Male bumblebees (Bombus terrestris) are more active and behaviourally flexible than workers.

doi:10.21203/rs.3.rs-8285569/v1

Download PDF

Research Article

Male bumblebees (Bombus terrestris) are more active and behaviourally flexible than workers.

https://doi.org/10.21203/rs.3.rs-8285569/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Sex role differences can influence ecological and evolutionarily important traits like activity level and behavioural flexibility. In bumblebees, females (workers) are the main foragers for the colony, whereas males (drones) have minimal responsibility. However, males become solitary foragers once they leave the colony, suggesting that increased activity level and behavioural flexibility are crucial for their survival. Here, we conducted a laboratory experiment to compare male and female bumblebees’ active time in a novel environment (using an ‘activity’ task) as well as their colour-reward associative learning ability and behavioural flexibility (with a simultaneous two-choice discrimination-reversal colour learning task). As predicted, males were more active in the novel environment than females. Males and females showed comparable performance in learning the colour-reward association, but males demonstrated enhanced behavioural flexibility when the reward contingency changed. Males’ active time may reflect their exploratory behaviour (e.g., pre-mating patrolling), and their enhanced flexibility suggests their readiness to find new profitable flowers when exploited flowers decrease in quality. These results highlight the importance of these behavioural and cognitive traits for males, which may increase their chance of finding mates and improve their foraging efficiency.

pollinator

eusocial

insects

colony life cycle

insect cognition

The expression of ecological and evolutionary behavioural and cognitive traits may vary within a species depending on the roles of the sexes in different fitness-related contexts (e.g., (Lucon-Xiccato and Bisazza 2014; Videlier et al. 2015; Ogurtsov et al. 2018; Fuss and Witte 2019). For example, in species of bumblebee, Bombus spp. which shows classic sex role differences (Beekman and van Stratum 1998; Duchateau 2004; Huth-Schwarz et al. 2011; Del Castillo and Fairbairn 2012; Zhao et al. 2021a), females are workers who forage and care for the broods during the colony development whereas males do not forage for the colony because they do not have pollen baskets, and they are more specialised for reproduction (Thomson and Plowright 1980). In other words, males have minimal responsibility before dispersal (Goulson 2003; Baer 2003; Yadav et al. 2016; Belsky et al. 2020; Zhao et al. 2021b). When males leave their nest, they rarely return home (Goulson 2003; Kraus et al. 2009; Robert et al. 2017); they then begin solitary foraging and search for mates for reproduction. In this context, some behavioural and cognitive traits may vary, such as activity level and behavioural flexibility, because they have fitness consequences (Réale et al. 2007; Dukas 2013), and thus investigating whether trait varies in expression is crucial for understanding the ultimate significance of these traits.

Active time reflects time not spending on immobile activities (e.g., resting, sleeping) (van Dixhoorn et al., 2024). Activity time has also been related to exploration, e.g. (Montiglio, Garant, Thomas, Réale 2010; Tanaka, Young, Halberstadt, Masten, Geyer 2012), which allows individuals to discover resources and gain information about their environment (Reader 2015). More active bumblebee males can explore their environment, increase their chances of finding an unrelated queen for reproduction and discover profitable flower patches (Kraus et al. 2009; Amin et al. 2012). Indeed, bumblebee males have been shown to fly longer and travel farther than workers (Bertsch 1984; Osborne et al. 1999; Chapman et al. 2003; Kraus et al. 2009; Wolf et al. 2012). As males leave their nest, they explore the outside world the first time. The proxy of locomotion or overall activity level in the first exploration can be measured using adapted, established behavioural and cognitive tasks (e.g., analogue to those like the ‘open field test', see review by (Gould et al. 2009). This type of test commonly involves placing an animal in a novel, open arena, which is typically circular, square, or rectangular in shape (Walsh and Cummins 1976). The arena may be divided into equal-sized units or marked by distances to quantify the animal’s overall locomotion and spatial utilisation. Measurements are generally taken over a predetermined time period or until the animal ceases movement (Walsh and Cummins 1976).

Behaviourally flexible, or being able to quickly adjusting to environmental changes, demands and opportunities (Dukas and Bernays 2000; Chapman et al. 2003; Coppens et al. 2010; Lea et al. 2020), is another trait that is expected to be crucial as it has also been shown to have fitness consequences (e.g., Dukas and Bernays 2000; Nicolakakis et al. 2003). When bumblebee males become a solitary forager, the cost of not being behaviourally flexible is arguably higher for them than for females who have access to resources in the colony. Accordingly, this trait may increase bumblebee males’ survival by quickly responding to changes in their environment (e.g., adapting their foraging behaviour when flower quality decreases), maximise energy gain (Bertsch 1984, Brown and Brown 2020), and in turn, gain the time searching for mates. To measure behavioural flexibility, the use of the traditional discrimination-reversal learning task may be used. This task typically included two learning phases (see Shettleworth, 2010), namely, a discrimination learning phase (DLP) and a reversal learning phase (RLP). The DLP, involves presenting two stimuli to individuals, with one stimulus associated with reward whereas the other stimulus was not. Individuals have to learn this reward contingency until they meet a stringent learning criterion (typically 80% across consecutive sessions). Once this criterion is met, the RLP begins whereby the reward contingency is switched so that the previously rewarded stimulus becomes not rewarded and the previously non-rewarded stimulus becomes rewarded. The individuals have to learn the changed reward contingency until they meet the criterion. Performance of associative learning ability in the DLP and behavioural flexibility in the RLP can be measured by the number of errors made before reaching the criterion in the respective learning phases (e.g., Tapp et al. 2003; Izquierdo and Jentsch 2012; Wascher et al. 2021).

Here, we asked whether sex differed in active time and behavioural flexibility. To do this, we designed an ecologically relevant experiment by creating a new environment (an exploration task) to measure the overall activity time of male and female bumblebees on their first time leaving the nest and using the traditional two-colour discrimination-reversal learning task to assess their associative learning ability and behavioural flexibility. Previously, a few studies have compared females' and males’ associative learning of profitable flowers’ colour and location (Dyer and Chittka 2004; Wolf and Chittka 2016; Robert et al. 2017; Frasnelli et al. 2021; Muth et al. 2021; Manning et al. 2021), and results of these studies showed that females and males show comparable performance in associative learning. Despite this, it is yet to explore differences in active time when bees first leave their nest and performance in behavioural flexibility. Based on previous findings, we hypothesised that males would show a higher overall active time on their first time leaving the nest than females. We also expected that males and females would show a comparable performance in associative learning, but males would be more behaviourally flexible than females.

