Evaluating predatory arthropods for the biological control of the invasive Thrips parvispinus (Thysanoptera: Thripidae) in greenhouse sweet pepper Authors: Estefanía Rodríguez, Mario Porcel, Antonio Robledo, Lidia Lara, María del Mar Téllez, María del Mar Montoro, Clemente Martínez, Apostolos Pekas Published: 2026 ABSTRACT AbstractBACKGROUND Thrips parvispinus is an invasive pest of increasing concern in European greenhouse crops, especially sweet pepper. Its recent spread in southeastern Spain has disrupted integrated pest management protocols relying on inoculative releases of Orius laevigatus and Amblyseius swirskii. Although these predators effectively control Frankliniella occidentalis, their effectiveness against T. parvispinus is unclear. We evaluated the efficacy of O. laevigatus, A. swirskii, and the thrips‐specific Franklinothrips vespiformis, alone and combined, against T. parvispinus under semi‐field greenhouse conditions, also assessing predator compatibility, pest and predator distribution, and crop productivity.RESULTS O. laevigatus, alone or in the three species combination, provided the most effective pest suppression. F. vespiformis was compatible with O. laevigatus, although A. swirskii abundance declined in the combined treatment, suggesting intraguild interactions between predators. T. parvispinus was primarily a leaf‐dwelling thrips, with only 5.5% of individuals found in flowers. However, this proportion varied depending on predator presence, suggesting predator‐induced movement. Notably, pest abundance in flowers was higher in the A. swirskii treatment compared with the other predators and the control, suggesting predator‐induced behavioral avoidance. Fruit productivity mirrored overall pest suppression, with significantly higher yields observed in the O. laevigatus and combined treatments.CONCLUSIONThese findings refine our understanding of predator interactions and support the inclusion of compatible thrips‐specific predators as promising components of biological control strategies for T. parvispinus in sweet pepper crops. © 2026 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry. INTRODUCTION Thrips parvispinus (Thysanoptera: Thripidae) (Karny, 1922) is an invasive pest of increasing global concern, affecting a wide range of crops including ornamental plants, fruit trees, and vegetables. 1 Likely native to India, 2 it has expanded its distribution over the past three decades, with confirmed reports in Africa, North America, and Europe. 2 , 3 In Europe, sweet pepper (Capsicum annuum L.) is one of its primary host plants. 4 In Spain, T. parvispinus was first reported in 2017 on ornamental crops, 5 and its presence in sweet pepper was confirmed in 2022. 6 Since then, it has caused substantial damage to this economically important crop, which accounts for more than 1.09 million tonnes of annual production in Spain and generates more than €1.28 billion in turnover. 7 The continued spread of T. parvispinus represents a major threat to the sustainability of sweet pepper cultivation and disrupts established integrated pest management (IPM) strategies. Therefore, new species of natural enemies and strategies are needed against this pest. Southern Spain constitutes one of the largest contiguous areas worldwide where augmentative biological control is systematically applied, with sweet pepper cultivation as a prime example. For decades, IPM strategies in the region have reduced reliance on chemical pesticides through the use of natural enemies. 8 , 9 Today, nearly all of the approximately 12 500 ha of sweet pepper grown in Southern Spain rely on biological control as the main pest management strategy, 7 predominantly through the release of predatory arthropods including the generalist predator Orius laevigatus (Hemiptera: Anthocoridae) and species of predatory mites of the family Phytoseiidae. O. laevigatus has long been considered a cornerstone in the control of flower‐dwelling thrips, especially Frankliniella occidentalis Pergande (Thysanoptera: Thripidae), which was until recently the dominant thrips pest in sweet pepper. 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 However, the emergence of T. parvispinus has exposed limitations in this approach. Unlike F. occidentalis, which primarily inhabits flowers, T. parvispinus tends to dwell on leaves, a habitat less frequently patrolled by O. laevigatus. The effectiveness of O. laevigatus against leaf‐dwelling species may be reduced because of its floral foraging habits and pollen‐based nutrition. However, O. laevigatus has demonstrated certain capacity to control other leaf‐dwelling thrips, such as Echinothrips americanus Morgan and Thrips setosus Moulton. 18 Among the biocontrol agents used in IPM programs in sweet pepper, phytoseiid mites such as Amblyseius swirskii Athias‐Henriot are widely applied for their capacity to suppress whitefly and thrips populations, particularly the flower‐dwelling F. occidentalis. 