FLORAL BIOLOGY OF Peter Goldblatt,2 Ingrid NaÈnni,3 Peter 4 5 HESPERANTHA (IRIDACEAE: Bernhardt, and John C. Manning CROCOIDEAE): HOW MINOR SHIFTS IN FLORAL PRESENTATION CHANGE THE POLLINATION SYSTEM1 ABSTRACT Field observations, ¯oral dissections of a representative range of Hesperantha species, and pollen load analyses of insects captured on many of them indicate that ¯owers of this African genus are cross pollinated by a relatively broad range of insects. The pollination ecology of Hesperantha can be divided into four overlapping systems that exploit insects of four orders (Coleoptera, Diptera, Hymenoptera, and Lepidoptera). Species of the H. falcata type have erect or nodding, salver-shaped, strongly fragrant, white ¯owers that open in the mid to late afternoon and evening and are pollinated by long-tongued apid bees and/or noctuid moths. Species of the H. pauci¯ora type have a virtually identical ¯oral morphology, but the perianth is yellow or pink to mauve or blue and the ¯owers are usually unscented and are open during the day, closing between midday and late afternoon, ca. 16:30 H. Flowers of this type are also pollinated by apid bees, but in the southern African winter-rainfall zone other effective pollinators include nemestrinid ¯ies (Prosoeca) with relatively short probosces and hopliine scarab beetles. In H. latifolia type ¯owers the perianth is pink to magenta or red (rarely pale yellow), odorless, opens during the day but has an elongate perianth tube exceeding 18 mm in length. These ¯owers are pollinated mainly by long-proboscid ¯ies in the genera Prosoeca and Stenobasipteron (Nemestrinidae) or Philoliche (Tabanidae), but the red ¯owers of H. coccinea are pollinated by a guild of large butter¯ies including Papilio and the satyrid Aeropetes. Lastly, H. vaginata has odorless and nectarless, short-tubed yellow ¯owers, usually with contrasting dark markings, that open only during the day and are pollinated exclusively by the hopliine scarab beetle, Clania glenlyonensis. The taxonomic distribution of plant species with these pollination systems makes it clear that shifts in pollination systems have occurred repeatedly across Hesperantha, although ¯oral morphology and nectar biochemistry are relatively conservative. Whether ¯owers are nocturnal, crepuscular, or diurnal, only four vari- ables affect the ¯oral ecology: length of the perianth tube, presence or absence of ¯oral fragrance, timing of anthesis, and the closely associated trait of perianth color. Thus, species with pink, magenta, red, or yellow ¯owers close at night and are rarely fragrant, whereas those with white or pale yellow ¯owers are nearly always fragrant and either open late in the day or only at sunset and remain open for most of the night. Species show considerable variation in nectar volume and sugar concentration, closely correlated with pollination system, while two long-tubed species with ¯oral characters typical of long-proboscid ¯y pollination produce no nectar and are inferred to be Batesian or guild mimics that achieve pollination by deception. Key words: bees, ¯oral ecology, Hesperantha, hopliine beetles, Iridaceae, long-proboscid ¯ies, nectar, pollination systems, settling moths. The radiation and diversi®cation of the African For example, Lapeirousia is pollinated by long-pro- Iridaceae depend to a greater or lesser extent on boscid ¯ies, or bees and butter¯ies, or moths (Gold- the plasticity of pollination mechanisms, and all the blatt et al., 1995). Romulea exploits scarab beetles, larger genera of the family exhibit a wide range of pollen-collecting bees or, in at least two species, ¯oral adaptations and correlated sets of insect or long-proboscid nemestrinid ¯ies (Goldblatt et al., avian pollinators (Bernhardt & Goldblatt, 2000). 1998a; Manning & Goldblatt, 1996, and unpub- 1 Support for this study by grants 5408-95, 9554-97, and 6704-00 from the National Geographic Society is gratefully acknowledged. We thank Robert Brooks, Holger Dombrow, Douglas Kroon, H. Kruger, and Kim E. Steiner for their help with the identi®cation of insects; Margo Branch and John Manning for the illustrations; John Donaldson and his group for sharing ®eld facilities at Nieuwoudtville; Neil MacGregor for allowing us to conduct our research on his property; and Cameron McMaster and Lendon Porter for assistance with ®eldwork. Collecting permits were provided by the Nature Conservation authorities of Mpumalanga, Western Cape and Northern Cape Provinces, South Africa. 2 B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166- 0299, U.S.A. [email protected]. 3 National Botanical Institute, P. Bag. X7, Claremont 7735, South Africa. [email protected]. 4 Department of Biology, St. Louis University, St. Louis, Missouri 63103, U.S.A. [email protected]. 5 Compton Herbarium, National Botanical Institute, P. Bag. X7, Claremont 7735, South Africa. manning@nbict. nbi.ac.za. ANN.MISSOURI BOT.GARD. 91: 186±206. 