Orchid bee collects herbicide that mimics the fragrance of its orchid mutualists

Male orchid bees participate in a well-known mutualism with approximately 700 neotropical perfume orchids (Dressler 1982). Instead of producing their own pheromones for reproduction, they collect volatile oils from the surfaces of orchids and other sources by secreting lipids from their labial glands to capture these volatile oils, and then collecting the dissolved oils with brushes on their front feet (Eltz et al. 2005). They then hover and transfer the collected oils into storage tanks on their hind tibia, and later expose the oils during their courtship (Eltz et al. 2005). These volatile oils are positioned on surfaces of the orchid flowers so their collection by the bees often results in pollination of the flowers. To expose the volatile oils, the bees remove the collected oils from storage tanks and place them on velvet pads on their middle tibia and then fan their wings to create a fragrance cloud that attracts females. The females of the approximately 250 orchid bee species mate only after their males display a blend of collected chemicals, which is unique to each orchid bee species (Henske et al. 2023). Males also collect volatile oils from a few perfume flowers in other plant families such as Araceae, Solanaceae, and others, as well as from leaves such as basil and from decaying wood (Pemberton and Wheeler 2006; Whitten et al. 1993). Ramírez et al. (2010) observed male naturalized orchid bee, Euglossa dilemma Bembé & Eltz (Hymenoptera: Apidae), collecting triclopyr herbicide in Florida, and verified that large amounts of this compound in its ester form as 2-butoxyethyl (3,5,6-trichloro-2-pyridinyl) oxyacetate (sometimes known as triclopyr-BEE, hereafter simply triclopyr) was found in the tibial storage tanks of the bees. Bioassays with the herbicide and naïve male E. dilemma in the laboratory found that the bees responded quickly and exhibited collecting behavior (Ramírez et al. 2010). Exposure of the herbicide to E. dilemma in its native Mexico also found that they were attracted to triclopyr (Ramírez et al. 2010). In this scientific note we will elaborate on this unusual and interesting phenomenon and consider if the herbicide or its derivatives might be harmful to the bee and why the bees habitually and intensively collect this herbicide.

E. dilemma is native to Mexico and Central America and was first detected in Broward county in southeastern Florida in 2003 (Pemberton and Wheeler 2006; Skov and Wiley 2005). This bee has spread and now occurs from the Florida Keys north to central Florida (Pemberton and Escalona 2023). E. dilemma has been observed to collect pollen, nectar, and floral resin for its nest construction, and fragrance chemicals needed for its courtship from the flowers and a few leaves of 259 plant taxa, including 237 species belonging to 156 genera and 56 families in Florida (Pemberton 2023).

On 19 May 2005, at Fern Forest, a Broward county nature preserve west of Ft. Lauderdale, the first author noticed four males flying around leaves of Brazilian pepper (Schinus terebinthifolius Raddi: Anacardiaceae), a highly invasive tree that had been sprayed with an orange substance. The males exhibited their stereotypic fragrance collecting behavior (Figure 1). At another site in Fern Forest, many male orchid bees were flying around another invasive weed, air potato (Dioscorea bulbifera L.; Dioscoreaceae), that had been sprayed with this orange substance, and four bees were collecting the substance. From the preserve manager, Nikki Hochberg, it was learned that the orange substance was the herbicide, Garlon^® (triclopyr; Dow AgroSciences Inc, Midland, Michigan, USA) that had been applied to the invasive weeds 3 d earlier on 16 May 2005. Returning to the park on 27 May and 28 May, 11 and 12 d after the spraying, an estimated 18 and 17 males were flying near to or collecting the same sprayed herbicide. Preserve personnel said that when the herbicide was being prepared, prior to the orange colorant or the surfactant being added, about a cup of Garlon^® was spilled on an asphalt walkway. An absorbent was added to soak up and remove the herbicide. No bees were evident at or near the spill site until the next day when about “a hundred” came to herbicide remnant or its breakdown products.

Figure 1:

Euglossa dilemma males collecting triclopyr herbicide from a sprayed leaf of Brazilian pepper, Schinus terebenthifolius. The bee secretes lipids from its labial palps to dissolve desired volatile oils and then collects the dissolved volatiles with brushes composed of dense hair on the fore tarsi of its feet and then puts the volatile-containing lipid into storage tanks on its hind tibia for use during its courtship.

About 50 ml triclopyr was obtained from the preserve personnel, from the same container, without the colorant or the surfactant. At 10 AM on 29 May, this herbicide was put into a shallow dish with a cotton cloth and placed on a concrete wall of a Ft. Lauderdale residence where E. dilemma was common. The herbicide was checked every few hours during that day, but no bees were seen until 6:30 PM after a rain, when approximately 20 males were observed flying near to or collecting the herbicide. The colorless triclopyr became white in color after the rain. About a week later, on 4 June, approximately 15 males were hovering around the dish and eight were actively collecting the herbicide and/or its byproducts. A week later, on 11 June, after a rainy period, approximately 15 males were flying around and still collecting. The response of the bees to triclopyr was subsequently monitored every day either in the early morning or in the evening before dark. The bees continued to collect the herbicide for the next 15 consecutive days through 26 June, when eight bees were seen. During this period, the number of bees seen each day varied from single bees up to approximately 15 bees, and they were observed at the herbicide as early as 6:30 AM and as late as 7:15 PM. These male E. dilemma bees visited and collected this exposed triclopyr for 29 d.

