Floral Biology of Some Clusia Species from Central Amazonia

Floral Biology of Some Clusia Species from Central Amazonia Author(s): V. Bittrich and Maria C. E. Amaral Source: Kew Bulletin, Vol. 52, No. 3 (1997), pp. 617-635 Published by: Springer on behalf of Royal Botanic Gardens, Kew Stable URL: http://www.jstor.org/stable/4110290 Accessed: 06/03/2010 15:22 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=kew. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Royal Botanic Gardens, Kew and Springer are collaborating with JSTOR to digitize, preserve and extend access to Kew Bulletin.

http://www.jstor.org

Floral biology of some Clusia species from Central Amazonia V. BITTRICH'& MARIA C. E. AMARAL1 and 3 species of Clusiasect. Summary.The floral biology of 3 species of Clusiasect. Chlamydoclusia from CentralAmazoniawasinvestigated.The flowersof both sexes offer resin as the reward Cordylandra for pollinatingbees. The male flowersof the speciesof sect. Chlamydoclusia mayadditionallyoffer pollen. In the male flowers Principalpollinatorsobservedwere bees, mainlyof the tribesEuglossiniand Trigonini. of sect. Chlamydoclusiathe pollen is kept separately from the floral resin, but in sect. Cordylandrait is mixed with the resin. The elongated connectives of the fertile stamens in sect. Chlamydoclusiasecrete

droplets of an oily fluid which probablyserves as accessory pollenkitt. The stigmas of the pistils and secrete an oily fluid; possiblefunctions of this are discussed. C. pistillodesof flowersof sect. Cordylandra has male plantswith two differentfloralmorphs;pollen of both producedfertile seeds when renggerioides used to pollinate female plants. The taxonomic significance of the different pollination strategies in Clusiaand their evolution is discussed.

INTRODUCTION

The neotropical

genus Clusia L. (Guttiferae) comprises

in a number of more-or-less natural taxonomic

about 250 species grouped

sections. Life forms include

lianas, shrubs and trees. As in other (sometimes stranglers), hemiepiphytes Guttiferae (Clusiaceae), most tissues of the plants contain latex channels. While the

vast majority of Clusia species are dioecious, a few are hermaphrodite or include races with hermaphrodite flowers (Maguire 1959, Hammel 1986, Bittrich & Amaral 1996). The genus is notable for its wide range of floral morphology, especially in the androecium (e.g., Planchon & Triana 1860, Engler 1888). This morphological variability is paralleled by a great diversity in reproductive biology (Armbruster 1984, Bittrich 1993, Bittrich & Amaral 1996, Correia et al. 1993). Rewards for pollinators include pollen, nectar and floral resin. In a couple of species (e.g., of sect. Criuva and sect. Clusiastrum) the female flowers are rewardless and pollinated by "mistake-pollination" (cf. Baker 1976). Bees of various sizes belonging to several taxonomic groups, beetles, hummingbirds and flies have been observed as flower visitors. Apomixis is known in C. roseaJacq. and races of C. minor L. (Maguire 1976, Hammel 1986). Floral resin is an extremely rare reward for pollinators in angiosperms; apart from Clusia spp., it is only known from Dalechampia L. spp. (Euphorbiaceae; e.g.,

Armbruster 1984) and ClusiellaPlanch. & Triana (Guttiferae).The chemistry of the floral resin of some Clusia species of sect. Chlamydoclusiaand sect. Cordylandrawas Accepted for publication June 1997. 1Depto. de Botinica, IB - Universidade Estadul de Campinas, C.P. 6109, 13081-970 Campinas, SP, Brazil.

617

618

KEW BULLETINVOL. 52(3)

investigated by Oliveira et al. (1996); various polyisoprenylated benzophenones were found as major components. Hammel (1986) defined the so-called "C. minor group" within Clusia by the presence of floral resin. There is still no evidence, however, that this group is monophyletic; floral resin has evolved at least twice in Guttiferae:in Clusia and in Clusiella. These genera are considered by Stevens (in press) not to be very closely related. Resins and the thickened latex secreted from wounds of plants are used by certain female bees for nest construction (Armbruster 1984, Roubik 1989, pers. obs., Fig. 1A), perhaps because of their plastic, slow-drying, waterproof properties and possible antimicrobial/antifungal activity (Armbruster 1984, Oliveira et al. 1996). The enormous variety in androecial morphology of male flowers in Clusia is particularly striking in resiniferous flowers. In a recent paper (Bittrich & Amaral 1996) we put forward the hypothesis that this variation may be partly due to the fact that pollen which gets stuck to resin may be lost for pollination, and that this problem triggered the evolution of mechanisms that avoid such a loss. In the present paper we investigate this point further and describe the floral biology of 6 species from two different sections of Clusia, sect. Chlamydoclusia and sect. Cordylandra,offering floral resin as reward. Observations on floral biology from other Clusia species are also included and discussed. MATERIALS

AND METHODS

Our observations were made in Central Amazonia near Manaus (03'07'S, 60'01'W) between May 1991 and December 1992. The populations of the various Clusia species are all located within 60 km of Manaus (see Table 1). The abundance and behaviour of flower visitors during anthesis were recorded. Some flower visitors were captured and mounted for later identification. In some cases, bees were marked on their back with coloured correction fluid to check for movements between different plants. Voucher specimens of the bees are deposited at the Dept. of Entomology of INPA, and plant specimens were deposited in the Herbaria INPA and UEC (see Table 1). was followed throughout the The phenology of the population of C. renggerioides on and fruits the trees. for flowers year by checking monthly open of for the were checked glucose using glucose test paper, and presence Stigmas for oily substances with Sudan III. FI(;. 1. A female Melipona bee collecting latex from a wound in a branch of Symphonia globulifera L. f. (scalebar = 10 mm); B male plant of C. grandiflora, showing the large pendulous flowers (scalebar = 50 cm); C open male flower of C. grandiflora with the resin plate in the centre and the crown-like fertile androecium around (scalebar = 20 mm); D Euglossa bee with some resin on the corbicula landing of the stamens of a male flower of C. grrandiflora.The connectives are strongly prolonged and show droplets of an clear oily fluid (scalebar = 5 mm); E Euglossa bee hovering below a flower and moving the just collected resin ball to the corbiculae (scalebar = 5 mm); F female flower of C. grandiflora with the large umbraculiform stigma in the centre and the resin secreting staminodes around the ovary (scalebar = 20 mm); G Euglossa bee collecting resin from the staminodes while sitting on the stigma of a female flower (scalebar = 10 mm); H female flower of C. insignis showing the lemonyellow stigma in the centre and a resin-collecting Eufrieseapurpurata bee (scalebar = 10 mm);J male flower of C. nemorosa showing the yellow resin in the centre with a resin-collecting Trigona bee on it and the brownish stamens with much elongated connectives around (scalebar = 20 mm).

