Ecological Entomology (2004) 29, 52–59

Accumulating wing damage affects foraging decisions in honeybees (Apis mellifera L.) A . D . H I G G I N S O N and C . J . B A R N A R D

Animal Behaviour and Ecology Research Group,

School of Biology, University of Nottingham, U.K.

Abstract. 1. Nectar-foraging honeybees (Apis mellifera) on lavender (Lavandula stoechas) appear to forage so as to maximise net energy return from foraging bouts; however, evidence from other studies suggests that foraging has a detrimental effect on survival, due at least in part to physiological deterioration of the flight mechanism. But foragers also acquire wing damage during foraging, which may increase foraging effort and reduce foraging lifespan. 2. The accumulation of damage over time and its effects on foraging flight and flower choice were studied in the field using a system in which the criteria for flower preference by foragers was known from previous work. Wing damage accumulated exponentially over time and resulted in foragers becoming less choosy about the flowers they visited. 3. Damage added experimentally contributed independently to the effect on choosiness. Effects of wing damage (natural and added experimentally) were also independent of those of a relative measure of age, which related in an inconsistent way to changes in foraging preferences. Key words. Ageing, Apis mellifera, foraging preference, inflorescence, Lavandula stoechas, wing damage. Introduction Optimal foraging theory predicts that organisms forage in such a way as to maximise net reproductive benefit subject to constraints, where constraints may be environmental, such as predation risk, or constitutional, such as mechanical constraints and limited memory capacity. In haplodiploid social insects, like honey bees (Apis mellifera), reproduction is vested in the colony queen, and foraging decisions by the worker caste are shaped by the reproductive interests of the queen rather than the individual forager (e.g. SchmidHempel & Wolf, 1988; Seeley, 1989; Camazine, 1993), prompting interpretations of the colony as a functional superorganism (e.g. Seeley, 1989; Moritz & Southwick, 1992). Whether or not foraging decisions are best viewed in terms of individual foragers or a superorganismal colony, however, foraging workers are subject to constraints in their decision making like any other predator.

Correspondence: A. D. Higginson, Animal Behaviour and Ecology Research Group, School of Biology, University of Nottingham, University Park, Nottingham NG7 2RD, U.K. E-mail: [email protected]

52

One such constraint in foraging-phase worker honeybees appears to be deterioration of the flight mechanism, which ultimately limits foraging lifespan. This has two components. The first is physiological deterioration of the flight muscle, as a result of which honeybees often return to the hive with a partially filled crop (Schmid-Hempel et al., 1985). By doing this, they fail to maximise net energy delivered per unit time as expected under classical optimal foraging theory, but instead they appear to maximise net gain taking into account the trade-off between foraging and mortality (Houston et al., 1988). Foragers survive for an average of 7 days, and usually their maximum foraging lifespan is not more than 2 weeks; however, high foraging rates appear to lead to early death (Neukirch, 1982; SchmidHempel & Wolf, 1988). In contrast, wintering workers that rarely leave the hive can survive an entire winter (Page & Peng, 2001). That mortality among foragers may be due, at least in part, to physiological deterioration is suggested by Tofilski (2000), who found that, just before death, the time taken to collect food from an artificial feeder increased, the number of visits to the feeder per foraging flight decreased, and time taken to fly to the hive increased dramatically. Neukirch (1982) also found that the amount of time spent foraging per day affected total foraging

