Impact of a shared sugar food source on biological control of Tuta absoluta by the parasitoid Necremnus tutae

Honeydew is a sugar-rich food source produced by sap-feeding insects, notably by major pests such as aphids and whiteflies. It is an important alternative food source for the adult stage of various key natural enemies (e.g., parasitoids), but it may be used also as food by agricultural pests. Necremnus tutae is an idiobiont parasitoid, and it is the most abundant larval parasitoid associated with the South American tomato pinworm, Tuta absoluta, in recently invaded European areas. The impact of N. tutae on T. absoluta populations was evaluated under greenhouse conditions with and without the presence of a honeydew producer, the aphid Macrosiphum euphorbiae. In addition, laboratory experiments were performed to evaluate the longevity of N. tutae and T. absoluta adults when fed with water, honey or honeydew produced by the aphid. In the greenhouse, N. tutae effectively reduced T. absoluta population by the end of the experiment, and this independently of the presence of the aphid; still the presence of M. euphorbiae led to delayed and reduced T. absoluta population peak when controlled by the parasitoid (there was a fourfold increase in parasitoid density in presence of aphid). The longevity of both N. tutae and T. absoluta females increased in the presence of honeydew (when compared to water only) under laboratory conditions; it hinted that honeydew could be used by an herbivore as T. absoluta. The interactions between the two phytophagous species showed contrasting effects, and this mostly independently of parasitoid presence. Tuta absoluta had no impact on aphid population dynamics. By contrast, the production of T. absoluta larvae was higher in the first part of the experiment as the adults lived longer (feeding on aphid honeydew) and produced more larvae. In the second part of the experiment, there were fewer T. absoluta larvae in the presence of the aphid, likely owing to resource competition and/or negative plant-mediated indirect interactions. We demonstrated that honeydew producers could impact parasitoid-host population dynamics both through direct and indirect interactions and that potential positive effects of a sugar food source toward a pest can be nullified in the long term by other negative effects occurring simultaneously.


