SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Flowers exhibit symmetrical patterns, and innate preferences for symmetry in pollinators like honeybees are documented. Most previous studies of symmetry preferences in honeybees, Apis mellifera, tested levels of asymmetry using artificial flowers or stimuli. Here we investigated the effect of flower asymmetry on flower preferences of honeybees in a novel approach using real flowers, incorporating their spectral properties and how the receivers process the visual signals. Importantly, we also tested the response of an ‘eavesdropping’ predator, the crab spider Thomisus spectabilis, that also utilizes the same flower to prey on honeybees. Flowers (Chrysanthemum frutescens) were manipulated to contain asymmetrical and symmetrical patterns, excluding olfactory cues. Both crab spiders and honeybees exhibited a significant preference for symmetrical flowers. Moreover, honeybees exhibited a significant preference for radial symmetry over bilateral symmetry, but no corresponding effect was recorded in crab spiders. Further analyses demonstrated that flower reflectance and orientation of the axis of symmetry did not affect crab spider decisions. Field observations on T. spectabilis revealed that the natural variation in C. frutescens symmetry had no effect on the choice of crab spiders. This indicates that spiders and honeybees may use other flower characteristics, for example, olfactory cues, together with flower symmetry, to make their foraging decisions.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Symmetry, and its perception, has been studied in a variety of insects (e.g. earwigs, Forficula auricularia, Tomkins & Simmons 1998), including a range of pollinators (Møller & Eriksson 1995). Three main forms of symmetry have been identified: monosymmetry (one axis of symmetry), polysymmetry (more than one axis of symmetry), and asymmetry (no axis of symmetry; Endress 2001). Recently, flower symmetry has been proposed as a response by plants to a perceptual bias for symmetry in pollinators (Lehrer et al. 1995; Enquist & Johnstone 1997; Neal et al. 1998).

In nature, asymmetry may be caused by a range of interacting environmental (e.g. nutritional deficiencies) and genetic (e.g. inbreeding) factors (Møller & Swaddle 1997). Fluctuating asymmetry describes small variations from perfect, often bilateral, symmetry (Van Valen 1962; Møller & Swaddle 1997) caused by developmental instability. Thus the level of symmetry may be a reliable cue of flower quality, and flowers that exhibit a high level of symmetry may offer higher rewards to pollinating insects compared with less symmetrical flowers within a flower species (e.g. Møller & Sorci 1998; Møller 2000). However, even if fluctuating asymmetry in flowers has the potential to influence pollinator choice, there is little evidence to date that insects possess the visual resolution to detect these fine-scale differences. However, floral senescence or herbivory has the potential to create flowers that differ greatly in shape, colour and size. Bees are known to be influenced in flower choice by colour (e.g. Chittka & Menzel 1992; Lunau 2000), odour (e.g. Galen & Kevan 1983; Wells & Wells 1985) and size (e.g. Ohara & Higashi 1994). Importantly, pollinators of Lantana camara discriminate between the different colour phases associated with flower age and reward quality (Weiss 1991).

Additionally, signals must not be considered only in terms of communication between signaller and receiver (in our case, flower and honeybee). Many signalling systems are exploited by eavesdropping predators which should also be taken into account. Honeybee and flower communication is exploited by the crab spider predators (Heiling & Herberstein 2004; Heiling et al. 2004), which we also test for symmetry preferences.

Crab spiders (Thomisidae) ambush pollinating insects by sitting and waiting on flowers (Foelix 1996; Schmalhofer 1999). Studies have shown that crab spiders respond to flower odours (Aldrich & Barros 1995; Krell & Krämer 1998) as well as visual and tactile cues (Morse 1988; Greco & Kevan 1994) when selecting hunting sites. Experiments on the crab spider Misumena vatia revealed that prey abundance and flower quality (Morse 1988), in addition to previous experience (Morse 1999, 2000), affected flower choice. The Australian crab spider, Thomisus spectabilis, increases its foraging success by responding to visual and olfactory signals generated by flowers to attract honeybees (Apis mellifera, Apidae; Heiling & Herberstein 2004; Heiling et al. 2004). However, they may also respond to flower symmetry if this is a reliable predictor of pollinator visitation rates.

