Ultraviolet Floral Patterns in the Native Hawaiian Flora: What Do They Mean for Island Biogeography?l

We examined 104 species (13%) of the approximately 784 species of biotically pollinated plants native to Hawai'i and found 14 (13.5%) that have an ultraviolet (UV) floral pattern. However, detailed examination revealed that 32% of the Hawaiian strand species have UV floral patterns, whereas only 8% of the upland species did. All of the flowers with UV patterns measured 1 cm or more in diameter and all but two appear yellow to humans. We discuss several possible explanations for the apparent paucity of UV floral patterns in the native Hawaiian upland flora. MOST PLANTS HAVE FLOWERS that either completely absorb or very weakly reflect ultraviolet light (UV), and the foliage of most plants is also slightly reflective (Kevan 1983). Many plants are now known to have floral patterns that reflect ultraviolet light and that evidently function to attract and orient pollinating insects (Jones and Buchmann 1974). These patterns of reflection range from being almost completely reflective to possessing a few reflective streaks against an absorptive background. Earlier investigations, mostly of continental flora, have found that the proportion of species with reflective flowers within a given region varies from about 13% in the Canadian Arctic (Kevan 1974) to as high as 41% of the nonnative species naturalized in the California flora (Guldberg and Atsatt 1974). We examined the flora of the Hawaiian Islands to see if the reflective component of its remote, oceanic island flora would fall within this continental range. Also, Guldberg and Atsatt's analysis of the California flora indicated that certain floral characteristics, such as corolla size and visible color, have a greater propensity to significantly reflect UV. We recorded several floral, taxonomic, 1 Manuscript accepted 18 March 1998. 2 Department of Biological Science, California State University, Fullerton, California 92834. 3Department of Botany, University of Hawai'i at Manoa, Honolulu, Hawai'i 96822. 82 and biogeographic characteristics either to determine agreement of the Hawaiian flora with the mainland model or to help establish independent trends. MATERIALS AND METHODS From January 1979 to June 1979, native biotically pollinated flowers were collected, taken to the University of Hawai'i at Manoa, and videotaped using a portable television camera and videotape recorder (Sony model AVC 3400) with a Rodenstock UV-Rodagon F5.6j60 mm UV transmitting lens corrected for wavelengths between 5400 and 3400 A and fitted with a Tiffen 18A filter (Eisner et al. 1969). In addition, during summer months from 1986 to 1993 flowers were photographed in situ using a 35-mm camera (Canon Fl) equipped with a Tiffen 18A UV transmitting filter. Photographs were taken with Kodak Pan X or Plus X film at f 22, and multiple exposures from 5 to 9 sec were made of each flower. A total of 117 images, of 104 species from 46 families, was made. We noted flowers that reflected UV moderately to strongly, notably distinct from absorptive or lightly reflective backgrounds, and also flowers having distinctive patterns of reflective and absorptive areas. We did not use a gray scale and thus did not attempt to note the percentages of reflectance or reflection coefficients. In addiUltraviolet Floral Patterns in Native Hawaiian Flora-JoNES ET AL. 83 tion, we noted the flower color visible to humans, the corolla diameter, taxonomic affinities, whether the species were native or endemic, woody or herbaceous, and whether the species were strand or upland plants. For the last categorization, we used the definition of strand as discussed in Sohmer and Gustafson (1987) and Whistler (1980): that circumtropical community of plants dispersed onto island shores by ocean currents and affected by shoreline conditions such as salt spray, brackish ground water, and occasional seawater inundation. (This plant community is also referred to as "littoral" but because that term is also used of freshwater lakes, we prefer "strand" to distinguish the seashore community.) RESULTS AND DISCUSSION Of 104 species examined, only 14 (13.5%) were found to reflect UV or have a floral pattern of reflectance and absorbency. This is considerably lower than the values reported for most mainland floras and agrees with the value of 13% found in Arctic Canada (Kevan 1974). However, when the species were sepa. rated into strand flora and upland flora, an interesting pattern emerged: 32% of the strand species reflected UV, whereas only 7.9% of the upland species did so. Table 1 presents the taxa we sampled with their locations and observed characteristics. Seven families (15.2% of the 46 sampled families) were found to have some reflective members: Asteraceae, Fabaceae, Goodeniaceae, Malvaceae, Myoporaceae, Sterculiaceae, and Zygophyllaceae. In contrast, of the 61 families sampled by Guldberg and Atsatt (1974) in California, nearly 80% had at least one reflective species. In corolla size our findings agree with those of Guldberg and Atsatt. They reported that flowers greater than 1 cm in diameter are likelier to reflect UV and that the likelihood of reflectance increases with size. In our study, all but one (Myoporum sandwicense) of the reflective flowers were greater than or equal to 1 cm in diameter. Our results partially disagreed with those of Guldberg and Atsatt concerning the propensity of some visible colors to reflect. Theirs and earlier studies, which they cited, found that yellow and purple flowers were likeliest to reflect UV and that white and green flowers usually did not. Eleven of the 14 reflective Hawaiian flowers were yellow and none of the green flowers sampled reflected UV, but the other three reflective flowers were white. Hawai'i has relatively few species with purple or violet flowers and none that we sampled reflected UV. We consider our sample size to be quite robust. There are approximately 980 species of angiosperms in Hawai'i (Sohmer 1994) and about 20% of these are wind-pollinated (Carlquist 1974), leaving 784 biotically pollinated species. Our sample of 104 species represents 13.3% of the biotically pollinated species and 10.6% of the total angiosperms. In contrast, Guldberg and Atsatt (1974) sampled 300 species of California plants, but this represents only about 5% of the approximately 5700 species of angiosperms there (Hickman 1993). We are confident, therefore, that our survey presents an accurate picture of UV floral reflectance in Hawai'i. What can account for Hawai'i's apparent dearth of UV-reflecting flowers? There are two probable explanations: either few reflective species successfully colonized the Hawaiian Islands or a number of reflective species did establish but subsequently lost this ability. Each of these explanations presents dilemmas. In the case of failure to colonize, why should UV reflectance limit the ability of a species to disperse or establish new populations? In the second case, what factors present in Hawai'i could select against flowers reflecting UV? There are several avenues of investigation that need to be explored. An examination of the nonreflective species' closest relatives or the mainland populations from which the colonists originated might yield clues. Does Hawai'i's pattern hold true on other oceanic islands? Do continental islands also follow this pattern or are their species more likely to resemble those on the mainland? Because UV reflective patterns are known to be of importance in plant-pollinator interactions, what