Eliciting maize defense pathways aboveground attracts belowground biocontrol agents

Plant defense pathways mediate multitrophic interactions above and belowground. Understanding the effects of these pathways on pests and natural enemies above and belowground holds great potential for designing effective control strategies. Here we investigate the effects of aboveground stimulation of plant defense pathways on the interactions between corn, the aboveground herbivore adult Diabrotica speciosa, the belowground herbivore larval D. speciosa, and the subterranean ento-mopathogenic nematode natural enemy Heterorhabditis amazonensis. We show that adult D. speciosa recruit to aboveground herbivory and methyl salicylate treatment, that larval D. speciosa are relatively indiscriminate, and that H. amazonensis en-tomopathogenic nematodes recruit to corn fed upon by adult D. speciosa. These results suggest that entomopathogenicnematodes belowground can be highly attuned to changes in the aboveground parts of plants and that biological control can be enhanced with induced plant defense in this and similar systems.

consequences for belowground herbivores 18 . These effects need not be reciprocal or equivalent; in some cases, belowground herbivory can stimulate responses in both the root and the shoot, while aboveground herbivory elicits shoot only responses 19 .
The role for plant defense pathways in mediating tritrophic responses to pests and natural enemies both above and belowground has been little explored, particularly in regard to behavioral effects of defense induction at higher trophic levels. To explore these interactions, we investigate the effect of aboveground induction of plant defense pathways in corn (Zea mays L.) on recruitment of adults of the corn rootworm pest Diabrotica speciosa (Germar) (Coleoptera: Chrysomelidae) aboveground, recruitment of larval D. speciosa belowground, and recruitment of the entomopathogenic nematode Heterorhabditis amazonensis belowground (Fig. 1).
While the adult beetle D. speciosa is a widespread, prominent, and polyphagous pest of corn and soybeans in South America, much of the damage to corn results from D. speciosa larvae feeding on corn roots 20,21 . H. amazonensis is a natural enemy of D. speciosa present endemically and sometimes applied for biological control 22,23 . Corn, D. speciosa, and H. amazonensis constitute a representative model system; the effects of plant defense pathway stimulation on the above and belowground multitrophic interactions present in this system hold ramifications and implications not only for our understanding of such systems generally, but also for designing intelligent biological control strategies in the field.

Results
To investigate the effect of aboveground induction of plant defensive pathways in corn on recruitment of pests and natural enemies above and belowground, we presented undamaged, mechanically damaged, pest damaged, and elicitor treated plants to adult beetle D. speciosa, larval D, speciosa, and H. amazonensis entomopathogenic nematodes in multiple-choice and two-choice olfactometers.
Adult D. speciosa Response. To determine effects of plant defense induction on pest behavior aboveground, adult D. speciosa were presented with combinations of undamaged, mechanically damaged, pest damaged, and elicitor treated plants in two-choice, Y-tube olfactometers. Pest damaged plants had been fed upon aboveground by five adult D. speciosa for 48 hours prior to the start of the experiment. Elicitor treated plants received foliar applications of solutions containing methyl salicylate or methyl jasmonate; soils and roots did not receive treatment. Responses to treatment combinations were monitored, as was the time spent in making the choice with the idea that preference would be revealed not only by choosing a particular treatment, but also faster response times to that treatment.
Indeed, multivariate analysis of variance revealed that treatment had a significant (F = 2.63; df = 15,146; P = 0.002) effect on adult D. speciosa preference and response time in Y-tube bioassays. Data conformed to assumptions of normality (visual inspection; P > 0.05, Shapiro Wilk Test) and homoscedasticity (P = 0.18, Levene's Test). The linear combination of preference and response time identified by MANOVA and used as the  With this index, adult D. speciosa demonstrated a trend to preferring higher levels of damage; corn fed upon by other adult D. speciosa significantly (P = 0.045, Dunnett's Test) recruited more adult D. speciosa, faster (Fig. 2a). Interestingly, methyl salicylate (MeSA) treated corn seedlings recruited more adult D. speciosa, faster when contrasted with undamaged (P = 0.001, Dunnett's Test) and mechanically damaged corn seedlings (P = 0.01, Dunnett's Test) (Fig. 2b). Contrasts with methyl salicylate treated plants versus pest damaged plants and methyl salicylate treated mechanically damaged plants seemed to negate that effect (P > 0.17, Dunnett's Test). Likewise, methyl jasmonate (MeJA) contrasts did not elicit significant effects (P > 0.14, Dunnett's Test; Fig. 2c). The preference and faster response of adult D. speciosa to pest damaged and methyl salicylate treated corn seedlings in y-tube olfactometers (Fig. 2d) suggests that these adult beetles are attuned to volatile changes in the host plant. This similarity could indicate similarities in plant defense pathway induction between D. speciosa feeding and induction of the salicylic acid pathway. It appears that adult beetle feeding and methyl salicylate application elicit similar behavioral responses in recruitment of other adult D. speciosa.
