Are Botanical Biopesticides Safe for Bees (Hymenoptera, Apoidea)?

Simple Summary Synthetic pesticides are among the main threatening factors for wild and managed bees. In recent decades, botanical biopesticides have been gained increasing interest and use in agriculture due to their high selectivity and short persistence in the environment. To date, however, little has been discovered or researched about the adverse effects of these substances on bees. This paper reviews studies in the literature reporting the lethal and sublethal effects of botanical biopesticides on social and solitary bees. Although botanical products are considered safer than chemical pesticides, some of them can cause lethal and several sublethal effects on bees. We suggest that more research is needed on this topic, especially increasing knowledge about certain groups of bees such as solitary bees. Abstract The recent global decline in insect populations is of particular concern for pollinators. Wild and managed bees (Hymenoptera, Apoidea) are of primary environmental and economic importance because of their role in pollinating cultivated and wild plants, and synthetic pesticides are among the major factors contributing to their decline. Botanical biopesticides may be a viable alternative to synthetic pesticides in plant defence due to their high selectivity and short environmental persistence. In recent years, scientific progress has been made to improve the development and effectiveness of these products. However, knowledge regarding their adverse effects on the environment and non-target species is still scarce, especially when compared to that of synthetic products. Here, we summarize the studies concerning the toxicity of botanical biopesticides on the different groups of social and solitary bees. We highlight the lethal and sublethal effects of these products on bees, the lack of a uniform protocol to assess the risks of biopesticides on pollinators, and the scarcity of studies on specific groups of bees, such as the large and diverse group of solitary bees. Results show that botanical biopesticides cause lethal effects and a large number of sublethal effects on bees. However, the toxicity is limited when comparing the effects of these compounds with those of synthetic compounds.


Introduction
Bees (Hymenoptera, Apoidea) are the main group of pollinating insects, represented by about 20,000 described species in the world, with the greatest biodiversity in Mediterranean and xeric climate regions of the globe [1,2]. Due to its biological and ethological characteristics, this group provides the ecological service of pollination for spontaneous and cultivated plants. In particular, the pollination service by animals includes 87% of the world's spontaneous flowering plants [3], and about 75% of cultivated crops [4,5]. Fatty acids have also been used in some commercial biopesticides where they have a stabilizing function. However, some of them can have a secondary toxic effect on insect pests, for example, conjugated linoleic acid (CLA) or pelargonic acid [46]. Furthermore, preliminary studies on the fatty amid pellitorine have shown promising results for the control of mosquitoes [58] and Coleoptera post-harvest pests [59].
Limonoids are natural compounds mainly present in plants of the Rutaceae (Citrus spp.) and Meliaceae (Neem tree, Azadirachta indica A. Juss.) families. Azadirachtin from the Neem tree is one of the most widely used and studied biopesticides [49], isolated from all the parts of the plants, in particular from seeds [40]. Considered as a safe and selective product, Azadirachtin is very effective against several groups of pests, causing acute toxicity and anti-feeding and physiological effects [60], and it can also be used as a fungicide [40]. However, several studies question its safety as regards beneficial insects [61][62][63].
Phenolics, the largest group of plant secondary metabolites, perform various essential functions, from the regulation of physiological processes to defence against herbivores [64]. Abundantly present in Thymus spp. (Lamiaceae) plants, thymol can be used as an effective fungicide, bactericide, and also acaricide, it being effective against Varroa spp. (Varroidae), an important ectoparasite of honeybees [65]. The isoflavone rotenone, extracted from the roots of some species of Derris, Lonchocarpus, and Tephrosia (Fabaceae), is a neurotoxic compound used against a wide spectrum of insects and in the control of fish populations [40]. The safety of rotenone for humans and the environment has been questioned due to its high toxicity towards mammals [47].