#

2004 The Royal Entomological Society

Wing damage and foraging in bees lifespan, suggesting that the total lifetime flight effort is limited by energy supply because the enzyme mechanism of carbohydrate metabolism is exhausted after a certain flight performance and ageing bees are less able to synthesise glycogen to fuel the flight muscles. Worker lifespan in Neukirch’s study depended almost entirely on the length of the hive phase, with the length of the foraging phase being relatively constant. A second way in which the flight mechanism deteriorates is through mechanical wear and tear of the wings (Alford, 1975; Rodd et al., 1980). Foragers lose wing area as they age, especially when foraging (Rodd et al., 1980). Intuitively, this should render bees less manoeuvrable and increase the energetic cost of flight. In keeping with this, experimentally increasing wing wear in bumblebees (Bombus melanopygus) either caused them to stop foraging altogether or significantly reduced their life expectancy (although the cause of death in the latter case was not known) (Cartar, 1992). The rate of mortality in bumblebees increases with age, although this trend may merely represent the switch from in-nest activities, which are safe from predators and parasites, to the more risky foraging activities (Morse, 1986; Visscher & Dukas, 1997). Reduced wing area or increased wing asymmetry as a result of damage should also impair the ability to escape from predators (Rodd et al., 1980) as well as reducing flight efficiency. To see whether damage-related mortality was associated with increased energy costs, Hedenstro¨m et al. (2001) clipped the wings of bumblebees (Bombus terrestris). While they found an increase in wing-beat frequency and the coefficient of lift, there was no increase in metabolic flight costs, and hence it seemed there was no significant impact of wing wear on flight energy costs. However, Cartar (1992) clipped almost twice as much wing area as Hedenstro¨m et al. (2001) (18% against 10%), so it is possible the threshold for significant effects on energy use may be somewhere between the two percentages. Since there are no data for the effects of natural wing wear in foraging bees, the significance of these findings is as yet far from clear. In this paper, a field study of foraging behaviour in honeybees is reported in which the effects of both natural and manipulated wing wear on foraging decisions are examined, and the hypothesis that increasing wing wear reduces flight ability and foraging efficiency tested.

Materials and methods The study was carried out at the Quinta de Sa˜o Pedro research station (38
390 N, 9
110 W), Sobreda de Caparica, Portugal in March and April 2002. The Quinta is a suburban smallholding of approximately 3 ha comprising a mixed habitat of shrubland and woodland (see Duffield et al., 1993). Eight hives of honeybees are maintained in the grounds, whose workers forage for nectar and pollen (Herrera, 1990; Gonzalez et al., 1995) on a neighbouring (10–20 m away) stand of lavender (Lavandula stoechas). Previous work on this system has shown that nectar-foraging #

53

bees discriminate among inflorescences on the basis of morphological cues (mainly the number of open flowers and terminal bracts), probably maximising their energy return during foraging bouts (Duffield et al., 1993; Gonzalez et al., 1995). Choosing inflorescences, however, requires bees to inspect flower heads on the wing prior to alighting (Duffield et al., 1993; Gonzalez et al., 1995; see below). If flight ability is compromised through wing wear, choice might be expected to become less efficient. The prediction here therefore is that the difference in morphological characteristics between accepted and rejected inflorescences will decline as wing wear increases and bees become less discriminating.

Foraging observations Foraging honeybees were located opportunistically in the lavender stand and followed for a sequence of 10 decisions (see below) at 10 different inflorescences. Each sequence was recorded verbally in real time using a Sanyo TRC1148 cassette recorder and transcribed at the end of each day. The 10 inflorescences were labelled with a numbered piece of masking tape as the bee left each one. Several measures previously shown to reflect nectar availability (Duffield et al., 1993; Gonzalez et al., 1995) were taken from each inflorescence. These were: height of the inflorescence from ground, length of the inflorescence, length of the longest and shortest terminal bracts, the number of bracts, and the number of open flowers. Flowers were counted in each vertical quartile of the inflorescence in order to calculate the total number and standard deviation in the number of open flowers. Following Duffield et al. (1993) and Gonzalez et al. (1995), the inflorescences were divided into three categories according to the bees’ response to them: Accepted inflorescences were those on which a bee alighted and probed one or more flowers. Rejected inflorescences were those that appeared to be inspected by a bee (and sometimes touched by the antennae or legs) in a brief hovering flight but on which the bee did not alight. Ignored inflorescences were those that bees approached but at which they did not visibly pause or make any physical contact. It was assumed that bees detected but chose to ignore these inflorescences. The time in flight between Accepted inflorescences, and the time spent on each inflorescence were calculated from the cassette recordings of each sequence.