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

619

KEWBULLETIN VOL.

620

52(3)

TABLE 1. Species of Clusia investigated Voucher no.: Amaral & Bittrich C. grandifloraSplitg. C. insignis Mart. C. nemorosaG. Mey. C. lepranthaMart. Planch. & Triana C. renggerioides C. pana-panai (Aubl.) Choisy

91/8 90/22 91/26 91/15 91/25 92/3 91/28 91/42 91/29

Locality Manaus - Caracarai km 60 Reserva Ducke near Manaus airport Manaus - Caracarai km 60 tributary of Rio Tarumai-agu Manaus - Caracarai km 60 near Manaus airport near "cachoeira alta do Rio Tarumf" Campus of the "Universidade do Amazonas"

in C. nemorosa and C. renggerioides, buds were For the crossing experiments anthesis. These proved better under humid the before in on enclosed day nylon bags or than conditions closed paper bags plastic bags. On the day of anthesis the stigmas with a counted number of of some flowers of C. nemorosa were hand-pollinated the flowers hand lens and a 10x bagged again. In C. renggerioides pollen grains using were mixture of the put on the stigma on the day of resin-pollen tiny droplets the flowers and or to two anthesis, bagged again. Fruit set was days afterwards, up recorded several weeks afterwards and seeds of ripe fruits of C. renggerioides plants were later collected (c. 12 - 14 months after anthesis) and sowed in wet sand to check for germinability. As all populations investigated by us had male and female plants, we excluded the possibility of apomixis in them (cf. Richards 1986) and did not specifically test for it. The abortion of flowers bagged for other experiments in C. insignis, C. nemorosa, and C. renggerioidesconfirmed this assumption. Flowers and flower visitors were photographed using a Nikon 601 camera with flash and Fuji Velvia (50 ISO) colour reversal film. Pollen and anthers were investigated by light microscopy, and very young stages of flower development with the help of a Philips 505 SEM. For the SEM study, flowers were fixed in FAA, critical-point-dried, and sputter-coated with gold.

FIG;.2. A male flower of C. leprantha showing the conic-crater-shaped androecium and the large stigma of the pistillode in the centre. A visiting Ptilotrigona lurida bee is visible on the left hand. Two petals were removed (scalebar = 10 mm); B female flower of C. leprantha showing one row of staminodes with resin on their apex and the huge pistil in the centre. A Ptilotrigona lurida bee is collecting resin while sitting on the stigma (scalebar = 10 mm); C flowers of C. renggerioidesfrom different trees of the population at Manaus airport showing the three flower morphs and colour variation. Female flowers of two colour types on the left hand, male flowers without a pistillode in the center, and male flowers with pistillode on the right hand (scalebar = 20 mm); D male flowers of C. renggerioidesof the morph without a pistillode (scalebar = 20 mm); E androecium of C. renggerioides(morph without pistillode) showing the resin mixed with the pollen and the anthers lying horizontally on the filament apex (scalebar = 2 mm); F female flower of C. renggerioides.The stout staminodes have droplets of clear resin at their apex. The stigma is wet and acutely papillose (scalebar = 2 mm); G centre of a male flower of C. renggerioidesof the morph with pistillode, note the narrow lower part of it. The stout stamens have resin droplets at their apex. The pollen is visible in the interior of the resin droplets (scalebar = 2 mm).

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

621

622

KEWBULLETIN VOL. 52(3) RESULTS AND DISCUSSION

1. Sect. ChlamydoclusiaEngl. (= sect. Euclusia Planch. & Triana nom. inval.). The flowers of this section have numerous fertile stamens at the periphery and resinproducing staminodes in the centre of the male flowers (Fig. IC +J). Their primordia develop centrifugally on a ringwall. A pistillode is absent in the male flowers, but some ovary development is visible in the flower centre in the early stages of flower development. The stamens show a prolongation of the connective above the anthers (Fig. ID). In all species investigated the connectives produce droplets of an oily fluid (Bittrich & Amaral 1994). For details of the morphology of the species studied see Table 2. 1.1 Clusia grandiflora.

The population studied grew near a small creek in moist to wet sandy to loamy soil. Flowering lasted from February to June. We counted 5 male and 3 female plants flowering in the population. The distance between these plants varied between 10 and 50 m.

TABLE2. Flower and fruit characters of the species studied of Clusiasect. Chlamydoclusia Engl. C. grandiflora

C. insignis

C. nemorosa

II - VI

VIII - XI

VI - X

inflorescence

1 - 3 (d), 1 (9)

1 - 3 (d), 1 (9)

3- 14(d), 1-5(9)

Flowerdiameter

up to 17 cm

8.5 - 10.5cm

4.8 - 5 cm

Flowerform

bowl-shaped

tulip-or cup-shaped

bowl-shaped

Flower colour

white to pinkish

white outside/

white with red centre

wine-redoutside

Flowering period

Flowersper

Stamens

7 - 10 series

8- 12 series

(in Amazonia) 3- 5 series

Prolongation of the connective

4 - 6 mm

1.5 - 2.5 mm

2 - 3 mm

Resinplate formed by staminodes (d' fl.)

18 - 20 mm in diameter

15 - 18 mm in diameter

5 - 7 mm in diameter

Staminodes (9fl.)

