The selection of appropriate food resources by bees is a critical aspect for the maintenance of their populations, especially in the current context of global change and pollinator decline. Wild bees have a sophisticated ability to forage selectively on specific resources, and can assess the quality of pollen using contact chemosensory perception (taste). While numerous studies have investigated the detection of pollen macronutrients in bees and their impact on bee health and reproductive success, only a few studies have described the gustatory responses of bees towards specialized metabolites. In addition, these studies mostly focused on the response to nectar and neglected pollen, which is the main food resource for both bee imagines and larvae. Whether bees have the ability to detect specialized toxic metabolites in pollen and then rapidly adapt their foraging behavior to avoid them is very little studied. In this study, we tested whether pollen specialized metabolites affect bumblebees at both the micro-colony and individual levels (i.e. bioassays using supplemented pollen), and whether foragers detect these specialized metabolites and potentially display an avoidance behavior (i.e. preference tests using supplemented syrup). Bumblebees were fed with either amygdalin-, scopolamine- or sinigrin-supplemented pollen diets in ratios that mimic 50%, 100% and 200% of naturally occurring concentrations. We found no effect of these specialized metabolites on resource collection, reproductive success and stress response at the micro-colony level. At the individual level, bumblebees fed on 50%-amygdalin or 50%-scopolamine diets displayed the highest scores for damage to their digestive systems. Interestingly, during the preference tests, the solution with 50%-scopolamine displayed a phagostimulatory activity, whereas solution with 50%-amygdalin had a deterrent effect and could trigger an active avoidance behavior in bumblebees, with a faster proboscis retraction. Our results suggest that regulation of toxin intake is not as well-established and effective as the regulation of nutrient intake in bees. Bees are therefore not equally adapted to all specialized pollen metabolites that they can come into contact with.

Developmental_parameter.txt and Histo_scoring.txt

How specialized metabolites can impact pollinator behavior, performance and health was investigated by the use of a control diet as well as amygdalin-, scopolamine- and sinigrin-supplemented diets (i.e. test diets). The test diets contained chemicals mixed with the control diet in ratios that mimic 50%, 100% and 200% of the naturally occurring concentration. The control diet consisted of ground pollen loads with a dominance of Salix sp. mixed with inverted sugar syrup (BIOGLUC®, Biobest). 

The experiments were conducted at the University of Mons from February 2015 to May 2016. A first run of bioassays was performed in 2015 for amygdalin (i.e. four treatments; control, 50%-amygdalin, 100%-amygdalin and 200%-amygdalin), and a second run in 2016 for scopolamine and sinigrin (i.e. seven treatments; control, 50%-scopolamine, 100%-scopolamine, 200%-scopolamine, 50%-sinigrin, 100%-sinigrin and 200%-sinigrin). Ten queenless B. terrestris micro-colonies were established for each treatment using workers from five different colonies (Biobest bvba, Westerlo, Belgium) that were equally distributed among the treatments to ensure homogeneity of origin. A total of 110 micro-colonies were then monitored for all experiments. Each micro-colony was composed of five two-day-old workers placed in different plastic boxes (10 cm × 16 cm × 16 cm) in a dark room at 27°C and 76% relative humidity. The micro-colonies were fed ad libitum with sugar syrup (BIOGLUC®, Biobest) and pollen candies that were freshly prepared and renewed every two days (0.5 g, 1.0 g or 1.5 g depending on the age of the micro-colony) to avoid nutrient alteration and drying out during the experiment. Pollen and syrup collections were measured by weighing pollen candies and syrup container before their introduction into the micro-colony and after their removal. Ejected larvae were removed from the micro-colony; workers that died during the experiment were removed and replaced. Syrup and pollen supplies as well as micro-colonies monitoring were done in the darkroom under red light during the 35-day period following the first episode of egg laying of a worker. At the end of the experiment, workers were weighed. The total mass of workers was expressed as the sum of the weights of the five workers in each micro-colony, taking into account the time they spent in the micro-colony in case of death and replacement. The nest was then carefully dissected, and the number and mass of individuals were recorded for each brood stage. 

