BUMBLEBEE FORAGING PREFERENCES: DIFFERENCES BETWEEN SPECIES AND INDIVIDUALS

2 MATERIALS AND METHODS

2.1 SITE

FIGURE 1

(FIGURE 1) The site is in north-east Scotland about 10 km south of Stonehaven, at Ordinance Survey Grid Ref. NO 816 795, on a south-west slope at 150 m above sea level. There are few trees in the area, and none near the site, so the effects of the wind on the vegetation can be severe. The bee walk route was about 410 m long. 400 m followed one edge of a lane between two fields, the edge varied in width from 50 cm to 1.5 m. A 10 m strip of garden was also included. Both sides of the lane contained the same mixture of plant species and in similar proportions. The flowers used by bees during the bee walk are listed in TABLE 2. Other flowers present in the lane, but not seen to be foraged by bees were, Veronica sp., Campanula rotundifolia, Ulex europaeus and various grasses. In the garden, flowers used by bees are listed in TABLE 2, other flowers included Geranium sp. and some old-fashioned Rosa sp.

The bee walk side was free from the influence of cattle grazing, but parts of the other side of the lane were within reach of cattle which were put into Field B at the end of June. Field A was set-aside this year and grass last year. Field B was grass this year and set-aside last year. The most common flower in the set aside field was Capsella bursa-pastoris, there was also a large number of Lamium purpureum and various grasses.

2.2 PILOT STUDY
A pilot study was done on the 17th, 18th and 20th of June 1995. This enabled bee marking and measuring techniques to be tried out and perfected, bee identification to be mastered, and the length of the bee walk to be decided.

2.3 MEASURING ENVIRONMENTAL VARIABLES
Temperature was measured by hanging a thermometer from the barbed-wire fence that separated the field from the lane. The thermometer was positioned about 50 cm from the ground surface, approximately bumblebee foraging height for the site, and in full sun, as the bees forage in full sun. Light conditions were recorded either as sun, sun\cloud or cloud. The sky was quartered, then each quarter halved. Sun was recorded when there was cloud in only 0-2 segments, "sun/cloud" when cloud was present in 3-6 segments, and cloud when cloud was present in 7 or 8 segments. Mist and rain were also recorded. Wind speed was recorded using a Ventimeter made by Elvometer Co., Sweden. The ventimeter was held 1.5 m from the ground surface for 20 s, or more if required. The wind speed was taken at approximately the 90 m mark on the bee walk, and the ventimeter was held over the bee walk vegetation. Environmental variables were recorded at the start of every bee walk.

2.4 BUMBLEBEE IDENTIFICATION
Bumblebees were identified using keys in Prys-Jones and Corbet (1987) and Alford (1975). While Bombus terrestris queens and B. lucorum queens are easily distinguishable in the field, their workers are not. So, following the example of Brian (1957), all of the workers were recorded as B. lucorum, since B. lucorum queens were more numerous than B. terrestris queens in 1995. Males of B. lapidarius and B. pratorum are fairly easily distinguishable and so their caste was recorded; the caste of other species is not so easily distinguished in the field, so was not recorded. The six species regularly seen during the bee walk are on PLATE 1.

2.5 THE BEE WALK
The bee walk was done every 2 hours from 6 a.m. until 8 p.m. on five consecutive days; each session done once in June, July and August 1995. Each walk consisted of:
• Measuring the environmental variables.
• Walking down to the start of the bee walk in the lane.
• Walking the 400 m of the lane bee walk, recording each bee foraging, and catching unmarked bees in numbered jars.
• Walking down to the garden area, recording bees foraging and catching unmarked bees in numbered jars.
The bee walk was repeated for one extra day in mid-August 1995 to assess how well the marking had stood up to time. On that day only marked bees were recorded. As the walk progressed, the species of each individual bee and the flower it was foraging from, were recorded on a Sony Professional Walkman. When possible the caste of bee was also recorded. The walk started and ended at the same point every time, and the direction followed was always the same. The walk was done at a slow, regular pace, approximately 0.5 m s for the 6 am walk, when bees were less abundant, 0.4 m s for all other walks. The barbed wire fence was used as one boundary, and the bare ground of the lane as the other. All bees actively foraging in a strip of approximately 20 cm wide stretching from the fence to the bare ground were recorded. If a bee landed and foraged on a flower that had just been passed it was not recorded. When a marked bee was recorded more than once in the same walk, only its first sighting was used in data analysis. Bees foraging on flowers outside the fence around Field A were counted, regardless of whether the roots were also outside the fence. The flower species on either side of the fence were the same, but Centaurea nigra was slightly more common outside (bee walk side) the field than inside, and Cirsium arvense was more common inside the field than outside. It was not possible to include the flowers inside the fence of Field A as the ground in the field sloped sharply downwards.