Here, we use a series of choice experiments to investigate the preferences of honeybees for different types of symmetry and compare these with the preferences of their crab spider predators. The majority of experiments testing the symmetry preferences of bees have been performed using artificially generated stimuli, such as black shapes presented on a bee-white background (Giurfa et al. 1996). Møller & Sorci (1998) presented paper models of flowers consisting of a circular shape with a black dot in the centre, and Lehrer et al. (1995) used white and grey discs and checkerboard patterns to demonstrate that honeybees display an innate preference for flower-like patterns. Unlike these previous studies, we used real flowers for experimental manipulations as we anticipated that honeybees and crab spiders use a multitude of cues when selecting flowers.

Functionally, we predicted that honeybees would display a significant preference for symmetrical flowers over asymmetrical, as the former indicate flowers unaffected by age or damage. As crab spiders depend on the attractiveness of flowers to honeybees, we expected their flower choice to correspond to that of their prey.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Subjects

Honeybees, A. mellifera, were maintained in an outdoor hive on Macquarie University grounds. They were trained to visit a feeder (sucrose solution). There are no Chrysanthemum frutescens on the campus ground or in a nearby national park, where many bees from the hive forage. It is thus unlikely that subject bees had any experience with such flowers prior to testing. Crab spiders, T. spectabilis (Thomisidae), were collected from the field (Brisbane, Australia) in 2001 and therefore have experienced flower selection prior to our experiments. The native T. spectabilis has been observed in the field ambushing honeybees on C. frutescens, the daisy used in this study. Spiders were housed indoors at Macquarie University on a 12:12 h fluorescent light cycle, given water daily, and fed once a week with live crickets or flies. Field observations of crab spiders were performed in suburban gardens in Brisbane, Australia (October 2002).

Stimuli

Each C. frutescens (white flower variety ‘Summer Angel’) flower pair was matched for size and stage of development. Petals were removed by cutting 5 mm from the reproductive flower centre to avoid ‘edge effects’, which might otherwise emerge when unequal ratios of white petals to yellow pollen to black background were presented (Lehrer et al. 1990; Craig 1994). Flowers were placed against a round black background (5 cm diameter) to provide a constant background. The flower and background were covered with Glad WrapTM (The Clorox Company, Oaklands, CA, USA) to remove olfactory cues. Glad WrapTM is a thin clear sheet of polyethylene, which is permeable to all wavelengths of light above 300 nm, with <5% attenuation.

Procedures

Experimental procedures were identical for each choice experiment, with all experiments conducted outdoors. Naturally foraging honeybees were trained to visit the experimental set up by being provided with a feeding station. The feeding station provided sucrose solution (up to 25%), which encouraged honeybees to visit it. This ensured that we had enough naturally foraging animals for the experiments. We replaced the feeding station with the experimental arrangement to observe flower choice in honeybees. Honeybees were presented with a flower pair positioned horizontally on a rectangle (18 cm × 13 cm) of black cardboard. The horizontal positioning was chosen as it corresponds to the way honeybees encounter flowers naturally. Typically, honeybees circled the experimental set up before attempting to land on one of the flowers. The first contact with the Glad WrapTM covering the flowers was considered the choice of a flower. We allowed up to 10 min for the honeybees to make a choice. The honeybees were removed from the population after testing to avoid pseudoreplication.

Crab spiders were anaesthetized with carbon dioxide and placed in the centre of a circle of black cardboard (diameter 18 cm). Anaesthetizing the spiders enabled us to place them gently equidistant from the flowers and prevented an escape response. A flower pair was presented so that the flower heads were adjacent and in a vertical position exactly 4 cm from the spider. This arrangement ensured that spiders were able to perceive both flowers. Spiders were allowed 30 min to select a flower. A choice was defined as the first contact with the Glad WrapTM covering the daisies. Each flower pair was tested first with a spider, then a honeybee. Olfactory cues are known to influence the choice behaviour of honeybees (e.g. Pelz et al. 1997; Laska et al. 1999). Thus, after having made a choice, the spider was placed on the rejected daisy for an equal amount of time to exclude any influence of olfactory cues generated by the spider before the flower pair was re-tested using honeybees.

