Plants emit volatile organic compounds (VOCs) that function as signals in plant-plant communication, defense, and attraction of pollinators, seed dispersers, and other beneficial organisms (Dudareva and Pichersky, 2008). Insects, and in particular bees, are the most important group of plant pollinators (Knight et al., 2005; Ghazoul, 2006; Klein et al., 2007). With respect to insect-pollinated plants, pollinator visitation (or artificial pollination) is required to obtain maximal seed and fruit production (Klein et al., 2007; Klatt et al., 2014). Bees possess over 130 distinct antennal olfactory receptors and exploit olfactory cues when foraging and pollinating flowers (Chittka and Raine, 2006). The olfactory cues are important in differentiating between plant species, and in signaling the appropriate phenological stage for pollinator visits (Chen et al., 2009; Hossaert-McKey et al., 2010; Song et al., 2014). Indeed, bees exhibit innate preferences for certain floral odors (Raguso, 2008), and VOCs have been shown to act as long-distance flower location signals for pollinators (Kunze and Gumbert, 2001; Burger et al., 2010; Suchet et al., 2011) and can trigger an immediate pollinator decision to land on a flower (Lunau, 1992). However, the complex effects of non-floral VOCs or floral/non-floral VOC combinations on bee foraging and other insect behaviors are still not well understood (Junker et al., 2017), and in particular the influence of microorganisms (including viruses) on plant VOC emission is relatively under-investigated (Beck and Vanette, 2017).

Phaseolus vulgaris L. (common bean) is a plant that benefits from insect pollination with increased seed production (Ibarra-Perez et al., 1999; Kingha et al., 2012; Ramos et al., 2018; Elisante et al., 2020) and is also host to a number of viruses that can reduce seed yield. Bean common mosaic virus (BCMV) and bean common mosaic necrosis virus (BCMNV) are closely related potyviruses that are among the most prevalent viruses infecting legumes and both result in serious crop loss (Morales, 2006). Cucumber mosaic virus (CMV) is a cucumovirus with a very wide host range (Yoon et al., 2019) that causes serious epidemics and losses in cultivated common bean (Morales, 2006; Thompson et al., 2015; Worrall et al., 2015). Aphids transmit CMV, BCMV, and BCMNV, particles of which are carried between hosts on the stylets (probing mouthparts) of these insects, but the viruses are also transmitted efficiently through seed of infected bean plants (Morales, 2006; Worrall et al., 2015; Yoon et al., 2019). Viruses induce changes in plant metabolism that influence insect vector behavior (Carr et al., 2018; Mauck et al., 2016). For example, CMV, BCMV, and BCMNV induce biochemical changes in common bean that alter the feeding behavior of aphid vectors (Wamonje et al., 2020), and CMV induces alterations in the emission of VOCs that attract aphid vectors to infected hosts (Mauck et al., 2010). Such virus-induced changes in VOC emission by infected hosts are thought to contribute to an increase the rate of aphid-mediated virus transmission (Mauck et al., 2016; Carr et al., 2018; Donnelly et al., 2019). Currently, there is a paucity of data about virus-induced changes on other community members, such as pollinators.

There is no evidence that CMV, BCMV, or BCMNV are transmitted by bees, or by other pollinators. Indeed, comparatively few plant viruses are vectored by pollinators (Hull, 2014). It was therefore surprising to discover that bumblebees (Bombus terrestris) could perceive differences in the VOC blends emitted by uninfected and CMV-infected plants of tomato (Solanum lycopersicum), and Arabidopsis thaliana (Groen et al., 2016). In the same study it was also found that bumblebees enhanced seed production by CMV-infected tomato plants to a greater extent than for mock-inoculated plants (Groen et al., 2016). Mathematical modeling indicated that if a pollinator bias in favor of virus-infected plants in wild host populations occurred, it could outweigh the benefits of resistance, and natural selection would instead favor inheritance by plants of alleles for virus susceptibility, rather than those conditioning resistance (Groen et al., 2016). In this study we investigated whether the innate preference of bumblebees for VOC blends emitted by virus-infected plants is peculiar only to the previously investigated tomato-CMV pathosystem, or if this can occur in other virus-host combinations, such as bean plants infected with CMV, BCMV, or BCMNV.

