Predators and Parasitoids-in-First: From Inundative Releases to Preventative Biological Control in Greenhouse Crops

Repeated mass introductions of natural enemies have been widely used as a biological control strategy in greenhouse systems when the resident population of natural enemies is insufficient to suppress the pests. As an alternative strategy, supporting the establishment and population development of beneficials can be more effective and economical. The preventative establishment of predators and parasitoids, before the arrival of pests, has become a key element to the success of biological control programs. This “Predators and parasitoids-in-first” strategy is used both in Inoculative Biological Control (IBC), and in Conservation Biological Control (CBC). Here, we provide an overview of tools used to boost resident populations of biocontrol agents.


INTRODUCTION
Biological control in greenhouses dates back almost 100 years, when Speyer (1927) at the Cheshunt Experimental Station first described the control of greenhouse whitefly Trialeurodes vaporariorum (Westwood) by the specialist parasitoid Encarsia formosa Gahan . This example was followed in the fifties by the use of the natural enemies against mealybugs (Doutt, 1951) and in the sixties, by the introductions of the specialist predatory mite Phytoseiulus persimilis Athias-Henriot for the control of spider mites (Bravenboer and Dosse, 1962). Biological control in greenhouses has since been extended by the addition of generalist biocontrol agents to complement the specialist beneficials (Janssen and Sabelis, 2015). The release of generalists, that can feed on a range of prey, but may also exploit non-prey food, made it possible to maintain populations of natural enemies in crops in absence of the target pest, thus facilitating the preventative use of natural enemies.
Thirty years ago, Ramakers (1990) proposed the concept of "predator-in-first" and stated that the availability of supplementary foods, such as pollen and nectar, is essential for early establishment of generalist natural enemies. However, many cropping systems lack these floral resources . As a result, most biological control programmes rely on inundative release strategies, where natural enemies are periodically introduced in large numbers to control pest problems (Stinner, 1977;Collier and Van Steenwyk, 2004). The pest control in inundative strategies often relies on the released individuals, rather than their progeny (De Bach, 1964). In ornamentals especially, cheap predators and parasitoids are released weekly in crops without necessarily accomplishing establishment (Hoddle et al., 1997(Hoddle et al., , 1998Buitenhuis et al., 2014Buitenhuis et al., , 2015. Manual and automatized blowers of beneficials have been specially designed for that purpose (Van Schelt et al., 2008). Inundative biological control strategies have among others the drawback to exclude more clostly, but more efficient natural enemies, like predatory bugs.
In recent years, there has been an increased interest in strategies to allow a preventative establishment of natural enemies (standing army) Pijnakker et al., 2017). Both Inoculative Biological Control (IBC), which aims at establishing mass-reared natural enemies, and Conservation Biological Control (CBC), that seeks to conserve or enhance naturally occurring biocontrol organisms, can benefit by providing natural enemies with missing resources (Hagen, 1986;Zemek and Prenerová, 1997;Wäckers, 2005). In biological control programs substantial advancements have been made in the use of factitious prey, pollen (extrafloral) nectar, and honeydew as food supplements (Van Rijn et al., 2002;Lundgren, 2009;Messelink et al., 2014). In addition, some predators can feed on plant tissues, which facilitates their establishment in periods of prey scarcity or in the absence of prey (Eubanks and Denno, 1999;Lucas and Alomar, 2001;Pappas et al., 2017). Besides the role of non-prey food, establishment of predators can also be supported through the provisioning of additional non-food resources, like shelter and oviposition sites Pekas and Wäckers, 2017). The advances in the use of supplementary resources to support early establishment of natural enemies in greenhouse crops are the focus of this review. Microbial control is not developed in this review, as preventative use of insect-pathogenic and antagonistic fungi is complex, there are few studies and results are inconsistent (Elliot et al., 2000).

