Spontaneous flower constancy and learning in honey bees as a function of colour

Anim. Behav., 1997, 54, 615–627

Spontaneous flower constancy and learning in honey bees as a function of colour PEGGY S. M. HILL*, PATRICK H. WELLS† & HARRINGTON WELLS* *Faculty of Biological Sciences, The University of Tulsa †Department of Biology, Occidental College (Received 27 March 1996; initial acceptance 31 May 1996; final acceptance 23 December 1996; MS. number: 7571)

Abstract. When presented with an artificial flower patch of blue and yellow pedicellate flowers, individual honey bees, Apis mellifera L., became constant to one of the two flower colours, rarely even sampling the alternative colour. Some bees visited only blue flowers while others visited only yellow flowers. This paper describes the onset of constancy for bees that had had no experience with the experimental apparatus. In 3020 visits, bees failed to land on or drink from the flower colour on which they first landed only 17 times. This behaviour was not modified by quality or quantity of reward, training to the experimental site, group effects or presence of odour during trials. However, when we trained bees to a target painted with two colours and then forced them to sample monomorphic flower patches in sequence, all bees visited the only colour present: yellow or blue. When we subsequently offered these same bees yellow and blue flowers simultaneously (rewarded choices), they became constant. Eleven of 23 bees showed constancy to the less rewarding flower morph without even sampling the alternative. Those bees failed to sample even though they had previously been forced to visit the alternative flower morph, which offered a reward with twice the calories/volume. Constancy is thus spontaneous in honey bees, but it can be hidden by some experimental protocols designed to study ? 1997 The Association for the Study of Animal Behaviour learning. The constancy of honey bees, Apis mellifera L., where individuals restrict their visits to a single plant species even when equally rewarding alternatives are available (Waser 1983), has been well documented since the time of Aristotle (e.g. Ribbands 1953; Heinrich 1975a; Wells & Wells 1983; Waser 1986; Real 1991). Reports show that up to 95% of honey bees foraging at a site visit a single flower species (e.g. Free 1963; Menzel 1990), and that honey bees are poor pollen vectors for agricultural production of hybrid seed (e.g. Free & Williams 1983) because of flower constancy. Likewise, selective reactions to colours have long been studied. Honey bees can be trained to discriminate among colours (Frisch 1914/1915) and are capable of discriminating wavelengths of Correspondence: P. S. Hill, Faculty of Biological Sciences, The University of Tulsa, 600 South College, Tulsa, OK 74104, U.S.A. (email: biol–psh@ centum.utulsa.edu). P. H. Wells is at the Department of Biology, Occidental College, Los Angeles, CA 90041, U.S.A. 0003–3472/97/090615+13 $25.00/0/ar960467

colour without interference or distortion caused by variation in intensity (Daumer 1956). How honey bees use colour cues when foraging is, however, still a lively topic. Wells & Wells (e.g. 1983, 1984, 1986), using artificial patches of blue and yellow ‘flowers’ to which honey bees were attracted with odour cues only (i.e. they had no opportunity to learn to visit both colours during training), showed that an individual bee spontaneously limits visits to one colour morph. Some bees show constancy to blue and others to yellow, even though the bees are from the same hive and are foraging at the same time. The behaviour is persistent, even when rewards differ dramatically in quality, quantity and frequency. Wells & Wells (1983) used the term ‘individual constancy’ to describe this foraging strategy to emphasize a bee’s persistence in visiting a single colour, irrespective of what hive-mates are doing. Both experienced foragers and naive bees (caged indoors since eclosion) will show individual constancy to blue and yellow (C q akmak & Wells 1995).

? 1997 The Association for the Study of Animal Behaviour

615

616

Animal Behaviour, 54, 3

Differences in choices of experimental protocols, however, including which colour pairs are used, seem to lead to a variety of foraging behaviours in honey bees. For example, if experiments are designed to study learning mechanisms in a general way ‘in biologically arbitrary situations’ (Kamil 1983, page 293), constancy may not be seen. In fact, tests using yellow (or orange) and blue colour pairs have shown that honey bees prefer coloured targets with greater volumes of reward or higher frequency of reward irrespective of colour (e.g. Couvillon & Bitterman 1993). The methods used, however, centred on associative learning by animals in the experimental situation and forced sampling of a singly represented (temporally and spatially) reward, followed by extinction trials when no reward was present. Honey bees were not allowed to choose spontaneously between two simultaneously offered rewarded choices of colour, and did not show constancy to either colour morph. Furthermore, when honey bees are forced to sample from both yellow and blue flowers prior to testing, they may or may not later show constancy to one of the colours. Bees presented with equally spaced, equally rewarding choices became constant to blue or yellow (Marden & Waddington 1981). If alternatives were at different distances, however, most bees (but not all) tended to maximize energy by visiting the closest flower morph (Marden & Waddington 1981). Banschbach (1994) reported that bees often maximized energy by choosing a reward of higher sucrose concentration, regardless of colour. Consistency of choices was more pronounced (>84%) when the differences in reward quality were greater, but when the differences in concentrations of sucrose were small, individuals made as few as 21% of their visits to the better rewarding colour. Gould (1987), in a complex experiment designed to test learning of flower-handling, incidentally noted that honey bees can learn to visit both blue and yellow flowers at different times of the day. Bees switched from foraging on blue to foraging on yellow, depending on which choice was rewarded at specific times of day. We have shown that honey bees make foraging choices rapidly and precisely (e.g. Wells et al. 1992), but sampling visits were recorded when a honey bee drank from a reward, not when it landed without drinking or tasting a reward. In the present study, we determined whether naive

honey bees arriving from the hive landed and drank from only one colour in a yellow and blue flower patch, or whether they visited both colours in a brief sampling period before becoming individually constant to one flower morph. In other words, if honey bees are not taught to sample both colours in a choice experiment, do they actually sample on their own? We further determined whether this behaviour is modified by quantity or quality of reward, presence of floral odour during trials, training to the site, group effects or by previous extensive experience with both flower choices. We then discuss the basis of flower constancy and the adaptive significance of sacrificing calories for constancy.

