Combined use of zoophytophagous mirids for sustainable biological protection of greenhouse tomato crops

Abstract

Tuta absoluta and Bemisia tabaci are two key worldwide tomato pests. In response to chemical control limits, the use of zoophytophagous mirid predators as biocontrol agents is being increasingly studied and proposed for the management of these pests. However, the use of some mirid species and in particular Nesidiocoris tenuis, which naturally occurs in European tomato crops, is controversial due to its ability to inflict significant damage on tomato plants (necrotic rings) in parallel of its high predation capacity. Combining other mirid species such as Macrolophus pygmaeus and Dicyphus errans with N. tenuis could reduce the population growth of T. absoluta and of B. tabaci, while maintaining N. tenuis populations below a critical crop damage threshold. In this study, we tested the effect of combining N. tenuis with M. pygmaeus (at two densities: 10 and 15 individuals per plant) and with or without D. errans (five individuals per plant) on the abundance of all insects (predators and pests) for 8 weeks. We also measured leaf damage by T. absoluta and plant and flower damage (necrotic rings, flower abortion rate) by N. tenuis. We found that the combined presence of D. errans, M. pygmaeus and N. tenuis rapidly and significantly reduced the densities of B. tabaci adults, and T. absoluta larvae and damage compared to when M. pygmaeus (with both densities) was only present with N. tenuis. The presence of D. errans was critical to reduce N. tenuis population growth and necrotic rings on tomato plants, while higher densities of M. pygmaeus density reduced the rate of aborted flowers by N. tenuis. The manipulation of M. pygmaeus densities added to the presence of D. errans could provide a sustainable solution to control multiple pests simultaneously while reducing the damaging phytophagous activity of N. tenuis in tomato crops.

Introduction

Insects are considered as agricultural pests if crop damage leads to reduced yield and/or quality of the harvested product to an unacceptable level for farmers. Tomato is the most important vegetable crop worldwide, and both greenhouses and open field productions are attacked by many insect pests, among which are the invasive species Tuta absoluta (Meyrick) (Lepidoptera: Gelechiidae) and Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae) (De Barro et al. 2011; Desneux et al. 2011). These two species are extremely damaging and attack both greenhouse and open field tomato crops and often cohabit on the same plants (Desneux et al. 2010; Biondi et al. 2018; Ramos et al. 2018; Shen et al. 2021).

The tomato leafminer, T. absoluta is now considered as the most devastating pest of tomato in the world (Campos et al. 2017; Mansour et al. 2018; Han et al. 2019; Desneux et al. 2022). Its larvae attack all aerial parts of plant and can cause yield losses up to 100% (Desneux et al. 2010). The whitefly, B. tabaci is a very polyphagous and invasive crop pest worldwide which causes significant yield reduction through direct plant feeding but most importantly by vectoring many viral diseases that negatively impact crop yield (De Barro et al. 2011). Methods based on the spraying of synthetic chemical insecticides to control B. tabaci and T. absoluta have been progressively limited by the rise of population resistances to many molecules (Yao et al. 2017; Biondi et al. 2018; Guedes et al. 2019; Horowitz et al. 2020). In addition, these practices have serious negative impacts on beneficial insect fauna, human health, and the environment (Desneux et al. 2007; Kim et al. 2017; Palma-Onetto et al. 2021; Zafeiraki et al. 2022). Therefore, more sustainable alternative control methods are needed such as the use of natural enemies (Bompard et al. 2013; Campos et al. 2020, Zhang et al. 2022) or botanical insecticides (Ricupero et al. 2022). One of the most widely investigated options is the use of natural enemies, mainly of which are omnivorous or generalist predators, and can contribute to the simultaneous control of multiple pests.

