Acylsugar tomato lines suppress whiteflies and Amblyseius swirskii establishment

Plant defense traits such as trichomes along with biocontrol agents may provide alternatives to insecticide use in tomatoes, Solanum lycopersicum L. (Solanaceae). However, plant‐herbivore‐natural enemy interactions are not always complementary. In a series of greenhouse and field experiments, we explored whether augmented defense traits (i.e., production of acylsugars) in tomato plants could reduce sweet potato whitefly, Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae), populations and aid the establishment of the predatory mite Amblyseius swirskii Athias‐Henriot (Acari: Phytoseiidae). In the field experiment, commercial tomato cultivars and acylsugar‐producing tomato lines received no predatory mites or mites released via three methods: dusting on top, dusting on bottom, or slow‐release sachets. In the first greenhouse experiment, predatory mites were released onto the commercial and acylsugar‐producing tomato plants via sachets. In the second greenhouse experiment using a similar design, we augmented the mite diet with an alternative non‐prey resource (i.e., cattail pollen). Our results indicated that acylsugar‐producing tomato plants supported significantly fewer whiteflies than the commercial lines in all experiments. However, in the field, despite lower whitefly numbers, Tomato yellow leaf curl virus (TYLCV) (Geminiviridae, Begomovirus) was detected at higher frequencies in acylsugar‐producing lines. Few mites were recovered from all commercial and acylsugar‐producing lines in the field or greenhouse experiments suggesting A. swirskii does not establish well on tomatoes, and acylsugar lines successfully decreased whitefly populations but not a viral disease transmitted even at low whitefly abundance.

, including the development of whitefly-or disease-resistant tomato varieties. Currently, multiple commercial tomato cultivars are resistant to whitefly-transmitted viruses; however, these varieties are not immune to whitefly infestation (Srinivasan et al., 2012).
Plant defense traits, such as acylsugars, are non-selective and may reduce pest populations in addition to reducing the efficacy of natural enemies (Farrar & Kennedy, 1991;Coll & Ridgway, 1995;Schmidt, 2014). Therefore, breeding for traits that deter pests and coincidentally natural enemies may present a trade-off between host plant resistance and biological control (van Houten et al., 2013). Multiple biocontrol agents are documented to provide whitefly suppression in greenhouse and field conditions (Kheirodin et al., 2020;Sani et al., 2020). The phytoseiid mite, Amblyseius swirskii Athias-Henriot (Acari: Phytoseiidae), is an effective generalist predator and is considered an excellent biocontrol agent of whiteflies in numerous vegetable systems (Calvo et al., 2010(Calvo et al., , 2012(Calvo et al., , 2015Tellez et al., 2017). However, the predatory mites may have difficulty establishing on tomato plants (Sakamoto et al., 2012;van Houten et al., 2013;Paspati et al., 2021), especially in varieties with heightened expression of defense traits (Paspati et al., 2021). In this study, our objective was to evaluate whether the tomato plants with enhanced acylsugar production interact with whiteflies to suppress their abundance. We hypothesized that enhanced defense traits in acylsugarproducing tomato plants will lead to significantly lower whitefly abundance compared to commercial cultivars. Our other objective was to understand whether the defense traits expressed in acylsugar tomato plants play a negative role in A. swirskii establishment. To test our hypotheses that the establishment of predatory mites might be hindered by the defense traits, we evaluated the efficacy of combining A. swirskii with tomato plants that had either TYLCV resistance, or experimental lines with enhanced acylsugar production in a series of greenhouse and field experiments.