Another large class of plant secondary metabolites is that of polyketides, biosynthesized from acetyl-CoA. Annonins, classified as acetogenins, comprise an important group of polyketides that show a wide range of biological activities such as antimicrobial and pesticidal activities [66]. Effective against Coleoptera pests [67], annonins are extracted from the seeds of neotropical Annona (Annonaceae) trees [47].
Pyrethrin is one of the most marketed bioinsecticides. It is a mixture of compounds (Pyrethrins I, and Pyrethrins II) biosynthesized from Tanacetum cinerariifolium (Trevir.) Sch. Bip. (Asteraceae) [68]. Pyrethrin has neurotoxic action that interferes with the Na + /K + exchange pump, causing paralysis and resulting in toxicity for several groups of pests [46]. The characteristic of pyrethrins, of being particularly labile to UV from sunlight, led in the 1970s and 1980s to the development of synthetic derivatives, the pyrethroids, which are widely used nowadays [49].

Risk Assessment of Biopesticides on Bees
Most of the regulatory risk assessment for plant protection products (PPPs) (pesticides and biopesticides) uses the western honeybee as a surrogate species for ecotoxicological testing of pollinators [69,70]. In recent years it has been realized that this approach is not enough for pollinator conservation [70,71]. Sensitivity to pesticides varies according to the bee species and other factors such as body size, level of sociality, seasonality, voltinism, floral specialization, nesting behaviour, food consumption rate, overwintering strategies, sex, and caste [27,33,72,73]. This leads to different ecological impacts from the use of pesticides. Therefore, risk assessments have recently been expanded to include other bee species such as bumblebees (Bombus spp.) [26,44,74,75], solitary bees (Osmia spp. and Megachile rotundata (Fabricius)) [8,26,44,76,77], and stingless bees [78]. Currently ground-nesting bees, which represent about 70% of bee species [1], have not been taken into consideration in the PPP risk assessment protocols due to their difficulty in breeding, management, and use in laboratory protocols [8]. Still, today, however, knowledge regarding the ecotoxicology of the non-Apis bee species is scarce and certainly needs to be increased [33,79,80].
To date, there are no specific regulations and protocols for testing the effects of biopesticides on bees. Therefore, the same protocols for the chemical pesticides developed by the Organisation for Economic Co-operation and Development (OECD) are used [37]. These protocols [74,81] include laboratory chronic and acute oral/contact toxicity tests to measure ecotoxicological parameters, such as LC 50 and LD 50 , and to evaluate sublethal effects such as paralysis, movement alterations, and the presence of moribund specimens. However, sublethal effects caused by pesticides on bees are commonplace and still little studied, particularly regarding ecotoxicological tests about the effects of biopesticides on non-Apis bees [82][83][84].
In addition, there are few guidelines for the risk assessment of pesticides at field and semi-field levels with honeybees, bumblebees, and solitary bees [26,85], and no specific protocols for biopesticides. In general, the number of higher tier risk assessment studies on bees is low, both for synthetic pesticides (excluding neonicotinoids) and biopesticides.
We therefore highlight the need to (I) develop specific protocols to assess the lethal and sublethal effects of biopesticides on bees from different species; (II) increase the knowledge about species sensitivity distribution regarding chemical and biological pesticides for Apoidea, including ground-nesting solitary bees in the studies; and (III) develop new and better field and semi-field protocols both for synthetic and biopesticides.