Introduction
The South American tomato pinworm, Tuta absoluta Meyrick (Lepidoptera: Gelechiidae), is considered one of the main agricultural threats to world tomato production (Desneux et al. 2010Biondi et al. 2018). Tuta absoluta was first observed outside South America in Eastern Europe in 2006 and spread rapidly throughout Africa, the Middle East and parts of Asia (Campos et al. 2017;Sankarganesh et al. 2017;Han et al. 2018Han et al. , 2019Mansour et al. 2018;McNitt et al. 2019), as well as in central America in a lesser extent (Biondi et al. 2018;Verheggen and Fontus 2019). Chemical control is the method used most to prevent the outbreaks of T. absoluta in newly invaded areas due to the lack of effective alternatives (Biondi et al. 2018). However, insecticide efficiency may be compromised due to the cryptic nature behavior of the larvae that mine leaves, and occasionally stems, flowers and tomatoes (Guedes and Picanço 2012). Moreover, wide use of insecticides can cause extensive side effects, such as the development of resistance to insecticides in T. absoluta populations (Campos et al. 2015;Roditakis et al. 2018; and see Guedes et al. 2019) as well as a detrimental impact on non-target organisms such as natural enemies (Lu et al. 2012;Biondi et al. 2015;Jam and Saber 2018;Mohammed et al. 2018;Passos et al. 2018;Soares et al. 2019a, b) through multiple potential sublethal effects (Desneux et al. 2007). Indeed, an alternative to conventional pest control is to increase the activity of naturally occurring enemies within integrated pest management (IPM) programs (Ragsdale et al. 2011;Biondi et al. 2018;Perovic et al. 2018;Jactel et al. 2019).
Several predators and parasitoids spontaneously attacking T. absoluta in tomato crops in Europe have been identified (Castañé et al. 2011;Mollá et al. 2011;Zappalà et al. 2013). However, in tomato systems, T. absoluta coexist with a large number of pests including aphids, whiteflies and fruit worms (Kennedy 2003), that may indirectly interact with potential biological control agents in complex food webs. Indirect interaction is an important force driving insect communities, species density and diversity, including direct and indirect effects inside and outside the agroecosystem (van Veen et al. 2005). Indirect interactions may occur (i) between two prey or host species sharing the same natural enemy through predation or parasitization, and (ii) between two species separated in time or space and interacting in different ways with natural enemies (Chailleux et al. 2014b). Among the other pests present in the tomato systems, aphids such as Macrosiphum euphorbiae Thomas (Hemiptera: Aphididae: Macrosiphini) have been described as pests able to produce honeydew (Hullé et al. 2019). Honeydew is a carbohydrate-rich source that may be used as a food source by adults of natural enemies and phytophagous pests to enhance their fitness (Wäckers et al. 2008). Hence, the availability of suitable food sources (e.g., intrinsic plant quality in terms of nutritional status, and other alternative food sources) could affect the outcome of pest-natural enemy interactions and may be critical for the efficiency of biological pest control (Bompard et al. 2013, Chailleux et al. 2014bJaworski et al. 2015;Biondi et al. 2016;Desneux et al. 2019). The study of the multitrophic interactions among natural enemies and invasive pests is required to design effective biological control within the framework of the tomato management program.
Augmentative biological control relies on periodic releases of natural enemies that mainly attack eggs and other small arthropod instars (Desneux et al. 2010;Chailleux et al. 2013;Zappalà et al. 2013). However, other biological control agents that impact different pest stages, mainly in T. absoluta, as well as in different phenological stage and architecture of tomato plants have to be implemented (Perdikis et al. 2014;Mirhosseini et al. 2019). The eulophid parasitoid Necremnus tutae Ribes & Bernardo (Hymenoptera: Eulophidae) has been observed and is widespread in Europe and Africa parasitizing T. absoluta (Gebiola et al. 2015). It has been identified as a promising biological control agent on T. absoluta (Biondi et al. 2013a;Chailleux et al. 2014a;Naselli et al. 2017;Bodino et al. 2019). Previous laboratory trials showed that N. tutae reproductive and longevity outcomes were enhanced by the simultaneous presence of the sap feeder Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae), a honeydew producer (Dong et al. 2018). However, more field studies are needed to test the ability to control pest population.
In this context, we assessed the effect of the aphid presence on both T. absoluta and N. tutae population dynamics in a greenhouse experiment (i) to confirm the potential role of N. tutae as biological control agent of T. absoluta; and (ii) to test the impact of the presence of the honeydew, as a sugar food source, on parasitoid longevity and its ability to control the pest population. In addition, we tested the impact of honeydew as a food source for these two insects on the parasitoid longevity in a laboratory experiment.

Biological material
Tomato plants cv. Nano were grown in a climatic chamber until the seedlings emerged (25 ± 3 °C, 65 ± 5% relative humidity [RH] and a photoperiod of 16:08 [L:D] h) in cubic plastic pots (7 × 7 × 6.5 cm). After emergence, the tomato seedlings were placed under greenhouse conditions (20 ± 2 °C, 65 ± 5% RH and a photoperiod of 10:14 [L:D] h). At 6 days after seeding (DAS), the tomato seedlings were washed to remove soil and then transferred to new pots containing inert substrate (Perlite Italiana srl, Corsico, Italy) (see schedule Table 1). At DAS 24, the plants were transferred to larger pots (diameter: 10 cm, height: 9 cm) filled with the same substrate and placed in four separate compartments in the greenhouse. All the plants were watered using nutrient solution, and the latter was adapted for all the treatments to pH 6.0 ± 0.2 by using H 2 SO 4 (0.2 M) according to the compartment.
The T. absoluta colony was maintained in cages (55 × 75 × 80 cm), in regular chambers using pesticide free potted tomato plants (Biondi et al. 2013b). A solution containing 10% honey was provided ad libitum to adults in rearing cages. The N. tutae parasitoids originated from field-collected material and were reared in cages (45 × 45 × 55 cm) on T. absoluta kept on tomato plants in a climatic chamber. The M. euphorbiae aphid colony was maintained in a cage (90 × 50 × 60 cm) placed in the greenhouse using tomato plants as host (22 ± 5 °C, 60 ± 10% [RH] and during a photoperiod of 15:09 [L:D] h).