We found that naïve bumblebees have an innate preference for VOC blends emitted by common bean plants infected with three different viruses over non-infected plants. It was previously established that CMV-infected tomato plants are more attractive to bumblebees than uninfected plants, and benefit more from bumblebee-mediated pollination (Groen et al., 2016). Since CMV is not transmitted by bumblebees, it was hypothesized that this could be a “payback” or compensation to susceptible hosts (Groen et al., 2016). Mathematical modeling further suggested that in wild populations of plants that were dependent upon pollinators for reproductive success, increased pollinator visitation to virus-infected plants might outweigh the benefits of resistance and favor the persistence of alleles for susceptibility in the population (Groen et al., 2016). Our current work shows that virus-induced attraction of bumblebees is not unique or peculiar to the interaction of CMV and tomato, but extends to interactions of CMV and two other viruses (neither of which are pollinator-vectored), the potyviruses BCMV and BCMNV, with common bean. These observations in two different plant species suggest that modification by viruses of the emission of bee-perceivable plant VOCs may have arisen more than once.

Common bean and tomato have very different bumblebee pollination mechanisms. Tomato and several other solanaceous plants exhibit “buzz” or “pepper pot” pollination, in which foraging bumblebees grasp the fused anther cone and vibrate it to release pollen. Excess pollen provides bumblebees with a protein-rich reward but no nectar reward is provided (Glover et al., 2004). In contrast, the papilionate floral morphology of beans and many other legumes requires, almost exclusively, Hymenopteran pollinators with sufficient weight and strength to mechanically “trip” the hull and wing petals to expose the stamens and pistil. This enables bees to carry out pollination, and access nectar (a rich sucrose source) as the predominant reward (Leppik, 1966; Córdoba and Cocucci, 2011; Bailes et al., 2018). CMV is a cucumovirus with a broad host range encompassing many plant families (Yoon et al., 2019). Contrastingly, BCMV and BCMNV are legume pathogens that belong to one of the largest family of plant-infecting viruses, the Potyviridae (Ivanov et al., 2014), and are unrelated to CMV, which lies within the family Bromoviridae (Palukaitis and García-Arenal, 2003). Thus, we have shown that infection by diverse viruses can alter host VOCs to be more attractive to bees, and that the effect occurs in distantly related plants with distinct pollination mechanisms: bean (this study), and tomato (Groen et al., 2016).

Bumblebees are generalist pollinators. This is made possible by their ability to use several sensory channels to perceive multimodal cues for flower detection such as visual and olfactory stimuli (Katzenberger et al., 2013; Glover, 2014; Lawson et al., 2017) coupled to their learning capacity that allows effective association of particular floral features with nectar and pollen rewards (Cnaani et al., 2006; Gomez et al., 2008; Eisenhardt, 2014; Konzmann and Lunau, 2014). In this study, we have discovered an innate preference of bees for odors emitted from virus-infected bean plants in the absence of visual stimuli. Although this is consistent with the idea that viruses may benefit their plant hosts by encouraging pollinator visits, further studies will show how bee foraging behavior is affected by the combination of visual and olfactory cues. In multimodal experiments, Telles et al. (2017) found that flower color had the strongest effect on bumblebee foraging but that familiar floral scents enhanced the ability of bees to locate flowers with a novel color. Virus-infected plants are often stunted making their flowers less conspicuous, and it is conceivable that odor cues may be important to make them more evident to foragers.