Capturing and marking bees Focal bees were caught opportunistically while foraging on inflorescences of lavender by trapping them in a clear plastic vial. They were then chilled in a domestic freezer for 5–6 min, depending on the ambient temperature, until they stopped moving. Once sedated, bees were marked using coloured numbered disks [Opalithpla¨tchen; E. H. Thorne

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

54 A. D. Higginson and C. J. Barnard (Beehives) Ltd, U.K.] attached by Superglue, and their body length measured along a dorsal midline with a ruler from the front of the eyes to the tip of the abdomen. The amount of damage on each forewing was then quantified using a Wild M5 binocular microscope at 16 magnification to map missing tissue onto a 1:24 scale gridded template of the wing (Michener, 2000), and the area of damage (number of 0.4 mm2 squares) calculated (Fig. 1). The measure was thus standardised with respect to wing size. Damage appeared to be confined to the forewings, with none being noticed on the hindwings of any of the bees examined. In addition, since it is known to influence flight ability in other insects (Srygley & Ellington, 1999), asymmetry in the forewings was measured as the left-minus-right difference in the length of one of the subcostal cells (Fig. 1). After being marked and/or measured, bees were allowed to recover fully in a clear plastic vial before being released at the site of capture. When marked bees were subsequently observed foraging, a variable number of further sequences of 10 decisions were recorded. Bees were then recaptured and their wing damage measured once again, along with wing cell asymmetry (to provide an estimate of repeatability) and body length. Marked bees were recaptured between 0 and 5 times.

Experimental manipulation To test effects of damage experimentally, a second cohort of bees was marked and observed as previously, and the forewings of randomly selected individuals for which foraging sequences had previously been recorded were clipped using fine dissection scissors. Damage was added to bees that had already accumulated a degree of damage (mean ¼ 0.94% of wing area, range ¼ 0–8.31%). Approximately equal areas were clipped from each wing. The amount of damage of each wing after manipulation was mostly within the range of naturally accumulated damage (mean added damage ¼ 10.24% of wing area, range ¼ 3.48–23.97%; range of naturally accumulated damage ¼ 0–20.57%).

Fig. 1. A map of the forewing of a honey bee showing the cell measured for asymmetry (dashed line) and peripheral damage (hatched area) recorded from a focal bee. See text.

#

Statistical analysis The data were analysed using GLIM (Royal Statistical Society, London), SPSS (SPSS Inc., Chicago, Illinois) and R (Free Software Foundation, Inc., Vienna, Austria). All tests controlled for the within-subject nature of the dataset by incorporating bee identity as a random factor. Parametric tests were used throughout, with normalising transformations as necessary.

Results Natural wing damage accumulation Data from bees first captured when they had zero wing damage were used to analyse the natural rate of wing damage accumulation, plotted as the percentage damage with time since first capture in Fig. 2. Inspection of the figure suggested some form of second-order increase over time, so polynomial regression was used to test the relative goodness of fit of different plausible relationships (linear, quadratic, and exponential). Analysis showed that all three yielded significant positive relationships (linear: F1,103 ¼ 48.43, P < 0.001; quadratic: F2,102 ¼ 36.91, P < 0.001; exponential: F1,103 ¼ 82.42, P < 0.001) but that the exponential model provided the best predictor of total damage [additional variance explained ¼ 22.3% (vs. linear) and 7.6% (vs. quadratic)]. There was no discernible effect of capturing the bees on wing damage accumulation (t286 ¼ 1.615, NS).

Effects of wing asymmetry Left–right asymmetry in the length of the wing cell was normally distributed (Kolmogorov–Smirnov one sample Z ¼ 1.229, n ¼ 197, NS) but with a significant left bias (mean  SE left–right asymmetry ¼ 0.051  0.003 mm, t196 ¼ 14.405, P < 0.001). Wings therefore appeared to show directional asymmetry (Palmer & Strobeck, 1986, 2003; Møller & Swaddle, 1997). Repeated measures analysis of variance (ANOVA) over successive captures within bees caught at least three times tested for the repeatability of asymmetry measures and revealed no significant difference across captures (F1,21 ¼ 2.84, NS). Individuals that were first captured with zero damage and that had been recaptured within four days were used to analyse the effect of wing asymmetry on damage accumulation rate. Polynomial regression showed that there was no linear effect of absolute wing asymmetry on damage accumulation rate (F1,107 ¼ 0.27, NS), but the quadratic was significant (F2,106 ¼ 4.39, P < 0.05) and had a minimum at 0.057 mm, close to the population mean (0.051 mm). Subsequent analysis therefore used the deviation from the population mean rather than the absolute asymmetry in taking the effects of asymmetry into account. There was no significant effect of wing cell asymmetry on left–right asymmetry in damage accumulation.

