4.1 Species observed and comparative sizes of foraging populations
4.2 Marking and measuring of bees
4.3 Species flower preference
4.4 Individual flower preference
4.5 Constancy

Of the six species of bumblebee foraging on the site, Bombus lapidarius was the most common. There are no previous data on bumblebees for the site, however B. lapidarius with its distinctive colouring is easily recognisable and was certainly the most common bee in 1994 (pers. obs.) and possibly in previous years. B. pascuorum was also present on the site in 1994 and was seen foraging on Digitalis purpurea (pers. obs.).

On June 24th (FIGURE 3) Bombus pascuorum was not seen at all during the bee walk, whereas it had been seen on every previous day. The reason for this may be that Field B was being cut for silage that day and, as B. pascuorum is a surface nester preferring to nest in tussocky grass (Sladen, 1912; Alford, 1975), some of its nests were possibly being destroyed. The nests would have been at a very early stage, probably without adult workers. In the surrounding area this field was the best site for tussocky grass, the only other sites being roadside verges and field edges.

B. pascuorum bees did return to the site, but were never very numerous. The bee does not seem to have been resource limited, as Lathyrus pratensis, one of its flowers, was brimming with nectar during July, and so it may be that the limiting factor is safe nest sites. B. pascuorum is very useful for pollinating various clovers, so provision of safe nest sites should be considered important in areas where clover is an important agricultural crop.

In July all bees almost disappeared from the bee walk area, though many were seen flying overhead. It is believed that the bees were visiting nearby fields of oil seed rape, which must be considered an almost limitless resource during the period it is in flower. One field which had been sown with oil seed rape in 1994 and was fallow in 1995, but a sea of self-seeded rape was situated less than 1 km from the study site. In the same direction but about 1.5 - 2 km away were two fields sown with rape in 1995. Bumblebees forage as far as 2 km from their nests (Alford, 1975), so all of these fields would probably have been within foraging distance.

On the 9th of July almost two hours were spent walking in the set aside field, during that time only three bees were seen, all were Bombus lapidarius, two were foraging from Lamium purpureum, the other was flying low as if it were about to forage, or had just finished. So it appears that the bees had almost entirely deserted the site.

Bombus lapidarius was by far the most common bee in 1995, and possibly in previous years too; this may be because the site is ideally suited to it both in resources and nesting sites. As its name implies B. lapidarius likes to nest under stones (Sladen, 1912; Alford, 1975). In the study area and surrounding countryside most fields have dry stone dykes as their boundaries, many of these are no longer repaired but they still provide excellent nest sites for B. lapidarius.

In the study site bumblebees always outnumbered honeybees, and apart from late April and early May, when honey bees were rare, there always seemed to be sufficient resources, no incidents of aggression or competition were witnessed during the whole study period.

Index - Return to top

To investigate flower preferences and constancy of bees it is necessary to have a method for marking the bees safely, so that this will not interfere with their future behaviour, and will enable the bee to be recognised for at least the length of the intended observation period. It is also useful to have the opportunity of measuring the tongue and head length.

Brian (1957) had difficulty in persuading B. hortorum to drink, and could only get tongue measurements by killing the bees. She may have used sugar and water instead of honey and water, for B. hortorum readily drinks honey and water. When the first queens emerged in April and were foraging on heather, both honey and water and sugar and water were offered to them. None of the species could be persuaded to drink sugar and water, but all accepted honey and water. The smell of honey is perhaps more attractive.

Tongue length can be estimated using tongue-wing length regression equations (Morse, 1978) however this requires the tongues of a number of bees to be dissected out and measured when the bee is dead. This method was not suitable for this study, and in any case may not be the length actually extended during natural feeding. The connective tissue in the proboscis can enable the bee to extend its tongue making it 5-10% longer than the measured length when dissected (Inouye, 1980), also many bees do not always extend their tongues fully, especially if their honey stomach already contains a reasonable quantity of nectar (pers. obs.). If the bee is forced to regurgitate its nectar load, perhaps when too much force has been used when marking it, it will often extend its tongue further up the pipette to reach further into the honey and water. This was noticed about five or six times when measuring bees. Consequently it is believed that the correct and most "natural" tongue length is that measured using the pipette method in this study.