Study species and housing

We conducted this experiment daily between May and August 2024. Bees from 5 colonies (3 colonies and 2 pocket hives purchased from commercial suppliers (BIOBEST from Biobest Belgium N.V., Westerlo, Belgium, and Koppert, UK). Two colonies were housed in a natural environment based at Toyota Manufacturing Deeside and later brought to the lab for the study when there were few individuals left in the colony. The other colony and pocket hives were housed in the lab. Bees were kept in a black box, mirroring their underground nest. The nest was attached to a transparent circular tunnel (diameter 4 cm), which had three shutters and an open end to capture the bees using a plunger for the experiment. The lab (room temperature 21–23^∘c) was lit with natural daylight and artificial lights. The bees had ad libitum access to syrup and pollen during the exploration task but reduced amounts during the learning task to increase their motivation to participate in the experiment. Bees (N = 69) were individually marked with a numbered tag (Toppa et al. 2021).

Ethics

While there is no general animal welfare and husbandry guideline for invertebrates, this study strictly followed ASAB and ABS animal welfare guidelines and was approved by the Division of Psychology Ethics Committees, University of Chester (no: SDPC130624). Animal welfare and husbandry were carried out daily, whereby bees were fed syrup and pollen. The bees had ad libitum access to syrup and pollen before the exploration task, but we decreased the amount of syrup and pollen given to them during the learning task to increase their motivation. Enrichments (e.g., paper sticks and tissue tiles) were provided for the bees.

Apparatus

The experiment used two apparatuses that share similar features (Fig. 1). Both apparatuses were rectangular-shaped boxes that had 10 equally divided compartments. For the exploration apparatus (Height x Width x Length: 3.3 cm x 21.8 cm x 50.9 cm, Fig. 1A), the floor of each compartment (W x H: each 4.5 x 1.2 cm) was covered with a random mosaic white-and-red square (1 cm x 1 cm) pattern. Each cell had a hole in the middle of the wall separating it from the next one, and a shutter was used to block this entrance between compartments, and at one end of the apparatus. This allowed the experimenter (PC and SD) to lift the shutter and let a bee enter/exit the experiment or go to the next compartment through the holes.

For the learning apparatus, it was used to conduct the discrimination-reversal learning task (H x W x L: 3.1 x 15.6 x 53.6 cm, Fig. 1B). Each compartment had a pair of colour stimuli (each 4 cm x 4 cm). The colour and size of the stimuli were yellow (Perspex® Yellow 250) and blue (Perspex® Blue 727); these two colours are preferred by bees, and they can clearly distinguish them from each other (e.g., Raine et al. 2006; Raine and Chittka 2008, 2012; Ings et al. 2009; Strang and Sherry 2014; Evans et al. 2017). The stimuli were horizontally positioned, one on the left side and one on the right side of the compartment, equidistant from the entrance. The design of the box and the way of presenting the stimuli horizontally aimed to enhance visual information processing during the task (Rother et al. 2023). Aside from the boxes, there was a video camera, attached to a tripod, used to record the behaviour and performance of the bees. The tripod was placed around 30 cm away from the apparatus, and the video camera was 1.5 m above the box.

Procedure

Bees went through the same standardised procedure; they participated alone in every task daily during their active time. When a bee emerged from its colony and went to the end of the tunnel that connected to their nest, s/he was brought to the task with a plunger. Bees first participated in the activity task, followed by the discrimination learning phase (DLP) and finally a reversal learning phase (RLP). Testing a bee was discontinued if s/he did not emerge from the colony on a test day due to a loss of motivation or dead, which led to a decrease in the sample size across tasks; this procedure controlled/avoided confounding variables (e.g., fluctuation in motivation, memory decay during the learning task).

Sixty-nine bees (33 females, 36 males) left their nest for the first time to freely move in a novel environment (‘the activity box’). This task was designed to measure bees’ active time in a novel environment when they first experienced it. The task started when a bee entered his/her full body in the box and ended when the bee’s full body exited the box. During the task, no shutter was used; the bee visited any compartment as s/he wished. If a bee was inactive for 15 minutes in the box, the session was terminated, and the bee was brought home using the same plunger. A bee can reattempt a session if s/he emerged from the colony 40 minutes or more after the last session. If the bee completed the session (i.e., left the last compartment), s/he was fed with 50 w/w sucrose ad libitum (a food source preferred by both sexes) (Bailes et al. 2018; Pamminger et al. 2019; Brown and Brown 2020); this aimed to assess whether they were responsive to sucrose, and the amount drunk was measured.

The bees that emerged from the colony after the activity task had an additional ‘training phase’. This training phase included two sessions that served two purposes: 1) to allow the bees to familiarise themselves with the procedure of the learning task, and 2) to ensure the bees had high motivation for the learning task. In both training sessions, a shutter was added to block the hole so that the bees had to explore both the right and left sides of the box. The shutter between the compartment and the next one was lifted every 30 seconds, and the bees had to go through all 10 compartments to be considered as passing the training phase. After each training session, the bees were rewarded with 50 w/w sucrose ad libitum (Bailes et al. 2018; Pamminger et al. 2019; Brown and Brown 2020) when s/he left the last compartment. The box was thoroughly cleaned using water and alcohol wipes before the next bee was trained with the following steps: 1) we used tissue papers to absorb bees’ dejections and remaining solutions on the flowers; 2) we used alcohol wipes to clear each compartment and all sides of the box as well as the flowers; 3) we used hot water to clean the compartments and the flowers; and finally, 4) we used tissue papers to dry all compartments and flowers and reset a session for the next bee. These cleaning steps are aimed at removing any scent left from the previous bee. After cleaning, the box was reset for the next bee.