3 - 4 series

6 - 8 series

2 - 4 series

Stigmas

10-15

15-21

4-10(-12)

Stigmatic area

c. 20 mm in diameter

12 mm in diameter

c. 7 mm in diamter

Fruitmaturity

afterc. 12 months

afterc. 5 months

afterc. 6 months

Fruit size

28 - 30 cm in diameter

14 - 16 cm in diameter

3 - 5 x 2 - 3 cm

Fruitcolour

greenish-white red

darkviolet (in Amazonia)

Seed colour

greenish-white cream

Arilcolour

orange

yellowish

orange-red

red

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

623

FIG. 3. A androecium of a male flower of Clusia hilariana Schltdl. (sect. Phloianthera), with the fertile part forming a semi-globular structure in the centre and the resin-producing staminodes surrounding its base (scalebar = 10 mm); B male flower of Clusiella axillaris Engl. with resin secreted by staminodes at the lower part of the androecium, while the fertile anthers invest the elevated knob-like structure (scalebar = 10 mm).

The large hanging flowers (Fig. 1B + C) emit a rather strong, pleasant odour reminiscent of Magnolia grandifloraflowers. The stamens are connate at the base for c. 2 cm and form a crown-like structure (Fig. IC). The connectives are produced into long awn-like tips which project inwards and thus narrow the access for flower visitors to the interior of the flower. On the day of anthesis, the staminodes in the centre of the flower form a plate covered with yellowish opaque resin. The female flowers show resin-producing staminodes surrounding the ovary (Fig. 1F). The huge green stigmatic area in the flower centre protrudes well above the staminodes. It is covered at anthesis by large quantities of sticky mucilage which contains hexose (c. 0.1 - 0.2 %). Flowers of either sex last one day; by noon the petals start to become brownish, and are generally shed from the male flowers during the afternoon. The flowers opened at about 3.00 a.m. Resin production started at about 23.00 hours on the previous day, so some resin is present already on the staminodes. From dawn onwards the flowers were visited regularly by resin-collecting bees. The main visitors were three or more unidentified species of Euglossaand occasional Eufriesea pulchra; less frequent were small bees of the tribe Trigonini (Partamona gr. cupira, Partamona sp., Trigona (Trigona) williana, Ptilotrigona lurida) and Meliponini (Melipona

lateralis).The morphology of the male flowers does not generally permit the bees to land directly on the resin plate; rather, they are forced to land on the stamens (Fig. 1D). This is made more difficult by the strongly elongated connectives. The bees generally need more than one attempt to land successfully, and during these attempts they are well dusted with the dry powdery pollen. Additionally, the connectives produce droplets of an oily liquid (Fig. 1D), which probably serves to make the dry pollen stickier, thus functioning as accessory pollenkitt (cf. Vogel 1984, Steiner 1985). After reaching the resin plate, the bees form little balls of resin with their mandibles and eventually cut them loose from it. Subsequent behaviour differs

624

KEWBULLETIN VOL. 52(3)

in Euglossini and Meliponini/ Trigonini. The latter can move the resin ball from the

mandibles to one of the corbiculae while remaining seated on the resin, but Euglossinihave to leave the flower to achieve this operation while hovering below the flowers (Fig. 1E). If there is space left on the corbiculae the bees return to the flower to collect more resin. During the hovering phase, however, they are often attacked and chased away by other euglossine bees and especially by Trigonasensu lato. This behaviour might promote pollen transfer, as the expelled bees will visit other flowers to collect more resin. This is difficult to prove, but it is noteworthy that of 10 euglossine visitors observed visting a certain flower during one morning, four arrived with some resin present on their corbiculae (see also under C. insignis below). When visiting a female flower, the bees always land on the large protruding stigma area and often collect the resin from the apex of the staminodes while sitting on it (Fig. 1G). Pollination is thus certain if the bee has already visited a male flower. The sticky mucilage on the stigmas probably facilitates the removal of pollen from the bee's body. The strong-flying euglossine bees are good pollinators of C. grandiflora, but the small Trigonini bees, although common vistors, are much less efficient. Only extremely rarely did they arrive at a flower with resin on their corbiculae, and marked bees regularly returned to the same flower, or to another on the same tree. The relatively large amounts of resin offered by the flowers and the distances between the plants are probably responsible for this behaviour, which identifies the small bees as resin thieves of C. grandiflora. Only Trigona sensu lato bees collected

pollen from male flowers of C. grandiflora,

and then very rarely. However, when we fixed a branch with male flowers of C. nemorosa (see below)

in a male plant of C. grandiflora, several Trigona bees soon

started to collect pollen from these flowers. It is not clear why bees showed so little interest in the pollen of C. grandiflora. 1.2 Clusia insignis.

In the population near Manaus international airport we counted 25 male and 18 female plants. Distances between these plants varied between 0 and c. 50 m. Some male plants started to flower a few weeks earlier than the female plants. The morphology of the flowers is generally very similar to that of C. grandiflora,but the form and colour are different (Fig. 1H and Table 2). The resin is lemon-yellow and much more opaque than that of C. grandiflora,as if it were mixed with latex. Unlike the resin of C. grandiflora,that of C. insignis does not dissolve completely in organic solvents like chloroform. The female flower has a large lemon-yellow stigma, possibly mimicing the resin plate of the male flowers (1H). This may serve to orientate the bees in the rather dark interior of the flowers. Female flowers were slightly smaller than male flowers and more deeply coloured inside. The petals of male flowers measured on average 4.9 cm in length (n=37, S=0.33 cm) and 3.0 cm in width (n=37, S=0.38 cm), while those of the female flowers measured 4.2 cm in length (n=40, S=0.31 cm) and 2.2 cm in width (n=40, S=0.23 cm). The flower aroma is faint. Flower visitors observed at the airport site were bees of the tribe Euglossini: various unidentified

Euglossa species, Eufriesea purpurata, Eulaema sp., with Eufriesea

purpurata (Fig. 1H) being clearly the most common visitor. Of the tribe Trigonini: Trigona (Trigona) fulvicentris, Trigona (Trigona) williana, Duckeola ghilianii; Meliponini:

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

625

Melipona fulva. At the Reserva Ducke bees of the tribe Trigonini: Trigona (Pissania) cf. cilipes, Partamona sp., Cephalotrigona femurata, Ptilotrigona lurida and Melipona

rufiventris(Meliponini), also visited the flowers, which started to open at about 6:00 a.m. The behaviour of the visiting bees when collecting resin was similar to that observed in C. grandiflora.Due to the more frequent visits, however, the resin was already depleted by about 10.00 a.m. after which the frequency of visits diminished considerably. The relatively large Eufrieseapurpuratabees were able to collect up to 0.068 - 0.070 g (n=3) of resin during one visit. The total resin produced by one flower during anthesis (measured on bagged flowers), was 0.272 - 0.311 g in male flowers (n=5) and 0.236 - 0.289 g in female flowers (n=4), so that E. purpuratabees would need only about four visits to deplete a flower. As the secretion continues after the flower opens, and because there is much competition between bees, the real number of visits is much higher. In contrast to C. grandiflora,some of the resincollecting bees visiting C. insignis also collect pollen from the flowers, generally after the resin was depleted. By marking bees we were able to confirm that the same individuals of Eufriesea purpurata that collected resin during the first 2 - 3 hours of anthesis collected pollen later. Other E. purpurataindividuals collected resin on one day and pollen on the next. Rarely we observed Eufriesea bees that started by collecting pollen and only later switched to resin. The bees of one Euglossasp. were observed to collect only pollen. During the pollen-collecting phase the visits of bees to female flowers may result in mistake-pollination. As in C. grandiflora,bees often arrive at a flower with some resin already in their corbiculae. Of 67 visits of euglossine bees to one particular flower between 6:00 a.m. and 10:00 a.m., the bees arrived 23 times (c. 34%) with resin on their corbiculae, 32 times (c. 48%) without resin and 12 times (18%) to collect pollen. On another day five flowers were observed from 6:00 a.m. until 10:00 a.m. Of 89 bees (Euglossini) recorded, 33 (17%) had resin on their corbiculae, 38 (43%) arrived without resin, and 18 bees (20%) collected pollen. During this period three bees collected resin first and pollen later. Fruit production in C. insignis at the Manaus airport site was 77.2% (n = 259), showing that the bees were efficient pollinators. 1.3 Clusia nemorosa.

The populations investigated near Manaus were exclusively dioecious. Gynodioecious populations of the race with hermaphrodite flowers from Bahia (NE Brazil) were investigated recently by A. Lopes (in prep.). Flower visitors to C. nemorosa in a white sand campina near Manaus were recorded by Mesquita & Franciscon (1995). In the population of Tarum~i-acuthe plants mostly grow on top of a large dune on the edge of a low forest. We counted 49 male and 32 female plants in the population, mostly growing very close together (the distance between male and female plants generally varied between 0 m and 10 m, very rarely between 25 m and 50 m (X = 5.3 m; S = 8.9 m). The male and female

plants grew close

together without obvious pattern. Many seedlings of C. nemorosawere observed below the plants. Flowering lasted from June to October. The floral morphology of C. nemorosais very similar to that of C. grandiflora,but the flowers are much smaller. The resin is golden-yellow, the stigma is green and covered with mucilage as in the other species of the section. In contrast to the large-

626

KEW BULLETIN VOL. 52(3)

flowered species C. grandifloraand C. insignis, the resin plate protrudes above the anthers (Fig. 1J). Bees may land directly on the resin or on the petals. The tips of the awn-like prolongated connective are more or less on the same level as the resin and the bees are powdered with pollen when touching them. As in the other two species of section Chlamydoclusia,the flowers last only one day. Visiting bees started their activity before sunrise (5.45 a.m.) and collected until c. 15.30 - 16.00. The most common visitors to the flowers were female bees of Trigonini (Trigona williana, Cephalotrigona femorata, Duckeola ghiliana, Partamona gr. cupira sp., Partamona sp., Ptilotrigona lurida) and Meliponini (Meliponafulva, Melipona

gr. marginata sp.). Euglossine bees (Euglossa sp.) were only rarely observed, and mainly higher up in the trees. Marking of bees showed that in this population of C. nemorosa the small bees moved between the trees, and may act as pollinators. Meliponini are probably better pollinators than Trigonini, as the latter tended to return to the same flower, especially if the tree was isolated. During one whole morning, the flower of a female tree, some 50 m from the next flowering male tree, was visited by the same Trigona bee (marked on the back) and the flower was aborted some days later suggesting lack of pollination. The difference in pollinating bees between C. nemorosa and C. grandiflora or C. insignis is most probably due to the

generally very small distances between the trees. The smaller amount of resin in the flowers may be responsible for the rarer visits of Euglossini which may prefer the plants of the large-flowered species occurring in the same region. Hand-pollination of flowers bagged before anthesis with a counted number of pollen grains yielded the results listed in Table 3. We found that a single developing seed may be sufficient to trigger the development of a fruit, although such fruits were smaller than normal. The relation of pollen grain number to number of seeds obtained varied (see Table 3), but we found a ratio of approximately 2 pollen grains to each seed developed. Pollination of only one stigma with sufficient pollen mostly leads to the fertilization of ovules in all locules: of 11 flowers pollinated, 8 formed normal fruits and 3 were aborted. The transmission tissues of the different stigmas unite in the ovary and must allow the distribution of pollen tubes to all locules. Fruit-set in the population was 88.7% (n = 186) suggesting that the small Meliponini/ Trigonini bees were efficient pollinators.

2. Sect. CordylandraPlanch. & Triana (emend. P. F. Stevens ms.) This section is characterized by a mixture of pollen and resin in the male flowers (Fig. 2A, E, G), acutely papillose or hairy stigmas (Fig. 2F), the absence of sepals, petals and staminodes in young fruits, and the bony endocarp of the capsules (P. F. Stevens ms.). Pistillodes are present or absent in male flowers. There are nearly always five stigmas, but their number may be reduced in the pistillodes of the male flowers. For details of the morphology of the species studied see Table 4. 2.1 Clusia leprantha.