Larval D. speciosa Response. To determine effects of aboveground plant defense induction on pest behavior belowground, larval D. speciosa were presented with combinations of elicitors, undamaged, pest damaged, and elicitor treated plants in two-choice, sand-filled olfactometers.
To first test whether larval D. speciosa would recruit to the elicitors themselves, solutions containing either methyl salicylate (MeSA) or methyl jasmonate (MeJA) were placed on filter paper in opposing ends of an inverted-T olfactometer. Larvae were released into the center then, after 24 hours, responding larvae were removed from the end caps and counted. Larval D. speciosa significantly (P = 0.002, t-test) preferred germinated corn seeds (positive control), but did not significantly (at α = 0.05) respond to elicitor solutions (Fig. 3).
To test whether larval D. speciosa would recruit to corn seedlings in which defense pathways had been induced, D. speciosa larvae were presented with treated seedlings placed in opposing ends of a two-choice olfactometer. Treated seedlings received foliar applications of elicitor solutions containing either methyl salicylate (MeSA) or methyl jasmonate (MeJA) while pest damaged seedlings were fed upon aboveground by three adult D. speciosa for twenty-four hours prior to the start of the experiment. While larval D. speciosa recruited readily to corn seedlings versus blank controls (P = 0.009, t-test), elicitor treated and pest damaged plants did not significantly (at α = 0.05) recruit additional larvae (Fig. 3).
The response of larval D. speciosa to corn seedlings suggests that, while larvae can detect and recruit to the presence of a food source (corn roots), induction of plant defense pathways aboveground, whether by herbivory or elicitor application, have little effect on larval behavior. This could potentially be due to a lack of plant defense pathway mediated changes belowground on the part of the plant, or due to an inability to detect and respond on the part of the larvae. (d) Y-tube apparatus used for assays. Clean humidified air is pumped into the chambers containing treatments then subsequently into the Y-tube. Adult beetles are released in the bottom of the Y and allowed to choose between treatments. D. speciosa Time Preference index is constructed from average response times and preferences for the top treatment in a pair (i.e. for Corn in Corn vs Air); higher numbers indicate a faster response and a greater preference for the top treatment in a pair. Points and error bars denote mean D. speciosa response (n = 10) and standard error respectively. A single asterisk denotes significant differences (P < 0.05) in response as compared with a baseline of Air vs Air. Double asterisks denote significance at P < 0.01.

Entomopathogenic Nematode H. amazonensis Response.
To further explore the effects of aboveground plant defense induction belowground, we examined behavior of infective juveniles of the entomopathogenic nematode H. amazonensis in sand-filled multiple choice olfactometers (Fig. 4). These natural enemies, which hold great potential for biological control, were presented with combinations of undamaged, mechanically damaged, pest damaged, and elicitor treated plants. Pest damaged plants were fed upon aboveground for 48 hours by five adult D. speciosa; elicitor treated plants received foliar applications of solutions containing methyl salicylate (MeSA) or methyl jasmonate (MeJA).
Combined, theses results suggest that the entomopathogenic nematode H. amazonensis is highly attuned to differences in status of corn plants upon which their insect hosts feed; they recruit to herbivorized plants and can demonstrate differential responses to mechanically damaged and pest damaged seedlings. Interestingly, stimulation of plant defense pathways through foliar elicitor application seems to mimic much of the same effects suggesting similarities between herbivory by adult D. speciosa and induction of plant defense pathways.
Multitrophic Implications. Taken together, these results suggest a role for plant defense pathways in mediating the interactions between plants, herbivores, and natural enemies above and belowground. Indeed, while adult D. speciosa are somewhat responsive to herbivory and induction of plant defense pathways, H. amazonensis entomopathogenic nematodes belowground seem to be particularly attuned to aboveground induction of plant defense pathways (Table 1). In contrast, larvae do not seem to discriminate or respond to elicitor treatments in our assays (Table 1).