The bees that had passed the training phase went to the learning task (full completion as an indicator of motivation). The learning task was used to measure bees’ colour-reward associative learning ability in the DLP and behavioural flexibility in the RLP. A bee participated in 1–2 sessions daily (as a motivation measure and to avoid over-feeding). Each session included 10 flower pairs (i.e., 1 pair per compartment, bees made 10 choices in total per session). The flower pair was blue and yellow. Each bee was randomly assigned to associate one flower colour with a reward (0.01ml 50 w/w sucrose) and the other flower colour with a control (0.01ml water control). The drop of each flower was placed at the centre of the flower. We decreased the size of the drops to < 1 mm in diameter, which required the bees to walk close to the stimuli and use their antenna to detect whether the drop was water or sucrose. The presentation of the flower pairs was pseudo-randomised within and across sessions. Within sessions, the same flower colour was presented on the same side for no more than two consecutive compartments to avoid bees developing side bias. The same flower colour was shown on the left side five times (out of 10) and another five times on the right side. Across sessions, the same colour pair was not presented in the same compartment more than two consecutive times. All the bees experienced the same colour sequence to control the order effect on the experience.

Bees could explore both flowers, and their first choice (indicated by at least half of his/her body on a flower) was marked as either ‘correct’ (reward) or ‘incorrect/error’ (control). When a bee chose the rewarded flower, the shutter in the middle of the divider was lifted, and the bee passed to the next compartment. When the bee made an error, s/he had to visit the other flower (i.e., correctional choice) before going to the next compartment. Choices were recorded both by direct observation and by the video camera. The criterion for completing a learning phase was that a bee had participated in the task daily and that the bee made ≥ 8 out of 10 rewarded first choices (i.e., ≥ 80% correct) in two consecutive sessions. The bee was brought back home when s/he completed the task and went to the second session if s/he emerged from the colony 40 minutes or more after the last session. After each session, the box was cleaned in four steps mentioned above in the activity task.

Bees went to the RLP the day after they had completed the DLP. The RLP had the same protocol and learning criterion as the DLP. In the RLP, the bees had to unlearn the previous colour-reward association (e.g., B + Y-), and relearn that the previously unrewarded colour became rewarded (e.g., B-Y+) until they reached the same learning criterion set in the DLP (80% choices on rewarded colour).

Behavioural Measurement

Active time in the compartments when the bee first experienced the novel apparatus was analysed (reflected the most when males leaving their nest the first time). Active time in the compartments was measured in seconds, and the recording started when a bee’s full body entered the apparatus (the first compartment) until its full body left the apparatus (the last compartment). From these recordings, we obtained the active time of each bee in each compartment, total active time across compartments (i.e., the sum of active time in each compartment), the frequency of visiting each compartment, and mean active time in each compartments (i.e., total active time in each compartment divided by the number of visits to that compartment).

The associative learning and behavioural flexibility performances were the number of errors made before reaching the learning criterion in the DLP and RLP, respectively (e.g., Tapp et al. 2003; Izquierdo and Jentsch 2012; Wascher et al. 2021). We also adjusted the learning criterion to include bees that had not returned to the task but showed significant learning, when a bee had 80% of correct responses in a single session with at least 5 consecutive correct responses or more in that session.

We additionally measured the size of each bee, which was estimated using the inter-tegular span (ITS) (Cane 1987; Hagen and Dupont 2013). We used a digital calliper to obtain the distance (mm) between the tegulae. Sucrose consumption was measured as the amount (ml) consumed after the activity task, using a syringe with a 0.02 ml graduation.

Statistical analysis

All analyses were conducted using R (version 4.4.1), and the significance level was set as two-tailed p ≤ 0.05. Data were analysed using Generalised Linear Model and Generalised Linear Mixed Model (GLMM) from the ‘glmmTMB’ package (Magnusson et al. 2017), pairwise contrasts with Tukey correction from the ‘emmeans’ package (Lenth et al.), a binomial test and individual analysis (Sokal and Rohlf 1995). All GLMM models included bee ID as a random variable to maximise model convergence. Model fits were checked with the package ‘DHARMa’ (Hartig 2025). Multicollinearity was checked after running each model using the Variance Inflation Factor (VIF) (< 5) and tolerance (0.25). These packages generated estimates, standard errors, z and p values for the results.

Activity task. a GLMM with gamma log link distribution was used to model the positively skewed, non-negative but continuous variables (e.g., active time) (Ng and Cribbie 2017). Main analyses included between-group level: 1) sex predicted the total active time in all the compartments (Table 1a); 2) sex, compartment number, and their interaction influenced the total active time in each compartment (Table 1b); 3) sex, compartment number, and their interaction in relation to the mean active time in each compartment (Table 1c). Additional analyses for the activity task included: 1) a GLMM gamma log link distribution test on sex difference in body size (Table 1e); 2) a Poisson log link distribution test to examine sex, compartment number and their interaction on the frequency of visits to each compartment, (Table 1d), with posthoc analyses uisng pairwise contrasts with Tukey corrections to compare sex differences in frequency of visits to each compartment (Table 2); 3) a GLMM gamma log link distribution test to analyse sex differences in drinking amount (Table 1f); 4) within-sex analyses using a GLMM to examine the effects of body size (ITS) and the total active time in all compartments on sucrose consumption (Table 1g-h).

Discrimination-reversal learning task. For each learning phase, two models were run to examine learning performance using GLMM Poisson log link distribution. In the first model, we only included bees that had met the learning criterion (80% correct responses for two consecutive sessions). This model included a fixed factor, sex, and the response variable was the number of errors made before reaching the learning criterion (Table 3a, 5b). We then included bees that had met the adjusted learning criterion (i.e., met the 80% learning criterion with 5 or more consecutive correct responses in a session) and reran the model (Table 3b).

Between-group analyses of each learning phase. In the DLP, we conducted between-group analyses using pairwise contrasts with Tukey corrections for multiple comparisons to examine whether females and males who had and had not completed the learning phase differed in body size (inter-tegular span) (Table 4a), sucrose consumption (Table 4b), and total active time in the activity task (Table 4c). For RLP, all bees met the (adjusted) learning criterion (i.e., completed the task), and thus, between-group analyses reflected sex differences. We ran GLMM with gamma log link distribution to examine body size in one model (Table 5c), sucrose consumption (Table 5d) in another model, and total active time in all compartments in the final model (Table 5e).

