The authors have declared that no competing interests exist.
‡ These authors are joint first authors on this work.
Insect pollinators such as bumblebees play a vital role in many ecosystems, so it is important to understand their foraging movements on a landscape scale. We used harmonic radar to record the natural foraging behaviour of
Recent advances in animal-tracking technology have brought within reach the goal of tracking every movement of individual animals over their entire lifetimes [
Foraging behaviour in bees has long been studied [
Only in the last 20 years has harmonic radar technology made it possible to observe and record the flight paths of insect pollinators at ecologically relevant scales [
The studies described above revealed a great deal about the structure of exploratory and foraging flights, but opened up a number of key questions that are unanswered as yet. Does the change in flight structure from inexperienced to experienced bees occur gradually or as a sudden transition? When and how do bees discover the forage sources they go on to exploit? No prior study has been able to track the activity of individual insects throughout their entire life history, or even a significant portion of their life, making it impossible to address these questions. Studies of foraging behaviour have tracked a few flights from each bee but had no knowledge of their previous or subsequent experience. Lihoreau et al. were the first to track individual bumblebees over many consecutive foraging bouts, using a combination of harmonic radar and motion video capture to detail the formation of efficient and repeatable foraging routes between artificial forage sources over time [
Harmonic radar allows us to combine finer-scale temporal data than is usually obtained from satellite tracks with the ability to track every movement of our focal bees over the several week timescale of their lives as foragers. This continuous monitoring of flight patterns allows us to pinpoint when and how particular sites were first discovered; to identify the exact point at which switches in behaviour take place; to tell which parts of the environment are being utilised by individuals at any one time; and to address a number of questions that would be beyond reach without detailed knowledge of an individual’s history and development. Through the analysis of this dataset on the lifelong tracking of four individual forager bees, detailing 31 days of flight activity and 244 flights, we provide the most detailed study to date of the dynamics of resource exploitation versus exploration in a foraging insect pollinator in a field setting. To the best of our knowledge, this dataset represents the first lifetime track of any individual animal in sufficiently high temporal resolution to examine foraging routes.
Field work took place from June to September 2015 on arable farm land at Rothamsted Research (Hertfordshire, UK, 51’48”13N 0’22”8W,
Each panel represents the lifetime activity of a single bee. The position of the nest is marked by a blue circle. Each individual flight is shown in a different colour, the earliest flights undertaken by each bee in green, changing smoothly through yellow until the last flights in each bee’s life are shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain. A-D): flights of Bee 1–4 respectively.
We used commercially sourced colonies of
Before transferring colonies to the field we allowed them free access to a small flight cage (45 cm x 45 cm x 45 cm high) containing a gravity feeder filled with 30% sucrose solution. Bees that fed from the feeder were marked using numbered tags (Opalith Zeichenplättchen Leuchtfarben, Bienen-Voigt & Warnholz, Ellerau, Germany). We monitored these bees for 1–2 days, choosing one bee that fed often and regularly as our focal bee.
