NAT. NAT. HIST. BULL. SIAM Soc. 39: 19-52 ， 1991

STENCH AND FRAGRANCE: UNIQUE POLLINATION LURE OF THAILAND'S LARGEST FLOWER ， RAFFLESIA KERRll MEIJER

Hans Banzige r*

ABSTRACT

R 蹴 Ra jJ1 esia kerrii is de 削除d for 肱伽t time from bl<∞ ming male and female female flowers. Females can be distinguished from males by 血.e presence of 釦 obliquely set band band along the lower rim of 出 e disk. R. kerrii's dis 住ibution is extended farther south and north 由an hitherto reported ， to Perak (W. Malaysia) and Pr achuab Kh irikhan (S. 百凶land) where where 12 colonies with 60 buds were found 泊 an area ca. 2 km 2 at 500 - 850 m a.s. 1.百 le habitat habitat was unusually rich in Tetrastigma (Vi 阻ceae) lian ぉ血ough only T. quadrangulum Crai b et Gagnep. is confinned as host among 10 spp. present. A s 回 ng smell of carrion is emitted 貨 om rafflesia's perigone lobes ， while a weaker 仕uity fragranc e-ー出血 lown 泊 other rafflesi as- is produced in 出e 佃 be's cavity ，血 es 印目白 be 泊g kept separa 飽 by 血e diaphragm. C 創出 n flies Chrysomya villeneuvei Patton ， C. rufifacies (Ma 叫 uart) ，Lu cilia porphyrina (W alker) 釦 d Hypopygiopsis tumrasvini K 町油鎚hi (Ca lliph 凶 dae) 卸 a町 acted by the stench and and the ‘fes 飽:ring sore' apμarance (lurid colour ，cra 旬r shape ， white blotches an d/ or whitish mold on processes) of 血e flower. 百le 釦nctions of 白 e fragrance (e.g. to advertise carbohy- dra 旬 s，essential to adult flies) is disc 凶鈎dωge 血.er with the physiology of 白 e flies. Surplus pollen pollen mush which has dropped 台。 m 血.e anthers and a slimy secretion coating the p釘 ts

below below the disk c組 be sucked by the flies. Th ey climb channels leading to the anthers ， which are are positioned in such a way 白紙 pollen cannot be stolen but only smeared onωthe back of the the flies' 曲orax. 1n血 e female flower 出e process is similar ， the pollen clot being rubbed off onto 也 e stigmatic surface set between the oblique band and the column. is It proposed that R. R. kerrii is not a deceptive flower and 白紙白 .e flies might act 鎚 long distance pollinators between between rafflesias flowering up to several weeks apart. 明le re l

hosts hosts indicates 出at rafflesi 舗 may be able to survive parent host death and spre 唱d their pop 叫ation by vegetative infection of the host's lateral rumJ ers.

INTRODUCTION

Rafflesias Rafflesias have baffled scientists since early in the last cen 加 ry when reports 合om Suma 回 about R伊 esia arnoldii R. Br. ，由 e 伽 t of some 14 spp. of the genus (MEuE R ， 1984; 1984; MAT SALLEH & LAT 町， 1989) ，caused a sensation in the botanical world. 百le flowers' flowers' phenomenal size ，ephemeral blooming ， unlikely parasitic existence ，vivid colours ， and lack of chlorophyll ，leaves ，stem and roots make them a living p 紅 'adox. Because of their their rarity and concealed existence as filaments within the host liana (Tetrastigma spp.) for for much of the time ， their biology remains largely unknown. Seed dispersal 釦 d infection of the roots of a new host 釘 'e an enigma ，but possibly 由e most disconcerting aspect of these ‘flowering beauties' is their bouque t: a cadaveric stench of ro 剖ng snakes. Th e con- comitant buzzing of blowflies ，well-known pests associated with excrement ， festering sores

市De par 加lent of Entomololgy ，Faculty of Agriculture ， Chiang M 副University ， Chiang Mai ，50 ∞2，司副 land