Within-sex analyses of each learning phase. For bees that had completed each learning phase, we carried out within-sex analyses to examine predictors for learning performance. We used GLMM Poisson log link distribution to examine whether their body size (inter-tegular span), sucrose consumption, and total active time in all compartments predicted their DLP learning performance in one model (Table 3c-d) and RLP performance (behavioural flexibility) in another model (Table 5f-g).

Additional analysis. For bees that had met the unadjusted DLP learning criterion (n = 15), we further used GLMM Poisson log link distribution to examine whether their DLP performance was affected by their initial choice (Y/B) in one model (Table S4e), and their assigned initial reward colour (Y/B) in another model (Table S4f). A binomial link distribution was used to examine whether inter-tegular size (ITS), active time in the activity task, and sucrose consumption would predict whether a bee completed DLP or not (Table S4g).

Performance differences between learning phases. A GLMM Poisson log link distribution was used to examine the number of errors in the last session of DLP and the first session of RLP (Table 5a), the total number of errors made before reaching the learning criterion of DLP and RLP (Table 5b). A binomial test was used to examine whether bees’ first choice on the previously rewarded colour was significantly above chance. Individual analyses were also conducted to understand whether bees had learned the DLP colour-reward association.

Sex and Activity Levels in a New Environment

The total active time was significantly longer and showed a wider variation in males (median = 11 minutes) than in females (median = 5 minutes) (GLMM: p < 0.001, Fig. 2A, Table 1a). Active time per compartment was also significantly higher in males than in females (p = 0.002, Fig. 2B, Table 1b). Bees’ active time significantly decreased over successive compartments (p < 0.001), but no interaction effect was detected between sex and compartment number (p = 0.899, Table 1c).

The additional between-group analyses showed that bees significantly decreased their mean active time in successive compartments (p < 0.001, Table 1c). Males and females had comparable mean active time in each compartment (p = 0.171). The interaction of sex and compartment number on mean active time was not significant (p = 0.103, Fig. 2E). Frequency of visits significantly decreased over successive compartments (p < 0.001). Males visited each compartment more often than females (p < 0.001, Fig. 1F, Table 1d), and the interaction effect between sex and compartment number significantly affected the frequency of visits (p < 0.001). Pairwise contrasts of sex differences in the frequency of visits to each compartment further showed that males visited the compartments close to the entrance and in the middle of the box more often (Fig. 2F, Table S1). Males and females who participated in this task had a comparable body size (ITS) (p = 0.317, Fig. 2C, Table 1e) and sucrose consumption (p = 0.828, Fig. 2D, Table 1f). The additional within-sex analyses showed that males’ sucrose consumption was not associated with body size or total active time (Table 1g), whereas females’ sucrose consumption was associated with less active time (p < 0.001, Table 1h).

Table 1

Activity task. This table shows fixed variables, estimates (est), standard error (S.E), z and p values for the response variable (a) total active time, (b) active time in each compartment, (c) mean active time in each compartment and (d) frequency to each compartment. Between-group analysis of (e) Inter-tegular span (mm) and (f) amount dank (ml). Within-sex analysis of (g) males and (h) females on their sucrose drinking amount. Bold values indicate significance, p < 0.05.
	Response variable	Fixed variable(s)	Est	S.E	z	p
a	Total active time	Sex (Male)	1.13	0.30	3.77	< 0.001
b	Active time in each compartment	Sex (Male)	0.90	0.30	3.03	0.002
		Compartment number	-0.20	0.02	-10.14	< 0.001
		Sex (Male)* compartment number	< 0.01	0.03	0.13	0.899
c	Mean active time in each compartment	Sex (Male)	0.26	0.19	1.37	0.171
		Compartment number	-0.12	0.02	-7.05	< 0.001
		Sex (Male)* compartment number	0.04	0.02	1.63	0.103
d	Frequency to each compartment	Sex (Male)	0.80	0.20	4.00	< 0.001
		Compartment number	-0.10	0.01	-8.17	< 0.001
		Sex (Male)* compartment number	-0.05	0.01	0.01	< 0.001
e	Inter-tegular span(ITS)	Sex (Male)	0.04	0.04	1.00	0.317
f	Amount drank	Sex (Male)	-0.06	0.25	-0.22	0.828
	Within-sex analysis
g	Males: Amount drank	Inter-tegular span (ITS)	0.60	0.31	1.94	0.053
g	Males: Amount drank	Active time	0.01	0.01	1.37	0.170
h	Females: Amount drank	Inter-tegular span (ITS)	0.93	0.50	1.86	0.063
h	Females: Amount drank	Active time	-0.31	0.09	-3.49	< 0.001

Sex and Performance in Learning a Colour-Reward Association

Out of 69 bees, 29 bees (13 females, 16 males) emerged and participated in DLP the next day after the activity task. 14 bees did not return after a few sessions (median = 4), and fifteen bees (8 females, 7 males) met the DLP learning criterion in 1 to 10 sessions (median = 4). Of these 15 bees, 7 bees first chose blue and 8 first chose yellow, which may reflect their innate colour preference (Gumbert 2000). However, the number of errors made was neither affected by the bees’ first choice (p = 0.413, Table 2a) nor the initial assigned reward colour (p = 0.372, Table 2b). Males made significantly fewer errors than females (p = 0.022, Table 2c) but this significance disappeared after including a few more bees that had stopped returning to the task but showed significant learning (adjusted criterion: a single session with 80% correct and ≥ 5 consecutive correct choices, p ≤ 0.031) (n = 21, n_Female = 11, n_Male = 10, p = 0.139) (Fig. 3A, Table 2d).