Several species of mirid predators in the genera Dicyphus, Macrolophus, and Nesidiocoris in particular, are naturally present in tomato crops (Alomar et al. 2002; Ferracini et al. 2019) and their regulation role against pests such as whiteflies, aphids, and lepidopteran eggs and larvae is widely recognized (Calvo et al. 2009; Ingegno et al. 2019; Thomine et al. 2020; Pérez-Hedo et al. 2021a). Their effectiveness can be attributed to their voracious consumption of immature stages (Sylla et al. 2016) and their zoophytophagous feeding behaviour, which allows them to subsist on plants during periods of prey shortage (Thomine et al. 2020; Pérez-Hedo et al. 2021b). Nesidiocoris tenuis (Reuter) and Macrolophus pygmaeus (Rambur) (Heteroptera: Miridae) are currently mass-produced and commercialized, while European Dicyphus species are now also included in biocontrol strategies, mainly through conservation (Castañé et al. 2004; Ingegno et al. 2017). Despite their importance and wide use, the phytophagous feeding mode of these omnivorous predators is a double-edge sword: although it allows the maintenance of populations during periods of prey shortage, they may cause important plant damage, especially at high population densities, which calls for an optimization of their use as biocontrol agents. In particular, the intense feeding of N. tenuis on tomato plants causes serious plant damage, including necrotic rings, blisters, and flower abscission (Sánchez 2009; Calvo et al. 2009; Pérez-Hedo and Urbaneja 2016; Moerkens et al. 2020). This phytophagous behaviour has significant negative effects on plant growth and productivity (Arnó et al. 2010; Sánchez and Lacasa 2008; Castañé et al. 2011; Chinchilla-Ramírez et al. 2021), which limits its potential to be used as a biocontrol agent (Abraços-Duarte et al. 2021; Ingels et al. 2022). In many European regions, it is considered as a serious tomato pest regardless of the biocontrol services it provides. It pushes growers to repeatedly control their populations with chemical insecticides, which disrupts their biocontrol capacity (Mouratidis et al. 2022). Instead of N. tenuis, Integrated pest management (IPM) strategies based on the use of mirid predatory biocontrol agents use M. pygmaeus, which has rarely been reported to cause plant damage and is generally considered safe to use (Castañé et al. 2011). However, this species is slow to establish abundant populations on crops (De Backer et al. 2014, Sánchez et al. 2021). In addition, the use of M. pygmaeus alone is not always effective in managing T. absoluta and B. tabaci populations (Calvo et al. 2009; Mollá et al. 2014; Jaworski et al. 2015; Michaelides et al. 2018, Sánchez et al. 2018). Its combination with other biocontrol agents has the potential to improve biocontrol programs (Konan et al. 2021; Yao et al. 2022). Unfortunately, this strategy can be compromised because M. pymaeus population levels tend to decline in favour of N. tenuis, which is naturally present and can feed on M. pygmaeus nymphs (Moreno-Ripoll et al. 2012; Perdikis et al. 2014; Mouratidis et al. 2022). Intraguild predation (IGP), that is the predation of a natural enemy species upon another one sharing a prey species (Chailleux et al. 2014; Mirande et al. 2015), is frequent among mirid species (Moreno-Ripoll et al. 2012; Perdikis et al. 2014; Abraços-Duarte et al. 2021; Mouratidis et al. 2022). Dicyphus errans Fieber (Heteroptera: Miridae) has been studied in the past as potential biocontrol agent but has not been used commercially in augmentative biological control (Madeira et al. 2019; Mouratidis et al. 2022). The high voracity of this predator on different prey combined with low phytophagy could make it an ideal biocontrol agent in commercial greenhouses (Ingegno et al. 2017). In addition, D. errans tends to be the higher-order predator in intraguild predation (IGP) interactions with N. tenuis (Mouratidis et al. 2022), which could help limit N. tenuis population growth and damage. In tomato agroecosystems in Mediterranean regions several species of the genus Dicyphus, Macrolophus and Nesidiocoris co-occur naturally (Ferracini et al. 2019; Mollá et al. 2014; Perdikis et al. 2014). So far, experiments testing the impact of the combined use of N. tenuis and M. pygmaeus and/or D. errans on the growth of pest populations, N. tenuis and their damage to plants have been mainly conducted in the laboratory or in the presence of immobile Ephestia eggs (Mollá et al. 2014; Perdikis et al. 2014; Konan et al. 2021; Mouratidis et al. 2022). Therefore, it is necessary to evaluate these interactions at a larger scale including under commercial greenhouse conditions and in the presence of pests.