GENERAL METHODS We used Italian honey bees, A. m. ligustica, from an 18-frame hive in experiments with an artificial flower patch simulating pedicellate flowers (for details see Wells et al. 1981, 1992). This patch design allowed us to control flower morphology, quality, quantity and frequency of nectar rewards, morph distribution and abundance, odour and number of foragers. Bees were free flying and had had no previous experience with the experimental apparatus, but they had not been constrained from foraging prior to the testing, and thus were only naive in terms of the experimental design. We made all observations outside in natural lighting. Experiments were conducted in more than one floral environment. Experiment 1 was conducted on the campus of the University of Tulsa (post oak-black jack forest), Oklahoma, in July 1990, and experiment 2 was conducted on the campus of Occidental College (chaparral), Los Angeles, California, in September 1992. Experiment 3 was also at the University of Tulsa; part A was conducted during the first week of March 1994 and part B in June 1995. Each artificial flower consisted of a 30#30-mm Plexiglas square, mounted on a 90-mm pedicel of 6-mm dowelling, and fitted into brown pegboard. We painted the underside of flowers with blue and/or yellow enamel (Testors No. 1208 or 1214: spectral reflectance functions shown in Fig. 1). In each flower we drilled a ‘nectary’, 3 mm deep and 2 mm in diameter, on the upper surface to hold the reward. We arranged the artificial flowers randomly in a 6#6 Cartesian grid with an interfloral

Hill et al.: Honey bee spontaneous flower constancy

Reflectance (%)

100 80

White Yellow

60 40

Blue

20 0 300 350 400 450 500 550 600 650 700 Wavelength (nm)

Figure 1. Reflectance spectral analysis of colours used to paint artificial flowers: blue, yellow and white are Testor’s paints No. 1208, 1214 and 1245, respectively. Redrawn from Wells & Wells (1986 J. Anim. Ecol. 55, 881–889) with permission.

distance of 75 mm, and we refilled the flowers with the same quantity and quality of reward present when each was emptied by a forager. When we changed nectar rewards between sections of an experiment, a new flower patch was used. We washed flowers in unscented detergent, rinsed them with distilled water and air-dried them after each use. Sections of an experiment were performed sequentially and without interruption. We tested data for homogeneity of behaviour among bees and for random visitation of flower colour morphs by individual bees, using the replicated goodness-of-fit test, or G-statistic (Sokal & Rohlf 1981).

EXPERIMENT 1: SELECTIVITY VERSUS REWARD QUALITY FOR FORAGERS TRAINED TO A SITE Methods We trained honey bees to fly approximately 30 m to a clear petri dish containing lightly peppermint-scented 1  (34% weight/weight) sucrose solution (honey bees typically forage on nectars ranging from 10 to 70% sugar: Seeley 1985). The dish was removed and replaced with the artificial flower patch, and bees were allowed to forage freely. Foragers were thus trained to the site, but not to the artificial flowers. When the scented feeding dish was removed, we allowed the bees to chose between blue flowers

617

provisioned with 10 ìl of unscented 2  (68% weight/weight) sucrose or yellow flowers with 10 ìl of unscented 1  sucrose (part A). We marked each bee as she landed on the first flower. We collected data on the foraging of 10 individuals for 1 h, including the flower visited each time the bee touched down (landed), or extended her mouth parts and exhibited the bobbing of the abdomen associated with drinking (drank). In a second section (part B), we conducted the same experiment with 10 new bees, but reversed the provisioning conditions: the yellow flowers offered 10 ìl of 2  sucrose and the blue flowers 10 ìl of 1  sucrose. Results The colour of the first flower visited by each arriving bee was the colour to which it returned repeatedly for the duration of the experiment. In 2281 visits, bees failed to land on or drink from the flower colour on which they first landed only 16 times. When blue flowers held a 2  sucrose reward and yellow flowers held a 1  sucrose reward (Table I), six of the 10 bees were constant to yellow. Although the caloric reward offered by yellow flowers was half that of blue flowers, during 642 visits only one of the six bees preferring yellow even sampled the reward on blue. That bee visited a blue flower only once. When reward quality was reversed and yellow flowers offered the better reward (Table II), five bees were constant to blue, never even landing on yellow flowers. The other five preferred yellow, landing on blue nine times out of 550 total visits. Honey bees did not forage homogeneously as a group in either part of experiment 1 (part A: GH 9 =1458.1, P<0.001; Table I; part B: GH 9 =1539.8, P<0.001; Table II). Furthermore, they did not forage randomly as a group in either section, neither when blue flowers offered a 2  sucrose reward (part A: GP 1 =28.861, P<0.001; Table I), nor when yellow flowers offered the 2  reward (part B: GP 1 =6.056, P<0.025; Table II). In fact, the data as a whole did not fit the 1:1 ratio expected with random foraging (part A: GT 10 =1486.979, P<0.001; Table I; part B: GT 10 =1545.858, P<0.001; Table II). Not even one bee foraged randomly. Each bee showed a marked preference for a flower colour, some to blue and others to yellow (individual constancy).

