We used the buff-tailed bumblebee (B. terrestris) as a model species even though it is considered a resistant species (Martinet et al., 2021a) because it is easy to manage and available all year. Colonies were bought at Biobest NV (Westerlo, Belgium). We reared the different colonies in plastic boxes (8 cm × 16 cm × 16 cm) provided by Biobest NV (Westerlo, Belgium) at their optimal temperature of 26°C with 50–60% of relative humidity (e.g., Vanderplanck et al., 2019). Colonies were maintained in constant darkness and were fed with BIOGLUC (Biobest NV, Westerlo, Belgium) as sugar resources and pollen of Salix sp. as lipid/protein resources (Vanderplanck et al., 2019). Pollen loads were purchased from the “Ruchers de Lorraine” company. New pollen candy (pollen mixed with BIOGLUC) was provided every 2 days, with previous candies removed to prevent decay. Males were produced from queen-right colonies (colonies developing normally with a queen) and were isolated in plastic boxes without workers. To ensure that all males were exposed to the exact same conditions before our experiments, we maintained all males at 26°C and 50–60% relative humidity directly after emergence until their maturity. All males were 7–10 days old at the time of the experiment. This period corresponds to the peak of pheromones production of mature males (Baer, 2003; Coppée et al., 2011; Ayasse and Jarau, 2014).

All queens and males were conditioned in standard plastic box (16 cm × 11 cm× 9 cm) with ad libitum pollen candies and sugar under controlled lab conditions (26°C and 50–60% of humidity). Queens were isolated in a standard plastic box (16 cm × 11 cm × 9 cm) with ad libitum pollen candies and sugar solution (Biogluc, Biobest, Waterloo, Belgium) 1 day before the mating experiment to improve the mating success (Röseler, 1985).

Bumblebee males (n = 75) were thermally exposed following the protocol of Martinet et al. (2015). Briefly, males were placed in an incubator at 40°C (typical heatwave temperature) until the “Heat Stupor” state. This state is recognizable by some behaviors: (i) the bumblebee falls upside down and is unable to turn back on its legs, (ii) it loses all its reflexes, and (iii) it has muscular spams in its legs. After heat stress, survival males recovered 1 day in plastic boxes at 26°C with ad libitum access to nutritive resources (i.e., sugar) before being presented to virgin queen for the mating assays.

A subset of the control and heat stressed males (n = 15 and n = 15, respectively), were killed for pheromones quality check. Cephalic labial gland secretions (CLGS) were extracted in 400 mL of heptane following a protocol adapted from De Meulemeester et al. (2011). We used gas chromatography-mass spectrometry using a Finnigan GCQ quadrupole system (GC/MS) with a non-polar DB-5 ms capillary column [5% phenyl (methyl) polysiloxane stationary phase; column length 30 m; inner diameter 0.25 mm; film thickness 0.25 μm] to determine the qualitative composition of CLGS. The identification was confirmed with analytical standards. We performed a relative quantification for CLGS samples with a gas chromatograph Shimadzu GC-2010 system (GC-FID) equipped with a non-polar SLB-5 ms capillary column [5% phenyl (methyl) polysiloxane stationary phase; column length 30 m; inner diameter 0.25 mm; film thickness 0.25 μm] and a flame ionization detector.