The present study aims to test the following hypotheses: (i) interactions between predators (M. pygmaeus, D. errans and N. tenuis) increase the biological control of B. tabaci and T. absoluta, (ii) increasing the density of M. pygmaeus helps limit plant damage caused by N. tenuis by diverting N. tenuis plant feeding towards feeding on M. pygmaeus young nymphs in case of prey scarcity, and (iii) the presence of D. errans helps maintain N. tenuis populations below damaging threshold through intraguild predation. Therefore, we measured the effects of varying the density of M. pygmaeus (10 versus 15 individuals released per plant) and of the presence or absence of D. errans on the population dynamics and damage on tomato of T. absoluta, B. tabaci and N. tenuis.

Material and methods

Insect rearing

Tuta absoluta and B. tabaci colonies were reared in climatic chambers on tomato, Solanum lycopersicum L. and tobacco, Nicotiana tabacum L. (Solanaceae) plants, respectively. The predators N. tenuis and M. pygmaeus were supplied by Koppert Biological Systems (France). A colony of D. errans was established from individuals collected in Southern France, and identified by Jean-Claude Streito (INRAE Montpellier). The three mirid species were reared on tomato plants in cages (30 × 30 × 60 cm) made of insect-proof nylon mesh. They were fed with E. kuehniella Zeller (Lepidoptera: Pyralidae) eggs. Eggs of E. kuehniella were provided as needed every 2 days. All experiments were performed at the French National Research Institute for Agriculture, Food and Environment (INRAE), Sophia Antipolis (France) under laboratory conditions with 24 ± 1 °C, 65 ± 5% RH and 16:8 L:D.

Plant growing

Tomato seeds, S. lycopersicum L. (cv. Nano) were sown in plastic pots (7.9 × 7.9 × 6.5 cm, TEKU, Rixheim, France) in a climate chamber (24 ± 1 °C, 65 ± 5% RH and 16:8 L:D). Twenty-eight days after sowing, tomato plants with four true leaves were transplanted into plastic pots of 10 cm diameter and 9 cm height before being placed in a greenhouse.

Experimental design

The experiment was established in a glass greenhouse divided into four 40 m² compartments (24 ± 1 °C, 65 ± 5% RH). We established a full factorial experimental design with two crossed factors. The pests T. absoluta and B. tabaci and the predator N. tenuis were used in all treatments. The first two-level factor varied the density of M. pygmaeus, i.e. 10 individuals per plant (Mp10) and 15 individuals per plant (Mp15). The second factor was the presence or absence of D. errans (De). Thus, four treatments were tested: Mp10, Mp10De, Mp15 and Mp15De; such labelling is used thereafter to refer to the four treatments. The treatments were replicated four times each, and arranged in a randomized complete block design in the four greenhouse compartments, treated as four blocks. Treatments in each compartment were isolated in tunnels covered with fine mesh netting (height: 2 m, width: 1 m, length: 5 m) to prevent insect transfer (Jaworski et al. 2015). Each tunnel contained two rows of seven tomato plants. To have heterogeneous generations and avoid rapid population decrease, two releases of 30 B. tabaci adults per plant were conducted 24 h after tomato plant transplanting and 25 days later. One couple of T. absoluta adults per plant (28 individuals per tunnel) and the two densities of M. pygmaeus (10 and 15 individuals per plant) were released the same day as the first release of whiteflies. The predators N. tenuis and D. errans were released 7 days after plants were transplanted, at the following densities: N. tenuis and D. errans 2 males, 2 females and 1 nymph per plant each; M. pygmaeus (Mp10) 4 males, 4 females and 2 nymphs per plant; M. pygmaeus (Mp15) 6 males, 6 females and 3 nymphs per plant.

Insect sampling

The population dynamics of each insect species (N. tenuis, M. pygmaeus, D. errans, B. tabaci, and T. absoluta) were monitored weekly for eight weeks starting one week after predators introduction. Each week, ten plants were selected in each tunnel, excluding the first and last plants of each row to avoid border effects. Adults of N. tenuis, M. pygmaeus and D. errans were observed and counted on the whole plants. Nymphs were not counted, due to the difficulty in identifying them unambiguously at the species level at these stages. Adults of B. tabaci were visually observed and counted on both sides of three leaves (1 upper leaf + 1 middle leaf + 1 lower leaf) randomly selected on each of the ten selected plants, and a mean per leaf was calculated for each replicate. In addition, T. absoluta young larvae were counted on the first five apical well-developed leaves, and summed to obtain one number of larvae per five leaves for each replicate. This choice was made for two reasons: (i) the five apical leaves are considered as females preferential oviposition area, and host the vast majority of young T. absoluta larvae, while only older larvae (counted in previous weeks) may be found in older leaves; (ii) while older leaves may be damaged and bear galleries dug by T. absoluta larvae, these galleries are most often empty, since such larvae already underwent pupation (Cocco et al. 2015; Konan et al. 2022).

