NAT. NAT. HIST. BULL. SIAM Soc. 43: 337-365 ， 1995

ECOLOGICAL ，MORPHOLOGICAL AND TAXONOMIC STUDIES ON THAILAND'S FIFTH SPECIES OF : ZIPPELll (BLUME) SPACH

Hans Banziger*

ABSTRACT

A1 1 22 c1 usters of Rhizanthes zippelii studied in τbai1and (where it is confirmed for the

first first time) and in W M a1 aysia pぽ asitized exc1usively pedunculare (W all. 巴x Laws.) P1anch. P1anch. () a1 though other Tetrastigma spp. were a1 so present ，some reported as hosts

elsewhere.τbe mo 中 hology and eco1ogy of the host 釘 ed 巴tai1 巴d. Most1y its only

connection connection with the ground was by aerial roots and R. z伊'pelii gener a1 1y grew on 血ese ne 紅 ground ground leve l. Morphologic a1 evidence indicates 出at R. lowi (B 巴cc ぽ i) H 紅 ms ，described from buds ，is best synonymized with R. zippelii ， a rather variable species. From extrapo1ation of growth growth measurements ， buds would need some 200 days to reach a circumferenc 巴 of 3.7 -4 .1

cm (when 出ey break through the host b訂 k) from a swelling of 0 .5 cm diameter ， and another

200 200 to maturity when 14.7-20.5 cm in circumference. Fl owers start to open 釘 'ound midnight and and are fully open arQund noon the following day (a1 though insects first visit during ear1y moming) and remain fresh for 2-3 days. Th e mush-like pollen is gener a1 1y dep1eted in 白 e first day day but pollinating f1i es (C a1 liphoridae ，Diptera) continue to visit for many days. 百le pollen mush c1 0ts and hardens once smeared by f1i es onto their bac k. Experiments show 白紙 h釘 d-

ened ened pollen mush retains substanti a1 gen 凶nability for m 釦 y days 組 d 白紙c1 0tting is readily

reversed reversed when swept over 出 e fluid-soaked stigma. Nectar is exuded ne 紅 the tip of 白e

perigones. perigones. A ‘sωffy room' odour comes 仕om 血e perigonelcolumn 飢 da weaker excrementl cheese-like cheese-like smell from the caudate appendages. R. z伊>p elii is a sapromyophilous flower which acts acts main1y by brood ・site deception.

INTRODUCTION

An encounter in a steamy rain forest with a flowering Rhizanthes z伊'pelii makes 姐 intruder intruder wonder.τ 'h e look of 出e flower is more 北 in to a tentacled animal-a starfish or medusa-than to a member of 出e kingdom. But at 出es 創 ne time the reddish globe with crater ，embedded 泊 a tangle of rufous hairs ，reminds one of the blood-shot orifice of a furry mammal. Th is pe 甲lexing aspect is p制 of 組 intricate set of lures: visual ，tactile ， olfactory olfactory and gustatory ， as treated in a second study on 血e plant' s pollination syndrome (BANZIGER ， 1996). The present state of know1edge of Rhizanthes Dumortier has been aptly reviewed by MEUER & VELDKAMP (1988) ，with data on its nomen c1 a加ra1 history ，mo 中 hology and anatomy (e.g. SOLMS-LAUBACH ，1876 ，1898; HE 町 RICHER ，1905; CAMME 乱 OHER ，1920) ， seeds (SOLMS-LAUBACH ，1874 ，1898) ， pollen morphology (TA Kl汀 AJAN ET AL. ，1985) ， cytology (町凡， 1933) ，geographic dis 凶 bution 阻 d， as far as the sometimes deteriorated and

*Department *Department of Entomology ，Faculty of Agriculture ， Chiang Mai University ， Chiang Mai ，Thailand ， 50200

337 338 HANS BANZIGER incomplete incomplete museum material allowed ，detailed taxonomic description. Pr eviously known ，among else ， mainly under Blume ， at present two species 紅 e 閃 cog 凶zed ， viz. R. zippelii (Blume) Spach and R. lowi (Becc 訂 i) Harms ， di 姐 ibuted in Suma 回， W ，Bomeo and ， for R. z伊llelii ， also Java (MEUER & VELDKAMP ，1988). 百le present study proves 血e presence of R. z伊llelii in Thailand for the the first time ，a finding which is not entirely unexpected as 由e species is included in a list of 百凶plant n創 nes(SMIτ 1N AND ， 1980). However ， Dr. T. Smitinand informed me 白紙 he had never seen 出e pl 叩 t ，despite his life-long field experience; 白紙 the record originated from his predecessor; and 出at 白ere is no herbarium material from Thailand (if any had ever ever been collected). Furthermore ， the record remains dubious also because one of the local local names listed was 出e same as that for kerrii M 吋民 b，ua phud (前朋) with which it might have been confused ， while another is bua khrang (u 'lf'l f.:J). The present study increases increases to four the numb 巴r of genera of Rafflesiaceae present in Thailand ， the other three being being Gri 白血 (wi 血 two included species) ，Mitrastemma Makino and Rafflesia R. Brown (wi 白 one 百聞 species each) (HOSSEUS ， 1907; HANSEN ，1972a ， b; MEUER ， 1984; BANZIGER ， 1991). Less Less is known about the biology of Rhizanth 自由an the still poorly understood Rafflesia Rafflesia and Sapria while Mitrastemma has been comprehensively researched by WATANABE (1 936 ， 1937). The bud phenology of Rafflesia and Sapria is known 旬 some extent extent (MEIJER ， 1984; ELLIOπ ，1990 ，1992) ， as is 白 e pollination biology of Rafflesia (BEAMAN ET 札.， 1988; B 加 ZIGER ，1991) 出ough it is poorly known in Mitrastemma 叩 d unknown in Sapria. Seed dispersal and host infection are st i1l a matter of sp 田 ulation for all all of them. For Rhizanthes ，among the least known of the nine genera of Rafflesiaceae ， even mo 中hological aspects and 紅 'e far from elucidated. Thi s is perhaps due to its its variability ， complicated by the peculi 紅 presence of bisexual as well as male and female flowers ， as a1r eady noted by SOLMS-LAUBACH (1 876) and HE 別問CHER (1 905). Otherreasons are are its secretive mode of life and a disconcerting ， to some people even repulsive aspect ， with with hairs which look like they could sting 姐 d appendages which ， during the ope 凶ng phase phase of the flower ，resemble 出e trap spikes of a carnivorous plan t. This did not endear it it to local people or the occasional collector. The most detailed biological notes so far have been on the longevity of 2 or 3 flowers and 白eir odour ，made on four occasions by F. Bartel (HE 問問C田 R， 1905). Several authors have have commented on the smell produced by the flower. Visiting insects were identified as carrion carrion flies ，仕 uit flies ，gnats ，ants ，etc. DELP 町 o (1868) attempted a 缶百 t interpretation of of the pollination mechanism.

STUDY SITES AND HABITATS

百】 e study 紅 eas consisted of 白ree sites in evergreen rain fores t. The most northerly was in Su 拙註in Di s凶ct ， Narathiwat Province ， South Th ailand. The other two were between Gopeng and Chenderiang ， South Perak ，Peninsular Malaysia ， the general 紅 'ea mentioned by MOLESWOR' 百 I ALLEN (1 968). Site Site (a) 也 S Th ail 叩 d was along a 5-15 m wide s悦組n in a2 0- 50 m broad gully meandering meandering southwards to westwards ， at 26 0- 340 m msl (mean sea level). 13 clusters of STUDIES ON RHlZANTHES ZIPPEUI (RAFFLESIACEAE) 339

R. R. zippelii-four only with dead buds or desiccated flowers but possibly with live thalli inside inside the host-were dis 位ibuted over a distance of about 1 km. The clusters were less 出an 3 metres from the stream though two were about 10 m from it.百 le forest (Fig. 1) was variably shady with estimated total daily sunflecks reaching the clusters from a minimum of of 112 to a maximum of 2 h. Site Site (b) (shown to me by Mr. M. Wong) was a north-flowing ， 15 m broad ， shallow 抑 'eam bed s位ewn with stones ，gravel and various at 330 m ms l. The wa 町 flowed between between and below them; after heavy showers 白ey would be briefly be inundated. Much of of the 訂 ea was covered by a species of Ar aceae. Despite careful checking of 血e surroundings surroundings only two clusters of R. zippelii were found ，30 m ap 紅 t， both in the stre 創 n bed ，where the vegetation was most open. Thus ， while the soil was perpetually wet ， the habitat habitat received more light 白血血e average understorey vegetation ， with sunflecks reaching the the flowers for about 2-3 h daily.

Site Site (c) was in a sep 紅 ate valley about 2 km from site (b) and consisted of 7 clusters (2 (2 shown to me by an Orang Asli ，5 found by mysel f) s住'ewn over about 112 km ， in a n紅 TOW gully along a stream ，35 0-4 50 m msl ， exposed SE to SW.τ 'h e habitat was essentially the the same as that of (b) but more shaded. Sunflecks nevertheless reached the clusters for up to 112 h daily. One cluster was on rocks in the stream bed ， the others on adjacent gravelly gravelly b創 lks a few metres from the stream. τ'h e 訂 ea of study in S Th ailand is under the influence of the east coast weather regime -heaviest -heaviest rain in November and December ， followed by a relatively dry period in March- May with low rainfal l. During 30 March to 17 April 1995 only 4 late aftemoon showers were experienced. Mid-day temperaωres were 26- 30 "C and relative humidity (RH) was 65-90 % (quite high considering that it was the ‘dry' season ， obviously the result of humid ， nearly constant ， though slow wind drifts down along the stream). 百le minimum night night temperature was 21 "C at 100% RH. Humidity is of paramount importance for the clotting clotting of the pollen mush (cf. chapter on pollen). At the sites in Perak ， exposed to the west coast weather regime ， there is a period with low rainfall in January and February and ag 泊n in June and July when there 訂 e generally m 組 y days without rain. However ， the periods dぽ ing my studies in J佃 U 釘 y and February 1994 1994 and 1995 were unusually wet with most days cloudy at least in the aftemoon and with rainfall rainfall during more 出an half 出e 26 days of research (凶 ght rains not considered). Relative humidity humidity was generally 85-100% but when sunflecks reached the clusters it was reduced to to about 75% ， and occasionally was as low as 659 もwhen wind came from sunny ridges instead instead of down the humid gully. Between 0930 and 1930 h temperatures were 22-25 0 C. Th e rainy weather experienced was unfortunate as a single shower destroys the reproductive ability ability of male R. zippelii flowers by washing away the pollen; furthermore ，pollinator activity activity is reduced. 百le quest for R. zippelii in Th ailand began on 18 March and the last observation was on 18 April ，1995. 百le Malaysian sites were visited on 9 days during 24 January to 3 February ，1994 ，阻 d 16 days during 15 to 31 Janu 訂 y， 1995. Fl ower watching for pollinators was c紅 ried out during 29 days for a total of 140 h， about half of which were at the Th ai sites. sites. Unfortunately ，none of the three sites is in a protected area. As during other studies on endangered species in Th ailand ，1 worked alone and refrained from asking locals about 340 HANS BANZIGER

the the possible presence of the plant in their 紅白to avoid giving 白epl 佃 ta marketable value. My discovery of 出e plant was the result of systematic search for 也e appropria 飽 host liana in in suitable habi ぬts. Th e study sites in Malaysia were relatively safe thanks to M 工M. Wong and the Orang Asli who were aw 紅 'e of the advantages of prot 田伽g 血em. Collected Collected ma 回 ial is at present at 血e author' s dep 紅回路nt but eventually it will be deposited deposited at the Forest Herbarium ， Bangkok.

