Pollination of Ravenala madagascariensis (Strelitziaceae) by Lemurs in Madagascar: Evidence for an Archaic Coevolutionary System?

by W. John Kress, George E. Schatz, Michel Andrianifahanana, Hilary Simons Morland
Pollination of Ravenala madagascariensis (Strelitziaceae) by Lemurs in Madagascar: Evidence for an Archaic Coevolutionary System?
W. John Kress, George E. Schatz, Michel Andrianifahanana, Hilary Simons Morland
American Journal of Botany
Start Page: 
End Page: 
Select license: 
Select License



Department of Botany, NHB-166, National Museum of Natural History, Smithsonian Institution,
Washington, DC 20560; Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63 166-0299;
Centre National de Recherche sur I'Environnement, B.P. 1739, Antananarivo 101, Madagascar;
and Department of Anthropology, Yale University, P.O. Box 21 14 Yale Station,
New Haven, Connecticutt 06520

Investigations of the floral biology of the traveler's tree (Ravenala madagascariensis) and the ecology of the ruffed lemur (Varecia variegata), both endemic to the island of Madagascar, suggest a plant-pollinator relationship. Ravenala exhibits many specializations for visitation by large nonflying animals: inflorescences placed below the crown of the plant and easily accessible to arboreal animals; large flowers enclosed in tough, protective bracts that require a strong pollinator to open; stiff, rodlike styles that withstand the rough handling of the visitors; and copious, sucrose-dominant nectar that provides a renewable reward for a sizable animal for a long time period. Our observations in the field also show that Varecia variegata: consistently and almost exclusively visit the flowers of Ravenala; cany pollen on their fur between flowers on the same plant and between conspecific plants; do not destroy the flowers while obtaining the nectar; and appear to be highly dependent on nectar as a food source during specific times of the year. The basal phylogenetic position of Ravenala in the family Strelitziaceae, as indicated by molecular sequence data, and the distribution of reproductive traits in the three extant genera are consistent with the hypothesis that pollination by nonflying mammals is an archaic system, whereas bird and bat

pollination are derived.

The origin and explosive diversification of flowering plants during the Cretaceous is often attributed to the simultaneous radiation of various groups of animals, es- pecially insects, which served as pollen dispersal agents (Crepet and Friis, 1987; Crepet, Friis, and Nixon, 199 1; but see Labandeira and Sepkowski, 1993). Vertebrates are thought to have played a negligible role in the early evo- lution of angiosperms because the main groups (e.g., birds and bats) that serve as important pollinators of extant plant taxa originated no earlier than the Eocene (Baker and Hurd, 1968). However, Sussman and Raven (1 978) suggested that nonflying mammals, such as primates and marsupials, could have been significant pollinators since the uppermost Cretaceous, but were outcompeted in the Tertiary by nectar-feeding birds and bats. Any coevolved relationships between flowering plant species and non- flying mammal pollinators that persist at the present would appear to be ". . . 'living fossils', which have a great deal

I Manuscript received 2 1 May 1993; revision accepted 23 December


The authors thank R. Sussman, D. Overdorff, L. Emmons, D. Baum,

A. Yoder, R. Hoffmann, L. Hirsch, F. Berkowitz, and L. Rakotovao for discussion, suggestions, sponsorship, and assistance with the logistics of this research; S. Love for invaluable help in the field; J. F. Smith and

E. A. Zimmer for collaborating on the molecular phylogenetic analysis;

E. Freeman for analysis of nectar sugar composition; and A. Tangerini for the pen and ink illustrations. This research was supported by grants from the Research Opportunity Funds and Scholarly Studies Program of the Smithsonian Institution to WJK, the W. Alton Jones Foundation to GES, and NSF grant BNS-8612154, National Geographic Society, and Wenner-Gren Foundation to HSM.

Author for correspondence. Current address: NYZS The Wildlife Conservation Society, Inter- national Programs, 185th and Southern Blvd., Bronx, NY 10460-1099.

to tell us about the evolution of both the mammals, in- cluding some of our own antecedents, and ofthe flowering plants" (Sussman and Raven, 1978).

Since the early work of Porsch (1934a, b, 1936), many nonflying mammals, which feed on nectar or flowers, have been suggested as possible pollinators of angiosperms, including rodents (Carpenter, 1978; Wiens and Rourke, 1978; Lumer, 1980; Recher, 198 l), genets (Lack, 1976), treeshrews (Nur, 1976; Faegri and van der Pijl, 1979), procyonids (Janson, Terborgh, and Emmons, 198 l), mar- supials (Vose, 1973; Rourke and Wiens, 1977; Wiens, Renfree, and Wooller, 1979; Steiner, 198 1 ;Turner, 1982; Goldingay, 1990; Goldingay, Carthew, and Whelan, 199 l), primates (Coe and Isaac, 1965; Tattersall and Sussman, 1975; Sussman and Tattersall, 1976; Hladik, Charles- Dominique, and Petter, 1980; Prance, 1980; Janson, Ter- borgh, and Emmons, 198 1 ;Torres de Assump~lo, 198 1 ; Garber, 1988; Overdorff, 1992), and even giraffes (du Toit, 1990). Most of these reports concentrate on the feeding behavior of the animals and provide only casual observations on the attributes of the plants; little data are presented to support the contention that the plants and animals have coevolved (Wiens et al., 1983). The best documented cases are those between Australian marsu- pials and the plant families Proteaceae and Myrtaceae (Rourke and Wiens, 1977; Wiens, Renfree, and Wooller,
1979; Hopper, 1980; Turner, 1982; Goldingay, 1990;
Goldingay, Carthew, and Whelan, 1991) and between
South African rodents and the genus Protea (Wiens and
Rourke, 1978).