19 19 20 HANS BANZIGER

組 d carrion ， around the flowers dispels 佃 Y1 泊gering doubt about the proven 佃 ce of the stench. stench. Yet ， in a new surprise revealed in this study ，合 om 仕1e very heart of some rafflesias a sweet 企'agr 創 1ce em 鉱 1ates. Th e genus occurs mai 凶y wi 出in 白eW. 組 d S. Malesian floristic region ， with Th ailand the only coun 町 ou 凶 de Malesia S 'O far with a rafflesia (sensu stricto) ， R. kerrii Me りer. R. kerrii has been reported mainly from 5l 'O calities in Ranong and Surat 百四1i Pr ovinces (MEDER ， 1984; N 町 OMDHAM & KUBAT ， 1987: MEUER & E 比， 10π ，1991). 百1e known southernmost 胎凶t extends somewhat south of the Kangar .， Pattani line ， the bound- 的 of the Malesian flora on the Malay Peninsula (WlllTM O阻， 1975) ， with a collection from a site on Bukit Tepoh which marks the border area of Narathiwat Pr ovince of 百凶land wi 由 Kelantan State ， Malaysia (MEDER & ELLI O' π， 1991). Th e present study extends the rafflesia's dis 凶bution sigt 泊白 cantly f: 紅白erno 吋1， to Pr achuab Kh irikhan ， not contiguous with other known populations. Buds acquired by the au 血or in Upper Perak also extend its presen 回 f紅白.er south 凶'0 Malaysia but ，部 men- tioned tioned in Appendix 2，出 e buds' ori 匝n is not certain. In the absence of precise da 偏食'O m Malaysia ， R. kerrii can still be considered a 百凶 endemic ， but it is virtually certain that 出.e species ~ distribution reaches 凶 o northem W. Malaysia and also S. Burma. Kerr's col- lection lection ne 紅 Ban Lam Li eng (R anong Pro v.) is 0 叫Y 10 km from the Burmese border (Te- nasserim nasserim Division) ，釦d new findings reported here put it even ne 釘 'er to Burma. If If rafflesi 槌 are considered in the broad sense ， then Th ailand is home to at least two more species: Sapria himalayana Griff. 泊 northwest and west Th ailand (HOSSEUS ， 1907; 1907; HANs 聞， 1972; SMITINAND ， 1980; BANZIGER ， 1988; ELLIOπ ， in prep.); and S. poilanei poilanei Gagnep. in 血e southeast (S 班 TINAND ， 1980). With an exp 佃 sion reaching 0.7 m ， R. kerrii is not q凶te 踊 large as R. arnoldii's flower flower of close to 1 m ， but it is still the 1釘 'gest flower of Th ailand. Yet only 槌 recently 酪 1984 has it become known to science with a valid description by MEDER. Botanists had acωally sporadically collected it since 1927 but it was ‘forgo 伐.en' until S班問AND (1 980) mentioned mentioned it in a list of 百凶plants under the name R. patma B 1. (initially identified by Kerr) ，a species known only from Sumatra and Java. Furthermore ，no one s関 ms to have acknowledged 也at from time immemorial l'O cal forest dwellers ， ethnic Th ais and other peoples peoples of 血e Malay Peninsula ， have known and used it. As dok bω tuum 組 d bua phut (Thai: 悶 n仇ド in Ranong Prov. and 'U 1 朋 in Surat Thani Prov. ，respectively) 組 d bunga pa ωω (pakma) (Malay ， applied also to other rafflesias) ，it has been ascribed medicinal medicinal properties. Possibly because of the r副 ty of bl 'O oming flowers it is believed to possess possess magical and mystic-religious powers ， to confer invulnerability and assist in a仕組 n・ ing ing Nirvana (pers. comm. by a gum tapper of Ranong). h 白e most detailed s加dy of R. kerrii so 伽， MEUER &E 比10' π (1 991) summa- rized rized previous information and added new taxonomic and ecological data obtained 仕om 合eshly cut buds. Fruits and seeds remain w 北nown. R. R. kerrii was unexpectedly found by the author while investigating the interrela- tionship tionship between rare moths and lianas (e.g. BλNZIGER ， 1989). Fl y activity on a bloom- ing ing flower prompted an investigation 'O f the p'O ll 加ation mechanism ， as a comparison for 佃 ongoing similar project on dipteran pollination of orchids. Additional details not di- rectly rectly connected with p'O llination 田 ology are also included in 由.e present research ー由e first first of bl 'O oming flowers and living bud colonies 'O f R. kerrii in the wild. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEIJER 21

At the time of my research 1 was not aware of the excellent study of the polli- nation nation mechanism of R. pricei Meijer by BEAMAN et a1. (1 988) in which for the first time the the details of rafflesia pollination are described. While the pollen transfer mechanism is the same in the two rafflesias ， there appear to be appreciable differences in the way the two rafflesias ‘manipulate' fly behaviour for their own ends. This will be discussed in a later pape r.

