The western honey bee, Apis mellifera L. (Hymenoptera), is arguably the most important pollinator worldwide. While feeding, A. mellifera uses a rapid back-and-forth motion with its brush-like mouthparts to probe pools and films of nectar. Because of the physical forces experienced by the mouthparts during the feeding process, we hypothesized that the mouthparts acquire wear or damage over time, which is paradoxical, because it is the older worker bees that are tasked with foraging for nectar and pollen. Here, we show that the average length of the setae (brush-like structures) on the glossa decreases with honey bee age, particularly when feeding on high-viscosity sucrose solutions. The nectar intake rate, however, remains nearly constant regardless of age or setae length (0.39±0.03 μg s−1 for honey bees fed a 45% sucrose solution and 0.48±0.05 μg s−1 for those fed a 35% sucrose solution). Observations of the feeding process with high-speed video recording revealed that the older honey bees with shorter setae dip nectar at a higher frequency. We propose a liquid transport model to calculate the nectar intake rate, energy intake rate and the power to overcome viscous drag. Theoretical analysis indicates that A. mellifera with shorter glossal setae can compensate both nectar and energy intake rates by increasing dipping frequency. The altered feeding behavior provides insight into how A. mellifera, and perhaps other insects with similar feeding mechanisms, can maintain a consistent fluid uptake rate, despite having damaged mouthparts.
The underlying physical mechanisms by which organisms acquire and transport liquids for feeding is of significance to a wide variety of disciplines (Gillett, 1967; Kim and Bush, 2012; Yang et al., 2014). Several fluid-uptake mechanisms have been described, which often depend on material properties, including morphology, chemistry and physiology (Kim and Bush, 2012; Crompton and Musinsky, 2011; Lehnert et al., 2013; Harper et al., 2013). Fluid-feeding insects are of particular interest because they have mouthparts that are adapted to acquire and transport nanoliter amounts of liquids (Kim et al., 2011; Lehnert et al., 2017; Hischen et al., 2018). The western honey bee, Apis mellifera L. (Hymenoptera), for example, rapidly dips floral nectar using a tongue (glossa) covered with brush-like setae (Snodgrass, 1956; Simpson and Riedel, 1964; Krenn et al., 2005; Wu et al., 2015).
The first drinking model to elucidate the viscous-dipping feeding mechanism of A. mellifera simplified the glossa as a bald rod (Kim and Bush, 2012). Subsequently, Yang et al. (2014) proposed a model that considered the effects of the setae (erectable, brush-like structures on the glossa) and used experimental data to validate theoretical predictions on volumetric flow rate and energy intake rate. Considering that the back-and-forth movements of the glossa occur at a frequency of ∼5 Hz (Li et al., 2015) (similar to a sewing needle), we hypothesize that the high-intensity work and fast dipping frequencies cause wear or damage to the glossal setae, which could result in the gradual deterioration of nectar-loading capabilities. This situation, however, creates a paradox because it is the older worker honey bees (i.e. those likely most prone to setae damage) that forage pollen and nectar (Amdam and Omholt, 2002). We hypothesize that if mouthpart damage does occur, A. mellifera employ a method of mechanistic or behavioral compensation to overcome the structural wear of the glossa in order to maintain optimum fluid uptake rates (Abrams et al., 2015).
MATERIALS AND METHODS
Western honey bee rearing and colony maintenance
Approximately 2000 western honey bees, A. mellifera, were collected from Guangzhou, China (22°N, 112°E), where no specific collecting permits were required, and were housed in a hive with drones and a queen. The entire system was maintained at 25°C at 50% humidity, and bees were fed a 35% (w/w) sucrose solution and an inorganic salt solution (Kim et al., 2011) (Fig. 1). Pupae were removed from the hive and placed into a container (28–30°C). Upon adult emergence, individual A. mellifera workers were color coded with a unique mark on the tergum (solution composed of acetone and oil painting dye) that was used to identify their age (Huang et al., 1991).