Table S2. DLP performance. This table shows fixed variables, estimates (est), standard error (S.E), z and p values for response variable (a) number of errors with the stringent learning criterion (80% correct response for consecutive two learning sessions), (b) number of errors with adjusted learning criterion (80% correct response in a session with 5 or more consecutive correct responses in the session). Within sex analyses: predictors of DLP performance (number of errors), the model includes fixed factors, inter-tegular span (ITS), amount of sucrose drunk (ml) and total active time for (c) males (n = 10) and (d) females (n = 9, two females lost their tag before measuring body size, and thus reduced one sample size for females). Additional analysis at the group level to examine whether DLP performance was affected by (e) a bee’s very first choice, (f) initially assigned rewarded colour, and predictors for completing DLP or not. Bold values indicate significance, p <0.05.

	Response variable	Fixed variable(s)	Est	S.E	z	p
a	Number of errors (with adjusted criterion)	Initial choice (very first choice) in DLP	0.71	0.87	0.82	0.413
b	Number of errors (with adjusted criterion)	DLP Assigned rewarded colour (blue or yellow)	0.82	0.92	0.89	0.372
c	Number of errors	Sex (Male)	-1.79	0.78	-2.30	0.022
d	Number of errors with adjusted criterion	Sex (Male)	-0.98	0.66	-1.48	0.139
e	Completion of DLP	Inter-tegular span (ITS)	1.29	1.10	1.17	0.241
		Active time	0.16	0.21	0.73	0.468
		Amount drank (ml)	23.43	11.63	2.01	0.044
Within-sex analysis
f	Males: Number of errors (with adjusted criterion)	Inter-tegular span (ITS)	-0.54	0.52	-1.04	0.298
		Amount drank (ml)	4.15	2.36	1.76	0.079
		Active time	-0.05	0.07	-0.70	0.482
g	Females: Number of errors (with adjusted criterion)	Inter-tegular span (ITS)	0.88	0.86	1.03	0.304
		Amount drank (ml)	1.87	1.34	1.39	0.163
		Active time	-0.61	0.26	-2.32	0.021

For the 29 bees that had participated in DLP, higher sucrose consumption increased the likelihood of successful completion (amount drank: p = 0.044, Table 2e). We further explored whether body size, sucrose consumption, and active time in the activity task differed across males and females who did or did not complete the DLP (Table S2). Between-group pairwise comparisons with Tukey corrections for multiple comparisons showed that the male completed group had a significantly larger body size than the male not-completed group (ITS: p = 0.008, Fig. 3B, Table S2A). Notably, the male not-completed group significantly consumed less amount of sucrose than the other three groups (Fig. 3C, Table S2B). The four groups did not differ significantly in total active time in the activity task (Fig. 3D, Table S2C). The bees that had completed the learning phase allowed us to conduct analyses regarding the possible predictors for the DLP performance within each sex. Within-male analysis showed that none of the factors predicted males’ DLP performance (n = 10, Table 2f), whereas within-females analyses showed that females with a higher total active time made fewer errors (active time: p = 0.021, Table 2g).

Sex and Behavioural Flexibility

Fifteen bees (8 females, 7 males) that had met the DLP (unadjusted) learning criterion participated in the RLP on the next day. Thirteen bees (8 females, 5 males) met the (unadjusted) RLP criterion, and all bees (8 females and 7 males) met the adjusted RLP criterion. In the first RLP session, most females (88%) and males (86%) first chose the previously rewarded colour significantly above chance (binomial exact p = 0.003, n = 15). Both females and males made significantly more errors than random on the first five choices (pooled consecutive binomial tests: females: χ2₁₆ = 41.59, p < 0.001; Males: χ2₁₄ = 30.50, p < 0.01, Note S1). Bees also made significantly more errors in the first RLP session than in their last DLP session (p < 0.001, Table 3a). Overall, bees made significantly more errors before reaching the criterion in the RLP than in the DLP (p < 0.001, Table 3b), showing that the RLP was more difficult and suggesting that bees had learned the DLP colour-reward association and were not relying on any hypothetical cues from the reward itself. In the RLP, males made significantly fewer errors than females (p = 0.023, n = 15, Fig. 4A). These results are retained if we only included bees that met the unadjusted RLP criterion (p = 0.013, n = 13) or excluded a female who made extensive errors before reaching the criterion (p = 0.049, n = 14).

All bees completed the RPL, and the between-group comparisons reflected sex comparisons. Additional between group analyses showed that no significant differences were detected between sexes in body size (ITS: p = 0.592, Fig. 4B, Table 3c), sucrose consumption (p = 0.251, Fig. 4C, Table 3d) or total active time during the activity task (p = 0.266, Fig. 4D, Table 3e). Additional within-sex analyses showed that none of these factors predicted males’ RLP performance (ITS: p = 0.085; amount drank: p = 0.093; active time: p = 0.335, Table 3f) whereas females who had a larger body size made more errors (ITS: p < 0.001; amount drank: p = 0.430; active time: p = 0.596, Table 3g).

Table 3

RLP. This table shows fixed variables, estimates (est), standard error (S.E), z and p values for response variables. (a) Comparison of the number of errors of the last DLP session and the first RLP session (reference group), including the adjusted learning criterion (80% correct response in a session with 5 or more consecutive correct responses in the session) (n = 15). (b) Comparison of the number of errors made in each learning phase (RLP or DLP (reference group)), including the adjusted learning criterion (80% correct response in a session with 5 or more consecutive correct responses in the session) (n = 15). All bees (8 females, 7 males) completed the RLP. One female lost her tag before measuring body size, and thus reduced the sample size for females. Between-group comparisons reflected sex comparisons (female as reference group), comparisons included (c) inter-tegular span (ITS), (d) amount of sucrose drank (ml), and (e) total active time in the activity task. Within-sex analysis to examine the predictors of RLP performance, for (f) males (n = 7) and (g) females (n = 7). Predictors included inter-tegular span (ITS), amount drank (ml) and total active time (minutes). Bold values indicate significance, p < 0.05.
	Response variable	Fixed variable(s)	Est	S.E	z	p
a	Number of errors	Learning session (last DLP session)	-0.98	0.10	-9.45	< 0.001
b	Number of errors	Learning phase (RLP)	0.96	0.12	7.83	< 0.001
c	Inter-tegular span (ITS)	Sex (Male)	0.04	0.07	0.54	0.592
d	Amount drank (ml)	Sex (Male)	-0.93	0.81	-1.15	0.251
e	Active time (secs)	Sex (Male)	0.43	0.38	1.11	0.266
Within-sex analysis
f	Males: Number of errors (with adjusted criterion)	Inter-tegular span (ITS)	-0.25	0.14	-1.72	0.085
		Amount drank (ml)	-2.59	1.54	-1.68	0.093
		Active time	0.02	0.02	0.96	0.335
g	Females: Number of errors (with adjusted criterion)	Inter-tegular span (ITS)	1.32	0.32	4.17	< 0.001
		Amount drank (ml)	0.37	0.47	0.79	0.430
		Active time	-0.07	0.12	-0.53	0.596

