Conospermum

Abstract: Striking convergent evolution for a hidden (cryptic), ground flowering (geoflorous) habit in distantly related, low shrubby Australian and South African Proteaceae is interpreted as an adaptation for pollination by nonflying mammals. The cryptic, geoflorous habit is especially well developed in species groups of Dryandra in southwestern Australia and Protea in the Cape region of South Africa. Considerable circumstantial evidence exists in both regions for pollination by mouselike, often arboreal marsupials in Dryandra and true rodents in Protea. Evidence from inflorescence structure suggests the cryptic, geoflorous habit is derived from bird-pollinated species, possibly in response to fires common in the sclerophyllous communities where these genera grow. A number of floral characteristics and the occurrence in Australia of mouselike marsupials adapted to a nectar (and pollen?) diet suggests that a class of flowers has evolved for pollination by nonflying mammals. This postulated floral class possibly also extends to other Australian arboreal proteaceous and also myrtaceous genera, but in South Africa is probably restricted to Protea. Pollination by nonflying mammals is largely ignored or given little credence in current treatments of pollination ecology (Faegri & van der Pijl, 1971; Proctor & Yeo, 1972). There is, however, good reason for this; all the available evidence relating to this phenomenon is either circumstantial, inferential, or anecdotal. Nonetheless, field observations in Australia and South Africa and a subsequent search of the literature have led us to believe that true rodents and marsupials may, in fact, be the normal pollinators of several southern hemisphere proteaceous genera. Furthermore, various floral characteristics in these genera and the special adaptations for nectar feeding in some of the putative pollinators suggest structural coadaptations by both flowers and apparent pollinators. Although plans are underway to conduct definitive studies, no unequivocal evidence can be presented at this time for regular pollination by nonflying mammals, and Kirstenbosch Botanic Garden, Newlands, Cape Province, South Africa 7700. 2 Department of Biology, University of Utah, Salt Lake City, Utah 84112, U.S.A. ANN. MISSOURI BOT. GARD. 64: 1-17. 1977. This content downloaded from 157.55.39.153 on Mon, 19 Sep 2016 04:39:11 UTC All use subject to http://about.jstor.org/terms 2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VOL. 64 that is not the intent of this paper. We hope, however, that our comments will help to reopen and stimulate research in this fascinating area of pollination biology pioneered by Porsch (1934, 1935, 1936a, 1936b) and subsequently neglected for 40 years. The purpose of this paper is fourfold: (1) to elucidate our observations and ideas on inferred pollination by nonflying mammals in the South African and Australian Proteaceae, (2) to point out the striking convergent evolution of flowering habits between southwestern Australian and South African Proteaceae of the Cape region, (3) to review some of the rather scattered and fragmentary literature on the subject, and (4) to evaluate the evidence for the existence of a class of flowers adapted to pollination by nonflying mammals. References to the subject of pollination by nonflying mammals usually mention the arboreal Australian marsupials which apparently feed on nectar (e.g., the honey possum, Tarsipes spencerae) and to introduced rats suspected of pollinating a climbing pandan (Freycinetia arborea Gaudich.) in Hawaii. Faegri & van der Pijl (1971) and Proctor & Yeo (1972) furthermore state that no flowers appear to be adapted for pollination by nonflying mammals, although Faegri and van der Pijl mention the classic papers by Porsch (1934, 1935, 1936a, 1936b) in which he builds a case for floral adaptations to pollination by nonflying mammals in several Australian genera. Grant (1950) mentions, without comment, a "marsupial" pollinated flower class based on the proteaceous genus Dryandra. In addition to rodents and mouselike marsupials, some primates may also be regular pollinators. For example, according to Coe & Isaac (1965) the baobab (Adansonia digitata L.) is pollinated in East Africa by the lesser bush baby (a lorisid primate). This characteristic African tree is generally considered to be bat pollinated. Petter (1962) mentions that several arboreal, mouselike lemurs (Lemur, Varecia, Hapalemur, Microcebus) visit flowers seeking nectar and are generally attracted by sweet liquids in captivity. More recently Sussman & Tattersall (1976, and personal communication) demonstrate that Lemur mongoz mongoz is apparently an important pollinator of introduced kapok (Ceiba pentandra Gaertn.) in Madagascar. F. L. Carpenter (personal