HABITAT

In In contrast to MEIJER & ELLIOπ (1 99 1) 1 ca 汀 ied out my research alone without telling telling locals what 1 was studying. Furthermore ， in order to protect this rare rafflesia from unscrupulous unscrupulous collectors and visitors ，1 have chosen not to reveal the exact location of the plants plants at this stage. The site is some 300 km north of what heretofore was the most northerly northerly population known. A 3 ・hour ascent from the nearest road leads one to the main colonies colonies at 700 ー750 m. The climb is through notoriously forest tricky (confirmed by local people people )--d espite marks left on my way ，1 am not sure if 1 ever managed to follow the same track track twice. The rafflesias grew on the southeast-facing river catchment， on slopes of up to about about 30. inclination ，between 475 and 840 m. At elevations lower than ca. 300 m and higher higher than 900 m the vegetation was rather xeric and darker ，respectively ， less suitable for for host Tetrastigma spp. 1 did not find any R. kerrii on the northwest-facing slopes above the the same stream ，where lianas were scarce in genera 1. Unlike some other rafflesia species (e.g. (e.g. MAT SALLEH & LATIFF ，1989) ， R. kerrii was not close to streams ， the nearest being some 50 m away. Big granitic boulders and smaller rocks were frequent but mostly not adjacent adjacent to the rafflesias. The forest (Fig. 3) was evergreen and the vegetation remained green green even during the exceptionally arid dry season of 1991 when no rain fell from December until Apri 1. Fire does not seem to reach the sites normally though it rages through through more peripheral ， drier sections of the mountain. Leaf litter covered the soil completely. completely. Sm aI l trees predominated ，interspersed with unusually abundant lianas but few large large trees. The c佃 opy allowed some rays to reach the ground and the blooming rafflesias of of colonies 3 and 4 possibly received direct sunlight some 5-10% of the day. Th e ground flora flora was quite sparse ， mainly fems and seedlings; no Gramineae were seen at the study sites. sites. Many rattans and other spiny plants made passage arduous. Plants Plants in the vicinity of colony No. 4 included ， besides unusually common Tetras- tigma tigma spp. ， several Menispermaceae (Stephania reticulata Forman ，Cycle αα tjehensis Forman ，Diploclisia glaucescens (B 1.) Diels) ， Cissus sp. ，Cyc αs circinalis L.， Justicia val- ida ida Rid 1. and a number of unidentified herbs. Relevant Relevant vertebrates observed were squirrels such as the common Callosciurus flavimanus flavimanus G. S. Hillaire ， C. caniceps (Gray) and the mouse deer (Tragulus javanicus (Osbeck)). (Osbeck)). They may play a role as seed dispersers of rafflesias. Gibbons (Hylobates lar L.)， L.)， leaf monkeys (Presbytis sp.) and macaques (Macaca nemestrina (L.) or M. arctoides (Geoffroy)) (Geoffroy)) were seen or heard on most days; head ，skin ，intestines of macaques and leaf monkeys were occasionally found at hunters' camps. They were covered with egg-laying and feeding blowflies ，rafflesias' pollinators. These mammals may also contribute to seed dispersal dispersal of Tetrastigma spp. 22 H ANS B ANZ IGER

ss c 1

Figures 1- 2. Cross secti on of blooming fl ower of R. kerrii, showing main morphological charac­ teristics. I, Female; 2, Male (part).a = anther; ae = annu­ lu s exterior; ai = annulus in­ te rior; c = column; cd = col­ lar of disk; cl =disk; dp = dia­ phragm; ob = oblique band; p = process; pi = peri gone a lobe; r = ramentae; s = sul cus; ss = sti gmatic surface. 2 STENC I-I AND FRAGRANCE: RAFFL ESIA KERRII ME lJ ER 23

Fi gur e 3. I-I abitat of R. ke rrii. A f ema l巴 tlow er can be see n ne ar th 巴bottom ce ntre .The two arrow s point to th e5 t巴m ofth 巴 h0 5t，th e li ana Telra sli gm αqlladr all glllu m 24 HANS BANZIGER

Figure 4. Ma.le R. kerrii, 0.5 m across, in full bl oom. Two blowflies have settled on the right perigone lobe, and two are in fli ght above the diaphragm on the left.

Figure 5. Female R. kerrii just past full bloom, the distal parts of the perigone lobes already wilting. No blowflies present. STENCH AND FRAGRANCE: RAFFLESJA KERRIJ MEIJER 25

STUDY POPULATION

Th e sωdy 紅開W 剖 visited 11 times during 4 periods: 26 ー27 J釦 U 紅 y， 12 Febru- ary ，5 -8 March ，7-10 April 199 1. Host 田 sessment w 槌 not always successful as 血e digging digging had to be carried out with gr 芭at c紅 'e or avoided altogether in order to m 泊加lize damage; damage; the roots sometimes extended deep down or between rocks.

Colony cluster A:

At 475 m ，3-4 km (1. 5h walk) 合om base. Colony No. 7 wi 出 1 dead bud of 12 cm across and 21 釘 ger buds or flowers 泊 very advanced stage of decay ，30 cm 合om each other. other. Host almost certain T. quadrangulum of 6x 8 cm s旬m size. Colony No. 8， 15 m further further on ， with 2 healthy buds of 8 and 14 cm ， in close contact. Host T. quadrangulum ascertained ascertained by following the root to the liana stem of 4x 5 cm (Banziger coll. No. 901).

Colony cluster B:

At 500 m ， ca. 250 m further on 企om A. Colony No. 1 with 13 dead buds up to 10 10 cm ，a moribund and a healthy one of 14 cm ，over 釦 area of 5x 6 m. Host almost certain certain 1-3 T. quadrangulum of 4x 5 cm stem size. A very small Tetrastigma sp. 1 was present present further away. R ∞ ts with most buds decaying ， hence high bud mortality. Colonies No. 10 ，11 ，12 ， at 30 ，50 佃 d 80 m from No. 1，respectively; with 5，5 佃 d4 buds ， respectively ， dead or rotten; about medium size and some 0.5 m 企'om each other. Host almost almost certain T. quadrangulum ，血e larg 回 t5x6cmins 飽m size.