Animal species with strong sex roles may exhibit different behavioural and cognitive traits that hold ecological and evolutionary significance (e.g., Lucon-Xiccato and Bisazza 2014; Videlier et al. 2015; Ogurtsov et al. 2018; Fuss and Witte 2019). By using a bumblebee species that shows clear sex role differences, we examined whether female workers and male drones differed in active time and behavioural flexibility. Key findings include bumblebee males were more active as they showed a higher active time in a new environment the first time, bumblebees males showed comparable colour associative learning ability to females, and males were more behaviourally flexible as they made fewer errors in RLP than females.

The fact that bumblebee males show higher active time than females could not be attributed to body size or sucrose consumption, but likely come from the fact that drones generally fly longer than workers during their dispersal (Bertsch 1984; Osborne et al. 1999; Chapman et al. 2003; Kraus et al. 2009; Wolf et al. 2012). Though it remains to be examined, it could also be due to age difference between males and females as age has been negatively related to active time and flight distance in bumblebee females (Gilgenreiner and Kurze 2024). Nevertheless, active time in a novel environment the first time may be related to other repeatedly traits, such as exploration tendency or anxiety-related traits (Montiglio, Garant, Thomas, Réale. 2010; Tanaka, Young, Halberstadt, Masten, Geyer 2012). Considering the frequency of entering each compartment, males visited most compartments more often than females in their first experience, possibly suggesting exploration. This possibility is strengthened by the result that males made equally long but more frequent visits to each compartment than females. Relating these results to ecological contexts, bumblebee males generally dispersing in a random direction from their nest and travelling far away from it (Osborne et al. 1999; Chapman et al. 2003; Kraus et al. 2009; Wolf et al. 2012) and they patrol after dispersal; males create a flight path, marking different places with their scent to attract potential mates, and regularly patrol these places (Alcock et al. 1978; Valterová et al. 2019). These behaviours could allow males to explore their environment more than workers, and increase their chances of discovering profitable flowers after their dispersal (Kraus et al. 2009) as well as potentially increasing their likelihood of finding an unrelated queen for reproduction (Amin et al. 2012).

In the DLP, males showed comparable associative learning performance to females, which is consistent with previous studies that have used different setups (e.g., stimuli presented in an array or two-choice task) and conditions (e.g., natural or laboratory) (Wolf and Chittka 2016; Muth et al. 2021). When examining the factors that affected DLP performance, within-females analysis showed that active time was positively correlated with DLP performance. As discussed above that active time may reflect exploration tendency, a previous study has been shown that such tendency is positively correlated with associative learning ability, which can benefit the colony when workers efficiently exploiting newly emerged food sources (Raine and Chittka 2008). In contrast, within-male analysis showed no relationship between associative learning and active time or other tested factors (i.e., body size, sucrose consumption). Despite this, the between-group analyses revealed that males who completed the DLP had a larger body size and drank more sucrose than the males who never completed the DLP. These results suggest that a larger body size may bring an advantage for males in learning (Del Castillo and Fairbairn 2012). It is possible that larger body size may increase sucrose consumption, however, further investigations (e.g., increased energy expenditure in the activity task, sucrose sensitivity, or responsiveness) are encouraged to clarify the potential link between active time (activity) and associative learning abilities.

In the RLP, most bumblebees’ first choice was on the previously rewarded colour in the first session. Bees also made more errors before reaching the criterion of RLP than in the DLP. These results suggest that the previously rewarded stimulus became a ‘distractor’ when bees had to relearn the reward contingency (Dyer and Chittka 2004). As predicted, males were more behaviourally flexible (i.e., made fewer errors) than females, which could not be attributed to differences in body size, sucrose consumption or active time in the activity task. Within-female analysis showed that body size was a predictor of their RLP performance. Workers’ size was negatively correlated with RLP performance, which may be explained by large and small workers differing in their foraging roles; larger bees invest more in learning highly rewarding flowers (like in our experiment) while small bees were more readily accepting flowers of all qualities (Frasnelli et al. 2021). Unlike females, none of the tested factors can predict males’ RLP performance. Despite this, the males who participated in RLP were those who had completed the DLP and were larger in body size than their male counterparts who never completed the DLP, suggesting a possible link between body size and behavioural flexibility. Nevertheless, males’ overall enhanced flexibility may reflect their readiness to respond to changes in their environment, which could be ecologically adaptive, for example, adjusting their foraging decisions in response to a change in flower quality could help them to maximise energy gain, and the time spent searching for mates. Arguably, the costs of not being flexible are higher for males than for females; females can continue to access resources in the colony. There is also an upside for workers to make more ‘mistakes’ as it can lead them to discover and exploit new profitable sources and collect more resources for the colony (Evans et al. 2017). Future studies could conduct a second (or more) reversal to further assess the learning strategy of males, and whether sexes differ in relearning the colour-reward association across several reversals.

In conclusion, we provided evidence that longer active time when males first leaving their nest and behavioural flexibility are two key behavioural and cognitive traits for male bumblebees in the context of their dispersal. Drones outperforming workers in their activity and behavioural flexibility could be related to the pressure of being solitary foragers. These results overall contribute to the understanding of how the expression of ecologically and evolutionarily important traits varies between sexes with different roles.

Conflict of interests.

We declare there is no conflict of interest.

Funding.

This study is supported by the Sustainability and Environmental Research Knowledge Exchange Institute, funding PKYC and KH (PS00054).

Author Contribution

PKYC designed, conducted, analysed, and wrote the first draft of the study. SD conducted the experiment and preliminary analyses. KH and TR made significant contributions to conceiving the project, analyzing it, discussing it and editing the manuscript.