HOST -PARASITE REL A: τ10NSHIP

Hosts Hosts infected by Rh izanthes zippeui

SOLMS-LAUBACH (1 876) mentioned Tetrastigma papillosum 但I. )Pl 組 chon (as Ciss 凶 papillosa papillosa (V itaceae)) as the host of R. zippelii in Java. 町凡(1 933) reported it on the root of of a Villebrunea sp. (U凶 caceae) 泊 Suma 回. MEIJER & VELDKAMP (1988) added T. lanceolarium lanceolarium (Roxb.)Planchon ，which is also 出e species most often infected by Rafflesia 血ough ， according to LATIFF & MAT-SALLEH (1 991) and Latiff (pers. comm よ白 is name is is aju 凶or synonym of T. leucostaphylum (Denns t.)A1 ston. However ，more recen t1 y Lati 質 (pers. (pers. comm.) concluded that the appropriate n釘 ne for 由is most 食'equent host species of rafflesias rafflesias is T. tuberculatum B 1. 1 have used Tetrastigma sp. 12 for 白is taxon (e.g. BANZIGER ， 1991). MOL 回 WOR' 百 ALLEN (1 968) reported R. lowi ー corrected to R. zippelii byME 田 R&VE 印 KAMP (1 988) - on Kadsura lanceolata K 泊 g (Schisandraceae) a fmding which is questioned by MEIJER & VELDKAMP (1 988). In fact ，Mo 回 SWORTH ALL 聞 'S 凶 cing of 血e host 加出e tangle of was uncertain. Fu 地 ermore ， as shown 泊 a very different different s佃 dy 但ANZIGER ，1989 ， Fig. 15) ，some Schisandraceae like Kadsura heteroclita (Roxb.)Craib ， have stems with corky ridges ，a feature typical of some Tetrastigma spp. ， which might have caused confusion. R. R. lowi was mentioned by MEIJER & VELDKAMP (1 988) as occurring on T. dubium (Laws.) (Laws.) Planchon ， T. glabratum (B l. )Planchon and T. papillosum ，白ough not on T. lanceolarium. lanceolarium. In In my study are 出， both 泊百lailand and Malaysia ， the hosts of all 22 clusters of R. zippelii zippelii were T. pedunculare (W al l. ex Laws.)Planchon (e.g. coll. No. 1202 ，1307 ，1308 ， 1328 ， 1330)σ'igs. 15-17). 百lI s species ，known from Malaysia ，Suma 仕a and Bomeo ，is herewith herewith newly recorded for Th ailand ， where 1 found it in Yala ，Pa 位制 and Nara 由iwat provinces provinces coll. (e.g. No. 1321 ， 1327). F町出ermore ，it is also a new host record. At 血 e sites in Malaysia ， an estimated 1 泊 2-4 T. pedunculare lianas were infected wi 由 R. zippelii; in Thailand 白is was much lower.

Morphologi 伺 1 and Ecological Notes on the Host Plant

Below is a description of T. pedunculare in the live state. Special attention is given to to stem characteristics ， a feature often omitted 泊 previous descriptions. Stems 釘 ep 紅白叫arly useful useful for species recognition by 也 e ecologist in the field. The stem of the liana ，round 泊 cross section and up to 4 cm in diameter ， is relatively soft soft and lacks corky ridges or plates so often seen in other Tetrastigma. 百le bark is very ST U D IES O N RHI ZAN TH ES ZIPP E LII (R A FFL ES IACEAE) 34 1

Fi g llI 巴 1. R. zi pp elii in it s typica lh abitat in S Thai land .A n owe r in full bloo l11 is at l eft bott o l11 ，f ive

lar ge blld s 31 巴 in th e bac kg rollnd 342 H ANS B ANZIGER

Fi gure. 2. A freshly opened male fl ower of R. zippelii. S Pcrak, Malaysia.

Fi gure. 3. G lillering droplets o f nectar soak th e ramenta and pan of 1he tuft hairs near the tip of three pcrigoncs. ST U DI ES ON I? HI ZANT H ES Z IPP E LII (R AFFLES IACEAE) 343

Fig山 e.4 . A l1l plllla w ith rin g 0 1' anth 巴!日 of l1l ale I? zippelii .No t巴 th e in 巴glll ar b ancl of ye ll ow poll 巴n 11l1l sh co ns isti ng of l1l e rgecl cl rop lets

Figur e. 5. A l1l pllll aa ncl sti g l1l u tic f，日 cia in fCl1l alc I? zipp e/ ii . Note th ew hit e papill a巴 344 H ANS B ANZIGER

Fi gure. 6-11. Opening phases of R. zippe!ii. 6. Bud 12 h before opening; note the diverging bracts. 7-11. Buds/flowers about 2, 4, 6, 8 and 9 h after the opening started at midnight. STUDIES ON RHIZANTHES Z1 PPE llI (RAFFLESIACEAE) 345

血in ， of greyish to brownish colour except where covered by lichens (infrequent in other species species of 出e ) and is more or less lenticellate. Tissue immediately below the bark is is red. 1 have found 白is feature 0凶y in few Tetrastigma ，e.g. in T. laoticum Gagnep. ， Tetrastigma Tetrastigma sp. 22 (coll. No. 1199). 百le stem is more elastic than in other species of 血e genus. 百le sound emitted by the liana when bent ， the ‘crac k1 ing' so typical for Tetrastigma (and (and some other Vitaceae) ，is somewhat less distinc t. A single large liana may produce dozens of aeri a1 roots hanging down in s凶 ight threads threads to the ground ， from as high 出 20 m 泊血ec 佃 opy ， to just a few cm above the ground. 百le aeri a1 roots (Fig. 17) ，often on1 ya few mm in diameter ， can grow as thick 出 the the stem itself ， for which 血ey 紅 'e then often mistaken. If they are tom ， secondary aeri a1 roots roots develop at or ne 紅 the wound. Typic a1 1y ，aeri a1 roots do not emit any ‘crac k1 ing' sound sound when bent ， are rather more reddish below the bark and gener a11 y more strongly lenticellate 由加 the stem. The lenticels can be quite conspicuous ，occasion a1 1y like sm a1 1 tuber c1 es which might lead to confusion with the stem of T. papillosum by the non-speci a1i st. 百le aeri a1 root system resembles 白紙 found in Tinospora crispa (Lour.) Merr. ， Tinospora baenzigeri baenzigeri Forman (Menispermaceae) (BANZIGER ， 1982) but in T. pedunculare it is more impressive: impressive: large 01d specimens have lost a1 1 contact with the ground via the prim 紅 y root system ，出 e stem growing high up and extending over sever a1 tree crowns; the only connection connection wi 白 the ground is by 血 e roots. aerial None of 白e many specimens found in S Th ailand had direct contact with the ground through prim 紅 y roots 組 d in M a1 aya on1 y very very few did. An other important feature is the production of later a1 runners ， present to v紅 ying extent extent a1 so 泊 a number of other Tetrastigma spp. ，e.g. qu 日2drangulum Craib & Gagnep. ， obovatum Gagnep. ，laoticum ，tuberculatum ，curtisii (Ridley) Suesseng. ， hookeri (Laws.)Planchon ， as well as other lianas such as Parvatia brunoniana Decne. (Lardizab a1 aceae) (BANZIGER ，1989 ，1991 ，組 d unpub l.). In a single growing season T. pedunculare pedunculare 's lateral runners can cover a distance of 25 m. At various interv a1 s a1 0ng the runner ，rootlets develop and if the runner is severed when old enough ，it develops into an independent independent plant. If it is assumed that after 5 ye 訂 s or so a new plant is large enough to produce produce its own runners 組 d 白e seq 田 nce is repeated for 50 years ，出回出e plant may have 'travelled' 'travelled' some 250 m 出 a separate but genetic a1 1y identic a1 p1an t. It is probable that 血e p紅 asite ‘hitch-hikes' a1 0ng with it and that nearby c1 usters of R. zippelii belong to genetic a1 1y identic a1 individu a1 s， derived from a single origin a1 th a1 1us. 百le leaves 釘 e3 ・foli01ate (Fig. 15) ，crenate ， with very broad le af1 ets (wid 血 more 白血 ha1 f 白e length) ， with asymmetrical later a1 1eaflets and short petiolules (less 由佃1. 5cm long). long). The flowers (Fig. 16) are greenish-yellow ， very fragrant ， the cyme generally dichotomous ， often bom on the old stem ， but mos t1 y high up. The fruits (Fig. 15) are slightly slightly compressed dorso ・ventr a1 1y ， dark red when mature 姐 d，atι8 mm in diameter ， among the sm a11 est in Tetrastigma. τ'h ey tasted sour. Th e seeds ，yellowish to greyish to dark dark brown ，紅 'e 3-5 mm in diametre ，more or less cordate ， with v紅iably crenate m 紅g泊s 組 d typic a1 1y with aT -shaped endosperm. Further taxonomic features are as mentioned in LA' 百FF (1 984). In In South Thailand many individuals bore profuse fruit during late March to Apri l. In Perak Perak on1 y one pl 組 t had fruits but it a1 so had flowers at ぬe same time in mid-J 釦 U 紅 y. 346 HANs BAN Zl GER

It It would therefore appear 白紙 in bo 血Th ailand and Malaya the liana fruits in 血e 合y season season and 也at the simultaneous presence of 企uits and flowers in the Perak specimen may be due to 血e presence of two 官 y' se 酪 ons 血ere ， the second being in June-July ， presumably when the January flowers will bear 白血企凶t. 1 found the liana at 200 ー700 m elevation but but could not check higher up. It w 出 more common at 20 0-4 00 m. It lives 泊 a very specific specific habitat ，namely along peren 凶al streams of small to average size. Individuals often crossed crossed the 抑 eam in the canopy ， the aerial roots hanging down on bo 白 sides of ， and into ， the the stream.