Colony cluster C:

Possibly Possibly some 2 km (ca. 1. 5 h walk) further up from B. Colony No. 3 at 720 m ， with with a rotten and a healthy bud of 17 cm which developed inωflower No. 2 (Figs. 3 ，5). Hostw 儲 T. quadrangulum ，ascertained by digging and following the root to the liana stem of of 2x 3 cm (Banziger coll. No. 896). T. papillosum and T. sp. 1 were also present ne 釘 by. Colony Colony No. 4， at 750 m ， ca. 200 m fl 町出er up from No. 3， with 2 small buds of 7 cm ，2 1紅 ger ones of 19 cm (Fig. 11) ，one of which became flower No. 3 (Fig. 19) ，and the bl ∞ming flower No. 1 (Fig. 4) ， and 3 long since rotten flowers. Are a just 1x 1. 5m ，some buds buds in close contac t. Host was T. quadrangulum ascertained by digging and following the r ∞It to the liana stem of 3x 5 cm which bo 白金uits (B 沿lZ iger coll. No. 864 ， 876). T. papillosum papillosum and T. sp. 1 were also growing ne 釘 by. Colony No. 9， at 765 m ， ca. 70 m f町出er up from No. 4， with 2 buds of 4.5 and 9 cm and 4 dead ones ， about 0.5 m from each each other. Host probably T. quadrangulum but T. sp. 1 and T. sp. 11 were also nearby.

Loc ation of colonies No. 2，5， 6 not certain as 1 lost my track. No. 2， at 710 m ， wi 曲 2 aborted and 2h 伺 l血y buds of 10 (Fig. 10) 佃 d 12 cm ， over an 釘 ea of 2x 3 m. Host possibly possibly T. sp. 1 but T. quadrangulum w 槌 present fur 白er away. No. 5， at 780 m ， with 2 open ， long since withered flowers. Host possibly T. sp. 1 but T. quadrangulum w 出 not far off. off. No. 6， at 840 m ， with 4 a:加 Irted buds. Host w 削 own. 26 HANsBAN Zl G殴

Th e roots infested were about 1.5-2 cm in width (Fig. 20). Th e par 加 itized lian 邸 were were ne 釘 Iy all of modest size ，usually 3x5 cm across ，b凶 one w 酪 6x8 cm. Th e basal 2-5 cm of buds and flowers were embedded in the so i1 and were about one to several m from where the stem of the host emerged f凶m 血.e ground. Because of their brownish colour ， buds did not s旬nd out much against the background of leaf litter ， the young ones being being completely submerged in it. Even the reddish ， blooming flowers which lack the large whi 飽 spots of other rafflesia species ， were not t∞ conspicuous against dead leaves. In one instance instance I was alerted by the s加 lch of the flower before seeing it. Most conspicuous were dead dead buds and flowers due to their charcoal-like aspec t. Wi 血 a maximum of some 15 'buds (possibly growing on mo 問血an one host) ， colonies colonies of R. kerrii are smaller 白血those of S. himalayana in which more 也an 100buds sometimes sometimes emerge 合'O m a single host during the co 町田 of a flowering se 踊 on (E LLIO 'IT， pers. pers. comm. 加 d pers. obs.). However ， the latter is a much smaller species and its main host host rather 1釘 'ge.

OBSERVATIONS Hωts Very li tt1 e attention has been given to host plants in studies of rafflesia although 出ey 創-e of paramount importance to their conservation. 百le hosts of rafflesia res are 凶c凶 to Tetrastig 刷 spp. (Vi 飽.ceae). BROWN'S (1 821) record record of R. arnoldii on Ciss 凶 sp. appears to be a misiden' 凶 cation (LA1 買 F ，1984). In白紙 taxonomic taxonomic sωdy 12 spp. occurring in (戸'O bably the Malaysian part ot) the Malay pe 凶n- sula sula are revised; the total number given for Malesia is 57 spp. GAGNEPA 町(1 912 ， 1930) lists lists 39 spp. for Indochina and CRAIB (1926) 16 for 官lailand.

At least 7 and probably as many 邸 10 species of Tetrastigma were found in my S加dy area (coll. deposited at 恥 Dep t. Entomology ， Faculty of Agriculture ， Chiang Mai University University (DEFACU)):

T. T. cruciatum Craib et Gagnep ・ co l1. No. 877 ，892 ，931 ，936 T. T. lanceolarium Planch. co l1. No. 898 ，929 ，930 ，942 ，943 T. T. papillosum (B 1.) Planch. coll. No. 890 ，895 ，944 T. T. quadrangulum Ga. 伊 ep. et Craib coll. No. 864 ，894 ，896 ，900-902 ，913 Tetrastigma Tetrastigma sp. 1 coll. No. 876 ，888 ，889 ，891 ，893 ，897 ，899 ， 938 ，939 Tetrastigma Tetrastigma sp. 6 coll. No. 937 Tetrastigma Tetrastigma sp. 7 (aff. siamense Gagnep. et Craib) coll. No. 927 ，928 Tetrastigma Tetrastigma sp. 8 co l1. No. 940 Tetrastigma Tetrastigma sp. 10 coll. No. 933 Tetrastigma Tetrastigma sp. 11 (aff. garrettii Gagnep.) coll. No. 934 ， 935)