Acknowledgement

We thank Toyota Manufacturing Deeside for providing full support for conducting this experiment. A special thank to M Vasey for supporting all logistics for the project. Thanks also to M MacDonald, O Reece, and L Chumper, for their logistical help.

Data Availability

Metadata and r code are freely accessible on OSF ( [https://osf.io/3j75g/?view\_only=9d7891976466476d8c721e4a4812bd13](https:/osf.io/3j75g/?view_only=9d7891976466476d8c721e4a4812bd13) )

Alcock J, Barrows EM, Gordh G et al (1978) The ecology and evolution of male reproductive behaviour in the bees and wasps. Zool J Linn Soc 64:293–326
Amin MR, Bussière LF, Goulson D (2012) Effects of Male age and Size on Mating Success in the Bumblebee Bombus terrestris. J Insect Behav 25:362–374
Baer B (2003) Bumblebees as model organisms to study male sexual selection in social insects. Behav Ecol Sociobiol 54:521–533
Bailes EJ, Pattrick JG, Glover BJ (2018) An analysis of the energetic reward offered by field bean (Vicia faba) flowers: Nectar, pollen, and operative force. Ecol Evol 8:3161–3171
Beekman M, van Stratum P (1998) Bumblebee sex ratios: why do bumblebees produce so many males? Proceedings of the Royal Society of London Series B: Biological Sciences 265:1535–1543
Belsky JE, Camp AA, Lehmann DM (2020) The Importance of Males to Bumble Bee (Bombus Species) Nest Development and Colony Viability. https://doi.org/10.3390/insects11080506. Insects 11.
Bertsch A (1984) Foraging in male bumblebees (Bombus lucorum L.): maximizing energy or minimizing water load? Oecologia 62:325–336
Brown M, Brown MJF (2020) Nectar preferences in male bumblebees. Insectes Soc 67:221–228
Cane JH (1987) Estimation of Bee Size Using Intertegular Span (Apoidea). J Kans Entomol Soc 60:145–147
Chapman RE, Wang J, Bourke AFG (2003) Genetic analysis of spatial foraging patterns and resource sharing in bumble bee pollinators. Mol Ecol 12:2801–2808
Coppens CM, de Boer SF, Koolhaas JM (2010) Coping styles and behavioural flexibility: towards underlying mechanisms. Philos Trans R Soc Lond B Biol Sci 365:4021–4028
Del Castillo RC, Fairbairn DJ (2012) Macroevolutionary patterns of bumblebee body size: detecting the interplay between natural and sexual selection. Ecol Evol 2:46–57
Duchateau MJ (2004) Sex ratio variation in the bumblebee Bombus terrestris. Behav Ecol 15:71–82
Dukas R (2013) Effects of learning on evolution: robustness, innovation and speciation. Anim Behav 85:1023–1030
Dukas R, Bernays EA (2000) Learning improves growth rate in grasshoppers. Proc Natl Acad Sci U S A 97:2637–2640
Dyer AG, Chittka L (2004) Fine colour discrimination requires differential conditioning in bumblebees. Naturwissenschaften 91:224–227
Evans LJ, Smith KE, Raine NE (2017) Fast learning in free-foraging bumble bees is negatively correlated with lifetime resource collection. Sci Rep 7:496
Frasnelli E, Robert T, Chow PKY et al (2021) Small and Large Bumblebees Invest Differently when Learning about Flowers. Curr Biol 31:1058–1064e3
Fuss T, Witte K (2019) Sex differences in color discrimination and serial reversal learning in mollies and guppies. Curr Zool 65:323–332
Gould TD, Dao DT, Kovacsics CE (2009) The Open Field Test. Mood and Anxiety Related Phenotypes in Mice. Humana, Totowa, NJ, pp 1–20
Goulson D (2003) Bumblebees: Their Behaviour and Ecology. Oxford University Press
Gumbert A (2000) Color Choices by Bumble Bees (Bombus terrestris): Innate Preferences and Generalization after Learning. Behav Ecol Sociobiol 48:36–43
Hagen M, Dupont YL (2013) Inter-tegular span and head width as estimators of fresh and dry body mass in bumblebees (Bombus spp). Insectes Soc 60:251–257
Hartig F (2025) DHARMa: Residual Diagnostics for Hierarchical. Multi-Level / Mixed) Regression Models
Huth-Schwarz A, León A, Vandame R et al (2011) Workers dominate male production in the neotropical bumblebee Bombus wilmattae (Hymenoptera: Apidae). Front Zool 8:13
Ings TC, Raine NE, Chittka L (2009) A population comparison of the strength and persistence of innate colour preference and learning speed in the bumblebee Bombus terrestris. Behav Ecol Sociobiol 63:1207–1218
Izquierdo A, Jentsch JD (2012) Reversal learning as a measure of impulsive and compulsive behavior in addictions. Psychopharmacology 219:607–620
Kraus FB, Wolf S, Moritz RFA (2009) Male flight distance and population substructure in the bumblebee Bombus terrestris. J Anim Ecol 78:247–252
Lea SEG, Chow PKY, Leaver LA, McLaren IPL (2020) Behavioral flexibility: A review, a model, and some exploratory tests. Learn Behav 48:173–187
Lenth S, Love et al Emmeans: Estimated marginal means, aka least-squares means. R package version
Lucon-Xiccato T, Bisazza A (2014) Discrimination reversal learning reveals greater female behavioural flexibility in guppies. Biol Lett 10:20140206
Magnusson A, Skaug H, Nielsen A et al (2017) glmmTMB: generalized linear mixed models using template model builder. R package version 0 1 3
Manning TH, Austin MW, MuseMorris K, Dunlap AS (2021) Equivalent learning, but unequal participation: Male bumble bees learn comparably to females, but participate in cognitive assessments at lower rates. Behav Processes 193:104528
Montiglio P-O, Garant D, Thomas D, Réale D (2010) Individual variation in temporal activity patterns in open-field tests. Anim Behav 80:905–912
Muth F, Tripodi AD, Bonilla R et al (2021) No sex differences in learning in wild bumblebees. Behav Ecol 32:638–645
