Flight plays crucial role in the fitness of insect pollinators, such as bumble bees. Despite their relatively large body size compared to their wings, bumble bees can fly under difficult ambient conditions, such as cooler temperatures. While their body size is often positively linked to their foraging range and flight ability, the influence of age remains less explored. Here, we studied the flight performance (distance, duration, and speed) of aging bumble bee workers using tethered flight mills. Additionally, we measured their intertegular distance (ITD) and dry mass as proxies for their body size. We found that flight distance and duration was predominantly influenced by age, challenging assumptions that age does not play a key role in foraging and task allocation. From the age of 7 to 14 days, flight distance and duration increased six-fold and five-fold, respectively. Conversely, body size primarily impacted the maximum and average flight speed of workers. Our findings indicate that age substantially influences flight distance and duration in bumble bee workers, affecting foraging performance and potentially altering task allocation strategies. This underscores the importance of considering individual age and physiological changes alongside body size/mass in experiments involving bumble bee workers.

Experimental overview

The experiment was designed based a pilot study, in which the flight performance of differently aged workers from two colonies had been examined in the previous year. In this study, we repeatedly measured the individual flight performance of 61 workers from four colonies (n = 14, 18, 8 and 21 workers per colony) at the age of 7, 14 and 21 days. Due to occasional difficulties in locating every marked worker on each day of measurement or potential mortality (n = 13/77), or instances where a few workers had lost their tags during flight measurement (n = 5/77), these individuals were excluded from data analysis. We selected workers with a wet body mass ranging from 100 mg to 320 mg, primarily during the colony’s exponential growth phase. 

Bumble bee husbandry

We used four commercial, queenright Bombus terrestris colonies (Natupol Research Hives, Koppert B.V., Netherlands) with approximately 20-30 workers. They were housed in their standard plastic nest boxes (27 [l] x 24 [w] x 14 [h] cm). Each colony was connected to a small foraging arena (60 x 40 x 28 cm), where bees were required to perform foraging flights to access ad libitum 40% w/v sucrose solution provided through 50 mL gravity feeders. Additionally, colonies were provided with 3-6 g of pollen candy (2:1 honey bee collected organic pollen: 75% sucrose syrup) daily, depending on their consumption. Colonies were kept under laboratory conditions with a relative humidity of about 40 % RH and an average room temperature of 23 ± 1 °C. Foraging areas were illuminated by two flicker free daylight-like LEDs (each 2400 lm, CRI98, 5500 K, True-Light International GmbH, Germany) under 14:10 h light:dark regime, but nest boxes were kept dark by covering them with cardboard. The foraging arenas served to allow aging bees to ‘practice’ flying and become accustomed to the same lighting conditions used during tests (described below).

Tagging and marking

Newly emerged workers (< 1 d old), identified by their silvery appearance and unfurled wings [43], were collected from their colonies. Their sex was determined by counting antennal segments (females have 12, males 13), under a stereo microscope. Then bees were immobilized with metal pins without harming them to carefully shave off the thorax hairs between the tegula, where a circular stainless-steel tag (Æ 2 mm, thickness = 0.1 mm, weight (mean ± sd) = 2.43 ± 0.02 mg, n = 20, Sartorius Micro SC2, Sartorius AG, Germany) was attached using superglue (Supergel, UHU GmbH & Co. KG, Germany). The weight of the tag accounted for only 1% of the mean weight of all tested bees (231.84 ± 52.62 mg, analytic balance A210P-OD1, Sartorius AG, Germany), significantly lighter than typical pollen and nectar loads carried by foragers (up to 90% of body weight [28]). Therefore, we assume that this tagging method is unlikely to significantly affect flight performance as measured here. These metal tags were colour coded to differentiate cohorts. For reliable identification of each individual bee, the middle legs were colour coded using water-based permanent paint markers (5M Uni-Posca, Mitsubishi pencil, Japan). Afterwards, tagged bees were placed in a separate plastic cup for 30-60 min before returning them to their natal colony.

Although precautions were taken to guarantee that attached tags would not interfere with wing movements, tag positions were measured using a digital microscope (figure S1a, CHX-500F, Keyence GmbH, Germany) on frozen specimens at the end of the experiment. The tag deviation from the centre between the tegulae was calculated, but tag position did not significantly affect their flight performance (figure S1b,c).