Marking bees on the dorsal surface of the abdomen would have given a bigger area to mark, and made identification on re-sighting easier. However, it was decided not to use the abdomen because wax is extruded from between the ventral tergites (Alford, 1975: Heinrich, 1979), and the hardening of the Tipp-Ex on the dorsal surface might have interfered with wax production and/or movement of the abdomen.

The marking technique appeared to work quite well and did not seem to harm the bees in any way, as many were re-sighted over long periods. Laverty (1993) glued plastic discs with numbers on to the thorax with quick drying glue, but never saw any of his marked bees again. Plastic discs with numbers already on them would probably be more easy to read than hand written numbers, so if a method could be found of sticking them to the thorax with a water based glue, or even with Tipp-Ex, this might prove to be an even better method.

The marking seemed to stand up well to wear and tear, and numbers written on 2nd August could still be read on 15th August, although the writing had faded a little. Unfortunately the weather during August was very dry so it is not possible to say how well the marks would stand up to prolonged wet weather.

The B. lapidarius worker that would not fly away (TABLE 1) was examined with a x10 hand lens to see if the Tipp-Ex was touching the tegulae or impeding wing movement. Neither was the case. The queens kept in artificial nest boxes were set free on 26th June. Two, a Bombus terrestris, and a B. pratorum, had reached the stage of building a honeypot, but neither had laid eggs.

Index - Return to top

Each species had its own distinct flower preferences for each of the three sessions (FIGURES 4, 5 and 6). Preferences may be linked to the morphology and method of foraging flowers, as well as the length of corolla.

The Compositae provide the bee with a good landing platform and have many nectar sources, so that bees remain on one flower head for a long time. While on such a flower the bee can allow its temperature to drop, so saving energy, but it must then raise its body temperature before it can fly to the next flower. The usual method of raising the temperature is to shiver, this produces heat without moving the wings, and costs about the same in energy terms as flying. However there is another way the bumblebee can produce heat and this is by substrate cycling (Prys-Jones & Corbet, 1987), the enzyme required is fructose bisphosphate, and the process uses less energy for a given amount of heat produced than shivering does. B. lapidarius has over 60% higher fructose bisphosphatase activity than the bee with the next highest activity (B. lucorum) (Prys-Jones & Corbet, 1987). So it is possible for B. lapidarius to keep warm with greater energy efficiency than other bees, while it is foraging on flowers like Compositae, which require a lot of probing for small amounts of reward. In FIGURE 6 B. lapidarius, a short tongue species, shows greater than 75% preference for Compositae. In FIGURE 7 this percentage is increased for males, for whom foraging is not their only task. In FIGURE 4 Allium schoenoprasum is very strongly preferred by B. lapidarius, and the behaviour of the bees on this flower is similar to that on Compositae, as each head provides a good landing platform for multiple probing before flying off to the next head. Sladen (1912) lists the Composite Centaurea nigra as being one of the favourite flowers of B. lapidarius and also of B. terrestris.

In June (FIGURE 4) B. lucorum foraged from Vicia sepium by biting through the corolla, or calyx and corolla, to get at the nectar, this was also found by Brian (1957). B. lapidarius also foraged from V. sepium, but in July. Although the tongue lengths of both species of bee are similar, B. lapidarius always foraged from the flower in the correct manner.

The shorter-tongued bees, Bombus lucorum and B. lapidarius, prefer clumps or elevated flowers that they can land on and probe at random, even though the rewards per probe may be low. Brian (1957) noticed that B. lucorum preferred flowers in more exposed places, while B. pratorum and B. pascuorum preferred flowers in more sheltered places. In this study B. pascuorum was the only forager on Vicia cracca, and also foraged from V. sepium and Lathyrus pratensis, which were lower down in the vegetation. Brian (1951) found that B. pascuorum specialised in Vicia spp. and Trifolium pratense. In this study Vicia spp., Trifolium spp. and Lathyrus pratensis have the same handling difficulty (TABLE 2) as they have the same type of flower. B. hortorum could have foraged from these flowers, but preferred flowers growing in higher or more clumped situations.

B. hortorum (FIGURE 6) visited the fewest flower species, with its long tongue enabling it to specialise on long-corolla flowers such as Digitalis purpurea. The study area does not seem to be well suited to it in that there were rather few long corolla flower species. Tongue length appears to influence flower preference in that short tongues exclude certain species from certain flowers, and the awkwardness of moving a long tongue while standing on a flower make it impracticable for B. hortorum to forage from low reward Compositae. Before landing on a flower B. hortorum partially extends its tongue (Prys-Jones & Corbet, 1987), and when flying from flower to flower it keeps its tongue extended for longer trips than other species do. B. hortorum preferred the more difficult to handle flowers in this study, which may have offered larger rewards. However the range of flower difficulty was not very great in the study area, so this relationship might not exist in other sites with a more diverse range of flowers.

If long-tongued bees could forage profitably on short corolla flowers they would be generalists, and their food niche would overlap with all other bumblebee species, they would also be expected to be more numerous, but that was not the case in this study nor in other places (Ranta & Lundberg, 1980).

B. pratorum, the smallest sized bee, preferred the smaller gullet-shaped flowers (Faegri & van der Pijl, 1979), that possibly provide a greater reward per probe than do the Compositae, but require more movement between probes. During August one B. pratorum nest was located inside a water pump cover. The area was surrounded on three sides by clumps of Centaurea nigra, yet the bees from this nest flew through, around or over these plants to reach the clumps of Lavandula angustifolia and Stachys lanata that were located about 4 m away.

It appears that tongue length influences foraging choice by preventing short tongued bees foraging on long corolla flowers, and long tongued bees from foraging on inflorescences that are landing platforms, or open flowers. So tongue length can explain which flowers the bees will not forage on, but not which flowers they will forage on.

In New Zealand the forage crop clover (Trifolium pratense) did not set seed when introduced last century, so in 1884 some B. terrestris and B. ruderatus queens were sent over from England. Both species were well established within a few years (Sladen, 1912). In 1906 and 1907 B. lapidarius queens were sent, but five years later no B. lapidarius bees could be found (Sladen, 1912). Yet in this study B. terrestris (grouped with B. lucorum) never foraged on T. pratense or any of the other Leguminosae. T. pratense was foraged in this study only by B. pascuorum.

It was expected that the bees would forage from Ulex europaeus (Knuth, 1906) which was found along the bee walk, but they did not. When a bee forages from U. europaeus its weight triggers the release of the style and stamens which rise up to hit the underside of the bee, it is fairly easy to see whether a flower has been triggered or not. Neither of the two large U. europaeus on the bee walk route had any triggered flowers.

Index - Return to top

The preferences of the individuals are less easy to rationalise in relation to tongue length, they form a subset of the species preferences, but apart from that they do seem to be individual preferences. Why one individual chooses one species of flower, whilst another individual of the same species chooses a different species of flower is not clear. It may be just chance. Laverty (1980) observed that bumblebees found the location of nectar in flowers by trial and error, so the bee might decide to major on the first flower on which it located a reward above a certain threshold, after visiting both rewarding and unrewarding flowers, as Heinrich (1979a) observed. One or two bees decided that the pipette was a suitable resource, and seemed determined to revisit it. It did not resemble a flower, but it did provide a large reward. Perhaps within the guild of preferred flowers of a species, the choice of the individual is as Darwin (1891) said "The cause probably lies in insects being thus enabled to work quicker; they have just learnt how to stand in the best position on the flower, and how far and in what direction to insert their proboscides." Bumblebees ability to discriminate between different types of rewarding flowers decreases as the number of types increases (Dukas & Real, 1993), which would make it more profitable for them to restrict their foraging to only one or two species of flower, bypassing others even if they are also rewarding, as it would save handling time and also learning. This leads to the belief that once the major flower has been chosen it will influence or even limit the choice of the minor flowers, which in time may become majors. Laverty (1994) found that when bees in enclosures were forced to switch from one morphologically similar flower to another, they took only one third as long as naive bees to learn to handle the new flower species. So it is possible for a bee to transfer previous learning experience when switching to a flower that is similar in morphology and handling difficulty. Naive bees may be influenced to forage from a flower that has an odour they recognise from the nest, they will have been drinking nectar brought back by other foragers, so may have become conditioned to one or a group of odours. More time may be spent trying to learn to handle a flower with a recognised odour. Insufficient data on major and minor flowers were gathered in this study to test this in the field. Waser, (1986) found that constancy to one type of flower increased as the array of flowers on offer increased in dissimilarity of morphology and colour. And Laverty, (1993) found that when bees were offered bouquets of real flowers, they switched to flowers of similar corolla length more often that could be expected if the choice were random.