Among the primates, several neotropical tamarins (Cal- litrichidae) and monkeys (Cebidae) visit the flowers of some rain forest trees and lianas (Prance, 1980; Janson, Terborgh, and Emmons, 1981; Torres de Assump~Bo,

198 1 ; Garber, 1988). Prosimians, including bush babies (Lorisidae; Coe and Isaac, 1965) in East Afrjca and lemurs (Cheirogaleidae and Lemuridae; Richard and Dewar, 199 1) in Madagascar, have also been suggested to be nectar- feeders. Jumelle and Perrier de la Bathie (1 9 10) were the first to report that lemurs visited flowers to take nectar and eat petals. Sussman and Raven (1978) listed six spe- cies of nocturnal lemurs that had been observed to feed regularly on flowers during some part of the year. Most of the reports of flower-visiting, however, showed that lemurs destroy the flowers during feeding (Charles-Dom- inique and Hladik, 197 1; Martin, 1972; Hladik and Charles-Dominique, 1974; Petter, 1978; Hladik, 1979) or did not provide conclusive evidence of pollination (Petter, 1962; Petter, Schilling, and Pariente, 197 1 ; Hladik, Charles-Dominique, and Petter, 1980). Until recently the most detailed report of pollination by lemurs (Tat- tersall and Sussman, 1975; Sussman and Tattersall, 1976) showed that Eulemur mongoz mongoz spent much of its feeding time licking nectar from the flowers of Ceiba pen- tundra, the kapok tree, and probably transported pollen between trees. Ironically the kapok tree is a plant recently introduced to Madagascar and hence not a candidate for coevolution with the native lemurs. Evaluation of the published evidence for actual pollination by the 12 taxa of lemurs reported to visit flowers reveals that only one new study provides sufficient documentation to suggest a specialized plant-pollinator relationship (~verdorff,


In this paper we present new data supporting an eco- logical relationship between a plant endemic to Mada- gascar (Ravenala madagascariensis) and an endemic, nec- tar-feeding lemur (Varecia variegata variegata). Our observations support Sussman and Raven's (1978) orig- inal hypotheses that 1) lemurs "play a significant role in the pollination of certain plant species in Madagascar," and 2) that this "association . . . is archaic, rather than recently derived." However, the evidence does not suggest a tightly "coevolved" plant-pollinator relationship.


We conducted two separate investigations of floral bi- ology and animal behavior in Madagascar. The primary investigation on the floral biology and flower visitors to Ravenala madagascariensis Sonn. (Strelitziaceae: Zingi- berales: Monocotyledonae) was carried out during the rainy season in February 1992, at which time we made observations at two sites in Madagascar. The principal site was along the eastern coast of Madagascar in primary wet forest on the island ofNosy Mangabe (1 5"30rS, 49'46%; Madagascar Special Reserve) in the Bay of Antongil near the town of Maroantsetra. At this site Ravenala is a rel- atively common forest species often found in light gaps in primary and old secondary growth. Our observations on Nosy Mangabe concentrated on a hillside population of 44 individual plants that had a total of 7 1 inflorescences. We specifically monitored visitation rates to four inflo- rescences on three plants during the day and night over a 6-d period (59 hr of total observation). A night scope (PS-910 STARSCOPE from Fujinon, Inc., Wayne, NJ) was used to observe nocturnal visitors to the flowers. More limited observations were made at a second site (2 1'20' 12"S,48'07'2 1 "E)in Madagascar in open savannah and secondary growth about 40 km inland from the coastal town of Mananjary where Ravenala is extremely abun- dant.

One of us (Morland, 199 1) conducted an additional 1 1 -mo field study between July 1987 and January 1989 that focused specifically on the social organization and ecology of the black and white ruffed lemur, Varecia var- iegata variegata (Lemuridae: Primates). The study in- cluded 1,560 hr of observations of 26 individual animals and was also carried out at Nosy Mangabe.

During the 1992 study, nectar was sampled from 14 continuously bagged flowers on three inflorescences with 5- or 10-ml syringes at 2 hr intervals for up to 48 hr after anthesis. The needle was inserted into the nectar chamber, which is formed by the bases of the three overlapping petals, and all nectar was extracted. Concentration was measured with a temperature compensated hand-held re- fractometer. For analysis of sugar constituents, nectars from 14 flowers (four samples/flower) were spotted on filter paper, allowed to air dry, and characterized using high-performance liquid chromatography (Freeman et al., 1984). The energy value of total mean nectar produced was calculated as M sucrose x volume in ml x 1.3 5 kcal/ ml (Bolten et al., 1979). The level of stigma receptivity was crudely estimated by the amount of tissue discolor- ation advancing from white (receptive) to brown (non- receptive). Phenological observations were made on a total of 53 flowers in 36 bracts on three inflorescences over the course of 4 d. Plant voucher specimens (W. J. Kress et al. 92-3402) are deposited at Parc de Tsimbazaza (Antananarivo; TAN), United States National Herbarium (Washington, DC, US), and Missouri Botanical Gardens (St. Louis, Mo).

In order to investigate the evolutionary origin of the pollination system of Ravenala and understand the di- versification of floral features in the family, a cladistic analysis ofthe three genera of the Strelitziaceae was carried out using a data set independent from reproductive char- acters. The other two genera of the family are Strelitzia, with three to five species endemic to southern Africa, and Phenakospermum, a monotypic genus found throughout tropical South America east of the Andes (Kress, 1990). DNA sequences from the chloroplast-encoded gene rbcL served as characters for a parsimony analysis that included both monotypic genera of the family and a single species of Strelitzia. This analysis of the Strelitziaceae was part of a larger study of the phylogenetic relationships of the Zingiberales that encompassed 18 genera and 2 1 species of the order plus proposed outgroups in the monocots (Smith, Kress, and Zimmer, 1993). The other seven fam- ilies of the Zingiberales, therefore, functioned as the out- group for ascertaining relationships within the Strelitzia- ceae. For details on materials and methods ofthe molecular investigation see Smith, Kress, and Zimmer (1 993).

Once the most parsimonious cladogram of the family was obtained, ten plant features specifically associated with pollination were mapped onto the cladogram (Table 2). Information on pollinators of the other two genera were taken primarily from Frost and Frost (1981) and Kress and Stone (1993). Polarization of the pollination characters to the out-group of the Strelitziaceae was not


Figs. 1-7. Vegetative and reproductive morphology of Ravenala madagascariensis. 1. Vegetative habit of single shoot. 2. Massive axillary inflorescence with congested, overlapping bracts, some with protruding open flowers; note position of ruffed lemur approaching flowers. 3. Cutaway view of inflorescence bract with floral bracts removed to show one open flower after anthesis (a) and second mature, but unopened flower (b). 4. Cut-away of entire flower with one-half of free sepal (c), one of two fused petals (d), and four of six anthers (e) removed. Note the stiff, rodlike style (f), nectar chamber formed by the overlapping bases of the three petals (g), and solid perianth tube with pollen tube canal (h) leading to ovary (i). 5. Cross section of ovary showing axillary attachment of two rows of ovules in each of the three locules. 6. Longitudinal segment of

possible because many of the in-group states are absent or variable in the out-group (Lowiaceae). The characters were therefore optimized on the cladogram following the principle of parsimony for the in-group only, i.e., char- acter state changes were mapped such that each trans- formation had the fewest possible number of steps min- imizing evolutionary reversals and parallelisms.