The population investigated by us grew near a small lake along the Manaus Caracarai road. It consisted of one female and 2 male trees, all growing within a few metres of each other. The flowers are open upwards. The morphology of male

627

FLORALBIOLOGYOF CLUSIA FROM CENTRALAMAZONIA

TABLE3. Relationship between number of pollen grains used to pollinate the stigma of flowers of Clusia nemorosa(coll. nr. 91/25) and the number of seeds obtained from the mature or submature fruits. Number of pollen grains

Number of seeds

2 3 4 5 6 7 8 12 21 24 27

0 0/1 3 2/2 2/3/4 2 6/7 5 10 9 15

TABLE 4. Flower and fruit characters of the species studied of Clusia sect. Cordylandra Planch. & Triana C. leprantha

C. renggerioides

C. pana-panari

Flowering period

XI - II

IX - III

V - VII

Flowers per inflorescence

9- 15 (d), 1(- 4)(9)

5 - 30 (),

Flower diameter

3 - 5 cm

2 - 2.5 cm

1.6 - 1.8 cm

3 - 7(- 13)(9)

3 - 14(d),

1 - 5(9)

Flower form

cup-shaped

bowl-shaped

bowl-shaped

Flower colour

white outside/ brown-violet inside

white with red centre or reddish

yellowish or tinged with orange

Stamens

very numerous

20 - 30(- 40) (Morph 1) (9 -) 10(- 12) (Morph 2)

c. 50 (in C-Amazonia)

Form of androecium

conic-crateriform

disciform (Morph 1) 1 series (Morph 2)

disciform

Staminodes (9 fl.)

10(- 12)

(7-)10(-

5

Pistillode

present

absent (in C-Amazonia)

Fruit maturity

after c. 12 months

absent (Morph 1) present (Morph 2) after c. 12 - 14 months

Fruit size

4.5 x 4 cm

2.0 x 1.5 cm

9 - 16 x 10 - 12 mm

Fruit colour

green

blackish-red

green

Seed colour

red

whitish

whitish

Aril colour

orange-red

orange-red

orange-red

11)

after c. 4 months

628

KEW BULLETINVOL. 52(3)

and female flowers is quite different, especially in the androecium. In the male flowers the stamens are connate and form, possibly with the participation of the receptacle, a conic-crateriform androecium around the fungiform pistillode which has a large stigma on top (Fig. 2A). It is noteworthy that this strange form of androecium has evolved in parallel in C. mexiensis P. F. Stevens ined. (= C. arrudea Planch. & Triana ex Engl. nom. illeg.) of sect. Phloianthera, although in this species the pollen is not mixed with the resin. In female flowers of C. leprantha only one row of staminodes is arranged around the large pistil (Fig. 2B). The stigmas of the pistil and the pistillode are acutely papillose. The stamens of the male flowers secrete a rather fluid yellowish resin in which the pollen is suspended, forming clouds. On very hot days the resin-pollen mixture may run down to the base of the androecium or even into the flower base. The staminodes of the female flowers secrete a clear viscous resin at their apex, in which the antherodes are visible (Fig. 2B). The flowers started opening only after sunrise, and about one hour after sunrise the visiting bees were able to enter the flower. At this time, the male flowers already offer considerable amounts of resin, while the female flowers are still almost without it. Resin secretion of male flowers ends around noon, but in female flowers secretion continues until about 16.00. The male flowers were very frequently visited by Ptilotrigonalurida bees which generally land on the stigma of the pistillode and often collect resin while they are sitting on it. A smaller Trigonasp. and a very tiny unidentified bee visited much more rarely. Ptilotrigona lurida visited the female flowers much less often than the male flowers. In contrast, the small Trigonabees collected preferentially from the female flowers, possibly because they were chased away from the more rewarding male flowers by the larger Ptilotrigona lurida bees. Only in the afternoon, when the Ptilotrigonalurida bees had left the plants, were the smaller bees observed on the male flowers. Only around noon and even then rarely, were the Ptilotrigona lurida bees observed to visit the female flowers, where they generally collect resin from the staminodes while sitting on the stigma (Fig. 2B). We never observed euglossine bees as flower visitors at any site, Trigoninibees were the only ones. The rather small amount of resin, and its more fluid consistency in the male flowers, are possibly responsible for the absence of larger bees. A single female tree near the "Cachoeira alta do Tarumfi" set fruit although the nearest male plant was about 500 m away. As this is rather a large distance for Trigonabees, which were the only visitors, we suspect that the fruit set was due to pollination with pollen from C. renggerioides,of which several trees visited by the same Trigona bees grew much nearer to the female plant of C. leprantha. From experiments with cultivated plants in Campinas (Bittrich & Amaral unpubl. data) we know that Clusia species even of different sections are easy to hybridize. All crossing experiments involving species of sect. Cordylandraresulted in fruit set, with seeds generally germinating without problems. 2.2 Clusia renggerioides.

Plants of this species are very common near and even in Manaus at places where the primary forest has been destroyed, and can often be found along roads. They occur in wet and dry sandy places, e.g. in white sand campinas, often forming large

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

629

populations with the trees generally growing close to each other. In the population near Manaus airport, we counted 116 females and 146 males (see below for more details) in an area of 500 x c. 30 m. No obvious pattern for the occurrence of female and male trees was observed. The distance between a female plant to the closest male plant varied from 20 cm to 15(- 28) m (X = 7.2 m, S = 6.0 m, n = 95). The flowering season was from September to the end of February. Older trees may flower for more than 4 months. The flowers are reddish or white with a red centre of varying size (Fig. 2C). They emit a smell reminiscent of Indian incense. Male flowers only last one day and a few fall early in the morning before the resin has been removed. Female flowers last up to three days, although the petals become increasingly brownish and the receptivity of the stigma diminishes (see below). Visiting bees recorded were Trigonini:Trigona (Trigona) fulviventris, Ptilotrigona lurida.