Animal Behaviour, 54, 3

618

Table I. Foraging behaviour of honey bees on a yellow–blue dimorphic flower patch when blue flowers held the better quality reward (experiment 1) Yellow (1  sucrose) Bee number 1 2 3 4 5 6 7 8 9 10 Subtotals Yellow preference Blue preference Totals

Blue (2  sucrose)

Drank

Landed

Drank

Landed

Total

103* 93* 108* 100* 108* 90* 4 0 0 0

3 6 6 9 5 10 2 0 0 0

0 1 0 0 0 0 98* 118* 115* 94*

0 0 0 0 0 0 15 9 11 7

106 100 114 109 113 100 119 127 126 101

602 4 606

39 2 41

1 425 426

0 42 42

642 473 1115

The values are numbers of actual visits of each of 10 individuals over a 2-h period. The asterisk indicates the colour on which a bee landed on its first visit from the hive. P<0.001.

EXPERIMENT 2: SELECTIVITY VERSUS REWARD QUANTITY FOR SINGLE FORAGERS Methods On 2 consecutive days, we squirted a lightly clove-scented 1.5  sucrose solution on the

entrance board of the hive, at half-hour intervals, for 2 h. Many bees collected the scented sucrose from the entrance; others carried it into the hive on their bodies after having landed in the solution as they returned from foraging. During this time, neither artificial flowers nor reward dishes were present.

Table II. Foraging behaviour of honey bees on a yellow–blue dimorphic flower patch when yellow flowers held the better quality reward (experiment 1) Yellow (2  sucrose) Bee number 1 2 3 4 5 6 7 8 9 10 Subtotals Yellow preference Blue preference Totals

Blue (1  sucrose)

Drank

Landed

Drank

Landed

Total

80* 94* 0 0 0 123* 0 0 105* 96*

14 6 0 0 0 8 0 0 10 5

5 1 136* 112* 137* 0 125* 70* 0 0

2 1 9 9 8 0 5 5 0 0

101 102 145 121 145 131 130 75 115 101

498 0 498

43 0 43

6 580 586

3 36 39

550 616 1166

The values are numbers of actual visits of each of 10 bees over a 2-h period. The asterisk indicates the colour on which a bee landed on its first visit from the hive. P<0.025.

Hill et al.: Honey bee spontaneous flower constancy

619

Table III. Foraging behaviour of honey bees on a yellow–blue dimorphic flower patch when the two flower morphs differed in quantity of reward (experiment 2) Bee number 1 2 3 4 5 6 7 8 9 10 Total

Yellow (20 ìl 1.5  sucrose)

Blue (2.5 ìl 1.5  sucrose)

Total

0 0 40 0 1 0 0 34 44 31 150

131 116 0 40 42 149 111 0 0 0 589

131 116 40 40 43 149 111 34 44 31 739

The values are actual numbers of visits of 10 individual bees. P<0.001.

On the third day, we placed the artificial flower patch on the experimental site, 60 m from the hive, before the hive’s entrance board was squirted with the reward solution. Blue flowers were provisioned with 2.5 ìl, and yellow flowers with 20 ìl, of the clove-scented 1.5  sucrose solution. We marked the first naive bee arriving from the hive and recorded its foraging choices. All other bees landing on the flower patch were captured and removed. We also removed the marked bee at the end of that day’s observations to control for prior site experience of foragers and possible group effects. We tested 10 honey bees singly in this manner. Results Six bees were constant to blue flowers. Only one of the bees preferring blue sampled yellow flowers, and it visited yellow only a single time. Four bees preferred yellow flowers, which held a 20 ìl reward, and none of these visited blue flowers. In 739 visits, honey bees failed to choose the colour morph first visited only once. Single honey bees visiting the artificial flower patch did not forage homogeneously as a group (GH 9 =736.2, P<0.001; Table III) nor did they forage randomly as a group (GP 1 =278.8, P<0.001; Table III). Again, not even one bee foraged randomly as an individual, and the data as a whole did not fit the 1:1 ratio expected with random foraging (GT 10 =1015.0, P<0.001).

EXPERIMENT 3: SELECTIVITY VERSUS REWARD QUALITY FOR FORAGERS HAVING EXPERIENCE WITH TWO TRAINING REWARDS Methods We trained honey bees to fly approximately 30 m to a clear petri dish containing lightly clovescented 1  sucrose solution. We removed the scented dish and replaced it with the artificial flower patch provisioned with 10 ìl of the same scented 1  sucrose solution. Half of each flower was painted blue and the other half yellow, so that bees encountered both colours every time they visited a flower. Bees were marked and allowed to forage freely for 75 min. We recorded flower visits for 11 individuals in part A and 12 in part B. First, we replaced the dichromatic flowers with 36 blue flowers containing 10 ìl of an unscented 2  sucrose reward, and we recorded flower visits by the 11 marked bees for 75 min. We then replaced the blue patch with a grid of 36 yellow flowers containing 10 ìl of an unscented 1  sucrose reward and observed the same 11 bees for another 75 min. The last section of the experiment involved replacing the yellow patch with a flower patch containing 18 blue flowers and 18 yellow flowers, randomly arranged in the Cartesian grid. We filled yellow flowers with the 1  sucrose and blue flowers with 2  sucrose, as in the previous

Animal Behaviour, 54, 3

620

Table IV. Foraging behaviour of honey bees visiting blue and/or yellow artificial flowers of different reward quality, after being pre-trained to visit a blue/yellow target (experiment 3A)

Bee 1 2 3 4 5 6 7 8 9 10 11 Total

Dimorphic patch

Blue/yellow target 1  sucrose

Blue patch 2  sucrose

Yellow patch 1  sucrose

2  sucrose blue

1  sucrose yellow

Total

36 40 35 11 25 16 12 36 30 30 32 303

29 23 20 25 20 24 11 13 17 18 18 218

30 44 23 39 33 36 25 33 35 27 24 349

0 0 45 0 33 17 0 0 20 43* 46 204

44* 39* 1* 42* 16* 15* 31* 36* 10* 2 6* 242

139 146 124 117 127 108 79 118 112 120 126 1316

The values are numbers of actual visits of each of 11 bees in a single observation period of 5 h. The asterisk indicates the flower colour on which a bee landed on its first trip in the fourth part of the experiment.

