Oecologia (2003) 136:571–573 DOI 10.1007/s00442-003-1336-y

PLANT ANIMAL INTERACTIONS

Renee M. Borges · Vinita Gowda · Merry Zacharias Butterfly pollination and high-contrast visual signals in a low-density distylous

Received: 24 July 2002 / Accepted: 12 June 2003 / Published online: 4 July 2003 Springer-Verlag 2003

Abstract In low-density butterfly-pollinated structures are a vital advertising strategy for this distylous frondosa (), flowers attract pollinators at short plant, which occurs in very low densities and produces distances while conspicuous, non-rewarding accessory very few open flowers daily. High-contrast visual signals bracts are detectable at long distances by long-ranging have heightened detectability (Lehrer and Bischof 1995; pollinators such as the birdwing butterfly Giurfa et al. 1996; Ne’eman and Kevan 2001), facilitating that did not detect flower-bearing in the absence of associative learning between signal and reward that is these bracts. However, even in the absence of flowers, the vital for successful advertising. white, ultraviolet-absorbing bracts attracted butterflies that visited flowerless plants. Although flower visits by short-ranging territorial butterflies declined significantly Materials and methods on removal of bracts, they did not cease completely. Nectar-robbing carpenter bees and birds did not change The plant and study site their behaviour following bract removal. Bract removal is distylous, individual plants bearing either caused a significant decline in fruit set, indicating their long- or short-styled flowers. Fruit is set only with pollen from the importance as visual signals to pollinators. opposite stylar morph, though pollen of both morphs germinates (unpublished data). The plant occurs in extremely low densities Keywords Advertisement · Heterostyly · Reproductive along forest edges (e.g. 15 plants along a 2-km linear edge), producing small, tightly closed, tubular, bright orange flowers. strategy · Spectral reflectance Usually one sepal in a group of flowers is enlarged into a white, leaf-shaped, vertically presented bract (Fig. 1a). Flower longevity is 24 h while individual bracts persist through the flowering season. Introduction The experiments were performed at the edges of a low elevation semi-evergreen forest at Karinja (12520N, 75E) in Karnataka State of the Western Ghats of . Animal-pollinated plants find mates by attracting polli- nators. Low-density plants can suffer from a minority disadvantage in the pollination market (Silander 1978) Experiments and spectral measurements and may need to increase attractiveness of their floral We obtained a spectral reflectance of bracts and leaves between displays possibly by deploying long-lasting, non-reward- 330 and 721 nm relative to a ceramic, highly ultraviolet-reflecting ing accessory structures as visual signals. white standard under an angle of 45 (minimising surface In butterfly-pollinated Mussaenda frondosa (Rubi- reflection) with a Zeiss spectrophotometer MCS230, using a Zeiss aceae) we show that high-contrast secondary sexual CLXII xenon light source. We conducted experiments in which we removed either bracts or flowers from entire plants and compared visitation rates between R. M. Borges ()) · M. Zacharias pre-treatment (control) and treatment conditions for the same Centre for Ecological Sciences, plants. Plants in the control condition had both bracts and flowers. Indian Institute of Science, We recorded visitation to flowers on plants with and without bracts Bangalore 560 012, India (n=11 plants) for 5 days each in the pre- and post-bract removal e-mail: [email protected] states from 0700 to 1730/1800 hours in 30-min durations and Tel.: +91-80-3602972 similarly to plants with and without flowers (n=12 plants) for 2 Fax: +91-80-3601428 days each in the pre- and post-flower removal states. We used different sets of plants for each experiment, which may explain the V. Gowda difference in visitation rates between the two experiments. We Department of Biological Sciences, pooled data across days for each plant for each experiment, and George Washington University, calculated flower visitation rates (events/min) for each visitor type per plant. A flower visit consisted of probing into or hovering Washington, DC 20052, USA 572