Ng VKY, Cribbie RA (2017) Using the gamma generalized linear model for modeling continuous, skewed and heteroscedastic outcomes in psychology. Curr Psychol 36:225–235
Nicolakakis N, Sol D, Lefebvre L (2003) Behavioural flexibility predicts species richness in birds, but not extinction risk. Anim Behav 65:445–452
Ogurtsov SV, Antipov VA, Permyakov MG (2018) Sex differences in exploratory behaviour of the common toad, Bufo bufo. Ethol Ecol Evol 30:543–568
Osborne JL, Clark SJ, Morris RJ et al (1999) A landscape-scale study of bumble bee foraging range and constancy, using harmonic radar. J Appl Ecol 36:519–533
Pamminger T, Becker R, Himmelreich S et al (2019) The nectar report: quantitative review of nectar sugar concentrations offered by bee visited flowers in agricultural and non-agricultural landscapes. PeerJ 7:e6329
Raine NE, Chittka L (2008) The correlation of learning speed and natural foraging success in bumble-bees. Proc Biol Sci 275:803–808
Raine NE, Chittka L (2012) No trade-off between learning speed and associative flexibility in bumblebees: a reversal learning test with multiple colonies. PLoS ONE 7:e45096
Raine NE, Ings TC, Ramos-Rodriguez O, Chittka L (2006) Intercolony variation in learning performance of a wild British bumblebee population Hymenoptera: apidae: Bombus terrestris audax. Entomol Gen 28:241
Reader SM (2015) Causes of Individual Differences in Animal Exploration and Search. Top Cogn Sci 7:451–468
Réale D, Reader SM, Sol D et al (2007) Integrating animal temperament within ecology and evolution. Biol Rev Camb Philos Soc 82:291–318
Robert T, Frasnelli E, Collett TS, Hempel de Ibarra N (2017) Male bumblebees perform learning flights on leaving a flower but not when leaving their nest. J Exp Biol 220:930–937
Rother L, Müller R, Kirschenmann E et al (2023) Walking bumblebees see faster. Proc Biol Sci 290:20230460
Sokal RR, Rohlf FJ (1995) Biometry: the principles and practice of statistics in biological research. WH Freeman, New York New York, USA
Strang CG, Sherry DF (2014) Serial reversal learning in bumblebees (Bombus impatiens). Anim Cogn 17:723–734
Tanaka S, Young JW, Halberstadt AL, Masten VL, Geyer MA (2012) Four factors underlying mouse behavior in an open field. Behav Brain Res 233:55–61
Tapp PD, Siwak CT, Estrada J et al (2003) Size and reversal learning in the beagle dog as a measure of executive function and inhibitory control in aging. Learn Mem 10:64–73
Thomson JD, Plowright RC (1980) Pollen carryover, nectar rewards, and pollinator behavior with special reference to Diervilla lonicera. Oecologia 46:68–74
Toppa RH, Arena MVN, da Silva CI et al (2021) Impact of glues used for RFIDs on the longevity and flight muscles of the stingless bee Melipona quadrifasciata (Apidae: Meliponini). Apidologie 52:328–340
Valterová I, Martinet B, Michez D et al (2019) Sexual attraction: a review of bumblebee male pheromones. Z Naturforsch C 74:233–250
Videlier M, Cornette R, Bonneaud C, Herrel A (2015) Sexual differences in exploration behavior in Xenopus tropicalis? J Exp Biol 218:1733–1739
Walsh RN, Cummins RA (1976) The Open-Field Test: a critical review. Psychol Bull 83:482–504
Wascher CAF, Allen K, Szipl G (2021) Learning and motor inhibitory control in crows and domestic chickens. R Soc Open Sci 8:210504
Wolf S, Chittka L (2016) Male bumblebees, Bombus terrestris, perform equally well as workers in a serial colour-learning task. Anim Behav 111:147–155
Wolf S, Toev T, Moritz RLV, Moritz RFA (2012) Spatial and temporal dynamics of the male effective population size in bumblebees (Hymenoptera: Apidae). Popul Ecol 54:115–124
Yadav S, Yadav S, Sharma D, Sangwan N (2016) Bumblebees, Life Cycle and their role in Pollination-A Review. Int J Res Sci
Zhao H, Liu Y, Zhang H et al (2021a) Worker-born males are smaller but have similar reproduction ability to queen-born males in bumblebees. Insects 12:1008
Zhao H, Mashilingi SK, Liu Y, An J (2021b) Factors influencing the reproductive ability of male bees: Current knowledge and further directions. Insects 12:529
Coppens CM, de Boer SF, Koolhaas JM (2010) Coping styles and behavioural flexibility: Towards underlying mechanisms. Philosophical Trans Royal Soc B: Biol Sci 365:4021–4028. https://doi.org/10.1098/rstb.2010.0217
Lea SEG, Chow PKY, Leaver LA, McLaren IPL (2020) Behavioral flexibility: A review, a model, and some exploratory tests. Learn Behav 48(1):173–187. https://doi.org/10.3758/s13420-020-00421-w
van Dixhoorn I, Aubé L, van Zyl C, de Mol R, van der Werf J, Lardy R, Mialon M-M, van Reenen K, Veissier I (2024) From data on gross activity to the characterization of animal behaviour: which metrics for which purposes? Peer Community J 4:e105. https://doi.org/10.24072/pcjournal.489

No competing interests reported.

Supplementarymaterials.docx

Download PDF

Reviewers invited by journal
17 Dec, 2025
Editor assigned by journal
08 Dec, 2025
Submission checks completed at journal
08 Dec, 2025
First submitted to journal
05 Dec, 2025

You are reading this latest preprint version

Male bumblebees (Bombus terrestris) are more active and behaviourally flexible than workers.

Status:

Version 1

Abstract

Figures

Introduction

Methods

Study species and housing

Ethics

Apparatus

Behavioural Measurement

Statistical analysis

Results

Sex and Activity Levels in a New Environment

Sex and Performance in Learning a Colour-Reward Association

Discussion

Declarations

Conflict of interests.

Funding.

Author Contribution

Acknowledgement

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1