Populatioo Populatioo Structure of Rhizanthes zippelu 00 its Host

In In S 官1ail 組 d R. z伊'P elii grew exclusively on aerial roots of the host ，all at ground level 'except a few which were not more 白血 30 cm above ground. In Malaysia all buds ofthe ofthe two 1紅 'gest clusters similarly developed only on 白e very large aerial roots (白 e stem of of bo 白 hosts ， without prim 訂 y root connection with the ground ，was no less 白血 15-20 m above the ground at the lowest point). Most of the 15 buds of another cluster developed on 血e stem of the host and some on the aerial roots. In the remaining 6 clusters all buds developed developed on roots but it could not be determined whether on prim 紅 y or aerial ones. Most buds buds were just below ， at or just above ground level ， but some were nearly 1m above ground ground and 5-6 m from the po 泊t where the aerial root entered the ground (after coming down in a wide c町 ve and following the ground for several meters). Two small clusters were were on roots in close contact with the rock facies of large bolders.τ 'h e cluster largest ， near near a ta1 1 tree ，was so thickly covered by leaf litter 白紙 only the flowers which had pushed aside aside the leaves while ope 凶ng ，and 由e tip of some large buds ， emerged. 百1e size of the host root on which R. zippelii grew ranged 企'om 1 t04cm 泊 diameter. Buds could be so close each other 出 to mutually impair growth and flowering. Two of the 22 clusters were solitary ， the others occurred 加 groups of 2-12 clusters ， 出e dist 佃 ce between clusters being 1ι30m.τ 'h e smallest recog 凶zably live cluster had o凶y one live bud (besides withered flowers) ， the cluster largest (site (c)) had ， over an 釘 'ea of of 1m 2， at least 30 small to large buds and 10 出1y ones still embedded in host tissue; nearby nearby satellite clusters of the s創 ne host had about h叫f as many additional buds. To ta1 number of buds seen 泊 Thailand was more 血阻 100 組 d in Malaysia well over 170. Four Four of the 13 Th ai clusters had only old ， long since withered flowers and it is not clear clear whether the parasite 血allus was still alive inside the hos t. Th e remaining 9 clusters had live buds and four had also flowers (only fresh enough ones still visited by flies considered considered here). All 9 Malaysian clusters had live buds and 6 had flowers. Skeletonized ， black black remains of flowers which must have withered months before were present in most clusters. clusters.

Considerlng Considerlng all flowers (fresh and old: 29) and buds (15) analyzed ， the sex ratio w 儲 1 female to 3 males. In 5 clusters both sexes were present ， in one there were 2 males ， in 4 only one individual could be sexed and in the remaining clusters the sex of the parasite could could not be established. If reinfection of a host is indeed as r紅 'e as is generally assumed-but this is by no means sure 一白e sex 他国bution would indicate 白紙 R. zi] フ'P elii is is likely to be monoecious. A similar siωation apparently applies to S. himal の， 'ana 但山0π ， 1992 ，組d own observ.) but possibly not to Ra .fJl esia where clusters tend to be of one sex only. only. STUDIES ON RHlZANTHES ZIPPEUI (RAFFLESIACEAE) 347

MORPHOLOGICAL AND PHYSIOLOGICAL NOTES ON RHI; 日NTHES ZIPPELII

Functional Functional Morphology of the Flower

The flower morphology is treated mainly from the functional aspect because of its importance importance for the second p訂 tofmy sωdyon 出e pollination. Nevertheless ，mo 中home 凶 c data data (Tables 1-3) 紅 e included because they might prove useful for the clarification of the taxonomic taxonomic status of R. zippelii and R. lowi (cf. discussion). No significant difference was noted noted between the Thai and Malaysian plants ，especially when considering the overall variability variability of the species. Diarneter Diarneter of the whole flower (Fig. 2) is 11-17 cm ， not including the brownish-red ， filiform ， caudal appendages 2.5-5.1 cm long attached to the tip of the perigone lobes at a right angle (geniculation). The 16 perigones are generally detached from adjacent ones about about in the middle of their length though some do not sep 紅 ate at all while others sep 紅 ate nearer nearer the base. Various swellings (calli) may be present on the distal half of the perigone. Near Near the tip of 出e perigones there is a narrowly triangular ，nectariferous pad 0.8- 1. 1 cm long long and 0. 6- 1 cm wide on which the rarnenta are se t. On the basal half 出e perigones form form a camp 佃 ulate tube which intemally merges with the circumarnbulator ， a circular channel channel (cf. Figs. 4， 5) running round the base of the column. (The circumambulator is so so called because pollinators walk in it around the column.) 31-52 brownish-red radial lines lines stretch from the base of the pale circumarnbulator to the upper tube where they fade into into the more brownish coloration of the distal p訂 t of the perigones. At the center of the flower flower is the pale yellow column consisting of a stalk about 0.5 cm long and a globular head head 1. 2-2.1 cm wide and 0.7 ・1. 8 cm high. The upper half of the globe is formed by the brownish-red brownish-red arnpulla (Figs. 4， 5) with a crater depth of 0.8-1.6 cm. The lower half of 白e globe consists of the reproductive organs (described below). The flower is connected to to the host via the ‘neck' ，a section containing the ov 訂 y (rudimentary in males) ， set between between the insertion of 出e perigone bases and the cupula (the disk-like attachment of flower flower and host). The neck is covered by 3 whorls of 5 black bracts (scales). Females Females are distinct from males in the following characters (cf. Tables 1-3): Generally shorter shorter caudate appendages (2-3.3 cm vs. 2.5-5.1 cm) ，shallower crater depth (0.7-1 cm vs. vs. 1.1-1. 6 cm) ，larger (higher and wider) globular head (1. 4ー1. 8 and 1. 6-2.1 cm vs. 0.7- 1.1 and 1. 2- 1. 7 cm) ，叩d more massive neck (3.2 -4.4 cm vs. 2. 0- 2.9 cm) which is also differently differently shaped. Narnely ， in the female the walls are parallel or ，more often ，slightly converging converging upwardly so that the neck is narrowest just below the perigone insertion. In the male male neck the walls converge downwardly (cf. Fig. 6) ， thus it is n紅 rowest further down near near where the last whorl of scales is attached. There There are four types of hairs. (1) Finely pointed ，straight ，sp 紅白 hairs up to about 1 cm long. Brown in colour ，they 訂 e on the outer wall of the 紅 npulla (Fig. 4) and span 白e gap gap between this and the furry hairs of the tube wall. It would appear that their main function function is ， thanks to a certain stiffness ， to ke 巴p ‘unauthorized' insects - mainly flies such as as Muscidae ，Platystomatidae ， etc.-from intruding into the gap where they might adversely affect affect pollination ，e.g. getting sωck ，sc ぽ ing off pollinators ， or stealing pollen. Blow flies are are robust enough to deflect these hairs and deeper. penetrate (2) ‘Fuπy' hairs ， cinnarnon brown in colour ，up to 1. 1 cm long ， with hooked ，bifid ， 348 HANS B 馴 ZIGER

Table 1. Morphome 凶c data of R. zippelii 企om site (a) ，S Th ailand. Sizes in cm.

Specimen code Female Female Male Male Male Male Male Male flower flower flower flower flower flower flower flower bud 5.W 9.1 3.V 5.Z 5.1 5.2 5.8 11. 2

Diameter Diameter of flower (gen- 17 11. 5 16 14 .5 14 13 .5 14 岨・・圃ーー iculation iculation to geniculation)

Circumf 巴rence of bud ーーー司胃 ー畠晶ー・・ 13.6

Length Length of caud a1 appenda ・ 2-3 .3 2.3-2.9 ?-5 .3 3-4 3-4 3-4 2-3 2-3.2 ges ges (geniculation to tip)

Width Width of perigone lobes 1.1 -2.0 1ル1. 5 1. 9 1.1 -2 1.3-1. 6 1.3-1. 9 1- 1. 6 ーーーーー

Column height (base to 2.2 1. 9 :damaged 1. 6 1. 45 1. 4 1.3 1. 25 crest crest of ampulla)

Wid 血 of stalk of column 1. 2 0.8 0.8 0.8 0.8 0.8 0.8 0.6

Am pulla height (reddish p紅 t) 0.8 0.7 ~amaged 0.6 0.5 0.5 0 .5 0.6

Extern a1 diameter of ampulla 1. 81 1. 6 ドamaged 1. 7 1. 6 1. 5x 1. 7 1. 4x 1. 5 1. 35 (at (at maximum width)

Diameter Diameter of ampulla crest 0.95xl damaged 0.95xl 0.9xl 0.8x 1.1 0.75xO.9 0.7 (crater (crater aperture)

Crater Crater dep 白 (crest to bottom ， 0.9 damaged 1. 5 1. 5 1. 5 1. 2 1. 2 measured measured intemally)

Wid 血 of stigmatic fascia or 0.75 d創 nag 吋 0 .4 0.3 0.3 0 .4 0.35 annular annular row of anthers

Number of an 血ers ーーーーー ー・・ーーー damaged 48 51 50 45 ーーーーー

Wid 由。f neck 4.4 3.5 2.6 2.9 2.5 2.8 2 .4 1. 9

Remarks: Remarks: 1 stigma wider 白an ampulla: 2.1 cm. Specimens 5.W ，5.Z ，5.1 ，5.2 飢 d 5.8 紅巴 組合'om 血e same cluster cluster while 3.V ， 9 .1 and 11. 2 are a1 1 from different clusters. Table 2. Morphometric data of R. zippelii from site (b) ， Malaysia. Sizes in cm.

Specimen Specimen code Female Male Male Male Male Male Male Male Female Female Female flower flower flower flower flower flower flower flower flower bud bud bud 0.3 0.3 0.1 0.2 0 .4 0.5 0.6 0.7 1. 5 1. 2 0.8 0.9

Diameter Diameter of flower (gen- 13 4 12.5 15 .5 13 14.5 12 .5 13.5 ーーーーー ーー骨ー・・ ーーーーー iculation iculation to geniculation) ∞同 ddu Circurnference Circurnference of bud ーーー圃ー ーーーーー 国・・ーーー ーーーーー 圃ーーーー 胃ーーーー ーーーーー ーーーーー 17 .4 13.6 14.5 目的。 Le ngth of caudal 2.8 3.5 -4 .5 damaged 3.5 -4 .5 damaged 3-4 2.5-3.5 3.5 -4 .5 2.7-3.1 1. 9-2.3 1. 5-2 appendages appendages (geniculation to tip) Z 』自民主『均一白 Width of perigone lobes 1.1-1. 8 1.1-1. 8 0.7- 1.3 1. 2- 1. 9 1- 1. 9 1. 2- 1. 8 1- 1. 7 1- 1. 8 ーーーーー ーーーーー ーーーー-

Colurnn Colurnn height (base to 1.8-2.0 1. 55 1. 5 1. 6 damaged 1. 45 1.3 1. 6 1. 9 1. 75 1. 6 crest crest of ampulla)

Wid 出 of stalk of colurnn 0.9 0.7 0.7 0.8 0.7 0.8 0.7 0.7 0.9 0.95 0.95 旬、、同ト旬、(河〉明白弘明白〉の開〉開)N Ampulla height (reddish part) 0.8 0.6 damaged 0.6 damaged 0.5 5 0.5 0.5 0.7 0.6 0.6

Extemal Extemal diameter of ampulla 1. 8x2 1. 5 1. 2 1. 5 damaged 1. 5 1. 3x 1. 4 1. 4 1. 75 1.7 5 1. 8 (at (at maximum width)

Di ameter of ampulla crest 0.9x 1.1 0.9xl damaged 0.8 damaged 0.9 0.7xO.85 0.85 0.95 0.85 0.9 (crater (crater ape 目ure)

Crater Crater depth (crest to bottom ， 1. 0 1. 3 1. 4 1. 5 damaged 1. 3 1. 2 1. 2 0.8 0.7 0.6 measured measured intemally)

Width of stigmatic fascia 0.7 0.3 5 damaged 0.3 5 damaged 0 .3 0 .3 0.35 0.65 0.63 0.6 or or annular row of anthers

Number of anthers ーーーーー 36 damaged 41 damaged 39 46 41 .幽ーー・・ー 帽ー・・ー" ー・・"ーー

Width of neck 3.5 2.3 2.0 ー・・ーーー 2.0 2.0 2.0 2 .4 3.1-3 .4 3 .4 3 .4 い品。

Remarks: Remarks: aI1 specimen 釘 e from the same cluster. wuo一 Table 3. Morphome 凶c data of R. zippelii from site (c) ， Malaysia. Sizes in cm.

