Potential of in vivo- and in vitro-cultured entomopathogenic nematodes to infect Lobesia vanillana (Lepidoptera: Tortricidae) under laboratory conditions

Entomopathogenic nematodes (EPNs) have been successfully applied as biological control agents against above ground and soil stages of insect pests. However, for commercial application, it is crucial to mass culture these nematodes using in vitro liquid culture technology, as it is not attainable when using susceptible insects as hosts. Lobesia vanillana (Lepidoptera: Tortricidae) is regarded a sporadic pest of wine grapes in South Africa. The in vivo- and in vitro-cultured South African EPNs, Steinernema yirgalemense and Steinernema jeffreyense (Rhabditida: Steinernematidae), were evaluated against larvae and pupae of L. vanillana in laboratory bioassays. For larvae, high mortality was observed for all treatments: In vitro-cultured S. yirgalemense (98%) performed better than S. jeffreyense (73%), while within in vivo cultures, there was no difference between nematode species (both 83%). No significant difference was detected between in vivo- and in vitro cultures of the same nematode species. The LD50 of the in vitro-cultured S. yirgalemense, was 7.33 nematodes per larva. Mortality by infection was established by dissecting L. vanillana cadavers and confirming the presence of nematodes, which was > 90% for all treatments. Within in vitro cultures, both S. yirgalemense and S. jeffreyense were able to produce a new cohort of infective juveniles from L. vanillana larvae. Pupae, however, were found to be considerably less susceptible to EPN infection. This is the first study on the use of EPNs to control L. vanillana. The relative success of in vitro-cultured EPN species in laboratory assays, without any loss in pathogenicity, is encouraging for further research and development of this technology.


Introduction
Lobesia vanillana (De Joannis) (Lepidoptera: Tortricidae) is regarded as sporadic pest of wine grapes in the Western Cape province of South Africa, occurs throughout the Afrotropical region [1] and appears to be polyphagous. Apart from limited information relating to taxonomy and locality, very few records exist in literature. It has been reported as a pest of Mangifera sp. (Anacardiaceae) and Vanilla planifolia (Orchidaceae) [2], from which it gets its name [3]. Natural enemies for L. vanillana are unknown. Biological control agents, including Bacillus thuringiensis (Bt) (Bacillalus: Bacillaciae) and mating disruption products, have been registered and used in Europe on Lobesia species [4]. Natural enemies such as Trichogramma parasitoid wasps (Hymenoptera: Trichogrammatidae) have been identified in Europe [5].
Diapausing pupae of Lobesia botrana overwinters under the bark of vines and emerge the following spring [6], and L. vanillana is suspected to follow a similar life cycle. Observations by the author suggests that larvae and pupae of L. vanillana do not have a soil-dependent life stage.
Insect parasitic nematodes, or entomopathogenic nematodes (EPNs), occur in soils across the world and are natural enemies of many insect species. EPNs of the family Steinernematidae (Rhabditida) are associated with the symbiotic bacteria of Xenorhabdus (Enterobacteriales: Enterobacteriaceae) [7]. Infective juveniles (IJs) of these species release their symbiotic bacteria shortly after penetrating the haemocoel of their target insect, causing mortality within 48 h [8], depending on the number of nematodes that penetrated, and the size of the insect host [9]. The virulence of these pathogens is usually correlated with host abundance and the associated microclimate [10]. In agricultural applications, they can be used as inundative control (similar to chemical insecticides), but in favourable conditions, they can successfully establish to provide persistence [9].
Steinernema yirgalemense Nguyen, Tesfamariam, Gozel, Gaugler & Adams and Steinernema jeffreyense Malan, Knoetze & Tiedt have been successfully mass-produced using in vitro liquid culture methods [19,20]. Promising results were obtained against false codling moth, using both in vivo-and in vitro-cultured nematodes, with no significant difference between culture types in laboratory and field trials [18]. The in vitro liquid mass production of EPNs is much more cost and labour effective than in vivo production, the latter of which is better suited for small-scale experiments and insecticidal applications [21].
Insects without a soil stage may be more susceptible to EPNs, as they may not have had the opportunity to evolve the resistance necessary to protect themselves from nematode infections. This weakness of above-ground pest defence mechanisms against microbiological pathogens can thus be exploited to provide biological control, for example, as with previous research on mealybugs, and the addition of adjuvants to nematode suspensions [18,[22][23][24][25][26] has shown.
The aim of this study was to evaluate the pathogenicity of in vivo-and in vitro-cultured S. yirgalemense and S. jeffreyense against the larvae and pupae of L. vanillana. Screening was conducted under optimum laboratory conditions and mortality by infection confirmed. The LD 50 of the most efficient species was determined. Additionally, reproduction of the nematode in the insect cadaver was investigated.