Fig. 1 a The flowers and bracts of Mussaenda frondosa. b Spectral probing — butterflies: 2.70, birdwing butterflies: 2.20, carpenter reflectance of leaves and bracts. c, d Comparison of pre-treatment bees: 0.42, birds: 1.60. d Flower-removal experiment: (n=12 control and treatment observations for visitation rates (flower plants). Z-values for Wilcoxon matched pairs signed ranks tests. probing or hovering) by each group of visitors to plants Flower probing — butterflies: 3.06, birdwing butterflies: 2.20, (means€SD) using a matched pairs test. Control plants have both carpenter bees: 2.66. Hovering — butterflies: 0.75, birdwing bracts and flowers. *P<0.05, **P<0.01. Butterflies Butterflies butterflies: 2.36, carpenter bees: 1.09. Solid triangles represent excluding the birdwing. c Bract removal experiment: (n=11 plants). values=0 by definition, but shown to indicate that probing visits to Z-values for Wilcoxon matched pairs signed ranks tests. Flower flowers in control state were significantly greater than 0 around (measured only in the flower removal experiment) flowers birdwing Troides minos, which we placed in a separate or former flower locations. Extremely low plant densities con- category because it is a long-ranging canopy butterfly, strained sample sizes. We recorded the number of open flowers on the experimental plants each day and morphometric parameters of which descended to feed on flowers. all bracts harvested from experimental plants in the bract removal Following bract removal, visitation by all butterflies to experiment. flowers on such plants declined significantly. However, territorial butterflies continued to visit such bractless plants, whereas long-ranging birdwing butterflies made Results no visits to such plants, suggesting that they did not detect them in the absence of bracts (Fig. 1c). Flower visitation Plants presented very few flowers (mean€SD=7.7€5.2, by nectar-robbers was not significantly affected by bract n=11 plants) but many bracts (mean€SD=49.4€35.4, removal (Fig. 1c). Birds did not visit the set of plants n=11 plants) per day. Average (SD; maximum) measure- chosen for the flower removal experiment; consequently ments of individual bracts per plant (n=11 plants) were: their response to flower removal is unknown. Importantly, length 7.2 cm (0.9; 10.2), width 4.2 cm (1.1; 7.0), contour fruit set (% flowers setting fruit) of plants with bracts outline or perimeter 18.5 cm (3.8; 27.5) and area 20.2 cm2 (mean€SD=71.3€32.6%) was significantly higher than (4.3; 29.5). The bracts absorb ultraviolet (Fig. 1b). that of those without bracts (mean€ SD=24.3€18.1%) Butterflies, carpenter bees (Xylocopa sp.) and sunbirds (Wilcoxon matched pairs signed ranks test; Z=2.49, (Nectarinia zeylonica) visited the flowers. Butterflies, P<0.05; n=10), indicating the importance of bracts as representing 57.4% of visits, were legitimate visitors; the pollinator signals. Territorial butterflies continued to others with 36.1% and 6.5% of visits respectively were hover over plants from which flowers had been removed nectar robbers. Most butterflies were territorial under- and birdwing butterflies also visited these plants (Fig. 1d), storey species, which made 65.3% of visits. The other showing that bracts were effective signals for birdwing 34.7% of butterfly visits were made by the southern butterflies. 573 Discussion which the flowers alone would not have been able to achieve. Low-density plants must be individually conspicuous as they cannot rely on the combined advertisement of Acknowledgements We thank Andreas Gumbert for the spectral neighbours. Low-density Mussaenda frondosa advertises reflectance measurements; Veena C.P. for field assistance; Lars Chittka, Andreas Gumbert, Natalie Hempel de Ibarra, Almut using non-rewarding bracts that increase reproductive Kelber, Peter Kevan, Michiyo Kinoshita, Doekele Stavenga and success indicated by the significant decline of fruit set in Misha Vorobyev for useful suggestions; and Almut Kelber, Natalie bractless plants. Although sterile accessory structures can Hempel de Ibarra, and the reviewers for valuable comments on the increase attractiveness (Bell 1985; however see Herrera manuscript. This research was funded by the Department of 1997), we demonstrate differential responses of various Science and Technology, Government of India. flower visitors to their removal. Birdwing butterflies did not visit flowering bractless plants although they visited flowerless plants with intact bracts. Since butterflies are References capable of associative learning (Papaj 1986; Weiss 1995), Bell G (1985) On the function of flowers. Proc R Soc Lond Ser B this could be because of a learned association between 224:223–265 bracts and flowers. Territorial, short-ranging butterflies, Giurfa M, Vorobyev M, Kevan PG, Menzel R (1996) Detection of however, continued to visit flowers of bractless plants coloured stimuli by honeybees: minimum visual angles and although such visits declined significantly, suggesting that receptor specific contrasts. J Comp Physiol A 178:699–710 Herrera J (1997) The role of colored accessory bracts in the they use other cues (Stanton 1984), or trapline (Waller reproductive biology of Lavandula stoechas. Ecology 78:494– and Gilbert 1982) to re-visit flower locations. Nectar- 504 robbing carpenter bees and birds, on the other hand, did Lehrer M, Bischof S (1995) Detection of model flowers by not alter their behaviour following bract removal. Since honeybees: the role of chromatic and achromatic contrast. Naturwissenschaften 82:145–147 there was no visitation by birds in the flower-removal Ne’eman G, Kevan PG (2001) The effect of shape parameters on experiment (even to control plants), their response to maximal detection distance of model targets by honeybee flower removal is unknown. workers. J Comp Physiol A 187:653–660 Bracts of M. frondosa contrast strongly with leaves in Papaj DR (1986) Conditioning of leaf-shape discrimination by chemical cues in the butterfly, Battus philenor. Anim Behav all spectral ranges but the ultraviolet. This together with 34:1281–1288 their size and undissected outline should make them Silander JA (1978) Density-dependent control of reproductive detectable at large distances (Weiss 1991; Lehrer and success in Cassia biflora. Biotropica 10:292–296 Bischof 1995; Giurfa et al. 1996; Ne’eman and Kevan Spaethe J, Tautz J, Chittka L (2001) Visual constraints in foraging 2001; Spaethe et al. 2001). Furthermore, M. frondosa bumblebees: flower size and color affect search time and flight behavior. Proc Nat Acad Sci USA 98:3898–3903 occurs only in high light situations where visual signals Stanton ML (1984) Short-term learning and the searching accuracy are most exploitable. The flowers seem to signal to of egg-laying butterflies. Anim Behav 32:33–40 butterfly pollinators at short range while non-rewarding Waller DA, Gilbert LE (1982) Roost recruitment and resource bracts appear to specifically provide long-range cues to utilization: observations on a Heliconius charitonius L. roost in Mexico. J Lepid Soc 36:178–184 long-distance fliers such as the birdwing butterfly. This Weiss MR (1991) Floral colour changes as cues for pollinators. could be a strategy for the acquisition of long-distance Nature 354:227–229 genes and for the export of genes to further distances, Weiss MR (1995) Associative colour learning in a nymphalid butterfly. Ecol Entomol 20:298–301