The entire colony was then transferred to the field where it was placed on a wooden frame (98 cm high) and covered with the lid to a honeybee hive. Once in the field, the tunnel was opened and the entire colony was allowed to forage freely. This allowed us to track the behaviour of the focal bees while the colony operated as naturally as possible, with all foragers able to forage
Movements of the focal bees outside the nest were tracked using 32 mm harmonic radar (previously described in [
A radar transponder consisting of a 16 mm vertical dipole was attached to the numbered tag on the thorax of each focal bee using superglue (Loctite Power Flex Gel, Henkel Ltd., Hemel Hempstead, UK). Transponders weigh around 15 mg. The transponder represents only 8–10% of a typical worker’s mass of 175–200 mg and bumblebees are known to carry nectar loads of up to 90% of their body mass [
We monitored the movements of the focal bee for 7–14 hours per day, depending on the weather. Tracking of the focal bee was terminated only if the bee did not return to the nest for 48 hours after which it was presumed to be dead or lost to the colony. The longest absence from which a bee ever returned to the nest was 20h12m. Overnight and on days when it was too rainy or windy to operate the radar, we blocked the tunnel entrance using a divider with an 8mm hole in it. Non-focal bees quickly learned to pass through this hole and foraged normally, but the length of the transponder prevented the focal bee from passing through and ensured that no flight could occur when the radar was not running. On five occasions the focal bee did not return to the nest before sunset so her movements could not be accounted for overnight. On these occasions we returned with the radar soon after sunrise the next morning and it is unlikely that we missed any flight, since bumblebees do not fly in the dark. No bee that had stayed out overnight was ever found to be already flying or attempting to re-enter the nest when we began recording the following day, and the directions from which they were first sighted were consistent with our last sightings the previous evening. On one occasion Bee 2 was observed in the field after sunset, sitting on the stalk of a thistle, just below the flower, and she was still in place when we set up the radar the following morning, resuming foraging at 8:15 am, 1h30m after we had begun recording.
When a bee remained static for a period, an experimenter watching the radar screen could use a custom-written Matlab script to convert the radar coordinates of the last known position of the bee to GPS coordinates. This was communicated to a second observer in the field who used a handheld GPS device (Garmin (Europe) Ltd., Southampton, UK) to search the area around the last sighting of the bee. Using this method, it was sometimes possible to get visual observations of the focal bee in the field. When observations were possible we recorded whether or not the bee was feeding on a flower and used individually numbered stakes to mark the locations of plants on which the bee had foraged. Visual observations were difficult to obtain since the radar only gives us the bee’s position accurate to approximately 2 m, and bees can sometimes make short flights between flowers that do not show up on radar due to obstructions by other plants or the landscape topography, meaning that the observer had to search within a radius of at least 5 m, in a complex landscape of tall stems and flowers where a single bee is difficult to spot. Observations could not be made systematically since they were only successful when the bee remained in the same location long enough for the observer to arrive and search, and because it was often not possible to locate the bee even if it had not moved.
We extracted a number of variables from the flight tracks. The
Flights were categorised as either
We analysed differences between flights in the two categories using four GLMMs (using the
Two bees switched the destinations of the flights we categorised as exploitation flights during their foraging careers. To investigate whether doing so conferred benefits in foraging performance, we reanalysed the data for just the exploitation flights of these two bees. GLMMs were performed on the same four dependent variables as above, with the flight destination (first or second) as predictor and bee ID as a random factor.
We tracked four bees for periods of 6–15 days each (see
Bee 1 | Bee 2 | Bee 3 | Bee 4 | |
---|---|---|---|---|
25/06/15–06/07/15 | 11/07/15–16/07/15 | 23/07/15–06/08/15 | 21/08/15–03/09/15 | |
12 | 6 | 15 | 14 | |
10 | 6 | 8 | 7 | |
156 | 31 | 26 | 31 | |
3 | 5 | 6 | 8 | |
142 | 7 | 17 | 16 | |
0.91 | 0.23 | 0.65 | 0.52 | |
15.6 ± 8.8 | 5.2 ± 1.9 | 3.3 ± 0.9 | 4.4 ± 3.2 | |
35:56 ± 119:41 | 133:10 ± 255:26 | 169:53 ± 205:33 | 31:59 ± 20:42 | |
1:27 ± 2:15 | 3:43 ± 3:35 | 7:09 ± 10:52 | 2:32 ± 1:35 | |
1:01 ± 1:14 | 25:44 ± 59:41 | 5:35 ± 14:46 | 26:14 ± 40:15 | |
442 ± 493 | 1362 ± 1366 | 1661 ± 2478 | 825 ± 485 | |
179 ± 71 | 217 ± 120 | 168 ± 72 | 257 ± 125 | |
2 | 2 | 1 | 0 | |
0 | 7 | 4 | 2 | |
0 | 15 | 12 | 2 | |
0 | 50 | 34 | 4 |
Tracking terminated for two bees (Bee 1 and 3) when they failed to return from an otherwise typical foraging trip, suggesting they may have died during the foraging bout, perhaps as a result of predation by crab spiders or birds. Bee 2’s last recorded flight was a fast, straight flight in a direction that she had not previously exploited for forage and which took her beyond the range of the radar, never to return. Bee 4’s final flight was likewise made in a novel direction.