11 , 14 , 19 In particular, the combined application of O. laevigatus and A. swirskii has been effective in reducing F. occidentalis populations in sweet pepper. However, the efficacy of A. swirskii against T. parvispinus appears limited. Behavioral studies have shown that while A. swirskii can prey on first‐instar larvae, second‐instar larvae and adult females of T. parvispinus exhibit defensive behaviors and may even prey on A. swirskii eggs. 20 Although this interaction is unlikely to severely impact mite populations, 1 , 20 it raises concerns about the overall reliability of A. swirskii in controlling T. parvispinus. The ant‐mimicking thrips Franklinothrips vespiformis Crawford (Thysanoptera: Aeolothripidae) is a generalist predator known to feed on thrips and other small arthropods. 21 , 22 , 23 It has gained attention as a potential biocontrol agent for a variety of greenhouse pests and is commercially available in Europe for use in IPM programs. For instance, F. vespiformis has been used in greenhouse crops such as roses, for the management of Thrips tabaci Lindeman and F. occidentalis. 24 Its efficiency in preying on the leaf‐dwelling thrips E. americanus and whiteflies, suggests potential for broader use in protected vegetable crops, including sweet pepper. 21 , 23 Despite these promising attributes, its efficacy against T. parvispinus under Mediterranean greenhouse conditions remains untested. In this study, we investigated in a greenhouse experiment the efficacy of three thrips predators (O. laevigatus, A. swirskii, and F. vespiformis), applied individually and in combination, in suppressing T. parvispinus populations on sweet pepper. We also assessed the impact of predator releases on several variables of crop productivity. Because little is known about the distribution of T. parvispinus across different plant organs, thrips populations and predator densities were monitored in both leaves and flowers. Although research on T. parvispinus remains limited compared with other thrips species, this study provides essential insights into the efficacy of potential biological control agents under Mediterranean greenhouse conditions, representing a necessary contribution to this emerging field. These results will shed light on potentially new strategies for the management of T. parvispinus and safeguard crop production in greenhouse sweet pepper. MATERIALS AND METHODS Plants and insects Pepper plants (Capsicum annuum L., cv. Kation; F1 hybrid, Clause commercial house) were cultivated in 150‐cell trays filled with commercial substrate (Brill, Gebr. Brill Substrate GmbH & Co., Georgsdorf, Germany) in a climate‐controlled growth chamber [26 ± 1 °C, 60% relative humidity (RH), 16:8 h light/dark photoperiod, photosynthetic photon flux density (PPFD) of 67.2 μmol m−2 s−1]. Sowing was carried out on 21 February 2024, and seedling emergence occurred on 1 March 2024. Plants were watered twice weekly using a nutrient solution containing 12 mmol L−1 NO₃−, 1.5 mmol L−1 PO₄3−, 5.5 mmol L−1 K+, 4 mmol L−1 Ca2+, 0.6 mmol L−1 NH₄+, and 30 mg L−1. After 35 days, the seedlings were transplanted into pots (10.5 × 10.5 × 14 cm) in the same growth chamber. Twenty‐five days later, plants (~30 cm tall), bearing only leaves (~15 fully expanded true leaves), no flowers, or fruits, were moved to the greenhouse and transplanted into 5‐L pots for the experiment. Colonies of T. parvispinus were initially established from individuals provided by Professor P. Bielza (Polytechnic University of Cartagena, Spain). The rearing was maintained in plastic containers (18 × 12 × 6.5 cm) under the same climate conditions as previously described for plant growth (26 ± 1 °C, 60% RH, 16:8 h light/dark photoperiod, PPFD 67.2 μmol m−2 s−1). Each container contained three or four fresh green bean pods (Phaseolus vulgaris L. cv Helda) serving as both food and oviposition substrate, with 20 adult thrips per container. To prevent direct contact with moisture at the base of the containers, the pods were placed on a plastic mesh elevated above a thin layer of substrate (~1.5 cm), the same as used for the seedlings and pots. Container lids were fitted with fine metallic mesh to ensure ventilation. The bean pods were replaced two or three times weekly. Before use, to prevent fungal growth and ensure pods duration, pods were washed in water for 15 min, disinfected in 2% bleach [35 g L−1 sodium hypochlorite (NaOCl)] for 5 min, rinsed thoroughly, and air‐dried on filter paper. To increase colony size before the experiments, two pepper plants were introduced into the rearing chamber 17 days before the start of the experiment, allowing thrips to colonize them naturally. The colonies were periodically supplemented with individuals collected from a commercial pepper greenhouse in Almería, Spain. All field‐collected specimens were taxonomically confirmed as T. parvispinus by the Plant Health Laboratory of the Junta de Andalucía. All predators used in the experiments, namely F. vespiformis, O. laevigatus, and A. swirskii were supplied by Biobest Sistemas Biológicos S.L. (Almería, Spain) as