sections. We changed the distribution of flower morphs every 15 min to avoid a bee’s association of reward with a particular spatial pattern. We observed the same 11 bees for one more 75-min period. Part B was a reverse experiment of the above. The dichromatic patch was replaced by yellow flowers, provisioned with 2  sucrose, and these were replaced by a blue patch provisioned with 1  sucrose. In the last section of the experiment, 18 blue flowers (1  sucrose) and 18 yellow flowers (2  sucrose) were randomly arranged in the Cartesian grid. We observed 12 new bees in this section in the same way as in part A. Results All 11 bees trained to a blue/yellow target in part A subsequently made repeated visits to monomorphic patches of first blue and then yellow flowers (Table IV). When blue and yellow flowers were presented simultaneously, 10 of the 11 bees visited yellow flowers on their first trip to the flower patch after the simultaneous choice was offered; they continued to visit the last colour presented singly. One bee sampled blue on its first visit from the hive after both colours were presented simultaneously. She sampled yellow twice, but showed constancy to blue from the beginning. Four other bees switched to blue after initially

visiting yellow, one having visited yellow only on its first visit from the hive. Only one bee indiscriminately visited both colours. Five of the 10 bees never sampled blue when both colours were presented simultaneously, even though blue flowers contained twice the caloric reward as yellow, and even though these same bees had experienced this higher reward only a short time previously. Single honey bees visiting the dimorphic artificial flower patch in part A did not forage homogeneously as a group (GH 10 =407.5, P<0.001; Table IV). Bees as a group did forage in a 1:1 ratio (GP 1 =3.2, P>0.05; Table IV), even though the data as a whole differed significantly from random foraging (GT 11 =410.8, P<0.001; Table IV). Total visits to blue and to yellow were essentially the same (Table IV). Only three of the 11 bees (5, 6 and 9) appeared to have foraged randomly when the total G was partitioned (G1 <3.1, P>0.05; Table IV). Of these, bee 6 visited both colours throughout the section. Bee 5 made her last 31 visits to blue, and bee 9 made her last 20 visits to blue (Table V). These two bees became constant to the reward on blue after making several visits to yellow. In part B, the 12 bees trained to the blue/yellow target also visited monomorphic patches of yellow or blue (Table VI). When yellow and blue flowers were offered simultaneously in the last section, one bee failed to return, but eight of the remaining

Hill et al.: Honey bee spontaneous flower constancy

621

Table V. Sequence of visits of individual honey bees to the dimorphic flower patch in experiment 3A Bee

Colour sequence

1 2 3 4 5 6 7 8 9 10 11

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY YBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY YYYYYYBBYYYYYYYYYYBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB YYYYYYYBYBBYYBBBBYBBYYBBBBBBYYBB YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY YYYYYYYYYYBBBBBBBBBBBBBBBBBBBB BBBYYBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB YYYYYBBBBYBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB

Yellow flowers held a 1  sucrose reward and blue flowers held a 2  sucrose reward. All bees had been pre-trained to a blue/yellow target and then visited first a blue monomorphic patch and then a yellow monomorphic patch before being given blue and yellow flowers offered simultaneously with rewards.

11 visited blue, the last colour offered singly, on their first trip back from the hive. Two bees visited a yellow flower, which contained the 2  sucrose reward, on their first trip from the hive, but they switched to blue flowers with the 1  sucrose reward and never returned to yellow. A third bee visited yellow first and sampled blue only once out of 57 visits in that section. Eight bees were individually constant to blue flowers with a 1 

reward, and three were individually constant to yellow flowers with a 2  reward. Individuals visiting the dimorphic artificial flower patch in part B did not forage homogeneously as a group (GH 11 =479.8, P<0.001; Table VI) nor did they forage in a 1:1 ratio (GP 1 =112.1, P<0.001; Table VI), randomly as a group (GT 12 =591.9, P<0.001; Table VI), or randomly as individuals (partitioned G1 >15.2, P<0.001).

Table VI. Foraging behaviour of honey bees visiting blue and/or yellow artificial flowers of different reward quality, after being pre-trained to visit a blue/yellow target (experiment 3B)

Bee 1 2 3 4 5 6 7 8 9 10 11 12 Total

Dimorphic patch

Blue/yellow target 1  sucrose

Yellow patch 2  sucrose

Blue patch 1  sucrose

2  sucrose yellow

1  sucrose blue

Total

48 40 34 18 59 51 45 40 61 44 52 60 552

43 35 40 25 39 52 36 40 53 34 24 46 467

63 54 68 39 54 59 28 51 49 47 60 68 640

0 38 1* 0 44 0 0 0 1* 0 56* 0 140

56* 10* 55 11* 8* 43* 0 48* 43 49* 1 52* 376*

210 177 198 93 204 205 109 179 207 174 193 226 2175

The values are numbers of actual visits of each of 12 bees in a single observation period of 5 h. The asterisk indicates the flower colour on which a bee landed on its first trip in the fourth part of the experiment.

Animal Behaviour, 54, 3

622

Individuals showed a preference for the colour blue, which was the last colour they visited in the monomorphic patches.