The initial flight of Bee 1 (
The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain. A-D): flights of Bee 1–4 respectively.
Beginning with her next flight (#4, see S5 Fig), almost every flight she made for the remainder of her life fit our definition of exploitation flight: The 11 flights that were not considered exploitation flights all appeared similar to the exploitation flights, with a visit to a known foraging site, but the bee made a shorter second loop before returning to the nest. In these flights she repeatedly flew in the direction of the woodland corner, approximately 270m from the nest, making fast, straight trajectories that were extremely similar both between the outward and return portions of each flight as well as between flights. The portions of her flight furthest from the nest were not detectable by the radar but the flights appear to head toward a clearing at the corner of the woodland. This clearing was thickly overgrown with brambles which were in flower at that time and which attracted large numbers of bees and other pollinating insects. The position of the radar would have allowed us to detect her if her path had continued on the same trajectory into the next field, so it is likely that the clearing was her destination.
Bee 1 continued to visit the same location with remarkably little variation in her flight paths for 78 consecutive flights over the course of five days. After six days of tracking we had to confine her to the nest for two days due to bad weather. On the 9th day of tracking she made only a single flight, although her access to the field was unrestricted, which did not obviously differ from her other flights fitting the definition of exploitation flight.
On the 10th day she made one further flight, similar to her previous exploitation flights, in which she stopped at the corner of the woodland and returned to the nest. Her next flight, however, took an entirely new path, disappearing from radar behind a tree at the boundary between the experimental field and an adjacent field, Park Grass, to the Northwest of the nest, but reappearing from the edge of the woodland to the North-northwest of the nest. From this point on she never repeated her original trajectory but flew toward the edge of the woodland near to Rothamsted Manor, varying between North-northwest and North. The radar tracks typically lost the bee a few meters in front of the woodland, approximately 140 m from the nest, but it is likely that she was feeding on brambles growing at the woodland edge or on lime trees which were flowering along the treeline at that time. We cannot rule out the possibility that she flew over the trees and into the Manor gardens, but on three occasions we detected her with the radar at the woodland edge for 81–319 s, supporting the hypothesis that that was her destination. She continued to visit this new location for a further 69 flights over three days, finally failing to return from an otherwise ordinary looking outbound flight. Of particular note is that a single loop of her initial exploratory flight appears to have taken her to both the locations she later visited in exploitation flights (see
By the time Bee 2 was released (see
Starting with the first flight made the following day, Bee 2 made only seven (out of 31) flights that satisfy the description of exploitation flight. Each of these flights initially took a similar direction to the first exploited location of Bee 1, but the radar tracks of most were curtailed when the bee entered the radar-shadow of a tree at the corner of the study field and Park Grass. The most complete track (flight #15, see S172 Fig) shows the bee heading for the woodland edge about halfway along (≈ 210 m from the nest). In addition to these seven flights, Bee 2 visited the same area on a further 16 flights that were not considered exploitation flights as they included several distinct loops. These flights generally covered a larger area and took inconsistent flight paths by comparison to the flights we classified as exploitation flights. The bee was observed feeding on thistles in other locations on multiple occasions during this period and also made a further seven flights in which she did not visit the destination of her exploitation flights. Tracking of Bee 2 terminated at around 2pm on the 6th day of tracking when, after approximately 45 minutes of flight during which she was observed feeding on thistles, she made a fast, straight flight toward the Southeast, disappearing beyond the range of the radar and never returned. There was a severe rain storm that night, which it is likely that the bee did not survive.