Materials and Methods
The search for peer-reviewed English-language publications up to 2022 was conducted using Google Scholar, Scopus, and ResearchGate through the following keyword terms and their combination: "bioinsecticides", "biopesticides", "botanical insecticides", "Azadirachtin", "Essential oils", "EOs", "botanical extracts", "Pyrethrins", Pyrethrum", "Nicotine", AND "toxicity", "exposure", "effect", AND "bee", "social bee", "honeybee", "Apis", "bumblebee", "Bombus", "stingless bee", "Meliponini", "solitary bee", "Osmia", "Megachile". Additional studies from literature references were considered. In this review, papers about the toxicity on A. mellifera of botanical products used in beekeeping were not considered, as they are summarized in the recent revision of Ntalli et al. [39]. Table 1 includes and summarizes the studies and divides them following the different groups of Apoidea (social and solitary bees). Information is provided about the botanical substances tested, the category of assay (laboratory-assessed lethal (L) and/or sublethal (S) effects, field and semi-field), the type of treatment application (contact, topical, ingestion, fumigation, spray, crop spraying, crop granules, and "ingestion and topical" in the cases of tests with larvae), the target of the experiments (eggs, larvae, adults, colony, microcolony), the main effects reported by the results of the experiments, and the country in which the experiments were performed. The "Botanical substance" column in Table 1 includes the individual biopesticides tested in the different papers analysed. No papers were found in which synergistic effects between different substances were tested. Table 1 includes only article papers; however, some studies presented at conferences or symposia are discussed in the text. Most of the studies reviewed (53.7%) did not test a chemical insecticide as a positive control, and these studies are highlighted in Table 1 with a double asterisk (**). The toxicological parameters (LC 50 , LD 50 ) of botanical biopesticides extrapolated from the analysed papers are reported in Table 2.

Honeybees
Honeybees are eusocial bees and are among the best known and most widely studied insects. Of the ten Apis species [86], two are managed in the world [87]. The first is the eastern honeybee, A. cerana Fabricius, managed in South and East Asia [88], and the second is the western honeybee, A. mellifera, managed in Africa and Europe for several millennia [89]. Nowadays, A. mellifera is a cosmopolitan species and is considered the most important pollinator worldwide, it being also the most important species for honey, pollen, propolis, and wax production [90]. Colony collapse disorder syndrome (CCD), observed for the first time in the US in the spring of 2007, has sparked growing interest from the public and the scientific community regarding the conservation of honeybees, and consequently other bees [7].
In the literature on the effects of botanical insecticides, both species of Apis previously mentioned have been studied, although A. mellifera is present in a greater number of studies. A single study [91] includes also the giant honeybee, A. dorsata Fabricius, a wild honeybee found from India to Vietnam.
Some studies showed that alkaloids can be toxic to A. mellifera. Nicotine increases the mortality, specifically of honeybees, causing a mortality from 75 to 100%, and shows a LC 50 value of 60.15 ng/bee in contact exposure and 32.45 ng/bee in ingestion [92,93]. Sabadilla dust at its highest dosage shows 100% mortality within 48 h [94,95], and ryania dust extract at 40% caused 31% mortality 72 h after treatments [94].
Several studies have evaluated the impact of essential oils, with potential use in beekeeping for controlling Varroa spp. mites or other parasites, on A. mellifera, and the majority of these compounds showed low toxic effects for honeybees (reviewed by Ntalli et al. [39]). Many EOs and botanical extracts for use in crop protection also showed a lack of or low toxicity for honeybees at realistic field dosages [93,[96][97][98][99]. However, several oils and extracts, widely used as biopesticides, caused lethal and sublethal effects on the larvae and adults of honeybees. Andiroba oil (Carapa guianensis Aublet, Meliaceae) and garlic extract (Allium sativum L., Amaryllidaceae) caused high larval mortality and affected the development and body mass, while citronella (Cymbopogon sp., Poaceae) and eucalyptus oil (Eucalyptus sp., Myrtaceae) showed high mortality for adult honeybees [100]. In addition, eucalyptus and garlic oil decreased honeybee speed and movement in walking tests, and all botanical treatments showed repellent effects on worker honeybees [100]. Artemisia absinthium L. and Eupatorium buniifolium Hook. ex Hook. & Arn. (Asteraceae) EOs, potentially usable against tomato pests, were tested also on A. mellifera [98] in topical tests and in the "Complete Exposure Test" described in Ruffinengo et al. [101]. Results showed a high LD 50 in topical tests (respectively 197 and 252 µg/bee) but a low LD 50 in the Complete Exposure Test (respectively 0.26 and 0.15 mg/cm 2 ). This suggests that the use of these products may be minimally toxic for bees at the doses that can be used for the control of Trialeurodes vaporariorum (Westwood) (Aleyrodidae) (LD 50 : 0.08 and 0.02 mg/cm 2 , respectively) but toxic at the doses usable against Tuta absoluta (Meyrick) (Gelechidae) (LD 50 : 0.50 and 0.65 mg/cm 2 , respectively) [98]. EOs of Origanum vulgare L. and Thymus vulgaris L. (Lamiaceae) showed higher mortality in topical and contact tests with adult honeybees, and O. vulgare EO also reduced honeybees' mobility during the walking bioassays [102]. Spray and ingestion treatments with extracts of  [103]. The ingestion of Origanum majorana and P. granatum also reduced the mesenteric cells of the midgut of workers [103]. In addition, repellent effects on A. mellifera were observed using garlic and citronella extracts [104].