Discussion
The effects of plant defense pathway stimulation aboveground on herbivore and natural enemy behavior above and belowground has implications for understanding and managing the interactions between corn, D. speciosa, and H. amazonensis as well as similar interactions in other systems. The ability of plant defense pathway stimulation aboveground, whether through herbivory by adult D. speciosa or methyl salicylate application, to recruit adult D. speciosa aboveground suggests a role for plant defense pathways in mediating adult distributions in the field. Adult D. speciosa are well known to monitor and respond to plant volatiles in the lab and in the field 24,25 , responding particularly well to cucurbitacins which may influence progeny fitness 26 . In our case, herbivory and stimulation of the salicylic acid pathway may induce release of compounds recognized as favorable by the adult D. speciosa. Such recognition may signal presence of other individuals, more available resources, or reduced plant defenses. Adult D. speciosa likely recruit to volatiles released by the plant to take advantage of such conditions. Indeed, observations from collections in the field support that hypothesis; adult D. speciosa are seldom found individually on corn plants in the field. Corn plants typically host many adult D. speciosa, most of them mating pairs. Volatiles emitted by prior feeding or stimulation of plant defense pathways may contribute to aggregation and mate finding in adult D. speciosa. Points and error bars denote mean percent response (n = 4) and standard error respectively. A period denotes marginal significance at P < 0.1; a single asterisk denotes significant differences at P < 0.05; double asterisks denote significance at P < 0.01; triple asterisks denote significance at P < 0.001.

Table 1. Responses of herbivores (adult and larval D. speciosa) and natural enemies (the entomopathogenic nematode H. amazonensis) to undamaged, mechanically damaged, pest damaged (fed upon by adult D. speciosa) and elicitor treated corn seedlings. + indicates attraction; − indicates repellences;
ND inducates no significant difference detected; NA indicates the combination was not assayed.
Interestingly, stimulation of plant defense pathways aboveground seems to have little effect on larval D. speciosa behavior belowground. This observation could be a boon for strategies using induced plant defenses to augment biological control in conjunction with other strategies to repel D. speciosa larvae 27 . Because larvae do not preferentially respond to elicitor treated plants, induced plant defense will likely not attract belowground larval pest populations in the field while augmenting control through recruitment of entomopathogenic nematode natural enemies.
Our results suggest that entomopathogenic nematodes can respond to stimulation of plant defense pathways aboveground through herbivory or elicitor application. Likewise, because H. amazonensis does not recruit to mechanially damaged plants, our results suggest that mechanical damage alone is not sufficient to stimulate a plant defense response. Indeed, monitoring of plant defense pathway stimulation by aboveground herbivory on the part of the entomopathogenic nematode H. amazonensis may hold adaptive significance; feeding aboveground by the adult D. speciosa indicates a greater likelihood of eventually finding suitable larval hosts, especially for cruiser nematodes like H. amazonensis which can travel large distances belowground in search of food 28,29 . As adult D. speciosa recruit to herbivorized plants where they are likely to encounter other members of their species, they are apt to mate and reproduce on the same plant, letting the eggs fall to soil in the immediate area. After hatching, the larvae seem non-discriminatory and will recruit to the nearest plant root. Entomopathogenic nematodes seem to have evolved to take advantage of this system, responding to the decision makers aboveground -the adult D. speciosa -in the search for food.
This search for food on the part of entomopathogenic nematodes can be augmented in this and similar systems. Aboveground, elicitor application can be considered as a means of influencing D. speciosa distributions. Belowground, aboveground application of elicitors may also be considered for influencing distribution of entomopathogenic nematodes. Additionally, the volatiles involved in mediating entomopathogenic nematode attraction to aboveground herbivory can be used to augment biological control of larval D. speciosa belowground. We are exploring that possibility.

Methods
Plant Materials. Bt transformed Herculex I (Dow AgroSciences, Pioneer Hi-Bred International) corn seedlings expressing the Cry1F gene were used in all experiments. This variety was developed primarily to control Fall Armyworm (Spodoptera frugiperda) and is in widespread use in Brazil. Seeds were germinated in moist vermiculite, then grown for twenty days in organic substrate. Prior to use in belowground bioassays with larval D. speciosa and entomopathogenic nematodes, corn seedling roots were gently washed to remove substrate before placement in olfactometers.