Vespiformis‐System, Orius‐System 500, and Swirskii‐System in sachets, respectively. Greenhouse experiment The experiment was conducted from 23 April to 4 June 2024 at the IFAPA La Mojonera Research Center (Almería, Spain) within a single multi‐span polyethylene greenhouse facility. Inside this structure, 20 independent mesh‐enclosed experimental cages (2 × 3 × 2 m each) were arranged in parallel rows along the central corridor of the greenhouse. Each cage was constructed using thrips‐proof mesh (300 × 300 μm2), with individually sealed mesh‐front doors to prevent external contamination and cross‐treatment interference. The cages were spatially distributed into four blocks, based on a combination of north–south and east–west orientations (Supporting Information, Fig. S1). Each block contained five cages, enabling each treatment to be replicated once per block. Each compartment contained six sweet pepper plants (Capsicum annuum), grown in 5‐L pots, with a spacing of 1 m between plants (Supporting Information, Fig. S1). Temperature and RH were recorded hourly throughout the experiment using four HOBOware® temperature/RH data loggers, positioned inside the experimental cages in the northwest, northeast, southwest, and southeast compartments. No chemical treatments (insecticides or fungicides) were applied before and during the experiment. Five treatments were established in a completely randomized block design, each with four replicates: (i) control (only T. parvispinus and no predators), (ii) T. parvispinus + A. swirskii, (iii) T. parvispinus + F. vespiformis, (iv) T. parvispinus + O. laevigatus, and (v) T. parvispinus + all three predator species combined (Supporting Information, Fig. S1). One day after placing the plants in the compartments, 15 adult female T. parvispinus were released per plant. The thrips were collected by using an aspirator from the laboratory colony. Plants were maintained under pest‐only conditions for the first 3 weeks after thrips inoculation. Sampling began 1 week after T. parvispinus was introduced and continued weekly for 7 weeks. Fruit harvest was performed only at the end of the experiment (date 7). The predators were introduced 3 weeks later, at the onset of flowering, as occurs in commercial greenhouses to optimize the performance of O. laevigatus. Flowering started during the third week after thrips inoculation. Release rates were as follows: three commercial sachets of A. swirskii per compartment (~125 individuals per plant), and eight adult females per plant for both each of O. laevigatus and F. vespiformis. These release rates, particularly for O. laevigatus, reflect the doses currently applied in commercial greenhouse sweet pepper production under high T. parvispinus pressure in the study area. 25 On each plant, three leaves (from the upper, middle, and lower canopy) and three flowers were inspected using a ×10 hand lens. We recorded on both leaves and flowers: (i) T. parvispinus larvae and adults, (ii) A. swirskii eggs and mobile stages, (iii) O. laevigatus nymphs and adults, and (iv) F. vespiformis larvae and adults. To prevent cross‐contamination, each treatment was sampled by dedicated personnel. At the end of the experiment (date 7), all fruits were counted and 3 plants per cage (12 per treatment) were harvested. The plants (once uprooted) and all fruits from these plants were both fresh weight and dry weight after drying in an oven at 60 °C for 24 h, were measured. Statistical analysis The efficacy of the treatments relative to the control was assessed by analyzing the total abundance of T. parvispinus (adults and immatures), and their abundance on leaves and flowers, using three Generalized Linear Mixed Models (GLMMs). In each model treatment and block were included as fixed factors. Time, expressed as weeks from predator introduction (week 4) to plant collapse (week 6) (Fig. 1), was included as a continuous covariate, except in the model for T. parvispinus abundance in flowers, where it was treated as a categorical fixed effect due to non‐linearity (Fig. 1(C)). Cage and plant were included as random effects to account for spatial autocorrelation within cages and for repeated measures on individual plants. The variable block was excluded from the models when it was not significant (GLMM: likelihood ratio test (LRT), P > 0.05). An additional GLMM, using total T. parvispinus abundance as the response variable and the same model structure, was fitted to data from week 3 to verify that pre‐treatment infestation levels were homogeneous. Boxplots of the abundance of Thrips parvispinus (adults and larvae) recorded on sweet pepper plant in leaves and flowers (total) (A), leaves (B) and flowers (C) during the greenhouse experiment conducted under controlled conditions comparing different biological control treatments. Different letters denote statistically significant differences among treatments from weeks 4 to 6 (Generalized Linear Mixed Model: false discovery rate‐corrected multiple comparison, P < 0.05). Differences in predator