PLANT-TISSUE FEEDING
While plant feeding by omnivorous biocontrol organisms can potentially cause plant damage (see below), using tissue-feeding omnivores as biological control agents has many advantages. They have been traditionally underestimated in inundative release biocontrol strategies (Castañé et al., 2011); however, several recent studies emphasize its importance (Coll and Guershon, 2002;Eubanks and Styrsky, 2005;Castañé et al., 2011;Pappas et al., 2017). Plant-tissue feeding plays a major role in the survival of several omnivores, as it has been shown to occur broadly in heteropterans (Ridgway and Jones, 1968;Naranjo and Gibson, 1996), as well as in a number of phytoseiid mites (Tanigoshi et al., 1993). In Heteroptera it provides nutrients that are essential to successful development (Gillespie and McGregor, 2000;Sinia et al., 2004) and may help them persist in periods of drought. Plant-tissue feeding is also assumed to provide some crucial resources that facilitate prey consumption. When consuming prey, some predatory Heteroptera require a source of water to dilute the digestive enzymes they inject into their prey (Cohen, 1985). Plant-tissue feeding may allow the Heteroptera to balance nutrients, proteins, carbohydrates, vitamins and minerals that would otherwise be restricted in a carnivorous diet (Polis et al., 1989;Coll, 1998). Particular plant species allow a full development of omnivorous Heteroptera in absence of prey. As omnivores can rely on tissue feeding, the risk of dying or leaving the crop at low prey densities is probably limited (Crawley, 1975;Pimm and Lawton, 1978), but will depend on the plant species.
While the induction of plant resistance mechanisms in response to herbivore feeding has been widely studied, few studies have addressed the effect of plant-tissue feeding by omnivore natural enemies in terms of plant defense induction (Stout et al., 1997;Agrawal et al., 1999;Agrawal and Klein, 2000;Agrawal, 2005a,b;Pappas et al., 2017). Induced plant resistance mechanisms include the production of secondary metabolites, part of which are released as volatile chemicals (Herbivoreinduced plant volatiles, HIPVs) (Paré and Tumlinson, 1999). HIPV's play an important role in protecting the damage sites against entry by pathogens. The induced change in plant chemistry can reduce plant attractiveness to herbivores, as well as herbivore performance (Turlings et al., 1990, Bolter et al., 1997Karban and Baldwin, 1997;De Moraes et al., 2001;Kalberer et al., 2001;Wakefield et al., 2005), thus representing an important example of direct plant defense. An indirect defense mechanism is involved when HIPV's are used by the herbivore's natural enemies to locate their prey/hosts (Turlings and Wäckers, 2004). In addition, plant-tissue feeding also elicits the production of extrafloral nectar, an indirect defensive trait which allows plants to recruit ants and other nectar feeding omnivores, which in turn protect the plants by attacking the herbivores (Wäckers and Bonifay, 2004;Kost and Heil, 2005). Several studies have now demonstrated that plant-tissue feeding by predators also activates plant defense mechanisms (Pérez-Hedo et al., 2015a,b;Naselli et al., 2016;Pappas et al., 2016;Zhang et al., 2018). Plants with activated defense systems are less attractive to the tobacco whitefly Bemisia tabaci (Gennadius), but more attractive to the whitefly parasitoid E. formosa. Pappas et al. (2015) showed that the zoophytophagous predator Macrolophus pygmaeus Rambur induces defense of tomato plants, making them less susceptible to the two-spotted spider mite Tetranychus urticae Koch, but without affecting the greenhouse whitefly T. vaporariorum. Zhang et al. (2018) demonstrated that T. urticae and Western flower thrips Frankliniella occidentalis (Pergande) laid fewer eggs on sweet pepper plants previously inoculated with M. pygmaeus. As this also applied to newly produced leaves, which were not directly exposed to the omnivore, this suggests that the induced plant response is systemic. The development time of F. occidentalis larvae feeding on leaves previously exposed to M. pygmaeus was also prolonged.
The introduction of omnivores as biological control agents can create complex interactions. The ability of omnivores to feed on multiple trophic levels may not improve biological control. The possible benefits of plant-tissue feeding are omnivore specific and dependent on the developmental stage, prey availability and plant nutritional composition (Naranjo and Gibson, 1996). The complexity of food choice by omnivores remains poorly understood: in some cases, prey consumption is reduced when both prey and plant diets are available (Crum et al., 1998, Kiman and Yeargan, 1985. Feeding on highquality plant food may provide a highly nutritious preferred food source and decrease the consumption of a particular prey species (Abrams, 1987). Omnivorous bugs consumed fewer prey on plants presenting lima bean pods in a study by Eubanks and Denno (1999). Plant-tissue feeding often facilitates survival rather than reproduction probably because of low nitrogen contents (De Clercq and Degheele, 1992). Population dynamics can be strongly influenced by the developmental stage (Coll and Guershon, 2002), the period of phytophagy of the omnivores (Cisneros and Rosenheim, 1997) or by the competition for plant food between omnivores and prey (Polis and Holt, 1992;Coll and Izraylevich, 1997).
Concerning the impact of induced changes in host plant chemistry on pests and beneficials, Ode (2006) underlined that this aspect has been insufficiently explored and reviewed negative tritrophic effects of inducible plant defenses on natural enemies. In a study by Agrawal et al. (2002), predatory mites were less attracted to plants that produced cucurbitacins than cucurbitacin-free plants and had a reduced fecundity when feeding on herbivores that feed on defended plants when compared to those that fed on plants free of cucurbitacins. Induced plant resistance is thus not always favorable to biological control.
Using phytophagous beneficials also has the consequence that their plant feeding exposes them to systemic pesticides (Coll, 1998;Krischik, 1999, Arnó andPrabhaker et al., 2011;Put et al., 2015). Plant feeding by omnivores can also result in crop damage or reduced crop growth, in particular at high omnivore populations. Omnivores can cause direct mechanical feeding injuries, injuries to plant vascular tissues or damage through the salivary enzymes killing plant cells (Castañé et al., 2011). Plant-tissue feeding heteropterans can also disturb plant hormonal balances . In tomatoes, fruit damage by heteropterans is often reported, reflecting a preference for the more nutritious tomato fruit (Salamero et al., 1987;Alomar et al., 1991;Lucas and Alomar, 2002;Albajes et al., 2006;Castañé et al., 2011). Nesidiocoris tenuis (Reuter) is known to cause injuries on the aerial parts of tomato plants (necrotic rings on stems, shoots, leaf petioles and flower stalks), leading to flowers and fruits abortion, and, reduced growth (Arnó et al., 2010). Gillespie et al. (2007) reported damage on gerbera flowers, Castañé et al. (2003) and Sengonça et al. (2003) on cucumber and zucchini fruit.
Despite the above negative aspects, omnivores are crucial elements in biological control strategies, especially mirids in tomato crops and anthocorids in sweet pepper crops. Current biological control programmes are supported with different tools like smart-phone applications to register and follow crop injuries and sticky traps to monitor omnivores and the pests. This allows growers to maximize benefits of omnivores and avoid risks. Omnivorous predators are commonly used in greenhouse crops and their establishment is even stimulated by provision of alternative food (Lenfant et al., 2000;Castañé et al., 2006;Put et al., 2012;Moerkens et al., 2017;Brenard et al., 2019;Sade et al., 2019). Growers try to avoid applications of systemic pesticides, which are not compatible with omnivores. Resistance breeding does not yet take plant suitability for omnivores into account, but this might change as we gain further insights in the complex interactions involving omnivores.

PEST FEEDING (PEST-IN-FIRST)
The "pest-in-first" (PIF) strategy is one of the oldest strategies to allow the establishment of a biocontrol "standing army" in greenhouses. Here an early introduction of natural enemies is combined with a controlled (pre-) release of the pest. This concept can also be used to allow early establishment of specialist natural enemies, which cannot be supported by factitious prey or pollen. One of the first examples of a successful PIF strategy is the release of two-spotted spider mites to reinforce the establishment of the predatory mite P. persimilis (Hussey et al., 1965;Gould et al., 1969;Markkula and Tiittanen, 1976;Havelka and Kindlmann, 1984;Waite, 2001;Bolckmans and Tetteroo, 2002). Other pest-in-first strategies have been evaluated, such as the introduction of low numbers of greenhouse whitefly T. vaporariorum in tomato followed by timed releases of its parasitoid E. formosa (Parr et al., 1976). Growers, however, are typically reluctant to release pests due to the risks of causing crop damage (Parr et al., 1976;Stacey, 1977, Starý, 1993. Instead, they prefer to wait until the pest develops naturally before introducing biocontrol, as releasing P. persimilis in naturally occurring T. urticae hot spots is often cheaper. Alternatively, they opt for calendar introductions of biocontrol agents.

FACTITIOUS PREY
To allow early establishment of generalist predators, growers can use factitious prey (i.e., foods which the predators usually do not encounter in their natural habitat). Some factitious prey, such as eggs of the Mediterranean flour moth, Ephestia kuehniella Zeller, decapsulated cysts of the brine shrimp Artemia spp. and astigmatid mites Midthassel et al., 2013;Nguyen et al., 2014a;Delisle et al., 2015a;Labbé et al., 2018) can be excellent food sources for a wide range of generalist predators. Studies on these supplemental foods are summarized in Table 1. Some of these factitious prey, especially E. kuehniella eggs and astigmatid prey mites, are also used in the commercial production of biological control agents.