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

Wing damage and foraging in bees

55

6

% Wing loss

5 4 3 2 1 0

0

50

100 150 200 Time since first capture (h)

250

300

Fig. 2. The accumulation of wing damage (% wing area) over time in bees with initial damage scores of zero. The curve is a best fit exponential; equation: y ¼ 1.0768e(0.0168x).

Choice of inflorescence Since the morphological measures taken from inflorescences were variously intercorrelated, principal components analysis was used to derive independent axes of variation among inflorescences. The unmanipulated and experimentally damaged cohorts were analysed separately. Analysis of the unmanipulated cohort yielded five components (Table 1), of which the first three accounted for 78% of the variance. The first component (explaining 47% of the variance) mainly reflected the length of the inflorescence and the number and length of the bracts, the second (explaining 21%) mainly the height of the inflorescence above the ground, and the third (explaining 13%) contrasted the number of open flowers (negative) and the number and size of bracts (positive). Components 1 and 3 thus best accorded with previous work (Duffield et al., 1993; Gonzalez et al., 1995) which showed that number of open flowers, the size of the inflorescence, and the number and size of bracts were the most important factors influencing nectar reward and choice of inflorescence (bees preferred inflorescences with more flowers and fewer, smaller bracts). Multifactor ANOVA of each of the five components by decision category (Accept, Reject, Ignore) for undamaged bees (controlling for bee as an additional factor) revealed a significant difference between decision categories

Table 1. Component weightings from principal components analysis of inflorescence morphology for bees in the experimentally unmanipulated cohort. See text. Standardised component % variance accounted for:

147.32

220.89

313.44

Component weightings Height Length of inflorescence Length of longest bract Number of bracts Number of open flowers

0.030 0.339 0.347 0.313 0.298

0.932 0.190 0.159 0.157 0.041

0.200 0.028 0.208 0.694 0.961

#

only in components 1 (F2,497 ¼ 30.44, P < 0.001) and 3 (F2,497 ¼ 7.13, P < 0.01), showing that Accepted heads were significantly larger (Fig. 3a), had significantly more flowers, and fewer, shorter bracts than Rejected or Ignored heads (Fig. 3b). Undamaged bees thus responded as found previously by Duffield et al. (1993) and Gonzalez et al. (1995). The differences remained significant when all bees, damaged and undamaged, were included in the analysis (component 1: F2,1766 ¼ 95.33, P < 0.0001; component 3: F2,1766 ¼ 26.15, P < 0.0001). Subsequent analyses were therefore confined to these two components. Analyses of morphological measures (body length, asymmetry deviation, absolute wing cell length) and a relative measure of age (time since first capture) in bees with zero damage revealed no significant effect of morphology on either component 1 (F4,496 ¼ 1.84, NS) or 3 (F4,496 ¼ 0.09, NS). Neither (averaging data within bouts) was there any effect on time spent on Accepted inflorescences (t188, NS in all cases); however, body length and asymmetry deviation both showed significant positive relationships with the average time flying between Accepted inflorescences (t188 ¼ 2.31, P ¼ 0.022 and t188 ¼ 2.20, P ¼ 0.029 respectively).

Effects of wing damage on foraging behaviour Wing damage during bouts between any two successive captures was estimated from the measurements of damage at each capture by assuming an exponential increase between the two. The effects of accumulating damage and time since first capture (the relative measure of age used here) on components 1 and 3, nested within bees, was analysed separately for different decisions (Accept, Reject, Ignore). Nested analysis of covariance for Accepted inflorescences revealed significant effects of total wing damage and time since first capture on both components (for component 1: F36,1163 ¼ 1.43, P ¼ 0.050; component 3: F36,1163 ¼ 1.98, P ¼ 0.001; time since first capture for component 1: F48,1163 ¼ 1.62, P ¼ 0.006; component 3: F48,1163 ¼ 1.70, P ¼ 0.002). The modal relationships were negative for component 1 and positive for component 3 in the case of wing damage, but negative for both components in the case