Virgin queens of B. terrestris are known to be very reactive to male pheromonal secretions in solution (Lecocq et al., 2015). We performed ethological tests based on the protocol established by Terzo and Sinkevich (2005) and updated by Lecocq et al. (2015). Virgin queens were stored individually in breeding conditions (i.e., 26°C, 50–60% of relative humidity and food resources ad libitum). To minimize potential external disturbances, ethological tests were carried out in the breeding room. All ethological tests were filmed using a camera (HC-100, Panasonic, Osaka, Japan) positioned at 81 cm above the experimental device. Individuals were introduced one by one into the apparatus (70 cm × 70 cm × 8 cm). The device was divided into four square areas (35 cm × 35 cm), henceforth referred to as “quarters.” A plastic basket (85 mm × 85 mm × 24 mm, VWR International, Fontenay-sous-Bois, France) was placed in the corner of each quarter of the apparatus containing a piece of filter paper inside (1.5 cm × 1.5 cm) soaked with a 2 μl sample of pheromonal secretions extract from control or heat-exposed males previously used for the CLGS quality check. Filter papers soaked in heptane (the solvent used for CLGS extraction) or not soaked were used as controls to test the attractiveness of the heptane itself. Therefore, the virgin queens were able to choose between four quarters during a same experiment: two quarters with either heptane or an unsoaked paper and the two other quarters with pheromonal secretions from either a control or heat-exposed male (Figure 1). Each young virgin queen was tested with a different sample from control or heat-exposed males randomly selected among the extracted pheromones of the 30 males. After each test, baskets and filter papers were removed, the apparatus was cleaned with ethanol and then ventilated for 5 min to ensure complete ethanol evaporation. Filter papers were placed randomly in the quarters to avoid a potential quarter preference factor. Each test was performed for a 5 min period under red-light conditions. The first 2 min correspond to the time required for virgin queens to be receptive to the new environment and were therefore removed from the analysis (Lecocq et al., 2015). During the last 3 min, we recorded the time spent by the queens in each quarter of the apparatus (we considered that the queen was in a quarter when more than half of its body was in this quarter). Before experiment, neutrality tests (n = 30) were carried out to check that no quarter was more attractive than others. Ethological bioassays were carried based on the CLGS of 30 males of B. terrestris (heat-exposed group n = 15; control group n = 15) and with 30 virgin queens, both from different colonies to avoid brother-sister mating.

Reproduction and hibernation of bumblebee queens was performed following Lhomme et al. (2013). For the reproduction experiment, we used 50 virgin queens (2-day old) provided by Biobest (Westerlo, Belgium). Each queen was placed in a flight cage (35 cm × 35 cm × 60 cm) exposed to natural light at room temperature with three different males (7–10 days old) of the same condition (control or heat-stressed) to stimulate the mating. To avoid genetic and behavioral bias, we used non-related males to virgin queens from different standardized colonies provided by Biobest (Westerlo, Belgium). Once a queen was mated with a male, the other males were removed from the device to avoid any disruption during the mating. After a successful mating (±30 min), the queen was placed in a plastic box (16 cm × 11 cm × 9 cm) during 1 week with ad libitum pollen and sugar before hibernation.

For the hibernation, bumblebee queens (n = 50) were placed individually in a match box with a moist cotton to maintain humidity. Those boxes were maintained at 5°C for a period of 2 months (Beekman et al., 1998). After the hibernation, the queens were placed in a fly cage (Yoon et al., 2010) (30 cm × 30 cm × 30 cm) for 1 week at room temperature with ad libitum pollen and sugar. They were subjected to a photoperiod of 8 h of light and 16 h of dark for 3 days and then a phase of 24 h of full light to exit hibernation (Yoon et al., 2010). Finally, all the queens were placed individually in a rearing box (16 cm × 11 cm × 9 cm, Biobest NV, Westerlo, Belgium) to initiate their colony.

Queens were reared with optimum temperature conditions (26°C) and constant humidity (50–60%) in rearing plastic boxes (16 cm × 11 cm × 9 cm) in complete darkness. They were fed ad libitum with Salix sp. pollen (Ruchers de Lorraine), a diet of optimal quality (Vanderplanck et al., 2019) with sugar syrup (Biogluc, Biobest, Westerlo, Belgium). Prior to bumblebee feeding, pollen loads were crushed and mixed with sugar syrup (Biogluc, Biobest). Each pollen candy was weighed and replaced every 2 days and the removed candies were weighed to assess pollen consumption. During all the experimentations, syrup consumption was measured for each colony.