Like Bee 1, Bee 3 visited two main foraging locations during flights fitting the definition of exploitation flight, but whereas Bee 1 visited only one on each flight and switched between the two, Bee 3 made four flights that included stops in both locations. Interestingly, as with Bee 1, she appears to have discovered both exploitation locations on her first ever flight (see
The first flight on the third day of tracking was the first flight by Bee 3 that met our definition of an exploitation flight: on leaving the nest she stopped at the thistle patch she regularly visited, but then flew across a low hedge between fields to a spot South-southeast of the nest in another fallow field in which scattered thistles were the main potential source of forage (≈ 190 m from the nest). This would become her second exploitation location. The following two flights were also defined as exploitation flights but were anomalous in that, unlike every other exploitation flight we observed, they included detours to other, non-revisited, parts of the field, in addition to stops at the exploitation location in the thistle patch. These were the only flights made by this bee in which she stopped at the thistle patch but not the second exploitation location. Beginning the following morning, Bee 3 made 17 more flights over five days, all but two of them fitting the definition of exploitation flight and including a stop at the second exploitation location in the next field. Three of these flights also involved a stop at the first exploitation location during the outbound portion of the trip. All, including those with two destinations, were characterised by fast, straight flights with no detours, although the outbound and return trips appeared less similar to one another than was observed in Bee 1, the outward trips always passing over a low hedge to get to the foraging site, while the return trips often skirted around the end of the hedge.
At the time Bee 4 was recorded, the majority of the thistles in the field site had gone to seed. Around 10% of thistles were still flowering, scattered randomly among the seed heads. The Park Grass field to the Northwest had some scattered flowers, mainly dandelion. The initial flight of Bee 4 was unusual (
Beginning in mid-afternoon of the second day of tracking, Bee 4 made a series of nine flights that we classified as exploitation flights, heading directly Southeast from the nest. These flights took her further away from the nest than the other bees ever travelled and the end-point remains unknown since she travelled beyond the range of the radar in most of them. There was no available forage at the location at which we typically lost the signal and an observer in the field was unable to find the bee, suggesting that she had continued flying past the point where we lost her signal. Two flights recorded her crossing the road that forms the boundary of Rothamsted Research, losing the signal in the neighbouring residential area, ≈ 575 m from the nest (flights #17 and #19, see S231 and S233 Figs). We consider it likely that she was foraging in one or more domestic gardens. Unlike Bees 1 and 3, which never returned to exploratory flights after commencing exploitation, this series of nine flights was interrupted by two exploration flights consisting of loops within the main experimental field.
On the fourth day of tracking, Bee 4 made another exploration flight which included a stop in Park Grass (≈ 240 m from the nest). Over the next four days she made four flights to the same location that met the definition of exploitation flight, interspersed with another four explorations that included a stop in Park Grass and two that did not. Of particular interest, one flight (#27, see S241 Fig) included visits to both exploitation locations, which were approximately 800 m apart; first visiting Park Grass and then, in a separate loop, flying in the direction of the residential area she visited on her initial exploitation flights. Her final flight took her in an unfamiliar direction, from which she never returned.
We used GLMMs to compare flight characteristics of flights categorised as exploration and exploitation flights (
Boxes show median and interquartile range, and whiskers represent the range of the data. Outliers have been removed for clarity (outliers defined as data-points lying more than 1.5 times the interquartile range outside the quartiles). 1-4A) flight duration of exploration and exploitation flights for Bee 1–4 respectively; 1-4B) time in flight; 1-4C) flight distance; 1-4D) digressiveness scores. Note that the ordinates differ in scale due to high levels of inter-individual variation.