Biological material
Seeds of the experimental acylsugar-producing tomato lines (henceforth called acylsugar lines) were provided by the Tomato Breeding Program at Cornell University (Ithaca, NY, USA). Acylsugar lines CU0701026 and FA7/AS, and the commercial cultivars Skyway 687 and SV7631TD were evaluated in the field experiments. For the greenhouse experiments, we used the acylsugar lines FA7/ AS, CU071026, and QTL6/AS, and the commercial cultivars Florida 47 (purchased from Tomato Growers Supply Company, Fort Myers, FL, USA) and Amelia F1 Hybrid (purchased from Harris Seeds, Rochester, NY, USA). The commercial cultivars for the field experiments Skyway 687 (purchased from Johnny's Selected Seeds, Winslow, ME, USA) and SV7631TD (purchased from Seedway, Hall, NY, USA) represented TYLCV resistant and susceptible tomato cultivars, respectively. The acylsugar lines range from moderate to high acylsugar content and trichome density. CU071026 is a benchmark acylsugar line obtained by transferring increased levels of acylsugars from Solanum pennellii Correll accession LA716 into cultivated tomatoes reported to produce 15% of acylsugar levels produced by S. pennellii LA716 lines. FA7/AS was created by introgression of quantitative trait loci FA7 which contains acylsugar fatty acid profiles into CU071026. Introgression of QTL6 within CU071026 is associated with increased acylsugar level and trichome IV density (Leckie et al., 2012(Leckie et al., , 2014. We produced tomato seedlings in a greenhouse using nursery trays filled with organic in-ground soil blended with organic compost at a temperature of 29 ± 2 °C. In the field, drip irrigation was provided as needed. Weeds between beds were suppressed 3× during the sampling period by mowing. In addition to mowing, an organic herbicide, Biosafe Weed and Grass Killer (Arbico Organics, Oro Valley, AZ, USA) at the rate of 0.1 L/1 L of water was sprayed between beds 2× during the sampling period.
We obtained the adult whiteflies for greenhouse experiments from a colony maintained at the University of Georgia (Tifton, GA). The colony was maintained in 4-5 leaf stage cotton seedlings in a rearing room at 27 ± 2 °C and L14:D10 photocycle.
Two A. swirskii products were used: Swirski-Mite (50000 predatory mite nymphs and adults per L in loose bran) and Swirski-Mite Plus (250 mites in each branfilled sachet), both purchased from Koppert Biological Systems (Howell, MI, USA). We assessed the viability of the predatory mite products immediately after receiving. We placed three mite sachets in a plastic cup over a container filled with water and observed the water under microscope for the presence of A. swirskii after 2 days. For Swirski-Mite product, approximately 0.5 g bran was sprinkled in a Petri dish and A. swirskii was observed under a microscope.

Field experiment
We conducted field experiments in a certified organic field located at the University of Georgia Tifton Campus (Tifton, GA). Tomato seedlings grown in the greenhouse were transplanted into raised beds at 4-5 leaf stages. The field was divided into four blocks each containing 16 beds. Each bed was 4.6 m long, 0.91 m wide, and 0.2 m high, 0.31 m away from the adjacent beds, and covered with black plastic mulch. Ten plants of each commercial line (Skyway 687 and SV7631TD) and five plants of each acylsugar line (CU071026 and FA7/ AS) were transplanted in each bed with a plant-plant spacing of 0.46 m. A full factorial treatment combination of mite release methods (no mite, basal application, top application, and sachet application) and tomato lines (CU071026, FA7/AS, Skyway 687, and SV7631TD) were assigned to beds following a randomized complete block design. We introduced predatory mites 1 day after tomato plants were transplanted in the field. The four mite treatments included (1) not applying any mites (no mite), (2) sprinkling a dosage of Swirski-Mite at the base of the tomato plants (basal application), (3) sprinkling a dosage of Swirski-Mite to the canopy of the plants (top application), or (4) attaching two Swirski-Mite Plus sachets to each bed of tomato plants (sachet). When the plants reached 8-10 leaf stages, the mite sachets were moved to be hung in the tomato canopies.
We sampled predatory mites and whitefly abundance every 2 weeks for a period of four sampling weeks after the mite release. We randomly selected a leaflet from the middle section of each of the six tomato plants in each bed. Adult whiteflies on the abaxial surface of the leaflet were counted and recorded in the field (field scouting) following Diehl et al. (1995) before taking leaflet samples. The leaflets were collected, placed in individual plastic bags, transported to the laboratory, and stored overnight at −20 °C. Whitefly eggs and nymphs, A. swirskii eggs and adults, and any other herbivores on each leaflet were counted. Tomato yellow leaf curl virus presence was assessed based on visual symptoms exhibited as slight to pronounced yellowing of leaflet margins, curling and cupping of the leaflet, with a reduction in size and stunted plants (Friedmann et al., 1998). The number of plants in a bed exhibiting TYLCV symptoms was recorded.