Astigmatid Prey Mites
In greenhouses, breeding sachets of Neoseiulus cucumeris Athias-Henriot have been developed to allow slow releases of predatory mites in the crop and thus reduce handling costs (Sampson, 1998). These rearing systems consist of predators, astigmatid mites as food and carrier material. They allow for a release of predators for periods up to four (sometimes even eight) weeks. In crops like roses or potted plants, that do not feature pollen, astigmatid prey mites do not establish on the plant, and where pests cannot be tolerated, predatory mite populations cannot build up on the crop and sachets need to be renewed regularly.
In potted plants, the spread of the predatory mites released from the breeding sachet is limited because plants are widely spaced; the majority of the predators remain on the plant, which received the sachet . To tackle these problems, strategies like using one (small) sachet per plant have been developed to provide each plant an open rearing system (Valentin, 2017) or predators and prey are blown over the crop with purpose built automatized blowers. Supplying predatory mites with factitious (astigmatid) prey mites as food has recently gained popularity among growers. The following feeding mites could be used to improve the reproduction and survival of predatory mites: Tyrophagus putrescentiae (Schrank) (Pirayeshfar et al., 2020), Carpoglyphus lactis (L.) (Nguyen et al., 2013), Thyreophagus entomophagus (Laboulbene), Suidasia medanensis (Oudemans) (Sánchez et al., 2019), and Aleuroglyphus ovatus (Troupeau) (Xia et al., 2012;Ferrero et al., 2016;Rueda-Ramírez et al., 2018). However, very few trials have been performed at plant level Vila et al., 2017;Pirayeshfar et al., 2020) and studies on the actual effect of these strategies on pest control are limited. Pirayeshfar et al. (2020) succeeded in increasing Amblyseius swirskii Athias-Henriot numbers on chrysanthemum plants by providing living T. putrescentiae per plant, but failed to do so using frozen T. putrescentiae. Pirayeshfar et al. (2020) suggested that the different diets used to rear the prey mites may have an influence on the population increase of the predatory mites. Nowadays, some growers disperse the astigmatid mites C. lactis or T. entomophagus to supplement the released predatory mites with food. Ferrero et al. (2016) developed a gel product protecting astigmatid mites' eggs when they are dispersed, and increasing their shell-life on the crops. This product resulted in high population levels of A. swirskii on cucumber plants and a better biological control of sweet potato whitefly than with the use of breeding sachets. Using astigmatid mites as food supplement has the advantage that they are relatively cheap (Ramakers and van Lieburg, 1982;Castagnoli, 1989). The main drawback of mass application of astigmatid mites lies in the fact that they can cause health issues for users including dermatitis, allergies and anaphylaxis (Iglesias-Souto et al., 2009;Fernández-Caldas et al., 2014;Liu and Zhang, 2017;Mullen and O'Connor, 2019;Pirayeshfar et al., 2020). Furthermore, Tyrophagus spp. can cause plant damage (Czaikowska et al., 1988;Buxton, 1989;Fischer, 1993;Fan and Zhang, 2007;Yasukawa et al., 2011). Muñoz-Cárdenas et al. (2017) proposed another approach using astigmatid mites in combination with mulch layers to provide food to crop-inhabiting predatory mites. Prey mites were introduced in the litter to stimulate the predatory mite A. swirskii on aboveground plant parts. In cage experiments with rose plants this increased predatory mite numbers 4-fold, as compared to control plants. Thrips control was increased and leaf-and flower damage was reduced as a result. It was shown that A. swirskii, which is usually considered to be a leaf-inhabiting species, actually moved between rose foliage and mulch to feed on the astigmatid mites. Similarly, Grosman et al. (2014) increased biocontrol of thrips and whiteflies with A. swirskii using mulch layers in different ornamental crops. Adding mulches with, for example bran, yeast or Biotop R , waste product of the potato industry (Grosman and de Groot, 2011), organic matter (Settle et al., 1996;Neves Esteca et al., 2020), animal manure (Navarro- Campos et al., 2012) to boost fungi, decomposers and plankton feeders can increase predators' population levels. Despite promising results in small-scale experiments, this strategy has not been adopted by growers. This is probably due to the increased labor when mulching, and the fact that the system is often effective on the short term only. Grosman et al. (2014) found that this approach produced predators for up to 6 weeks, which is comparable to the longevity of breeding sachets. These methods carry also the risk that the predators switch from feeding on herbivores to soil organisms (apparent mutualism, Holt, 1977) as reported by Birkhofer et al. (2008), or, are out-competed by other soil predators, that also benefit from the substrate manipulation Van Holstein-Saj, 2007, 2011).

Mediterranean Flour Moth Eggs
To support establishment of predatory bugs, growers mainly use eggs of the Mediterranean flour moth E. kuehniella. The high nutritional value of E. kuehniella eggs ensures development, reproduction and survival of many arthropods ( Table 1). In M. pygmaeus, higher reproduction rates were found when the mirids were fed on the moth eggs, as compared to whitefly pupae (Fauvel et al., 1987;Alomar et al., 2006). Sprinkling of E. kuehniella eggs on crop plants was the first supplemental food strategy widely adopted by growers on a large scale. They are typically used on the points where predatory bugs Orius spp. (sweet pepper) and mirid bugs M. pygmaeus, N. tenuis, and Tupiocoris cucurbitaceus (Spinola) (mainly tomato) are introduced (Put et al., 2012;Moerkens et al., 2017;Brenard et al., 2018).
The main factor limiting the use of E. kuehniella eggs is their cost, with prices of 400 EUR/kg . They also need to be kept frozen. Furthermore, the eggs tend to dry out once applied on the crop, or when the relative humidity is too high, they become moldy on the leaves. Due to their cost, E. kuehniella eggs are not used to feed predatory mites. However, this food source was found to be suitable for oviposition of Iphiseius degenerans (Berlese) (Vantornhout et al., 2004), A. swirski and Amblydromalus limonicus Garman and McGregor Vangansbeke et al., 2014c), Gaeolaelaps aculeifer Canestrini and Stratiolaelaps scimitus (Womersley) (Navarro- Campos et al., 2016), as well as for the pest F. occidentalis. The performance of predatory mites on E. kuehniella eggs can vary (Vangansbeke et al., 2014c;Leman and Messelink, 2015) depending on egg storage conditions and ambient humidity in the crop. Liu and Zhang (2017) observed that immatures of A. limonicus exhibited difficulties to pierce the chorion of E. kuehniella eggs that hardened at low ambient humidity.