Currently, no field or semi-field studies have been conducted with EOs or their extracts to assess their effects on honeybees.
A single study evaluated the effects of fatty acid-based bioinsecticides on A. mellifera [105]. The amide pellitorine increased the mortality of larvae, newly emerged adults, and adults workers [105].
There is a considerable number of laboratory and field studies regarding the impact of Azadirachtin on honeybees (A. cerana, A. dorsata, and A. mellifera). Some of those papers reported only slight toxicity [97,106,107] or the absence of effects in field applications [108]. The first studies were conducted in the 1980s and reported high mortality and reduction in the survival of A. mellifera larvae, with morphological larval abnormalities, but no larval anti-feeding effects [109,110]. Young honeybees had malformations and were unable to hatch after the treatment of small hives with Neem seed extract spray [111]. However, these symptoms have not been observed in larger colonies, and field treatments carried out did not repel honeybees from flowers [111]. The absence of field repellence of honeybees was also observed after the treatment with a Neem seed extract of canola fields (Brassica campestris (L.), Brassicaceae) in Canada, despite the occurrence of repellent effects in laboratory choice tests [112]. The application of Neem oil on the cells of honeybee larvae caused high mortality at higher doses, with a higher LD 50 than those of the other insects [113]. High mortality [100,114,115], reduction in adults' survival [116][117][118], and larval survival and development [100,118,119] were found in laboratory tests with azadirachtin. The ingestion of azadirachtin altered the haemolymph amino acid composition [114]. Azadirachtin also impaired the flight ability [117] and walking activity [100] of honeybees. Furthermore, sublethal concentrations of azadirachtin decreased immune gene expression, and inhibited the activity of polyphenol oxidase (PPO), a midgut antioxidant enzyme of A. cerana cerana workers [115]. Field and semi-field studies highlighted the detrimental effects on honeybees after azadirachtin treatments. Although azadirachtin did not affect honeybee mortality, there was a reduction in the foraging activity and brood development of honeybees placed in tunnels with Brassica napus L. (Brassicaceae) in semi-field experiments [120]. Thompson et al. [121] observed a reduction in colony overwintering in azadirachtin-treated colonies but no apparent effects regarding the brood and queen development. Different formulations of azadirachtin affected the number of foraging honeybees of A. cerana indica, A. dorsata, and A. mellifera in mustard crops in India [91], as well as the number of A. mellifera forager bees in Brazilian melon fields [122].
Thymol and other phenolics such as carvacrol can be used in organic beekeeping, especially in the control of Varroa destructor [123]. These compounds have been tested on honeybees and their parasites by several authors, and the effects are summarized in Ntalli et al. [39]. These authors, considering them toxic to honeybees, reported however that their hazard ratios were lower than those of common synthetic alternatives. Studies on the impact of rotenone on A. mellifera have been conducted since the first half of the last century [94,[124][125][126]. Rotenone was evaluated as slightly harmful to A. mellifera, after the reduction in honeybee survival in topical contact tests [97]. Repellence, high mortality of adult honeybees, reduction in body mass, and modifications in walking activity of foraging honeybees were also detected after rotenone exposure [100].