Specimen code Male Male Male Male Male Male Female Female Male Male Male flower flower flower flower flower flower flower bud btid bud bud bud 2.2.0 2.2.0 2 .4 2.5.2 2.5.1 2.5.28 2.7.1 2.5.3 2.5.30 2.6 2. 1.3 2. 1.1

Diameter Diameter of flower (ger ト 14 14.5 14 13 13 13 ーーーー・・ ーーーーー ーー・・ーー ーーーーー ーーー-- iculation iculation to geniculation)

Circurnference Circurnference of bud ー---- ーーーーー ー・ーーーー ーーーーー ーー・・ーー ーーーーー 15.7 10.7 16 15.7 15.7 Le ngth of caudal 3-4 3.8-5.1 damaged 3-3 .5 2.5 -3.5 4 2.6 1. 7 3-3.8 2. 4- 3 .4 3. 0- 3 .4 appendages appendages (geniculation to tip)

Wid 出 of perigone lobes 1.1-1. 9 1.3-1. 9 1. 5 1. 2- 1. 5 1.1-1. 7 1.1-1. 6 ーーーーー ー-ーーーー 圃ーーーー

Colurnn Colurnn height (base to 1. 5 1. 5 1. 6 1. 35 1. 45 damaged ーーーーー ----ー ーーーー. ーーーーー ーー司ーー 国〉 crest crest of ampulla) Z 間切』 Wid 白 of sta 1k of column 0.95 0.7 0.7 0.8 0.7 damaged 0.8 0.65 0.7 0.8 0.7 ZN5 Ampulla height (reddish part) 0.7 0.5 5 0.6 0.4 3 0.5 0.6 0.6 0 .4 0.4 5 0.4 5 0 .4 5 開 MN Extemal Extemal diameter of 創 npulla 1.7 1. 4 1. 4 1. 6 1. 4 1. 5 1. 9 1. 65 1. 45 1. 5 1. 5 (at (at maximum width) Diameter Diameter of ampulla crest lx 1.1 0.9 0.8xO.9 0.8xO.9 0.9xl msect 1. 0 0.6 0.8 0.8 0.8 (crater (crater ape 巾同 d創 naged Crater Crater depth (crest to bottom ， 1. 6 1. 2 1. 2 1.1 0.95 msect 0.7 0.4 5 1. 2 1. 2 0.8 measured measured intemally) damaged Width of stigmatic fascia 0 .4 0 .4 0 .3 0 .3 0.3 5 0.35 0.6 0.55 0.3 0.35 0 .3 or or annul 訂 row of anthers

Number of an 出ers 47 40 damaged 48 46 damaged ーーーーー ーーーーー ーーー四ー ーー国圃曲圃 ーーーーー

Wid 白 of neck 2 .3 2.2 2.3 2 .4 2.1 2.2 3.2 2.5 2.0 2.0 1. 9

Remarks: Remarks: specimens 2.5.1 ，2.5.2 ，2.5.3 ，2.5.28 and 2.5.30 紅 'e from the same cluster ，as 紅 e 2. 1.1 and 2. 1.3， while 2.2.0 ， 2 .4， 2.6 and 2.7.1 are al1 from d江f巴rent clusters. clusters. STUDIES STUDIES ON RHlZANTHES ZIPPEllI (RAFFLESIACEAE) 351

trifid trifid or multi-branched endings. 百ley are wildly intertwined (Figs. 4， 5) in a tangle covering covering the perigone from the base of the circumambulator to 出e 加 ft hairs in a somewhat loose ，wholy mat 0.2 ー0.5 cm thick. The branched endings obvioulsy mutually anchor 血eir position. position. The tangle leaves a circular gap of 0. 4- 0.8 cm between it and the column. 1 propose propose that this tangle rnirnics ， at least tactically ， the fur of a mamma l. At the same time it it also works as a device for efficient pollen acquisition and delivery: it elastically presses intruders intruders of the circumambulator against the anthers or stigma when they negotiate the circular circular gap. (3) (3) Tuft hairs (BECCA 町 's (1869) ciuffi) ，somewhat darker ，shorter (0.8 cm) ， though thicker thicker than 出e fl 町可 h姐 S 組 d generally lacking branched endings so that they do not form an intertwined tangle but stand out more or less straight upw 釘 d as well as laterally ， somewhat like in a bundle of wheat (Fig. 2). 百ley are found between the furry hairs and the the ramenta and cover the width of 由e perigone for about 0.5 cm. Most authors have not distinguished distinguished these hairs from type (2) probably because they worked with preserved flowers; flowers; in this the hairs' peculiar posture is los t. 1 propose that they function as a barrier-not barrier-not very efficient perhaps-against visitors (except the strongly-built calliphorids) in in order to reduce attempts to proceed to the centre of the flower where they might cause damage or disrupt the pollination process. AIso ， since some of them get soaked in nectar ， they they may increase volatilization and hence improve attraction; many insect visitors fly only to to the nectar. (4) 百le r創 nenta (antler hairs) ， again essentially sirnilar to type (2) but very much shorter shorter (0.1 cm) ，訂 e somewhat thicker-walled ， darker and likewise provided with single or or multiple branchings. Being shorter ，出 ey 紅 e not so intertwined and form a dark brown velvety velvety carpe t. They are distributed distally to the ωft hairs ，on the nectariferous pad (Fig. 3). 3). Their function may be to soak up the nectar and possibly to improve its volatilization.

Anthers and Stigma

All All 29 flowers and 15 buds of R. zippelii exarnined proved to be either male or female. In In regions like Java most R. z伊'P e1ii are bisexual while a few are unisexual ，either purely so so (SOLMS-LAUBACH ， 1876) or with more or less s虻ongly reduced organs of the opposite sex sex (HE 町阻CHER ， 1905). In bisexual plants the anthers are located directly below the stigma stigma from which 白ey are separated by 佃 annular depression. In my populations the yellowish yellowish white anthers are set on the lower half of the globular apex of the column ，just below the ampulla (Fig. 4). The 31-52 anthers are 町 anged in a circular row 0 .3-0.4 cm wide. wide. They 紅 e curved radially as well as tangentially ，resulting in a rib-like appearance of of the row. They have a depression somewhat below the upper end where a drop of pollen mush collects and merges more or less completely with that of adjacent anthers. This conglomeration conglomeration gives rise to a continuous but irregularly thick ring (Fig. 4) of pollen mush which pro 住udes 仕om the anthers ， ready to be sweapt off by pollinators entering into ， proceeding proceeding along ， or leaving ， the circumambulator. Th e free gap between the anthers and 出e furry hairs is about 0. 4-0 .8 cm but it can be elastically widened by intruders as the tangle tangle of fu 町 hairs ，0.2 -0 .5 cm thick ， is not stiff. The correct location of the stigma was pointed out by SOLMS-LAUBACH (1876) but its position position was later rnisinterpreted by most authors until HEIN 町 CHER (1 905) 叩 d MEIJER & 352 HANS BANZIGER

VELDKAMP (1988) re 氾ognized the right position again. The accuracy of this location is also also proved by SOLMS-LAUBACH'S (1876) discovery there of germinated pollen and by my experiments experiments with pollen gen 凶nation (see below). Th e stigmatic surface is a 0.7-1 cm wide fascia arranged around the lower half of 血e globular globular apex of the column ，just below the ampulla (Fig. 5) ー→巴 ssentially in a position corresponding corresponding to 白紙 of 血e 朗自ers.τ 'h e stigma is bright white ， at close look somewhat glittering glittering due to the innumerable pap i1l ae which give it the appearance of a velvety mat ， rather rather similar to the stigmatic area in Ra Jjl es 加佃d Sapria ， though in the latter ones it is more horizontally set and less radially curved. 百le 仕ee gap intruders have to negotiate is is comparable to 白at in the males. An important finding is 出at ， while in most flowers the stigma is covered by a sticky film ， in R. zippelii 姐 d even more so in S. himalayana ， Ra. kerrii (Ra Jjl esia is shortened to to Ra. to distinguish it from Rhizanthes ， R.) and Ra. cantleyi Solms-Laubach ，there is 組 abundant abundant secretion which soaks up the papillae of the stigma ，a fundamental feature which has has not been mentioned by any worker on these pl 組 ts. Th e liquid is not oily as in many flowers flowers but a cle 肌 watery secretion. 百le function of such a stigma with unusual amounts of of fluid is obvioulsy 旬 lique 守合ied pollen clots (see below) on the back of the pollinators

佃 d to ‘sponge 佃 d brush' it off with the pap i1l ae as the pollinator crawls into ，along 佃 d out out of the circumambulator. Additionally ， the secretion stigmatic dilutes and spreads the pollen pollen over a wider area ，reducing crowding of the huge masses of clumped pollen grains ， thereby thereby improving fertilization. This works also when pollinators c紅 ry fluid pollen mush.

