Pollination of Cyphomandra endopogon var. endopogon by Eufriesea spp. (Euglossini) in French Guiana

** Cyphomandra endopogon var. endopogon has since been determined by specialist Lynn Bohs to be Cyphomandra endopogon Bitter subsp. guianensis Bohs


Carol Gracie
piclecpo.jpg - 4328 Bytes

Gracie, Carol (New York Botanical Garden, Bronx, NY 10458-5126). Pollination of Cyphomandra endopogon var. endopogon (Solanaceae) by Eufriesea spp. (Euglossini) in French Guiana. Brittonia 45(1):39-46. 1993.--The pollination systems of two closely related genera, Cyphomandra and Solanum, differ in rewards offered and pollinator activity. I suggest that the anther differences which define these two genera have evolved in response to the different pollination systems. Male euglossine bees of the genus Eufriesea were observed to pollinate the flowers of Cyphomandra endopogon var. endopogon while gathering aromatic compounds from the anther connectives. Samples of the compound collected from the flowers and the bees were tentatively identified as ocimene, a terpene.

Key words: Solanaceae, Cyphomandra, Eufriesea, pollination, euglossine bees, terpene, extrafloral nectar.

Cyphomandra is a genus of about 50 species of shrubs and small trees found from southern Mexico through South America. The best known member of the genus is C. betacea (Cav.) Sendtner, the source of an edible fruit of economic importance known as the "tree tomato" or "tamarillo." Cyphomandra endopogon Bitter var. endopogon is a small tree reaching 2-8 m in height. It is found in disturbed areas of tropical rainforest at 100 to 1000 m elevation. Its range extends from the eastern slope of the Andean cordillera in southern Colombia, through eastern Ecuador, northeastern Peru, and western Brazil, with disjunct populations in French Guiana and northeastern Brazil, close to its border with French Guiana. The other variety, C. endopogon var. parviflora Bohs, is found in rainforest between 100 to 800 m in the upper Amazon basin of southern Colombia, eastern Ecuador, and northeastern Peru (Bohs, 1986).

Male euglossine bees are known to visit the flowers of many species of Orchidaceae as well as those of some other families (e.g. Gesneriaceae, Araceae, Bignoniaceae, Solanaceae, etc.) to collect aromatic compounds which are thought to play a role in their courtship behavior (Dressler, 1982). Pollen collecting is done by female bees of many groups, including Euglossini (Buchmann, 1984; Roubik, 1989). A relationship between Cyphomandra and euglossine bees has been noted by other researchers. Soares et al. (1989) describes an interaction between Euglossa (Euglossella) mandibularis and Cyphomandra calycina Sendtner in Brazil. Mori (pers. comm.) and Dressler (in Williams, 1982) have observed Eulaema bombiformis gathering aromas from Cyphomandra hartegii (Miers) Walp. in Panama.

The purpose of this study was to determine what attractant was being provided by Cyphomandra endopogon var. endopogon for its euglossine visitors and whether the euglossines were effective pollinators.

Materials and Methods

Our study site was located in the vicinity of the village of Saül, in the geographical center of French Guiana (3° 37'N, 53° 12'W) at an altitude of approximately 200 m. Vegetation along the trail leading from the air strip to the village is frequently cut, thus allowing sunlight to strike the small Cyphomandra trees for several hours each day. Approximately twenty-five mature trees were located along a 200 m segment of the trail. Two of the larger trees (height = ca. 4 m, dbh = 19 cm and 14.5 cm on a double-trunked tree and 20 cm on a single-trunked tree), located across the trail from one another, were selected for detailed observation of floral visitors.

Observations began each morning between 0530 and 0550 hours and continued until dusk during the period of 18-24 August, 1988. Flowers and buds at different stages of development were tagged and monitored for growth, time of anthesis and other changes until the corollas fell off. Upon arrival at the study site each morning, we inspected all inflorescences within view on our two study trees. Open flowers were sampled for nectar throughout the day using an aspirator and micropipettes. A voucher collection was made of one of the study plants (Mori et al. 19054; NY, CAY). Stigmas from flowers of all different ages were tested for receptivity by placing them into a 3% solution of hydrogen peroxide and watching for bubbling activity, a test generally considered to be a valid indicator of receptivity (I. Baker, pers. comm.).

Ten minute counts of bee visits to flowers of the two Cyphomandra trees selected for observation were made at hourly intervals. The duration of bee visits was recorded, and in some cases, interfloral movements of bees were noted. This was facilitated by marking the abdomens of some of the bees with colored paint. Bees were subsequently captured, killed with ethyl acetate, mounted, and sent to a specialist for identification. Vouchers are deposited at the Harvard Museum of Comparative Zoology (Gracie 1, 2, 3, 12, & 14) and in the collection of the author (Gracie 11, 13, & 15).

Additional bees and freshly opened flowers were collected for aroma analysis. Bees and flowers were placed separately into glass chambers connected by flexible tubing to glass sorbent tubes packed with glass beads, Tenax-ta, Ambasorb, and activated charcoal (Envirochem model # ST-031). The sorbent tubes were connected by flexible tubing to a pump (Gilian model # HFS 113U) calibrated to draw air from the glass holding chambers into the aroma traps at a rate of 1 L/min with 25 inches of back pressure. Bee aromas were pumped through the tubes for 30 minutes. Flower aromas, which were much stronger, were pumped for 20 minutes.

The sorbent tubes were subsequently thermally desorbed and the aromas analyzed using gas chromatography and mass spectrometry in the laboratories of General Foods, Tarrytown, NY. The resulting aroma profiles were compared with those in the General Foods library of known compounds.

Pollen collected from the abdomens of the bee voucher specimens was examined with a scanning electron microscope and compared with that from the anthers of Cyphomandra endopogon var. endopogon.

Liquid droplets from the calyces were tested for sugar content using a Bausch and Lomb hand-held refractometer; this extrafloral nectar was spotted on Whatman #1 filter paper for later analysis. Buds were collected and preserved in FAA for examination with SEM.

In the following year, 1989, I obtained a sample of liquid ocimene (PQ). I returned to the study site in French Guiana on 1 Sep and found that the Cyphomandra trees were almost at the end of their flowering period. Several E. convexa bees were observed visiting the remaining open flowers. Three blotters were dabbed with ocimene and placed on a nearby fallen log ca. 65 cm above the ground and among the leaves of the Cyphomandra tree.


Cyphomandra trees were in flower and producing immature fruit in mid-August. The unbranched, pendulous inflorescences of coiled racemes each produced ca. 50 flowers, with only one or two opening per inflorescence each day. The flowers began to open shortly before 0600 hours, with many opening between 0545 and 0630. Opening was very rapid; petals were more than half reflexed within a few seconds, and fully reflexed within 1-1.5 minutes after they first showed signs of opening. A few flowers opened later in the day. Upon opening, the petals were a deep lavender color and strongly reflexed (Fig. 1A). Anther dehiscence occurred shortly thereafter. The green, capitate stigma was exerted 4-6 mm from the five appressed purple anthers. Stigma receptivity tests scored a positive response for stigmas of all ages, from those in unopened buds to those of withered flowers. By the end of the first day, the petal color had begun to fade to pale lavender, and the flowers closed before nightfall. Flowers receiving bright afternoon sun began to fade and relax their petals earlier than those in the shade. The flowers reopened on the second day, but the petals did not reflex as strongly and had increased in length by 1-4 mm (ave. 2.5 1.4 mm, n = 10). By day three, the petals had become greenish white and had increased in length by 4-9 mm (ave. 6.5 1.8 mm, n = 8) since anthesis. Anthers were no longer appressed to the style at this time (Fig. 1B). Corollas generally fell on the third day. Nectar was not present at any stage of floral development.

The first floral visitors appeared between 0615 and 0625 hours (Fig. 2). The principal visitors of Cyphomandra endopogon var. endopogon were observed to be males of two species of euglossine bees, Eufriesea convexa and E. elegans, that were collecting aromatic compounds from the flowers and incidentally serving as effective pollinators. Visits by E. convexa were far more numerous. Neither species is frequently collected, particularly E. convexa (Kimsey, pers. comm.). Two peaks of activity occurred, the greater one between 1100 and 1210 hours (35.7% of visits, n = 543) and a somewhat lesser one between 0700 and 0810 hours (30.7% of visits). Activity had virtually ceased by 1500 hours. The bees visited only freshly opened flowers, ignoring those with semi-reflexed, pale lavender petals and those with spreading, white petals. The bees would often hover in front of a freshly-opened flower momentarily before either landing on it or flying to another flower. When making an aroma-collecting visit, they would grasp the petals with their middle and hind legs, hang beneath the flower, and use their forelegs to stroke the bases of the anther connectives. Examination of the anthers after a flower had been visited by Eufriesea showed that the swollen dorsal surface of the anther connective was discolored, appearing brown where the bee had scraped with its foretarsal brushes. The bees continued to collect compounds from the same flower or flew to another. As the bees stroked the anthers, their ventral surfaces were dusted with clouds of fine, white pollen released from the terminal pores. They were observed to subsequently contact the stigmas of the same or other flowers with their pollen-covered abdomens (Fig. 1D). Using SEM, pollen taken from the abdomens of the bee voucher specimens was found to be identical to that from the Cyphomandra anthers (Fig. 3A). The pollen is small (< 20 mm) with a relatively smooth exine as is typical of pollen from poricidal flowers (Barth, 1985). Bees were observed sequentially visiting flowers on different trees. Bohs (1986) states that all species of Cyphomandra, with the exception of C. betacea, are self-incompatible. Thus, some means of transporting pollen from one plant to another is necessary to ensure fertilization. Although the bees are wary and difficult to net, once they had begun their aroma collecting activity they were easy to observe and capture, as has been noted by Soares (1989) and Dressler (1982). If a bee was disturbed by another Eufriesea during aroma collection, it would aggressively fly at the intruder, chasing it away before returning to the flower.

Floral visit duration averaged 26.3 32.0 seconds with a range of 1-180 seconds (n = 117). Flowers that had been recently visited often were inspected briefly by hovering bees which then generally left without landing, perhaps indicating their ability to recognize whether or not aromatic compounds were available. Previously visited flowers were usually visited for shorter periods of time than those receiving a primary visit.

One incident of a bee landing on a large bud and thus causing it to open was recorded. Subsequently, we learned that stroking mature buds (ca. 1.9-2.0 cm in length) could often prompt them to open. Soares (1989) has observed Euglossa (Euglossella) mandibularis deliberately forcing open the buds of Cyphomandra calycina to collect aromas.

Chromatograms of aromas from the Eufriesea bees and Cyphomandra flowers showed an identical strong terpene peak (Fig. 4). The aroma profiles were matched most closely with that for ocimene. However, there are other compounds with the same molecular weight and with profiles having peaks that match almost as well (e.g. camphene). No bees were attracted to blotters scented with ocimene (PQ). Several reasons may be postulated for this lack of visitation: (1) the bees might have been attracted to a component of the floral aroma other than the one producing the largest peak in the chromatogram, (2) the bees may be been attracted to the compound only in combination with others, (3) the ocimene may have been the wrong isomer, (4) the original attractant may have been a closely related terpene with a similar profile. Aromas must be recollected in order to identify the compound with certainty and to determine the particular attractant for the bees.

Small brown ants were observed to visit droplets of liquid, presumably extrafloral nectar, on the calyces of the youngest buds within the coiled portion of the raceme. The droplets were gone by ca. 1000 hours each day, but some ants continued to visit the inflorescences after that time showing an interest in the margins of the calyces where droplets had been found earlier. Larger red ants were observed patrolling the inflorescences at night. SEM revealed the presence of secretory glands on the exterior of the calyces of young buds (Fig. 3B). The stalked glands either atrophy or break off as the buds mature. The nectar droplets contained 0.7-1.0% sugar. Although readings for such dilute nectar are not highly reliable (I. Baker, pers. comm.), the calyx nectar was shown to contain sugars in the following proportions: sucrose 0.57, glucose 0.16, fructose 0.28, (ratio (S/G+F) = 1.29).Amino acids were present in small quantities (3 on the histidine scale), and no proteins, phenols or lipids were detected (I. Baker, pers. comm.). Further study is needed to determine if this extrafloral nectar is secreted in order to attract the ants that perhaps then serve to protect the young buds from herbivory. Occasionally, larger buds were observed to have insect damage.


Flowers of both Cyphomandra and Solanum do not produce nectar, have poricidal anthers, and are visited by bees which serve as pollinators. The principal difference between the flowers of the two genera is the enlarged anther connective in Cyphomandra. In Solanum, various species of female bees visit the flowers for the purpose of collecting pollen. The method used to dislodge the pollen grains from the anthers is termed "buzz pollination." Bees tightly clasp the stamens and rapidly contract and relax their indirect flight muscles producing strong vibrations; the wings are stationary during this activity (Erickson & Buchmann, 1984). An audible "buzz" may be heard which differs from the sound made by the bees in flight. The buzzing sound during pollen collection is a result of the interaction of the bee's cuticle with the boundary layer of air surrounding the bee. The position of the anthers is not important since the pollen is forcibly expelled (Buchmann, 1984). This vibratile pollination syndrome is found in many unrelated families and thus must have evolved independently several times (Buchmann, 1984). Poricidal dehiscence is found in at least 72 families and it is estimated that only 5-10% of plants with poricidal anthers are not buzz pollinated (Buchmann, 1984).

In the case of Cyphomandra, the flowers are more or less pendant so that the apical pores of the anthers face downward. Pollen is easily shaken out in a salt shaker manner when the anthers are touched. Male euglossine bees that visit the flowers of Cyphomandra do not visit to collect pollen, but rather to gather aromas. The foretarsi of euglossine bees are equipped with fine brushes which are able to absorb aromatic compounds through capillary action (Dressler, 1982). Their scraping movements on the greatly enlarged anther connectives during the process of aroma collection cause pollen to fall from the anthers onto their ventral surfaces. In both cases, an electrostatic differential between the flowers and the bees may play a role in increasing the precipitation of pollen from anther to bee. Plants, being grounded, carry a negative charge while bees in flight during fair weather have been shown to build up a significant positive charge (Erickson & Buchmann, 1984; Roubik, 1989). The bees often hovered in front of the flowers while they combed their aroma-saturated foretarsal brushes with stiff comb-like hairs on the midbasitarsus and then transferred the collected compounds into special slits in the enlarged hindtibias (Fig. 1C). The hindtibias of aroma-collecting euglossine bees are packed with spongy, glandular material that absorbs the aromatic compounds (Kimsey, 1982).

Pollen is thought to have been the original floral attractant for insects (Faegri & van der Pijl, 1966). The pollen of a great majority of flowers is yellow due to the presence of carotenoids and lipids. These pigments probably evolved to help protect the pollen tube and generative nuclei from damage by ultraviolet radiation (Buchmann, 1984) in pollen exposed as an attractant in anthers that dehise longitudinally. The pollen of poricidal anthers is usually white or cream-colored; UV protection is unnecessary since the pollen is not exposed, but rather enclosed in the anthers (Buchmann, 1984). In Solanum the external locule walls of the anthers are bright yellow, producing an image of abundant pollen even if the anthers are empty. The strong yellow of the anthers contrasts markedly with the usual purple or white corollas of Solanum as is typical in buzz pollinated flowers. Flowers with such strongly contrasting colors within the visible spectrum of light have been shown to be attractive to pollen-gathering bees even without reinforcing UV patterns (Buchmann, 1984). The anthers of Cyphomandra endopogon var. endopogon are purple and do not sharply contrast with the lavender corollas. The anthers of other species of Cyphomandra are variously described as being white or yellow, yellowish violet, purplish, pale yellow with purple spots, etc.; none are described as being bright yellow (Bohs, 1986). The pollinator attractant in Cyphomandra is not pollen, but aroma, thus visual attractants that mimic pollen are unnecessary. However, the aroma-producing anther connectives in Cyphomandra have become enlarged, thus providing euglossine bees with an expanded area on which to collect aromas. The interaction of different species of Euglossini with a number of species of Cyphomandra over a broad geographical range may indicate a coevolutionary relationship between the two groups of organisms. I suggest that the evolution of this relationship has contributed to the differentiation between Cyphomandra and the closely related Solanum. The relatively small differences between the genera, in particular the enlargement of the anther connective in Cyphomandra, that serve to reproductively isolate one from the other have most likely resulted from the coevolution of the flowers and their pollinators.


Special thanks go to my husband, Scott Mori, for sharing his earlier knowledge of the Cyphomandra-euglossine relationship, for his encouragement and suggestions throughout the study, and for reading the manuscript. Thanks are also due to the late Irene Baker for analysis of the extra-floral nectar, to Ron Belanger and Jerry Cohen of General Foods for technical advice and analysis of the aromatic compounds, to Don Black for taking the photomicrographs with the SEM, to Don Diaz of Gilian Instruments for the loan of equipment, to Gerhard Haas for facilitating the liaison between NYBG and General Foods, to Lynn Kimsey for identification of the bees, to Mick Richardson for advice on aromatic compounds, to Bob Rufe for obtaining the ocimene sample from Bush Boake Allen Division, Union Camp Corp., to John Mitchell for reading an early draft of the manuscript, and to Jackie Kallunki, Lisa Campbell and two anonymous reviewers for commenting on the manuscript. I am also indebted to the group of volunteers (H. Betros, L. Flynn, C. James, P. O'Malley, D. Vuillequez, and W. Pagels) whose patience and assistance were critical during the data gathering phase of the study. This project was funded in part by The Fund for Neotropical Plant Research of The New York Botanical Garden.

Literature Cited

Barth, F. G. 1985. Insects and flowers: the biology of a partnership. Princeton University Press, Princeton, New Jersey.

Bohs, L. A. 1986. The biology and taxonomy of Cyphomandra Solanaceae). PhD. dissertation. Harvard University. University Microfilms, Ann Arbor, Michigan; publ. no. 8620561.

Bohs, L. A. 1989. Ethnobotany of the genus Cyphomandra (Solanaceae). Econ. Bot. 43(2) 143-163.

Buchmann, S. L. 1984. Buzz pollination in angiosperms. Pages 73-113. In: Jones, C. E. & R. J. Little, editors. Handbook of experimental pollination biology. Scientific and Academic Editions, New York.

Dressler, R. L. 1982. Biology of the orchid bees (Euglossini). Ann. Rev. Ecol. Syst. 13: 373-394.

Erickson, E. H. & S. L. Buchmann. 1984. Electrostatics and pollination. Pages 173-184. In: Jones, C. E. & R. L. Little, editors. Handbook of experimental pollination biology. Scientific and Academic Editions, New York.

Faegri, K. & L. van der Pijl. 1966. The principles of pollination ecology. Pergamon Press, Oxford, England.

Kimsey, L. S. 1982. Systematics of bees of the genus Eufriesea (Hymenoptera, Apidae). Entomology. Vol. 95. University of California Press, Berkeley, California.

Roubik, D. W. 1989. Ecology and natural history of tropical bees. Cambridge University Press, Cambridge, England.

Soares, A. A., L. A. do O. Campos, M. F. Vieira & G. A. R. de Melo. 1989. Relações entre Euglossa (Euglossella) mandibularis Friese, 1899 (Hymenoptera, Apidae, Euglossini) e Cyphomandra calycina (Solanaceae). Ciência e Cultura (Revista da Sociedade Brasileira para o Progresso da Ciência) 41(9): 903-905.

Williams, N. H. 1982. The biology of orchids and euglossine bees. Pages 119-171. In: J. Arditti, editor. Orchid biology, reviews and perspectives. Vol. 2. Comstock Pub. Assoc., Ithaca, New York.

For additional informaiton on the pollination of Cyphomandra by euglossine bees refer to:

Sazima, M., S. Vogel, A. Cocucci & G. Hausner. 1993. The perfume flowers of Cyphomandra (Solanaceae): Pollination by euglossine bees, bellows mechanism, osmophores, and volatiles. Pl. Syst. Evol. 187: 51-88.

Go to    Figure 1 | Figure 2 | Figure 3 | Figure 4
Go to    Eufriesea convexa Photo Gallery
Go to    Cyphomandra endopogon var. endopogon Photo Gallery
Back to Pollination Biology Page
Back to Fungal and Plant Diversity of Central French Guiana Home Page