spp. require a minimum amount of stimulus prior to engaging in normal foraging behavior (Resznik and Umarova 1991). Given that a greater number of non-responsive females were consistently found at the lowest egg density is in general agreement with this hypothesis (Table 3). Further evidence in support of this hypothesis is the nearly significant trend of a greater number of eggs parasitized by T. deion and T. ostriniae for the nine four-egg patches treatment compared to the 36 single-egg patches treatment (p = 0.0914 and p = 0.0738, respectively) (Table 1). There may be different thresholds for density-dependent host-foraging responses in Trichogramma. For example, T. ostriniae exhibited greater parasitism only at the highest of the four host densities (Table 1).

The host densities and arrangements tested in this experiment were chosen because they provided an approximation of how P. interpunctella distributes eggs in warehouses and stores. A study by Arbogast and Mullen (1978) suggests that this distribution consists of single or small groups of eggs (< 10) per local group aggregated around a compromised package or food product spill. An experiment performed by Grieshop et al. (2006) testing T. deion's ability to manage P. interpunctella in small arenas containing either bulk corn meal or cornmeal in compromised packaging clearly indicated that the parasitoid was largely ineffectual at reducing P. interpunctella populations on bulk product, but reduced it by more than 75% in packaged products. Thus, trichogrammatids that respond to low densities of hosts and thoroughly search in areas where low numbers of eggs are found (i.e., on packages or adjacent shelf space,) would theoretically be better suited to manage this important stored product pest. In this case, both T. deion and T. ostriniae demonstrated a more aggregated search pattern at the highest host density, suggesting that they may perform better when foraging in finished stored product environments (Figure 2).

Two other experiments using the same three species, but conducted at different spatial scales, support the idea that T. deion is the best-suited species for biological control of P. interpunctella. The first was conducted at a smaller scale and examined the impact of microhabitat complexity on single females (Grieshop et al. 2008), while the second was conducted at a much larger scale and examined the foraging efficacy of mass-released parasitoids on retail shelving units (Grieshop et al. 2007). The failure of T. deion to outperform the other two species in this experiment may be due to either: 1) the difference in spatial scale among the present and previously published experiments, or 2) the homogeneous nature of the arenas used in this trial. Thorpe and Dively (1985) demonstrated that three species of Trichogramma responded differently depending on the size of arena used in laboratory tests. In these trials, two species of Trichogramma performed best at a smaller scale, while the third performed best at the largest spatial scale observed. Alternatively, multiple investigators have shown that habitat complexity affects Trichogramma foraging success in the laboratory (Andow and Prokrym 1990; Lukianchuk and Smith 1997; Gingras and Boivin 2002; Gingras et al. 2002, 2003; Andow and Olson 2003). Thus, it is possible that physical/structural homogeneity of the arenas used in this trial may have predisposed all three species to parasitize more hosts than when they were in a more heterogeneous habitat.

Although not entirely conclusive, these results suggest that T. deion is the best-suited species for biological control of P. interpunctella. T. deion parasitized slightly more host eggs than the other two species in all four treatments and significantly more than T. pretiosum in the 36 four-egg patches treatment (Figure 1). Furthermore, while intrapatch egg density appeared to increase patch discovery for T. ostriniae at the greater interpatch density, T. deion did not appear to be affected by this factor. Because the presence of a single P. interpunctella larva, pupa, or adult can result in the total economic loss of the product (Subramanyam et al. 2001), location of individual host eggs is of critical importance for Trichogramma used in retail stores or warehouses. Other studies comparing the three species' performance at either a finer spatial scale in response to microhabitat complexity or at a broader spatial scale examining the ability of mass-released Trichogramma to locate eggs on shelving units found that T. deion parasitized/killed significantly more P. interpunctella eggs than the other two species (Grieshop et al. 2007, Grieshop et al. 2008).