Trigona (Trigona) cf branneri, Trigona (Trigona) williana, and

The morphology of the female flowers is similar to that of C. lepranthawith a large gynoecium in the centre surrounded by a single row of staminodes. The stout columnar staminodes have antherodes at their apex and secrete a clear viscid resin around these (Fig. 2F). Two different types of male flowers were found in the populations near Manaus. Each male plant showed only one of the flower morphs: Morph 1. Flower without pistillode (Figs. 2D&E). There are generally (20 -)25 - 30(- 40) stamens forming a 5-sided, more-or-less flat androecium, although on one tree some flowers had as few as 7 stamens. The filaments are stout and the anthers lie almost horizontally on their apex. They open by longitudinal slits. The apex of the stamens secretes a rather fluid resin on the day of anthesis in which the pollen becomes intermixed (Fig. 2E). The visiting bees collect resin while sitting on the androecium. Morph 2. A pistillode is present. The flower closely resembles the female flower, as there is only one row of stamens around the central fungiform pistillode (Fig. 2G). The pistillode has a large stigma which is only slightly smaller than that of the female flower. The (9 -) 10(- 12) stout columnar stamens secrete resin at their apex which is mixed with pollen. In contrast to Morph 1, the stamens of Morph 2 secrete two distinct types of resin: the peripheral resin channels secrete a more viscous, and the central channels, a more fluid resin or oil. The pollen is mixed with the fluid resin which is almost completely enveloped by the viscous resin. This viscous resin may serve to prevent the more fluid resin-pollen mixture from running down from the apex and becoming unavailable for pollination. This problem does not exist in male flowers without pistillodes as there the anthers form a kind of plate on which the fluid resin spreads out (Fig. 2E). As in the female flowers, bees in a flower with a pistillode generally collect resin while sitting on the stigma. In the population near Manaus airport male plants with flower Morph 2 were slightly more frequent: we counted 81 plants of Morph 2 and 65 plants of Morph 1, while 116 females were recorded. In a population of the campus of the Universidade do Amazonas in Manaus, however, the male plants of Morph 1 were clearly more common (26 individuals) than those of male Morph 2 (10 individuals), and there were 43 female plants. The phenology of resin secretion was different in

630

VOL. 52(3) KEWBULLETIN

the three flower types: the male flowers of Morph 1 started secreting at c. 18.00 on the day before anthesis and finished at about 7.00 a.m. on the day of anthesis. The male flowers of Morph 2 started at c. 23.00 (secreting both types of resin) and stopped around noon. The female flowers started at c. 3.00 a.m. and also stopped at around noon. The behaviour of the visiting bees generally reflected the availability of floral resin. This was especially striking in those cases where the trees were frequently visited, probably when they were close to a nest of the visiting bees. In the case of male flowers without pistillodes (Morph 1) we observed that the bees made many visits to a flowering plant from sunrise until c. 9.00 a.m. and then almost disappeared. In male flowers with a pistillode and in female flowers, the bees collected generally less frequently, but until about 13.00 - 14.00. Female flowers were sometimes visited on the day after anthesis, although no new resin was produced, so that visits might result in mistake-pollination. Both the stigmas of the female flower and the pistillode of the male flower secrete small amounts of a clear fluid (Bittrich & Amaral 1994). This fluid does not contain sugar and is not aqueous, but more-or-less apolar. It most probably serves to dilute resin drops (with pollen inside) deposited by bees on its surface, thus improving the chances of pollen grains coming into contact with the stigmatic papillae. Pollen does not germinate inside the resin, but when resin is mixed with water, pollen starts to germinate at the contact surface of resin and water. The reason for the presence of the more-or-less apolar fluid on the stigma of the pistillode is unclear. If it has a function at all, it could be to place part of the pollen on the stigma of the pistilode, which would then serve as a place of secondary pollen presentation. The stigma of the pistillode in C. schomburgkiana(Planch. & Triana) Engl. possibly fulfils this function (Bittrich & Amaral 1996). Further investigation is required to determine if pistillodes in Clusiagenerally function in this way. Vegetative characters, as well as floral characters other than those of the androecium, left little doubt that the two different morphs of the male plants belong to the same species. Several crossing experiments substantiate this view. Bagged flowers of the same female tree were pollinated with pollen from male trees of the two morphs and the fruit set recorded. The female trees were certainly compatible with pollen of both morphs: of 169 flowers pollinated with pollen from male Morph 1, 159 (94%) set fruit and 10 (6%) aborted, of 157 flowers pollinated with pollen from male Morph 2, 139 (88.5%) set fruit and 18 (11.5%) aborted (for more details see Table 5). The percentage of aborted flowers is low overall, but clearly higher when pollen of Morph 2 was used. This may be either a real difference, or an artefact that arises because the presence of viscous resin makes handling of the pollen more difficult. Unfortunately, the crossing experiments do not prove conspecifity of the morphs conclusively. As mentioned above, recent crossing experiments undertaken by us with cultivated plants in Campinas showed that with very few exceptions crossing of different species of Clusia leads to fruit set and germinable seeds (Bittrich & Amaral unpubl. results). Only the investigation of flowers of the F1 or F2 generation could give a conclusive answer. During the crossing experiments we also investigated the duration of receptivity of the stigmas. Of 56 flowers pollinated on the second day of anthesis 39 (70%) set fruit, and of 23 flowers on the third day six (26%) set fruit.

631

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

TABLE 5. Results of crossing experiments

in C. renggerioides (coll. nr. 91/28), using from the two different male pollen morphs to pollinate the same female trees. Numbers of successful fruitset (+) and aborted flowers (-) are given.

Female tree

male Morph 1

male Morph 2

+

-

+

A B C D E F G H

13 17 6 21 27 18 8 49

0 0 0 3 0 2 1 4

11 6 10 11 35 18 7 41

1 0 1 3 2 2 2 7

total 326

159

10

139

18

Fruit set of C. renggerioidesin the population at the airport was 78% (209 successful fruit sets and 58 abortions). This number shows that the small Trigona sensu lato bees are quite efficient pollinators, at least in this population. 2.3 Clusia pana-panari.

The population on the Campus of the University of Manaus consisted of 7 plants, 3 male and 4 female. The distances between the trees varied between 3 to c. 20m. The flowering period lasted from May to July. The flowers are similar to those of C. but yellowish or tinged with orange and are somewhat smaller. renggerioides, The androecium of male flowers in the population studied by us consisted of c. 50 stamens with gross columnar filaments, but over the whole distribution area the number of stamens in this species varies between five and c. 80 (Bittrich unpubl. data). The anthers open by 4 short slits. We only found flowers without a pistillode, although flowers with a pistillode are reported in the literature (Eyma 1934, Maguire 1966) and were seen by us on herbarium specimens especially from eastern Amazonia. We do not know, however, if the two morphs may occur in a single population, as observed in C. renggerioides(see above). As in the male flowers of C. renggerioides without pistillodes, the male flowers of C. pana-panari secrete a rather fluid resin mixed with pollen at their apex. The female flowers have only five thick staminodes which secrete clear viscous resin at their apex around the antherodes. Resin production started at c. 23:00 in both male and female flowers. Bees of the tribe Trigonini,predominantly Ptilotrigonalurida, were almost the only flower visitors. The bees visited the flowers, sometimes in large numbers, from c. 7:30 a.m. until

17:30 - 18:00. Very rarely Melipona seminigra (Meliponini)

was

recorded. As in the two species described above, Trigonabees land on the stigma of the female flower and generally collect resin while sitting on it. Some bees also collected latex from wounds inflicted by us on the leaves of one of the trees.