Greenhouse experiment
The experiment was performed at the French National Institute for Agricultural Research (INRA), Sophia Antipolis, France, in 2016 and 2017. The performance of N. tutae was checked with and without the honeydew produced by M. euphorbiae in semi-field. The experiment was carried out in the greenhouse in four separated glass compartments. Each compartment had eight rows, and every row was composed of eight tomato plants (Fig. 1). Four tunnels were created for the treatments and consisted in 16 tomato plants that were covered with slim nylon mesh (5 × 1 × 2 m). The following four treatments were set up: (i) T. absoluta (negative control), (ii) T. absoluta + M. euphorbiae, (iii) T. absoluta + N. tutae (iv) T. absoluta + M. euphorbiae + N. tutae (Fig. 1). The treatments described above were replicated four times in each greenhouse compartment. A general overview of the experiment design is shown in Table 1. The first release was performed with 20 couples of T. absoluta in 30 DAS in all treatments (i-iv). 34 DAS, five M. euphorbiae were placed on top of each tomato plant in treatments (iii) and (iv). 44

Laboratory experiment
Honeydew collection. The honeydew was collected using parafilm (plastic packaging, Chicago, IL) strips in the M. euphorbiae colony reared on tomato plant. The parafilm was exposed for 24 h under the tomato leaves with high M. euphorbiae density. Once the parafilm strips were removed, the abundance of honeydew was checked using a stereoscope microscope (20x), and thereafter placed inside Petri dishes for storage at − 12 °C until needed. Necremnus tutae and T. absoluta survival. The experiment was performed to verify N. tutae female parasitoid survival. Three treatments were designed: (i) N. tutae + water (control), (ii) N. tutae + honey and (iii) N. tutae + honeydew. The same procedure was carried out with T. absoluta females. The parafilm was cut into strips (1 cm × 5 cm) containing honeydew, honey or water. For each treatment, a parafilm strip was inserted into a glass tube (3 cm diameter × 10 cm height) and the latter was closed using moistened cotton. Then, a 24-h-old female N. tutae or T. absoluta was introduced into this tube which was placed in the thermostatic cabinet (25 ± 1 °C, 60 ± 10% [RH] for a photoperiod of 16:08 [L:D] h), and checked daily to determine insect survival.

Statistical analysis
The statistical analyses were performed using R version 3.4.0. The impact of the treatments (with or without N. tutae) and of the time period (number of days after seeding) on the number of M. euphorbiae was analyzed using a linear mixed model with the treatments and the time as fixed effects and the compartment number as a random effect (the dependent variable follows a normal distribution confirmed by a Shapiro-Wilk test). Moreover, the impact of the treatments on each sampling day on the number of aphids was analyzed using the same method as described above. The impact of the treatments (T, TM, TMN, TN) and of the time period (number of days after seeding) on the number of T. absoluta larvae and adults, as well as the number of N. tutae adults, was analyzed using a generalized linear mixed model following a Poisson distribution. This model was built using the treatments and the time as fixed effects and the compartment number as a random effect. The impact of the treatments on each sampling day on the number of T. absoluta larvae and adults and N. tutae was analyzed using the same method as described above. Whenever necessary, a multicomparison test was done, using the package multcomp. A log-rank test and the Holm-Sidak multiple comparison procedure were used to identify significant differences among the survival curves of T. absoluta and N. tutae adults fed with water, honeydew produced by M. euphorbiae or honey. The figures were obtained using SigmaPlot v.12.5 (Systat Software Inc. 2011).