百lI s is about 5/6 of the species found in the whole of West Malaysia and also of Doi Suthe p-Pu i National Park ，we l1 -known for its high biodiversity (B ANZIGER ， 1988; E LLI O' πet al. ，1989) ， where about 12 species have been recorded (MAxwE LL ，pers. comm ふ Th is mountain has been s町 veyed for c1 0se to a cen 加ry and is ，wi 白 261 km 2，much wider STENCH AND FRAGRANCE: RAFFLESIA KERRII 勘偲UER 27

出組曲e area of R. kerrii of a few km 2 which 1 have visited 11 times. Two and even 白隠e species species of Tetrastigma were in the close vicinity of some R. kerrii colonies. However ， in my study area only T. quadrangulum (Fig. 3，12) ，a new host ， has been confmned to be par 田 itized by R. kerrii 由ough Tetrastigma sp. 1 is a candidate. Pr eviously reported ho 蜘 of of R. kerrii ar ち T. lanceolarium ~血日ER ， in 勘fE UER & ELLIO TI'， 1991) and T. papillosum (N 町 OMDHAM & KUBAT ，1987) ，bo 由 present in my study area. T. T. quadrangulum was the most common species in the belt where the rafflesia occurred ，closely followed by T. sp. 1. Elsewhere T. sp. 1 ，T. papillosum and T. lanceola- rium rium prevailed. Th us ，contrary to what 1 had written (B ANZIGER ， 1988) following BA 別& HUMPHREY (1 980) ，Tetrastigma are not rare ， at least 血e species known as ho 蜘 of R. kerrii. kerrii. Th ey are"-often m 吋町 components of 由e forest ，clearly dominating 泊 some a四回. 官le s包 ms of many Tetrastigma spp. conta 泊 signi 白cant amounts of water. 百lis rapidly rapidly dribbles out if one cuts a section out of the stem and holds it vertically ， an acti- vity vity unfortunately practiced by 白irsty forest people because .tT e water is po ぬble ， clear ...... " and tasteless. 百le local n釘 ne for Tetrastigma sp. 8 is ~"甘 ~tÏ1l1 (由 aowan nam =‘ water liana'). liana'). All the mentioned s戸cies show 血is feature (血 ough T. sp. 1 seems to have little water; water; spp. 6，10 ， 11 have not yet been tested). 明 le availability of large amounts of readily flowing flowing wa 飽r may be related to its parasitism by rafflesias. Its tissue is very spongy and could could be ‘pumped up' easily without undue loss of nu 凶ents from the host as long as not too too many flowers develop on it. Th e 合uits of the many species of Tetrastigma 釘 e quite diverse ， as must be their seed seed dispersal ecology. T. campylocarpum (Kurz) Planch. has no pulp ， but T. voirnieria- num Gagnep. and T. quadrangulum 釘 e said to be pulpy and commestible (GAGNEPA 町， 1912 ， 1930). Fruit size ranges from 0.7 cm (T. pedunculare (Wal l. ex Laws.) Planch.) to 3.5 3.5 cm (T. hookeri (La ws.) Planch. (LAl 官 F，1984)). 官le colour is white ，yellow ， red or black. black. 1 found ripe T. quadrangulum 企uitsωs U1p部 s3 cm in diameter. Th ey were shiny bright bright yellow ， strongly pe 拍Jmed ，and 血e pulp had plenty of sweet juice. Su 甲risingly ， such such a ‘delicacy' is not eaten by people in the study 釘 'ea because ， as they exp 凶 n ，it is not fed fed upon by animals and hence is not safe (al 白ough in N. 百lailand it is said to be edible). In In an experlment on myself 1 took a daily increased dose 企'oma 由ly drop up to the whole pulp pulp of a fruit (after which my supply ran out ー 1 took 血e precaution to undergo the 凶 al 泊 Chiang Mai where there 紅 'e hospitals). Except for a slight throat irritation no detectable ill-effects ill-effects ensued. 1 deduce 針。 m this 白紙 there must be a frugivore which eats 由is and 白紙 in in the study area it is likely to be either nocturnal ，relatively scarce ， or otherwise difficult to to de ぬct. Fruits with as 町ong scent are generally attractive to marnmals rather 曲an birds; but but being attached to relatively young and slender stem sections hinders 血e approach of primates primates and birds large capable of swallowing the whole fruit or at least large chunks con- taining taining the 1-3 big seeds. Th e most likely dispersers are 白us bats which may fly directly onto 血 e fruit and take off with the whole of it to eat somewhere else. It is also likely 白紙 fruits fruits fallen to the ground 紅 e eaten by mouse deer. In In a number of 部 pects Tetrastigma spp. 紅 e similar to the liana Parvatia brunoni- ana ana Decaisne (Lar dizabalaceae) ， such as the corky bark with prominent ridges or other structures. structures. Some of the biological features described for 出 .e latter (BANZIGER ， 1989) are likely likely to apply also to Tetrastigma ， such as vegetative reproduction by lateral runners. In time time these send out rootlets and later become sep 紅 ate plants if severed from the parent 28 HANs BANZIGER