Triclopyr is practically non-toxic to honey bees in acute contact exposure (lethal dose that kills 50 % of the sample, LD₅₀ > 100 µg/bee) (National Pesticide Information Center 2002), but a recent registration review cautioned that there is concern about chronic exposure of adult honey bees and larvae to the herbicide (EPA United States Environmental Protection Agency 2020). Some herbicides have been recently found to decrease survival in adult honey bees (Mohamed et al. 2023).

It is important to note that E. dilemma males do not ingest the triclopyr or its breakdown products. During its collection and storage, the herbicide contacts only the external surfaces of the bees even when inside the tibial storage tanks, because the storage tanks are invaginations of the cuticle (Vogel 1966). To preserve the lipids used to capture the collected compounds in their tibial tanks, the bees recycle the lipids from their tanks through their bodies and back into their lipid gland in their heads for re-use (Eltz et al. 2007). This implies that they have much more intimate contact with the collected herbicide. In the recycling process, however, the lipids are separated from the collected chemicals so that the chemicals do not enter the hemolymph, although the mechanism of this cleaving process is not known (Thomas Eltz, personal communication). In the Eltz et al. (2007) study, both a fragrance compound, methyl salicylate, collected by the test bee (Euglossa viridissima Friese) and lipids that the bee uses to collect the methyl salicylate were radiolabeled and added to the tibial storage tank. The labeled lipids began to appear in the labial gland after 1 h. After 2 days the amount of labeled lipid in their tibial tanks almost disappeared and appeared in the labial glands in the heads of the bees. In contrast, the amount of labeled methyl salicylate in the tibial tanks barely declined and the compound was not detected in the labial glands. The recycling process helps concentrate the collected compounds, while protecting the bees from their potential toxicity. Triclopyr, therefore, seems unlikely to harm male bees that collect and store it because it is not considered to be very toxic to bees and because of the ways in which these orchid bees collect and manage fragrance chemicals.

The bees probably collected this herbicide because male orchid bees are genetically programmed to collect particular odorant chemicals that are essential for their courtship and successful mating (Henske et al. 2023; Zimmermann et al. 2009). These compounds are major components of the fragrances of the orchid mutualists of the bees and other volatile sources (Pemberton and Wheeler 2006 and references therein). Some of these chemicals are used successfully to create baits to attract, census, and even discover new species of orchid bees (Dressler 1982; Janzen 1971). It may be that triclopyr and or its breakdown products may mimic compounds collected and used by male E. dilemma in their courtship.

The rate of breakdown of the ester form of triclopyr found in the applied herbicide is quite sensitive to conditions (Cessna et al. 2002), but under sunlight its estimated half-life is on the order of 12 h (McCall and Gavit 1986). Due to the delayed responses of the male orchid bees to the cleaned up spilled triclopyr at Fern Forest (bees were not observed to come to the spill until the following day), and to the triclopyr exposed in the Ft. Lauderdale residence (bees did not appear until 8 h after its placement), it appears that the bees were attracted to breakdown products of the herbicide more than triclopyr itself. Also, as the days passed smaller and smaller quantities of triclopyr would be left due to light-related breakdown. Even with a 1-d long half-life instead of 12 h for the triclopyr, less than 0.01 % would remain after 12 d when 17 bees still came to collect the material. The initial breakdown products of triclopyr are TCP (3,5,6-trichloro-2-pyridinol) and TMP (3,5,6-trichloro-2-methoxypyridine) (Cessna et al. 2002). TCP rapidly degrades into nonhalogenated, low molecular weight organic acids (Woodburn 1993).

Which of the triclopyr and its breakdown molecules provokes E. dilemma into collecting it is an essential question. Presumably, triclopyr and/or its derivatives bind to and activate the same chemical sensing proteins that are stimulated by its needed natural fragrances and that provoke the attraction and collecting behavior. The main classes of these proteins are the odorant receptors (OR) and odorant binding proteins (OBP) (Venthur and Zhao 2018). A study of E. dilemma antennal proteins found 86 versions of OR and 10 of OBP (Brand et al. 2015), which would include proteins with functions related to fragrance collection as well as other olfactory responses.