Plant damage

Plant damage due to T. absoluta was measured as the proportion of tomato leaves attacked by T. absoluta (presence of galleries) in the first five apical leaves inspected for the presence of T. absoluta. Plant damage due to the phytophagous activity of N. tenuis was measured by counting the number of necrotic rings on stems, young shoots, and petioles, as well as the proportion of aborted flowers (number of flowers which did not form fruits relative to the total number of flowers) from the start of the flowering period.

Statistical analysis

All statistical analyses were performed using R Core Team (2023) version 4.3.0. We analysed the effects of the treatments on insect densities separately over different periods coinciding with the completion of a development cycle at the experimental temperature (24 ± 1 °C; Jaworski et al. 2015; Yao et al. 2022). The development cycles of N. tenuis, T. absoluta and B. tabaci development cycles are 21, 21–22 and 19–20 days, respectively, at 24 °C (Bonato et al. 2007; Sánchez 2009; Desneux et al. 2010), so the corresponding density and damage data were analysed separately on Weeks 1–3, 4–6 and 7–8. The development cycles of M. pygmaeus and D. errans at 24 °C are 28 and 31 days, respectively (Perdikis and Lykouressis 2002; Ingegno et al. 2017), so the corresponding density data were analysed separately on Weeks 1–4 and 5–8. We performed independent generalized linear mixed models (GLMMs) with a negative binomial error distribution accounting for data overdispersion (function ‘glmer.nb’, R library ‘lme4’; Bates et al. 2105), except for N. tenuis density in Weeks 4–6 and 7–8, for which we used a linear mixed model (function ‘lmer’, R library ‘lme4’). The fixed effects were M. pygmaeus initial density, the presence/absence of D. errans, and their interactions. The random effect was the plant identifier nested in the greenhouse compartment. Similarly, GLMMs were used to analyse the effects of the treatments on the proportion of attacked leaves by T. absoluta, the number of necrotic rings and the proportion of aborted flowers due to N. tenuis phytophagous activity. The error distributions used were binomial, negative binomial, and binomial error distributions, respectively, and the fixed and random effects were the same as above. The statistical significance of fixed effects was determined through a stepwise regressive type-II model comparison with an ANOVA and a χ²-test. When both factors or the interaction were found to have a significant effect, pairwise mean comparisons between treatments were performed on the best model using the ‘emmeans’ R function (library ‘emmeans’; Lenth 2023). The absence of residual heteroscedasticity and overdispersion was verified in the best model using the functions ‘simulateResiduals()’ (R library ‘DHARMa’; Hartig 2022).

Results

Mirid population densities

The number of N. tenuis adults per plant declined from Week 1 to Week 3, then increased up to Weeks 5–6 and finally stabilized or declined again (Fig. 1A). Neither the initially density of M. pygmaeus nor the presence of D. errans had an effect on N. tenuis numbers during the first development cycle (Weeks 1–3; Fig. 1A; Table 1). However, the interaction between M. pygmaeus initial densities and the presence of D. errans significantly affected the number of N. tenuis per plant during the second cycle (Weeks 4–6): both higher initial densities of M. pygmaeus and the presence of D. errans increased N. tenuis densities, but the two effects combined resulted in lower N. tenuis densities than in treatments with either effect alone (Mp15 or Mp10De) and than in the control (Mp10) over this period (mean ± SE of adults per plant: Mp10 5.25 ± 0.21, Mp10De 5.62 ± 0.21, Mp15 6.23 ± 0.22, Mp15De 4.68 ± 0.21; Fig. 1A; Table 1). Finally, during the third cycle (Weeks 7–8), both M. pygmaeus initial densities and the presence of D. errans significantly affected N. tenuis densities, but not the interaction term: increasing M. pygmaeus initial density increased N. tenuis densities, while the presence of D. errans had the opposite effect (mean ± SE of adults per plant: Mp10 5.65 ± 0.21, Mp10De 3.93 ± 0.20, Mp15 5.96 ± 0.21, Mp15De 4.44 ± 0.22; Fig. 1A; Table 1).

