Visual signals communicated at ultraviolet wavelengths, which are invisible to humans and are therefore more difficult to analyse3,4, may be used by ambush predators to manipulate their prey's behaviour and increase capture success. We have investigated how T. spectabilis interferes with floral signals, and the effect of its visibility on the attractiveness of the flower to pollinating insects.

Under natural light conditions, we presented honeybees with pairs of randomly selected white daisies, one of which carried an anaesthetized spider, and recorded which of the two flowers the bee visited first. We then repeated the experiment using a plastic foil covering on each flower and spider; the cover blocked olfactory cues but was transparent to light of wavelengths greater than 300 nm.

Compared with empty flowers, the presence of white crab-spiders on the petals of daisies evidently attracted honeybees more, in both the presence and absence of olfactory cues (Fig. 1a). This indicates that the bees must have been guided by visual signals alone, and that the visual signal generated by the spider renders the flower more inviting to bees.

Figure 1: The effect of the presence of the crab-spider Thomisus spectabilis on the white daisy (Chrysanthemum frutescens) on flower visitation by honeybees (Apis mellifera).
figure 1

a, Proportion of bees that visit vacant daisies (yellow bars) and daisies occupied by spiders (blue bars) in the presence (binomial test, n = 33, P = 0.0045) and absence (n = 25, P = 0.0053) of olfactory cues. All spiders, flowers and bees were used only once. b, Difference in honeybee colour-receptor excitation values (mean ± s.e.; for methods, see ref. 2) between spiders' abdomens and daisy petals at different wavelengths (ultraviolet receptors, λmax = 345 nm; blue receptors, λmax = 440 nm; green receptors, λmax = 535 nm; ref. 7). Tukey test, asterisk denotes P < 0.001; NS, not significant.

To identify this signal, we measured the spectral reflectance from 300 to 700 nm of the flower petals and of the spiders' abdomens. We calculated the colour contrast5 of the spiders against the flower petals and computed the euclidean distances in the colour space of hymenopterans2. We found that, compared with the flowers, white spiders reflect a considerable amount of ultraviolet light.

There was also a pronounced difference in the honeybee receptor-excitation values generated by spiders and flowers at ultraviolet wavelengths (Tukey-tests, P < 0.001 and P < 0.001, respectively), but not in the blue and green regions of the spectrum (ANOVA, F2,74 = 136.8, P < 0.001; Fig. 1b), where receptor excitation is comparable for both (Tukey test, P = 0.901). Consequently, instead of being cryptic, as they are to humans, the spiders produce a strong colour contrast that is detectable by their hymenopteran prey (mean euclidean distance in colour space ± s.e., 0.14 ± 0.01; n = 25). The values for colour contrast are well above the detection threshold of 0.05 (ref. 2; one-sample t-test, t24 = 7.6, P < 0.001).

We conclude that T. spectabilis uses quite the opposite signalling strategy to that known to be used by other crab-spiders1,2. T. spectabilis is difficult to perceive from far away, when bees use only their green-receptor signal to detect objects6, but is highly conspicuous in the insect visual spectrum when seen at close quarters. Because ultraviolet-reflecting white flowers are extremely rare in nature5, the spider will contrast strongly with almost any natural flower.

T. spectabilis will also be just as conspicuous to other flower visitors, as all known pollinating insects, including stingless bees7 (which are the spider's most likely Australian native prey), perceive ultraviolet light. We propose that the presence of spiders on flower petals creates a colour pattern that is particularly effective because bees have a pre-existing bias towards it — an idea that is consistent with empirical data showing that bees innately prefer flowers with strongly contrasting markings8.