In the first experiment, animals were tested for a preference between a flower with many axes of symmetry and another with no axes of symmetry (radial symmetry and asymmetry). Ten patterns of asymmetry were designed by randomly selecting half of the petals for removal (Fig. 1a). To control for petal number, every second petal of the radially symmetric flower was removed (Fig. 1b). Each pair of flowers was tested using one crab spider and one honeybee, allowing choices to be directly compared. Initially, each pattern of asymmetry was tested eight times (i.e. with eight honeybees and eight spiders), with the honeybees trained to visit a radially symmetric feeder. As honeybees repeatedly visit the feeder, they may learn to associate a feeder design with reward, which may bias our experiments. We eliminated this problem by training honeybees to an asymmetric feeder, the shape of which was altered each day of the experiment. We then tested flower preference on a further 40 honeybees.

image

Figure 1. Flower stimuli. (a) Patterns of asymmetry used in experiment 1; (b) radial symmetry; (ci) transversal bilateral symmetry, with axis shown in dotted line; (cii) dorsoventral bilateral symmetry

Download figure to PowerPoint

In expt 2, we compared choices between many axes of symmetry (radial symmetry; Fig. 1b) and one axis of symmetry (bilaterally symmetrical flowers; Fig. 1c). Forty crab spiders and 40 honeybees were tested with a horizontal axis of symmetry in the bilateral flower (transversal bilateral symmetry; Fig. 1ci). A further 40 crab spiders were then tested using a dorsoventral axis of symmetry (Fig. 1cii) in the bilateral flower to control for bias in the orientation of the axis of symmetry.

Flower diameter and diameter of the central pollen disc were measured for both the chosen and rejected daisy. The spectral reflectance of flowers affects the choice behaviour of honeybees (e.g. Chittka & Menzel 1992; Gumbert et al. 1999; Briscoe & Chittka 2001) and crab spiders (Heiling et al. 2005). Thus, the reflectance of each flower was also measured using a USB 2000 spectrometer with a PX-2-pulsed xenon light source attached to a PC running OODBase32 software, calculating per cent reflectance from 300 to 700 nm (Ocean Optics Inc., Dunedin, FL, USA). The reflectance of both the petals and central disc was measured.

Field observations of crab spiders were made on bushes of C. frutescens daisies. This allowed the direct comparison of crab spider choices in the field with those made in controlled experiments. The length of each petal from the centre of the flower to its tip was measured on each occupied and a randomly selected unoccupied flower on the same plant. Standard deviations of petal lengths were analysed as a measure of flower asymmetry. In addition, flower and central disc diameter were measured.

Analyses

All data were tested for normal distribution (Kolmogorov–Smirnov test) and homogeneity of variances before being subjected to parametric analyses. Binomial probabilities were used to assess the flower preferences of honeybees and crab spiders. Contour lines for each treatment were measured by scoring for each flower the number of petal edges not overlapping another. These scores were then correlated to the choices of honeybees and crab spiders.

The colour analyses of flower reflectance refer to the visual system of honeybees, as they incorporate the spectral sensitivity functions of animals, which are available only for honeybees (Peitsch et al. 1992) but not for crab spiders. Our calculations included the following steps (for methods see Chittka 1996). First, we computed the sensitivity factor R of each photoreceptor type (UV, blue and green). Calculating R incorporates the sensitivities of honeybee photoreceptors in the UV, the blue and the green across the range of 300–700 nm. As the relative sensitivity of bee receptors depends on the reflectance spectrum of the background (Briscoe & Chittka 2001), the illuminating daylight spectrum and the reflectance spectrum of the background were incorporated. Background colour is a standardized green leaf background, which honeybees perceive when crossing vegetation in the search for flowers (Briscoe & Chittka 2001). The subsequent calculations incorporating the R-values as well as the illuminating daylight spectrum and the spectral reflectance curve of the flowers output the relative amount of light (quantum catch p) absorbed by each type of photoreceptor. Based on these, we calculated the relative receptor excitation values (EUV, Eblue and Egreen), which are the physiological receptor voltage signals for each type of photoreceptor and provide the basis for computation of the colour coordinates ‘x’ and ‘y’. These coordinates indicate the position of colour loci in the hexagon colour space of honeybees, which illustrates how the colour of objects is perceived (Chittka et al. 1994). We used paired t-tests (Bonferroni correction of α = 0.0063) to compare receptor excitation values and colour coordinates between chosen and rejected flowers.