Greenhouse experiment
The number of T. absoluta larvae depended on the different treatments ( 2 3 = 67.2, P < 0.001) and the sampling day ( 2 6 = 162.9, P < 0.001) (Fig. 2). The number of T. absoluta larvae was stable the first 85 DAS and in the assessments of 99-106 DAS; it increased between the 92 and 99 DAS; and it finally decreased between the 106 and 113 DAS. Taking into consideration each sampling day, the number of larvae in the treatment with T. absoluta only was the lowest until 92 DAS. Then, the number of T. absoluta larvae increased until 113 DAS, with a population peak at the 106 DAS (Table 2, Fig. 2). In the treatment with M. euphorbiae, the number of T. absoluta larvae was higher at the beginning of the assessments and decreased after 78 DAS. However, after 85 DAS the number of T. absoluta larvae increased gradually until the end of the assessments, reaching a density almost twofold lower than when T. absoluta was alone. The treatment with only the parasitoid N. tutae showed a population peak of T. absoluta larvae at 92 DAS and then a decrease until 113 DAS, reaching a very low level by the end of the experiment. In the treatment with the three species, the population dynamic of T. absoluta was similar but showed lower and delayed population peak density (at 99 DAS) as well as lower larvae densities by the end of the experiment (106-113 DAS) (Fig. 2).
The number of T. absoluta adults depended on the various treatments ( 2 3 = 56.6, P < 0.001) and the sampling day ( 2 6 = 391.9, P < 0.001) (Fig. 3). The numbers of T. absoluta adults were similar the two first weeks (71 and 78 DAS). Then they increased slightly between 85 and 92 DAS and then increase steadily by the end of the experiment. The number of T. absoluta adults was not statistically different among the treatments at 71 to 78 DAS and at 106 DAS (Table 2 and Fig. 3). The number of T. absoluta adults in the treatments with only M. euphorbiae + T. absoluta was higher than in the treatment T. absoluta alone (i.e., without M. euphorbiae) and with the three species at 99 DAS. Nevertheless, at 113 DAS in the end of the experiment, the number of T. absoluta adults was highest in the treatment with M. euphorbiae (Fig. 3).
The number of the parasitoid N. tutae depended on the presence of M. euphorbiae ( 2 1 = 12.7, P < 0.001) and on the sampling day ( 2 6 = 37.5, P < 0.001) (Fig. 4). The number of N. tutae adults was stable the first 85 DAS and between 106 and 113 DAS; it increased between 85 and 92 DAS and at 99-106 DAS; and then it finally decreased between 92 and 99 DAS. The number of N. tutae was not statistically different in the first three assessments at 71-85 DAS (Table 2 and Fig. 4). However, in the treatment with M. euphorbiae the number of N. tutae was threefold-fourfold higher than the treatment without M. euphorbiae at 85-113 DAS (Fig. 4).
The number of M. euphorbiae was not strongly impacted on the presence of N. tutae, although marginally significantly with slightly higher aphid number by the end of the experiment when the parasitoid was present (F 1214 = 3.5, P = 0.065). The main factor was impacting aphid density was sampling day (F 1214 = 30.3, P < 0.001) (Table 2, Fig. 5). The number of aphids increased on 85-99 DAS, and it was similar to the assessments at 71-85 DAS as well as those at 99-113 DAS.

Discussion
The importance of sugar food sources in biological control program has been largely reported in light of their positive effects on biocontrol agents (Wäckers et al. 2008;Tena Fig Winkler et al. 2009). Honeydew is the most commonly available sugar food source in agricultural systems (Wäckers et al. 2008) and is produced by sap-feeding insects, including by pests such as aphids and whiteflies (Hendrix et al. 1992;Hullé et al. 2019). In tomato crops, the biological control of the major pest T. absoluta by the parasitoid N. tutae may be modulated by honeydew presences as tomato plants are often infested by whiteflies and/or aphids. The N. tutae parasitoid provided effective biological control of T. absoluta, and this regardless of aphid presence (M. euphorbiae). However, the aphid presence had various effects on the system (i) the peak population of T. absoluta was both delayed and reduced when controlled by the parasitoid, (ii) T. absoluta adults fed with aphid honeydew lived longer, stressing that honeydew food source could be used positively by an herbivore, and more larvae were produced in first part of greenhouse experiment, (iii) in the second part, there were fewer T. absoluta larvae potentially owing to occurrence of resource competition and/or plant-mediated indirect interactions with the aphid. The present study hinted that honeydew producers may impact parasitoid-host population dynamics both through direct and indirect interactions and that potential positive effects of a sugar food source toward a pest can be nullified in the long term by other negative effects occurring simultaneously. Analyzing these interactions may provide data on the potential role of biological control agents on key pests in the framework of IPM in greenhouse tomato crops.