abundance between treatments were also analyzed using GLMMs. Five models were built with the abundance of adults and larvae of F. vespiformis, adults and nymphs of O. laevigatus and total abundance of A. swirskii as response variables. All models included treatment and block as fixed factors, and cage and time (weeks 4 to 7) as random effects. Block was excluded if found non‐significant. Fruit production (number of peppers per plant) was analyzed using a GLMM, again with treatment and block as fixed factors and cage as a random effect. Initially, all models were fitted using a Poisson distribution and a log‐link function suitable for count data. Where overdispersion was detected, we implemented a negative binomial distribution. Plant and fruit biomass (fresh and dry weights), were analyzed using Linear Mixed Models (LMMs). These models also included treatment and block as fixed effects and cage as a random effect. Fruit fresh and dry weight models exhibited heteroscedasticity in the residuals, indicating heterogeneous variances between treatments. We therefore specified these LMMs using the function ‘varIDent’ to account for this heterogeneity. The improved fit of the models was confirmed using likelihood ratio tests (GLMM: LRT, P < 0.001). Type II Wald tests were used for statistical inference in all models, and multiple comparisons of factor levels was carried out by means of a multiple comparison matrix with P‐values adjusted using the false discovery rate (FDR) method. Finally, model assumptions – that is, normality and homoscedasticity for LMMs, and overdispersion, zero inflation, and absence of patterns in the residuals for GLMMs – were verified. All the analyses were carried out with R v.4.5.1 for Windows using packages ‘lmer4’, 26 ‘nlme’, 27 ‘DHARMa’ 28 and ‘multcomp’. 29 RESULTS Thrips parvispinus abundance and population dynamics The mean ± SE temperature during the experimental period was 26.01 ± 0.5 °C, and the mean RH was 52.8 ± 0.8%. T. parvispinus abundance did not differ among treatments prior to predator release (Fig. 1(A)) (GLMM: χ2 = 4.8, d.f. = 4, P = 0.304), and lower abundance was registered in the southwestern block (5.4 ± 0.9; cumulative mean per plant ± SE) compared with the rest (10.6 ± 0.8) (GLMM: FDR‐corrected multiple comparison, P < 0.05). Predator treatments had a significant impact on T. parvispinus abundance relative to the control (Fig. 1(A)) (GLMM: χ2 = 93.6, d.f. = 4, P < 0.001). The control consistently exhibited the highest thrips densities throughout the experiment. Significant differences among treatments became apparent by week 4, with T. parvispinus populations in the control continuing to increase until week 6. A sharp decline was recorded in week 7 due to the collapse of the untreated plants. Among the predator treatments, O. laevigatus and the combined release achieved the most effective suppression of thrips populations. Both treatments significantly reduced T. parvispinus abundance by week 6, with no statistical difference between them (Fig. 1(A)). By contrast, individual releases of F. vespiformis and A. swirskii were less effective, and no significant difference in performance was observed between these two predators, although both differed significantly from the control. Notably, the combination of O. laevigatus with either A. swirskii or F. vespiformis did not enhance control beyond that achieved by O. laevigatus alone, suggesting an absence of additive or synergistic effects (Fig. 1(A)). Higher T. parvispinus abundance was observed in blocks oriented east (301.2 ± 24.5; cumulative mean per plant ± SE) than in blocks oriented west (193.9 ± 25.6) (GLMM: FDR‐corrected multiple comparison, P < 0.05). In leaves, T. parvispinus abundance mirrored overall population dynamics, reflecting the high concentration of the pest in this plant organ (Fig. 1(B)), with treatments differing significantly, as observed in the overall population analysis (GLMM: χ2 = 90.3, d.f. = 4, P < 0.001). The block also affected T. parvispinus abundance with higher values recorded for the eastern blocks (GLMM: FDR‐corrected multiple comparison, P < 0.05). However, different trends were observed in flowers (Fig. 1(C)). Although only 5.5% of total T. parvispinus individuals were found in flowers (Table 1), significant differences were found among treatments (GLMM: χ2 = 65.4, d.f. = 4, P < 0.001). The highest thrips abundance in flowers occurred in the A. swirskii treatment, even exceeding thrips abundance in the control (Fig. 1(C)). Amblyseius swirskii was the least observed predator in flowers with 0.4% detected in this organ (Table 1). Conversely, O. laevigatus was predominantly observed in flowers (59.4% of total individuals), where T. parvispinus populations were significantly reduced in both the O. laevigatus‐only and the combined treatments, with no significant difference between them (Fig. 1(C)). This suggests that O. laevigatus occupies flowers primarily for pollen feeding rather than predation. In the F. vespiformis treatment, thrips