Source of insects
Lobesia vanillana larvae were obtained from a laboratory colony, which was established in February 2018 from field-collected individuals, and artificially reared at a rearing facility in Stellenbosch. They were fed an agar-based modified codling moth diet [27] and kept at a constant 25˚C with a 18:6 h light-dark cycle. Adults were placed next to a window to receive natural indirect sunlight, and were given a cotton ball, dipped in a 2% sugar-water solution, for nourishment. Tenebrio molitor L. (Coleoptera: Tenebrionidae) larvae, used for the in vivo culture of EPNs, were cultured in the laboratory on bran and carrots, in vented containers according to the technique of Van Zyl and Malan [28].

Source of nematodes
In vitro-cultured S. yirgalemense (157-C) (GenBank accession number EU625295) [15,19] and S. jeffreyense (J194) (KC897093) [29] were cultured according to the technique of Dunn et al. [20]. The specific bacteria associated with the two nematode species were obtained from Eppendorf tubes stored in 15% glycerol at -80˚C. A volume of 200 μl was added to 30 ml tryptic soy broth in Erlenmeyer flasks and incubated on an orbital shaker for 48 h. A 5% v/v of bacteria culture was added to a 30 ml complex medium and left to incubate for 48 h on a shaker, after which flasks were inoculated with sterile-stored stock-cultured nematodes at a concentration of 2 000 IJs/ml. After 14 days on a shaker at 25˚C, the life cycle was completed and flasks with IJs in the medium were stored at 14˚C on a shaker in the dark, until needed. To use as inoculum, 1 ml of stored cultures of sterile nematodes were washed free of bacteria, by using a 32 μm sieve with running water and by washing the nematodes trapped on the sieve.
In vivo-cultured nematodes, of the two EPN species, were cultured using T. molitor larvae using standard procedures [30]. Petri dishes with T. molitor larvae were inoculated with IJs of the two nematode species, respectively. After infection, petri dishes were placed on water traps to collect the emerging IJs. These IJs were harvested within the first week of emergence, after which they were stored horizontally in vented tissue culture flasks at 14˚C until use. Concentrations were calculated using the method developed by Glazer and Lewis (2000) [31].

Bioassay protocol
The test arena consisted of 24-well bioassay trays (CellStar TC, Cat No 662160, Greiner Bio-One, Frickenhausen, Germany). Filter paper discs (12.7 mm diam., Grade FN 30, Cat No 3.526.012-7, Ahlstrom-Munksjö, Bärenstein, Germany) were added to each alternate well and alternate row, on which IJs were inoculated at a predetermined concentration [31], by pipetting 50 μl of the nematode suspension onto the filter paper. Each treatment consisted of four 24-well bioassay plates, with six L. vanillana individuals per tray, for a total of 24 larvae per treatment. For larval bioassays, glass rectangles were placed between the lid and the tray to prevent the escape of larvae during the incubation period.
Trays were stacked in closed 2-L plastic ice cream containers, each with moistened paper towels at the bottom, to provide an environment of high humidity, then incubated at 25˚C for 48 h in the dark. Mortality by EPN infection was assessed by gently prodding the insect with a dissection needle and evaluating the integrity and colour of cadavers.
Larvae were carefully rinsed using a handheld water jet to rid the cadaver of surface nematodes. Cadavers were then either placed on a modified White Trap [32] to evaluate in vivo nematode production, or dissected to evaluate nematode infection.