Impact of honeydew
Necremnus tutae is a synovigenic parasitoid species, i.e., females continuously mature eggs throughout their lifespan relying both on their own nutritional reserves and on additional nutrients acquired, for example, through host feeding (Jervis et al. 2001;Balzan and Wäckers 2013). In addition to feeding on host hemolymph, parasitoid females require sugar food sources such as floral nectar and/or honeydew in order to sustain biological traits linked to actual parasitoid fitness (Jervis et al. 2001;Wäckers et al. 2008;Benelli et al. 2017). Honeydew consumption has been shown to increase parasitoid longevity (Hogervorst et al. 2007;Hagenbucher et al. 2014;Tena et al. 2018) and may be involved in parasitoid performance, e.g., by increasing offspring production (Lahiri et al. 2017;Dong et al. 2018;Picciau et al. 2019); this has notably been reported for N. artynes feeding on floral nectar (Balzan and Wäckers 2013) as well as for N. tutae feeding on whitefly honeydew (Dong et al. 2018). We demonstrate further that tomato aphid honeydew enhances N.
tutae adult longevity under laboratory conditions (Fig. 6b); the presence of honeydew producers in greenhouse however provided positive effects on parasitoid longevity, although this effect was confounded with potential positive impact on parasitoid fecundity (Fig. 4. DAS 92-113).
In addition to its impact on parasitoid, honeydew also proved as a potential reliable sugar food source for T. absoluta adults notably increasing their longevity under laboratory conditions (Fig. 6a); it was also suggested by the greenhouse experiment when comparing T. absoluta alone vs. T. absoluta + aphid (significantly higher T. absoluta density in DAS 85, 99 and 113, Fig. 3). Still this positive effect of aphid honeydew on adults was offset in the long term by the multiple negative effects that pullulating aphids have on T. absoluta larvae (see below). In addition, it is known that lifespan increase does not cascade directly to fitness increase in lepidopterans (Jervis et al. 2007), as demonstrated for T. absoluta for which 97% of realized fecundity occurs in first 17 days (Lee et al. 2014). As honeydew led to an increase in T. absoluta longevity from 17 days (water only) to 26 days (honeydew), such increase could only provide marginal increase in fecundity. Honeydew still may provide other benefits to T. absoluta adults as rich-sugar food sources could enhance lepidopterans mobility and thus dispersion (Jervis et al. 2007).