Theplant -Ravenala madagascariensis, often called the traveler's tree, is the only species in the genus and is common at elevations below 1,000 m primarily in regions of eastern Madagascar with significant rainfall (Perrier de la Bathie, 1946; Andrianifahanana, 1992). The national tree of Madagascar, "ravinala" (Malagasy common name literally meaning "leaf of the forest"), is a tall, single- stemmed or suckering tree with a palmlike trunk that holds the large bananalike leaves high in the forest canopy to heights of 30 m (Figs. 1, 8). It occurs in both primary rain forest and open secondary growth, the former being its probable natural habitat before the influence of humans allowed it to invade large tracts of disturbed forest and cleared areas.

The flowers of Ravenala are borne in large, bracteate inflorescences (to 85 cm in height) situated below the crown of bananalike leaves in the axils of the leaf bases (Fig. 2). The inflorescence includes from five to 15 strong, congested, protective bracts enclosing ten to 16 flowers each. A flower (ca. 19 cm in length; Figs. 3, 4, 9) consists of an inferior ovary (Fig. 5), perianth tube, three yellowish white sepals fused at their base to the perianth tube, three white petals of which two are fused along their margins, six pollen-bearing stamens (Fig. 6) enclosed by the two fused petals, and a stiff, rodlike style terminated by a broad ovoid stigma (Fig. 7). The anthers are held under tension inside the perianth, which remains closed at flower ma- turity and must be forcibly opened by animal visitors seeking nectar (Fig. 3). Anthers have dehisced at the time of flower opening (Fig. 6). Virgin flowers may be first opened by pollinators during the day or night and begin to oxidize and turn brown 24-36 hr after anthesis. About 72 hr after anthesis an abscission layer has formed just above the ovary and the perianth with adnate anthers can be easily detached from the receptacle. No strong odors are produced by the flowers; however, a slight fragrance is detectable up to 30 cm from the flower.

The overlapping bases of the sepals and petals form a large nectar chamber (ca. 6.75 cm3) inside the flower (Fig. 4). At anthesis this chamber is filled with copious, sucrose- dominant nectar with a concentration ofabout 14% (Table 1). Over 60% of the nectar produced by a flower is present when the flower is first opened. After anthesis, nectar is c~ntinuously secreted into this chamber for about 24 hr (X = 23.57 + 7.65 hr; range 6-32 hr; N = 14) during which time the rate of production and concentration de- creases. When nectar production ceases the stigma ap- pears to be no longer receptive.

From the flower production rates that we observed in the field, we estimate that a typical inflorescence (with 13 bracts each containing 14 flowers; Figs. 1-3) will produce an average of 5.2 open flowers per day for 5 wk. This rate of flower production for a single inflorescence is equiv- alent to providing 2 1.7 kcal of energy per day (4.17 kcal/ flower x 5.2 flowerdday) for a total of 759 kcal over the life of the inflorescence. Each plant produces one to two inflorescences per season (2= 1.6 inflorescences/plant; N = 44 plants). Because of staggered flowering times of individuals, a population can be in flower for 2 to 3 mo of the year, thereby offering a rich energy reward to floral visitors for a considerable period of time.

Animal visitors -During our observations in 1992 on Nosy Mangabe three species of lemur were seen visiting the flowers. Eulemur fulvus albifrons, the white-fronted lemur, and Cheirogaleus major, the greater dwarf lemur, were only occasional visitors during the day and night, respectively. The black and white ruffed lemur, Varecia variegata variegata, was the primary and dominant ani- mal observed to repeatedly visit the flowers between dawn and sunset (Figs. 2, 10, 11). We never say Varecia visit the flowers at night. During the 40 hr of diurnal obser- vation, individuals of Varecia made 57 visits to the four inflorescences under watch (an average of 1.33 visits/ inflorescence/hr) for a total of 333 min of visitation. De- pending on the number of inflorescences and open flowers present, lemurs spent up to 7 min on an individual plant. Visits to one plant with two inflorescences lasted 5.35 min on the average (N = 37 visits), whereas visits to the two plants with a single inflorescence each lasted 3.57 min (N = 30 visits) and 1.47 min (N = 19 visits) on average. In general lemurs were observed to forage alone when visiting flowers of Ravenala. In two rare instances when two individuals approached the same inflorescence, ag- gressive encounters occurred between them.

Ruffed lemurs are 3-4 kg strepsirhines found only in the eastern rain forests of Madagascar (Harcourt and Thornback, 1990). The 1987-1 989 study showed that their diet consisted mainly of fruit and nectar supple- mented by leaves and other items from at least 67 plant species (Morland, 1991). Nectar was the second most important food item in the yearly diet, and on the average accounted for about 22% of feeding time. Nectar was not available to lemurs throughout the year, but when it was produced, it appeared to be their dominant food. Two species provided the main supply of nectar: Ravenala madagascariensis andLabramia costata (Hartog ex Bail- lon) AubrCville (Sapotaceae). When the latter was in flow- er, Varecia spent more time (74% of feeding records) taking nectar from these flowers than feeding on the fruits or leaves of any other plant species. Complete data on the dietary importance of Ravenala nectar for Varecia are not available because the total flowering season of Ravenala was not monitored during this I 1 -mo study nor during the 1992 observations. However, the consistent and attentive behavior of Varecia at the flowers oflabra

one of six anthers with two lateral sacs containing pollen at anthesis. 7. Stigma with fingerlike appendages surrounding the receptive area. Note position of pollen in receptive area.

Figs. 8-1 1. Habitat, morphology, and pollinator visitation to flowers ofRavenala madagascariensis. 8. Overall aspect of plant bordering primary forest west of Irondro, Madagascar (2lo23'19"S, 47O56'53"E; voucher: Kress et al. 92-3177). Note distichous leaf arrangement and position of inflorescences below leaf crown in axils of leaf bases. 9. Part of inflorescence of Ravenala with bract cut away to show large open flower at anthesis.