Pollen Pollen

Unlike Unlike in most plants ， the pollen gr 泊ns in Ra Jjl esia ， Rhizanthes and Sapria 紅 e suspended suspended in a fluid ma 位以. When attached to 白e anthers or freshly acquired ， the pollen occurs occurs as a mush with a consistency between mayonnaise and milk. In Rhizanthes it is paler and and somewhat more fluid 白血 in Ra Jjl esia 佃 d Sapria. ME 田 R (1 958) 叩 d BEAMAN ET AL. (1 988) assumed that for successful pollination ， male and female flowers must be present present at the same time so 白 at pollinators c叩 deliver pollen on 出e same day. However ， BλNZIGER (1991) proposed 白紙 calliphorids may deliver pollen weeks after 血ey had acquired acquired it. It It seems 白紙 nobody has been aware of some interesting features in the consistency and and gen 凶nability of the pollen mush of 血e 由ree genera: (1) once smeared off from 出e anthers anthers it go 自由rough a more or fast less coagulation ， drying and h紅白凶ng process depending depending on 出e humidity of the outside 泊r; (2) 血e process is reversible under appropriate conditions; conditions; (3) the pollen retains its viability for an extended period of time even in 白e 合 y state. In In S. himalayana ， under average e紅 ly afternoon humidity and temperature in December-Febru ぽ Y (75-85 % RH ，16- 18 0 C)， the surface of a drop ofmush on a microscope cover cover glass becomes covered with a dry film in 5-10 min. Af ter 15-20 min the drop does not not come off when gently wiped by dry fingers. An hour later the lower layers are also dry. dry. In this state the clot has hardened and remains firmly stuck to the back of 出 e pollinator lifelong lifelong unless it is liquefied again. In Ra. kerrii and Ra. cantleyi the pollen also clots but the the speed of 白e process is not the same due to the different microclimate of their respective STUD IES ON RHI ZANT HE SZ IPP E LII (R A FFLESIAC EAE) 353

Figll re. 12. Larg e blld (14 cm cir cll l1l fe ren ce) and two s1ll al1 blld s( lef t one 2.5 CI1l ，ri g ht on巴 3.0 C Ill Cir Clllll -

f巴! 巴nc巴) still enclo se d in hos t ti ss ue

Figllr e. 13. B lI d of 10. 6C I1l Cir ClI lllf el巴nc ec ut to show w hit e hair s Figul 巴. 14. Frllit of R. zipp eli i (22 c m cir c llm- feren ce) 354 H ANS BANZ IGER

Figure. 15. Trifoli olate leaves and ripe fruits of Terrasrigma pedunculare.

Figure. 16. Female inflo rescence of T. pedunculare.

Figure. l7. Aerial roots of T. pedunculare, I c m diameter. Note the lenti cels. STUDIES ON RHIZANTHES ZIPPELII (RAFFLESIACEAE) 355

~ 2.0 :g E ..§.. 1.8 .<:: i 1.6 ~ ..,"' 1.4 ::J .0 1.2

1.0

0.8

0.6

0 .4

0.2 --- 3 5 7 9 11 13 15 17 bud circumference (em)

Figure 18. Relationship between bud size and bud growth rate of R. zippelii at site (a).

e .!:!. 16 Q) c0 Q) 14 eQ; ::J 12 ·~o .., 10 ::J .0 8

6

4

2

100 200 300 400

time (days)

Figure 19. Average bud growth of R. zippelii based on growth rates derived from diagram I. 356 HANS BANZIGER

habitats. habitats. R. zippelii ∞C 町 S 泊 a wanner and generally more humid habitat (22-30 ・C ，85-100% RH) 血叩 s. himalayana and its pollen mush becomes rubbery and not readily smearable 泊 about 1h ，and hard in about 4 h. But the process is ra 由er faster when wind from drier crest 紅 eas (e.g. 65 % RH) reaches the flower ， or especially when exposed to s住ong 剖r cu 町 ents during the pollinators' fast flight ， or when 血ey bask in the sun or fly up 加to 也e canopy or other drier environments. Calliphorid fly pollinators were often caught with completely completely dry pollen smears ， probably acq 凶red 泊由e previous day (s) ， both ne 紅 R. zippelii 組 d Ra. kerrii. τ'h e process is reversible: addition of a drop of water or sme 紅 of stigmatic 自国 d to a clot dried on a slide or pollinator 出orax liquefies the clot and floats the pollen grains in in a matter of seconds. Si 伊 ific 佃 tly ，prelimin 紅 y gerr 凶nation experimen'ωshow 白紙也e pollen of all four species species retains substantial viability in 由e 命y state for many days. Viability progressively declines declines over 出e next 3 weeks (no experiments yet made with Ra. kerrii pollen older 白血 11 11 days). 百lI s is much longer 白an pollen types carried by honey bees which generally last for for less 出回 a day (KRAAI ， 1962; STANLEY & L 町 SKENS ， 1974). Since conventional gerr 凶nation methods were not successful even with 企esh pollen mush ， pollen dried for

S附 ific periods of 伽 e，剖 well 儲 fresh pollen ，on splinters of cover sl 恥 was applied onto 由e stigma of live female rafflesiaceous flowers in 悦 field ，left for 24 h and then examined microscopically microscopically for pollen tubes. To prevent natural poll 泊ation ， the flowers were covered with with screening before ope 凶ng. Gerrnination of R. zippelii pollen occurred on the stigma of of both Sapria and Rafflesia; gerrnination w 部 also successful across the latter two and among themselves. Experiments with the stigma of R. zippelii as gerr 凶nation medium could not not yet be carried ou t. For several reasons (e.g. long distance transportation) the samples of of pollen dried on cover slips could not be left at 出 e site. It is possible 白紙 the different environmental environmental conditions reduced the viability of 血e pollen. Th e remarkable ability of 血e pollen of the three genera to solidify and be reliquefied by profuse stigmatic fluid has two impo 白血 tconsequences. It ensures fmn attachment of large large pollen loa 也 to the vector for long periods. It allows selective delivery ，i.e. it is not lost lost on wrong stigmas when pollinators visit other flowers 一回calliphorids often do- but can can be sweapt off only by a fluid-soaked stigma. Virtually the only misdelivery which could could occur is between Rhizanthes and Rafflesia. In In a number of individuals of R. zippelii ，norrnal-looking anthers were exuding nothing nothing but a colourless ， clear fluid instead of 白e typical yellowish mush. Mi croscopic examination examination showed 也at such drops contained insignificant numbers of pollen grains. 百le drops drops dried 部 transp 釘 'ent clots and it is assumed 白紙出ey 紅 e 也e matrix 泊 which pollen grains 紅 e norrnally suspended. In In R. z伊'pelii pollen mush conglomerates do not drop from the anther to 白e circumambulator circumambulator to be sucked by pollinators ， unlike what 1 had previously assumed to be the the case with Ra. kerrii and S. himalayana (BANZIGER ， 1991). Having seen many more of of these flowers 1 now conclude 白紙泊Ra. kerrii pollen-dropping must be exceptional and in in S. himalayana it might have been an artifact due to handling when cutting open the bud. In In Mitrastemma yamamotoi Makino the pollen is also suspended in a fluid ma 回 x. However ，in 血is species it is not wa 町 soluble but of lipidic base (W ATANABE ，1936). Hence it it is not affected by rain ， unlike in R. zippelii where a single sustained shower will wash STUDIES ON RHlZANTHES ZIPPEUI (RAFFLESIACEAE) 357

off the the off pollen and destroy the flower's 閃 productive capability. In Rafflesia and Sapria the pollen pollen is well protected ，set below and off the margin of the disk ， and also well above the base base of the tube and thus it is not exposed to rain. In M. yamamotoi nothing is known about about pollen clotting and its viability over time. However ，it can be assumed that it does not not clot ， as is the case with the lady slipper orchid Paphiopedilum villosum (Lindley) Stein where where the pollen ma 位ix is also lipidic (B沿~ZIGER ， in press) and the pollen mass retains its its very viscous consistency for many weeks. It is interesting to note that 血e pollen of this orchid orchid retains its viability for at least 8 weeks while glued to the dorsum of its hoverfly pollinators ，leading to normal capsule development when smeared onto the orchid's stigma (B 白河 ZIGER ， in press).

Neclar Neclar

Another Another unexpected finding is the secretion of nectar (Fig. 3) on the pad covered by the the ramenta. Pr of. Dr. S. Vogel very kindly carried out histological analyses of the pads 1 had sent to him and confirmed that they contain nectar glands. The nect 紅 appears in clear ，colourless ，initially more or less discrete droplets which merge and soak up the ramenta ramenta and sometimes part of theωft hairs. It tasted sweet to my tongue and was sucked by a variety of insects r組 ging from the pollinating blow flies to non-pollinating visitors such such as muscid and sarcophagid flies ， honey and stingless bees ，wasps ，ants ，butterflies etc. (B 沿~ZIGER ， 1996). Since the nectar occurs ne 紅 the margin of the lobes ， hence far from where where the actual pollination occurs ， the nectar is functionally somewhat comparable to extrafloral extrafloral nectaries; pollination is not a direct consequence of nectar uptake. 百lI srelocation may be an adaptation to ward off nect 紅 thieves which might disrupt the activity of 出e 位ue pollinators ， and keep away 仕om the flower' s reproduction centre 血e most deleterious ants which which devour nectariferous tissues. Th e nect 紅 was not present in all flowers. For instance ，it was lacking in female 9.1 and and male 5.8 and its absence was not due to depletion by insects (which can occur in a very very short time) ， nor to ants (which c叩 obliterate the tissue in a few hours). It is clearly an an individual lack as observed in flowers kept under vigil from ope 凶ng time. Absence of nect 紅 in some flowers may indicate its reduced importance in the pollination process. Since Since pollinators are successfully lured to the circumambulator by other cues ，nectar has lost lost to a wide extent its pu 中ose. Yet Yet the nect 訂 probably retains two minor functions. As a nourishment it ‘wets' 出e appetite appetite of the pollinators and induces them to remain on the flower ， to crawl around and search search for more ， thereby increasing the chances of approaching the circumambulator. By offering offering a reward it induces calliphorids ， which are capable of a certain degree of flower constancy constancy (KUGLER ，1951) ， to visit another flower - important for pollen delivery onto a female. female. Recently Recently MEUER (in litt.) also noted the presence of a secretion at the antler hairs.