DISCUSSION When free-flying honey bees, naive to the experimental apparatus, arrive at an artificial flower patch of pedicellate yellow or blue flowers, they spontaneously choose one flower morph and become constant to that colour. We observed this behaviour whether multiple bees that were trained by odour to the site were offered rewards of different quality (experiment 1), or single bees following an odour trail were offered rewards of different quantity (experiment 2). Individuals not trained to visit both flower alternatives prior to testing often fail to sample (or even land on) the alternative flower type, a behaviour observed in honey bees collecting pollen (Free 1963; Moezel et al. 1987) and discussed in detail elsewhere (e.g. Marden & Waddington 1981; Wells et al. 1981; Wells & Wells 1983). When honey bees received extensive experience with both yellow and blue flower colours on the same experimental apparatus, by offering them both colours on one target and then each colour in sequence with differing caloric rewards, they visited the only flower morph presented. If they were then offered simultaneously blue and yellow flowers, they became individually constant, some to blue and others to yellow, even when reward quality differed (experiment 3). Bees often failed to sample both colours, even though they had just successfully extracted different caloric rewards from each colour. This behaviour is much different from that seen when colours are offered simultaneously but without a reward in extinction tests (Buchanan & Bitterman 1989). Bees in that environment discriminate between colours, preferring the association with the previously offered higher energetic reward rather than the colour. Our results support interpretations of our previous constancy reports (where we gave simultaneous rewarded choices without pre-training to colour), as well as those that show that bees will disregard colour cues to maximize energy (when extensive training to both colour morphs was employed prior to testing), depending solely on the experimental design. Our results also support predictions from the colour-learning literature

that honey bees can be trained to two colours (Menzel & Erber 1978; Menzel 1985) and that long-term memory can be established in less than a minute if several short rewards follow each other quickly (Erber 1975). When honey bees are trained to a target that is half blue and half yellow (experiment 3), they apparently learn that either colour is an acceptable cue representing food. When they are offered a choice of two colours that are both rewarded, they form a search image for one colour. On the other hand, if bees are not forced to sample alternatives, and alternatives all provide a reward, they continue to visit the flower on which they first landed (experiments 1 and 2). Three rewarded approaches are sufficient to establish long-term memory of the search image (Lindauer 1975), and Menzel (1990, page 251) stated: ‘A stable memory for a color signal lasting for at least 2 weeks is already established after 3 learning trials’. Unless we study a bee’s behaviour when it first encounters a foraging choice, as Kamil (1983) suggested, without prior associative learning, we fail to see the constancy shown by honey bees in experiments 1 and 2 (e.g. Couvillon & Bitterman 1993). Without pre-training and forced sampling of one morph at a time (to ensure experience with both yellow and blue), we do not see modification of constancy due to learning (e.g. Wells & Wells 1983, 1984, 1986). Experiments that stop after pre-training and forced sampling (without offering blue and yellow simultaneously with rewards) do not show constancy in honey bees, even after extensive experience with both morphs and even when one reward offers twice the calories (experiment 3). Banschbach (1994) showed that some honey bees with experience with both yellow and blue will ignore colour to visit the greater reward preferentially when given a binary choice, but she did not conduct a reverse experiment to verify that individuals visiting yellow would switch to blue (or vice versa) when rewards were switched. Our results suggest that her bees would continue to visit the colour they had associated with the better reward, as long as some reward was provided by these flowers. A learning disposition (Tinbergen 1951; Lindauer 1975) or preparedness (Menzel 1985, 1990) to associate a particular colour with food is probably a ‘hard-wired’ species characteristic of honey bees (C q akmak & Wells 1995; Giurfa et al. 1995). On the other hand, honey bees retain a

Hill et al.: Honey bee spontaneous flower constancy flexibility of choice in establishing the colour to which they become constant, just as they show flexibility in the timing of learning of flower or landmark cues (Lehrer 1993). We have shown that if the first flower morph chosen provides a series of rewards, that flower type will be visited repeatedly and with constancy. Based on learning theory, we predict that if the first flower type chosen does not provide three rewards in sequence, thus promising a full crop load after repeated sampling on the first foraging trip, the bee will sample other flowers (and colours) in the vicinity. It follows that, if we had omitted training to two colours in experiment 3 and first gave bees a blue monomorphic patch (all rewarding flowers), these same bees would not have continued to forage on a replacement patch of only yellow flowers. They would have developed a constant colour search image for that time and place for blue. Later in the day, the same bees might have switched to yellow flowers (e.g. Gould 1987) if blue ones were not available. We predict that blue-constant foragers continue their fidelity to colour unless that rewarded choice is not present. Not all colour pairs elicit the same behaviour from foraging honey bees, however. When flower colours have similar spectral reflectance, reminiscent of the variety of hues present intraspecifically in a natural flower patch, the difference in colour does not elicit constancy. Although Italian honey bees visiting a blue-yellow dimorphic patch of pedicellate artificial ‘flowers’ are constant to one colour, those offered a blue-white dimorphic patch (using the pedicellate flowers) forage randomly when rewards are equal and selectively visit the morph with the greater caloric reward when quality varies (Wells et al. 1992). Therefore, bees visiting blue or white flowers can discriminate between these two colours but choose between them based on calories rather than colour (Wells et al. 1992). If a blue-yellow-white trimorphic patch is used, a bee that visits yellow will not visit blue or white flowers. A bee that samples from either a blue or white flower on its first visit to the trimorphic patch will visit both blue and white flowers but will rarely even sample from yellow flowers (Wells & Wells 1986). These behavioural observations are supported by Chittka’s physiological studies of honey bee colour vision in that human-blue and human-white are close to each other, but are separated from human-yellow, in a

623

E(B)

450 * * *

*

*

400

* *

*

Blue 500 * White Background * Yellow *

*

350 ** **** 300 *

*

*

*

540

0.1:0.9

0.9:0.1

*

E(U)

*

*

0.5:0.5 *

*

*

*

*

*

*

E(G)

Figure 2. Colour hexagon diagram for blue (b), yellow (y), and white (w) colours of artifical flowers, showing the relative position of the colour loci of stimuli in the honey bee colour space. (The diagram was prepared by L. Chittka.)