Fig. 1

Mean (± SE) number of (A) N. tenuis, (B) M. pygmaeus, and (C) D. errans adults per plant depending on the initial density of M. pygmaeus introduced and with or without D. errans. ‘Mp10′ 10 initially introduced M. pygmaeus; ‘Mp15′ 15 initially introduced M. pygmaeus; ‘De’ D. errans present. Vertical dashed lines show the time periods used in statistical tests and coinciding with estimated insect generation times. Different letters show significant differences between treatments or groups of treatments within each time period. N = 40 replicates per treatment and per week (insects counted on 10 plants per tunnel with four tunnels per treatment)

Table 1 Response of the number of N. tenuis, M. pygmaeus and D. errans adults per plant to the initial density of M. pygmaeus introduced, the presence of D. errans, and the second-order interaction. (-) parameters not included in the model

The number of M. pymaeus individuals per plant declined from Week 1 to Week 3 and stabilised at low densities from Week 3 until the end of the experiment (Fig. 1B). Higher initial densities of M. pygmaeus led to higher M. pygmaeus densities over both periods tested (Weeks 1–4 and 5–8), but the presence of D. errans had no effect (mean ± SE of adults per plant: Weeks 1–4: Mp10 2.71 ± 0.23, Mp10De 2.36 ± 0.19, Mp15 3.71 ± 0.26, Mp15De 3.38 ± 0.24; Weeks 5–8: Mp10 0.29 ± 0.05, Mp10De 0.46 ± 0.06, Mp15 1.12 ± 0.10, Mp15De 1.14 ± 0.10; Fig. 1B; Table 1). The number of D. errans per plant declined from Week 1 to Week 3, then increased from Week 5 to Week 8 (Fig. 1C). However, the initial densities of M. pygmaeus had no effect on D. errans densities neither during the first (Weeks 1–4) nor the second cycle (Weeks 5–8; Fig. 1C, Table 1).

Pest densities

The number of T. absoluta larvae per five leaves declined through time overall, and it was not affected by the initial density of M. pygmaeus, but was affected by the presence of D. errans over each of the three periods tested (mean ± SE of adults per plant: Weeks 1–3: Mp10 2.14 ± 0.21, Mp10De 1.84 ± 0.19, Mp15 2.42 ± 0.27, Mp15De 1.73 ± 0.19; Weeks 4–6: Mp10 1.47 ± 0.14, Mp10De 0.56 ± 0.10, Mp15 1.11 ± 0.12, Mp15De 0.46 ± 0.08; Weeks 7–8: Mp10 0.80 ± 0.12, Mp10De 0.25 ± 0.07, Mp15 0.56 ± 0.05; Fig. 2A, Table 2).

Fig. 2

Mean (± SE) number of (A) T. absoluta larvae per five leaves, and (B) B. tabaci adults per leaf, depending on the initial density of M. pygmaeus introduced and with or without D. errans. ‘Mp10′ 10 initially introduced M. pygmaeus; ‘Mp15′ 15 initially introduced M. pygmaeus; ‘De’ D. errans present. Vertical dashed lines show the time periods used in statistical tests and coinciding with estimated insect generation times. Different letters show significant differences between groups of treatments within each time period. N = 40 replicates per treatment and per week (insects counted on 10 plants per tunnel with four tunnels per treatment)

Table 2 Response of the number of T. absoluta larvae per five leaves and of the mean number of B. tabaci adults per leaf to the initial density of M. pygmaeus introduced, the presence of D. errans, and the second-order interaction. (-) parameters not included in the model

Similar to T. absoluta larvae, the mean number of B. tabaci adults per leaf also declined through time overall, and was affected by the presence of D. errans, but not the initial densities of M. pygmaeus over the three periods tested, except that the interaction term was significantly negative in the first period (mean ± SE of adults per plant: Weeks 1–3: Mp10 5.29 ± 0.27, Mp10De 3.98 ± 0.29, Mp15 3.91 ± 0.27, Mp15De 3.99 ± 0.27; Weeks 4–6: Mp10 2.61 ± 0.16, Mp10De 1.78 ± 0.16, Mp15 2.16 ± 0.14, Mp15De 1.74 ± 0.14; Weeks 7–8: Mp10 1.20 ± 0.13, Mp10De 0.60 ± 0.08, Mp15 1.12 ± 0.14, Mp15De 0.50 ± 0.08; Fig. 2B, Table 2).