Odours

CAMMERLOHER (1920) studied the stomata of Rhizanthes 加 d Rafflesia and suggested 白at they 釘 e 血e places where the unpleasant odours 紅 'e released. However ，since they 紅 e 358 HANS BAN Zl GER found found exclusively on the underside (outside) of the perigone lobes they cannot be loci of odour odour emanation since 血is occurs on 由.e upperside (加 side). Al so ，insects attracted to 血e flower flower only settle on this side. An other function Cammerloher suggested is 血at they may release release physiological water. However ， because of the almost constantly high hur 凶dity of 出e microhabitat 1 do not think evaporation would be very efficient ， nor have 1 seen droplets droplets of water forming at 血e stomata. M 邸周R & VELDKAMP (1988) described the odour of R. z伊'[J elii ， reported by various authors ， as ranging from being virtually absent to being curiously acid ， to foul and cadaverous. cadaverous. According to HE 悶悶CHER (1 905) the odour was not unple 部組tand at any rate not not carrion- lik: e. Fresh flowers of my population emitted two very different ，locally cle 紅'l y sep 釘 able odours ， as assessed by close sme 1li ng. 百le s佐onger one originated from t1t e tube ， adjacent adjacent p紅 t of the lobe and possibly the colurnn. 1 found it a most perplex 泊g one and for for w 叩 t of a better description 1 would compare it to 恥則的air of a room wi t1t old furniture ，and of a crowded room needing ventilation. Th e weaker odour emanated from the the caudate appendages. It was rather more unpleasant and varied between cheesy to excrement excrement -like. 百le odours were weak and generally perceptible only close to t1t e flower but but occasionally a faint whiff could be felt as far away as 1- 1. 5 m. Th e odours 紅 e therefore quite different from 白紙 of Ra. kerrii (mostly rancid on perigone ，variably fruity in the tube but cadaverous from some distance) and Ra. cantleyi (weaker (weaker in general ， barely perceptible on 出e perigones ，weakly reminiscent of foul eggs in in the tube) (BANZIGER ， 1991 and unpub l.). S. himalayana odour is also cadaverous but rather rather weaker 白血 Ra fJl esia's. To the human nose R. zippelii' s odour was the weakest of the above but to flies it must be the most po 飽 nt as shown in two experiments comp 紅 ing R. zippelii and Ra. cantleyi ， all all four being normally visited by calliphorid flies before the experlments started. In both cases ，one R. z伊'[J elii w 田 cut an placed 50 cm 企om one Ra. cantleyi st i1l in situ. In the frrst 住iall 部 t泊g 35 minR. z伊'pelii was visited by 14 Chrysomya pinguis (W alker) 釦 dC. chani chani Kurahashi and Ra. cantl り，i by only one Hypopygiopsis iゆ， mata 但igot). In血 e second 住ial of 10 min ， 18 C. pinguis and C. chani settled on R. zippelii ，6 on Ra. cantleyi. 百le calliphorids calliphorids were not only more frequent but also more persistent on R. zippelii 血an on Ra. Ra. cantleyi 白ough 也is could be additionally due to 血e former' s perigone tex 細胞 which mimics the fur of am 阻 lma l.

PHENOLOGICAL OBSERV A:. τ10NS ON RHlZANTHES ZIPPEUI

Bud Growth

SOLMS-LAUBACH (1876) studied in great detail the anatomy and mo 叩hology of buds 加 various stages of development ，from its inception on the thallus filaments of the p紅 'asite 血rough to bud ma 加rity ， but its phenology has remained unk:n own. 1 monitored the grow 血 of of 101 buds from 31 March to 17 April 泊百回land and of 63 buds from 16 to 31 Janu 紅 y， 1995 ， in Malaysia. Size increments were measured as circurnference at 血e bud' s widest 臣民 h， but for practical reasons ， in small buds 白is had to be obtained 企'om measured diameters. diameters. STU Dl ES ON RHlZANTHES Zl PPE Ul (RAFFLESIACEAE) 359

Buds frrst appear as small swellings which develop ，still embedded in the brownish host host tissue ， into protruding ，spherical nodules (Fig. 12). When about 3.7 -4 .1 cm in circumference 血ey break 血rough the host tissue. 百ley remain enclosed in the fleshy and pale pale bracts (scales) until about 10 cm in circumference (about 2 months before flowering) when the ac 旬 al bud st 制 s to show below the receding tips of 由e bracts.τ 'h e meridians - the brownish red lines along the fissures where 血e perigones split from each other - become increasingly evident ， as does the bud coloration 錨 a whole from its pale base to the the brownish red top ， while very slowly the bracts darken to eventually become 企y， black 組 d papery 血in at least distally (Fig. 6 ，12). Intemally 出e buds have white hairs (Fig. 13) until until about 11 cm in circumference. Their change to brownish has progressed about half way when about 12.5 cm and is completed when 13.5-14 cm in circumference (some 20 days days before flowering) 出ough parts of the nect 紅 pads and antler hairs 紅 e still somewhat paler 血an the res t. From a circumference of about 13 cm onwards ，it is possible to establish establish the sex of the bud from withou t. By gently bending down the bracts ， one can see 血e shape of the walls of the neck. If it n創 TOWS downwardly ，it is a male; if it widens or remains remains the s鉱 ne it is a female.τ 'h e bud opens at 組 average circumference of 17.0 cm (min (min 14.7 ，max 20.5 cm) (16 buds measured 12 within hours before ope 凶ng ， except one measured measured 2 days prior to an 曲目is; its maωre size was assessed by ex 住apolation). Th e opening of a bud can be predicted within 1， sometimes 2 days ，from the position of of the last whorl of bracts ，viz. its distinct divergence from the neck (Fig. 6) ， at least in males. males. In the female the divergence is not so obvious ， probably due to the broader neck and and its less inclined walls typical of females ， which allow less space for divergence. Cracks along along the meridians ， formed first at 白e top of the bud long before an 出esis ，later widening and and extending down the sides of the bud ， are a further 白ough only very approximate indication 白紙 in a number of days the bud will open. Furthermore ， about two days before opening ，bud exp 佃 sion accellerates 企om 0.13 -0 .18 cm/ day to 0.3-0 .5 cm/ day. An increased bud expansion rate shortly before an 白esis is found also in other Rafflesiaceae. In Ra. kerrii in in S Chumphom Pr ovince the widening of three buds accellerated from a circumference increment increment of 1c m/ day to 3c m/ day 5 days prior to opening ，leading to a maximum bud circumference circumference of 75-85 cm (BλNZIGER ，unpubl ふ Elliott (pers. comm.) found faster expansion expansion also in the last bud stages of S. himalayana. Th e grow 血 rates of 百凶佃d Malaysi 佃 populations co 町'espond closely. From the curve curve (Di agram 2) it follows that when a bud breaks through the host tissue at a circumference circumference of 3.7 -4 .1 cm ，it needs some 200 ー225 days for flowering ， but for the smallest smallest measured buds of 1. 6 cm circumference an estimated 400 days are needed. However ， the the data for the small buds 紅 'e less reliable due to their slow grow 血 and the short monitoring period. period. This is very close to the growth rates measured by EL LlO' π(1992) for S. himalayana ，viz. 111 days from 7 cm circumference to flowering (105 days in R. zippelii). The giant Ra. amoldi R. Brown needs nearly 500 days to grow from the circumference of 4.7 4.7 cm to flowering at 107 cm (ME UER ， 1958).

Anthesis Anthesis

In In Th ailand the ope 凶ng of four buds was followed throughout the night at ho 町'l y intervals. intervals. In 血ree the perigones started to split ap 訂 t around midnight ， in the fourth around 360 HANS BANZIGER

0200 h， and this was completed around noon in all fou r. Th e other 17 buds (from all sites) were were found flowering in the morning after being closed the previous aftemoon; they also must must have opened at 凶ght and the degree of flower expansion of the majority of them was within within the same range. In three it was completed during 090 0- 1000 h. The final position of of the recurved lobes varied from one just below the horizontal plane to a much more recurved recurved position with the distallobe parts more or less vertical and the caudate appendag 回 horizontal horizontal instead of close to ve 凶 cal as in the former. In the latter ， which occurred in flowers flowers not in direct contact with the ground ，出 e expansion period lasted beyond noon. The splitting ap 制 started and was fastest on top of the bud where generally 3-4 wedge-like wedge-like gaps formed (Fig. 7， 8) between groups of fo 町 or more perigones still attached to to each other ，together with their caudate appendages. 百le groups split into s泊gle perigones at at various times later ， though some only split p紅 tly or not at al l. Th e tips of the caudate appendages ， which generally were the last to separate ，started to detach around 030 0-0 400 h when also the odour became more evident and the nectar droplets visible. By 0600-0700 h (Fig. 9) the caudate appendages were more or less horizontal and the lobes already overhanging ， the whole giving the impression of a ‘cage'. Ar ound 0800 h (Fig. 10) the caudate caudate appendages were inclined about 45 0 ， resembling somewhat the spikes of a camivorous camivorous plant ready to snap. By 093 0- 1030 h (Fig. 11) they were vertical and by noon they they were open still more and 白 e reflexing stopped or slowed down. As a comparison ， Ra. kerrii kerrii needs about 24 h to open fully; 2 flowers started to open in 白e morning ， one after 凶ght-fall (BANZIGER ，unpubl ふ The best Thai cluster of R. z伊'/)elii had 3 flowers during 11 days ， the best Malaysian had had 7 during 9 days. One cluster in Thailand had 2 flowers which opened on 白e same day ， both both males. In Malaysia one cluster had 3 males and another cluster had a male and a female female (deformed and incapable of reproduction) which flowered at the same time.

Flower Longevity

As already indicated by Bartel's notes (HE 凹 RICHER ，1905) ， flower longevity is short. Wh en no rain fell ， the flowers looked ‘fresh' for 2， at most 3 days and the odours were weaker weaker and started to deteriorate on the second or 出ird day ，although flies still visited 出e flowers flowers many days afterwards. Af ter 白ree days 批 odour was more mouldy. Pollen Pollen depletion depends on the number of flies entering the circumambulator and their their size and behaviour. 1 found that s位ong fly activity can exhaust pollen supply within a few hours and in all but one flower ，pollen was depleted by mid-aftemoon of the frrst day. day. Thus the chance of pollen washout by potential rain is not great since it is often already already depleted by pollinators. In In the only female flower 1 was able to vigil from opening time ， the stigma was wet during during the frrst day ，合 om e紅 ly morning to eve 凶ng. In another female ， checked in 血e morning morning of its second flowering day ， the stigma was slightly wet but it is not clear whether the the liquid was stigmatic fluid or moisture due to rain or dew of the previous night. In yet another another female ， found when some three to five days old but still visited by flies ， the stigma was 合y. It is conceivable 由at females may be pollinated and fertilized on the second or few few subsequent days with 仕esh pollen mush but not with dried pollen. Rain Rain also causes the tangle of 白汀y hairs to collapse and stick to the perigone surface ， STUDIES STUDIES ON RH/~弘NTHES Z1 PPEUI 侭AFFLESIACEAE) 361

impairing 出e proper functioning of the pollination mechanism. Rain also causes a faster darkening darkening and rotting of the fleshy p紅 ts of the flower which are 血en consumed by various nocturnal nocturnal invertebrates ， leaving only nor トfleshy black tissue behind which can persist for many weeks.

Fruit Fruit

At all 血ree sites there were very few remains of female flowers which rnight have been fertilized and developed fruit (Fig. 14). 1 left them in situ ， but collected two for the s加 dy of the seeds. However ， only one had what looked like not yet quite ripe seeds. Th ey look look very similar to Ra ffL esia seeds.