liana ，'O r when the latter dies. Since in s'O me species the runners attain a length 'O f many meters ， the liana can actually shift fr 'O m the 'O riginal place ， and spread. 百lI s is 'O f great significance significance t 'O rafflesias because ， if their filaments can infiltrate new runners ，出 ey can m 'O ve and expand 出ec 'O l'O ny (1 share this idea with Dr. S. Elli 'O tt). H 'O wever ， the 'O pp 'O si 飽 is is als 'O c'O nceivable ，namely ，出 at the h 'O st can rid itself 'O f the parasite if the latter cann 'O t infil 住ate new runners (企 om 由er 'O'O ts the filaments must be able t 'O penetrate 出e stem and fr 'O m there 由ey 白 en have t 'O infiltrate the runners). Large c'O l'O nies such as N 'O. 1 can be expected expected t 'O we ak: en the liana and be the likely cause 'O f death 'O f th 'O se r'O'O ts with all 出e rafflesia rafflesia buds. Th em 'O dest size 'O f rafflesia-parasitized T. quadrangulum c'O mpared with 出e maximum size 出ey can attain ，may be indicative 'O fa gr 'O wth sl 'O w-d 'O wn. H 'O wever ，it c'O uld als 'O mean that y'O ung Tetrastigma 釘 'e m 'O re fぬquently infected 出an hithert 'O sus- pected ，either vegetatively thr 'O ugh runner infiltrati 'O n an d/'O r thr 'O ugh new infecti 'O n by the r 'O'O tlets 'O f rafflesia seedlings (cf. discussi 'O n).

Morphology of the R. kerrii Flower (Figs. 1-9 ， 21-23)

Th e 'O riginal descripti 'O n (MEUER ，1984) ， based 'O np 'O'O rly preserved material ， has been been c 'O nsiderably improved by ME 田 R &ELLIOπ (1 991) wh 'O examined freshly cut buds. The present analysis adds further details based ，f'O r the first time ，'O n fully devel 'O ped and 'O pen ， fresh male and female fl 'O wers. Overall Overall c'O l'O ur dull red with br 'O wnish tinge. Warts and n'O dules 'O f perig 'O ne l'O be in in male N 'O. 1 pinkish ， in female N 'O. 2 as backgr 'O und 'O r even slightly darker (sexual dim 'O rphism ，individual variati 'O n，'O r just disc 'O l'O rati 'O n?). Pustules 'O n diaphragm pinkish white. white. Wall 'O f perig 'O ne tube appe 紅泊g dark ， due t 'O the black ramentae and n 'O dules. An nulus exteri 'O r， disk and pr 'O cesses shiny and lighter red th 'O ugh the extremities 'O f 出e latter latter are darker. Bristles 'O n pr 'O cesses yell 'O wish br 'O wn. Diameter Diameter 'O f wh 'O le fl 'O wer 50 cm (female 52 cm) but may attain 70 cm. Perig 'O ne l'O bes 13-18 cm l'O ng and 19 ー22 cm wide ，with 'O ut 山 white edge menti 'O ned by M 町田 & ELLIO 'IT f'O r buds. W ，訂 ts (Fig. 7) 'O n perig 'O ne l'O be c'O nsisting 'O f gr 'O ups 'O f2 -4 m 'O re 'O r less less c 'O alescing n'O dules ， with smaller 'O nes rand 'O mly dispersed in 白 e spaces between 由e warts. warts. Th ese 紅 'e smaller 出m 仕le spaces between the warts and tend t 'O decrease in size t'O wards the margins. Diaphragm (Fig. 4) curved inwardly with pentag 'O nal 'O uter rim (Fig. 5) 5) 'O f 65 cm circumference (69 cm in female) ，4- 5 cm wide (4.5-7 cm in female) ， with great great numbers 'O f 由ly ，irre 郡11 釘 ly and densely packed pustules about 1 mm acr 'O ss ， with centrally centrally ruptu 陀 d membrane (Fig. 8). At places the pusωles draw cl 'O ser t'O ge 出er f'O rming pinkish-white a pinkish-white aggl 'O merati 'O n (Fig. 5) surr 'O unded by a reddish hal 'O which can give 自己 impressi 'O n 'O fa wart (MEUER menti 'O ns 5c 'O ncen 凶c rings 'O fw 制 s 'O n 出e diap 胎 agm). Central Central 'O pening 'O f diaphragm slightly elliptic ， 12 by 17 cm (m 'O re roundish in female ，15- 16 16 cm) ， the underside with bright white ，r'O undish t 'O elliptic bl 'O ts (Figs. 4， 13) up t 'O m 'O re than than 10 mm diameter (7 mm in MEUER and MEUER & ELLIO 'IT) th 'O ugh m 'O stly quite smaller. smaller. N 'O clear arrangement 'O f the bl 'O ts in c'O ncen 住ic rings seen ，n 'O r were any raised 'O n stalks 血'O ugh there were a number 'O f raised ，crater-like s住uctures in the area where the bl 'O ts and the ramentae j'O in. 百lese are maximally 3 mm l'O ng (1 0 mm in MEUER) with enl 訂 ged bifid 'O r 出自 d tips. Dis 住ibuti 'O n 'O f the ramentae (Fig. 1) variable: wi 白血 aband 'O f2 -4 cm wid 出，出 e upper m 紅 gin 'O f which starts just bel 'O w the white bl 'O tches while the STENCH AND FRAGRANCE: RAFFLESIA KERR/I ME lJ ER 29