Two of our bees (Bee 1 and Bee 4) switched the destination of their exploitation flights and subsequently never returned to exploiting the original location. Comparison of the exploitation flights to each of the two locations for each bee revealed that the distance flown was lower after switching sites (t156 = -2.04, P = 0.04;
Boxes show median and interquartile range, and whiskers represent the range of the data. Outliers have been removed for clarity (outliers defined as data points lying more than 1.5 times the interquartile range outside the quartiles). 1-2A) Flight duration of exploration visits to the 1st and 2nd location exploited by Bees 1 and 4, respectively; 1-2B) time in flight; 1-2C) flight distance; 1-2D) digressiveness score. Note that the ordinates differ in scale due to high levels of inter-individual variation.
Thus, the flights we classified as exploitation flights differ systematically from all other flights in a variety of flight characteristics which are not a necessary consequence of the definition of exploitation but suggest a difference in function. When bees switched the destination of their exploitation flights, the only apparent difference in flight characteristics was a reduction in the distance flown in flights visiting the second location.
We were able in some circumstances to convert the radar position of a stationary bee into GPS coordinates which an observer used to look for the focal bee in the field. We were not able to make visual observations of Bee 1 outside the nest. Relatively frequent observations were made of Bees 2 and 3, and Bee 4 was observed twice. In every case, the bees were found on thistles and, with the single exception of the first flight of Bee 4, they were nectar feeding during every observation.
The radar data suggest that all of our focal bees returned to particular forage patches repeatedly during their exploitation flights. Visual observation of Bee 3 at her first exploitation location confirmed that she returned frequently to the same part of one particular large thistle patch, where she was observed feeding. However, the spatial resolution of the radar data is not fine enough to address the question of whether they return to individual plants or flowers. We used individually numbered stakes to mark every thistle plant on which any of our focal bees was observed to feed, marking 88 in total over 29 foraging events by three bees. No bee was ever observed to return to a plant on which it had previously fed, although we cannot rule out the possibility that they did so on other occasions.
It was very common for the bees to move from plant to plant within individual thistle patches, frequently moving to the nearest neighbours of the plant they were on, interspersed with occasional flights of 5–10 m to other parts of the same flower patch. We observed a mean of 3 thistles fed from in each thistle patch, but this is certainly an underestimate since it took several minutes for an observer to arrive at a patch after the bee stopped flying. We never observed any of our focal bees foraging for pollen.
In this study we have followed the life-long flight activity of bumblebee foragers (S2-245 Figs), in sufficient temporal and spatial resolution to examine foraging movements at a local scale. We have gathered a wealth of data on the movements of individual bees and shed light on several important aspects of their foraging behaviour, including quantifying the relative frequency of flights fitting our definitions of exploration and exploitation flights, examining when forage sites are first discovered and illuminating the high degree of inter-individual variation in foraging behaviour. Although we did not directly quantify foraging behaviour, analysis of our data on spatial movement can help to illuminate how exploitation is balanced with the need to explore the landscape for potential food sources.
Osborne et al. compared the flight characteristics of flights recorded from bees with differing levels of experience [
Because our flight categories are not defined by the timing or path structure of the flights there is no reason
The flights fitting our definition of exploitation flight covered a shorter distance than other flights, which can be explained by their lower digressiveness: it seems that the destinations visited during exploitation flights are just as far away as the furthest distances reached during exploration, but that by flying straight paths towards their goal, the actual distance travelled by each is reduced. The bees also spent less time in flight during exploitation flights. However the total bout duration (the time that the bee spent out of the nest), was not lower in exploitation flights. This discrepancy can be explained by the fact that the time in flight accounted for less than 5% of the duration of a typical foraging bout. It is likely that almost all the time spent out of the nest is spent actually feeding and that this determines the length of a foraging bout. Thus, bees exploit known resources by developing and following efficient routes which are likely to reduce the energetic costs of foraging, so may increase their foraging efficiency (the ratio of energetic gain to cost [
Osborne et al. noted that the flights of more experienced foragers were longer, straighter and less looping than their initial flights [
Because no prior study has been able to follow the entire foraging career of individual bees, it has never previously been possible to pinpoint when a bee first discovered a location that later became an important forage source. All bees in our study started their foraging career with several exploration flights and our data suggest it is during these flights that they discover most or all of the sites that they will return to for the rest of their lives. Bees 1 and 3 seem to have discovered both of their locations during their initial exploratory flight; Bee 4 visited her second exploitation location, in Park Grass, on her third exploration (seven days before she visited it as part of an exploitation flight), although we have no evidence that she had prior knowledge of her first location before her first exploitation flight; it is not clear from the radar tracks exactly where Bee 2 was foraging during her exploitation flights, so we cannot determine when she discovered that location.