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

56 A. D. Higginson and C. J. Barnard

Fig. 3. Means  standard errors from analysis of variance of the relationship between foraging decision (Accept, Reject, or Ignore an inflorescence) by experimentally unmanipulated bees and (a) component 1 (loading positively for inflorescence size) and (b) component 3 [contrasting numbers of flowers (loading negatively) with number and size of terminal bracts (loading positively)] from principal components analysis of inflorescence morphology. See text and Table 1.

of relative age (Table 3, part a). Thus, as wing damage increased, bees on average accepted smaller inflorescences with fewer flowers and larger bracts. This resulted in damaged bees as a group accepting lower quality inflorescences than undamaged bees (ANOVA of component 3 nested within bee: F67,1187 ¼ 1.51, P ¼ 0.006; Fig. 4a); however, bees accepted smaller but better quality inflorescences as they aged (Table 3, part a). No significant effects of wing damage emerged for Rejected inflorescences, but time since first capture showed a significant positive association with component 3 [F47,375 ¼ 1.48, P ¼ 0.027; modal within-bee slope ¼ 0.001 (min., 0.227, max., 0.327)]. #

Fig. 4. Means  standard errors from analysis of variance of the difference in the quality of Accepted inflorescences between damaged and undamaged bees (a) for component 3 (Fig. 3b) in the experimentally unmanipulated cohort and (b) for component 2 in the manipulated cohort. See text.

There was no significant relationship between component 3 and either wing damage or time since first capture for Ignored inflorescences, but both were associated positively with component 1 [wing damage: F27,127 ¼ 1.61, P ¼ 0.043; modal slope ¼ 0.003 (min., 0.0737, max., 1.335), time since first capture: F38,127 ¼ 1.73, P ¼ 0.013; modal slope ¼ 0.006 (min., 0.321, max., 0.500)]; that is, the average size of inflorescence ignored by bees increased with damage and age. Analysis of the average time spent by bees on Accepted inflorescences showed no significant tendency for bees to spend longer on inflorescences or flying between inflorescences as wing damage increased, but time spent flying between inflorescences increased significantly with age (F48,95 ¼ 1.53, P ¼ 0.040).

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

Wing damage and foraging in bees

57

Effects of manipulating damage Principal components analysis of flower morphology for inflorescences encountered by bees in the second, experimental, cohort once again yielded a component (component 1, accounting for 36.9% of the variance) with a positive weighting for measures of inflorescence size and bract length (Table 2), and another (component 2, accounting for 17.3% of the variance) contrasting the number and size of bracts (this time negatively; cf. component 3 for the first cohort; Table 2) and height above the ground and number of flowers (positively). Inflorescences encountered by bees in the second cohort were slightly later in the season and generally smaller with fewer open flowers than those encountered by the first cohort (t4258 comparing between cohorts for inflorescence length ¼ 15.06, P < 0.001, size of bracts ¼ 22.36, P < 0.001, and number of flowers ¼ 19.44, P < 0.001). ANOVA of the two components by decision category and whether or not damage was added to the wings (again controlling for bee as an additional factor) revealed the same difference between Accepted, Rejected, and Ignored inflorescences in components 1 (inflorescence size: F2,2029 ¼ 5.22, P ¼ 0.003) and 2 (the contrast between bracts and number of flowers: F2,2029 ¼ 34.86, P < 0.0001) as in the previous cohort (Fig. 5), except, of course, that the sign of the relationship between component 2 and bract number/ size and number of flowers was the opposite of that of component 3 in Fig. 3. There was no significant effect of adding damage per se on component 1 (F1,2029 ¼ 2.70, NS), and component 2 showed the expected decline in bees with added damage only at the 10% level (mean  SE in bees without added damage ¼ 0.04  0.03, with added damage ¼ 1.10  0.04; F1,2029 ¼ 2.81, P ¼ 0.094). Once again, nested analysis of covariance was used to analyse the effect of the amount of natural and experimentally added wing damage on components 1 and 2 for the different decisions. Both naturally accumulated damage and the amount of damage added experimentally were entered into the analyses. As in the first cohort, time since first capture (relative age) was also included. For Accepted inflorescences, both time since first capture and damage added experimentally showed significant negative effects on component 1 (time since first capture: Table 2. Component weightings from principal components analysis of inflorescence morphology for bees in the experimentally manipulated cohort. See text. Standardised component % variance accounted for:

136.84

217.34

Component weightings Height Length of inflorescence Length of longest bract Number of bracts Number of open flowers

0.138 0.317 0.400 0.262 0.098

0.664 0.009 0.202 0.067 0.660

#

Fig. 5. Means  standard errors from analysis of variance of the relationship between foraging decision (Accept, Reject, or Ignore an inflorescence) by experimentally manipulated bees and component 2 [contrasting number of flowers (loading positively) and the number and size of bracts (loading negatively)] from principal components analysis of inflorescence morphology. See text and Table 2.

F59,1345 ¼ 2.02, P < 0.0001; added damage: F18,1345 ¼ 2.91, P ¼ 0.001; Table 3, part b), but natural damage did not (F42,1345 ¼ 1.28, NS). All three variables, however, had a significant negative effect on component 2 (time since first capture: F59,1345 ¼ 2.69, P < 0.0001; added damage: F59,1345 ¼ 2.15, P ¼ 0.003; natural damage: F45,1345 ¼ 2.41, P < 0.0001; Table 3, part b). As in the first cohort, this resulted in damaged bees (including natural and added damage) accepting lower quality inflorescences than undamaged bees (nested ANOVA: F94,1404 ¼ 4.64, P < 0.0001; Fig. 4b). Added and natural damage also showed significant positive relationships with component 1 for Rejected inflorescences [added damage: F17,535 ¼ 2.21, P < 0.004; modal slope ¼ 0.0011 (min., 1.47, max., 9.062); natural damage: F41,535 ¼ 1.78, P < 0.003; modal slope ¼ 0.0013 (min., 0.0188, max., 0.0055)], but negative relationships with component 2 [added damage: F17,535 ¼ 4.11, P < 0.0001; modal slope ¼ 0.0005 (min., 0.016, max., 0.012); natural damage: F41,535 ¼ 1.78, P < 0.003; modal slope ¼ 0.002 (min., 2.84, max., 179.5)]. There was a significant negative effect of time since first capture on component 2 for Rejected inflorescences [F58,535 ¼ 2.73, P < 0.0001; modal slope ¼ 0.012 (min., 19.61, max., 0.911)]. There were no significant effects of any of the three variables on components 1 or 2 for Ignored inflorescences.

Discussion The results suggest that accumulating wing damage is an accelerating cost of foraging that has an important effect on the foraging behaviour of honeybees independently of the

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

58 A. D. Higginson and C. J. Barnard Table 3. Mode and range of within-bee slopes from nested analysis of covariance of the relationship between the morphology of Accepted inflorescences [principal components (see text)], age (time since first capture) and the amount of wing damage sustained by bees. (a) Unmanipulated cohort

Component 1(inflorescence size)

Component 3 [open flowers (–) and size/number of bracts (þ)]

Estimated natural damage Age

0.012 (min., 0.576; max., 0.910) 0.002 (min., 0.414; max., 0.107)

0.0053 (min., 0.970; max., 1.324) 0.003 (min., 0.427; max., 0.199)

(b) Manipulated cohort

Component 1(inflorescence size)

Component 2 [open flowers (þ) and size/number of bracts (–)]

Estimated natural damage Added damage Age

NS 0.001 (min., 0.051; max., 0.006) 0.0059 (min., 2.108; max., 8.59)

0.104 (min., 0.251 max., 162.47) 0.0001 (min., 0.037; max., 0.008) 0.0012 (min., 17.76; max., 1.609)

NS, relationship not significant.