Without knowing which proteins are specifically responsible for this behavior or their structures, we can look for clues in similarities between the structures of these synthetic molecules and the structures of natural fragrances. Similarity in structure is correlated with a similarity in binding affinity. Many measures of molecular similarity have been defined. Here we choose a simple measure, based on the maximum common substructure (MCS) metric (Cao et al. 2008), developed to predict which molecules in a library of candidates are likely to have similar biological activity as a known drug or endogenous compound. The MCS analysis identifies the largest contiguous fragment that two molecules share, which can be thought of as the intersection of the two molecular structures. Dividing the size (defined as the number of atoms excluding hydrogen) of this fragment by the size of the union of the two structures (the common structure, plus the dissimilar sites in both molecules) gives an MCS-Tanimoto score that ranges from 0 to 1, with 1 indicating identical structures. Two molecules of equal size whose maximum common substructure comprises 50 % of the structures of both molecules would have an MCS-Tanimoto score of 0.333; if the MCS constitutes 2/3 of each structure the MCS-Tanimoto score would be 0.5. The Similarity Workbench function of the online ChemMine Tools server (https://chemminetools.ucr.edu; Backman et al. 2011) was used for these calculations.

The structure of the derivative TMP (3,5,6-trichloro-2-methoxypyridine) was compared with structures of 24 fragrance molecules that E. dilemma is known to collect in Florida, which are major components of the bee’s orchid mutualists (Pemberton and Wheeler 2006). We focus on TMP and not triclopyr itself or other derivatives for several reasons. The time course for attraction and collection by the bee as described above suggests that a derivative, and not the parent compound, is responsible for the effect. The breakdown product TCP, as noted above, is known to decompose quickly into small molecules that are much simpler than the known fragrance molecules. TMP, as a smaller molecule than triclopyr, will tend to be more volatile and so will be more available in the air to be sensed by the bees. In any case, the fragrance compounds discussed below as having the greatest similarity with TMP also had high scores for similarity with triclopyr and TCP.

The structure of TMP also was compared with HNDB (2-hydroxy-6-nona-1,3-dienyl-benzaldehyde), for which E. dilemma and its sibling E. viridissima show a marked difference in degree of preference that has been traced to a single OR protein (Brand et al. 2018, 2020). Of these 25 compounds, 16 show negligible similarity to TMP (with a maximum common substructure of two or fewer sites). Eight out of the remaining nine, whose names and similarity indices relative to TMP are given in Table 1, contain an aromatic ring. Although there is no well-defined threshold of molecular structural similarity that translates to similarity in binding strength, it is plausible that molecules sharing over half of their molecular structure could have comparable binding affinities to the same protein. Four of these compounds share greater than half of their structures with TMP as indicated by MCS scores exceeding 0.333: 1,4 dimethoxybenzene (also known as hydroquinone dimethyl ether, HDE), eugenol, and the two isomers of methyl-p-methoxycinnamate, with index values of 0.5, 0.4375, and 0.3889, respectively. To put these values in some context, the two odorant molecules found to bind most strongly to the odorant binding protein OBP14 in Apis mellifera L. (Spinelli et al. 2012), eugenol and methyl cinnamaldehyde, share an MCS similarity score of 0.4375 and bind with nearly identical affinities. It is worth noting, however, that the OR tends to be more specific in their binding partners than the OBP (Brand et al. 2018). The experimental observations of E. dilemma males suggest that the attraction to TMP or another triclopyr by-product is quite strong relative to that of the natural fragrance molecules. However, it should be acknowledged that the behavior reflects not only the potency of triclopyr or other components as a mimic of fragrance molecules, but also the concentrations released into the air during these observations, which may exceed the quantities typically released by natural sources.

TMP, the principal breakdown product of triclopyr, is molecularly most similar to four chemicals collected and stored by male E. dilemma in Florida (hydroquinone dimethyl ether/HDE, eugenol, methyl-E-p-methoxycinnamate and methyl-Z-p-methoxycinnamate) (Pemberton and Wheeler 2006) (Table 1). Interestingly, HDE is a principal fragrance component of three E. dilemma orchid mutualists in tropical America (Lycaste aromatica Lindl., Lycaste cruenta Lindl., and Gongora armeniaca Rchb.f; Orchidaceae). Eugenol is a principal component of two of the orchid mutualists (L. aromatica and G. armeniaca; Orchidaceae). The methyl methoxycinnamates are important components of the fragrances of L. aromatica, L. cruenta, Stanhopea saccata Bateman, and S. radiosa Lem. (Orchidaceae). These same four chemicals also are important components of the perfumes of L. aromatica (Figure 2). So, it does seem that triclopyr, or rather its breakdown product TMP, could mimic fragrance compounds needed by male E. dilemma in their courtship, which explains why this orchid bee intensively and habitually collect triclopyr.

Figure 2:

Euglossa dilemma males pollinating Lycaste aromatica, one of its orchid mutualists (note the orchid’s yellow pollinaria on the abdomen of the bee). Triclopyr breakdown product TMP (3,5,6-trichloro-2-methoxypyridine) is molecularly similar to four fragrance compounds of this orchid, 1,4 dimethoxy benzene (HDE), eugenol, methyl-E-p-methoxycinnamate, and methyl-Z-p methoxycinnamate.

Another orchid bee was previously observed to collect a pesticide, when Eufriesea purpurata Mocsáry collected large quantities of the synthetic pesticide DDT (dichlorodiphenyltrichloroethane) in Brazil (Roberts et al. 1982). Molecular similarity between DDT (or its breakup products) and fragrance compounds needed by this Brazilian orchid bee probably also explains why this even stranger insecticide collection occurred.