6 7

8 9

Fig Ul巴 s 6- 9. Morphological details of female R. kerrii shortly past full bloom. 6， disk with processes; 7，w 紅白 of perigone perigone lob e; 8 ，pustules of diaphragm; 9， process of disk with rows of bristl 巴s on cres t. Note rapid for mation mation of mold which was near ly completely l acking 24 h 巴arli e r(Fig. 6). Length ofbar 25 mm in Fig. 6， 5 mm in Figs. 7 - 9 30 H ANS BiiNZIGER

11 12

Figures 10- 11. Buds of R. kerrii. 10, medium sized; !I , 19 cm across, some 2 weeks before opening.

Figure 12. Te rrastigma quadrangulum, host of R. kerrii. Note trifo li ate leaves, angul ated young stems, and frui ts soon ri pening. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEUER 31

l'O wer m 紅 'gin ends abruptly (Fig. 13) ，ab 'O ut where tiny n'O dules are f'O und d'O wn t 'O the annulus annulus exteri 'O r. Al ternately ，血 er 鎚 nentae may be 'O na br 'O ader band where they steadily sh 'O rten further d'O wn and gradually merge int 'O the tiny n'O dules menti 'O ned. Th e annulus exteri 'O r is ab 'O ut 2 cm wide; between this and the annulus interi 'O r runs a sulcus 2-3 mm wide. 百 e cen 回 1 disk (Fig. 6) is 12 cm across (up t'O 18 cm in ME 田 R & ELLIOIT) with a gr 'O'O ve al 'O ng the rim which is c'O nfined by the disk c 'O llar 'O f2 cm height al 'O ng the crest 'O f which is ar 'O w 'O f tiny bristles. On the disk are 29 pr 'O cesses (Figs. 6， 9) (32 in female

N 'O. 2， arranged in an 'O uter wh 'O rl 'O f 17 ，a sec 'O nd 'O ne 'O f 12 and 佃 inner gr 'O up 'O f 3; 40 pr 'O cesses in a male bud from Grik) (MEUER & ELLIOIT have 27 -44 processes). 官leyare up t'O 3 cm l'O ng ，elliptic at base ， tend t 'O be radially c'O mpressed ，increasingly f1 attened t'O wards the apex where 出ey are m 'O stly blade-like ， 0.5 -2 mm thick and up t 'O 2 cm wide ， spike-like; 'O r spike-like; s'O me anast 'O mized and bec 'O me rather wider. On the crests and tips 創ち r'O ws 'O f bristles nearly 1 mm l'O ng (Fig. 9) (n 'O t menti 'O ned by the ab 'O ve auth 'O rs). Th e underside 'O f the disk (Figs. 1，2，21-23) -d escribed in m 'O re detail because 'O f its its imp 'O rtance f'O rp 'O ll 泊ati 'O n-sh 'O ws a number 'O f differences in the tw 'O sexes. Unlike the female ， in the male the l'O wer rim is m 'O re 'O r less r'O unded ， at m 'O st with 'O ne edge 'O ns 'O me sect 'O rs 泊 s'O me specimens. The an 血ers ，5-6 mm in diameter and saddle-shaped ，number 28 (36 in a bud fr 'O m Grik; 26 ー31 in MEUER & ELLIOπ). Each is set in a cavity 'O fι10 mm width. 官lese are carved 'O ut 'O f the underside 'O f the disk and set in a circular row ar 'O und the c'O lumn. Th ey 'O pen t'O wards the interi 'O r and the base 'O f the f1'O wer ， the anther being being in a vertical p'O siti 'O n with the 'O nly p'O問 directed h'O riz 'O ntally t'O wards the c'O lumn. Between the cavities there is a ridge which extends in a radial spire fr 'O m the anther first up ， then in ，and finally d'O wn bef 'O re levelling 'O ff near the annulus interi 'O r as it curves 'O U t. The ridges are t'O pped with rows 'O f 1-2 mm l'O ng ，s'O mewhat stiff hairs (n 'O t menti 'O ned in MEUER & ELLIOπ) which make the ridges impassable 組 d als 'O shield the 組出er at 血e distal distal 'O pening 'O f the cavity. Sp 訂 sely distributed hairs 紅 'e als 'O 'O n the underside 'O f 出e disk and 'O n the annulus interi 'O r where the ridges level 'O ff d'O wn t 'O the sulcus but n'O t between the the ridges. MEUER & ELLIOIT menti 'O ned additi 'O nal ，‘ weak ， l'O ngitudinal ridges' in 血e anther anther cavities; they are lacking 'O r extremely faint in my specimens. In 'O ther w 'O rds ，血 e cavity cavity with the 加 ther is at the end 'O fa gr 'O'O ved ，6- 10 mm wide ， channel f'O rmed by radial ， hair-t 'O pped ridges starting ne 紅 the annulus interi 'O r and running up in an increasingly steeper steeper curve (Figs. 21-23). In In the female the l'O wer rim 'O f the disk has tw 'O well marked edges between which runs runs an 'O bliquely set ，2-5 mm wide ，f1 at 'O r slightly c'O ncave band (Fig. 1). 百lis fea 制民 is is missing in the male and is very useful f'O r sex distincti 'O n (b yt 'O uch) in 出 e field; it has n'O t been menti 'O ned by the ab 'O ve auth 'O rs. Th e an 由er cavities 釘 e replaced by a gr 'O'O ve 2- 4 mm wide running r'O und al 'O ng the c'O lumn. Th e ridges and hairs are c'O mpletely reduced and 'O f the channels 'O nly faint ，narr 'O w ，radial depressi 'O ns remain ，if any. Th e underside 'O f the the disk ， between the gro 'O ve and the narr 'O w ， 'O blique band ，is c'O vered by a velvety mat 'O f tiny white hairs 'O r papillae; this annular band represents what 1 believe is the l'O ng- ught-after s'O ught-after stigmatic surface (cf. discussi 'O n). It is set at an angle with the annulus interi 'O r and the c'O lumn ，which is with 'O ut hairs ，S 'O that the space between the first and the latter tw 'O n釦'O ws t'O ward the interi 'O r (Figs. 1，28 ， 29). 32 HANs BA NZl GER