Index - Return to top

The remarkable constancy shown by the bees when individuals were followed (FIGURE 13) was quite surprising. Most of the bees were foraging on clumped resources, but as FIGURES 9-12 show, they were aware of other good resources nearby. In particular B49, the Bombus pratorum forager who was followed twice within an hour or so, was on both occasions 100% constant to one of her preferred species, Lavandula angustifolia. Her other preferred species (FIGURE 11), Stachys lanata, was located only 1.5 m from the L. angustifolia clump, so it is unlikely that when the bee was lost sight of she was just moving to the S. lanata. Heinrich (1976) in a similar experiment of following bees found slightly lower rates of constancy, of sixteen foraging trips six were 100% constant, nine 85-98% constant, and one 57% constant, whereas in this study the figures were 100% for eleven trips, and 96% for one trip.

It was previously thought that bumblebees were relatively inconstant foragers (Brian, 1951; Free & Butler 1959; Real 1983), this belief arose largely through the analysis of pollen and the loose use of the term constancy. From data gathered in this small study it appears that on any one foraging trip, bumblebees are far more constant than was previously expected.

This study also shows that constancy varies according to how it is measured. Constancy has been measured at so many levels in the past that comparisons between different studies are difficult.
• Constancy of species was measured by Brian (1957) and in this study. Distinct preferences for certain species of flower were found, usually each species of bumblebee restricted itself to a few species of flower.
• Brian (1951) measured constancy at nest level by analysis of larval faeces. Again distinct preferences were found, similar to those of species level.
Caste constancy was measured here and reflected the species preferences.
Individual preferences were measured by Free, (1970) by pollen analysis of individual loads, who found that about half the loads were mixed, but many of these mixed loads were as much as 98% pure.
Individual preferences were measured by Heinrich (1976, 1979, 1979a); and in this study, and it was found that bees tend to specialise on one or two species of flower; these flowers are a subset of the species preferences.
Individual foraging trip was measured by Heinrich (1976, 1979, 1979a); and in this study, and it was found that there was much higher constancy than at any of the other levels.

At each level from species to individual trip the constancy increases. Consequently the use of the term constancy for anything other than an individual foraging trip is misleading. Free (1970) states that constancy should be measured over only one foraging trip, by either pollen analysis or direct observation. He then points out the main difficulty with direct observation, i.e. it is easy to lose sight of the bee. This difficulty can be partly overcome by marking bees and with miniaturisation and technological advances I am sure the day will soon arrive when a transmitter will be fitted to the thorax of a bee, making direct observation even easier. Pollen analysis will never be able to measure flowers visited for nectar, or pollen that has been discarded or eaten by the bee.

Waser (1986) also recognises the confusion caused by the words "constancy" and "preference". He defines two kinds of preference, and proposes using Bateman's index to measure constancy. Bateman's index can be used to measure accurately flower choice in an enclosure where there are two types of flower in equal numbers, of equal handling difficulty, both either with or without honey guides, and/or UV markings, and with equal quantities of reward at equal concentrations; in short, flowers that are identical in all but colour. Such a convenient situation is rarely found in the field, but where it is Bateman's index could be used.

"Constancy" should be restricted in use to the sequential movement of an individual animal from flower to flower, as it feeds or gathers resources. Measurements of flower usage that cover more than one foraging bout, or consider more than one individual, should be considered as "preferences". Waddington (1983) argues that the term "constancy" be replaced with "floral-visitation-sequences" as this term is "neutral", but surely this is just replacing one perfectly good, though misused term, that can be understood by both the scientific and non-scientific community, with a longer, less well understood term that will probably be misused in time.

Measuring constancy is a very difficult problem as there are so many variables that are constantly changing, and the very act of foraging causes some of the variables to change. Some form of sequential comparison index that can measure runs and breaks, but also takes into account the number of different species visited, and also the availability of the various flowers would be needed. Until such an index is available, constancy over one foraging trip can be assessed by presenting the data as a sequence of visits, as is shown by Heinrich (1976), and in FIGURE 13.

Index - Return to top

(C) Copyright 1999 L. Smith