honey bee’s colour space (e.g. Chittka 1992; Chittka et al. 1994; Fig. 2). Darwin (1895) knew that honey bees were not deterred by natural variations of hue within flowers of a species: ‘Humble and hive-bees are good botanists, for they know that varieties may differ widely in the colour of their flowers and yet belong to the same species’ (page 416). Chittka established a physiological basis for perception that might lead to this behaviour (Chittka et al. 1994). Honey bees perceive colours based on differential excitation of the three colour receptors in the bee eye. Colours are grouped together in a honey bee’s colour space (Chittka et al. 1994) based on spectral properties. Helversen (1972) showed that honey bee colour discrimination is poorest between close wavelengths in the ultraviolet, blue and yellow regions of their visible spectrum, but Lehrer & Bischof (1995) found that honey bees discriminate easily between any yellow pigment and any blue or violet tested. Our preliminary results show that Italian honey bees can readily distinguish between white, blue and violet, or between yellow and orange, but within these two groups, colour cues do not constrain foraging behaviour (P. Hill & H. Wells, unpublished data). Therefore, blue and white are treated as ‘similar’, and honey bees choose between the colours of similar spectral reflectance for the reward

624

Animal Behaviour, 54, 3

containing the most calories (Wells et al. 1992). Between the ‘blue’ and ‘yellow’ groups of colours, or between the ultraviolet and the ‘blue’ groups, differences are great enough to constrain foraging, and individuals become constant to one colour, regardless of calories. This connection between physiology and behaviour offers promise of, and indeed requires, a whole range of new experiments. Interpretation of previous studies, which have not compensated for the different treatment of blue–white versus blue–yellow colour dimorphism by Italian honey bees in the experimental design, may need to be reconsidered, although it does not necessarily follow that the conclusions will be falsified with this additional control. For example, Laverty (1994) concluded that more complex flowers elicited constancy in bumblebees but simpler flowers did not. His complex flowers were orange and purple, however, colours to which we would expect bees to become individually constant. The simpler flowers were pink/white, magenta, blue-violet or purple/white and may not have elicited constancy from the individuals. Greggers & Menzel (1993) concluded that flower constancy was ‘not a general, stereotyped behaviour’ (page 27) when they observed honey bees visiting flowers based on flow rate of nectar rather than with constancy to the distinctive colours used in their experiments. Specifically, honey bees visited both white and violet when these colour morphs delivered the same rate of reward, but more visited the violet morph when its flow rate was increased. These colour choices may not have been distinctive to honey bees and may not have elicited constancy, even though they appear distinctive to humans. Still, how can constancy have resulted from natural selection if honey bees sacrifice calories through constancy to a less rewarding flower morph? The eusocial honey bee has complicated individual behaviour patterns that must also be studied in terms of the whole hive. Individual constancy foraging may be viewed as another form of work partitioning, a behaviour well documented for other activities within the hive (Seeley 1985) that enhance bees’ inclusive fitness (Hamilton 1964). Even though colony-level selection does not work at the expense of individual success, group foraging is ‘designed to achieve high efficiency for the colony as a whole, not for each individual forager’ (Seeley 1985, page 80).

The probability that each foraging worker will return to the hive with a full crop load increases, however, and handling time decreases, once she has become constant to one flower colour. At the same time, since each bee individually associates a colour with food, foragers as a group should visit representatives of all profitable flowering species in the vicinity on any given day. In this way a single resource will not be overwhelmed by all the hive’s foragers competing for the same nectar (Heinrich 1975b), individuals will increase the efficiency with which they extract nectar from the single morph (Darwin 1895), and sibling interference in foraging will be minimized (Wells & Rathore 1994). Even though some workers obtain nectar of a lower volume or caloric reward than others, all workers forage so that the hive is efficiently provisioned. Heinrich (1975b) commented: ‘The most rewarding flowers are not necessarily the ones of greatest food production. Profit is also related to the number of other bees already foraging from the flowers, and on how well the foragers have learned to manipulate the flowers’ (page 326). Varju´ & Nu´n˜ez (1991) found that hive-mates return with a partially filled crop ‘when bees forage in team on a less favorable patch of flowers’ (page 735). Their model suggested that Italian honey bees could ‘afford’ this behaviour because of their frequent access to information on alternative nectar sources from sister scouts at the hive. Honey bees show great plasticity in using sensory cues (Giurfa et al. 1994), and colour is not the only cue used to make floral choices. Odour (Wells & Wells 1985), shape (Lamb & Wells 1995) and pattern (Petrikin & Wells 1995) influence foraging behaviour, but these factors were held constant in our current experiments to test the role of colour. We also chose not to vary distance between flowers in our protocol, even though numerous reports in the literature suggest that constancy generally declines with an increase in inter-floral distance. Italian honey bees remain constant to yellow or blue when the closest flower is the alternative colour (Wells et al. 1981), and individuals will fly over two or more flowers of the alternative to forage on the colour morph chosen when they first landed on the patch (Wells et al. 1986). These choices were made even though the rejected flower morph was more common and held twice the quantity of reward. Levin & Anderson’s (1970) general model assumed that