Our data on the preliminary flights of our bees suggest that there may be a greater range of exploration strategies in use by bumblebees than previously thought: Bees 1 and 3 explored their environment with a small number of arcing, looping flights, similar to those described by Osborne et al. [
There does not seem to be a clear temporal break between exploration and exploitation behaviour—all bees interspersed exploration flights with those we classified as exploitation flights—but exploration flights were rare after the onset of exploitation. Bee 4 made three flights that resemble the initial explorations, with looping paths covering large areas in less familiar parts of the field. As noted above, the availability of bramble and thistle was much reduced at the time Bee 4 was recorded and her initial exploitation location was much further from the nest than those favoured by the other bees. Although our data cannot determine what caused this bee to change its behaviour, one intriguing possibility may be that the familiar foraging location experienced a change in profitability, causing the bee to abandon the location in search of better forage conditions. Bee 2 interspersed flights fitting the definition of exploitation flight with others classified as exploration flights, visiting other parts of the field throughout its life. One hypothesis that may be worthy of future investigation is that a lack of suitable foraging locations motivated these bees to continue searching for new food sources when other bees had switched to an exploitation phase.
It is clear that exploration of the environment does not necessarily end with the initial orientation flights and that there is a wide range of variation in how individual bees allocate resources between exploiting known food supplies and exploring. Further work will be required to determine whether the amount of additional exploration is dependent on the quality and quantity of available resources or whether it represents inter-individual differences in foraging strategy.
All our bees apart from Bee 2 visited locations, likely to be forage patches, to which they were faithful for up to six days at a time. Each exploitation flight appeared to have only a single destination, with the exception of four flights by Bee 3. In most cases we did not have radar coverage of the places where feeding actually took place but the return tracks originated close to the end of the outbound journeys, suggesting that the bees were not performing a trapline between multiple distant patches of forage. Prior to the advent of harmonic radar, studies of the foraging patterns of bumblebees and honeybees within floral patches demonstrated that individual bees repeatedly visited a single foraging location, and it was inferred that foraging bees restrict their activity to a single patch in the environment [
The spatial movement patterns of pollinators have important implications for plant gene flow throughout the landscape [
As described above, it was only very rarely that flights we classified as exploration flights were interspersed with exploitations. There was no switching back and forth between foraging destinations. However, over the course of their lifetimes, two of our bees (1 and 4) made an abrupt switch of destinations, after several days of constancy. Interestingly, in the case of Bee 1, she had visited this second destination only once, on her initial exploratory flight nine days earlier. This suggests that forager bumblebees must be able to memorise potential foraging sites over at least that period, and that further exploration is not required in order to forage based on these memories. Bee 4 preceded her switch with an exploration flight that involved a stop at what became her second destination, but interestingly she had already visited that location during exploration flights several days previously, so might also have been acting on a stored memory of an alternative potential site.