counts were lower than in the control and A. swirskii treatments but did not significantly differ from the other predator treatments. Only 5.5% of F. vespiformis individuals were found in flowers, confirming its preference for vegetative plant parts (Table 1). Table 1Total counts of Thrips parvispinus and released predators for its control on leaves and flowers, and percentage of their presence in flowers throughout the experimentSpeciesStageWeekWeek 1Week 2Week 3Week 4LeafFlower% FlowerLeafFlower% FlowerLeafFlower% FlowerLeafFlower% Flower T. parvispinus Adult22552.29700.067012715.995731925.0Larva900.0127200.031161.92522993.8 F. vespiformis Adult—————————3240.0Larva—————————3240.0 O. laevigatus Adult—————————2777.8Nymph—————————121250.0 A. swirskii All—————————7811.3SpeciesWeekWeek 5Week 6Week 7TotalLeafFlower% FlowerLeafFlower% FlowerLeafFlower% FlowerLeafFlower% Flower T. parvispinus 341754313.787045055.541231132.71819316128.129551033.492773043.2173080.5180765202.8 F. vespiformis 9218.26333.35700.07578.52800.08400.035410.315063.8 O. laevigatus 01100.0136483.131885.7189083.36810460.5292546.317621354.828535455.4 A. swirskii 10165.633900.0143100.0194970.4 Total counts of Thrips parvispinus and released predators for its control on leaves and flowers, and percentage of their presence in flowers throughout the experiment Predator abundance and interactions The abundance of F. vespiformis adults did not differ significantly between the single‐species treatment and the combined treatment, with similar numbers of adults occurring in both cases (Fig. 2(A)). However, larval abundance was higher in the combined treatment than in the single‐species treatment (Fig. 2(B)). O. laevigatus exhibited comparable population levels between treatments, suggesting compatibility with F. vespiformis (Fig. 2(C), (D)). By contrast, A. swirskii was considerably more abundant in the single‐species treatment than in the combined treatment (Fig. 2(E)), where its population declined rapidly and was nearly absent from the early stages of the experiment. Mean ± SE number of individuals recorded per week for each treatment of Franklinothrips vespiformis adults (A), F. vespiformis larvae (B), Orius laevigatus adults (C), O. laevigatus nymphs (D) and Amblyseius swirskii, during the Thrips parvispinus biological control experiment in the greenhouse. Different red letters indicate statistically significant differences between treatments (Generalized Linear Mixed Model: Wald test, P < 0.001). Plant productivity Biological control treatments had a significant impact on plant productivity. The control treatment, which lacked predator introductions, yielded the lowest number of harvested fruits per plant (Fig. 3(A)) (LMM: χ2 = 14.0, d.f. = 4, P = 0.007). Although the number of fruits tended to be higher in the O. laevigatus and combined treatments, these differences were not statistically significant when compared with the A. swirskii and F. vespiformis treatments. However, fresh plant weights differed significantly among treatments (LMM: χ2 = 190.6, d.f. = 4, P < 0.001). The highest plant biomass was recorded in the O. laevigatus and combined predator treatments, which significantly outperformed all other treatments (Fig. 3(B)). Significant differences were also detected in fresh fruit weight (LMM: χ 2 = 186.3, d.f. = 4, P < 0.001). The control treatment differed from all treatments except A. swirskii, whereas the combined treatment produced the highest fruit weight, followed by O. laevigatus and F. vespiformis (Fig. 3(C)). Productivity per plant (total fresh fruit weight per plant) was highest in the combined treatment (LMM: χ 2 = 136.4, d.f. = 4, P < 0.001), which did not differ from the O. laevigatus treatment (Fig. 3(D)). The lowest productivity was observed in the control, which did not differ from the A. swirskii treatment. Blocks oriented west exhibited higher fresh plant weight (237.7 ± 14.2 g versus 187.3 ± 13.3 g), fresh fruit weight (20.0 ± 0.5 g versus 13.5 ± 0.4 g) and total fresh fruit weight per plant (599.0 ± 56.3 g versus 375.2 ± 44.2 g) (LMM: FDR‐corrected multiple comparison, P < 0.05). No differences were observed between blocks in fruits per plants (LMM: likelihood ratio test, P = 0.243). Dry weights of plants, fruits, and total fruit per plant followed the same pattern as the fresh weights (Supporting Information, Fig. S2). Boxplots of number of pepper fruits per plant (A), fresh weight per plant (B),fresh weight per fruit (control, n = 301; Amblyseius swirskii, n = 323; Franklinothrips vespiformis, n = 317; Orius laevigatus, n = 395; combined, n = 396) (C) and total fresh fruit weight per plant (sum of the fresh weights of all harvested fruits from an individual plant) (D), during the greenhouse experiment conducted under controlled conditions comparing different biological control treatments against T. parvispinus. Black diamonds in the boxes indicate mean values for each treatment. Different letters denote statistically significant differences among treatments (Linear Mixed