Hill et al.: Honey bee spontaneous flower constancy pollinators always visit their nearest neighbouring plant, so it does not apply to honey bees. Other studies with distance were with bumblebees (e.g. Waser 1983; Brown & Clegg 1984) which are variable, although non-random, foragers that select from a wide variety of flower choices and then restrict visits to a major and one or more minor species (Heinrich 1976, 1979; Oster & Heinrich 1976). Marden & Waddington (1981) simultaneously varied both colour and distance and concluded that honey bees switch strategies away from constancy to visit mostly closer flowers; they also forced sampling of both blue and yellow colour morphs prior to testing. Their results were further complicated by the use of scented reward, since honey bees also show odour constancy under some experimental conditions (Wells & Wells 1985). Our reward volumes are higher than those found in many natural flowers, but not higher than those used by others in experiments with honey bees (e.g. Marden & Waddington 1981: 2 ìl; Buchanan & Bitterman 1989: 5 ìl). We have also shown (Wells & Wells 1984) that constancy to blue and yellow flower morphs by Italian honey bees is not influenced by reward quantities in the range of 2.5 ìl–100 ìl. Greggers & Menzel’s (1993) report of decline in constancy with use of 1 ìl rewards, however, warrants further study using controls for flower colour. Compelling questions still remain to be answered in studying flower constancy in the genus Apis. The Indian hive bee, A. cerana indica, is not flower-constant when given the option of blue or yellow flowers (Wells & Rathore 1994). Since this species probably does not perceive colours differently from Italian honey bees (Peitsch et al. 1992; Chittka 1996), its behaviour suggests that interspecific differences in these two eusocial species have resulted in different selective pressures from the environment. The Indian hive bee is physically smaller and forms colonies of only about one-eighth the size of Italian honey bees, so behavioural constraints in foraging may not have developed in response to work partitioning and competition from hive-mates (Wells & Rathore 1994). Comparative studies across the genus and among races of A. mellifera (Varju´ & Nu´n˜ez 1991; i. C q akmak & H. Wells, unpublished data) show promise of clarifying issues of selection in different environments, which have hindered efforts to extrapolate behaviour across

625

taxa, as general predictive models of constancy behaviour are developed. Different information from the environment takes precedence in a bee’s flower choice, depending upon the experimental design used. What appear to be minor changes in protocol elicit greatly different behavioural responses, even without prior learning experience. Throughout a vast literature on foraging behaviour, it seems that honey bees will play the games we invite them to play, whether or not we understand the rules they establish for themselves. The challenge remains to discover which cues from the environment the honey bee uses to make choices, and of course, ultimately, the source of the choice.

ACKNOWLEDGMENTS We thank P. L. Schwagmeyer, V. H. Hutchison and L. Devenport for pre-submission comments and encouragement. We are grateful for comments of N. Waser, J. Gould and an unnamed referee. We are especially grateful to L. Chittka for review comments and preparation of Fig. 2. We thank undergraduate students H. Joost and B. Silkey for their help with observations. Research was supported in part by USDA grant 9303936 and NSF (REU) grant BIR-9300197 to H. W. and a Faculty Research Grant from the University of Tulsa to P.S.H.

REFERENCES Banschbach, V. S. 1994. Colour association influences honey bee choice between sucrose concentrations. J. comp. Physiol. A, 175, 107–114. Brown, B. A. & Clegg, M. T. 1984. Influence of flower color polymorphism on genetic transmission in a natural population of the common morning glory, Ipomoea purpurea. Evolution, 38, 796–803. Buchanan, G. M. & Bitterman, M. E. 1989. Learning in honey bees as a function of amount of reward: tests of the equal-asymptote assumption. Anim. Learn. Behav., 17, 475–480. C q akmak, i. & Wells, H. 1995. Honey bee forager individual constancy: innate or learned? Bee Science, 3, 165–173. Chittka, L. 1992. The colour hexagon: a chromaticity diagram based on photoreceptor excitations as a generalized representation of colour opponency. J. comp. Physiol. A, 170, 533–543. Chittka, L. 1996. Does bee color vision pre-date the evolution of flower color? Naturwissenschaften, 83, 136–138.

626

Animal Behaviour, 54, 3

Chittka, L., Shmida, A., Troje, N. & Menzel, R. 1994. Ultraviolet as a component of flower reflections, and the colour perception of Hymenoptera. Vision Res., 34, 1489–1508. Couvillon, P. A. & Bitterman, M. E. 1993. Learning in honey bees as a function of amount of reward: further experiments with color. Anim. Learn. Behav., 21, 23–28. Darwin, C. 1895. The Effects of Cross and Self Fertilisation in the Vegetable Kingdom. New York: Appleton. Daumer, K. 1956. Reizmetrische Untersuchung des Farbensehens der Bienen. Z. vergl. Physiol., 38, 413– 478. Erber, J. 1975. The dynamics of learning in the honey bee (Apis mellifica carnica): I. The time dependence of the choice reaction. J. comp. Physiol., 99, 231–242. Free, J. B. 1963. The flower constancy of honey bees. J. Anim. Ecol., 32, 119–132. Free, J. B. & Williams, I. H. 1983. Foraging behavior of honey bees and bumblebees on Brussels sprouts grown to produce hybrid seed. J. Apicult. Res., 22, 94–97. von Frisch, K. 1914/1915. Der Farbensinn und Formensinn der Biene. Zool. Jahrb. Physiol., 35, 1–188. Giurfa, M., Nun˜ez, J. & Backhaus, W. 1994. Odour and colour information in the foraging choice behaviour of the honey bee. J. comp. Physiol. A, 175, 773–779. Giurfa, M., Nun˜ez, J., Chittka, L. & Menzel, R. 1995. Colour preferences of flower-naive honey bees. J. comp. Physiol. A, 177, 247–259. Gould, J. L. 1987. Honey bees store learned flowerlanding behaviour according to time of day. Anim. Behav., 35, 1579–1581. Greggers, U. & Menzel, R. 1993. Memory dynamics and foraging strategies of honey bees. Behav. Ecol. Sociobiol., 32, 17–29. Hamilton, W. D. 1964. The genetical evolution of social behavior. J. theor. Biol., 7, 1–52. Heinrich, B. 1975a. Energetics of pollination. A. Rev. Ecol. Syst., 6, 139–170. Heinrich, B. 1975b. Bee flowers: a hypothesis on flower variety and blooming times. Evolution, 29, 325–334. Heinrich, B. 1976. The foraging specializations of individual bumblebees. Ecol. Monogr., 46, 105–128. Heinrich, B. 1979. ‘Majoring’ and ‘minoring’ by foraging bumblebees, Bombus vagans: an experimental analysis. Ecology, 60, 245–255. Helversen, O. von 1972. Zur spektralen Unterschiedsempfindlichkeit der Honigbiene. J. comp. Physiol., 80, 439–472. Kamil, A. C. 1983. Optimal foraging theory and the psychology of learning. Am. Zool., 23, 291–302. Lamb, J. M. & Wells, H. 1995. Honey bee (Apis mellifera) use of flower form in making foraging choices. J. Kans. entomol. Soc., 68, 388–398. Laverty, T. M. 1994. Costs to foraging bumble bees of switching plant species. Can. J. Zool., 72, 43–47. Lehrer, M. 1993. Why do bees turn back and look? J. comp. Physiol. A, 172, 549–563. Lehrer, M. & Bischof, S. 1995. Detection of model flowers by honey bees: the role of chromatic and