physiological ageing process. Analysis of damage accumulation in known individuals showed an exponential increase over time, suggesting a positive feedback effect of already acquired damage. The fact that damage also increased with the degree of asymmetry in wing size might suggest that accelerating damage relates to changes in the aerodynamic efficiency of the wings and the need for greater adjustment in their beat frequency and orientation to achieve manoeuvrability. Surprisingly in this respect, however, there was no effect of damage on the amount of time flying between Accepted inflorescences, though older bees appeared to spend longer in flight. Interestingly, natural damage accumulation increased with the degree of departure of wing size asymmetry from a left-biased population mean (rather than with the absolute degree of asymmetry). This suggests the small degree of left bias may have some functional significance, though what this may be is as yet unclear. A similar left bias has been found in a captive population of honeybees in the U.K. (A. D. Higginson, in prep.) and is known in other insects. In male speckled wood butterflies (Pararge aegeria), for example, a left wing bias appears to facilitate faster turning in the aerial territorial disputes (Windig & Nylin, 1999). It is known that honeybees and bumble bees show a degree of handedness in their spatial movements (Fergusson-Kolmes et al., 1992; Kells & Goulson, 2001), but relationships between this and morphological asymmetry remain to be tested. Perhaps surprisingly, wing size asymmetry had no effect on asymmetry in damage accumulation. Undamaged bees based their choice of inflorescence on the same contrast of floral characters (the size of the inflorescence, the number of open flowers, and the number and size of terminal bracts) as found to predict nectar reward in previous and ongoing studies of the lavender system (Duffield et al., 1993; Gonzalez et al., 1995; A. D. Higginson, in prep.; see also Herrera, 1990, 1991). As their wing damage increased, however, bees accepted smaller inflorescences and inflorescences with fewer open flowers, and larger and more numerous bracts, implying a reduction in choosiness as damage accumulated. Smallness, fewer flowers and more bract tissue are characteristics of ageing inflorescences with reduced nectar production (Duffield #

et al., 1993; A. D. Higginson, in prep.), so being less choosy implies reduced nectar foraging efficiency. Adding damage experimentally had a significant impact on foraging decisions independently of, but in the same direction as, naturally acquired damage, supporting the idea of a direct role of wing damage in the behaviour of bees. Damage varied in its relationship with the characteristics of Rejected inflorescences, showing no significant association in the first, unmanipulated, cohort, but a negative association with quality measures (open flowers and bract size/number) in the second cohort. Along with Ignored inflorescences, however, there was a significant positive relationship between damage and inflorescence size among rejected heads. These last relationships with inflorescence size are puzzling, but may partly reflect the difficulty of defining ignored heads unequivocally and changes in the flight behaviour and manoeuvrability in older, more damaged bees. The difference between cohorts in this respect may also have been a consequence of the changing size distribution of inflorescences during the study period. These possibilities have been the subject of further study currently in analysis (A. D. Higginson, in prep.). The inconsistent effects of bee age on the size and quality components of the morphology of Accepted inflorescences, with both positive and negative associations in different samples, are also difficult to explain. However, it is important to stress that the measure of age was relative to time of first capture, so it is likely that true age varied considerably with the relative measure and may even have differed between the two cohorts studied. Controlled studies of the effects of age are currently being undertaken in laboratorybased experiments with observation hives. Despite these outcomes, however, the results nevertheless suggest that age and accumulating wing damage have independent effects on the foraging behaviour of worker honey bees.

Acknowledgements We thank Francis Gilbert, Francis Ratnieks, and two anonymous referees for helpful comments on the manuscript, Francis Gilbert for statistical advice, Martin Hoyle

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59

Wing damage and foraging in bees and Mick Crawley for further helpful discussion, and Armin Pircher of the Quinta de Sa˜o Pedro field station for his support and hospitality in the field. The work was supported by a BBSRC Committee Studentship to ADH.