Cysts of the Brine Shrimp
Several cheaper options have been investigated to replace the use of Ephestia moth eggs in commercial production as well as in field applications ( Table 1). Out of the options tested, the most promising substitute of flour moth eggs for feeding generalist predators was found to be dry cysts of the brine shrimp Artemia spp. Having been used widely as fish food, Artemia cysts have the advantage that they can be stored for years in dry form, and do not require freezing as required for E. kuehniella eggs (Arijs and De Clercq, 2001). They keep their nutritional value longer than E. kuehniella eggs when applied on crops (De Clercq et al., 2005a;Messelink et al., 2016;Moerkens et al., 2017), they do not become moldy on plants (Vandekerkhove et al., 2009) and are, depending on the quality, up to 30 times cheaper than flour moth eggs . However, Artemia cysts vary substantially in nutritional quality, and the high quality product is roughly the same price as E. kuehniella eggs. When used at high densities, Artemia cysts can leave a "fishy smell" on the crop. Brine shrimp cysts have been tested as prey for several natural enemies for production purposes ( Table 1).
In field crops, Hoogerbrugge et al. (2008) and Leman and Messelink (2015) found either no or poor establishment of A. swirskii when fed with a commercial strain of Artemia sp. cysts alone in a chrysanthemum crop. In contrast, Vantornhout et al. (2004), Nguyen et al. (2014a) and Vangansbeke et al. (2014cVangansbeke et al. ( , 2016b showed in laboratory studies complete development of, respectively, I. degenerans, A. swirskii and A. limonicus on a diet of decapsulated cysts of a non-commercial Artemia franciscana Kellogg strain. Vangansbeke et al. (2016a) succeeded in establishing A. swirskii on chrysanthemum and ivy plants using this Artemia strain, but not when using a commercial decapsulated Artemia cyst product. Inconsistent results between studies might be explained by the incomplete decapsulation of the cysts or their level of hydration  and by the substantial variation in Artemia product quality. De Clercq et al. (2005a) showed differences of composition and nutritional quality between Artemia cysts of diverse origins, which can have an impact on the nutritional value as well (Bloemhard et al., 2018;Sade et al., 2019). Overall, Artemia cysts have become a valuable complement in biological control programs in greenhouse vegetable crops, as most tomato and sweet pepper growers release M. pygmaeus with this alternative food, either by itself or in combination with E. kuehniella.
The provision of brine schrimp cysts and Mediterranean flour moth eggs is now a common practice on introduction points of predatory bugs in vegetable crops. As astigmatid prey mites are concerned, only ornamental growers release them weekly or biweekly, but generally additionally to predatory mites. They tend to use them within their inundative release strategies of predatory mites instead of seeking for an early establishment of predators after a few release. Besides their use to stimulate predatory mites, astigmatid mites might also hold potential to support field populations of predatory bugs, as they were found to be a suitable food source for Orius spp. (El-Husseiniak and Sermann, 1992;Husseini et al., 1993;Nagai et al., 1998;Gomaa and Agamy, 2002;Yang et al., 2009;Bernardo et al., 2017;Bonte et al., 2017;Song et al., 2018). This application is still at an experimental stage at growers.

BANKER PLANT SYSTEMS
Supplementation of food resources and oviposition places for natural enemies can be done by providing secondary plants, so-called "banker plants" or "open rearing systems" (Bennison, 1992;Bennison and Corless, 1993). The principle of the banker plant system is the use of plants, usually different from the crop, to provide beneficials with alternative (non-pest) prey and/or plant-provided food resources. This method was developed for the introduction of parasitoid wasps to control aphids (Starý, 1969;Lyon, 1973) and whiteflies (Stacey, 1977). The banker plant method is often seen as a further development of the "Pest-in-first" method in greenhouses as the initial banker plant systems introduced additional crop plants carrying the target pest ( Table 2) (Parr and Stacey, 1975;Stacey, 1977).
The advantages of using banker plants have been widely described. They aid survival, reproduction and development of naturally occurring or introduced natural enemies even in absence of pests (Pratt and Croft, 2000). When introducing beneficials, they can be released onto the banker plants where they find essential resources for survival and reproduction. Once the target pest appears, they then move from the banker plants into the crop. Successful banker plant systems allow for early season augmentation of beneficials and can replace "repeated inundative releases" (Hansen, 1983) thereby reducing costs (Huang et al., 2011). Furthermore, when using potted banker systems, these banker plants can be moved for "hot spot treatment" of pest colonies (Ramakers and Voet, 1995). When selecting banker plants, one should consider the natural enemies' affinity for the plant, as well as the plant's capacity to carry suitable alternative prey or other (food) resources (Jacobson and Croft, 1998;Goolsby and Ciomperlik, 1999). Nutritional, allelochemical and plant morphological traits (Price et al., 1980;Grevstad andKlepetka, 1992, Desneux andRamirez-Romero, 2009) are taken into account when selecting candidate banker plant-prey combinations. It is known that morphological plant characteristics (e.g., toughness of leaves and stems, number of nectar glands, flowering period, number of flowers, plant pubescence, acarodomatia, trichomes) can be correlated with the dispersal, oviposition, developmental and reproductive success of arthropods (Walter and O'Dowd, 1992a,b;Pfannenstiel and Yeargan, 1998;Brodeur, 1999, Lundgren et al., 2008;Parolin et al., 2012a). In addition, banker plants must be capable to survive temperatures and light conditions as the ones used in greenhouse production (Van der Linden, 1992).
Banker plant systems can be divided into two groups: (1) plants providing non-pest prey/host (2) plants producing nonprey food sources (pollen and nectar). Diverse publications provide complete inventories of the banker plant systems and their potential Frank, 2010;Huang et al., 2011;Ying et al., 2012;Miller et al., 2017;Miller, 2018;Payton Miller and Rebek, 2018). An overview of banker plant systems is given in Tables 2-4.

Syrphids
Episyrphus sp. Fischer, 1997 Greater celandine Chelidonium majus L.   Bennison, 1992;Lamparter, 1992;Albert, 1995;Conte, 1998;Jacobson and Croft, 1998;Van Schelt, 1999;Nagasaka and Oya, 2003;Saito, 2005Nagasaka et al., 2010Jandricic et al., 2014;McClure, 2014;McClure and Frank, 2015;Miller et al., 2017;Miller, 2018;Payton Miller and Rebek, 2018 Wheat Triticum aestivum L. the system, due to inconsistent efficacy, labor (handling and maintenance), sink effects, and/or issues with hyperparasitoids (Jacobson and Croft, 1998;Van Driesche et al., 2008;McClure and Frank, 2015;Payton Miller and Rebek, 2018). Furthermore, the parasitoid species reared on banker plant systems are not necessarily efficient against all occurring target pest species, such as the potato aphid, Macrosiphum euphorbiae (Thomas), and the foxglove aphid, Aulacorthum solani (Kaltenbach) (Van Driesche et al., 2008;Nagasaka et al., 2010;Prado et al., 2015). Development of parasitoid wasps (Jandricic et al., 2014) on the banker plant systems may be insufficient due to the provided aphids being of insufficient size. This can lead to a reduction of survival and male biased sex ratio of the parasitoids (Hoddle et al., 1998;Chau and Mackauer, 2001;Henry et al., 2005). Gall midges produced on aphid species of poor nutritional value can also suffer in terms of size and fecundity (Kuo-Sell, 1989). Natural enemies may be reluctant to switch from aphids on the banker plant to the crop aphids (Lester et al., 2000;Coyle et al., 2011) which can hamper the establishment of beneficials on the crop. This can be due to an acquired adaptation to the banker plant aphid through associative learning processes (Hoddle et al., 1998, Keasar et al., 2001, Ode et al., 2005, Prado and Frank, 2014. For all these reasons, the use of non-pest prey banker plant systems is relatively limited, relative to other methods supporting preventative establishment.