communication) has data from Australia indicating that some species of Banksia are pollinated almost entirely by nonflying mammals, including an indigenous rat (Rattus fuscipes) and various marsupials. It is not our intent to evaluate the entire literature here. There are, however, numerous instances of various mammals being observed on or around flowers (Porsch, 1934), but the nature of their activities are, in fact, virtually unknown. As Faegri & van der Pijl (1971) point out with respect to pollination by nonflying mammals "much research remains to be done to establish relationship between possible regular pollinators and the blossoms in which they work." FLORAL CHARACTERISTICS AND CONVERGENT EVOLUTION OF PROTEACEAE PUTATIVELY POLLINATED BY NONFLYING MAMMALS The most obvious Proteaceae are trees and large shrubs, e.g., Grevillea and Banksia in Australia, and Protea in South Africa. Less known, however, is the occurrence of species groups on both these continents with inflorescences at or near ground level (geoflorous) and typically obscured from external view by This content downloaded from 157.55.39.153 on Mon, 19 Sep 2016 04:39:11 UTC All use subject to http://about.jstor.org/terms 1977] ROURKE & WIENS-NONFLYING MAMMAL POLLINATION 3 overlying foliage (cryptic). The taxonomic distribution of these cryptic, geoflorous species is limited principally to two distinct sections of Protea [Hypocephalae and Microgeantheae, sensu Phillips (1912) ] and some additional species of uncertain sectional classification in the Cape region of South Africa; in southwestern Australia, however, this flowering habit is associated with at least five genera (Banksia, Conospermum, Dryandra, Isopogon, and Petrophile) but is best developed in Dryandra [series Aphragmia and Niveae, sensu Bentham (1870)] and to a somewhat lesser extent in Banksia. In these equivalent infrageneric groupings in Protea and Dryandra the growth habit is low, tufted, and often rhizomatous. The flowers occur in heads, usually at ground level, or occasionally up to 30 cm high, but in either case the heads are typically deeply hidden within the foliage of the dense and widely spreading branch systems. The heads are generally visible only if the branches are forcibly parted and the base of the plant carefully examined (Figs. 1-6). The flowers are surrounded by a prominent series of overlapping bracts forming a cup-shaped involucre. The bracts vary in color through various shades of brown and are often flushed with different dull reddish tints. An inflorescence contains perhaps 100-200 flowers, but the large spikes of Banksia bear several thousand individual flowers. Many of the species produce copious amounts of nectar and the heads often emit a distinctive, "nutty" or "yeasty" odor. In the cryptic, geoflorous Cape species of Protea the basal portions of the bracts and flowers, particularly the styles, are also markedly succulent. Excellent illustrations of Protea flowers (but not necessarily the growth habits) can be seen in Rousseau (1970) for South African proteas and in Erickson et al. (1973) for Australian genera. Dryandra, as in most western Australian Proteaceae, develops no obvious succulence in the inflorescence or flowers. In general, the geoflorous habit, the cryptic positioning of the inflorescences, and the gross (though superficial) morphological similarities of the heads suggest strong convergent evolutionary tendencies. In fact, from a distance one would be hard pressed to distinguish between some species of Dryandra and Protea even though these genera represent the end points of evolution in two subfamilies of the Proteaceae, Grevilleoideae and Proteoideae, respectively, and occur on widely separated continents (Figs. 1-6). EVIDENCE FOR RODENT POLLINATION IN SOUTH AFRICAN PROTEAS Field observations over a period of years of the cryptic, geoflorous species of Protea in the Cape area show that considerable rodent activity is associated with these species (Table 1), but is especially obvious in P. subulifolia. The specific rodent activities associated with this species are: (1) freshly chewed involucral bracts and styles during and just prior to anthesis (Fig. 7); (2) clearly demarcated networks of heavily used runways linking different plants within populations, and which often intertwine around flowering and old fruiting heads; and (3) occasional burrows at the base of the plants. The runways and burrows are related to activities of the Cape striped field mouse (Rhabdomys pumilio pumilio). On various occasions and in different populations this animal (which is diurnal) was observed on runways between This content downloaded from 157.55.39.153 on Mon, 19 Sep 2016 04:39:11 UTC All use subject to http://about.jstor.org/terms 4 ~~~~~~ANNALS OF THE MISSOURI BOTANICAL GARDEN [VOL. 64