Susceptibility of larvae
Following the described protocol, bioassays consisted of two treatment groups (in vivo and in vitro), inoculated with S. yirgalemense and S. jeffreyense, and one control group treated with water only. Larvae were inoculated with 100 IJs/50 μl for all treatments. Controls received 50 μl distilled water only. The experiment was repeated on a different date, with a different batch of nematodes.

Susceptibility of pupae
Screening for the susceptibility of L. vanillana pupae was according to the described protocol, for the two treatment groups (in vivo and in vitro), inoculated with S. yirgalemense and S. jeffreyense, and one control group treated with water only. Pupae were inoculated with 200 IJs/ 50 μl for all treatments. Controls received 50 μl distilled water only. The experiment was repeated on a different date with a different batch of nematodes.

Dose-response for larvae
Following the results of susceptibility bioassays for L. vanillana larvae, the most effective EPN species was selected for calculating LD 50 estimates. The same inoculation procedure as described in the bioassay protocol was followed. Logarithmic dosages translated to 100, 50, 25, 12.5, 6.25 and 0 IJs/larva, respectively.

Nematode penetration and reproduction
To assess the infectivity of nematodes, half of the infected L. vanillana larvae from the susceptibility bioassay were incubated at 25˚C, at high humidity, and dissected 18-36 h after the initial 48 h incubation period. This allowed nematodes to grow within the cadaver, which improved their visibility for counting. Infected larvae were placed singly on a petri dish, with a droplet of distilled water to suspend the cadaver contents, and dissected with the aid of a Leica MZ75 stereo microscope. The presence and number of nematodes within the cadaver was recorded.
To assess the reproductive ability of nematodes within the infected L. vanillana larvae, the remaining cadavers of the susceptibility bioassay were placed in 90 mm diam. plastic petri dishes with one filter paper circle, (85 mm, Grade 1 Whatman, CAT no. 1001-085, GE Healthcare Life Sciences) moistened with 800 μl of distilled water. Approximately 10-12 cadavers were placed per petri dish. With the top removed, plastic petri dishes were transferred to 150 mm glass petri dishes, the bottom of which contained just enough distilled water to not float the plastic petri dishes, and covered with the glass lid. The resulting IJ suspension was harvested three times during the course of 45 days, and the concentration determined as nematodes per volume, divided by the number of larvae on the trap, to give the average number of nematodes produced per larva. Distilled water was added to the glass-and plastic petri dishes as needed.

Data analysis
Data were analysed in Microsoft Excel 2016 for descriptive statistics and processed in Statistica 13.3 [33] for comparative analysis. Probit analysis and LD 50 estimates were calculated using NCSS [34]. For larval susceptibility bioassays, residuals of the mortality response were considered normally distributed (Shapiro-Wilk's W = 0.967, p = 0.187), permitting the use of a factorial ANOVA and Fisher's LSD post-hoc test to evaluate responses between nematode production types and between nematode species. For pupal susceptibility bioassays, residuals of the mortality response were considered normally distributed, despite having a significant pvalue (Shapiro-Wilk's W = 0.923, p = 0.004), by examining relevant normality graphs. All analyses were evaluated for, and passed, Levene's test for homogeneity of variances (p > 0.05). Results are given as the mean response for all repetitions ± standard error, unless otherwise specified. The unprocessed dataset is available as S1 Dataset.
Low EPN performance in susceptibility bioassays against L. vanillana pupae resulted in an insufficient number of cadavers for dose-response, penetration and reproduction analyses.

Dose-response of larvae
The data fits the Probit model well (Chi-Square = 1.29; DF = 3; Prob. level = 0.73). Lethal dosage estimates for in vitro-cultured S. yirgalemense are indicated in Table 1. The Probit model may be expressed by the linear function P = 3.81 ± 0.466 + 1.375x ± 0.344, where P is Probit-Mortality and x is Log 10 -Dose (Fig 3).