Banker Plant Systems (Nectar and Pollen)
Natural enemies can also be boosted by banker plants providing pollen and nectar (     Grafius, 1995, 1996;Johanowicz and Mitchell, 2000 Yellow rocket Barbarea vulgaris R. Br. Parasitoids Diadegma insulare (Cresson) Grafius, 1995, 1996 were scarce (Ramakers and Voet, 1995). However, this banker plant requires a lot of maintenance due to its rapid growth and the predatory mites do not always move into the crop. Additionally, castor bean can become a source of certain pests. The usefulness of castor bean plants was also limited in sweet pepper crops, as pepper produces pollen and floral nectar. Only a limited number of ornamental growers is still using castor bean plants. Many banker plant systems were developed to maintain predatory bugs, especially the anthocorid Orius spp., primarily used for the control of thrips. Establishing this predator requires prey, and a range of plant derived food, including pollen (Kiman and Yeargan, 1985;Richards and Schmidt, 1996;Corey et al., 1998), nectar (Yokoyama, 1978), and plant sap (Kiman and Yeargan, 1985;Richards and Schmidt, 1996;Lundgren et al., 2008). In addition, the plant structures need to be acceptable as oviposition substrate (Lundgren et al., 2008). The predator establishes easily in sweet pepper crops where it utilizes the floral resources of the pepper plants. However, most ornamental crops lack these resources. This hampers establishment of beneficials as well as the short crop cycle of many ornamentals. As repeated introductions of Orius spp. are too onerous and expensive for growers, a range of banker plant systems have been designed to support establishment of the predator. Several studies have used ornamental peppers "Black pearl" or "Purple Flash" and the perennial Sweet Alyssum (Valentin, 2011;Wong and Frank, 2013). The use of Sweet Alyssum with its long lasting flowering period results not only in higher densities of Orius spp. (Picó and Retana, 2000;Alomar et al., 2008;Bennison et al., 2011;Hogg et al., 2011;Pumariño and Alomar, 2014), but also benefits hoverflies, predatory Heteroptera (Pease and Zalom, 2010) and several parasitoids (Johanowicz and Mitchell, 2000;Berndt and Wratten, 2005;Begum et al., 2006). However, the adoption of this method has been limited due to the fact that Sweet Alyssum is also exploited by pollen-feeding pest species, such as thrips. Several banker plant systems have also been designed to attract and sustain reproductive populations of predatory mirids.  Table 4. Addition of factitious prey on banker plants, as discussed earlier, is often suggested to increase survival, longevity and fecundity of the predatory bugs Alomar, 2012, 2014). The complexity of banker plant systems in greenhouse crops limited their adoption. Since the last decade, more straightforward methods have been developed to support the establishment and retention of beneficials in greenhouse. However, some sweet pepper growers still use them against aphids before the appearance of hyperparasitoids and there is an increased interest for banker plants to maintain predatory bugs in ornamental crops. Few attempts were made to use cover crops in or next to greenhouses and to grow combined crops, alternating rows of tomato, sweet pepper and egg plants in one greenhouse (Janmaat et al., 2014) to benefit from the vegetational diversity (Letourneau, 1983(Letourneau, , 1990. Despite good technical results, these strategies failed in practice because of the complexity of their management. In open fields, we recently see an increased interest for the use of banker plant systems.

POLLEN
Providing pollen as a protein rich food in crops is another approach to boost establishment of pollen-feeding natural enemies, especially predatory mites. Pollen has been used extensively to support populations of generalist predatory mites (McMurtry and Scriven, 1966;Kennett et al., 1979;Sabelis, 1990, 1993;Van Rijn and Tanigoshi, 1999a;Messelink et al., 2009;Nomikou et al., 2010;Hoogerbrugge et al., 2011;Maoz et al., 2014;Ranabhat et al., 2014, Vangansbeke et al., 2016a. In greenhouse crops, bee-collected pollen has been tested in a number of studies, either as dry pellets or in suspension (Ramakers, 1995;Kolokytha et al., 2011;Goleva and Zebitz, 2013;Montserrat et al., 2013;Duarte et al., 2015). However, as the grains absorb water and become moldy, bee pollen was judged inappropriate by growers for crop applications (Ramakers, 1995). Attempts of using pollen providing banker plant systems, like castor beans (Ramakers and Voet, 1995;Van Rijn and Tanigoshi, 1999b), also failed, for reasons described previously. Pollen started to be used on a large scale by growers in July 2013, when commercial supplements based on narrowleaf cattail pollen were made available for crops where pollen sources are lacking (Pijnakker et al., 2014). This allowed preventative establishment of predatory mites in crops like cucumbers (which are parthenocarp and do not produce pollen), as well as ornamentals where no pests are tolerated.