Flower and central disc diameters were compared between the chosen and rejected daisy using paired t-tests, with alpha levels adjusted using a Bonferroni correction (α = 0.0125). Relative receptor excitation values based on physiological sensitivities were calculated for the UV, blue and green receptors of the honeybee (for methods see Chittka 1996; Briscoe & Chittka 2001). The excitation values of chosen and rejected flowers were then compared using paired t-tests (Bonferroni correction of α = 0.0063). The position of chosen and rejected flowers within the honeybee visual colour hexagon (an illustration of the object in the honeybee colour space; see Chittka 1996) was then calculated and compared using paired t-tests (Bonferroni correction of α = 0.0063).

Floral asymmetry in daisies chosen in the field by crab spiders was calculated using the standard deviation of petal lengths from each flower and compared using a paired t-test (see Møller & Eriksson 1995 for methods). Flower diameter and central disc diameter between occupied and unoccupied flowers in the field were also compared using paired t-tests.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Honeybees trained to a symmetrical feeder and crab spiders both significantly preferred the symmetrical flower of a test pair to the asymmetrical (binomial tests: p = 0.01, n = 80 and p = 0.01, n = 77 respectively; Fig. 2). Crab spiders that did not select a flower were excluded from the analysis. Honeybees trained to an asymmetrical feeder also preferred symmetrical flowers (binomial test: p = 0.007, n = 39). Further analysis revealed that within the same pair of flowers, honeybees and crab spiders did not necessarily choose the same flower (irrespective of level of symmetry). In only 41 of the 77 trials did honeybees and spiders select the same flower (binomial test: p = 0.71, n = 77).

image

Figure 2. Frequency (%) of choices by crab spiders and honeybees between radially symmetric and asymmetric flowers. Honeybees were tested after being trained to symmetrical and asymmetrical feeders. Honeybees and crab spiders significantly preferred radially symmetric flowers (*p < 0.05)

Download figure to PowerPoint

There was no significant difference in flower or central disc diameter between the chosen and rejected flowers of honeybees trained to a symmetrical feeder or to an asymmetrical feeder (Tables 1 and 2). Similarly, there was no significant difference in the size of flowers that were chosen and rejected by crab spiders (Tables 1 and 2).

Table 1.  Flower size and petal reflectance
SubjectFlower characteristicChosen flowerRejected flowerStatistics
t(df)p
  1. Overall flower diameter (cm), receptor excitations of flower petals in the UV, the blue and the green, and X–Y coordinates in the colour hexagon for the petals of flowers chosen and rejected by honeybees and spiders. Receptor excitation values and coordinates in the bee colour space are dimensionless and do not carry a unit. All values presented are inline image ± SD.

Crab spidersFlower diameter2.62 ± 0.672.63 ± 0.67−0.33(70)0.74
Honeybees (symmetrical feeder)Flower diameter2.65 ± 0.702.66 ± 0.70−0.11(73)0.91
Honeybees asymmetrical feeder)Flower diameter4.27 ± 0.364.27 ± 0.390(38)1
Receptor (Euv)0.73 ± 0.020.73 ± 0.020.62(38)0.54
Excitation
 Eblue0.91 ± 0.000.91 ± 0.000.69(38)0.49
 Egreen0.88 ± 0.000.88 ± 0.000.96(38)0.34
(X)0.13 ± 0.020.13 ± 0.02−0.55(38)0.59
(Y)0.11 ± 0.010.11 ± 0.01−0.64(38)0.53
Table 2.  Pollen disc size and reflectance
SubjectFlower characteristicChosen flowerRejected flowerStatistics
t(df)p
  1. Central pollen disc diameter (cm), receptor excitations of pollen disc in the UV, the blue and the green, and X–Y coordinates in the colour hexagon for the pollen of flowers chosen and rejected by honeybees and spiders. Receptor excitation values and coordinates in the bee colour space are dimensionless and do not carry a unit. All values presented are inline image ± SD.