DISCUSSION

τ'h e populations of Rhizanthes of the three study sites ，which doubtlessly belong to the S創 ne species ， could not be identified unequivocally as either z伊'P elii or lowi with the key of of MEIJER & VELDKAMP (1988) nor by other descriptions of the two taxa. As there may not not be enough justification for upholding two species and since zippelii BLUME (1827) has priority ，1 have opted for 血is name. Synonyrnization had been previously proposed by HOOKER (1 873). In In an effort to resolve 血is problem 1s 旬 died Beccari's descriptions of lowi in 白e orig 泊al and also visited the well curated Herbarium Universitatis Fl orentinae. Unfortunately ，出 e relevant relevant type material is no longer there ，and could not be traced. BECCA 阻 'S (1 868 ， 1869) lowi lowi was not described from an open flower but 仕om mere buds ，and he could comp 紅 e these these only with the description and somewhat rnisleading illustrations of BLUME (1827). BECC 組1' S description (1869) of lowi buds is detailed but his comparison with z伊'P elii is somewhat confusing. He later (1875) reiterated his findings after studying zippelii material from the type locality and rejected HOOKER'S (1 873) merger of 白e two species; this was followed followed by most botanists. Still ， a description based only on buds ， in a taxon lacking vegetative vegetative parts ， is bound to be unreliable and problematic. Furthermore ，much new information information has accumulated during 批泊tervening 120 years.

My view is that 也is Rhizanthes is highly variable. Like m 佃 yp 紅 asites it has undergone undergone reductions ，a process which is probably still in progress as exemplified by the lack lack ofnec 旬r， or the secretion of pollen matrix without pollen ， in some of the individuals. Reductions Reductions resulted in additional and possibly more evident new characters 由加 in 叩 ordinary ordinary plan t. However ，泊 Rhizanthes the characters used for the distinction of zippelii from lowi occur in various combinations and sometimes with transitions. In the past this was not sufficiently appreciated because of lack of material. 針。 hably the main feature which induced Beccari to 白泊k his species w 邸 different from z伊'P elii is 白at according to BLUME (1 827) 也is appears to be 516 ・lobed with bi ・or 佐i・fissured caudate extensions. In reality the lobes and extensions simply did not separate completely completely into the normal 16 p訂 ts ， the fissures being the lines along which the caudate extensions extensions normally detach from each other. 百le presence of 加負 h剖rs in lowi and lack in z伊'P elii ，where Beccari says 由eyare 362 HANS BANZIGER

replaced replaced by the ramenta ， does not seem to be reliable. Th e various typ 邸 ofh 組r formations 紅 'e ， as pointed out by HE 町阻C田 R (1905) in his detailed studies of zippelii ，rather variable. From various illustrations 血e 旬ft hairs appe 紅白 be less conspicuous but not entirely lacking lacking and the ramenta merely somewhat more widely spread in some zippelii. Al so ，1 have have observed 白紙 the hairs 紅 'e sometimes bitten off by ants ，so 出at their absence may be be an artifact of which previous authors were not aware. 百le number of lowi's anthers ， 5ι60 ，is not significantly outside the variation found in z伊'J) elii of 38-50 (31-52 in my populations). populations). B 民 cari's mean deviation of ::l: 5 is suspiciously low (::1: 10.5 in my material) ， probably probably due to his very sc 組 .ty study material. Beccari's other important distinction ，viz. unisexual unisexual lowi and bisexual zippelii ， turned out not to be consistent as shown by SOLMS- LAUBACH'S (1 876) finding of pure males among the material from 出e type locali 旬- HE 町阻CHER (1905) described even transitions with more or less strongly reduced organs of 血e opposite sex. Al so colour seems unreliable taxonomically. According to MEIJER & VELDKAMP (1988) ， zippelii zippelii has pale yellowish-white perigosi 回 and hairs at frrst bloom which soon turn to various various shades of brownish red while lowi is so a1r eady before opening (M日 JER ， in litt.). 1 consider these mere different physiological stages of the same colour development p副 :em. Buds at my sites had white hairs inside while smaller than 11 cm circumference but had brown ones when more 出組 14 cm. MOLESWOR 咽 ALLEN (1 968) described flowers of lowi ，corrected to 均 'J) elii by MEIJER & VELDKAMP (1 988) ，from ne 紅 my populations 制 加itially very pale pink ， with white hairs before tuming d紅k. HE 町阻CHER'S (1905) 企esh zippelii zippelii from near 白e type locality had 'dirty white' perigones basally ， fleshy coloured ones ones distally but pale caudate extensions ， while the hairs were c泊namon. In a colour illustration illustration (ANo 附 MOUS ， 1995) zippelii from Sumatra has cinnamon as well 田 white hairs hairs (difficult to see) on perigones which are bright white throughout except for the nect 紅 pad with the ramenta and caudate extensions which 紅 ed 釘k. MEI 庖 R & VELDKAMP (1988) (1988) illustrated 組、berr 組 t' lowi where the basal half of 血e perigones and its hairs 紅 e white white and the distal ones 也rk. In some of my buds (circumference 11-14 cm) the darkening occurred occurred first on distal and basal perigone p紅 ts ， the middle section remaining pale somewhat longer ， while in others the nect 紅 pads and r釘 nenta stayed pale somewhat longer. To sum up ，evidently the time and the 紅 'ea where parts of the flowerωm from pale to brown are variable variable and occ 町 in different combinations. In conclusion ，it seems to me 白紙， based on the the data we have so far ， there is no clearly defmed ，consistent character which sep 紅 ates zippelii zippelii from lowi and 1 would propose to merge 出 e latter as a synonym into the former. Th e relationship between R. zippelii 叩 d T. pedunculare is of p紅 ticul 紅 interest ， not so so much because the liana is a new host 佃 da new Th ai record ， but because of the p紅出ite's res 凶 ction to a single host species over a wide 釘 ea in Th ailand and Malaysia. It It was not found 泊fecting any of the four other Tetrastigma 組 d two non-vitaceous species reported reported as hosts by other authors. Sign 出cantly ，among them was T. tuberculatum (my Tetrastigma Tetrastigma sp. 12) which was growing at all sites and 泊 some c邸 es only a few meters

from R. zij フ'J) elii. At Malaysi 佃 sites 7 individuals (e.g. coll. No. 1205 ，1206 ， 1232) were infected infected by clusters of Ra. cantleyi. Tetrastigma hookeri coll. (e.g. No. 1309 ，1310 ， 1335) was also present at all three sites. In 官 lailand there was additionally T. curtisii. Neither though though has been reported as hos t. Yet ，泊 spite of the finding of a new host it is well possible 出at future studies may STUDIES STUDIES ON RHlZANTHES ZIPPEUJ (RAFFLESIACEAE) 363

narrow narrow down rather than widen the host range of R. zippelii ， as happened in the case of Rafflesia. Rafflesia. In Peninsular Malaysia ，Latiff (pers. comm.) found that the hosts of Ra. cantleyi and and Ra. hasseltii were all T. tuberculatum which is also the host for most other Rafflesia in in Malaysian Bomeo. 1 made corresponding observations. Ra. cantleyi at sites (b) and (c) are are all on Tetrastigma sp. 12. North of the Isthmus of Kh ra ， Ra. kerrii infects exclusively T. T. quadrangulum (16 clusters seen) but south of Kh ra to the Malaysian border ， in a stretch of of 600 km ，1 have found it parasitizing only Tetrastigma sp. 12 (15 clusters of six populations in in four provinces (BλNZIGER ，unpubl ふ 1 have checked the identity of Ra. kerrii hosts No. 086164 佃 d 086168 (Forest Herbarium ，Bangkok) ， mentioned as T. papillosum (NIYOMTHAM & KUBAT ， 1987) and found them to be two different species ，neither being T. papillosum. The actual host is Tetrastigma sp. 12 while 086164 is virtually certain to be T. curtisii which in S Th ailand frequently occurs together with the actual hos t. Own field surveys in the the area of 086164 and 086168 showed Ra. kerrii on Tetrastigma sp. 12 ， T. curtisii also being being present but not infected. The two have similar stems and can easily be confused. Some Tetrastigma identifications in my paper (臥 NZIGER ，1991) 紅 e not flawless either (their (their primary identification though was not done by Prof. Latiff nor by mysel f): T. lanceolarium lanceolarium should be replaced by T. hookeri. There 訂 e some 57 Tetrastigma species recorded recorded from Malesia (LATIFF ，1984) ， 16 from Thailand (CRAIB ， 1926) and 39 from Indochina Indochina (GAGNEPAIN ，1912 ， 1930). A taxonomic revision of this difficult and st i1l poorly poorly known genus w i1l likely synonymize many of them. R. R. z伊l]J elii is a sapromyophilous flower since its pollinators are necro- 叩 d coprophagous Diptera ，viz. mainly Lucilia porphyrin α(Walker) and C. pinguis together with four further Calliphoridae Calliphoridae besides an additional four as exceptional or potential pollinators. The syndrome is based mainly on brood-site deception as the carrion flies are deceived into laying laying hundreds of eggs ， the hatchlings of which 紅 e doomed to starvation (B 必-l ZIGER ，1996). Th us the presence of a reward ，viz. nectar ， is somewhat unexpected because typically sapromyophylous sapromyophylous flowers do not offer any. In fact ， in the Rafflesiaceae there are only three three other genera with species known to produce a sweet secretion. Their pollination syndrome is not definitively known ， but evidence suggests that none is sapromyophilous.

One is Cytinus ， with hypocistis L. as the only European species of 出e far 国ly 創 nong several several other elsewhere. 百le other is Pilostyles ，richer in species and wider in dis 出 bution. Cytinus Cytinus hypocistis ， P. thurberi Gray and P. hamiltonii Gardner have floral nectaries lying at at the base of the column (HAYEK ， 1912; RUT 田町ORD ， 1970; DELL & BURBIDGE ， 1981) which are therefore ， unlike R. zippelii ，directly involved in the pollination. An unidentified native native wasp is the probable pollinator of P. hamiltonii (DELL & BURBIDGE ， 1981) while a long-tongued hymenopteran is thought to pollinate C. hypocistis (HA YEK ， 1912). In In the third genus ，Mitrastemma ， there is yamamotoi which produces large quantities of of a sweet liquid which is exuded not by nectaries but through hydathodes as a guttation process process (W ATANABE ，1937). 百le hydathodes are located high on the column and the liquid liquid collects mainly at the base of the perigones where it is taken up by such visitors as wasps ，flies and a bird ， Zosterops palpebrosa. W AT ANABE (1 936) considered the insects as as well as the bird to be pollinators but 出is was only inferred as no actual pollen acquisition or or delivery was observed. R. R. zippelii therefore is only a ‘half-de 四 ptive' sapromyophilous flower (sensu DAUM A1叩， 1971) 1971) which is 佃 uncommon type ， as found in some but by no means all Aristolochia ， 364 HANS BANZIGER

such such as grandiflora Swartz (CAMMERLO 田 R， 1923; HIuE， 1984).