Model: false discovery rate‐corrected multiple comparison, P < 0.050). DISCUSSION This study provides the first evaluation of the predatory performance of O. laevigatus, A. swirskii, and F. vespiformis against the invasive pest T. parvispinus in greenhouse sweet pepper. Among the tested treatments, O. laevigatus, both alone and in combination with the other two predators, consistently achieved the greatest reduction in pest populations. The predatory mite A. swirskii and the predatory thrips F. vespiformis offered lower but comparable levels of suppression when applied individually. This pattern was further mirrored in the plant productivity data with higher fruit yields and biomass recorded in treatments involving O. laevigatus or the triple combination. Previous studies across Europe have consistently identified O. laevigatus as the most effective biocontrol agent against thrips (e.g., F. occidentalis) in greenhouse pepper crops. 15 , 16 , 30 , 31 , 32 This anthocorid predator, which also feeds on pollen, predominantly inhabits floral structures, a behavior that likely enhances its efficacy in controlling flower‐dwelling thrips species. 33 , 34 This entails a theoretical spatial mismatch between O. laevigatus and the leaf‐dwelling T. parvispinus. For this reason, it is essential to identify additional natural enemies with different functional traits, specifically those that forage on leaves rather than flowers. Consistent with these findings, in the current study, O. laevigatus was predominantly recorded from flowers (59.4% of individuals) (Table 1), thereby increasing the probability of predation on the 5.5% of T. parvispinus individuals located in flowers. By contrast, A. swirskii and F. vespiformis were primarily detected on foliage, with only 0.4% and 5.5% of individuals, respectively, recorded within floral tissues (Table 1). These results highlight O. laevigatus as a potential well‐suited predator for managing T. parvispinus in greenhouse pepper, despite a substantial proportion of its population being located in flowers. However, further research is needed on plant‐foraging predators with significant potential to control T. parvispinus and complement the control capacity of O. laevigatus. The number of natural enemies is a critical factor in the efficacy of biocontrol agents and, eventually, the success of biological control programs. Although F. vespiformis and O. laevigatus were released at equal densities, the latter is substantially larger, with individuals weighing 0.12 ± 0.01 mg (males) and 0.17 ± 0.04 mg (females) dry weight (n = 10 each), compared with 0.04 ± 0.011 mg for F. vespiformis females (n = 19), approximately four times lighter. This marked size difference may translate into higher per‐capita prey consumption and should be considered when adjusting release rates. Given that predatory effectiveness is closely linked to body size, this discrepancy could affect their relative impact. For instance, Mendoza et al. 13 found that a strain of O. laevigatus selected for larger body size, consumed 66% more thrips larvae and 145% more adult thrips compared with a standard sized population. Consistent with this, recent comparative trials with other generalist predators [Orius insidiosus (Say) and Anystis baccarum (L.)] demonstrated that per‐capita consumption strongly determines control success: a single adult female predator killed ~20 adult T. parvispinus in 24 h, and releases at 1 predator per 2 plants achieved ~80% suppression in greenhouse cage studies. 35 Laboratory and greenhouse studies with the predatory mite A. swirskii confirmed predation on first‐instar larvae of T. parvispinus and successful population establishment on potted sweet pepper. 36 Although this study did not explicitly quantify the prey‐to‐predator ratio, the observed population trends are consistent with scenarios of stronger control in treatments involving O. laevigatus. In this context, the ability of O. laevigatus to feed on both larval and adult thrips, including the reproductive stage, likely contributed to a faster population suppression compared with A. swirskii, which mainly targets immature stages. 36 These results support the idea that equal numerical release rates between species do not necessarily provide equivalent predation pressure, highlighting the need to adjust release numbers based on predator capacity and size. 36 As such, the lower efficacy of F. vespiformis registered in our experiment might be attributed to a lower size compared with O. laevigatus, although additional factors such as prey handling time, predator voracity and predator digestion time may also play an important role. Future experiments should consider adjusting the release density, especially of F. vespiformis, to account for size‐related differences among predators and eventually their impact on the pest population, as well as conducting laboratory studies to examine additional predator–prey interaction parameters. Differences in the developmental rates among predators further determined predator performance. Throughout the experiment, T. parvispinus exhibited rapid population growth. The developmental time of T. parvispinus is relatively fast and comparable with that of F. occidentalis. 