Crab spidersPollen diameter0.79 ± 0.240.78 ± 0.220.70(70)0.49
Honeybees (symmetrical feeder)Pollen diameter0.78 ± 0.230.79 ± 0.23−0.41(73)0.68
Honeybees (asymmetrical feeder)Pollen diameter1.24 ± 0.131.24 ± 0.120(38)1
Receptor (Euv)0.06 ± 0.090.04 ± 0.072.23(38)0.03
Excitation
 Eblue0.18 ± 0.140.14 ± 0.112.23(38)0.03
 Egreen0.80 ± 0.010.80 ± 0.010.36(38)0.72
(X)0.65 ± 0.080.66 ± 0.06−2.26(38)0.03
(Y)−0.25 ± 0.10−0.27 ± 0.091.96(38)0.06

Analysis of relative receptor excitation by flowers chosen and rejected by honeybees trained to an asymmetric flower revealed no significant difference in petal reflectance in the UV, blue or green parts of the spectrum (Table 1). Similarly, no difference was detected in the reflectance of the central disc between chosen and rejected flowers (Table 2). Analysis of the position of the chosen and rejected flowers within the honeybee colour hexagon revealed no significant difference in petals or pollen (Tables 1 and 2). Among the asymmetric flowers (Fig. 1a), the proportion chosen by honeybees or crab spiders did not correlate with flower contour length (honeybees: Pearson's r = 0.1864, p = 0.6062; Spearman's ρ = 0.1801, p = 0.6185; crab spiders: Pearson's r = 0.0252, p =0.9448; Spearman's ρ = 0.2287, p = 0.5250).

Honeybees significantly preferred to visit radially symmetric flowers to bilaterally symmetric flowers (binomial test: p = 0.01, n = 40; Fig. 3) while spiders did not (binomial test: p = 0.63, n = 35; Fig. 3). Again, there was little agreement in flower choice between honeybees and spiders. In only 43% of trials was the same flower chosen by both honeybees and spiders (binomial test: p = 0.21, n = 35). An additional choice experiment testing crab spiders with a dorsoventral axis of bilateral symmetry again revealed no preference (binomial test: p = 0.29, n = 30).

image

Figure 3. Frequency (%) of choices by crab spiders and honeybees between radially symmetric and bilaterally symmetric flowers. Crab spiders were tested using both transversal and dorsoventral bilateral symmetry. Honeybees significantly preferred radially symmetric flowers (*p < 0.05)

Download figure to PowerPoint

Honeybees demonstrated no significant preference for the larger flower of a test pair, irrespective of symmetry, for either flower diameter or the central disc diameter (Tables 3 and 4). However, analyses revealed a significant preference in crab spiders for the flower with the largest diameter when testing a translateral axis of bilateral symmetry (Table 3) but not the largest central disc diameter (Table 4). This preference for the larger flower in crab spiders was absent when testing a dorsoventral axis of bilateral symmetry (Table 3).

Table 3.  Flower size and petal reflectance
SubjectFlower characteristicChosen flowerRejected flowerStatistics
t(df)p
  1. Flower diameter (cm), average petal receptor excitations and X–Y coordinates in the colour hexagon comparing radial and bilateral symmetry. Receptor excitation values and coordinates in the bee colour space are dimensionless and do not carry a unit. All values presented are inline image ± SD.

Crab spiders (translateral–bilateral symmetry)Flower diameter3.97 ± 0.493.86 ± 0.452.63(34)0.01
Receptor (Euv)0.72 ± 0.010.72 ± 0.031.1(34)0.28
Excitation
 Eblue0.91 ± 0.000.91 ± 0.011.43(34)0.16
 Egreen0.88 ± 0.000.88 ± 0.001.63(34)0.11
(X)0.14 ± 0.010.14 ± 0.02−0.96(34)0.35
(Y)0.11 ± 0.000.11 ± 0.01−0.75(34)0.46
HoneybeesFlower diameter3.91 ± 0.503.89 ± 0.450.38(39)0.71
Receptor (Euv)0.72 ± 0.020.72 ± 0.020.79(39)0.43
Excitation
 Eblue0.91 ± 0.000.91 ± 0.010.87(39)0.39
 Egreen0.88 ± 0.000.88 ± 0.000.85(39)0.40
(X)0.14 ± 0.010.14 ± 0.02−0.75(39)0.46
(Y)0.11 ± 0.010.11 ± 0.01−0.63(39)0.53
Crab spiders (dorsoventral bilateral symmetry)Flower diameter3.32 ± 0.503.20 ± 0.491.54(29)0.14
Receptor (Euv)0.71 ± 0.020.71 ± 0.021.08(29)0.29
Excitation
 Eblue0.91 ± 0.000.91 ± 0.000.36(29)0.72
 Egreen0.88 ± 0.000.88 ± 0.000.15(29)0.88
(X)0.14 ± 0.020.14 ± 0.02−0.75(29)0.46
(Y)0.11 ± 0.010.11 ± 0.01−0.75(29)0.46
Table 4.  Pollen disc size and reflectance
SubjectFlower characteristicChosen flowerRejected flowerStatistics
t(df)p
  1. Pollen disc diameter (cm), average pollen disc receptor excitations and X–Y coordinates in the bee colour hexagon comparing radial and bilateral symmetry. Receptor excitation values and coordinates in the bee colour space are dimensionless and do not carry a unit. All values are inline image ± SD.