Fifteen-day old adult A. mellifera were removed from the hive and randomly placed into beakers (170 mm×270 mm) with either 35% or 45% (w/w) sucrose solution. The sucrose solutions were based on sucrose concentration measurements from nectar (acquired with a polarimeter, Autopol IV) collected from three species of plants (Sophora japonica, Physostegia virginiana and Paulownia tomentosa) located near the bee hive. Each beaker was provided with 10 ml of their respective solution daily. The glossa of each honey bee was measured every 2 days using a light microscope (Eclapse 90i) at 4× magnification. During each measurement period, two individuals were randomly selected and placed into 100% ethanol for dehydration and further studied with scanning electron microscopy (SEM; FEI Quanta 200). Because setae length h differed along the length of a single glossa, we determined an average setae length for each individual by randomly selecting and measuring 10 setae from three uniform sections of the glossa (30 setae total): the tip (distal region), middle and base region (Fig. 1D). The average setae length of individuals within the same age class (15 days and older) was determined for both sucrose solution concentrations.
Nectar intake rate measurements
Honey bees of different ages were placed into glass beakers (1.357 cm3), sealed with a piece of wet gauze, and kept at 25°C and 50% relative humidity. Droplets of 35% and 45% sucrose solutions (10 or 20 μl volume per droplet) were dispensed using a microcapillary pipette (1–200 μl, DragonLab) onto dish feeders placed at the bottom of the glass beakers (Fig. 1F). A timer was used to determine the duration tF of ingesting a droplet and the average nectar intake rate was calculated as Q̇F = VF/tF, in which VF is the volume ingested (i.e. 10 or 20 μl).
The dipping frequency of the glossa was studied using a setup composed of a positioner, a high-speed camera (Phantom M110), a microscope (Axiostabilizer Plus, Zeiss) and an illuminant source (100 W) (Fig. 1E). A cuboid feeder fabricated with glass slides was placed between the LED light source and the high-speed camera. In addition, a 3-degrees of freedom motorized positioner (motion accuracy of 1.0 μm) was used to adjust the cuboid feeder position. During feeding observations, a live honey bee was glued via its thorax to the precision positioner so that the insect could be moved vertically, thereby allowing the mouthparts to reach the sucrose solution. The honey bees were fed through a feeder filled with either 35% or 45% sucrose solution, and the temperature was maintained at 25°C. We selected 20 bees for each nectar concentration at ages ranging from 17 to 25 days. All feeding cycles were recorded at 500 frames s−1, and the dipping frequency was calculated by first averaging five dipping cycles per individual, then using these values to determine the average dipping cycle per honey bee age. A Pearson's correlation was used to determine whether there was a relationship between dipping frequency and setae length.
Mouthpart morphology and nectar dipping rate
The average glossal setae length decreased with respect to age from 17 to 25 days (n=120) (Fig. 2). The absolute values of the slopes of the linear fits for honey bees fed 35% (k1) and 45% (k2) sucrose solutions were 3.82 and 4.34, respectively, and represent the rate of reduction of glossal setae length over time (measured as honey bee age). These values indicate that the deterioration rate of the average setae length of bees fed the 45% sucrose solution was greater than that of bees fed the less-viscous 35% sucrose solution (n=60 for both treatments). By in vivo then postmortem examination, we found that the dipping frequency increased with respect to deterioration of the glossal setae length (Fig. 2). Experimental data indicated a correlation between the average length of glossal setae h and dipping frequency f (R2=0.927), which can be fitted as h=−15.435f+212.04.
Mass intake rate
We measured the mass intake rate of honey bees of different ages and found that individuals imbibed the 35% sucrose solution at a rate of Ṁ35%=0.39±0.03 μg s−1 (n=15) (R2=0.004), and the 45% solution at Ṁ45%=0.48±0.05 μg s−1 (n=10) (R2=0.019). The mass nectar intake rate of honey bees for each sucrose solution concentration was approximately constant, independent of the average glossal setae length.
Model of nectar feeding with compensation
Model validation and energy intake and consumption
By combining the setae length measurements (Fig. 2) with Eqn 2, we calculated the theoretical volume intake rates with the dipping frequency compensation (Fig. 4). The theoretical intake rate without compensation for dipping velocity rapidly decreased from days 17 to 25 for both sucrose concentrations, during a time period when honey bees typically shift from cleaning cells and shaping combs to foraging outside the hive, when feeding efficiency is arguably most important (Seeley, 1982) (Fig. 4). The observations from the feeding trials, however, indicated that the velocity of glossa protraction and retraction increased during this time period. When considering the higher velocity of the glossa, we found that the theoretical nectar intake rate (0.39±0.03 μg s−1) when feeding on the 35% sucrose concentration stayed relatively consistent with the actual nectar intake rate recorded from the experiments (0.37±0.02 μg s−1). A similar pattern was observed when honey bees fed on the more viscous 45% sucrose concentration, where the theoretical volume intake rate, considering velocity compensation, matched the experimental value of 0.48±0.05 μg s−1. The calculations indicate that older honey bees can compensate for the lower nectar intake of shorter glossal setae by increasing dipping frequency.