As regards polyketides, we found a single study evaluating the effects of a squamocin (annonin)-based product on A. cerana in the laboratory and field [106]. In laboratory experiments, the product was considered to be slightly to moderately toxic and selective to honeybees according to the selectivity ratio (LC 50 of beneficial species (%)/LC 50 of pest species (%)) (LC 50 reported in Table 2). However, the field tests showed a significant reduction in the relative abundance and in the speed of foraging honeybees in Indian mustard crop (Brassica juncea L. Czern., Brassicaceae) [106].
As they were among the first botanical insecticides to be commercialized, pyrethrins have been tested on honeybees since the last century. Some studies reported no effects or low reductions in mortality [94,127,128], while others found high levels of toxicity [124][125][126] ( Table 2). This is likely due to the use of different formulations and different methods of product application (by contact, fumigation, spray, topical, and ingestion). Recently, it has been shown that a pyrethrum nanopesticide decreased the longevity and caused morphological alterations in the midgut of Africanized honeybees [129].

Bumblebees
Bombus Latreille includes about 250 species with annual colonies, [1], which are abundant in the Holarctic region. Several bumblebees species in Europe and North America are threatened, and their populations are in decline [12,23,[130][131][132]. There are nine species of social bumblebees managed for crop pollination [87], some of them widely used in laboratory risk assessment of synthetic pesticides. Bombus terrestris (L.) is the model species for ecotoxicological studies on bumblebees [26] and the only bumblebee species used in studies that have evaluated the risks of botanical insecticides ( Table 1). The buff-tailed bumblebee, B. terrestris, is one of the most abundant bees in the West Palaearctic [133], and it is a widely commercialized species used for pollination of several crops [134].
There is a small number of studies on botanical insecticides regarding B. terrestris, almost all on the effects of azadirachtin. Azadirachtin caused repellence on workers and caused high mortality in treated microcolonies of B. terrestris. Different azadirachtin concentrations also caused sublethal effects on the colonies, reducing the egg-laying, production of drones, the ovarian length, and the body mass of male offspring [61]. However, various formulations and technical powders of azadirachtin used at field-recommended doses in the laboratory did not elicit side effects on the treated colonies [135]. Colonies orally exposed to azadirachtin had a slight increase in worker and drone mortality, although the treatment was toxic for queens [136]. Other formulations of azadirachtin (nimbecidine) were highly toxic to B. terrestris in an acute oral experiment [137].
Sublethal concentrations of azadirachtin also reduced the foraging of pollen by B. terrestris in a legume field [138].
A blend of Perilla frutescens var. crispa extracts and phytoncide oil was found to be particularly toxic for B. terrestris, causing a 100% mortality after one hour in a contact experiment [139]. Sublethal doses of these botanical insecticides did not affect the walking distance and velocity of the bumblebees, but the angular velocity was significantly affected. Furthermore, the bioinsecticide reduced the levels of different genes involved in metabolism (NADH dehydrogenase 1 alpha subcomplex subunit 12, NDUFA12, cytochrome b-c1 complex subunit 9-like, UQCR10-like, ATP synthase subunit b, ATP5F1, and cytochrome b-c1 complex subunit 8, UQCRQ) at five and ten minutes after treatments [139].
To date, we are not aware of the LD 50 or LC 50 of botanicals biopesticides for any bumblebee species (Table 2).

Stingless Bees
Tribe Meliponini is a large group of bees found in the tropical and subtropical areas of the world with about 500 described species [1,140]. In Central and South America and Australia, stingless bees play important roles as pollinators and honey producers [141]. Pesticides are among the major threats for these pollinators [142][143][144][145][146], and only recently have different studies assessed the risks of pesticides and biopesticides on these pollinators [27,73,145]. Meliponini appear to be more sensitive to pesticides than A. mellifera and other pollinator species [33,147]. However, further studies are needed considering the large number of stingless bee species [27].