Crab spiders (translateral– bilateral symmetry)Pollen disc diameter1.28 ± 0.541.23 ± 0.461.82(34)0.08
Receptor (Euv)0.13 ± 0.160.10 ± 0.110.67(34)0.51
Excitation
 Eblue0.35 ± 0.200.33 ± 0.180.45(34)0.66
 Egreen0.82 ± 0.010.81 ± 0.011.21(34)0.24
(X)0.60 ± 0.130.61 ± 0.09−0.61(34)0.55
(Y)−0.13 ± 0.13−0.13 ± 0.130.23(34)0.82
HoneybeesPollen disc diameter1.18 ± 0.121.17 ± 0.150.54(38)0.59
Receptor (Euv)0.14 ± 0.150.09 ± 0.101.96(38)0.06
Excitation
 Eblue0.39 ± 0.200.30 ± 0.162.37(38)0.02
 Egreen0.82 ± 0.010.81 ± 0.012.65(38)0.01
(X)0.58 ± 0.120.63 ± 0.08−1.86(38)0.07
(Y)−0.09 ± 0.13−0.15 ± 0.122.29(38)0.03
Crab spiders (dorsoventral bilateral symmetry)Pollen disc diameter1.06 ± 0.181.06 ± 0.210(29)1
Receptor (Euv)0.05 ± 0.050.07 ± 0.08−1.49(29)0.15
Excitation
 Eblue0.22 ± 0.140.25 ± 0.15−1.8(29)0.08
 Egreen0.81 ± 0.010.81 ± 0.01−1.71(29)0.1
(X)0.65 ± 0.040.64 ± 0.071.27(29)0.22
(Y)−0.20 ± 0.13−0.18 ± 0.12−0.87(29)0.39

Comparisons of the receptor excitation values for each of the UV, blue and green parts of the spectrum demonstrated no significant differences between chosen and rejected flowers in either petals (Table 3) or the central disc (Table 4). However, as the p-values for honeybees were close to or at significance without the Bonferroni adjustment of alpha levels, an additional analysis compared the values of radially symmetric flowers and bilaterally symmetric flowers, but no significant difference was revealed (paired t-tests: petal – UV: t38 = −0.01, p = 0.49; petal – blue: t38 = −0.61, p = 0.27; petal – green: t38 = −0.06, p = 0.48; pollen – UV: t38 = 1.25, p = 0.89; pollen – blue: t38 = 1.74, p = 0.96; pollen – green: t38 = 1.61, p = 0.94). Similarly, no significant differences were detected in the colour hexagon values of chosen and rejected flowers (Tables 3 and 4).

Where crab spiders were found on C. frutescens in the field, no significant differences were found in overall flower diameter. Similarly, no differences were found in central disc diameter and flower asymmetry between occupied and unoccupied flowers (Table 5).

Table 5.  Flower characteristics in field
Flower characteristicOccupied flowerUnoccupied flowerStatistics
tdfp
  1. Overall flower and central pollen disc diameter (cm) between flowers occupied and unoccupied by crab spiders in the field. The degree of flower symmetry was calculated using the standard deviation of petal lengths within each flower. All values are inline image ± SD.