1 At the time of predator introduction, the population of T. parvispinus was already high across all experimental compartments. Therefore, an additional factor that could contribute to the greater predation by O. laevigatus is the time required for completion of its life cycle. At 27 °C the life cycle of O. laevigatus (13–15 days) 37 is considerably shorter than that of F. vespiformis (~21 days), 38 , 39 enabling a faster population buildup. This was reflected in our trial, where O. laevigatus displayed two distinct reproductive peaks (Fig. 2(D)), in contrast to a single peak for F. vespiformis (Fig. 2(B)). Consequently, the average temperatures recorded during the greenhouse experiment (26.1 ± 0.5 °C, mean ± SE) were likely more favorable for O. laevigatus, facilitating its faster population development. O. laevigatus is a species typically associated with the Mediterranean region. 32 , 40 In Spain, it is widely distributed across the Iberian Peninsula. 41 By contrast, F. vespiformis exhibits a pantropical distribution, presumably originating from Central America, and has been introduced for biological control in several regions worldwide, including Europe, with recent records from the Canary Islands. 42 , 43 Thus, the biogeographic origin and ecological adaptation of O. laevigatus suggest a greater suitability to coastal Mediterranean conditions, which may partly explain the results observed in this study. The presence of F. vespiformis in the mixed treatment suggests compatibility with O. laevigatus, the other species present until the end of the experiment. By contrast, the decline of A. swirskii abundance throughout the experiment indicates incompatibility in releases with either F. vespiformis, O. laevigatus, or both. Several studies have documented intraguild predation (IGP) between O. laevigatus and A. swirskii, 13 , 44 particularly under conditions of low prey availability. 13 Despite these interactions, both predators are successfully used together in greenhouse pepper cultivation, 44 , 45 suggesting that IGP does not necessarily undermine biological control efficacy when managed appropriately. In our study, an additional factor magnifying the incompatibility of A. swirkii with the other two predators might be the relative smaller size of the plants or the reduced leaf surface compared with a commercial sweet pepper greenhouse. We suggest that in a fully established sweet pepper crop, the IGP interactions between A. swirskii and the other two predators will be somewhat diluted. Therefore, it remains to be studied whether there is IGP between F. vespiformis and A. swirskii, and if it could affect biological control efficacy. Interestingly, the presence of A. swirskii influenced T. parvispinus behavior, with the pest relocating from leaves to flowers. Assuming this effect also occurred in the combined treatment, in a system with more complex behavioral interactions, it would have resulted in an inadvertent increase in exposure to O. laevigatus, which is more active in floral tissues. However, A. swirskii experienced a population collapse in the combined treatment, suggesting that no reciprocal benefit took place. This form of unidirectional, trait‐mediated interaction may represent a previously underexplored mechanism of indirect facilitation in arthropod biological control systems, potentially acting as a long‐term trait‐mediated effect across generations. 46 , 47 Predatory species exhibit distinct preferences for the developmental stages of their prey, which may have also influenced predator effectiveness in our study. F. vespiformis is capable of consuming larvae, pupae, and adults of various thrips species, 42 and O. laevigatus has been reported to target both larval and adult stages, with a preference for juveniles. 48 Similarly, A. swirskii primarily preys on first‐instar larvae of T. parvispinus, and its eggs may be vulnerable to predation by T. parvispinus itself, 1 , 20 which could compromise its performance in biological control programs. These differences in stage‐specific predation may have influenced the efficacy of each predator, and therefore, further studies under controlled conditions are needed to quantify the predation capacity of these predators across the T. parvispinus developmental stages. Despite promising suppression by O. laevigatus either alone or in combination with the other predators, none of the treatments fully contained T. parvispinus populations, largely because of the pest's high reproductive capacity and the late timing of predator introduction. Together with empirical observations, our results suggest that releasing predators early in initial pest foci could potentially reduce the risk of subsequent pest outbreaks under greenhouse conditions. Targeting the leaf‐dwelling early stages of T. parvispinus, poorly controlled