Greenhouse experiments
We conducted two greenhouse experiments. For both experiments, seedlings in nursery trays and plants in pots (grown from the seedlings) were maintained in bugdorms (60 × 60 × 60 cm with mesh size 150 × 150, or 160 μm aperture; MegaView Science, Taichung City, Taiwan) to exclude all insects prior to the experiments. The seedlings were potted in 12.7-cm-diameter plastic pots with in Miracle-Gro ground soil mix (Lowe's, Tifton, GA, USA). General purpose fertilizer (20:20:20 N:P:K; Peters Professional Brand, Allentown, PA, USA) was applied to the tomato plants immediately after potting and a week after the first fertilizer application.

Greenhouse experiment 1: Role of tomato lines and Amblyseius swirskii application in whitefly population suppression
A setup of a bugdorm with five whitefly-infested tomato plants of each of the five lines (Florida 47, Amelia, CU071026, FA7/AS, and QTL6/AS) was prepared for the experiment. The setup was replicated 6×. To prepare a bugdorm setup, we firstly infested the tomato seedlings with whiteflies at the 5-leaf stage. Cotton seedlings at (4-5 leaf stage) with approximately 15-20 whitefly adults each were placed inside the bugdorms with uninfested tomato seedlings. We allowed the whiteflies to establish on the uninfested tomato plants for 3 weeks, which is the amount of time whiteflies take to complete their lifecycle under favorable conditions (White, 2014). Once we observed whitefly establishment on the tomato seedlings, we divided the tomato plants into six bugdorm setups. The whitefly-infested tomato plants inside the bugdorms were undisturbed for 3 weeks allowing time for the new generation of whiteflies to develop. During these 3 weeks, we took leaflet samples from each plant line in the bugdorms weekly. A commercial mite sachet (Swirski-Mite plus, Koppert Biological Systems) was placed on each plant 3 weeks after the infested tomato plants were divided into bugdorm setups. We continued to take leaflet samples weekly for an additional 4 weeks after mite sachets were applied to the tomato plants. In total, we had 7 weeks of leaflet samples collected which included 3 weeks of leaflet sample collection before mite application and 4 weeks of leaflet sample collection after mite application. Samples were collected by taking a leaflet from the middle portion of each tomato plant weekly and placing them in a plastic bag. The leaflet samples were then transported to the laboratory for the count of whiteflies and A. swirskii at all life stages. Additionally, we removed all inflorescence from tomato plants inside the bugdorms to reduce the possibility of A. swirskii feeding on tomato pollen (Calvo et al., 2015).