2004. Volume 91, Number 1 Goldblatt et al. 187 2004 Floral Biology of Hesperantha lished). The majority of species of Gladiolus appear MATERIALS AND METHODS to be pollinated primarily by nectar-feeding apid INFLORESCENCE PHENOLOGY AND FLORAL LIFE and anthophorine bees (Goldblatt et al., 1998b), SPAN but some red-¯owered species are pollinated by the large butter¯y Aeropetes (Johnson & Bond, 1994), Direct observations are presented on 34 Hesper- while others are dependent on andrenid bees, a antha species made in the ®eld from 1993 to 2002 combination of bees and hopliine beetles (Goldblatt (Table 1) and supplemented by living collections at et al., 1998a), long-proboscid ¯ies, moths, or birds Kirstenbosch Botanic Gardens, Cape Town, and the (Goldblatt & Manning, 1998; Goldblatt et al., 1999, Missouri Botanical Garden, St. Louis. Observations 2001). were made in the southern spring, summer, or au- Hesperantha, a mid-sized genus of subfamily tumn at sites in southern Africa (August to April). Crocoideae (syn. Ixioideae), comprises approxi- Observations of insect foraging involved 4 to 10 hours per plant species (or occasionally more) and mately 80 species (Goldblatt, 1984, 1986, 1987, included recording of ¯oral attractants (pigmenta- 1993, 2003; Hilliard & Burtt, 1986; Goldblatt & tion, scent), rewards (nectar), the mode and timing Manning, 1996) distributed from the southern tip of anthesis (opening of individual buds), daily phe- of Africa through the eastern African mountains as nology, anther dehiscence patterns, expansion of far north as Ethiopia and Cameroon. Diversity is stigmatic lobes, the behavior of insects on the ¯ow- greatest in southern Africa, where there are two er, and the taxonomic diversity of ¯oral foragers. centers of diversity and regional endemism: the Floral scent was noted in the ®eld and in cultivated Drakensberg of South Africa and Lesotho; and the plants. Scents too weak to be immediately dis- west coast and near interior of Northern Cape and cerned by the human nose were recorded after in- Western Cape Provinces of South Africa. Compared dividual ¯owers were picked and placed in clean, with the approximately 260 species of Gladiolus, lidded glass jars and stored in a warm place. The the 50 species each of Ixia or Watsonia, or the 40 contents of each jar was smelled after a minimum species of Lapeirousia, Hesperantha species have a of 60 minutes (Buchmann, 1983). Plant voucher conservative ¯oral morphology. The relatively small specimens are deposited in the herbaria at the Mis- ¯owers (tepals typically 10±25 mm long, exception- souri Botanical Garden (MO) and Kirstenbosch Bo- ally to 37 mm) are radially symmetric, have sube- tanic Gardens (NBG) (Table 1). qual tepals, and are arranged in slender to compact spicate in¯orescences. The ¯owers of most species COMPATIBILITY have a symmetrical androecium and a style that Compatibility relationships were examined in diverges at the mouth of the ¯oral tube into three two species, Hesperantha quadrangula and H. pal- long, spreading branches. Interspeci®c ¯oral vari- lescens, both maintained in the laboratory and iso- ation is restricted to perianth tube length and de- lated from possible pollinators. Fruit and seed set gree of curvature, perianth color (uniformly white were compared in ®ve hand-selfed ¯owers and ®ve to cream vs. various shades of pink to purple, or ¯owers crossed with pollen of another individual of yellow, sometimes with dark markings), the pres- the same species. In the Iridaceae hand-sel®ng of ence or absence of fragrance, the type of ¯oral odor, as few as ®ve ¯owers is usually suf®cient, as results and the timing of anthesis (Goldblatt, 1984, 2003). are seldom mixed. Either all self crosses result in A particularly unusual aspect of ¯oral variation seed set or none do. Likewise, xenogamous crosses within this genus is the close linking of ¯oral pig- always succeed. mentation with the presence of ¯oral odor and tim- ing of anthesis. White- or cream-¯owered species NECTAR ANALYSIS of the southern African winter-rainfall zone are cre- Nectar volume measurements were taken pri- puscular or nocturnal and produce a strong ¯oral marily from unbagged ¯owers in the ®eld, soon after fragrance, whereas species with ¯owers of other they opened, re¯ecting both rates of secretion and colors, and the white-¯owered species of eastern depletion. To collect nectar, mature ¯owers were southern Africa, are typically odorless, at least to picked and nectar was withdrawn from the base of the human nose. Field studies of the pollination the perianth tube with 3 ml capillary tubes after systems of a selection of Hesperantha species were separating the ovary from the perianth base. The undertaken to de®ne and compare intrageneric percentage of sucrose equivalents in fresh nectar trends in the evolution of pollination mechanisms was measured in the ®eld or laboratory using a Bel- and the function(s) of ¯oral traits. lingham and Stanley hand-held refractometer (0± 188 Annals of the Missouri Botanical Garden Table 1. Study sites and voucher information for Hesperantha species studied.