2.6 MARKING BEES AND MEASURING TONGUE LENGTH, AND HEAD LENGTH AND WIDTH
Unmarked bees were caught in numbered glass jars during the bee walk, the species of bee and flower were recorded as mentioned above, and the jar number was recorded on tape. The bees were measured and marked in the order in which they were caught, as soon as the bee walk ended. After the bee walk, the bees were kept indoors in the shade in the glass jars in which they were caught while waiting to be marked.

Figure 2	It was decided not to anaesthetise the bees, as this might have caused behavioural changes (Prys-Jones & Corbet, 1987). As the bees varied in size and strength a device was made that could restrain any bumblebee, big or small. This was made from the top 3.5 cm of a 6.5 cm wide plastic flower pot and plastic net with a 4-5mm mesh secured over the top of the pot with a length of tape (FIGURE 2). This mesh size was small enough to prevent escape of even small bees and large enough to allow the bee to be marked between the threads. On just one occasion an extremely small Bombus pratorum worker managed to squeeze through the mesh, and, as she was obviously distressed, she was allowed to fly away. The plastic top of an aerosol cap was used to form a base that could be moved up inside the flower pot, to trap the bee between the mesh and the top of the aerosol cap. 1 mm graph paper was sellotaped to the aerosol cap, so that the head measurements could be made.

The conical shape of the restraining device made it possible to fit it over almost any size of glass jar. The bee was persuaded to fly up to the net by excluding light from the sides and base. Once the bee was on the net the jar was removed and the aerosol cap placed under the opening to prevent escape. It was possible to hold the aerosol cap in place with one hand, while offering the bee 30:70 honey and water in a plastic pipette with the other. Not all bees were willing to drink the nectar and, as time was limiting, if they ignored the pipette after it had been within reach for over 10 s it was removed and just their head measurements were taken and they were marked. There was no apparent difference between species in the proportion that drank from the pipette. One noticeable thing was that if their buzzing sound was higher pitched than normal they generally refused to drink and struggled more when being marked, this became known as the "angry buzz". It is not known what caused some bees to emit this higher pitched buzz, but there was certainly a link between it and how co-operatively the bee behaved.

The end of the plastic pipette was marked in 1 mm intervals with a Staedtler fine lumocolor permanent ink pen. With a little practice it was possible to move the meniscus of the nectar up and down the pipette so forcing the bee to extend its tongue to foraging length, which could be seen through the plastic and its length measured. This method is similar to that used by Brian (1957), except that she used capillary tubing, not a pipette. After the tongue length was measured the aerosol top was moved up the inside of the pot until it touched the back of the bee, when the bee turned to walk on the aerosol top the flower pot was moved down with just enough pressure to stop the bee moving its thorax. Its head length and width were now measured, and then the bee could be marked.

The time taken to mark bees varied greatly, the shortest time was about two minutes, but a few bees took as long as five minutes. Waiting for a bee to walk on the surface of the aerosol cap was the main reason for the variation in time taken. Some bees walked over every part of the mesh before they decided to walk on the aerosol cap.

Water based Tipp-Ex in four colours, white, blue, green and pink, was used to paint a disc on the bee, carefully avoiding the tegulae (wing bases). A small blob of Tipp-ex was placed on the thorax using the brush provided, the blob was then smoothed down with a blunted cocktail stick, this also flattened the hairs on the thorax. Since Bombus pascuorum has a very hairy thorax, the layer of Tipp-Ex was insufficient to flatten the hairs, so before marking these hairs were either trimmed using a very small pair of scissors, or scraped off carefully using a scalpel. In August the weather was much hotter and drier and the Tipp-Ex dried much more quickly, so a slightly thicker layer of Tipp-Ex was applied to B. pascuorum, this allowed the hairs to be flattened without having to trim or remove them. It was felt that this caused less disturbance to the bee, and made the marking process quicker. Once the Tipp-Ex had dried, a number from 1 to 100 was written on the Tipp-Ex with a Staedtler fine lumocolor permanent ink pen, and the bee was set free.