achromatic contrast. Naturwissenschaften, 82, 145– 147. Levin, D. A. & Anderson, W. W. 1970. Competition for pollinators between simultaneously flowering species. Am. Nat., 104, 455–467. Lindauer, M. 1975. Evolutionary aspects of orientation and learning. In: Function and Evolution in Behaviour: Essays in Honour ofProfessor Niko Tinbergen, F.R.S. (Ed. by G. Baerends, C. Beer & A. Manning), pp. 228–242. Oxford: Clarendon Press. Marden, J. H. & Waddington, K. D. 1981. Floral choices by honey bees in relation to the relative distances to flowers. Physiol. Entomol., 6, 431–435. Menzel, R. 1985. Learning in honey bees in an ecological and behavioral context. In: Experimental Behavioral Ecology and Sociobiology: in Memoriam Karl von Frisch 1886–1982 (Ed. by B. Holldobler & M. Lindauer), pp. 55–74. Sunderland, Massachussetts: Sinauer Associates. Menzel, R. 1990. Learning, memory, and ‘cognition’ in honey bees. In: Neurobiology of Comparative Cognition (Ed. by R. P. Kesner & D. S. Olton), pp. 237–292. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Menzel, R. & Erber, J. 1978. Learning and memory in bees. Scient. Am., 239, 102–110. Moezel, G., van der Delfs, J. C., Plate, J. S., Loneragaw, W. A. & Bell, D. T. 1987. Pollen selection by honey bees in shrublands of the northern sandplains of western Australia. J. Apicult. Res., 216, 224–232. Oster, G. & Heinrich, B. 1976. Why do bumblebees major? a mathematical model. Ecol. Monogr., 46, 129–133. Peitsch, D., Fietz, A., Hertel, H., de Souza, J., Ventura, D. F. & Menzel, R. 1992. The spectral input systems of hymenopteran insects and their receptor-based colour vision. J. comp. Physiol A, 170, 23–40. Petrikin, J. & Wells, H. 1995. Honey bee (Apis mellifera) use of flower pigment patterns in making foraging choices. J. Kans. entomol. Soc., 68, 377–387. Real, L. A. 1991. Animal choice behavior and the evolution of cognitive architecture. Science, 253, 980– 985. Ribbands, C. R. 1953. The Behavior and Social Life of Honey Bees. London: Bee Research Association. Seeley, T. D. 1985. Honey Bee Ecology: A Study of Adaptation in Social Life. Princeton, New Jersey: Princeton University Press. Sokal, R. R. & Rohlf, F. J. 1981. Biometry: The Principles and Practice of Statistics in Biological Research. 2nd edn. New York: W. H. Freeman. Tinbergen, N. 1951. The Study of Instinct. Oxford: Clarendon Press. Varju´, D. & Nu´n˜ez, J. 1991. What do foraging honey bees optimize? J. comp. Physiol. A, 169, 729–736. Waser, N. M. 1983. The adaptive nature of floral traits: ideas and evidence. In: Pollination Biology (Ed. by L. Real), pp. 241–285. Orlando, Florida: Academic Press. Waser, N. M. 1986. Flower constancy: definition, cause and measurement. Am. Nat., 127, 593–603.

Hill et al.: Honey bee spontaneous flower constancy Wells, H. & Rathore, R. R. S. 1994. Foraging ecology of the Asian hive bee, Apis cerana indica, within artificial flower-patches. J. Apicult. Res., 33, 219–230. Wells, H. & Wells, P. H. 1983. Honey bee foraging ecology: optimal diet, minimal uncertainty or individual constancy? J. Anim. Ecol., 52, 829–836. Wells, H. & Wells, P. H. 1986. Optimal diet, minimal uncertainty and individual constancy in the foraging of honey bees, Apis mellifera. J. Anim. Ecol., 55, 375–384. Wells, H., Wells, P. H. & Smith, D. M. 1981. Honey bee responses to reward size and colour in an artificial flower patch. J. Apicult. Res., 20, 172–179.

627

Wells, H., Wells, P. & Contreras, D. 1986. Effects of flower-morph frequency and distribution on recruitment and behaviour of honey bees. J. Apicult. Res., 25, 139–145. Wells, H., Hill, P. S. & Wells, P. H. 1992. Nectarivore foraging ecology: rewards differing in sugar types. Ecol. Entomol., 17, 280–288. Wells, P. H. & Wells, H. 1984. Can honey bees change foraging patterns? Ecol. Entomol., 9, 467–473. Wells, P. H. & Wells, H. 1985. Ethological isolation of plants 2. Odour selection by honey bees. J. Apicult. Res., 24, 86–92.