Flight mill setup

We used four tethered air flight mills (figure S2a), similar to previously described setups and methods [25, 41, 42]. The core of each flight mill is a lightweight and counter-weighted arm (length 32 cm) that floats by magnetic levitation and a needle that is inserted into a low-friction Teflon bearings at the centre of the arm. Individual bees were attached to a magnet (Æ 2 mm, 4 mm long) on one end of the arm and counter-balanced on the other arm, enabling tethered flights with their own power. An optical sensor transmitted a voltage pulse every half rotation (flight distance of 50 cm), recorded to a PC using the software guiBee [44]. Data extraction and calculations of flight distance, duration, and speeds were executed using the RScript FlightMillDataExtraction [45] in R (version 4.3.0). Each flight mill was positioned at the centre of a plastic cylinder (Æ 46 cm), keeping about 7 cm distance between the bee and the cylinder wall. The inside was decorated with 2.5 cm wide black and white vertical stripes continuously printed on paper to provide consistent visual feedback [46]. The walls also prevented interference from neighbouring flight mills and reduced any potential impacts of air currents [47]. The flight mills were illuminated by four flicker free daylight-like LEDs (each 2400 lm, CRI98, 5500 K, True-Light International GmbH, Germany) from a height of 70 cm.

Flight trials

Marked workers were gently collected from each colony in the morning using tweezers at the age 7, 14 and 21 days (i.e. each bee was repeatedly measured). They were kept separately per colony in a metal cage (9.5 x 8.5 x 5 cm) with ad libitum access to 40% w/v sucrose solution. After collection, workers were individually separated into flat-bottom glass vials (10 mL, 50 x 22 mm) with mesh lids, containing a 45 x 15 mm piece of cardboard to absorb any faeces. For 20 min each bee was individually fed with 40% w/v sucrose solution to satiation through the mesh of the lid. Subsequently, individual bees were weighed (d = 0.1 mg, analytic balance A210P-OD1, Sartorius AG, Germany) and attached to the magnet on the one side of the flight mill arm and kept in place on a launch platform. A counterweight (to ± 10 mg) was attached on the other side of the flight mill arm. The bees were then allowed to calm down and rest in the dark for 20 min by covering each flight mill with thick cardboard. Each bee was then positioned in the direction of flight and the launch platform was quickly removed to initiate flight (figure S2b,c; video S1). When bees stop flying, they would raise their wings and extend their legs, which resulted in a slowing of flight mill rotation. In addition to direct observations, stops can be detected in the raw data using our R script described above [45]. When a bee stopped flying (or did not initiate flight), it was allowed to rest on a handheld plastic Petri dish (Æ 46 cm) for approximately 20 s. Each bee was allowed 4 stops, i.e. 5 flight starts. We decided to allow multiple stops and pool individual flight data based on field observations showing that foragers periodically pause [48], and previous flight mill studies indicating that flight durations tend to be short after three stops [25, 49]. Our pilot study was consistent with those findings. To avoid biases in our dataset, we refrained from setting minimum or maximum flight durations or distances, even if a bee would not fly (e.g. one bee aged 21 days flew 0 m; see data in electronic supplementary material). The temperature during all flight was recorded at 5 min intervals (RC-5 temperature data logger, Elitech Ldt., UK) to calculate the average flight temperature and account for slight room temperature differences. At the end of each flight trial, bees returned to their natal colonies. After the last flight trial at age of 21 d, bees were frozen and stored at -20°C until further analysis.

Measuring intertegular distance and dry mass

In addition to the evaluation of the tag position, the intertegular distance (ITD) for each bee was measured using a digital microscope (figure S1a, CHX-500F, Keyence GmbH, Germany). The ITD serves as a proxy for workers body size, along with their dry weight (figure S1b) [50]. Prior to measuring their dry weight, the sternites (ventral abdominal segments) of each individual bee were cut open from the stinger to the fourth sternite without damaging their guts. The bees were then individually dried at 60°C for 3 d in drying cabinet (U40, Memmert GmbH & Co. KG, Germany) and subsequently weighed (d = 0.1 mg, analytic balance M-Pact AX224, Sartorius GmbH, Germany).

Statistical analyses

All statistical analyses and data visualizations were performed using R version 4.3.2 [51]. The complete code and output are provided in the electronic supplementary material. Briefly, to analyse the effects of the fixed factors age and body size (using dry mass as a proxy, figure S1b) as fixed factors and their potential interactions on the response variables flight distance, flight duration, average and maximum flight speed, generalized linear mixed effect models (GLMMs) were ran using the glmmTMB package. Flight distances were log-transformed using log₁₀ (x + 1) to improve model fit based on gaussian data distribution. To account for repeated measures of individual bee flight performance, Bee ID was included as a random factor. The covariates colony (figure S4) and temperature during flight (figure S3) were additionally included as random factors. Model selection was performed based on the Akaike information criterion (AIC) and likelihood ratio tests. The final models were compared with their respective null-models. Model assumptions and dispersion of the data were checked using the DHARMa package [52]. Significance (p < 0.05) of model terms was determined using the Anova function of the car package [53]. Pairwise comparisons between age groups were conducted using the function emmeans [54] with Bonferroni correction. The individual flight improvement was analysed using chisq-test function to perform a Pearson’s χ² test for count data with Yate’s continuity correction. The cor.testfunction was used to further describe the relationship between the flight duration and distance, dry mass and flight distance, dry mass and flight duration, and dry mass with mean and maximum speeds within each age class.