The theoretical analysis indicates that the energy intake rate is approximately 106 times as much as the power necessary to overcome viscous drag (Fig. 5). The energy dissipation caused by viscous drag, therefore, can be ignored. Notably, the energy intake rate of 45% sucrose solution is 1.31 times that for the 35% concentration; in other words, feeding on higher nectar concentrations provides a higher net energy intake (Fig. 5). We measured the sucrose concentration in the nectar of wild collected plants and found that the average concentration was approximately 35%, which is similar to the nectar concentration found in other bee-pollinated plants (Yang et al., 2014; Kim et al., 2011), but not as high as the theoretically preferred concentration, for which the energy intake rate would be higher.
Adult honey bees that emerge during spring in temperate regions have a mean lifespan of approximately 25–35 days, depending on complex dynamics involving biotic and abiotic factors (Seeley, 1982). Workers of A. mellifera nurse the brood and perform other tasks in the hive during the first 2 weeks; after that, they shift to foraging for nectar and pollen (Amdam and Omholt, 2002) (Fig. 4). As indicated in this study, the glossal setae, which are responsible for trapping nectar during the feeding process (Wu et al., 2015), are shorter and less effective at capturing nectar in older honey bees that are tasked with foraging for nectar. This situation is puzzling – why would individual workers with shorter setae be responsible for foraging for nectar? This study indicates that they have evolved a feeding mechanism that compensates for the damage to glossal setae by increasing dipping frequency.
We found a discrepancy between the theoretical results and natural nectar concentrations on which bees feed, which can be interpreted in at least two ways. First, nectar of higher viscosity causes a faster wear rate of the glossal setae, which likely occurs to a greater extent in the wild compared with the experimental results because of the high intensity of daily work. If glossal setae wear down at a higher rate |k|, the glossa would eventually degrade to a bald stick. For a bald stick, the Landau–Levich–Derjaguin theory predicts the volumetric intake rate will be Q̇T≈μ−1/6 (Kim et al., 2011), which is approximately 100 times less than for the hairy stick model (Yang et al., 2014); therefore, feeding on thicker nectar might have catastrophic consequences for honey bees. In addition, a reduced volumetric intake rate would impact the pollination rate because the bees would have to spend more time feeding and less time actively pollinating plants. A second reason for the discrepancy pertains to the hypothesis that natural selection favors flowers that maintain a lower sucrose concentration in order to keep their pollinators hungry, thus requiring a higher rate of flower visitation, which would increase pollination rates (Kim et al., 2011). These two aspects are not mutually exclusive, and ultimately result in natural selection favoring a nectar concentration that optimizes pollination.
Natural selection would favor honey bees that feed quickly and efficiently because of the threat of predators and other economic necessities (Roubik and Buchmann, 1984). Honey bees, therefore, have to meet the contradictive demands of keeping visiting time short and maintaining an optimal nectar intake rate. By increasing the dipping frequency, both demands could be satisfied. We are unsure whether foragers possess an adaptive neural mechanism to adjust the feeding frequency, as this requires further study. Considering that the dipping-regulation strategy might be important to the co-evolution of flowers and honey bees and other nectar-feeding insects that have mouthparts prone to wear, we anticipate this strategy could inspire maintenance plans for performance compensation in human-engineered devices that have easily worn appendages.
We thank Mr Lianhui Shi from China University of Geosciences (Beijing) for his contribution to capture SEM images.
Methodology: Y.C., Y.Y., M.S.L.; Software: C.L.; Validation: Y.C.; Formal analysis: Y.C.; Investigation: C.L., M.S.L.; Resources: S.Y.; Data curation: M.S.L.; Writing - original draft: J.W., M.S.L.; Writing - review & editing: M.S.L.; Supervision: S.Y.; Project administration: J.W., C.L., Y.Y., S.Y.
This work was supported by research grant from Sun Yat-Sen University for Bairen Plan (contract number 76200-18841223); the National Natural Science Foundation of China (grant no. 51905556); and the Open Project of Henan Key Laboratory of Intelligent Manufacturing of Mechanical Equipment, Zhengzhou University of Light Industry (No. IM201904).
The authors declare no competing or financial interests.