10. A black and white ruffed lemur (Varecia variegata variegata) arriving at inflorescence of Ravenala in primary forest at Nosy Mangabe. 11.The same lemur taking nectar from flowers of Ravenala with snout thrust inside of open perianth.

rnia and Ravenala, the pointed snout and long tongue, and the apparent dependence on nectar as a significant part of the diet all suggest that this frugivorous lemur is also specialized for flower visiting. Unfortunately no data are available on daily average caloric dietary requirements for ruffed lemurs or any related species.

During foraging, ruffed lemurs, which are totally ar- boreal animals, approached the inflorescences of Ravenala from the middle and upper branches of neighboring trees (Figs. 2, 10). They quickly found unopened or pre- viously opened flowers that were producing nectar. To open a newly emerging flower the lemur grasped the un- opened perianth with its teeth and roughly pulled it from the protective inflorescence bract, but did not break it OK This action sprang the perianth open, thus releasing the reflexing anthers that brush pollen onto the muzzle and head of the animal. While holding onto other bracts of the inflorescence with its hindfeet, the lemur pulled the lateral sepals apart with one or both forefeet to allow access to the nectar chamber. The snout was then thrust into the center of the flower and nectar was extracted with the tongue (Fig. 11). The lemur contacted both the stamens and stigma while feeding. We also saw the lemurs lick pollen directly from the anthers with their tongues and groom pollen off their fur. We never saw lemurs destroy the flowers they were visiting. The animals invariably visited all open flowers in an inflorescence and frequently moved between inflorescences on the same plant (33 of 37 observed visits = 89%) and between plants (20 of 49 observed visits = 4 1%) in the observation area. Although

TABLE 1. Summary of nectar analysis of Ravenala madagascariensis.

Sampling time"

Nectar At anthesisyrangi) Total' (range)

Amount (N = 14 flowers) 5.4 + 2.3 ml 8.6 + 2.7 ml

(1.9-9.0 ml) (3.9-12.8 ml) Concentration (sucrose 14.2 + 0.95% 11.8 + 1.2%

equivalents; N = 13 flow- (1 1.5-1 5.5%) (9.6-12.8%)

ers) Sugar composition (S/(G + 1.50 + 0.2 13

F); N = 14 flowers) (1.07-1.78)



Values in columns 2 and 3 are means and standard deviations.

Values for nectar present when flower first opens.

Values for total accumulated nectar sampled over a period of 48 hr following anthesis.

it was logistically impossible during this study to actually quantify the amount of pollen carried by the lemurs, the large amount of pollen observed on the fur and the move- ment of the animals suggest that pollen is transferred over significant distances between plants.

We saw no other vertebrates or invertebrates frequently visit the flowers of Ravenala on Nosy Mangabe. Both pteropodid bats (Pteropodidae) and sunbirds (Nectari- niidae) were observed or have been reported from the area (Jenkins, 1987; Langrand, 1990); however, only the latter occasionally visit these flowers. At the second study site we visited in 1992 near Mananjary, Ravenala is com- mon in open secondary growth, and we observed mature fertile fruits on many plants. However, lemurs are scarce in these open areas and we observed none visiting the flowers. Moreover, we saw no bats at the flowers during two evenings of observations made in this area during the typical early foraging period ofpteropodid bats (1 750- 2200 hr). In one case a single sunbird (not identified to species) was seen probing a flower of Ravenala, but did not contact the anthers or stigma.

Phylogenetic relationships and character evolution -The results of the cladistic parsimony analysis support the monophyly of the family Strelitziaceae in the Zingiberales; 12 apomorphic base pair mutations in the rbcLgene define this clade. Within the family, Ravenala is the most basal lineage; Strelitzia and Phenakospermum are terminal, sis- ter-taxa (Fig. 12). The branch including the latter two genera is supported by nine nucleotide substitutions. The

Madagascar lemurs South America phyllostomid bats South Africa sunbirds
  1. terminal Inflorescence 3. bright colors
  2. large flowers >30cm 4, pollen connected by
  6. nocturnal anthesis cellulosic threads
  8. non-replenishable nectar 5. non-explosive anthesls
  10. weak styles 7. flower longevity >48 hrs

9. hexosedominant nectar


1, axillary inflorescence 2, small flowers <20 cm

3. dull colors 4, pollen in monads 5, explosive anthesis

6. diurnal anthesis 7.flower longevity <24 hrs

8. replenishable nectar 9.sucrosedominant nectar

10. stiff styles

Fig. 12. Cladogram representing the phylogenetic relationships of the three genera of the Strelitziaceae. Tree derived from maximum parsimony analysis of sequence data of the chloroplast rbcL gene; the outgroup was 18 additional taxa of the order Zingiberales (for further details of the analysis see Smith, Kress, and Zimmer, 1993). Numbers in parentheses along branches indicate nucleotide mutations supporting each clade. Geographic distributions and primary pollinators for each genus are given under the generic names. Character state transformations for critical plant features associated with the pollination system of each genus are mapped onto the cladogram for maximum parsimony (see Table 2). Sucrose-dominant nectar (Character #9) can be placed equally parsimoniously in the Ravenala lineage.

distinctiveness (and possible ancient age) of each of the genera is emphasized by the high number of substitutions characterizing each of the three lineages: 23, 37, and 51 substitutions for Ravenala, Strelitzia, and Phenakospermum, respectively.

TABLE 2. Taxon-character matrix of pollination features of the Strelitziaceae optimized on the rbcL-based cladogram (see Fig. 12).


  1. Inflorescence position
  2. Flower size
  3. Color of flowers and bracts
  4. Pollen
  5. Anthesis mechanism
  6. Time of anthesis
  7. Flower longevity
  8. Nectar production
  9. Nectar sugars
  10. Styles


Axillary Small (<20 cm) Dull (green, yellow, and

white) Monads

Explosive Diurnal 524 hr Replenishable Sucrose-dominant Stiff and rodlike



Terminal Large (>30 cm) Dull (green, yellow, and

white) Monads

Explosive Nocturnal 524 hr Nonreplenishable Hexose-dominant Weak and flaccid


Axillary Small (<20 cm) Bright (orange, blue, and

white) Connected by cellulosic

threads Nonexplosive Diurnal 248 hr Replenishable Hexose-dominant Stiff and rodlike

Figs. 13, 14. Distribution maps of Ravenala madagascariensis and Varecia variegata in Madagascar. 13.Ravenala (from Andrianifahanana, 1992). 14. Varecia (from Harcourt and Thornback, 1990).