All studies on the impact of botanical insecticides on stingless bees were conducted in Brazil, involving a total of eight species (Table 1).
Most of the EOs tested on stingless bees resulted in low lethal effects [93,102,148,149], although other studies showed high mortality after EO treatment. However, topical contact tests with Corymbia citriodora EO (Myrtaceae) on Tetragonisca angustula Latreille resulted in 100% mortality of these pollinators [150]. EOs from Artemisia annua L. (Asteraceae) also increased the mortality of T. angustula [151]. The handling behaviour of Nannotrigona testaceicornis (Lepeletier) and T. angustula was evaluated through a videotrack system, after treatment with different EOs and extracts, which showed no effects in both species [152].
The EOs of Origanum vulgare and Thymus vulgaris reduced the walking speed and the travelling distance of Trigona hyalinata (Lepeletier) [102].
Several studies assessed the lethal and sublethal effects of azadirachtin on stingless bees in the laboratory, while few studies have been carried out in field and semi-field conditions. Topical tests with leaf and seed extracts from Azadirachta indica reduced the survival of T. spinipes [148]. The ingestion of azadirachtin during larval development increased the mortality and led to the production of deformed pupae and adults of Melipona quadrifasciata Lepeletier, despite not delaying the development time [153]. However, in adult stingless bees, the lethal toxicity of azadirachtin seems to be lower; it caused low mortality in contact and ingestion assays with Partamona helleri (Friese) and Scaptotrigona xanthotrica Moure [154]. In this study, the flight take-off of P. helleri was affected by oral ingestion [154]. Azadirachtin did not cause mortality in M. quadrifasciata and P. helleri in contact and oral exposure experiments. However, this biopesticide, in a concentrationdependent manner, caused a significant anti-feeding effect on P. helleri and repellence in M. quadrifasciata [62]. Azadirachtin proved to be particularly toxic for P. helleri queens reared in vitro during post-embryonic development, reducing survival at the higher doses and delaying development [63]. The treatment also deformed the specimens and reduced the reproductive system area of the queens [63]. Modifications in the gene expression of esterase isoenzymes (EST) and peptides were observed in T. angustula after contact tests in the laboratory and semi-field with different formulations of azadirachtin [155].
The ingestion of azadiractin caused a reduction in the gene expression of vitellogenin (Vg) of M. quadrifasciata workers, infected and uninfected with Escherichia coli [156]. The same study highlighted an increase in the number of haemocytes in both infected and uninfected bees due to insecticide ingestion [156].
Flower visitation rates of Plebeia sp. bees were not affected by azadirachtin treatment in a Brazilian melon field [122].

Effects of Botanical Biopesticides on Solitary Bees
The great majority of bee species in the world is solitary and belongs to seven different families (Stenotritidae, Colletidae, Andrenidae, Halictidae, Melittidae, Megachilidae, and Apidae) [7]. They exhibit a great variety of size, morphological characteristics, behaviour, nesting habitats, flight ranges, phenology, and nutritional requirements [1,7]. Eight species of solitary bees, mainly cavity-nesting species belonging to the genera Megachile Latreille and Osmia Panzer (Megachilidae), are managed for the pollination of crops around the world, and another 14 are potentially usable species [87]. Among these, there are a few ground nesting bee species such as the alkali bee, Nomia melanderi (Cockerell) (Halictidae), which are managed in North America, or Rhophitoides canus (Eversmann) (Halictidae) in Eastern Europe. The status of solitary bees is not well known throughout the world. In Europe, which hosts the best-known bee fauna, the latest IUCN Red List [23] assessed 60% of the species in the "data deficient" category, and the majority of the threatened species (45 spp.) are solitary. There is a clear need to improve our knowledge of the status of solitary bees in the world and to assess the risk to them of synthetic chemicals and alternative biopesticides.