Nematode penetration and reproduction
In all treatments, > 90% of the cadavers had nematodes present (Fig 4). A higher percentage of nematodes penetrated L. vanillana larvae in the in vivo-cultured treatments versus the in vitro counterparts, however this difference was not significant (p = 0.978).
Chi-square analysis revealed no significant differences between IJ penetration within cadavers and nematode culture type or nematode species. Residuals of nematode counts failed the normality assumption (Shapiro-Wilk's W = 0.771, p < 0.01), but relatively large sample sizes per group (n � 22) allowed for an ANOVA bootstrap analysis. There was a significant difference in the average number of nematodes per cadaver within the in vivo culture type, between S. jeffreyense (17.091 ± 0.187) and S. yirgalemense (4.917 ± 0.334) (Bootstrap p = 0.045), while within the in vitro culture type, there was no significant difference between them (Fig 5).

Control of Lobesia vanillana using EPNs
Both in vitro-cultured S. yirgalemense and S. jeffreyense nematodes were able to produce IJs from cadavers following the susceptibility bioassay. Cumulative production after 45 days totalled 2 130 IJs for S. yirgalemense and 2 356 IJs for S. jeffreyense, per insect cadaver (Fig 6).

Discussion
Mortality of L. vanillana larvae infected with EPNs was exceptionally high for all treatments and ranged between 72-98%. Specifically, in the case of in vitro cultures, S. yirgalemense performed significantly better than S. jeffreyense, while using in vivo cultures, there were no significant difference in mortality between species. There was also no significant difference found between in vitro and in vivo cultures of the same species. In the case of the mealybug, Planococcus ficus (Signoret), the dominant mealybug of grapevine in South Africa [35], inoculated with the same concentration of nematodes per insect, mortality rates of 63% for S. yirgalemense were found, compared to 40% for S. jeffreyense [24].
Low mortality was observed for all treatments against pupae. Only in vitro-cultured S. yirgalemense performed slightly better than the control. In general, lepidopteran pupae are less susceptible to infection, as indicated by Malan et al. [15] for false codling moth. However, Steyn et al. [18] suggested that the age of pupae have an effect on susceptibility, with older fully formed, hardened pupae, being less susceptible. Pupae, as the dormant survival stage of lepidopterans, are not normally targeted with the use of EPNs. However, in the case of false codling moth, the soil stages of the insect include the larvae, prepupae and emerging moth, which offers a window of opportunity for infection with EPNs [15]. The pupae of fruit flies in soil were found to be totally resistant to EPN infection [36].
Observations found L. vanillana to be relatively small (adults ± 6 mm in length), compared to other tortricid moths previously evaluated against EPNs, such as codling moth and false codling moth [37]. The LD 50 of in vitro-cultured S. yirgalemense against L. vanillana larvae was estimated as 7 IJs/larva. In a study by De Waal et al. [11], both in vivo-cultured S. yirgalemense and S. jeffreyense were evaluated against the above-ground diapausing codling moth larvae, at half the concentration used for L. vanillana in the present study (50 IJs/larva). Both nematode species resulted in a mortality of close to 100%. However, Odendaal et al. [14] found that in a semi-field spray trial, S. jeffreyense performed better than S. yirgalemense against codling moth larvae, contrary to what was expected. In most studies, especially those with micro insects, such as thrips and Bradysia spp., the large IJ size of S. jeffreyense prevents penetration [38].
Mortality, caused by nematode infection, was confirmed by dissecting L. vanillana larvae and evaluating the presence and number of nematodes. Within in vivo cultures, there was a significant difference between S. jeffreyense (17 nematodes per cadaver) and S. yirgalemense (five nematodes per cadaver), but no significant difference between the species of in vitro cultures. It was expected that the number of IJ penetrated would be higher in the case of S. yirgalemense, as it is a smaller IJ (± 635 μm) [39] when compared to the body length of S. jeffreyense (> 900 μm) [29].
Visual observation of larval cadavers directly after susceptibility bioassays, revealed that few of the final instar insect larvae treated with S. yirgalemense managed to pupate in their trays, with little webbing present, compared to those treated with S. jeffreyense, suggesting that S. yirgalemense is faster-acting than S. jeffreyense, in laboratory bioassays at least, but more research is needed to support this theory.