Plant damage

The proportion of leaves attacked by T. absoluta per plant was relatively stable through time (Fig. 3A). It was increased in treatments with increased initial densities of M. pygmaeus during the first and second T. absoluta cycles but not the third, and it was reduced by the presence of D. errans during the second cycle only (mean ± SE percentage of leaves attacked by T. absoluta per plant: Weeks 1–3: Mp10 26 ± 2, Mp10De 20 ± 2, Mp15 34 ± 3, Mp15De 27 ± 2; Weeks 4–6: Mp10 22 ± 2, Mp10De 14 ± 2, Mp15 21 ± 2, Mp15De 16 ± 2; Weeks 7–8, all treatments: 14 ± 1; Fig. 3B, Table 3).

Fig. 3

Damage by T. absoluta (A) and by N. tenuis (B, C) on tomato plants: A percentage of leaves attacked by T. absoluta per plant; B Number of necrotic rings per plant; C Percentage of aborted flowers per plant. ‘Mp10′ 10 initially introduced M. pygmaeus; ‘Mp15′ 15 initially introduced M. pygmaeus; ‘De’ D. errans present. Vertical dashed lines show the time periods used in statistical tests and coinciding with estimated generation times of the insect species causing plant damage. Different letters show significant differences between treatments or groups of treatments within each time period. N = 40 replicates per treatment and per week (insects counted on 10 plants per tunnel with four tunnels per treatment)

Table 3 Response of the proportions of attacked leaves by T. absoluta, and of the number of necrotic rings and the proportion of aborted flowers due to N. tenuis to the initial density of M. pygmaeus introduced, the time (date of the sampling), the presence of D. errans as well as all the pairwise interaction (GLMM following a negative binomial distribution). (-) parameters not included in the model

The number of necrotic rings caused by N. tenuis increased rapidly between Weeks 2 and 3, and slowly after that (Fig. 3B). It was lower under higher M. pygmaeus initial densities in Weeks 4–6 and 7–8 but not in Weeks 1–3, and it was also reduced by the presence of D. errans in Weeks 7–8 only (mean ± SE number of necrotic rings per plant: Weeks 4–6: Mp10 7.78 ± 0.36, Mp10De 7.57 ± 0.28, Mp15 6.78 ± 0.31, Mp15De 6.49 ± 0.21; Weeks 7–8, Mp10 9.63 ± 0.41, Mp10De 7.50 ± 0.29, Mp15 8.39 ± 0.33, Mp15De 6.59 ± 0.28; Fig. 3B, Table 3).

Finally, the proportion of aborted flowers, also caused by the phytophagous activity of N. tenuis, steadily increased from Week 1 to 4, and stabilised in Weeks 4–5 (Fig. 3C). Similar to the number of necrotic rings, it was reduced under higher initial M. pygmaeus densities, but only during the second period (Weeks 4–5), and it was not affected by the presence of D. errans ( mean ± SE percentage of aborted flowers per plant, Weeks 4–5: Mp10 27 ± 2, Mp10De 29 ± 2, Mp15 23 ± 2, Mp15De 217 ± 2; Fig. 3C, Table 3).

Discussion

In the present study, we evaluated how combining mirid predators can be used to reduce crop damage by N. tenuis, a biocontrol agent now considered a serious tomato pest in Europe (Moerkens et al. 2020), while increasing biological pest control. We found that increasing M. pygmaeus initial density increased the early population growth (second generation) of N. tenuis, while the presence of D. errans reduced N. tenuis population growth durably over a longer time. The presence of D. errans also reduced densities of both T. absoluta and B. tabaci, resulting in reduced leaf damage by T. absoluta. Plant damage by N. tenuis was also reduced in treatments combining mirid predators.