Colonies were dissected after 2 months of development (i.e., 2 months after the first eggs were laid). We selected this experiment time because it corresponds to a classic lifespan of commercial bumblebee colony and this time allows us to have a sufficiently grown colony to assess development parameters (Cnaani et al., 2002). We dissected the colonies using the following parameters defined by Taseï and Aupinel (2008): (a) total pollen and syrup consumption which gives a proxy on the larval development; (b) total mass of the brood; (c) number of new emergences (workers and possible reproductives) and (d) mortality (i.e., number of dead individuals divided by total individuals including initial ones and new-emerged ones) which gives indication on the workforce, brood care behavior and resilience (i.e., ability to respond to environmental stresses). We determined mass and number of each larval stages and emerged individuals (i.e., brood composition) considering separately the different brood stages, namely eggs, non-isolated larvae, isolated and pre-defecating larvae, isolated and post-defecating larvae, pupae, non-emerged and emerged males.

For ethological data, all the videos were analyzed by using the BORIS v.7.4.7 software (Friard and Gamba, 2016). The time spent in each quarter of the apparatus was analyzed by using a non-parametric Chi2 as the normality and homoscedasticity were not respected. A Kruskal–Wallis test was performed to test the difference between the different quarters followed by a post hoc Kruskal–Wallis multiple comparison test to identify which quarters are different from the others.

We performed statistical comparative analyses of the colony performances using R 3.5.1 environment (R Core Team, 2018). Significant tests (Wilcoxon Mann–Whitney) were run to assess differences between groups for each parameter. For the analysis of brood composition (relative mass of the different larval stages), we performed a permutational multivariate analysis of variance (perMANOVA) on arcsine-transformed data using the Bray-Curtis dissimilarity matrix and 1000 permutations (“adonis” command, R-package vegan, Oksanen et al., 2018). Prior to these tests, the multivariate homogeneity of within-group covariance matrices was verified using the “betadisper” function. When a significant difference was detected, we performed multiple pairwise comparisons with an adjustment of p-values (Bonferroni correction). Finally, separate t-test and Tukey post hoc tests were conducted to assess the effect of heat exposure on each developmental stage.

Quality check of CLGS revealed no difference among males and a sufficient load of pheromonal compounds including the main compound 2,3-dihydrofarnesol. Gas chromatography analyses revealed a typical chemical fingerprint of B. terrestris. Our results confirm that the 7–10 days-old males can be used to assess the attractiveness of their sexual attractive mixture.

Neutrality tests performed before the experiments showed no significant differences between quarters of the apparatus (Kruskal–Wallis multiple comparison test, p = 0.31), meaning that, without CLGS extract, young virgin queens have no preference for a specific quarter. During attractiveness tests, the time spent by the young virgin queens was not significantly different between the two control tests (i.e., performed with filter paper soaked in heptane or with unsoaked filter paper) (Kruskal–Wallis multiple comparison test, p = 0.19, Figure 2). However, a significant difference was detected between male secretions (from both control and heat-stressed groups) and heptane (Kruskal–Wallis multiple comparison test, p < 0.01) and between male secretions and non-impregnated filter paper (p < 0.01). Finally, no difference was detected between CLGS extracts from control and heat-exposed groups (p = 0.16).

Overall, our results show that heat exposure on males did not have an impact on mating time or copulation behavior. During nest development and colony growth, behavioral differences between both conditions (i.e., heat-exposed and control groups) as well as in colony dynamics and brood development were not observed. We evaluated the performance of bumblebee colonies based on colony growth, composition of brood, mortality, total pollen, and syrup collection. For both conditions (i.e., colonies from a heat-exposed male and colonies from a control male), all these parameters were not significantly affected by heat stress applied beforehand to the males (Kruskal–Wallis multiple comparison test, respectively, p = 0.67; p = 0.89; p = 0.73 and p = 0.37; Figures 3A,B). Finally, the dissection of the colony showed no significative difference in brood composition for all assessed developmental stages (Figure 3C, perMANOVA, p = 0.29).