With respect to the distribution ofreproductive features on the cladogram of the family, Strelitzia and Phenako- spermum are characterized by several autapomorphies associated with bird and bat pollination (Fig. 12). On the other hand the combination of reproductive traits found in lemur-pollinated Ravenala can be explained most par- simoniously as originating in the common ancestor of the family (Fig. 12), i.e., features of the pollination system in Ravenala are plesiomorphic and not uniquely derived in this lineage.


Pollination by lemurs-It is unquestionably difficult to prove that an animal is actually pollinating a plant, i.e., transporting viable pollen between flowers that results in successful fertilization and seed production (Cox and Knox, 1988). Evidence at best is usually indirect, especially for many arboreal taxa inhabiting remote tropical environ- ments. In the case ofRavenala we have shown that mem- bers of at least one species of lemur, Varecia variegata: 1) are consistent and possibly the only visitors to the flowers; 2) visibly carry pollen on their fur between flowers on the same plant as well as between flowers of conspecific plants; 3) do not destroy the flowers while obtaining the nectar; and 4) appear to be highly dependent on its nectar as a food source during specific times of the year. Fur- thermore the flowers themselves possess many obvious specializations for visitation by large, nonflying animals, such as: 1) inflorescences placed below the crown of the plant and hence more easily accessible to arboreal than to flying animals; 2) large flowers enclosed in tough, pro- tective bracts that require manual manipulation by a strong pollinator to be opened; 3) stiff, rodlike styles that with- stand the rough handling of the visitors; and 4) copious, sucrose-dominant nectar that provides an energy-rich,

renewable reward for a sizable animal for a 2-to 3-mo time period. This evidence coupled with the fact that we never observed significant visitation by animals other than lemurs during our study strongly supports the hy- pothesis that this plant species endemic to Madagascar has evolved with an endemic group of nonflying animals, the lemurs, as its primary pollinator.

Because the natural distribution of Varecia along the eastern coast of Madagascar (Harcourt and Thornback, 1990) is nearly equivalent to that of Ravenala (Figs. 13, 14), these lemurs may be common visitors to the flowers of this species throughout most of its range. However, in addition to Varecia variegata, we observed potential pol- lination by the diurnal Eulemurfulvusand nocturnal Chei- rogaleus major on Nosy Mangabe. These latter two genera of lemurs have previously been reported to visit flowers of other plant taxa (Tattersall and Sussman, 1975; Hladik, 1979; Hladik, Charles-Dominique, and Petter, 1980; Overdorff, 1992). The only other common lemur on Nosy Mangabe, the aye-aye (Daubentonia madagascariensis), has also been observed visiting the flowers of Ravenala at this site (E. Sterling, Yale University, personal com- munication). However, it is not known if aye-ayes are visiting the flowers for nectar or seeking insect larvae hidden within the bracts.

Our study was concentrated primarily at a single for- ested site in northeastern Madagascar and may not reflect the full extent of interactions which occur between Rav- enala and other floral visitors at other localities. We sus- pect that other frugivorous lemurs may serve as polli- nators of Ravenala in regions of Madagascar where Varecia is absent. For example, Eulemur macaco macaco, the Black Lemur, has been reported to visit the flowers of this plant in a disjunct population of Ravenala in north- western Madagascar (J. Andrews, Washington University, and I. Colquhoun, University of Western Ontario, per- sonal communications). At Ranomafana National Park in Madagascar south of the Nosy Mangabe site, Overdorff (1 992) provided documentation on pollination of several plant species by red-bellied lemurs (Eulemur rubriventer). She observed that this lemur preferred fruit, but spent a significant amount of time feeding on flowers. Although flowers of most of the species were destroyed during feed- ing, flowers were left intact in at least two plant species, and pollen was seen to be transferred between plants on the faces of the lemurs suggesting legitimate pollination. Unfortunately, Ravenala, which is a component of the primary forest at Ranomafana and may potentially be visited by red-bellied lemurs at this site, was not included in her study.

In sum, observations to date on flower visitation by lemurs to Ravenala and other plant taxa suggest that some plants may be specialized for pollination by nonflying mammals as hypothesized by Sussman and Raven (1 978), but the systems are not highly specific ones. The lemurs do not appear to be dependent on any one nectar source for food, and although some plant species, such as R. madagascariensis, may depend on lemurs for pollen trans- fer, the plant-animal relationships are not strict one-to- one species interactions. As is true of many animal pol- lination systems, flower visiting by lemurs probably arose through diffuse coevolution rather than close reciprocal evolution. Quantitative investigations of the extent of pollen transfer and successful fertilization following pol- lination of Ravenala by lemurs are planned.

Other flower visitors-The scarcity of lemurs in open secondary growth areas where Ravenala is very abundant and obviously reproducing argues for the presence of other floral visitors or pollination mechanisms. Of the limited number of nectar-feeding birds and bats that are known from Madagascar (Sussman and Raven, 1978; Jenkins, 1987; Langrand, 1990), several have been implicated as pollinators ofRavenala.Scott-Elliot (1 890), who provided a detailed description of the floral morphology and ex- plosive dehiscence of Ravenala, believed that sunbirds, which he observed visiting the flowers near Fort-Dauphin (Taolanaro), were important pollinators. Fruit bats in the genus Pteropus have been reported to visit (eat?) the flow- ers of Ravenala in Mauritius and Rkunion (Cheke and Dahl, 198 1) where this plant has been introduced and is now quite abundant in secondary growth (Marais, 1983). Visitation by pteropine bats has also been observed in northern Queensland (W. J. Kress, unpublished data) and Darwin (Calley, Braithwaite, and Ladd, 1993), Australia, where Ravenala is often cultivated as an ornamental plant. We believe that open, highly disturbed areas in Mada- gascar do not represent the natural habitat of Ravenala, and although our observation time was limited, we saw very few flower visitors in these regions. Even if fruit bats and sunbirds occasionally visit the flowers in these hab- itats, we believe they can only be considered secondary or facultative pollinators.

It is more likely that Ravenala is primarily self-polli- nating in such disturbed environments in Madagascar as well as in other parts of the world where it has been introduced. The proximity of the dehisced anthers and the stigma in mature, but unopened flowers would facil- itate autogamy. Calley, Braithwaite, and Ladd (1 993) have shown that in cultivated plants in Australia Ravenala is at least a facultative selfer even though autogamously produced fruits contain few seeds. Isozyme studies are currently underway to compare the amount and distri- bution of genetic variation present in populations of Ravenala in primary and secondary habitats in Madagascar (Kress, Andrianifahanana, and Roesel, unpublished data). Results from this genetic survey will provide comparative data on the levels of self-pollination and in-breeding in the contrasting environments occupied by these plants.