in this study, may be especially important to avoid outbreaks. Finally, although predator abundances did not differ between blocks, eastern blocks exhibited considerably higher populations of T. parvispinus, translating into reduced plant weight, fruit weight and total fruit weight per plant by the end of the experiment (Fig. 3; Supporting Information, Fig. S2). These differences may be partly explained by a small but consistent difference in temperature (26.8 ± 0.2 °C in the eastern blocks versus 25.4 ± 0.6 °C in the western blocks, mean ± SE). In southern Mediterranean greenhouse conditions, the daily trajectory of the sun generates greater diurnal variability in light exposure along the east–west axis than along the north–south axis, which likely contributed to the observed temperature differences. In addition, thrips show pronounced behavioral responses to light, and light‐dependent effects extending beyond behavior, such as variation in egg mortality, have also been documented. 49 Together, these behavioral and physiological responses to light may contribute to divergent population dynamics between eastern and western blocks. Also, studies conducted in commercial greenhouses suggest that external environmental factors and surrounding habitats can further influence predator–prey dynamics and pest control outcomes. 50 In conclusion, this study provides the first comparative assessment of three commercially available predators, O. laevigatus, F. vespiformis and A. swirskii, against T. parvispinus in greenhouse sweet pepper. Our results showed that O. laevigatus released alone or in combination, consistently achieved the highest suppression rates. Its effectiveness is attributed to a combination of biological traits, such as, a broad prey‐stage range, preferential use of floral structures, rapid life cycle, and strong adaptation to Mediterranean greenhouse conditions. F. vespiformis showed moderate efficacy attributable to the low release rates and appeared compatible with O. laevigatus in combined releases. A. swirskii, a cornerstone for IPM in sweet pepper, failed to establish in mixed treatments and likely suffered from IGP or competitive displacement most likely due to the relative low plant surface in our experiment. Nonetheless, O. laevigatus and A. swirskii have been successfully used for many years for controlling thrips and whiteflies in commercial sweet pepper greenhouses. 13 , 44 Future research should explore the potential of additional Franklinothrips species, such as F. megalops, that may be better adapted to Mediterranean greenhouse conditions, expanding the biocontrol toolbox. Moreover, adjustments the release dosages (taking into account predator body size, behavior, and ecological interactions) will be essential to maximize the effectiveness and sustainability of biological control programs targeting T. parvispinus. CONFLICT OF INTEREST The authors declare no conflict of interest. AUTHOR CONTRIBUTIONS ER: funding acquisition, conceptualization, experimental design, data collection, writing, editing. MP: conceptualization, experimental design, data analysis, writing, editing. AR: conceptualization, experimental design, data collection, editing. LL: conceptualization, experimental design, data collection, editing. MT: conceptualization, experimental design, data collection, editing. MM: data collection. CM: data collection. AP: conceptualization, experimental design, editing. Supporting information Figure S1. Layout of the experimental greenhouse. The schematic shows the interior of a single‐span greenhouse containing 20 thrips‐proof mesh cages (2 × 3 m each) arranged along a central corridor, with five cages per spatial block. Each cage contained six sweet pepper plants (Capsicum annuum). Treatments are indicated by cage labels: Control (C) only pest Thrips parvispinus, pest + Amblyseius swirskii (A), pest + Orius laevigatus (O), pest + Franklinothrips vespiformis (F), and pest + the combination of all three predators (Co). Arrows indicate cardinal directions; key structural elements (mesh front doors, central corridor, anteroom) are shown. Figure S2. Boxplots of (A) dry weight per plant, (B) dry weight per fruit (control: n = 301, A. swirskii: n = 323, F. vespiformis: n = 317, O. laevigatus: n = 395, combined: n = 396) and (C) total dry fruit weight per plant (sum of the dry weights of all harvested fruits from an individual plant), during the greenhouse experiment conducted under controlled conditions comparing different biological control treatments against T. parvispinus. Black diamonds within the boxes indicate mean values for each treatment. Different letters denote statistically significant differences among treatments (GLMM, FDR‐corrected multiple comparison, P < 0.050). Blocks oriented west exhibited higher dry plant weight (42.36 ± 2.56 vs. 33.11 ± 2.12; mean ± SE), dry fruit weight (1.61 ± 0.04 vs. 1.11 ± 0.03) and total dry fruit weight per plant (48.22 ± 4.59 vs. 31.3 ± 3.52) (GLMM: FDR‐corrected multiple comparison, P < 0.05).