632

KEW BULLETIN VOL.

52(3)

The Trigonabees were often observed to disturb each other during flower visits, but it is unclear if they belonged to the same nest. The behaviour of resin-collecting Trigoniniseldom appeared straightforward and it would be interesting to investigate if they follow some "optimal foraging strategy". GENERAL

DISCUSSION

Small bees are efficient pollinators of certain Clusia species (Bittrich & Amaral 1996). In the species of sect. Cordylandrabees of Trigonini seem to be especially important (Bittrich 1993). These bees are abundant on flowers of many tropical plants, but they are often nectar or pollen thieves rather than efficient pollinators (e.g., Renner & Feil 1993). This is true also in the case of the large-flowered C. grandifloraand C. insignis where the trees grow at greater distances from each other and offer large amounts of resin. Here the small bees mostly returned to the same flower or at least to the same tree. Their ability to chase away euglossine bees hovering below a flower may sometimes enhance pollination. For species with smaller flowers and small amounts of resin, but growing in more or less dense populations, the small bees become important pollinators. Larger bees may also find the rather fluid resin of male flowers of sect. Cordylandra difficult to collect. Unfortunately, we know nothing about pollination in those species of sect. Cordylandrathat grow as hemiepiphytes in dense forests. Do several plants of one species grow on one tree or at least on neighbouring trees? If the plants grow far from each other do other bees, stronger fliers than Trigonini,act as pollinators in this habitat? The resin-offering flowers of the sections Chlamydoclusia and Cordylandra show one basic difference in their male flowers. In the former section pollen and resin are clearly separated within the flower, the resin being secreted by special staminodes in the centre of the flower. In the latter, the stamens themselves secrete the resin and the pollen becomes mixed with it. It seems plausible to assume that the evolution of resiniferous Clusia flowers began with latex which may at first have leaked from accidental wounds in the stamens inflicted by pollen-collecting bees, while later the plants evolved to produce resin and secrete it in a controlled manner (cf. Armbruster 1984). It is not clear why latex was substituted by resin during the evolution of this reward. It appears possible that it was more difficult to secrete the fluid latex in sufficient quantities in a defined place since it tended to spread out inside the flower. Furthermore, it is probable that the simultaneous presence of latex or resin in the flowers soon resulted in pollen becoming stuck to these substances and thus lost for pollination (Bittrich & Amaral 1996). In different lines in Clusia, different solutions to this problem have evolved. In the first Clusia flowers offering floral latex/resin, both pollen and latex/resin were probably present on the stamens. One strategy to reduce loss of pollen would be to avoid contact between the sticky substance and the pollen. Examples can be found in C. schomburgkianaand C. columnarisEngl. (see Bittrich & Amaral 1996). Another strategy is described in this paper and is found in species of sect. Cordylandra:the pollen is mixed with rather fluid resin in the male flowers. The stigma may secrete a fluid which dilutes the resin further, improving the chances of pollen from inside the resin coming into contact

633

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

with the stigmatic papillae. In sect. Chlamydoclusiaresin-production is transferred from the stamens to staminodes, resulting in a wide separation of pollen and resin. The same strategy can be found in some species of sect. Phloianthera,but in this case the fertile stamens are in the centre of the flower and the resin-producing staminodes at the periphery (Fig. 3A). The central androecium protrudes well above the staminodes, so that the bees mainly land on the stamens, as seen in the male flowers of sect. Chlamydoclusia.In a few other species of the section Phloianthera,like C. microstemon Planch. & Triana, resin is mixed with pollen, paralleling the situation in the male flowers of species of sect. Cordylandra.In the other two genera of angiosperms where resin has evolved as a reward for pollinators, stamens and resinproducing organs are separated. In Clusiellathe resin is secreted by staminodes at the base of the androecium (Fig. 3B), while in Dalechampia it is secreted by modified bracteoles (e.g., Armbruster & Webster 1979). As we discussed in a previous paper (Bittrich & Amaral 1996), the variety of androecial morphology in Clusia is especially striking in flowers with resin as a reward for pollinators. The special problems connected with such a reward might have promoted an evolution of androecial morphology in the genus which is rare elsewhere in the angiosperms. On the other hand, it is also striking that in other Guttiferae,especially in the mainly palaeotropical genus Garcinia, there is a wide range of androecial morphology. It therefore seems probable that the variability produced by some genetic mechanism and offered to selection is especially pronounced in this family. TAXONOMY

Subdivision of the genus into sections was first proposed by Planchon & Triana (1860) and followed with some modifications by subsequent authors. The sections are mainly based on characters of the androecium, less often on characters of the staminodes of the female flowers, where number, arrangement, and presence or absence of sterile anthers are used. Taking into consideration the wide variation found especially in the androecium of Clusia male flowers, this is hardly surprising. As the androecial characters supposedly reflect different pollination syndromes or strategies, one could expect that these at least partly fit the sections of Clusia. Our observations mostly fulfil this expectation. However, Planchon & Triana (1860) and subsequent authors, while occasionally mentioning the presence of floral resin, were unable to use its presence and place of secretion systematically for their generic subdivison, because of the difficulty in observing such characters in herbarium specimens, and in distinguishing it from the latex present in many floral tisues. It is often impossible to observe in herbarium specimens whether the resin was secreted by staminodes or stamens, and if the pollen was mixed with resin, or separate. Parallel evolution additionally complicates the recognition of natural taxa. This explains, for example, the inclusion of Clusia leprantha(= C. purpureaEngl.) in sect. Phloianthera next to C. mexiensis (= C. arrudea) by Engler

(1888)

and Vesque

(1893). The male flowers of both species show an obconic-crateriform androecium with a pistillode in the centre (cf. Fig. 2A). In the first species, however, which in fact belongs to sect. Cordylandra(bracts, sepals and staminodes caducous; one series