Greenhouse experiment 2: Non-pest food effects on Amblyseius swirskii establishment and whitefly suppression
The second greenhouse experiment included three treatments: B. tabaci only, B. tabaci + A. swirskii, and B. tabaci + A. swirskii + cattail pollen, replicated over 12 tomato plants each. We used the commercial tomato cultivar Skyway 687 in this experiment at the 4-5 leaf stages. Amblyseius swirskii (Swirskii-System) and cattail pollen (Nutrimite) for this experiment were purchased from Biobest Sustainable Crop Management (Westerlo, Belgium). Like other experiments, we assessed the quality of the mites immediately after receipt. The mixture contained an average of 120 mites per 0.5 g of treatment calculated by averaging the number of mites in 0.5 g of products 5×. Aside from herbivores, A. swirskii can also feed on pollen as a supplementary food source (Calvo et al., 2015) and cattail pollen is observed as an alternative food source for predatory mites (Delisle et al., 2015a).
Twelve tomato plants were assigned to each treatment. The tomato plants were divided into three bugdorms with four plants in each bugdorm. Twenty-five whitefly adults were released in each bugdorm and were allowed to establish on tomato plants. We observed whitefly establishment on the tomato plants 2 weeks after introduction. Following whitefly establishment, we applied mites and cattail pollen to the tomato plants by sprinkling 0.5 g of products on the top of tomato plants. Both mites and cattail pollen were applied to the plants 2× at an interval of 2 weeks during the 4-week sampling period. We collected samples a week after the mites and cattail pollen were applied to the plants. One plant from each of the three bugdorms was sampled. Five randomly chosen leaflets per plant per bugdorm were selected and carefully observed under a stereomicroscope for the count of whitefly and A. swirskii at all life stages.

Statistical analysis
In the field-collected data, the whitefly abundance data were not normally distributed. Both whitefly abundance and TYLCV incidence were analyzed by the Kruskal-Wallis test. We conducted Bonferroni-adjusted Dunn posthoc tests for mean comparisons of all the whitefly stages (eggs, nymphs, adults) in relation to tomato lines. We excluded the mite application treatments from data analyses due to the poor recovery of predatory mites in our field experiment. In greenhouse experiments 1 and 2, whitefly counts and mite counts data for all life stages were logtransformed to fit normality and variance assumptions of ANOVA models, which was confirmed by the Shapiro-Wilk test. We then conducted ANOVA to test for the main effect of acylsugar lines followed by Tukey's honestly significant difference (HSD) test for comparisons of whitefly and mite counts between tomato lines. Whitefly egg and nymph counts before mite treatment and after the mite treatment in greenhouse experiment 1 were compared using t-test. JMP Pro v.15.0.0 (2019, SAS Institute, Cary, NC, USA) was used for all statistical analyses. All differences are determined to be significant at α = 0.05.

Field experiment -Whitefly population, mite population, and virus prevalence
We released approximately 8800 mites in the entire field including mite application by sprinkling the product on the top and bottom of the plants and mite sachet application to the plants. We took leaflet samples from the field at an interval of 2 weeks for 4 weeks. On each sampling date, we sampled approximately 352 leaflets from the entire field. However, we recovered only nine mites from all our sampling efforts, which suggests that mites did not successfully establish on the tomato plants.
The highest densities of whiteflies were observed during the first 4 weeks of the field experiment ( Figure 1). The number of whiteflies declined by the end of the sampling period (Figure 1). There were strong effects of tomato lines on whitefly eggs (χ 2 = 443.835), nymphs (χ 2 = 606.697), and adults (χ 2 = 627.941, all d.f. = 3, P < 0.001) abundance ( Figure 2A). Both commercial lines (Skyway and SV7631TD) accumulated higher abundance of all stages of whiteflies compared to acylsugar lines (CU071026 and FA7/AS). Whitefly abundance was not significantly different in both acylsugar lines (Figure 1). In contrast, mean comparisons for TYLCV disease incidence indicated lower virus incidence F I G U R E 1 Mean (± SE) number of whitefly (WF) eggs, nymphs, and adults observed per leaflet on all lines in open fieldgrown tomato plants. The whitefly adult counts represent field scouting counts across the sampling dates for all tomato lines. Field scouting was conducted by counting the whiteflies on the abaxial surface of tomato leaflets in the field. Adult whitefly counts were taken by selecting a leaflet from the middle section of each of the six tomato plants in each bed. on the commercial lines compared to the acylsugar lines (χ 2 = 47.143, d.f. =3, P < 0.001) ( Figure 2B).