Analysis of the characteristics of flights to each of the two destinations visited by each bee reveals that the second destinations were closer to the nest and that the bees’ flights to them were shorter and involved less flying time, although the total duration of flights did not decrease. Was reduction of the flight distance the motivation behind switching? The bumblebees’ condition may have declined with cumulative flight activity, so reducing the distance flown per trip may allow them to prolong their working life. Bumblebees with high levels of wing wear are more likely to forage in high density floral resources, and spend less time in flight [
One of the most striking results to emerge from these data is the large degree to which our bees differed from one another. Over 90% of the flights of Bee 1 fit our definition of exploitation flights, a much higher proportion than any of the other bees. In the case of Bees 2 and 3, this is partly explained by the fact that they spent longer exploring before beginning their exploitations. Bees 1 and 3 made few further exploration flights after switching to exploitation whereas Bees 2 and 4 interspersed continued explorations among the flights we categorised as exploitation. Bees 2 and 3 made exploitation flights to only one destination (although both were observed to feed in a variety of other locations during flights that could not be categorised as exploitation), whereas Bees 1 and 4 each switched the destination of the flights we classified as exploitation flights over the course of their foraging career, neither ever returning to their first destination afterwards.
Bee 1 made an average of 15 flights per day, while the other three bees only averaged 3–5 flights per day. Similarly large variations were seen in the duration of flights, the latency between flights and the distance travelled in each flight (see
Since each bee was from a different colony and each individual was tested in a different floral environment (because they had to be tested sequentially) we cannot be certain whether the heterogeneity of behavioural strategies was a result of heritable (colony) variation or the result of resource distribution (or a combination of both). A number of uncontrollable factors may have changed between the times we tracked each bee. The most important such factor is likely to be seasonal variation in what sources of forage were available. No two bees visited the same areas of the landscape on their exploitation flights (
Changing weather conditions might also account for some of the variation between bees. For example, Comba observed that the number of bumblebees visiting a forage patch was correlated with temperature [
Other potential sources of variation include the age and size of the bees, and the amount of competition they faced in exploring floral resources. We did not know the age of our bees when we began recording. We were able to track three of our focal bees for roughly two weeks apiece (and Bees 1 and 3 disappeared on otherwise typical seeming exploration bouts suggesting they might have been predated and would otherwise have continued to forage), but Bee 2 lasted only 6 days before flying away from the nest in a novel direction from which she never returned. It is possible that this bee was aged or sick, which might account for her unusual behaviour.
Although it is expected that randomly chosen individuals will tend to show variation in behaviour [
These are the data extracted from the raw tracking files and used in the analyses in this manuscript. The data are presented in an Excel spreadsheet with one row for every flight we recorded. The columns give the following information:
(XLSX)
The position of the nest is marked by a blue circle. Each panel shows every flight of a single bee that met the definition of an exploitation flight. Each individual flight is shown in a different colour, the earliest flights undertaken by each bee in green, changing smoothly through yellow until the last exploitation flights in each bee’s life are shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain. A-D): flights of Bee 1–4 respectively.
(TIF)
All flights made by Bee 1 during the period 25/06/2015–27/06/2015. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 1 during the period 28/06/2015–03/07/2015. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 1 on 04/07/2015. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 1 during the period 05/07/15–06/07/15. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 2 during the period 11/07/15–16/07/15. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 3 during the period 23/07/15–06/08/15. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
All flights made by Bee 4 during the period 21/08/15–03/09/15. Each figure represents a single flight. Bee ID, flight number, date of recording and the duration of each flight are shown in the bottom left corner of each figure. The position of the nest is marked by a blue circle. Colours represent the time from the start of each flight: initial portion of each flight is in green, changing smoothly through yellow until the end of the flight is shown in red. Grey dashed lines are used to join radar observations made more than 30 s apart, when the bee’s location was uncertain.
(ZIP)
The authors would like to thank Oscar Ramos Rodriguez for assistance in the field. We also wish to thank Andrew Riche and March Castle for creating the aerial orthomosaic image of the field site used in the figures. We are grateful to Kaz Ohashi and an anonymous reviewer for their insightful comments on an earlier version of this manuscript.