ACKNO 羽fL EDGMENTS

τ'h anks 紅 'e due to Dr. W. Brockelman ， Mahidol University ， and Dr. S. Elliott ， Chiang Mai University ， for criticizing and improving the manuscript ， to Mr. G. Li nder ， former librarian librarian at the Swiss Federal In stitute of Technology ，Zurich ，who was able to fmd most obscure obscure references. 1 am p紅白ul 紅 'ly indebted to Prof. Dr. S. Vogel ，Wien U 凶versity ， for his his histological verification of the nectaries. Pr of. Dr. A. Latiff ，Universiti Kebangsaan Malaysia ，Bangi ， carried out the most important identifications of Tetrastigma spp. Ar ac 加。 logist Dr. P. Schwendinger ，Inn sbruck U 凶versity ， had to muster exceptional patience with with my seemingly unending observations at the Malaysian sites but his extensive soil seaving seaving was duely rewarded with many spider novelties. M. Wong showed us site (b).百 le late late Dr. T. Smitinand 佃 d Dr. T. Santisuk ， both of the Forest Herbarium ，Bangkok ， commented on R. zippelii in Thailand and suggested places of botanical interest respectively.I 創 n particularly grateful to Mr. P. Sukumalanand who generously let me use his his computer facilities ， and for his and my other Department colleagues' support of 血e projec t. Miss S. Panthi helped with the table layou t. Dr. W. Meijer ，U 凶versity of Kentucky ， gave gave valuable information about Rafflesiaceae. Pr of. G. Moggi and Mr. G. Padov aI語 kindly let let me see 血e Rhizanthes material at the Herbarium Universitatis Fl orentinae ，Fl orence ， lta1 y. Th e National Research Council ，Bangkok ， extended research approvals (No. 125/35 and and 39 /3 8).

REFERENCES REFERENCES

ANoNYMOUS ， 1995. Rafflesiaceae ，in: 百le Wo r1 d of Pl 組 ts ，part 40. Asahi Shimbun ，22 January ，p 98-102. BANZIGER ，H. 1982. Fruit-piercing mo 由s(L ep. ，Noctuidae) in 百lail 叩 d: A gener a1 survey and some new perspec- tives. tives. Mitt. Schweiz. Ent. Ges 55: 213-240. BANZIGER ，H. 1989. Lardizab a1 aceae: New plant family for Th ailand ‘predicted' by rare mo 白 on Doi Suthep. Nat. Hist. Bull. Bull. Hist. Siam Soc. 37: 187-208. BANZIGER ， H. 199 1. Stench and fragrance: u凶que pollin 組 on lure of 百凶land's 1紅 gest flower ， Meij 釘 . Nat. Hist. Bull. Siam Soc.39: 19-52. BANZIGER ， H. 1996. Pollination of a flowering oddity: Rhizanthes zippelii(Blume) Spach (Raf flesiaceae). Nat. Hist. Hist. Bull. Siam Soc. (in press). BANZIGER ，H. (in press.). 百 le mesmerizing w副: pollination ruse of epiphytic lady slipper orchid Paphiopedilum villosum villosum (L ind.) Stein (Orc hidaceae). Bot. よ Linn. Soc. No. 120. BEAMAN ，R. S. ， P. J. Decker AND J. H. Beaman. 1988. Pollination of rafflesia (Rafflesiaceae). Am er. J. Bot. 75: 1148-1162. 1148-1162. BECCA 則， O. 1868. Descrizione di 出 nuove specie pi di 組胞 Bomens i. Atti Soc. ltal. Sci. Nat .1 1: 197-198. BE<叉 ARI ，O. 1869. Dl u抑留ione di nuove specie di pi 佃旬Bomens i. Rafflesiaceae. Nuovo Giom. Bot. ltal.l: 84- 91 (+ (+ Table 5). BECCARI ， O. 1875. Osservazioni so 戸富a1 cune Rafflesiacee. Nuovo Giom. Bot. Ital.7: 70 ー75. BL 剛 E ，C. L. 1827. Over een nieuw Plan 蜘 geslacht ，de Brugmansi a， uit na de 畑町Ii jke f: 醐 lie der 貼 izan 由eae. In In H. C. van Hall ，Bijdr. Natuur k. Wetensch. 2: 41~22. CAM 阻 RLO 田 R， H. 1920. Der Spaltoffnungsapparat von Brugmansia und Rafflesia. Osterr. bot. Ze itschr. 69: 69: 15 3- 164. CAMME 阻 β 田 R ，H. 1923. Zur Biologie der Blute von Aristolochia grandiflora Swartz. Osterr. bot. Ze itschr. 72: STUDIES ON RH lZANTH ES Z1 PPEUI 侭AFFLESIACEAE) 365

180 ・198 (+ Table 5-7). CRAIB ，W. G. 1926. Florae Siamensis Enumeratio. Siam Soc. ， Bangkok. DAUMANN ， E. 197 1. Zum Pro blem der Tauschblumen. Preslia (Praha) 43: 304- 317. DEu..， DEu..， B. AND A. H. BURBIDGB. 198 1. No 回 on 臨 biology of Pilos. 砂les (R afflesiac 倒的 in Wes 加盟 Australia. West. West. Austral. Herb. Res. Notes5: 71-79. Dm.P INO ， F. 1868. Ul teriori osservazioni e ∞， nsiderazioni sulla Di cogamia nel regno vege 旬le. Atti Soc. Ital. Sci. Nat. Nat. 11: 26 5- 332. E 比 Iσπ ，S.19 卯.百 le dis 凶butio n， status and ecology of Sapria hi. 閥均ana Gri ff. (R afflesia 印刷泊百ail 組 d. British British Ecol. Soc. Bull.11: 24 6- 249. E 比 Iσπ ，S. 1992. 百 e sta 伽s， ecology and conservation of Sapria himalayana Gri ff. 侭affI esiaceae) in 百 ailand. よ Wildlife in li 加 i加 ul2: 44- 52. GAGNBPAIN ， F. 1912. Flore generale de I' Indo-Chine. M 錨 son ，Paris. GAGNBPA 町， F. 1930. Flore generale de l'Indo ・・ Chine. Supplemen t. Masson ，Paris.

HANS 剖， B. 1972a. No 回 on some p' 釘舗itic phanero 脚 ns from 白e In do ・0 泊nese Pe 凶ns 叫a. Bo ιTidsskr.67: 146- 15 1. HANs 副， B. 1972b. Raffl esiaceae. Flora 01 Thailand ， 2: 182-184. App 1. Sc i. Res. Corp. Th凶 lan d， Bangkok. HAYB K， A. von. 1912. Ube r die Blutenbiologie von Cy tin 則的1p ocistisL. Osterr. bot. Ze itschr.62: 238-240.

HB 別問CHBR， E. 1905. Bei 凶 .gez 町 Kenntnis der Raffl esiac 伺 e. 1. Denksch κ K. Akad. 附 ss. 附叫mathe 肌 -naturw. Kl. Kl. 78: 57 -8 1 (3 Tables). HluB， HluB， L. 1984. Fenologia y ecologia floral de Aristolochia grandiflora Sw 釘包(Ari stolochiac 倒的 en Costa Ri ca. Brenesia Brenesia 22: 1-44. H ∞阻R， 1873. Reference in Heinricher ，1卯 '5. HωSBUS ， C. C. 1907. Eine neue Raffl esiaceenga 伽 ng aus Siam. Bot. Jb. 41: 5 5-61. KUG 回 R，H. 195 1. Blutenokologische Untersuchungen mit Go ldfliegenιucilien). Ber. Dtsch. Bot. Ges. 64: 327-34 1. KRAAI ，A. 1962. How long do honey-bees c釘 ry gen 凶nable pollen on them? Euphytica11: 53-56. LATIFF ，A. 1984. Studies in Malesian Vitaceae. VlI.百 le genus Tetrastigma in the Malay Peninsula. Gard. Bull. Sing. Sing. 36: 213-228. L.-; 賀町'， A. 釧 DK.MAT-S 札回H. 199 1. Rafflesia. In R. Ki ew (ed.) ， The state 01 nature conservation in Malaysia ，

89- 94. Malay 組 Nature Society ，Kuala Lumpur. MBuBR， MBuBR， W.1958.Acon 凶butionω 白etaxonomy 組 d biology of Rafflesia amoldi in West Suma 回 . Ann. Bogor. 3: 3: 3 3-44. MB 皿 R，W. 1984. New species of Rafflesia (R afflesiac 伺 .e). Blumea30: 却は15. MB 田 R，W. AND J. F. VBLDKA 岬. 1988. A revision of Rhizanthes (Raffl esiaceae). Blumea 33: 32 9- 342. MOLBSWORTH ALLBN ， B. 1968. In脚 esting jungle plan 飽. MaL Nat. よ 21: 29- 33. 町江，L. VAND 殴 .1933. Ko 出 mededeelingen. Myce 畑出.e (8 rugmansia) Zippe 脳(81.) Hochr. Trop. Natuur.22: 51-52. 51-52. NIYOMDHAM ， C. AND R. KUBAT. 1987. Balanophora 白紙組d Raffl esiaceae of τ 'h ailand. InH. Weber and W. Fo 路出U胞:r (eds.) ，Parasitic Flowering Plants ，Proc. 4th ISPFP ，M 釘 b町 'g. Ru 百fBRFO RD ， R. J. 1970. 百 eanaωmyandcyωlogy of Pilo 岬 les thurberi Gra y 侭affI副総飽 e). Aliso 7: 263 ー288. SMITIN 馴 D， T. 1980. 同 aipl 側， t names. Funny Pu blishing ， Bangkok. SOLMS ・LAUBACH ，H. GRAFI 到 ZU. 1874. Ub er den Bau der Samen in den Familien der Rafflesiaceae und

Hydnoraceae. Hydnoraceae. Bot. Ze iω， ng 32: 337-342 ，353-358 ，369-374 ，385 ・389 (+ Table 8). So 凶 S-LAUBA c:H， H. GRAFBN zu. 1876. Di e Entwicklung der Bluthe bei Brug 制 'nsia Zip peli iB1. und A 巾 'tolochia Clematitis Clematitis L. Bot. Zei，伽~g 34: 34 9- 361 ，365-370 ，481 -4 89 ，497-503 (+ Table 8). So 凶 S-LAUB Ai佃， H. GRAFBN zu. 1898. Die Entwicklung de 唱 Ovulum und des Samens bei Rafflesia und Brugmansi a. An 払 Jard. Bot. Buitenwrg ，Supp 1. 2: 11-2 1.

ST. 釧 LBY ，R G. AND H. F. LINSK 聞 s. 1974. Pollen: Biology ，Biochemist η"， Management. Springer ，Berlin. T雌町'AJAN ，A. L. ，N. R. MBYBR AND V. N. Kc 栂BNKO. 1985. Pollen morphology and cl 鎚 sification in Rafflesiaceae 1. s. 1. Bot. D 1，i. 70: 15 3- 162 (血 Russian ， English suntmary). WATAN 組B， K. 1936. Morphologiscl トbiologische Studien uber die Ga 伽 ng Mitraste 脚， 'n.I ・IV. J.Jap. Bot. 12: 603 -6 18 ，698 ー711 ，759 ー773 ，84 8-8 58. WATANAB 巳 K. 1937. Morphologisch-biologische Studien uber die Gattung Mitrastemon. V ・VlI. J.J ap. Bot. 13: 14- 24 ，75 -8 6，15 4- 162.