DISCUSSION
Our study shows that the experimental acylsugar tomato lines with enhanced defense traits had lower whitefly numbers compared to commercial cultivars. We observed similar results in both our field and greenhouse experiments. The tomato lines used in our study (CU071026, FA7/AS, and QTL6/ AS) are bred from a wild relative S. pennellii LA716 for high acylsugar production. Although the level of acylsugar production in the experimental lines is not as high as the wild relative (Leckie et al., 2012;Smeda et al., 2016), our results are showing a consistent trend of significantly reduced whitefly oviposition and development (Leckie et al., 2012;Smeda et al., 2016Smeda et al., , 2018. This is likely a result of reduced insect settling and phloem feeding in piercing/sucking pests Li et al., 2019;Blanco-Sánchez et al., 2021).
The results suggest that the effects of acylsugar lines may be context dependent and perform better in some environmental conditions than in others. The abundance of whitefly eggs, nymphs, and adults was not significantly different among both acylsugar lines (CU071026 and FA7/AS) in our field experiment. In our greenhouse experiment, however, we observed the lowest whitefly abundance on the tomato line FA7/AS followed by CU071026 and QTL6/AS. The presence of acylsugars on tomato leaves is associated with the deterrence of piercing and sucking insects (Goffreda et al., 1989;Rodriguez et al., 2011;Li et al., 2019). The experimental acylsugar lines express enhanced densities of type IV glandular trichomes which exude the acylsugars compared to a trace amount of acylsugar found in S. lycopersicum (Lawson et al., 1997;Leckie et al., 2012;Smeda et al., 2016). Differences in acylsugar production potentially help explain differences in the abundance between some of the acylsugar lines. The responses of whiteflies to the acylsugar lines appear context-dependent (i.e., field or greenhouse led to different outcomes). Although we did not measure the acylsugar levels in our study, it is known that the accumulation of secondary metabolites such as acylsugar is highly dependent on environmental factors (Shapiro et al., 1994). Leckie et al. (2012) report higher acylsugar levels in greenhousegrown (compared to field-grown) CU071026. Similar results are reported for the wild tomato S. pennelii in which higher acylsugar content was observed from greenhouse-grown vs. field-grown wild tomatoes (Shapiro et al., 1994). Smeda et al. (2016) report a higher accumulation of acylsugars in line FA7/AS than in CU071026, and combined results suggest further work is needed to optimize and clarify the differences in the production of acylsugars in relation to variable environmental conditions.
In our field experiments, although whitefly abundance was lower on acylsugar lines, we observed high TYLCV incidence when compared to commercial cultivars. The lowest incidence of the virus was observed in TYLCV-resistant and susceptible cultivars. Bemisia tabaci feeding is detrimental to tomato plants as they are capable of transmitting viruses such as TYLCV that cause serious damage to the crop (Marchant et al., 2020). The experimental tomato lines used in this study have moderate to high acylsugar content and high trichome density; however, currently, none of the lines possess virus-resistance genes (Smeda et al., 2016(Smeda et al., , 2017Ben-Mahmoud et al., 2018). Given the whitefly pressure in the field -a leaf is considered infested if it has >5 immature whiteflies (Barman et al., 2020) -, acylsugar lines are not completely immune to whitefly feeding and resultant virus acquisition and transmission. This is also true for the TYLCV-resistant cultivars available commercially (Srinivasan et al., 2012). Lower whitefly populations on acylsugar lines can help lower TYLCV infections overall, but once the plants are infected with the virus both acylsugar and non-acylsugar lines acquire similar viral loads capable of transmission (Marchant et al., 2020). Rodriguez et al. (2011) conducted a study to understand the level of viral loads and transmission in acylsugar lines compared to the commercial cultivar Moneymaker. The study demonstrated that, although virus symptoms and load were similar in acylsugar and nonacylsugar tomato lines, the virus spread from acylsugar lines was lower, which indicates that acylsugar effects go beyond influencing within-plant populations of whiteflies.