Developmental_parameter.txt - Feeding response and micro-colony development were evaluated based on: (i) composition (i.e. number of eggs, non-isolated larvae, isolated larvae, pupae, non-emerged and emerged drones) and fresh weight of offspring, (ii) larval ejection (i.e. number of larvae, alive and dead, removed from the nest by workers), (iii) pollen collection (i.e. amount of pollen consumed and stored) (fresh matter), (iv) pollen efficiency (i.e. the weight of hatched offspring divided by the total pollen collected per micro-colony), (v) syrup collection (i.e., amount of syrup consumed and stored) and (vi) pollen dilution (i.e. the total syrup collected divided by the total pollen collected per micro-colony). All weight parameters (i.e. brood weight, pollen collection, and syrup collection) were standardized by the total mass of workers in the micro-colonies to avoid potential bias from worker activities (i.e. consumption and brood care).

Histo_scoring.txt - Tissues for histological evaluation were prepared following the method described by Vanderplanck et al. (2020). For each treatment, four bumblebee individuals were randomly collected from the different micro-colonies and cold-anesthetized. Their abdomens were cut and incised to facilitate the fixation (Duboscq–Brazil fluid), dehydration and paraffin-embedding processes. Transverse serial sections of 5 µm thicknesses were performed with a microtome (Reichert-Jung® 2040 microtome) with the use of a softening agent (Mollifex^TM), and placed on silane-coated glass slides. After rehydration, the sections were stained with Masson's Trichrome staining method.

A single-blind microscopic evaluation was carried out using a research optical microscope (Leitz® Orthoplan). This allowed for eliminating biases due to knowledge of treatment. The parameters evaluated for damage score were the common histopathological alterations in the digestive tract, namely: (i) disorganization or loos of the brush-like border, (ii) vacuolization of the epithelial cells (hydropic degeneration), (iii) interstitial edema, (iv) apoptosis, and (v) necrosis. All parameters were scored from 0 (no damage) to 5 (extensive changes), except necrosis parameter that was scored from 0 to 6 (see original publication for criteria and score details). When necrosis parameter was set to at least 4 (i.e. sublethal damage), all other parameters were automatically set to the maximal value (5). Analysis was made of the damage score for each of the parameters on one hand, and of the total sum of damage scores (TDS) of the five parameters on the other hand. Thus, the TDS had a minimum possible total damage score of 0 and a maximum possible total damage score of 26.

Detection.txt

We tested the hypothesis that bumblebees can detect the specialized metabolites using preference tests following the protocol from Ma et al. (2016). For each treatment, 15 bumblebee individuals were randomly collected from five different colonies (i.e. three bumblebees per colony) and starved for 2-4 h in plastic vials (70 mm long, 25 mm inner diameter) in the rearing dark room at 27°C and 76% relative humidity. After this starvation period, bumblebees were transferred into a holding tube where they were able to move freely. The holding tube consisted in a modified 15 mL centrifuge tube fixed on a polystyrene holder as described in Ma et al. (2016). After a habituation phase of 3 min, the trial started and was recorded with a digital Dino-lite USB microscope camera fixed 5 cm above the tip of the holding tube. The trial was recorded using the software Dinocapture 2.0, with a 26.7 frames.sec^-1 and a 25X magnification rate. A drop of sugar syrup (BIOGLUC®, Biobest) was presented to the bumblebee using a 1-mL syringe. Individuals that did not consume the syrup within 5 min were discarded. For responsive individuals, test solutions were presented using a 100 µL micro-capillary tube connected to a pumping system to ensure the presence of a permanent droplet of test solution at the top of the micro-capillary tube (Ma et al., 2016). Test solutions were prepared by diluting the commercial powders directly in sugar syrup (50%, 100% and 200% of the naturally occurring concentrations). The control solutions consisted of pure sugar syrup (negative control) and a 1mM quinine solution (positive control) that was proven to have a deterrent effect (Ma et al., 2016). A total of 165 workers (i.e. 15 workers per treatment and 11 treatments namely, negative control, positive control, 50%-amygdalin, 100%-amygdalin, 200%-amygdalin, 50%-scopolamine, 100%-scopolamine, 200%-scopolamine, 50%-sinigrin, 100%-sinigrin and 200%-sinigrin) have been tested.

The 2-min test phase started as soon as the bumblebee's proboscis contacted the test or control solution inside the micro-capillary tube. The lengths of liquid inside the micro-capillary tube were measured before and after the test phase to calculate the volume of solution consumed. The volume of solution consumed as well as the number of feeding bouts, the cumulative duration of the feeding bouts, the total duration of effective feeding (i.e. contact with test or control solution) and the duration of the first contact (i.e. before the first proboscis retraction) were used to evaluate the phagostimulatory or the deterrent activity of the compounds tested. A feeding bout was defined as a contact between the extended proboscis and the test solution for at least 5 sec (French et al., 2015).