Effects of the secondary pest on biotic interactions at play in the system
Various biotic interactions are usually at play in complex agroecosystems involving multiple arthropod pests and natural enemies, thus identifying potential mechanisms underneath and the demographic consequences of such key interactions is crucial for increasing both applied and fundamental knowledge (Chailleux et al. 2014b;Naselli et al. 2017). In our system, aphid may have affected biotic interactions through three main non-excluding effects on T. absoluta: (i) parasitoid-mediated interactions, (ii) resource competition, and (iii) plant-mediated indirect interactions.
The aphid did provide food source (honeydew) benefiting to the parasitoid N. tutae (at least increasing its longevity), and so, a negative indirect interaction toward T. absoluta occurred through increased parasitoid-induced mortality in the leafminer. Therefore, such effect could be classified as apparent amensalism (−, 0) (Chailleux et al. 2014b) as there was no parasitoid-mediated obvious effect of T. absoluta toward the aphid (Fig. 5). Indeed, albeit T. absoluta larvae density was strongly reduced by parasitoid N. tutae, there was no change in the M. euphorbiae population growth; therefore, any positive effect of N. tutae on aphid population growth (e.g., preventing leaf being destroyed by T. absoluta and reducing the resource competition between the two pests) might have been nullified by the occurrence of other effect(s) (e.g., intraspecific competition could be a main driver in aphid population dynamic, Dixon et al. 1993;Mooney et al. 2008;Mohammed et al. 2019).
Resource competition between the two herbivores showed unilateral effects, with mostly T. absoluta being affected by aphids. The production of T. absoluta larvae was higher in the first part of the experiment, suggesting that no resource competition from aphid was occurring toward the leafminer larvae. However, this increase could be attributed to the positive effect that honeydew had on T. absoluta adults (i.e., increased longevity); living longer they produced more larvae. By contrast, in the second part of the experiment (when most adults from the first released cohort had died) there were fewer T. absoluta larvae in the presence of the aphid, independently of parasitoid presence. Thus, negative effects may have occurred through resource competition between leafminer larvae and aphids, notably owing to pullulating aphids on the plants. Tomato leaves heavily infested by aphids may be unsuitable for the leafminer (e.g., food quality and/or availability is decreased, Villemereuil and Lopez-Sepulcre 2011; Barabas et al. 2016). In addition, potential physical impairment in T. absoluta larvae may result from aphid presence, i.e., a leafminer larvae outgoing from a leaf mine could not dig a new mine owing to overcrowding aphids on the leaf. By contrast to the impact of aphid on T. absoluta through resource competition, the lack of effect of the leafminer toward aphid is somehow surprising as T. absoluta is known to have a major impact on tomato plants (being a defoliator herbivore). For example, previous studies reported that T. absoluta negatively affected biological traits and population dynamic of the whitefly Bemisia tabaci, another major sap-feeder insect on tomato (Bompard et al. 2013;Jaworski et al. 2015;Dong et al. 2018). Such discording results may be linked to the mobility habits of the two pests (aphid and whitefly) which differ drastically and could have its importance in their survival when competing with a defoliating insect. Despite being both sap feeders, mobility of whiteflies is quite limited in non-imaginal stages: only the first nymph stage can move (and barely), the other nymph stages are totally immobile and they die if the hosting leaf is destroyed. By contrast, all aphid stages are well mobile and the destruction of leaves by T. absoluta larvae does not lead to aphid death; they can move and start feeding on another part of the plant. Finally, T. absoluta may simply not have reached densities able to negatively impact aphid population through resource competition in our present study.
Aphids can also be involved in plant-mediated indirect interactions (Stout et al. 2006;Mouttet et al. 2011;Moreira et al. 2018) through triggering salicylic acid-pathway and subsequent plant chemical defenses (Walling 2000). Through this mechanism, M. euphorbiae may have negatively affected leafminer population dynamic (by lowering down population growth rate). Multiple studies documented that sap feeders could affect negatively the performance of leaf chewers (Zhang et al. 2005;Inbar and Gerling 2007;Soler et al. 2012;Mouttet et al. 2013) and our study is in concordance with these data. In addition, the lack of impact of T. absoluta on aphids also matched with results of a previous study from Mouttet et al. (2013) which demonstrated lack of plant-mediated effect of T. absoluta to another sap feeder (B. tabaci). Finally, because our experimental design did not separate spatially both herbivores on the plants, it was not possible to distinguish between direct and plantmediated indirect interactions between T. absoluta and aphid, although both interactions are likely confounded in realistic conditions.

Implication in biological control
Under microcosm and laboratory conditions, the ectoparasitoid N. tutae has proved to be a potential biological control agent of T. absoluta, targeting the larvae stage (Chailleux et al. 2014a;Biondi et al. 2018;Dong et al. 2018). In the present study we demonstrated that N. tutae can provide effective biological control of T. absoluta in tomato greenhouse, keeping the population at relatively low densities during the course of the experiment. Alternative food sources (e.g., honeydew produced by aphid) can help keeping parasitoids close to crops, in addition to providing and impacting the performance and longevity of parasitoids (Wäckers et al. 2008;Benelli et al. 2017) and this without promoting pest population growth (despite positive effect on the adult stage of the pest). However, aphids are pests that are targeted also by pest management program(s) and controlling their populations may compromise any benefits associated with their presence in the framework of N. tutae-based biological control. Other research has been carried out to select plant flowers, as alternative food sources, facilitating the establishment of N. tutae parasitoids and to improve biological control of T. absoluta (Balzan and Wäckers 2013;Arnó et al. 2018). Other alternatives such as the use of banker plants has been subjected to great interest as a means of biological control of invasive and/ or key pests (Bodino et al. 2018;Kruitwagen et al. 2018;Wang et al. 2018). Banker plant systems are based on the use of non-commercial plants infested with non-pest herbivores serving as alternative host-prey for predators or parasitoids (Frank 2010;Parolin et al. 2012;Li et al. 2015;Zhao et al. 2017;Jaworski et al. 2019). When such plant banker systems are infested by aphids (or more generally by sap feeders), they may serve as sources of alternative foods, notably honeydew. Studies determining the factors limiting or potentially optimizing the biological control through manipulating honeydew producers are still needed to develop further biological control and IPM programs.