The European Food Safety Authority (EFSA) suggested including the red mason bee, Osmia bicornis L., and the European orchard bee, Osmia cornuta (Latreille), as model organisms of solitary bees in the EU pesticide risk assessment scheme [26]. However, standardised test protocols to assess acute toxicity for solitary bees are still in development. The US EPA [44] has suggested the blue orchard bee, Osmia lignaria Say, and the alfalfa leafcutting bee, Megachile rotundata (Fabricius). The latter, to date widely managed for crop pollination in North America, is a Eurasian bee accidentally introduced into the US in the 1940s.
Knowledge regarding lethal and sublethal effects of botanical compounds on solitary bees is very scarce, and the few studies carried out show non-uniform laboratory protocols, with the use of different methods of application, and different life stages (eggs, larvae, newly emerged, adults of females and males). A product based on an extract of a small tropical tree, Quassia amara L. (Simaroubaceae) (Tecomag ® ), was found to be particularly toxic at field doses for Osmia cornuta eggs and larvae, with a mortality of more than 80% in a preliminary study conducted through the application of a drop of test solution in the provision of the eggs/larvae [157].
Studies conducted in North America, with adults of Osmia lignaria, showed low reduction in mortality with topical and ingestion treatment of Neem oil [116]. Slightly increased mortality was also registered in adults of Osmia cornifrons Radoszkowski with a treatment of wintergreen oil (Gaultheria procumbens L., Ericaceae) as a fumigant used for the control of Chaetodactylus krombeini Baker (Chaetodactylidae) [158]. Another study, conducted in Canada [57], tested botanical insecticides that could potentially be used to control a natural enemy of solitary bees, such as Pteromalus venustus Walker (Pteromalidae), a parasitoid of the alfalfa leafcutting bees, Megachile rotundata. This work tested fifteen plant powders against parasitoid and adult male bees in a contact experiment, highlighting a higher bee mortality with nutmeg powders [57].
The only field study [122] was conducted in Brazilian melon (Cucumis melo L.) fields, investigating the visitation rates of Halictus spp. (Halictidae) after treatments with azadirachtin. Halictus Latreille is a wide genus of bees that includes a scale of different social behaviours from solitary and semi-social to social species. Since the species was not specified in Tschoeke et al. [122], we reported this in the solitary bees category. Halictus bees showed reduced visitation intensity after treatment with a neem-based insecticide.
As with bumblebees, the LC 50 or LD 50 of botanical biopesticides for solitary bee species were not calculated in the studies reviewed in the literature ( Table 2).

Conclusions
Increasing awareness of the risks associated with synthetic pesticides is leading to a revaluation and increased production of studies on botanically derived products. These studies are undergoing a renaissance, especially in some countries such as India, China, and Brazil, where the number of papers on this topic has grown, particularly in recent decades [159]. Although botanicals are presented as eco-friendly alternatives with high selectivity and low persistence in the environment, the knowledge regarding their posing of risks to non-target organisms is still scarce, and studies in the literature indicate several detrimental effects on pollinators. Brazil appears to be the country from which the largest number of papers analysed came (n = 23), followed by the United States (n = 10) and India (n = 4). For several countries, especially in Europe, the efforts made to date on this topic are rather limited ( Figure 1A). Brazil has the greatest biodiversity in the world, so many botanical biopesticides are constantly being tested [160][161][162][163]. In Brazil, beekeeping activities stand out, in addition to a vast population of native stingless bees. In this sense, efforts have been made to carry out selectivity and toxicity tests of these products on these pollinators [62,63]. In general, the toxicity of botanical biopesticides is lower than that of synthetic products. However, this review highlights how the products from different classes of botanical biopesticides can cause lethal effects and a wide variety of sublethal effects (Table 3) on different groups of bees, ranging from social to solitary species, although studies found in the literature focus on just a few model species. Indeed, the great majority of the analysed papers focused on honeybees, especially A. mellifera, while very few works focused on a few other model species, such as bumblebees, and stingless bees ( Figures 1B and 2). Despite neotropical stingless bees only recently being the subject of risk assessment regarding pesticides and biopesticides, there is a growing number of stud-ies on botanical substances. On