Predatory mites
Neoseiulus cucumeris (Oudemans); Amblyseius swirskii (Athias-Henriot) Ranabhat et al., 2014;Goleva et al., 2015 Turkish pine Pinus brutia (Ten.) Predatory mites Amblyseius swirskii (Athias-Henriot) Kütük and Yigit, 2011 (Continued) Frontiers in Sustainable Food Systems | www.frontiersin.org  McMurtry and Scriven, 1964;Engel andOhnesorge, 1994 Broufas andKoveos, 2000;Papadopoulos and Papadoulis, 2008;Kolokytha et al., 2011;Rezaie and Askarieh, 2016;Rezaie, 2017;Fadaei et al., 2018 Willow Salix sp. Coccinellids Adalia bipunctata L. Blackman, 1967Blackman, 1996Cocuzza et al., 1997;Coll, 1998;Lundgren, 2009). However, the development of Orius spp. was prolonged and reproduction and survival reduced on pollen only diet (Salas-Aguilar and Ehler, 1977;Kiman and Yeargan, 1985;Funao and Yoshiyasu, 1995;Richards and Schmidt, 1996;Cocuzza et al., 1997;Vacante et al., 1997,?). As with the phytoseiids, Orius performance is dependent on pollen type . Pollen feeding also provides benefits for M. pygmaeus (Perdikis and Lykouressis, 2000;Vandekerkhove and De Clercq, 2010;Portillo et al., 2012;Put et al., 2012). Cattail pollen doubled the longevity of M. pygmaeus females compared to bugs provided only broad beans (Portillo et al., 2012) and promoted its establishment on tomato plants (Put et al., 2012). Also larvae of green lacewings can benefit from (bee) pollen (Patt et al., 2003). For syrphids, sexual maturation requires proteins that can be provided by pollen, allowing the females to mature successive batches of eggs (Schneider, 1948;Pineda and Marcos-García, 2008). In biological control, pollen has been reported to support population densities of natural enemies when prey densities are low and improve control efficacy against pests. Most of these studies concern predatory mites. In the presence of pollen, predatory mites provided better control of phytophagous mites (McMurtry and Scriven, 1966;Kennett et al., 1979;Ferragut et al., 1987;Pina et al., 2012Pina et al., , 2015Saber, 2013;Maoz et al., 2014;Duarte et al., 2015;Pijnakker et al., 2016), thrips (Ramakers, 1990;Van Rijn and Sabelis, 1993;Van Rijn et al., 1999) and whiteflies (Nomikou et al., 2001(Nomikou et al., , 2002(Nomikou et al., , 2010Nomikou, 2003). Van Rijn et al. (1999) were the first to provide clear experimental evidence that supplementing pollen can be an efficient strategy to boost the biological control potential of predatory mites. By supplying the predatory mite species A. limonicus and I. degenerans with cattail pollen they were able to enhance population levels of the predatory mites, which resulted in increased thrips control and reduced plant damage. Kütük and Yigit (2011) succeeded in pre-establishing A. swirskii on sweet pepper by spraying suspensions of pine pollen, thereby maintaining F. occidentalisnumbers under the desired threshold. Provision of cattail or corn pollen increased densities of A. swirskii and improved thrips control on potted chrysanthemum (Leman and Messelink, 2015). Muñoz-Cárdenas et al. (2017) succeeded in establishing A. swirskii in roses before thrips release by weekly provision of cattail pollen and thus managed to realize a decrease of thrips numbers and damage. Nomikou et al. (2002Nomikou et al. ( , 2010 showed improved suppression of tobacco whitefly populations by A. swirskii on single cucumber plants treated with cattail pollen. Skirvin et al. (2006) achieved a higher density of N. cucumeris on chrysanthemum plants with an application of castor bean pollen. Still, it led to an increased infestation of western flower thrips, as only a few predatory mites were recovered. These examples give an idea of the potential of pollen as alternative food but the studies were often performed over short temporal scales and at pest densities higher than those found in commercial greenhouses. Since cattail pollen has been made commercially available in 2013, growers can now apply pollen as part of biocontrol programs. They typically apply pollen weekly (blown at 250 g/ha) or biweekly (at 500 g/ha) (Pijnakker et al., 2016). These dosages correspond to 2.5 to 20 mg per plant. Various devices are available to apply pollen to enhance the establishment of predatory mites on crops (Gan-Mor et al., 2003Weintraub et al., 2009;Pijnakker et al., 2016). Other arguments in favor of the use of cattail pollen, next to its high nutritional value, are its low allergenic character (Weber and Nelson, 1985), its relatively low attractiveness to pests and poor nutritional suitability for thrips (Hulshof et al., 2003). The fact that cattail pollen is not attractive to (bumble) bees, means that it does not distract pollinators used in the crop (Schmidt et al., 1989).
However, the use of pollen can have some drawbacks as it can also benefit herbivores (Kirk, 1987;Van Rijn et al., 2002;Chitturi et al., 2006;Wäckers et al., 2007, Leman andVangansbeke et al., 2016b). Hulshof et al. (2003) showed at laboratory scale that F. occidentalis feeds on pollen, resulting in an enhancement of its growth rate and fecundity. Van Rijn et al. (2002), using simulation models, showed that availability of pollen benefits predators more than thrips and improved predator-prey ratios resulting in enhanced thrips control. This has since been repeatedly confirmed in studies showing the efficacy of the pollen supplementation at plant or crop level (see above). The high reproduction by A. swirskii on plants treated with pollen prevented the pest from developing, even if they can feed on the pollen. This represents an example of apparent competition, where the population development of a prey is suppressed by a shared predator when an additional prey or food supplement is present (Nomikou et al., 2010). Despite the fact that Typha pollen has been widely shown to be effective in supporting biological control, growers could still run a risk of damage when thrips are too numerous at the start of the crop or when predators do not establish because of pesticide residues. When applying any food supplement, including pollen, predation per predatory mite will go down due to satiation effects (Skirvin et al., 2007). In addition to predator satiation (Holt andLawton, 1993, 1994;Shakya et al., 2009) a feedingswitch to pollen may occur. Both mechanisms may result in a temporary reduction in predation rate of the pest (so-called apparent mutualism, Abrams and Matsuda, 1996). High dosage of cattail pollen reduced the predation of thrips by individual A. swirskii by 50% on laboratory scale (Leman and Messelink, 2015). However, these effects are typically short term and are soon outbalanced by the augmentation of natural enemy numbers (Van Rijn et al., 1999).
The use of supplementary food, and pollen in particular, is a powerful tool to help establish predatory mites and improve biocontrol efficacy. The application of exogenous pollen as supplemental food source can be optimized by avoiding excessive doses, as overly high pollen levels may result in satiation of predatory mites and stimulation of thrips (Sabelis and van Rijn, 2006). The choice of the pollen type and the match with the predatory mite can also affect the outcome. However, Typha pollen has proved to be suitable for a broad range of predatory mites ( Table 5). Growers succeed in building strong populations of predatory mites with feeding their predators before pest appearance and performing adjustments in their irrigation systems and their spraying data. Some persue the development of the application by automatizing the blowing of this alternative food.
Floral nectar has evolved as a food reward in the mutualism between plants and their pollinators. Even though natural enemies, with a few noticeable exceptions , are most likely not necessarily effective in pollinating flowers, they, nevertheless, can collect floral nectar to provide for their energetic needs. As natural enemies tend to have short mouthparts, their nectar foraging is restricted to those plant species with open and exposed floral nectaries, such as Apiaceae, Euphorbiaceae, or buckwheat (Campbell et al., 2012;Wäckers and van Rijn, 2012;Van Rijn and Wäckers, 2016).
While accessibility of floral nectar can be a bottleneck for natural enemies, some plants also produce nectar outside of the flowers. These so-called extrafloral nectaries tend to be exposed and thus highly accessible. Furthermore, the nectar is often secreted over prolonged periods of time (Wäckers and Bonifay, 2004). These adaptations fit their ecological function, as extrafloral nectar is part of a defensive strategy, allowing plants to recruit ants and other sugar feeding natural enemies. They, in turn, protect the plants when attacking herbivores. Extrafloral nectaries have been described in more than thousand plant species, including a number of important crops, such as cotton, cassava, peaches, plums, cherries, pumpkins, roses, field beans. In a number of plant systems, it has been demonstrated that the presence of extrafloral nectar can translate into both reduced plant damage and increased plant reproductive fitness (Heil, 2015).
Honeydew is a generic term for sugar-rich excretions of phloem-feeding Sternorrhyncha. In agricultural ecosystems, honeydew is often the most prevalent sugar source (Wäckers and Steppuhn, 2003;Hogervorst et al., 2007;Tena et al., 2015). However, honeydew differs from the above-mentioned sugar sources, as it is primarily a waste product. This can reduce the nutritional value of honeydew (Wäckers et al., 2008).
Overall, the nutritional suitability of the above sugar sources depends on composition and concentration of carbohydrates (Wäckers, 2001;Azzouz et al., 2004;Fratoni et al., 2020). Sugar concentration is an important factor determining sugar uptake. At low concentrations, gustatory perception might be impeded (Wäckers, 1999), whereas viscosity at high sugar concentrations can interfere with sugar uptake (Wäckers, 2000;Winkler et al., 2009). Upon the time of nectar secretion, sugar concentrations can already range from 5 to 75% (Dafni, 1992). Environmental conditions may further affect nectar concentrations both indirectly through their effects on the nectar producing plant, and directly through evaporation, hygroscopy or rain dilution (Winkler et al., 2009). Sugar concentrations of undiluted extrafloral nectar range from 5 to more than 80% (Koptur, 1992a,b;Wäckers, 2001). In general, extra floral nectar shows much more variation in terms of sugar concentration than floral nectar from the same plant. Extrafloral nectar tends to be more concentrated, probably due the fact that its exposed nature increases evaporation. The fact that honeydew is typically available as little droplets or as a thin film on the substrate, means that it is even more subjected to evaporation. As a result, sugar concentrations are often at saturation. This is likely to be a limiting factor in honeydew uptake. This problem is accentuated by the specific tendency of the honeydew sugars: raffinose and melezitose, to crystallize rapidly (Wäckers, 2000).