We expected a faster early population growth of N. tenuis when M. pygmaeus individuals were released at a higher density. This is because N. tenuis can feed on M. pygmaeus nymphs (Mouratidis et al. 2022), representing an alternative prey to the primary prey T. absoluta and B. tabaci. Therefore, increasing the initial density of M. pygmaeus would increase the prey resources available to N. tenuis, and thereby increase its population growth. We also hypothesized that the higher availability of M. pygmaeus nymphs would divert N. tenuis from feeding on plant resources, which in turn would decrease plant damage, since low prey densities in crop critically increases the phytophagy of N. tenuis (Calvo et al. 2009; Castañé et al. 2011; Yao et al. 2022). Indeed, we found that the number of necrotic rings and the percentage of flower abortion was slightly lower in treatments with higher M. pygmaeus initial densities, confirming the results of Yao et al. (2022). Feeding on plants is an essential requirement for N. tenuis to optimize the digestion and assimilation of consumed prey (Gillespie and McGregor 2000), and higher availability of M. pygmaeus prey likely compensates only partially the need to feed on plants in case of main prey shortage.

We also expected that the presence of D. errans would decrease N. tenuis population growth through IGP, since D. errans preys on N. tenuis nymphs (Abraços-Duarte et al. 2021; Mouratidis et al. 2022). The three predator species in the same environment inevitably compete for the shared resource (prey), likely increasing the frequency of encounters with heterospecific individuals, and therefore the likelihood of IGP. A reduction in N. tenuis population growth due to IGP by D. errans has been previously observed (Mouratidis et al. 2022). Here, the decrease in N. tenuis population densities from Week 6 in treatments with D. errans could be related to the increase in population densities of D. errans from Weeks 3–4 (Fig. 1). This reduction in N. tenuis population densities due to D. errans predation on N. tenuis nymphs also led to reduced necrotic rings by N. tenuis, since they have been shown to directly relate to N. tenuis density (Sánchez et al. 2009). Reducing densities of N. tenuis nymphs is especially important to limit tomato plant damage (Arnó et al. 2010; Calvo et al. 2012). However, the presence of D. errans did not influence the proportion of aborted flowers in our experiment, despite a consequently reduced population density of N. tenuis. This could be due to the still low D. errans densities during the flower sampling period (Weeks 3–6). Madeira et al. (2019) and Mouratidis et al. (2022) reported the necessity of D. errans prior establishment in crops for the efficient control of N. tenuis and reduced plant damage.

Our results showed a decrease in the three mirid population densities from Weeks 1 to 3 regardless of the treatments applied, due to a slow population establishment after the release. This may be related to increased mortality due to disturbance, and to the lack of prey availability. Indeed, predators were released one week after prey release, and remained at low densities throughout the experiment. The predator population increases from Week 4 coincided with the emergence of new generations. Indeed, between 24 and 25 °C (near experimental temperature 24 ± 1 °C) periods of about 21, 28 and 31 days were required to complete N. tenuis, M. pygmaeus and D. errans developmental cycles, respectively (Perdikis and Lykouressis 2002; Sánchez et al. 2009; Ingegno et al. 2017). The increase in D. errans densities from Week 4 onward would therefore coincide with the emergence of a new adult generation (only adult predators were counted), or with the introduction of new prey (second release of B. tabaci) 24 days after the first one (i.e. 4 days before the sampling of Week 3). For Dicyphus species, prey availability is an important factor for good establishment on crops (Madeira et al. 2019; Abraços-Duarte et al. 2021; Mouratidis et al. 2022) in contrast to N. tenuis, which is able to reproduce on plants and grow populations in absence of prey (Arnò et al. 2010; Castañé et al. 2011; Pérez-Hedo and Urbaneja 2016).