Acylsugar production has the potential to alter the compatibility of biocontrol agents, which could alter biological control efficacy. We tested the effectiveness of pairing A. swirskii with the acylsugar lines of tomatoes for the management of whitefly in field and greenhouse experiments. In the field and greenhouse, even though we repeatedly released mites or used time-release sachets, we recovered few A. swirskii on leaf samples. We T A B L E 1 Mean (± SE) number of Amblyseius swirskii eggs and adults and whitefly (Bemisia tabaci) eggs, nymphs, and adults per five leaflets per treatment. attribute this to the possibility that glandular trichomes are non-selective plant defenses and hinder the growth and foraging of A. swirskii (Paspati et al., 2021), a result that is consistent with the recent review of natural enemies (Peter et al., 1995;Simmons & Gurr, 2005;Tian et al., 2012;Riddick & Simmons, 2014). Currently, plant breeding efforts to accommodate natural enemies and predatory insects are minimal (Bergman & Tingey, 1979;Cortesero et al., 2000;Agrawal, 2000). For example, trichomes negatively impact the oviposition rate (Koller et al., 2007) and searching efficiency of predatory mite Neoseiulus californicus (McGregor) for the control of twospotted spider mites (Cédola et al., 2001). Studies also demonstrate reduced movement, oviposition, and predation by phytoseiid mite such as Phytoseiulus macropilis (Banks), Phytoseiulus longipes Evans, and Amblydromalus limonicus (Garman & McGregor) on tomato leaves (Sato et al., 2011;Davidson et al., 2016). Two other studies show specifically that A. swirskii survival was significantly impacted by trichomes and secondary metabolite (acylsugar) accumulating in the bodies (Buitenhuis et al., 2014;Paspati et al., 2021). Hence, the presence of trichomes and their exudates appears to have negative effects on A. swirskii in tomato, a top-performing predator of whiteflies in other crops (Nomikou et al., 2001;Calvo et al., 2015;Sakamoto et al., 2012). Amblyseius swirskii appears to avoid or have difficulty establishing on all the tomato lines we tested. In an attempt to stimulate biological control by A. swirskii, we provided a non-prey food source, cattail pollen. We hoped that the application of cattail pollen would provide an additional food source to help establish the predatory mite because A. swirskii can feed on pollen as an alternative food source (Ragusa & Swirski, 1975;Calvo et al., 2015). For example, Nomikou et al. (2002) observed aggregation of mites on leaves where pollen was applied, which improved the survival of A. swirskii in the absence of whiteflies and only pollen. Cattail pollen and apple pollen were observed as good food sources of A. swirskii (Delisle et al., 2015a;Vangansbeke et al., 2016). Delisle et al. (2015b) also reported improved control of thrips when A. swirskii were supplemented with apple pollen. In our greenhouse experiment, we observed slightly higher numbers of A. swirskii adults on tomato plants sprinkled with cattail pollen. We observed that the whitefly egg abundance was lower on the plants treated with cattail pollen. However, we observed no significant effect on whitefly nymphs and adult numbers. Our results indicate that although the addition of non-prey resources can help establish A. swirskii on tomato plants, mite establishment was still poor and, in this instance, non-prey food did not improve whitefly control.

Conclusions
The current study highlights the success of higher acylsugar-producing tomato lines for whitefly resistance.
We also conclude that the predatory mite A. swirskii is not compatible with tomato as a host plant. We only used mites as biocontrol agents for whiteflies, but there are various biocontrol agents available commercially. Future studies should test other arthropod compatibility with acylsugar lines to potentially synergize the pest control benefits of plant defenses and biological control.

AC K N O W L E D G M E N T S
We thank our experimental farm manager Andy Carter for helping with setting up plots, in addition to Michael Foster and Melissa Thompson for field and laboratory technical support. We thank two anonymous reviewers for their comments and suggestions on an earlier draft that greatly improved the manuscript. This research work was supported by the Sustainable Agriculture Research and Education (SARE) program (project LS19-305).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.