The in vitro cultures of both S. yirgalemense and S. jeffreyense had the ability to produce a new cohort of IJs, and after 45 days produced an estimated 2 130 IJ and 2 356 IJs per cadaver, respectively. Generally, the larval stages of lepidopterans were found to support nematode infection and reproduction [37], and more so when insects are believed to not have a soil stage. In rare cases, especially where insects have soil stages, a type of resistance against nematodes can develop, such as the case with woolly apple aphid, Eriosoma lanigerum (Hausmann) (Hemiptera: Aphididae) [40].
Local research established in vitro liquid culture methods for H. zealandica, S. jeffreyense and S. yirgalemense [19,20,41] while research on the formulation, packaging and storage of these species is still ongoing [42][43][44]. In vivo-cultured IJs can provide affordable, high-quality nematodes that are easy to culture, but only on a small scale [21]. Increased complexity, risk, labour and running costs are prohibitive when scaling towards mass-production [45]. The start-up capital and complexity of in vitro production methods are excessive for small-scale use, but for mass-production and augmentative releases where a large number of nematodes is required, it is the most cost-effective solution [21]. Results from this study indicate that the quality of in vitro-cultured nematodes are comparable to those cultured in vivo, the latter of which is considered the "more natural" method. Previous studies by Ferreira et al. [19,41] evaluated the efficacy of in vitro-and in vivo-cultured S. yirgalemense and H. zealandica against the greater wax moth, Galleria mellonella (L.) (Lepidoptera: Pyralidae), and found in vivo-cultured nematodes to cause significantly higher mortality than their in vitro-cultured equivalents.
Steyn et al. [18] evaluated the same nematode species, both in vivo and in vitro, cultured under the same conditions as the present study, and assessed their mortality against false codling moth, both in the laboratory and in the field. Using half the concentration of IJs used in the current study, high mortality of false codling moth was found in the laboratory, compared to semi-field applications which resulted in a mortality of � 70%. Similar to the present study, no difference was found between in vivo-and in vitro-cultured nematodes of these two species [18]. The isolate of S. yirgalemense demonstrated high efficacy against other key insect pests of various fruit crops in South Africa, including banded fruit weevil [46], fruit fly [36] and mealybug [24], and has been prioritised for commercialisation [43,44].
More nematode species, especially H. zealandica and other native species that show effective control against lepidopteran pests, can be evaluated in future research to establish a nematode susceptibly profile for L. vanillana. In addition, the application of nematode formulations to the canopy and soil of orchards may have the ability to control multiple pests simultaneously, which include in the case of table grapes: mealybugs, different weevil species, other lepidopteran insects, fruit fly and thrips [22,25,26,[47][48][49], especially when used in an integrated pest management programme. This is the first study on the use of EPNs to control L. vanillana by comparing in vivo-and in vitro-produced nematodes, without any loss of pathogenicity during the culture process, which is highly promising for the future commercial production of these biocontrol agents.
Results also indicate that in vitro-and in vivo-cultured S. yirgalemense and S. jeffreyense nematodes are able to infect and kill the larvae of L. vanillana, and that these in vitro cultures are able reproduce within this host and produce a new cohort of IJs, capable of finding and infecting new hosts. The relative success of S. yirgalemense, S. jeffreyense and other local EPN species against South African tortricid species in laboratory and field assays, the ability to produce nematodes using in vitro liquid culture techniques, and the industry demand for such products, encourages further research and development of this technology.
Supporting information S1 Dataset. Workbook containing the unprocessed results of this study. This workbook contains four datasheets: 1_Screening: Unprocessed results of the 'susceptibility of larvae' and 'susceptibility of pupae' screening assays; 2_DoseResp: Unprocessed results of the 'dose response of larvae' bioassay; 3_NemPresCadav: Unprocessed results of the 'nematode penetration' analysis; 4_NemProdWT: Unprocessed results of the 'nematode reproduction' analysis. A legend to acronyms and shortened words used in the dataset is provided in the worksheet 0_Legend at the start of the workbook. (XLSX)