The origin of lemur pollination-A main point of the Sussman and Raven (1978) hypothesis was that these nonflying mammals represent archaic coevolved plant- pollinator systems that have only persisted in areas of the world where more recently evolved, volant floral visitors are absent or less prominent (e.g., Madagascar with only three species of birds and three species of bats that are known to feed on nectar; Jenkins, 1987; Langrand, 1990). Although lemur pollination of Ravenala does not appear to be a tightly coevolved relationship, the pollination system may be ancient, at least in the Strelitziaceae. Ravenala is the only member of the family that has six fertile anthers (a plesiomorphic state for the order), suggesting that this genus is the most primitive of the three genera of the family (Tomlinson, 1962; Kress, 1990). The results of the phylogenetic analysis based on gene sequence data (Smith, Kress, and Zimmer, 1993) also indicate that Ravenala is the most basal lineage in the family; Strelitzia and Phenakospermum are derived (Fig. 12). Although no fossils of the Strelitziaceae have been reported, taxa rep- resenting some of the more advanced (e.g., Zingiberaceae) as well as the more primitive (e.g., Musaceae) families of the order Zingiberales are known from the late Cretaceous and early Tertiary (Kress, 1990; Manchester and Kress, 1993). It is therefore likely that the Strelitziaceae, a prim- itive family closely related to the Musaceae, had also originated during the Cretaceous.

Studies on the reproductive biology of the other two genera of the Strelitziaceae have shown that Strelitzia is pollinated by sunbirds (Scott-Elliot, 1890; Darwin, 1895; Rowan, 1974; Skead, 1975; Frost and Frost, 198 1) and Phenakospermum by phyllostomid bats (Kress and Stone, 1993). Many of the plant characters associated with these pollination systems appear from their position on the cladogram to be uniquely derived (Fig. 12). For example, in the genus Strelitzia, pollination by sunbirds corre- sponds with the evolution of 1) brightly colored flowers which serve as attractants, 2) pollen grains connected by cellulosic threads, 3) a nonexplosive type of flower open- ing that allows deposition of the aggregated pollen grains on the feet of the avian visitors that perch on the petals and stigmas, and 4) long-lasting flowers that produce nec- tar and reward pollinators for up to 3 days (Frost and Frost, 198 1). The bat-pollinated Phenakospermum has 1) large terminal inflorescences held high above the leaves and easily accessible to volant mammals, 2) larger flowers, 3) a single "dose" of copious nectar available only at the time of flower opening, 4) nocturnal anthesis, and 5) weak styles that are placed out of the path of the pollinators after the first visit to the flower (Kress and Stone, 1993). Each of these two genera is characterized by autapomor- phies (i.e., specializations) for pollination by particular pollen vectors.

In contrast, the basal lineage of the family composed of the monotypic Ravenala is distinctive in its lack of apomorphic character states associated with pollination. Optimization of reproductive features on the three-taxon cladogram places most of the states found in modem Ravenala, and hence states characterizing lemur polli- nation, in the common ancestor of the family at the base of the tree (Fig. 12). Features such as 1) axillary inflores- cences, 2) relatively inconspicuous bract and flower color, 3) pollen in monads, 4) explosive flower opening, 5) di- urnal anthesis, 6) 24-hr flower longevity, 7) replenishable nectar, and 8) stiff, durable styles are primitive character states for the Strelitziaceae. Sucrose-dominant nectar, the only pollination-related state unique to Ravenala (Fig. 12), could be placed equally parsimoniously at the base of the tree or in the Ravenala lineage. Although nectar has been analyzed in only a very few plant species pol- linated by nonflying mammals, "a trend towards sucrose- richness" has been reported (Baker and Baker, 1983).

The above evidence indicates that the reproductive traits associated with lemur pollination in Ravenala originated in the ancestral taxon of the family. The plesiomorphic traits have been retained in Ravenala while more spe- cialized pollination features have evolved in the two de- rived genera Strelitzia and Phenakospermum. Although evidence is meagre, the primate lineage containing lemurs most likely diversified from East African relatives in the early to mid-Tertiary (Szalay and Delson, 1979; Tatter- sall, 1982; Savage and Russell, 1983), which is roughly equivalent to the estimate ofthe origin ofthe Strelitziaceae (see above). We therefore suggest that some ancestor of the lemurs or other early mammal lineage present in Af- rica during the late Cretaceous or early Tertiary may have been the visitors to flowers of primitive members of the earliest lineage of the family. Although modern Ravenala appears to have a distinct ecological association with pres- ent-day lemurs as pollinators, the actual features of the plants that are specialized for pollinationmay have evolved to ensure flower visiting by an ancient ancestor of the lemurs, if not the earliest lemurs themselves.

Our data, although not supportive of a tightly coevolved system, are consistent with the hypothesis of Sussman and Raven (1 978) that pollination by nonflying mammals is archaic (at least in this family of flowering plants). In the Strelitziaceae, pollination by nonflying mammals, now represented in Madagascar by lemur pollination of Rav- enala, appears to have persisted in an isolated geographic region that today remains depauperate in significant flying vertebrate floral visitors. The two more advanced lineages ofthe family radiated into areas where birds (i.e., Strelitzia in Africa with over 70 species of nectar feeding birds; Clements, 199 1) and bats (i.e., Phenakospermum in South America with over 30 species of nectar feeding bats; Do- bat, 1985) evolved as pollinators of plants. The obser- vations that bush babies (Galago: Lorisidae: Primates) and vervet monkeys (Cercopithecus: Cercopithecidae: Pri- mates) occasionally visit the flowers of Strelitzia (Frost and Frost, 198 1) and that woolly opossums (Caluromys: Didelphidae: Marsupialia) are nectar robbers on Phena- kospermum (Kress and Stone, 1993) may reflect the be- havior ofthe ancestral, nonflying pollinators ofthese more derived taxa.


ANDRIANIFAHANANA, 1992. Contribution a l'etude biologique (sys-

M. tCmatique+cologie) de Ravenala madagascariensis Adans. (Stre- litziaceae) dans la region de Mananara-nord. Memoire de Diplomes #Etudes Approfondies des Sciences Biologiques Appliquees, Uni- versite d'Antananarivo, Antananarivo, Madagascar.

BAKER, H. G., AND I. BAKER. 1983. Floral nectar sugar constituents in relation to pollinator type. In C. E. Jones and R. J. Little [eds.], Handbook of experimental pollination biology, 1 17-14 1. Van Nos- trand Reinhold, New York, NY.

,AND P. H. HURD, JR. 1968. Intrafloral ecology. AnnualReview ofEntomology 13: 385-414. BOLTEN,A. B., P. FEINSINGER, H. G. BAKER, AND I. BAKER. 1979. On the calculation of sugar concentration in flower nectar. Oecologia

41: 301-304. CALLEY,M., R. W. BRAITHWAITE, AND P. G. LADD. 1993. Reproductive

biology of Ravenala madagascariensis Gmel. as an alien species.

Biotropica 25: 61-72. CARPENTER,F. L. 1978. Hooks for mammal pollination? Oecologia

35: 123-132. CHARLES-DOMINIQUE, Le Lepilemur du

P., AND C. M. HLADIK. 1971.

sud de Madagascar: ecologie, alimentation et vie sociale. La terre

et la vie, revue d'histoire naturelle, Paris 25: 3-66. CHEKE, A. S., AND J. F. DAHL. 1981. The status of bats on western

Indian Ocean islands, with special reference to Pteropus. Mammalia

45: 205-238. CLEMENTS,J. F. 1991. Birds of the world: a check list. Ibis, Vista, CA. COE, M. J., AND F. M. ISAAC. 1965. Pollination of the baobab (Adan

sonia digitata L.) by the lesser bush baby (Galago crassicaudatus

E. Geoffroy). East African Wildlife Journal 3: 123-124.

Cox, P. A., AND R. B. KNOX. 1988. Pollination postulates and two- dimensional pollination in hydrophilous manocotyledons. Annals of the Missouri Botanical Garden 75: 8 1 1-8 18.

CREPET,W. L., AND E. M. FRIIS. 1987. The evolution of insect pol- lination in the angiosperms. In E. M. Friis, W. G. Chaloner, and

P. R. Crane [eds.], The origins of angiosperms and their biological consequences, 18 1-20 1. Cambridge University Press, Cambridge.

--,AND K. C. NIXON. 199 1. Fossil evidence for the evo- lution ofbiotic pollination. Philosophical Transactions ofthe Royal Society, London, Series B 333: 187-1 95.

DARWIN, C. 1895. The effects of cross and self fertilization in the vegetable kingdom. D. Appleton, New York, NY. DOBAT,K. 1985. Bliiten und Fledermause. W. Kramer, Frankfurt am Main, Germany. DU TOIT,J. T. 1990. Giraffe feeding on Acacia flowers: predation or pollination? African Journal of Ecology 28: 63-68. FAEGRI,K., AND L. VAN DER PUL. 1979. The principles of pollination ecology, 3d ed. Pergamon Press, Oxford.

FREEMAN,C. E., W. H. REID, J. E. BECVAR, AND R. SCOGIN. 1984. Similarity and apparent convergence in the nectar-sugar compo- sition of some hummingbird-pollinated flowers. Botanical Gazette

145: 132-135. FROST, S. K., AND P. G. FROST. 198 1. Sunbird pollination of Strelitzia nicolai. Oecologia 49: 379-384.

GARBER,P. A. 1988. Foraging decisions during nectar feeding by tam- arin monkeys (Sanguinus mystax and Sanguinus fuscicollis, Callitrichidae, Primates) in Amazonian Peru. Biotropica 20: 100-106.

GOLDINGAY,R. L. 1990. The foraging behaviour of a nectar feeding marsupial, Petaurus australis. Oecologia 85: 19 1-1 99. ,S. M. CARTHEW, AND R. J. WHELAN. 1991. The importance of non-flying mammals in pollination. Oikos 61: 79-87.

HARCOURT,C., AND J. THORNBACK. 1990. Lemurs of Madagascar and the Comoros. The IUCN Red Data Book. IUCN, Gland, Switzer- land.

HLADIK, C. M. 1979. Diet and ecology of prosimians. In G. A. Doyle and R. D. Martin [eds.], The study of prosimian behavior, 307-

358. Academic Press, London.

1974. The behaviourand ecology ofthe sportive lemur (Lepilemur mustelinus) in relation to its dietary peculiarities. In G. A. Doyle, R. D. Martin, and A. C. Walker [eds.], Prosimian biology, 23-37. Duckworth, London.

--, AND J. J. PETTER. 1980. Feeding strategies of five nocturnal prosimians in the dry forest of the west coast of Mada- gascar. In P. Charles-Dominique, H. M. Cooper, A. Hladik, C. M. Hladik, E. Pages, G. F. Pariente, A. Petter-Rousseaux, J. J. Petter, and A. Schilling [eds.], Nocturnal malagasy prosimians, 41-73. Academic Press, New York, NY.

HOPPER, S. D. 1980. Bird and mammal pollen vectors in Banksia communities at Cheyne Beach, Western Australia. Australian Jour- nal of Botany 28: 61-75. JANSON,C. H., J. TERBORGH, AND L. H. EMMONS. 1981. Non-flying mammals as pollinating agents in the Amazonian forest. Biotropica

14: 1-6 (Supplement). JENKINS, M. D. [ed.]. 1987. Madagascar, an environmental profile. IUCN, Gland, Switzerland.

JUMELLE,H., AND H. PERRIERDE LA BATHIE. 19 10. Fragments biologi- ques de la flore de Madagascar. Annales du Music Colonial de Marseille 8: 452-454 (2strie).

KRESS, W. J. 1990. The phylogeny and classification ofthe Zingiberales. Annals ofthe Missouri Botanical Garden 77: 698-721.

, AND D. E. STONE. 1993. Morphology and floral biology of Phenakospermum (Strelitziaceae), an arborescent herb of the neo- tropics. Biotropica 25: 290-300.


C. C., AND J. J. SEPKOWSKI, 1993. Insect diversity in the fossil record. Science 26 1: 3 10-3 15. LACK, A. 1976. Genets feeding on nectar of Maranthes polyandra in northern Ghana. East African Wildlife Journal 15: 233-234. LANGRAND,0. 1990. Guide to the birds of Madagascar. Yale Uni- versity Press, New Haven, CT.

LUMER,C. 1980. Rodent pollination of Blakea (Melastomataceae) in a Costa Rican cloud forest. Brittonia 32: 5 12-5 17.

MANCHESTER, Fossil bananas (Musa-

S. R., AND W. J. KRESS. 1993. ceae): Ensete oregonense sp. nov. from the Eocene of western North America and its phytogeographic significance. American Journal of Botany 80: 1264-1272.

MARAIS, W. 1983. Musaceae. Flore des Mascareignes, vol. 174, 1-7. The Sugar Industry Research Institute, Mauritius. MARTIN, R. D. 1972. A preliminary field-study of the Lesser Mouse Lemur (Microcebus murinus J. F. Miller 1977). In P. Charles-Dom- inique and R. D. Martin [eds.], Behaviour and ecology of nocturnal prosimians, 43-89. Paul Parey, Berlin. MORLAND, H. S. 199 1. Social organization and ecology of Black and White Ruffed Lemurs (Varecia variegata variegata) in lowland rain forest, Nosy Mangabe, Madagascar. Doctoral dissertation, Yale University. New Haven, CT.

NUR, N. 1976. Studies on pollination in Musaceae. Annals of Botany

40: 167-177.

OVERDORFF, D. J. 1992. Differential patterns in flower feeding by Eulemur fulvus fulvus and Eulemur rubriventer in Madagascar. American Journal ofPrimatology 28: 191-196.

PERRIERDE LA BATHIE,H. 1946. Musackes. In H. Humbert [ed.], Flore de Madagascar et des Comores, 1-6. Imprimerie Officielle, Tanan- arive, Madagascar.

PETTER, J. J. 1962. Ecological and behavioral studies of Madagascar lemurs in the field. Annals of the New York Academy of Science

102: 267-281.

. 1978. Ecological and physiological adaptations of five sym- patric nocturnal lemurs to seasonal variations in food production. In D. J. Chivers and J. Herbert [eds.], Recent advances in pri- matology, vol. 1, Behaviour, 21 1-223. Academic Press, New York, NY.

, A. SCHILLING, AND G. PARIENTE. 197 1. Observations eco- ethologiques sur deux lemuriens Malgaches nocturnes: Phanerfurcifer et Microcebus coquereli. La terre et la vie, revue d'histoire naturelle, Paris 25: 287-327.

PORSCH, 0. 1934a. Saugetiere als Blumenausbeuter und die Frage de Saugetierblume. I. Biologia Generalis 10: 657-685. . 1934b. Saugetiere als Blumenausbeuter und die Frage de Sau- getierblume. 11. Biologia Generalis l l: 17 1-1 88. . 1936. Saugetierblumen. Forschungen und Fortschritte 12: 207. PRANCE,G. T. 1980. A note on the probable pollination of Combretum by Cebus monkeys. Biotropica 12: 239.

RECHER, H. F. 198 1. Nectar feeding and its evolution among verte- brates. In A. Keast [ed.], Ecological biogeography of Australia, 1639-1648. W. Junk, The Hague.

RICHARD, A. F., AND R. E. DEWAR. 1991. Lemur ecology. Annual Review of Ecology and Systematics 22: 145-175.

ROURKE,J., AND D. WIENS. 1977. Convergent floral evolution in South African and Australian Proteaceae and its possible bearing on pol- lination by non-flying mammals. Annals of the Missouri Botanical Garden 64: 1-17.

ROWAN, M. K. 1974. Bird pollination of Strelitzia. Ostrich 45: 40. SAVAGE,D. E., AND D. E. RUSSELL. 1983. Mammalian paleofaunas of the world. Addison-Wesley, Reading, MA. SCOTT-ELLIOT, 1890. Note on the fertilisation of Musa, Strelitzia

G. F. reginae, and Ravenala madagascariensis. Annals ofBotany 4: 259-

263. SKEAD, C. J. 1975. Weaverbird pollination ofStrelitzia reginae. Ostrich

46: 183-185.

SMITH, J. F., W. J. KRESS, AND E. A. ZIMMER. 1993. Phylogenetic analysis of the Zingiberales based on rbcL sequences. Annals of the Missouri Botanical Garden 80: 620-630.

STEINER,K. E. 198 1. Nectarivory and potential pollination by a neo- tropical marsupial. Annals of the Missouri Botanical Garden 68: 505-5 13.

SUSSMAN,R. W., AND P. H. RAVEN. 1978. Pollination by lemurs and marsupials: an archaic coevolutionary system. Science 200: 731-

736. ,AND I. TA~ERSALL. Cycles of activity, group compo-

1976. sition, and diet of Lemur mongoz mongoz Linnaeus 1776 in Mad- agascar. Folia Primatology 26: 270-283.

SZALAY, F. S., AND E. DELSON. 1979. Evolutionary history of the primates. Academic Press, New York, NY.

TATTERSALL,I. 1982. The primates of Madagascar. Columbia Uni- versity Press, New York, NY. ,AND R. W. SUSSMAN. 1975. Obsemations on the ecology and behavior of the Mongoose Lemur Lemur mongoz mongoz Linnaeus (Primates, Lemuriformes), at Ampijoroa, Madagascar. Anthropological Papers of the American Museum ofNatural History 52: 193-

216. TOMLINSON, 1962. Phylogeny ofthe Scitamineae-morphological

P. B.

and anatomical considerations. Evolution 16: 192-2 13. TORRES DE ASSUM~AO, I98 I. Cebus apella and Brachyteles ar-

C. achnoides (Ceibidae) as potential pollinators of Mabea ,fistulifera (Euphorbiaceae). Journal of Mammalogy 62: 386-388.

TURNER, V. 1982. Marsupials as pollinators in Australia. In J. A. Armstrong, J. M. Powell, and A. J. Richards [eds.], Pollination and evolution, 55-66. Royal Botanic Gardens, Sydney.

VOSE, H. 1973. Feeding habits of the Western Australian honey pos- sum, Tarsipes spenserae. Journal of Mammalogy 54: 245-247.

WIENS, D., M. RENFREE, AND R. 0. WOOLLER. 1979. Pollen loads of honey possums (Tarsipes spenserae) and non-flying mammal pol- lination in Southwestern Australia. Annals oftheMissouri Botanical Garden 66: 830-838.

,AND J. P. ROURKE. 1978. Rodent pollination in southern Af- rican Protea spp. Nature 276: 71-73.

--, B. B. CASPER, E. A. RICKART, T. R. LAPINE, C. J. PETERSON,AND A. CHANNING. 1983. Nonflying mammal polli- nation of southern African proteas: a non-coevolved system. Annals ofthe Missouri Botanical Garden 70: 1-31,

  • Recommend Us