Providing Sugar Sources to Boost Biological Pest Control
It has long been recognized that the lack of sugar sources in agricultural systems can strongly undermine the efficacy of biological control. This problem could be overcome by introducing food sources into our agricultural systems. Recently, we have seen an increasing interest in the use of (flowering) noncrop plants in field margins as a tool to sustain predators and parasitic wasps. Specifically selected seed mixtures are available that are intended to attract natural enemies and provide them with nectar sources. When introducing nectar plants, the use of sugar sources is not restricted to beneficial insects. Many pest insects thrive on sweets as well . By choosing plants that primarily benefit natural enemies, the positive impact on pest control can be maximized (Gurr, 2005;Winkler et al., 2010). Banker plants can also be used to provide sugar sources, either in the form of (extra-) floral nectar as in the case of castor bean or broad bean or by providing honeydew (considering the caveats mentioned above). As an alternative to the introduction of nectar-or honeydew providing plants, sugar can also be applied as such. Sugar can be either sprayed onto the crop, or provided in so-called "feeding stations." Spraying sugar has the advantage that it is an easy and cheap application method resulting in an even coverage and providing an easily accessible sugar source for the predators and parasitoids. The quantity of the sugar applied can be controlled through the choice of sugar concentration, by adjusting the spray volume/spray nozzle, and by varying the walking/driving speed. Yet, there are also obvious drawbacks to the use of sugar sprays. Blanket sprays quickly result in the crop getting sticky. Certain sugars can cause phytotoxicity when sprayed directly on the foliage. In addition, sugars on the plant surface are prone to growth of sooty mold. These drawbacks can be avoided, either by using very weak sugar concentrations (e.g., 0.1-1%) or by applying a higher concentrated solution in a very fine and light mist. For the target arthropods, having minute sugar droplets is often better than having to deal with a sticky sugar layer, as the latter interferes with arthropod mobility. Biological control practitioners have attempted to incorporate artificial sugar sprays as a strategy to cater to the nutritional needs of parasitoids (Mandour et al., 2007;Wade et al., 2008). However, the efficacy of this form of sugar provisioning in biological control programmes has been limited and inconsistent (Heimpel and Jervis, 2005;Wade et al., 2008). Tena et al. (2015) studied the use of sugar sprays in combination with the release of the parasitoid Aphytis melinus DeBach (Hymenoptera: Aphelinidae) for the control of the California red scale Aonidiella aurantii in a commercial citrus orchard. They demonstrated that sugar supplementation increased the parasitoid population density 2-fold. Parasitoid fecundity on sugar treated trees was shown to be enhanced as well. Higher population densities and the increase in realized fecundity translated in a 2-fold increase in parasitism under crop conditions. The successful examples show that under certain conditions the use of sugars can be an effective element in a conservation biological control strategy.
Some plants have special morphological structures, called domatia (from the Latin for home "domus"), that are targeted to either ants, or predatory/fungivorous mites. Ant domatia are represented by hollow thorns or stems and rolled leaf margins. Acarodomatia may take the form of either pits or (dense) tufts of leaf hairs, in which small arthropods may reside (Romero and Benson, 2005). Many studies have demonstrated that the presence of domatia increases the population of predators, that in turn protect the plant against herbivores and in some instances pathogens and weeds (Lundstrom, 1887;Rozario, 1995;Kreiter et al., 2003;Ferreira et al., 2008Ferreira et al., , 2011Shenoy and Borges, 2010;Parolin et al., 2013). Predatory mites tend to occur more abundantly on plants bearing acarodomatia (O'Dowd and Willson, 1997;Norton et al., 2000;Roda et al., 2001;Romero and Benson, 2005;Avery et al., 2014) and their survival and reproduction on such plants is enhanced (Pemberton and Turner, 1989;Karban et al., 1995;Agrawal, 1997;Cortesero et al., 2000;Avery et al., 2014;Bresch et al., 2015Bresch et al., , 2018. Domatia also act as oviposition substrate. Many predatory mite species show a preference to lay eggs on trichomes. Egg clusters can be frequently seen inside domatia, where predatory mites can molt protected from predators. Besides serving as shelters and oviposition sites, the hair-tufts may trap pollen and fungal spores that the mites can then consume (Roda et al., 2000;Romero and Benson, 2005;Loughner et al., 2008). Thus, acarodomatia can serve a function in providing food as well. Pekas and Wäckers (2017) showed a strong synergistic effect between the availability of fibers and food (pollen and sugar water) in affecting population growth of predatory mites on citrus plants. Romero and Benson (2004) demonstrated the protective role of domatia on the tropical tree Cupania vernalis L. By blocking domatia on part of the experimental trees, they showed that domatia increased predatory mite abundance and lowered herbivore damage from eriophyid mites. However, very few studies have investigated their long-term effect on predatory mites, pest or fungi under natural conditions (Norton et al., 2000;Monks et al., 2007;Ferreira et al., 2010). Norton et al. (2000) showed that acarodomatia increased the abundance of the mycophagous tydeid mite, Orthotydeus lambi (Baker), which resulted in the reduction of 48% of grape mildew infestation on the riverbank grape.
Domatia can be provided to beneficials by (1) selecting crop variety with the appropriate properties, (2) by adding suitable non-crop plants to crops (Skirvin and Fenlon, 2001;Van Rijn et al., 2002;Osborne and Barrett, 2005;Frank, 2010;Huang et al., 2011;Parolin et al., 2012b;Kumar et al., 2015), (3) or by using artificial structures (Loughner et al., 2011;Adar et al., 2014;Pekas and Wäckers, 2017). Crop varieties can differ substantially in domatia characteristics. Choosing crop varieties for their domatia traits could thus help to support establishment of natural enemies. Breeding programmes have actually started to include traits that determine suitability for beneficial organisms (Bottrell et al., 1998;English-Loeb et al., 1999). Some tomato varieties, for example, have been selected for their distorted trichomes to facilitate biological control of tomato russet mites (Van Houten et al., 2013;Legarrea et al., 2020). Another method for applying domatia can be the interplanting of domatia bearing banker plants between crop plants lacking domatia. Parolin et al. (2013) showed an increased spider mite control in roses using the predatory mites P. persimilis and Neoseiulus californicus (MCGregor) when adding lauristinus and frost grape bearing acarodomatia as bankerplants. Adding the banker plants in this system enhanced the establishment of predatory mites and increased their efficacy. The method can only be successful if natural enemies disperse from the manipulated habitat to the crop. In practice, growers tend to focus on commercial crops and are reluctant to introduce non-crop (banker) plants. A further strategy could be to use artificial domatia. Various types of artificial domatia have been used in studies to assess the benefits of natural dolmatia, but also, to improve biological control on crops (Loughner et al., 2010(Loughner et al., , 2011Pekas and Wäckers, 2017). Rozario (1994) brought tufts of polyester fibers to grape varieties with low natural domatia to augment populations of Galendromus occidentalis Nesbitt. Agrawal and Karban (1997) supplemented cotton plants with artificial domatia, which enhanced spider mite control and fruit yields. However, thrips numbers were also shown to benefit from domatia. Kawashima et al. (2006a) reported laboratory experiments showing that textured urethane foam and polyethylene shading nets were suitable sites for the reproduction of N. californicus. Kawashima and Jung (2011) suggested using urethane foam as ground cover in apple orchards to increase the survival of N. californicus populations during the winter. Loughner et al. (2011) mimicked domatia on glabrous beans seedlings and Impatiens plants by adding cotton fiber patches and chopped acrylic yarn fibers and found more A. swirskii on plants provided with pollen and artificial domatia than on plants given solely pollen or fibers. In this study, cotton patches and paper pulp supplements augmented and maintained A. swirskii populations, whereas jute and celluflo were not effective. Adar et al. (2014) tested the "pollen on-twine" method (Gan-Mor et al., 2011), using fibers (rayon/viscose 80% and jute 20%) coated with pollen and succeeded in enhancing E. scutalis populations. Oviposition occurred on rayon rather than on jute. Bresch et al. (2018) found wool, silk, polyamide, viscose and polyester to be equally suitable as natural domatia for oviposition by N. californicus. However, none of them improved spider mite control by N. californicus and polyamide even benefitted the pest. Pekas and Wäckers (2017) showed that the combined use of fibers, pollen, and sugar generate synergistic benefits to population growth of Euseius stipulatus (Athias-Henriot) on bitter orange. In orchards, Koike et al. (2000) designed a Phyto trap, which contains wool yarn mimicking the microstructure of spider mites colonies. This tool aims at collecting predatory mites in pear trees to monitor their densities during the seasons and maintain populations (Kawashima and Amano, 2006;Kawashima et al., 2006b). This tool inspired Messelink et al. (2016) to test among others further material with Velcro tape in greenhouse crops, but found that using millet husk or jute both combined with pollen were more effective to enhance the establishment of A. limonicus on anthurium. In spite of these examples showing the potential of using artificial domatia, there are no commercial applications of these artificial domatia yet.