Unlike D. errans, population densities of M. pygmaeus declined continuously in our experiment, which is likely mainly due to interactions with its competitors N. tenuis and D. errans. First, M. pygmaeus has been shown to have a low capacity of adaptation and establishment (De Backer et al. 2014, Sánchez et al. 2021). Unlike N. tenuis, M. pygmaeus reproduction is more difficult under most common temperature regimes in greenhouses tomato crops (Ingegno et al. 2021). Notably, it is identified as the most thermophile species of mirids in the Mediterranean area (Sánchez et al. 2009; Martínez-García et al. 2017). Second, M. pygmaeus is less competitive for prey resources: indeed, it is considered less aggressive and preys on T. absoluta and B. tabaci at a lower rate compared to other mirid species (Perdikis et al. 2009; Mollá et al. 2014; Sylla et al. 2016; Michaelides et al. 2018). For example, N. tenuis is known to occupy the upper plant strata (Perdikis et al. 2014) with a higher prey consumption rate than M. pygmaeus (Perdikis et al. 2009; Mollá et al. 2014). This limits the access of M. pygmaeus to meristematic plant tissues of higher nutritional quality (Arnó et al. 2010) and to prey such as T. absoluta and B. tabaci, which prefer the upper strata of tomato plant ( Cocco et al. 2015). Montserrat et al. (2004) reported that M. pygmaeus was energetically more conservative in foraging than Dicyphus predators, leading to a comparably lower encounter and predation rate on whiteflies. Finally, M. pygmaeus has a position of IGP prey: predation of M. pygmaeus young nymphs by N. tenuis females has already been reported (Moreno-Ripoll et al. 2012; Perdikis et al. 2014), and predation by Dyciphus predators is also likely (Abraços-Duarte et al. 2021). However, in our experiment we observed that D. errans preferred preying on N. tenuis rather than M. pygmaeus nymphs, which has been demonstrated in other studies (Mouratidis et al. 2022).

We found that an increased total predator pressure—either through high M. pygmaeus densities or through the presence D. errans along with N. tenuis populations—was necessary to improve the control of B. tabaci and T. absoluta populations, leading to lower leaf damage by T. absoluta. The combination of predators can have a complementary or additive effect in controlling pest populations. In fact, Yao et al. (2022), under the same experimental conditions showed a beneficial effect of the joint use of N. tenuis and M. pygmaeus on B. tabaci whitefly and T. absoluta population densities reduction. This may be due to complementary ecological niches of coexisting predators and / or avoidance of IGP (Perdikis et al. 2014; Aparicio et al. 2020; Konan et al. 2021; Chen et al. 2022; Cruz-Miralles et al. 2022). According to Lucas and Alomar (2002), the level of B. tabaci predation was higher in the lower part of the plant when Dicyphus tamaninii (Hemiptera: Miridae) and M. pygmaeus coexisted than when each predator species was alone. Similarly, a high predation of T. absoluta eggs was observed by Konan et al. (2021) when N. tenuis and M. pygmaeus were combined. The rapid reduction in B. tabaci adult and T. absoluta larval populations when D. errans was present could result from the effectiveness of the D. errans—N. tenuis combination in controlling these pests (Calvo et al. 2012; Ingegno et al. 2019; Mollá et al. 2014; Madeira et al. 2019). This could explain why pest density decreased when D. errans density increased. In the absence of D. errans, the number of T. absoluta larvae still declined, although more slowly and progressively.

The controversial status of N. tenuis – both a predator and a pest causing extensive damage on tomato – questions its use as biocontrol agent. This calls for new solutions for the integrated management of N. tenuis and other tomato pests. Our study demonstrated that combinations of biocontrol agents offers interesting avenues by manipulating interactions between predators (competition for food, complementarity, IGP, etc.). The combined use of M. pygmaeus and D. errans where N. tenuis is naturally present provides an efficient control of B. tabaci and T. absoluta populations, while limiting tomato plant damage by N. tenuis. Future studies should focus on investigating the optimal densities and release times of predators in these multiple biocontrol agent approaches. Such approaches could also be integrated with the use of service plants or alternative host plants to reduce N. tenuis phytophagy and damage on crop plants (Biondi et al. 2016, Konan et al. 2021; Chailleux et al. 2022; Castillo et al. 2022). Since top-down effects of predators on pest populations can be strongly modulated by bottom-up effects such as the manipulation of inputs to crop plants, it would be interesting to investigate how bottom-up effects affect intraguild interactions and the pest control efficacy of predator combinations (Becker et al. 2021; Han et al. 2022). The optimization of these new IPM tools should provide more sustainable perspectives in the management of zoophytophagous mirids and other pests in vegetable agrosystems.