Insect Rearing. Adult D. speciosa were collected from corn fields maintained by the Federal University of Lavras (Lavras, MG, Brazil) and taken directly to the laboratory for rearing following previously established methodology 30 . Adults were maintained on bean leaves (Phaseolus vulgaris) while eggs were collected from black gauze strips placed with the adults to induce oviposition. Eggs were washed from the gauze strips every two days then placed in petri dishes with moistened filter paper to maintain humidity until eclosion. Larvae were maintained on recently germinated corn seedlings in vermiculite until pupation. Eight days after eclosion, larvae were used in bioassays. Adult D. speciosa used in bioassays were taken directly from the field and used within a week of collection.
Nematode Rearing. Entomopathogenic nematode H. amazonensis infective juveniles used in multiple choice olfactometers were obtained from cultures maintained at the Federal University of Lavras where the nematodes were reared on greater wax math (Galleria mellonela) larvae. Wax moth larvae were reared in the laboratory 31 and maintained on artificial media 32 . When nematode infective juveniles were needed, G. mellonela larvae were inoculated with entomopathogenic nematodes 33 and infective juveniles subsequently collected on White traps 34 . After collection, nematodes were maintained in culture flasks in aqueous suspension at 16 ± 1 °C and used in bioassays within a week of collection.
Elicitor Preparation. For preparation of elicitors for application to the aboveground parts of maize seedlings, Tween 20 and ethanol were added to water such that final concentrations were 0.1 mL/L and 2.5 mL/L respectively. Concentrations of methyl salicylate and methyl jasmonate were adjusted to 0.5 mM then added to solution such that the amount used per plant per experiment was 65 μ L and 109 μ L respectively. Aliquots of 30 mL elicitor solution were applied to the aboveground foliage of corn seedlings using a spray bottle, a quantity sufficient for the corn seedling to become wet and to ensure homogeneous application. Contact of elicitor solution with the roots was prevented by an aluminum foil barrier.
Adult D. speciosa Bioassays. The responses of adult D. speciosa to treated and untreated corn plants were evaluated in two-choice Y-tube glass olfactometers (Fig. 2d). Aboveground portions of plants with the desired treatments were placed in glass chambers where filtered, humidified air was introduced then pumped via teflon tubes to the olfactometer, one treatment per arm. Adult D. speciosa were introduced at the base of the Y-tube olfactometer and allowed five minutes to choose an arm; response time and treatment choice were monitored. Treatments were rotated every three insects to avoid positional effect. Ten replications of each treatment combination were conducted with each replicate consisting of ten insect choices. Clean glassware and new plants were used for each replicate.
Treatment combinations consisting of undamaged, mechanically damaged, pest damaged, and elicitor untreated and treated plants were used to determine the effect of plant defense pathway stimulation on D. speciosa response. Twenty day old corn seedlings were used in all experiments; undamaged corn seedlings were taken directly from the greenhouse 20 days after germination and immediately used in the experiment. Mechanically damaged corn seedlings were cut with a scalpel using a template replicating foliar damage by adult D. speciosa 48 hours prior to use in the experiment. Pest damaged plants were fed upon by five adult D. speciosa placed in mesh bags on the foliage of each plant 48 hours prior to use in the experiment. Elicitor treated (either methyl salicylate or methyl jasmonate) plants were treated 48 hours prior to use in experiments.
Larval D. speciosa Bioassays. To examine larval D. speciosa responses belowground, two-choice sand filled olfactometers were used (Fig. 3b,d). Two choice olfactometers were constructed from 3/4 inch (1.9 cm) PVC pipe filled with washed autoclaved sand maintained at 12% moisture by volume. Larval responses were evaluated to treatments consisting of undamaged seedlings, damaged seedlings, elicitor treated seedlings, and elicitors.
For evaluating larval responses to volatile chemicals, treatments were applied to filter paper placed in the end caps of sand-filled, inverted T olfactometers (Fig. 3b). The above elicitor solutions were added in 30 ml aliquots to filter paper and allowed to dry. As above, controls consisted of the Tween 20 and ethanol solution without elicitors. Germinated corn seeds were used as positive controls. After placing filter paper or corn controls in the end caps of the sand-filled olfactometer, five 8 day old D. speciosa larvae were released into the center of the olfactometer. After an additional 24 hours, responding larvae were collected from the elbows and counted. Ten replications of each treatment combination were conducted.