References Alford, D.V. (1975) Bumblebees. Davis-Poynter, London. Camazine, S. (1993) The regulation of pollen foraging by honey bees: how foragers assess the colony need for pollen. Behavioral Ecology and Sociobiology, 32, 265–272. Cartar, R.V. (1992) Morphological senescence and longevity: an experiment relating wing wear and life span in foraging wild bumble bees. Journal of Animal Ecology, 61, 225–231. Duffield, G.E., Gibson, R.C., Gilhooly, P.M., Hesse, A.J., Inkley, C.R., Gilbert, F.S. et al. (1993) Choice of flowers by foraging honey bees (Apis mellifera): possible morphological cues. Ecological Entomology, 18, 191–197. Fergusson-Kolmes, L., Kolmes, S.A. & Winston, M.L. (1992) Handedness and asymmetry in the waggle dance of worker honey bees (Hymenoptera, Apidae). Journal of the Kansas Entomological Society, 65, 85–86. Gonzalez, A., Rowe, C.L., Weeks, P.J., Whittle, D., Gilbert, F.S. & Barnard, C.J. (1995) Flower choice by honey bees (Apis mellifera L.): sex-phase of flowers and preferences among nectar and pollen foragers. Oecologia, 101, 258–264. Hedenstro¨m, A., Ellington, C.P. & Wolf, T.J. (2001) Wing wear, aerodynamics, and flight energetics in bumblebees: an experimental study. Functional Ecology, 15, 417–422. Herrera, C.M. (1990) Daily patterns of pollinator activity, differential pollinating effectiveness, and floral resource availability, in a summer-flowering Mediterranean shrub. Oikos, 58, 277–288. Herrera, J. (1991) Allocation of reproductive resources within and among inflorescences of Lavandula stoechas (Lamiaceae). American Journal of Botany, 78, 789–794. Houston, A., Schmid-Hempel, P. & Kacelnik, A. (1988) Foraging strategy, worker mortality, and the growth of the colony in social insects. American Naturalist, 131, 107–114. Kells, A.R. & Goulson, D. (2001) Evidence for handedness in bumblebees. Journal of Insect Behaviour, 14, 47–55. Michener, C.D. (2000) The Bees of the World. The Johns Hopkins University Press, Chicago.

#

59

Møller, A.P. & Swaddle, J.P. (1997) Asymmetry, Developmental Stability, and Evolution. Oxford University Press, Oxford. Moritz, R.F.A. & Southwick, E.E. (1992) Bees as Superorganisms: an Evolutionary Reality. Springer-Verlag, Berlin. Morse, D.H. (1986) Predatory risk to insects foraging at flowers. Oikos, 46, 223–228. Neukirch, A. (1982) Dependence of the life span of the honeybee upon flight performance and energy consumption. Journal of Comparative Physiology, 146, 35–40. Page, R.E. Jr & Peng, C.Y.-S. (2001) Aging and development in social insects with emphasis on the honey bee, Apis mellifera L. Experimental Gerontology, 36, 695–711. Palmer, A.R. & Strobeck, C. (1986) Fluctuating asymmetry – measurement, analysis, patterns. Annual Review of Ecology and Systematics, 17, 391–421. Palmer, A.R. & Strobeck, C. (2003) Fluctuating asymmetry analyses revisited. Developmental Instability: Causes and Consequences (ed. by M. Polak), pp. 279–319. Oxford University Press, Oxford. Rodd, F.H., Plowright, R.C. & Owen, R.E. (1980) Mortality rates of adult bumble bee workers (Hymenoptera: Apidae). Canadian Journal of Zoology, 58, 1718–1721. Schmid-Hempel, P., Kacelnik, A. & Houston, A.I. (1985) Honeybees maximise efficiency by not filling their crop. Behavioral Ecology and Sociobiology, 17, 61–66. Schmid-Hempel, P. & Wolf, T.J. (1988) Foraging effort and lifespan of workers in a social insect. Journal of Animal Ecology, 57, 509–521. Seeley, T.D. (1989) The honey bee as a superorganism. American Scientist, 77, 546–553. Srygley, R.B. & Ellington, C.P. (1999) Estimating the relative fitness of local adaptive peaks: the aerodynamic costs of flight in mimetic passion-vine butterflies Heliconius. Proceedings of the Royal Society of London Series B, 266, 2239–2245. Tofilski, A. (2000) Senescence and learning in honeybee workers. Acta Neurobiologiae Experimentalis, 60, 35–39. Visscher, P.K. & Dukas, R. (1997) Survivorship of foraging honey bees. Insectes Sociaux, 44, 1–5. Windig, J.J. & Nylin, S. (1999) Adaptive wing asymmetry in males of the speckled wood butterfly (Pararge aegeria)? Proceedings of the Royal Society of London Series B, 266, 1413–1418.

Accepted 15 October 2003

2004 The Royal Entomological Society, Ecological Entomology, 29, 52–59