BCMV and CMV induced changes in volatile emission by flowering bean plants that rendered the VOC blend more attractive to bumblebees. Work with yellow-faced bumblebees (Bombus vosnesenskii Radoszkowsk: Byers et al., 2014) and black bean aphids (Aphis fabae Scopoli: Webster et al., 2010) showed that neither are attracted to plants by individual VOCs but rather by specific VOC combinations. Yellow-faced bumblebee workers preferred a combination of limonene, myrcene and (E)-ocimene over any single compound (Byers et al., 2014) while volatiles within a blend that function as host cues to the black bean aphid become non-host cues (or repellent) when presented individually (Webster et al., 2010). Our data on VOC emission from virus-infected bean plants supports the idea that the innate olfactory preferences of insects are driven by VOC blends rather than by single VOCs. Total VOC emission from virus-infected whole plants was roughly half that from healthy plants due in part to the negative effect of virus infection on plant growth, but emission of individual VOCs was also differentially affected at both whole plant and tissue dry weight levels. Two of the most abundant compounds in the VOC blend emitted by flowering bean plants, (E)-ocimene and 4-hexen-1-ol acetate, were significantly lower in all virus-infected plants. Diminishing the proportion of these VOCs in the blend from flowering bean plants is unlikely to account for the innate attraction of bumblebees per se, as the insects found the VOCs from mock-inoculated and BCMNV-infected plants equally attractive in olfactometry tests. The similar trajectories of the loading vectors for the monoterpenes (E)-ocimene and D-linalool (Figure 2) indicate that these VOC outputs are positively correlated and illustrate their closely related biosynthetic pathways derived from geranyl diphosphate (Dudareva and Pichersky, 2000). D-Linalool, which is widespread among floral scents (Knudsen et al., 2006), influences various diurnal (Dötterl and Vereecken, 2010), and nocturnal bees (Krug et al., 2018) and other nocturnal visitors, including moths or bats (Dobson, 2006). However, less D-linalool was emitted by BCMV-infected and CMV-infected plants relative to healthy and BCMNV-infected plants, which is puzzling. However, variation in a few key VOCs in a floral scent blend can profoundly benefit plant competitiveness and reproductive success (Parachnowitsch et al., 2012), which leads us to speculate that (as is the situation for aphid-bean interactions: Webster et al., 2010) it is the blend of VOCs, rather than the amount of specific VOCs, that is most important in driving olfactory attraction.

VOCs emitted by all virus-infected non-flowering plants attracted bumblebees indicating that changes in volatile emission by vegetative tissues provide some olfactory cues in bean-bumblebee interactions. This is consistent with work with CMV-infected tomato plants where bumblebees showed similar innate preferences for flowering and non-flowering plant VOCs (Groen et al., 2016). Although it may seem counterintuitive that bees are attracted to vegetative tissue, Kárpáti et al. (2013) demonstrated that for Manduca sexta, conspecific leaf and flower-derived volatiles combined were more attractive than the flower blend alone from the nectar sources in Nicotiana attenuata and Datura wrightii, suggesting this could be a strategy for foraging insects to optimize food location using odor cues. It is also known that leaf volatiles act in tandem with floral volatiles to attract pollinators from long and short distances. Studies in dwarf palm (Chamaerops humilis) showed that green leaf VOCs attract pollinators over longer distances than floral scents (Caissard et al., 2004; Dufaÿ et al., 2004), and green leaf VOCs may also reinforce floral visual clues (Kunze and Gumbert, 2001; Kulahci et al., 2008). Also, a recent paper has shown that bumblebees can accelerate flowering by interacting with and wounding leaves (Pashalidou et al., 2020). It will be interesting to see if this wounding behavior is influenced by plant VOCs, and if it would be affected by a plant’s infection status, since in nature plants are affected by multiple biotic and abiotic stresses (Gruden et al., 2020).

Finally, revealing further insights into the effects of virus infection on the emission of pollinator-perceivable VOCs by plants may be useful in devising strategies to increase pollination services for crops, and for understanding how pathogens or other stresses may affect pollinated wild plants. Around 87.5% of flowering plants depend upon animal pollinators including many crops (Ollerton et al., 2011), and populations of pollinators, in particular those of bees, are under threat (Potts et al., 2010; Godfray et al., 2015). Studies such as ours on bee-attracting VOCs may aid in improving pollinator service for crops through plant breeding or other methods.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

NM, JCr, AM, FW, and BG conceived the work. NM and JCu collected the data. AM and NC carried out statistical analysis. NM and AM conducted the data analysis. NM, AM, and JCr drafted the manuscript. All authors contributed to manuscript editing and gave final approval for submission.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.