634

KEW BULLETIN VOL. 52(3)

of staminodes; acutely papillose stigma; bony endocarp; cf. P. F. Stevens ms.), the resin is secreted by stamens and the pollen is mixed with the resin. In the second species secretion is by staminodes at the base of the androecium, and the pollen remains separated from the resin. On the whole, however, the sections defined in the last century seem to be natural taxa and the pollination syndromes and strategies so far observed agree well with them. We have described (Bittrich & Amaral 1996) different pollination strategies in the sections Polythecandra, Omphalanthera, and Clusiastrum, and here we have described those of sect. Chlamydoclusiaand Cordylandra.Furthermore, beetles were observed as pollinators on the white or yellowish, very aromatic pollen flowers of sect. Criuva (sensu lato), some of which are reminiscent of certain Annonaceae(Correia et al. 1993, pers. obs.), and we have observed flies taking nectar from the small yellowish flowers of Clusia melchioriiGleason of sect. Anandrogynein Bahia. The pollination biology of species of many sections, however, is still unknown or needs to be studied in more detail. Likewise, the phylogenetic relationships of the sections are completely unresolved. Thus we know neither which reward is primitive in the genus, nor whether the species with resin, pollen or nectar flowers form monophyletic groups. ACKNOWLEDGEMENTS

We are very grateful to the INPA and especially the staff of the Botany Department for their hospitality and support of various types during our work. We are especially indebted to Marcos Antonio dos Santos for additional observations and for collecting fruits resulting from our crossing experiments. We wish to thank Dr U. Diihrsen for critical reading of the manuscript and and corrections to the English. We thank Francisco J. A. Peralta (INPA) for identifying most of the bees. This work was supported by a grant of the Volkswagen Stiftung for V. B. and a CNPq grant for M. C. E. A. REFERENCES

Armbruster, W. S. (1984). The role of resin in angiosperm pollination: Ecological and chemical considerations. Amer. J. Bot. 71: 1149 - 1160. 1 & Webster, G. L. (1979). Pollination of two species of Dalechampia in Mexico bees. (Euphorbiaceae) by euglossine Biotropica 11: 278 - 283. Baker, H. G. (1976). "Mistake" pollination as a reproductive system with special reference to the Caricaceae.In: J. Burley & B. T. Styles (eds.). Tropical trees: variation, breeding and conservation, pp. 1"61- 169. Academic Press, London. Bittrich, V. (1993). Pollination biology of some Clusia species from South America. Meeting of the Association for Tropical Biology, San Juan, Puerto Rico, pp. 45 46 (abstract). & Amaral, M. C. E. (1994). O papel de liquidos apolares na biologia de polinizaao de Clusiaceae.XLV Congresso Nacional de Botinica do Brasil, 1994, p. 258 (abstract). (1996). Flower morphology and pollination biology of some Clusia &- species from the Gran Sabana (Venezuela). Kew Bull. 51: 681 - 694.

635

FLORAL BIOLOGY OF CLUSIA FROM CENTRAL AMAZONIA

Correia, M. C. L., Ormond, W. T., Pinheiro, M. C. B & de Lima, H. A. (1993). Estudo da biologia floral de Clusia criuva Camb. Um caso de mimetismo. Bradea 6: 209 - 219. Engler, A. (1888). Guttiferaeet Quiinaceae.In: C. P. F. von Martius et al. (eds.), Flora Brasiliensis 12 (1), pp. 381 - 486, Mfinchen. Eyma, P.J. (1934). Guttiferae.In: A. Pulle (ed.) Flora of Surinam. Vol. 3(1): 65 - 108. De Bussy, Amsterdam. Hammel, B. (1986). New species of Clusiaceaefrom Central America with notes on Clusia and synonymy in the tribe Clusieae.Selbyana 9: 112 - 120. Maguire, B. (1959). A revision of Clusia sect. PolythecandraP1. & Tr. of the Guttiferae. Bol. Soc. Venez. Ci. Nat. 20: 363 - 370. S(1966). The genus Clusia (Guttiferae)in Surinam. Acta Bot. Neerl. 15: 63 - 75. (1976). Apomixis in the genus Taxon 25: 241 - 244.

Clusia (Clusiaceae) -

a preliminary

report.

Mesquita, R. C. G. & Franciscon, C. H. (1995). Flower visitors of ClusianemorosaG. E W. Meyer (Clusiaceae)in an Amazonian white-sand Campina. Biotropica 27: 254 - 257. Oliveira, C. M. A., Porto, A. M., Bittrich, V., Vencato, I. & Marsaioli, A. J. (1996). Floral resins of Clusia spp.: Chemical composition and biological function. Tetrahedron Lett. 37: 6427 - 6430. Planchon, J. E. & Triana, J. (1860). Memoire sur la famille des Guttiferes. Ann. Sci. Nat. Bot., S6r. 4, 13: 306 - 366, 14: 226 - 367.

Renner, S. S. & Feil, J. P. (1993). Pollinators of tropical dioecious angiosperms. Amer. J. Bot. 80: 1100 - 1107. Richards, A. J. (1986). Plant breeding systems. George Allen & Unwin, London. Roubik, D.W. (1989). Ecology and natural history of tropical bees. Cambridge University Press, Cambridge, New York. Steiner, K. E. (1985). The role of nectar and oil in the pollination of Drymonia serrulata (Gesneriaceae) by Epicharis bees (Anthophoridae) in Panama. Biotropica

17:

217 - 229. Stevens, P. F. (in press). Clusiaceae.In: K. Kubitzki (ed.), The Families and Genera of Vascular Plants, Vol. III Dilleniidae.Springer Verlag, Heidelberg & New York. - (ms.). On the relationships of DecaphalangiumMelchior and RenggeriaMeissner, with comments on androecial morphology and evolution in the Clusiaceae. Vesque, J. (1893). Guttiferae. In: A. & C. de Candolle (eds.), Monographiae Phanerogamarum, vol. 8. G. Masson, Paris. Vogel, S. (1984). Blfitensekrete als akzessorischer Pollenkitt. In: F. Ehrendorfer (ed.), Mitteilungsband Botanikertagung in Wien 1984, p. 123. Institut fir Botanik, Wien.