the other hand, for other groups such as solitary bees, the number of studies, and the number of species and substances tested, are still scarce (Figure 2). In the literature we found toxicological parameters (LC 50 and LD 50 ) of the different botanical pesticides only for honeybees and stingless bees, with a gap for bumblebees and solitary bees ( Table 2). The toxicity of botanical biopesticides for bees varies greatly among different classes of botanicals and different formulations, and in this regard, the majority of the papers analysed focused on limonoids (azadirachtin) and essential oils ( Figures 1C and 2). Essential oils in general have been shown to be less toxic than other botanical products, although there are several exceptions. Alkaloid products such as sabadilla or ryania extracts cause lethal effects on honeybees and have not been tested on other bee species. Azadirachtin has proven to be one of the most studied botanical insecticides concerning bees, reporting lethal effects for several bee species, and a massive presence of sublethal effects (Table 3), this despite the detrimental effects varying significantly depending on the formulations used. The toxicity of some products for bees deserves further investigation with a focus on the sublethal effects, and an increase in field and semi-field studies, which have currently been carried out in a small proportion ( Figure 2). None of the trials with alkaloids, EOs, or phenolics from the analysed papers were conducted under field or semi-field conditions ( Figure 2). No synergistic effects between botanicals and between botanicals and chemicals pesticides have been investigated in bees, although this is an area of recent interest and attention [164][165][166]. Furthermore, of the total number of the studies reviewed (n = 54), a great proportion (53.7%, n = 29) do not include a chemical insecticide as the positive control in the experimental procedure. Including tests with a chemical insecticide group (control) can increase the potential for analysing and considering the data and can also facilitate understanding of the results. For this reason, it is important to highlight the need for protocols. In general, due to non-uniformly used methodologies, the results often cannot be compared. In addition, we have almost no information on their residues and persistence in bee matrices and thus the potential exposure level for bees in the field. The combination of these factors makes it complex to assess and discuss the actual safety of some of these products for bees. Therefore, the need has emerged for the development of new standardized protocols for the risk assessment of plant protection products regarding the different groups of bees. This is particularly true for stingless bees and solitary bees, there being protocols for honeybees in the literature but very few protocols for bumblebees (Table 4). New protocols are also required to evaluate the great variety of sublethal effects that could affect bees. In general, botanical biopesticides seem safer for bees than synthetic pesticides. However, some products need further evaluations, with the adoption of standardized biopesticide protocols, in order to assess the risks for different bee species, from social to solitary. the different groups of bees. This is particularly true for stingless bees and solitary bee there being protocols for honeybees in the literature but very few protocols fo bumblebees (Table 4). New protocols are also required to evaluate the great variety o sublethal effects that could affect bees. In general, botanical biopesticides seem safer fo bees than synthetic pesticides. However, some products need further evaluations, wit the adoption of standardized biopesticide protocols, in order to assess the risks fo different bee species, from social to solitary.   Table 4. Overview of the research gaps in the literature regarding the effects of botanical biopesticides and bees and future directions.

Research Gaps
Future Directions Lack of specific protocols to assess lethal and sublethal effects of biopesticides on bees Use of non-uniform methodologies Lack of inclusion of a positive control (chemical pesticide) in most studies Lack of studies focusing on groups, such as bumblebees, and solitary bees Toxicological parameters (LC 50 and LD 50 ) available only for honeybees and stingless bees Few field and semi-field studies Limited efforts of European countries on this issue Develop specific protocols to assess the lethal and sublethal effects of biopesticides on bees from different species with the inclusion of a positive control Increase the knowledge of species sensitivity distribution regarding chemical and biological pesticides for Apoidea, including in the studies the ground-nesting solitary bees Develop new and better field and semi-field protocols for both synthetic and biopesticides

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.