Flower diameter5.03 ± 0.145.03 ± 0.12−0.02(49)0.99
Pollen disc diameter1.08 ± 0.141.08 ± 0.02−0.20(49)0.84
Degree of flower symmetry0.20 ± 0.590.14 ± 0.05−1.36(49)0.17

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

As predicted, honeybees and crab spiders demonstrated a preference for symmetrical flowers over flowers that are asymmetrical. Furthermore, honeybees, unlike crab spiders, distinguished between radial and bilateral symmetry, preferring the former. Studies of symmetry preferences in the past have generally focused on the use of artificial stimuli (e.g. Giurfa et al. 1996). Our results demonstrate that this preference holds for real flowers in honeybees and crab spiders. Our field observations showed that crab spiders did not select flowers that showed less variation in petal length. This measure of asymmetry corresponds to fluctuating asymmetry, something different from asymmetry caused by the loss of entire petals because of herbivory or senescence.

The results indicate that gross asymmetry, exemplified by flowers with entire petals missing, is discriminated from radial symmetry by honeybees and crab spiders, although far from perfectly. This level of perception allows these arthropods to reject senescent flowers with missing petals. The visual resolution of honeybees is limited (Gould 1987). Although a high level of symmetry in flowers may indicate high quality (Møller 1995; Møller & Sorci 1998; Goulson 1999), honeybees probably cannot discriminate fine-scale fluctuating asymmetry. Crab spiders’ visual resolution is also probably unable to discriminate fine differences. However, we have grounds to believe that characteristics other than symmetry (e.g. odours) influence honeybee and crab spiders’ foraging decisions (Heiling et al. 2004).

Two, non-mutually exclusive, mechanistic explanations may be considered for why honeybees exhibited a significant preference for the radially symmetric flower over the bilateral. First, honeybees may possess an innate preference for radially symmetric patterns. Innate preferences have been demonstrated in a variety of organisms, resulting in the evolution of secondary sexual characteristics and behavioural traits in mating systems (e.g. Ryan & Rand 1990; Rosenthal & Evans 1998). However, these biases may arise from signal responsiveness important in other contexts (Endler & Basolo 1998). Indeed, an innate preference for shapes that mimic flowers has been demonstrated in honeybees by Lehrer et al. (1995), although it remains unclear whether honeybees prefer radial symmetry or many axes of symmetry. The preference for radial symmetry in flowers may be maintained by higher pollinator rewards if more symmetrical flowers indeed provide higher rewards (Møller & Sorci 1998; Møller 2000). Secondly, honeybees may learn that radial flowers possess the best rewards through positive reinforcement in a manner similar to that suggested by Møller & Eriksson (1995) for symmetry preference over asymmetry.

Random removal of petals in the asymmetric flowers also creates differences in the overall contour line of the flowers (Dafni et al. 1997), in contrast to the systematic removal of alternate petals in the symmetric flowers. Thus, honeybees or spiders may have chosen flowers for their specific contour line rather than level of symmetry. However, no correlation was found between the amount of contour line and honeybee or crab spider choices.

While we have eliminated odour from our experimental treatments, under natural conditions, honeybees may use odour in addition to symmetry to make foraging decisions. For example, honeybees and crab spiders will choose (at a significant level) the same flower from two randomly chosen C. frutescens daisies (Heiling et al. 2004). Flowers that were chosen did not differ significantly in age, size or reflectance from those that were rejected. The use of olfactory cues is implicated, because the elimination of olfactory cues, by covering flowers with Gladwrap™, led to random levels of agreement between the two species (Heiling et al. 2004).

Senescence of flowers may be predicted through loss of petals (and thus, asymmetry), colour (e.g. Weiss 1991), and potentially, odour. A comparison of choices made between symmetrical and asymmetrical flowers with olfactory cues included is required to test the role of olfaction. Previous research has already demonstrated that the relative importance of colour and pattern perception of flowers differs with the distance the bee is from the flower (Giurfa et al. 1995). Similarly, we predict that flower symmetry is less important than olfactory cues in the patch choices of honeybees and crab spiders.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

We are grateful to Lars Chittka for help with the analyses of reflectance data. This study was supported by the Australian Research Council (DP0449673 to ME Herberstein, K Cheng and L Chittka), and the Austrian Science Foundation (grant no. J2249 to AM Heiling). K Cheng also thanks the Wissenschaftskolleg zu Berlin (Berlin Institute for Advanced Study), at which he was a fellow while this manuscript was written.

Literature Cited

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited