Studies of the floral biology of neotropical Nymphaeaceae. 3

Ghillean T. Prance (") Anthony B. Anderson (*")

Abstract STUDY AREA

A fíeld study of the floral bio1ogy of two In the vicinity of Manaus Nymphaea rud­ species of Nymphaea (Nymphaeaceae) w:>.s made. N. rudgeana G .F.W. Meyer, a night nowering geana was observed at the two localities cited species, was studied around Manaus and Belém. above, on the Manaus-ltacoatiara road. The It is visited by the scarab beotle Rio Preto site is a natural area where the plant castanea Oliv. in the Manaus vicinity and by ·C. grows in sha llow water (20-100 em depth) at verticalis Burro. in the Belém vicinity. N. rudgeana the river margin. The km 39 site is an arti­ is protogynous and cross-pollination is quite fre. quent. N. ampla (Salisb.) DC. , a day flowering ficial lake caused by the road embankment in species, was studied near Salvador, Bahia. This an area that was formerly forest on h!gh species is visited by tbe bee species Trigona spini. grouna. pes. N. ampla is not protogynous and is largely Here N. rudgeana completely covers the sur­ self.fertilized, although a small amount of cross­ .fertilization may occur through the activity of the face of the lake, which has a small stream Trigona bees. running through it.

lNTRODUCTION OBSERVATIONS

Following the study of the biology of N. rudgeana is a night flowering species in (Poepp.) Sowerby made by which the flowers open for two consecutive the senior author (Prance & Arias, 1975; Pran­ nights and are visited by the scarab ce et a/. 1975), we decided to continue Cyclocephala castanea Oliv. This species of the work with other Brazilian species of beetle was commonly found in the flowers of Nymphaeaceae. The present paper presents Victoria amazonica, see Prance & Arias (1975), the results of. ou r observations on two species Prance et a/. (1975). ot the , Nymphaea, with the ultimate ai_m The f lower opening sequence is given ot providing enough R"aterial to discuss evo· below for one night of observations. The same lutionary \rends in the' family on the basis ot sequence with minar time differences (de­ floral biology. pending on cloud cover etc.) was observed on ali 10 nights of field work. 1. Nymphaea rudgeana G.F.W. Meyer The opening of the flowers is directly This species was studied in the vicinity of linked w ith the onset of darkness as in Victoria Manaus and Belém. As the observed species amazonica (Gessner, 1960; Prance & Arias, ot pollinating were different in the two 1975). Ali the flowers on the lake open at areas, the results from each area are presented the same time, as it becomes dark. At that separately. time they are strongly scented with a fruity smell. When the light is still bright in the I. VICINITY OF MANAUS West the flowers open unilaterally, opening on Herbarium Material: Prance & Ramos 23384, the shaded (East) side first. Rio Preto da Eva, Manaus-Itacoatiara road, km 79, Amazonas, Brazil. (INPA, NY); Prance By marking flowers with colored tape we & Ram os 23521, Manaus-Itacoatiara road, km established that they open for two consecutive 89, Amazonas, Brazil (INPA, NY). nights only. On both nights they give off the

( •) _ The New York Botanical Garden, Bronx, New York, 10458, and Instituto Nacional de Pesquisas da Amazônia, Manaus. ("") - Instituto Nacional de Pesquisas da Amazônia. Manaus .

ACTA AMAZONICA 6(2) : 163-170 - - 163 TABLE 1 ly forest. Beyond the lake are areas which Flower opening sequence for Nymphaea rudgeana are still forest. N. rudgeana forms a dense in the vicinity of Manaus (Ma.naus-Itacoatiara km 39) cover near the margins of the lake in water depths from 1 to 2 meters. Time Light Observations on 25 _06 . 75 OBSERVA TIONS 17:45 Daylight All flowers closed, buds with fruity smell. As in the Manaus region, flowering is nocturnal and the flowers open for two or 18:15 Sunset Sepals slightly open, strong three consecutive nights. They are visited by fruity small over lake. the scarab beetle Cyclocephala verticalis Burm. 18:30 End of Sepals more open, white of This is a smaller species of the same genus as twilight petals now showing. the beetle which visits N. rudgeana in the vi­ 18:45 Dark Sepals and outer petals rea.ch cinity of Manaus. C. vertica/is is anot i,Gr 45<> a.ngle, central cone of petals species that was found occasionally in the still closed. flowers of Victori·a amazonica, see Prance & 18:55 First beetle ( Ciclocephala cas­ Arias (1975). Prance et (1975). tanea) observed entering a fio· a/. wer by forcing central cone of The sequence of flower opening for two petals open. nights at Belém is given below: 19:20 Flowers with sepals and outer petals wide open, seprus at 75o TABLE 2 angle, central cone, of petals Flower opening sequence for Nymphaea rudgeana closed or <>lightly open in some in tbe vicinity of Belém flowers. Time Light Observations on 14 .07 . 75 s3me smell at equal intensity, but only on the and 16.07 .75 second night is the pollen released from the 17:45 Daylight linear anthers. This suggests that the stigma Ali flowers closed. is receptive on the first night. Thus, like 18:15 Sunset Sepals slightly open, strong Victoria, this species exhibits protogyny. fruity smell ali over lake. At both sites studied, the flowers of N. 18:30 End of Sepals more open, white of rudgeana are visited by the scarab beetle twilight petals now showing. Cyclocepha/a castanea Oliv. The beetle popu­ Fruity smell more intense. Jation was not dense at the time of our study 18:45 Dark Sepals and outer petals wide (25 June - 17 July): with numerous flowers open, central cone of petals closed. open (ca 300) on the lake at km 39, we could only find an average of six beetles in any onc 18:55 First beetle ( Cyclocephala ver­ evening. The beetles appear to come from ticalis) observed landing on the surrounding forest shortly after nightfal \. sepals and entering flower by forcing open the central cone Only one beetle at a time was observed in the of petals. flowers. 19:15 Central cone of petals slightly open in some flowers. i i . VICINITY OF BELÉM Herbarium material: Anderson 159, small The tables show that the behavior of the stream near beginning of Estrada Coqueiro, flowers is similar at both the Manaus and Be­ Belém, Pará, Brazil (INPA, NY). lém sites. By marking the flowers with colored tape STUDY AREA we established that they open for either two The Belém study area, like one of the Ma­ or three consecutive nights at Belém. By the naus sites, is an artificial lake caused by a third or fourth night, the flowers have a rosy road embankment in an area which was former- pink color and emit no scent. On the first

164- Prance & Anderson the flowers open and the beetles emerge covered with pollen, whereupon they are attracted to receptive first day flowers, which are white and strongly scented. In Nymphaes ruu'geana, the beetles are not trapped inside the fiowers during the day, nor is there any change in scent or calor between first and second day flowers. In arder to effect cross­ pollination, a beetle must move by chance from a second or third day flower, which is shedding pollen, to a receptive first day flower, and not to another second or third day flower. Furthermore, the relatively small size of the flowers of N. rudgeana does not allow for more than one or at the most two beetle visitors at a time; in contrast, the large flowers of Victoria accommodate half a dozen or more beetles at one time. As a result, the pollina­ tion mechanism of N. rudgeana is far less efficient than that of Victoria. However, a certain amount of gene exchange is ensured Fig. 1. Open flower of Nymphaea rudgeana. within the species. Beetles which visit second day flowers always emerge covered with pol­ len which sticks to them, and when by two or three nights ot opening, the ftowers chance they visit a first day flower with d give off the same smell at equal intensity, but receptive stigma, cross-pollination is likely. only on the second and third nights does the In the Manaus area, flowers which were pollen appear to be released from the mear bagged during their three days above water anthers. This suggests that the stigma is receptive on the first night. Thus, like V. amazonica, this species exhibits protogyny. The flowers are visited by the scarab beetle Cyclocephala verticalis. The beetle population was not dense at the time of our observations (July 14 - 16, 1975): with numerous flowers open (ca 50), we could only find a maximum of four beetles in any evening. The beetles appear to come from the adjacent forest shortly after nightfall and enter the flowers. They move freely from one flower to another during the night. Usually only one beetle, or at the most two, enter a flower at any one time.

DISCUSS!J:ON

The pollination of Nymphaea rudgeana differs considerably from that of Victoria ama­ Fig. 2. Flower section of Nymphaea rudgeana, zonica. In Victoria, the beetles are trapped showing the inner recurved st::-.minodes which are inside the flowers during the day. At nightfall, eaten by the beetles.

Studies of the fioral . .. - 165 consecutive days, whereupon they sink be­ neath the surface of the lake. The species was observed in flower during July ( 1975) and January ( 1976).

TABLE 3 Flower opening sequence for Nymphaea ampla in the vicinity of Salvador

Time Ligbt Observations on 13. 01. '16

5:00 Twilight Ali flowers closed. 5:45 Sunrise 6:00 Full daylight 7:00 Sepals of some flowers begin to open, slight scent emitted from flowers. 7:15 Some flowers with sepals and petals wide open, others with sepals beginning to open; ali flowers emitting slight scent. Arrival of first bees, which enter open flowers and land Fig. 3. Flower of Nymphaea rudgeana with the on closed flowers, chcwing the beetle Cyclocephala castanea. on the petals. tips of the closed petals. 8:00 Ali flowers open. Bee activ1ty intense. set seed freely, showing that N. rudgeana 1s 10:00 Bee activity declining. self-compatible. lt is probable that autogamy 12:00 Very few bees in vicinity of flowers. is the predominant form of pollination in N. 15:00 JFlowers begin to close. rudgeana. 17:00 All flowers closed. 18:15 Sunset 2. Nymphaea ampla (Salisb.) DC. Herbarium material: Anderson 162, Lagoa do Abaeté, approximately 10 krn NE of Salvador. Bahia, Brasil (INPA, NY); Anderson 225, Same locality (!NPA, NY).

STUDY AREA

N. ampla was studied in a natural black­ water lake surrounded by dunes of white sand within a few hundred meters of the ocean. The dunes are covered by a low, àense vege­ tation known as restinga. N. ampla occurs in low concentrations over scattered are::~s near the margin of the lake, in depths of water from 1-3 meters.

OBSERVATIONS

N. ampla is a dayflowering species which is visited by the bee Trigona (Trigona) spinipes ( = T. ruficrus). The flowers are white and give off a sweet scent which is not as pungent Fig. 4. Flower of Nympbaea rudgeana with the or fruity as the flowers of N. rudgeana. In N. beetle Cyclocephala castanea inside the flower among the stamens. ampla, the flowers open for three to four

166 - Prance & Anderson Fig. 5 . Flower of Nym phaea ampla.

Studies of tbe floral. .. - 167 gests that the flowers may only be receptive on the fi rst day. The f lowers are visited by the bee Trigona (Trigona) spinipes ( = T. ruficrus). During the three mornings of observations, as many as 9 bees could be seen visiting the 30 or more opened flowers in one area of the lake. The bees arrive at the time th3t the flowers are just beginning to open. They land on the first, second or even third day flowers and begin to ch.ew the upper tips of the closed petals in an apparent attempt to gain entry. /\s these bees are extremely slow and sporadic workers, they probabfy would never penetrate the thick cone of petals on their own if the Fig. 6. Flower of Nymphaea ampla. f lowers did not then commence to open, thL:S al lowing the bees to enter inside. Once inside. the bees begin to collect pollen from the inner The flower opening sequence is given rows of fertile stamens, which are curled over above for one day of observations. The same the s'dgmatic area. During this process, con· essential sequence was observed during the siderable amounts of pollen fali fron~ the entire 3 days of field work. anthers onto the stigmatic area. One bee was The opening of the flowers does not appear observed visiting a single flower for 17 minu­ to be directly linked with the approach of tes, during which time it caused far greater quantities of pollen to spill from the stamens daylight. Nor is the timing of flower opening onto the stigmatic area than it collected. The as synchronized among the population as it is, bees generally emerge from the flowers co- for example, in populations of N. rudgeana . lt vered with pollen, which sticks to them. No is possible that temperature is a more effective other were observed actualiy entering cue than sunl ight for the opening and closing the flowers. ot the tlowers in this species.

By marking the sepals of the flowers wi th DISCUSSION indelible ink, we determined that they open for three to four consecutive days. By the thi rd The Trigona bees are probably responsible or fourth day the flowers have a rosy pink for a small amount of out-crossing in /iJ. ampla. color and emit no scent. On the first two or The bees emerge from each flower coveretl three days of opening, the flowers give off the w ith pollen. and visits to adjacent flowers ssme delicate scent at equal intensity. The enable cross-pollination to occur. However, flowers generally appear on the surface ot the the amount of pollen transfer effected by bees lake a fui I day prior to their initial opening. is minimal compareci to the large quantities of Examination of these unopened flowers showed pollen which spill from the anthers onto the that polfen had already been released from the stigmatic area throughout the flowering period anthers and had fallen onto the stigmatic area, The small amount of pollen transfer which was still moist with lake water. The effected by the bees cannot account for the stamens were observed to be continually fer­ abundance of capsules bearing apparently tile throughout the flowering period, releasing viable seeds which were found in the areas copious amoLmts of pollen, much ot which where N. ampla occurs on the Lagoa de Abae­ falls onto the stigmatic area. By the second té. No mechanisms such as protogyny (ob­ or third day, the moisture on the stigmatic served in N. rudgeana) or protandry occur in area has completely evaporated, which sug- this species. Furthermore, no chemical or other

168- Prance & Anderson _, barriers to self-pollination have been reported flowers remain open for 3-4 days and. as we in other species of Nymphaeaceae. Therefore, have shown, outcrossing is rare and fertílíza• it seems safe to conclude t hat autogamy is tíon occurs predominantly via autogamy. In N. the predominant form of pollination in N. am­ rudgeana , flowers remain open for shorter pla, with occasional outcrossing effected by perioéls (2-3 days) and outcrossing appears to Trigona bees. be more frequent, although autogamy remains predominant. Finally, in Victoria amazonica, Nymphaea ampla, ranging from Texas to the period of flower opening is the least ex­ central Brazil, is one of the most widespread tensive (1 day) and ferti lization occurs pre­ .L\merican species of its genus. The species domínantly via outcrossing. In summary, V. is predominantly autogamous in the area stud­ amazonica, with its extremely effective cross­ ied , which is at the extreme southern part of pol lination system, achieves outcrossing in a its range. lt is likely that the autogamous re­ minimal period of time; by contrast, the spe­ production system enables this species tu spread beyond the range of its normal pollina­ cies w ith less effective cross-pollination tors. The importance of autogamy in the mi­ systems, N. rudgeana and N. ampla, maintain gration of species is well-known, and has been their flowers open for longer periods, appar­ much commented upon especially in relatiGn ently thereby increasing the probability of a to islend species. certain amount of outcrossing. Further stuoies of floral biology in other species of Nymphaea­ ceae would reveal whether there is a general CONCLUSIONS correlation between the length of time that flowers remain open and the effectiveness of The two species studied belong to different outcrossing. Such a correlation may prove to seci:ions of Nymphaea and have completely have evolutionary significance within the different f loral biology. N. rudgeana belongs family. to sections Syncarpiae of Conard ( 1905) and N. ampla to section Apocarpiae. These sections differ in the amount of fusion of the carpels: ACKNOWLEDGEMENTS in section Syncarpiae, the carpels are com­ pletely fused, while in Apocarpiae the upper The fieldwork was supported by the lnsti· part of the c::~rpe l s is free. The evolution of tuto Nacional de Pesquisas da Amazônia, Ma­ the carpel of Nymphaea was d1scussed in some naus, and by the National Science Foundation detail by Moseley ( 1961), based on vascular Grant GB-32575 X 3, and the work was carried anatomy and ontogony. He outlines the two out while the junior author held a Thomas J. trends to complete fusion of the ca rpels or to Watson Fellowship: ali are gratefully acknow­ fusion on the lower portion only, but he does ledged. We th::tnk José F. Ramos and Elza not derive the Syncarpiae from the Apocarpiae. In at least the two species studied here, the Suely Anderson for help with the fieid1 work, member of Syncarpiae, N. rudgeana, ís night Prof. Dr. Warwick E. Kerr for identification o ~ flowering and beetle pollinated, and the Apo­ t he bees, and Dr. Sebo· Endrõdi for identific?.­ carpíae, N. ampla, is day floweríng, vísited by tion of the beetles. bees, but largely autogamous. Further studies of other species are necessary to determine Resumo whether any correlation exists between carpel type and pollination or fertilization system. Foi feito um estudo da biologia floral de duas Our studies to date of floral biology in the espécies do gênero Nymphaea. (Nympha.eacea.e)_ N. rudgea.na G.F.W. Meyer floresce durante a noite Nymphaeaceae suggest that there exists a e foi estudada perto de Manaus e Belém. Na vizi· correlation between the length of time that nhança de Manaus foi encontrada a espécie de be· flowers remaín open and the effectíveness of souro Cyclocephala. castanea Oliv. (Sca.raba.eidae: outcrossíng. For example, in Nymphaea ampla Dynastinae) nas flores de N. rudgea.na, porém na

Studies of the floral. .. - 169 vizinhança de Belém foi encontrada outra espécie GESSNI:R, F. de besouro, Cyclocephala verticalis Burro. N. rud­ 1962 - A abertura das flores de Victoria regia geana é protogínica e polinização cruzada é fre­ em relação a luz. Boi. Mos. Goeldi, N. qüente. Nymphaea ampla (Salisb.) DC. floresce Sér . Botânica 17 : 1-13. durante o dia e foi estudada perto de Salvador, MOS!!LtY, M. F. Bahia. Nesta espécie foram encontradas abelhas Trigona, T. spinipes. N. ampla não é protoginica e 1961 - Morphological studies of the Nymphaea­ é predominantemente autogama. porém um pouco ceae II. The Flower of Nymphaea. Bot de polinização cruzada pode acontecer devido a ati­ Gaz., 122 : 233-259. vidade das abelhas. PRANCC, G. T. & J. R. ARIAs 1975 - A stuey of the Floral Biology of Victo. ria amazonica (Poepp.) Sowerby (Nym­ LITERATURE CITED phaeaceae). Acta Amazonica, 5 (2) : 109-139. CONRAD, H. S. PRANCI·, G.T., PR<\NCf:, A.E. & J .R . ARIAS 1905 - The water lilies. A monograph of the 1975 - A biologia floral de Victoria amazonica genus Nymphaea. Carnegie Inst. Was. (Poepp.) Sowerby (Nymphaeaceae) hington Publ., 4. Ciência e Cultura, 27: 1293-1297.

N O TE

While this article was in press, we received a paper on Nymphaea spp. in Surinam. (Cramer, J. M., A. D. J . Meeuse & P . A. Teunissen

Cramer et al. 0975) propose that the pollination system of the two Nympha~a species is very much the same as in V. amazonica. However, our observations reveal important differences. Thc pollination system of v. amazonica an increase in temperatura within the flowers, a sbort flowering period (1 da.y), and a higb degree of outcrossing. Conversely, the pollination system oí N. rudgeana is far less sophisticated: tempera­ tura increases were not observed in the flowers, flowering periods are relatively long (2-3 days), and outcrossing is much less common. Our study oí floral biology in species of Nympbaea supports the conclu­ sion that the pollination system of V. amazonica represents a highly evolved and specialized development within the Nymphaeaceae.

170 - Prance & Anderson