For evaluating larval responses to elicitor treated plants, inverted olfactometers connected to PVC elbows were used (Fig. 3d). Seedlings were placed in opposing arms, appropriate treatments applied, and the olfactometer filled with sand. Insect damaged seedlings received three adult D. speciosa for twenty four hours prior to use in the experiment. For elicitor treated plants, elicitor solutions were applied to the aboveground parts of maize seedlings. Elicitor solutions were prepared as described above. Control plants (undamaged seedlings) received the Tween 20 and ethanol solution without elicitors. Twenty-four hours after application, five 8 day old D. speciosa larvae were released into the center of the olfactometer. After an additional 24 hours, responding larvae were collected from elbows and counted. Four replications of each treatment combination were conducted.
Nematode Bioassays. Entomopathogenic nematode H. amazonensis response to treated and untreated corn plants was evaluated in sand-filled multiple-choice olfactometers consisting of a central chamber connected to eight arms into which corn seedlings were inserted (Fig. 4d). Olfactometers were constructed from 30 cm diameter plastic containers (Tupperware) perforated at equally spaced intervals to which eight 4 cm diameter PVC elbows were connected. Seventy two hours prior to the start of the experiment, olfactometers received corn seedlings and were filled with washed, autoclaved sand adjusted to 12% moisture by volume. After 72 hours of acclimation to the new environment, the aboveground foliage of corn seedlings was treated as described above to evaluate the ability of undamaged, mechanically damaged, pest damaged (above ground feeding by adult D. speciosa), and elicitor treated plants to recruit entomopathogenic nematodes belowground.
Treatment contrasts were arrayed around an olfactometer in an alternating pairwise fashion. For example, Blank vs. Undamaged contrasts were arranged as depicted in Fig. 4d. Four controls were alternated with four undamaged plants around the olfactometer for one replicate. Forty eight hours after treatment application, 2500 H. amazonensis infective juveniles were released in the center of each olfactometer. An additional 24 hours later, the olfactometers were disassembled, the responding nematodes extracted via Baermann funnel, and evaluated. Tests of known amount of nematodes placed in Baermann funnels and replicated 20 times yielded an extraction efficiency of 13.1 ± 1.4%. Nematode counts were adjusted accordingly; four replications of each treatment combination were evaluated. Statistical Analysis. Response times and preferences for adult D. speciosa in two-choice, Y-tube bioassays were analyzed using multivariate analysis of variance (MANOVA). Mean response time and proportion of adults responding to treatment of interest were calculated for each replicate; bioassay treatments were used to model variation in those two parameters. Diabrotica Time Preference Indices were constructed from the first linear discriminant function and evaluated with Roy's greatest characteristic root test. Conformation to assumptions of normality and homoscedasticity was verified through visual examination of residual plots, the Shapiro-Wilk test, and Levene's test. Significant results from MANOVA were further explored with Dunnett's test (to avoid inflated Type I error), comparing treatments of interest to baseline adult D. speciosa responses to Air vs Air trials.
Larval D. speciosa responses in two-choice olfactometers were analyzed using one-sample t-tests after subtracting number of larvae responding to controls from responses to treatment of interest. Responses were examined for adherence to assumptions of normality by visual inspection with quantile-quantile plots and the Shapiro-Wilk test.
Nematode responses in eight-arm, sand-filled olfactometers were analyzed using linear models and analysis of variance with treatment and contrast (treatment combination) as factors. To avoid effects of aggregation (presented with equivalent treatments, nematodes will often aggregate in one treatment arm [35][36][37] , responses to individual treatments were summed within replicates. Assumptions of normality and heteroscedastity were examined by visual inspection of residual diagnostic plots, the Shapiro-Wilk test and Levene's test. Post-hoc comparisons of nematode responses were conducted with Dunnett's test to avoid α inflation.
All data were collated in Microsoft Excel then read into R version 3.2.2 38 for analysis. RStudio version 0.99.484 39 was used as a development environment. Various supplementary packages were used in R for additional functionality: xlsx 40 for interface with Microsoft Excel, tidyr 41 and dplyr 42 and tidyr 43 for data arrangement and summary statistics, ggplot2 44 for graphics capabilities, car 45 for MANOVA statistics, and multcomp 46 and lsmeans 47 for multiple comparisons using Dunnett's test.