Marking and measuring were done indoors, the bees were set free out of a window which was about 10 m at right angles from the 100 m mark of the bee walk. Nearly all the bees flew away immediately, but a few sat on the windowsill or below the window. This area was checked periodically and any bees still there were brought indoors, fed and released when they decided to fly.

The number and Tipp-Ex colour of all marked bees seen on the bee walk were also recorded. Occasionally a marked bee was seen but the number could not be read; this was recorded as an unmarked bee.

2.7 PREFERENCE AND CONSTANCY
A highly constant bee ignores or passes over apparently rewarding flowers to forage from flowers of a single species; only occasionally will such a bee forage from another flower species, in other words, the bee is showing a preference for a particular species of flower. Constancy is very important to plants as it means they will have a greater chance of receiving compatible pollen (Faegri & van der Pijl, 1979).

Flower preferences were calculated using the program PREFER in Krebs (1989) which calculates Manly's alpha index of preference. The variables required by the program are:
• The number of species of flower present in the area.
• The number of flowers of each species in the area.
• The number of visits to each species of flower by bumblebees of a particular species.
• Whether the resource was renewable or not.

The number of flowers was calculated by observing how the bees used each flower species, so as to find what could be considered a "functional individual unit", i.e. an area where one bee could forage but not two.

The full list of flowers used by bees during the bee walk is shown in TABLE 2 in the results. During each five day bee walk period, flowers were counted on the morning of the third day. For flowers with inflorescences and umbels, heads were counted; spikes were counted in the Labiatae; Hypericum perforatum flower clusters were counted; racemes of Vicia spp. and Lathryus pratensis were counted; and for all others individual flowers were counted or estimated. For Stachys lanata, originally spikes, made up of several individual flowers, had been counted but, as the bee walk proceeded, it became obvious that it was common to find two, or even three bees on a spike, but it was also possible to find only one bee on a small spike. Consequently it was eventually decided that on average a spike comprised two functional units. The number of visits to each species of flower by bees was taken from the bee walk data, and the PREFER program was run for each species of bumblebee. It was decided to class the resources as renewable, since few flowers secrete nectar at just one time period and so do renew their nectar. Pollen is not a renewable resource, but no single bumblebee is likely to remove the entire quantity of pollen from a flower in just one visit.

Originally the program was run with each spike of Stachys lanata counted as one unit, the results appeared to greatly over estimate the preference for S. lanata and, it was felt, did not reflect what had been seen in the field. Consequently the program was re-run with each spike counted as two "flower units".

To measure flower constancy, bees were picked at random, by following the first bee sighted after stepping outside. The species, caste, number (if marked), and sequence of flowers visited was then recorded on tape. When one bee was lost the next bee sighted was followed as a new sample run. Most of the bee following was within the area of the bee walk, but was not restricted to this area; the bee was followed for as far and as long as was possible.

2.8 MEASUREMENTS AND CHARACTERISTICS OF FLOWERS FORAGED BY BEES DURING THE BEE WALK
The flowers foraged by the bees were counted as mentioned in 2.7, and on the same morning between five and twelve flowers of each species were then examined and vernier callipers used to measure the length the bee would have to extend its tongue to reach the nectar. The presence of nectar was also recorded, as was the accessibility of the nectar, and the width of the opening leading to the nectar. For Digitalis purpurea the corolla opening is large enough for even the biggest bee to enter, so the distance to nectar was taken to be the distance from the ovary to nectar, as this is how far the bee must extend its tongue. The handling difficulty of the flowers was estimated, and the flowers that were foraged for nectar were put into four groups, group one being the easiest to handle, and group four the most difficult.
• Group 1. Flowers that had visible nectar.
• Group 2. Flowers that provided a landing platform on which nectar (not visible) could be reached by random probing. • Group 3. Flowers which do not provide a landing platform, but where the path from the corolla opening to the nectar was straight.
• Group 4. Flowers like Group 3 but where the path from the corolla opening to the nectar was not straight.

The height range of the flowers, the height of the vegetation surrounding the flowers, and the effects of wind on both the flowers and the surrounding vegetation were recorded. The surrounding vegetation was noted as the site is very windy and the movement of the vegetation in the wind might affect foraging preferences. Where most of the vegetation was lower and only the taller grasses were higher, than the flower species being recorded, the surrounding vegetation was classed as mixed, otherwise the surrounding vegetation was classed as higher or lower.

Where possible flowers and inflorescences for identification and examination were picked from various heights and in various places throughout the bee walk. Identification of flowers was made using Rose (1981) and Clapham et al. (1993).

(C) Copyright 1999 L. Smith