CONCLUSIONS
This review presents the potential, advantages and risks of the concept of "beneficials-in-first" over augmentative biological control in greenhouses and aims at discussing tools thought to preserve natural enemies in biological control programmes. The studies described show that natural enemies' survival, development, reproduction and efficacy can be enhanced with the use of factitious prey, banker plant systems, pollen, sugars and manipulated habitats. This concept of "beneficials-in-first" is no longer a focus of researchers only, but has been explored by commercial growers over the past two decades. However, despite the trials demonstrating their benefit, some methods to enhance natural enemies' establishment are still excluded from biological strategies in practice, as they often lead to other issues at large scale. The development of hyperparasitoids or pests on banker plant systems, the possible molting of Ephestia eggs and Artemia cysts, the lack of survival of living factitious prey, the damages caused by Tyrophagus putrescentiae or by mirid predatory bugs when they are too numerous, are such examples. Overhead irrigation can hamper the application of alternative food and artificial domatia in some crops. Artificial domatia are also difficult for growers to introduce in their crop.
Further increasing the adoption of the "beneficials-in-first" concept by growers can be achieved by: 1. Clearly demonstrating efficacy on a larger scale and communicate if the selected resources benefit more natural enemies than the antagonists or pests, decrease pest damage and increase production quality or yield; 2. Increasing the involvement of growers in the process of such large-scale experiments, including costs and handling times; 3. Fine-tuning the amount of supplemental resources, the release techniques, the timing and the frequency with which the tools should be employed (Janssen and Sabelis, 2015;Madadi, 2018); 4. Automatization as it will be the key in expanding the scope of inoculative biological control, both in protected and openfield crops.
In the coming decade, we foresee more applications of feeding stations and nectar plants in crops to enhance the survival of natural enemies. Beside the tools of enhancement of beneficials, the use of methods limiting pest activity in greenhouses will gain more importance like tools of sexual confusion, distraction of pests such as "push-pull systems" or the use of "repellent substances."

AUTHOR CONTRIBUTIONS
JP: conceptualization, supervision, writing original draft, review, and editing. FW: conceptualization, supervision, writing draft, review, and editing. DV, RM, and MD: review and editing.
All authors contributed to the article and approved the submitted version.