Some of the differences mentioned are p釘 tly due to the variability of rafflesias and to the possibility 血at this most northerly population is subspecific a1 1y distinc t.百 le bris t1 es on 血e processes could be of taxonomic significance. According to KUIπ(1969) 出ey 紅 e absent 泊 R. arnoldii but present in other species.τ 'h ey can su 任er damage in preserved preserved materi a1， which explains why 由ey may have been overlooked in previous de- scriptions scriptions of R. kerrii ， and possibly in other sp 田 ies.

Phenological Phenological Notes

百le m a1 e flower No. 1 (Fig. 4) was discovered in full bloom on 26 J佃 U 紅 yand presumably presumably had been open for 2-4 days. It had slightly damaged perigone lobes ， probably nibbled nibbled by some insect. Most of the processes were blackish with extensive ，irregul 訂 patches patches of yellowish-white mold (Fig. 13); some still had reddish sections. 百lI s is not necessarily necessarily a sign of deterioration but may contribute to 血.e flower's mimicry of a fester- 泊g wound. Th e following day 由e perigone lobes were sta 此ing to curl at the edges and the overall overall colour seemed a bit darker. Paired marks were detected on the lobes ， probably caused caused by the teeth of a squirrel or ra t. Mter 16 days the lobes had completely shriveled ， become 也rk g陀 Y to blac k， and the inner p釘 ts of the tube were rotting. Th e fem a1 e flower No. 2 was found on 12 February as a bud 15 cm in diameter. On 6 March it measured 17 cm across and the pinkish ou 飽r side of the lobe was visible between between the gaping sc a1 es. By 7 April it was slightly past full bloom (Fig. 5). Perigone lobes lobes and diaphragm were more brownish 血an in m a1 e No. 1， the dis ta1 3-10 cm of the lobes lobes darker than the proximal ones and with some greyish white mold ， and s旬此ing to cur l. However ， the disk ，processes (Fig. 6) and annulus exterior were much brighter red 血m 加 No. 1，a1 though the latter was not as old. Sm a1 1 patches of whitish mold were present present on some of the processes. Th e following day most of the processes and disk were greyish greyish black with greyish white mold cover (Fig. 9) ， showing how fast darkening 佃 d mold grow 血 can be. 官le diaphragm' sm 釘 gin around the opening was greyish ， the lobes somewhat further dark. On the fourth day everything inside the tube w 制 blackish ， other p紅 ts a1 so dark ， yet the lobes st i11 had sections greyish brown-red. Th em a1 e flower No. 3 was detected on 26 January as a bud 15 cm in diameter. on on 12 February it measured 19 cm across and the pink between the sc a1 es was visible over about about 1/3 of 血e surface (Fig. 11). On 6 March it w 儲 a fully open ， but w i1 ting flower (Fig. 19) ， though still of the origin a1 shape ，completely grey-black ， with plenty of mold ，possibly aw 田:k or more old. Th e following day more w i1白 19 ，curling and mold. 1 have not yet been able to study the earliest Stages of a R. kerrii flower wh i1 e it IS IS openmg up. STENCH AND FRAGRANCE: RAFFLESIA KERRll MEI 厄 R 33

Scent Scent Emanation Rafflesias Rafflesias are notorious for their carrion-like smell and R. kerrii is no exception. Some authors ，however ， have contested the presence of a pu 凶d smell. HOSSEUS (1907) stated stated that S. himalayana does not smell but in my experience it does ， though not strongly so. so. Ku 町(1 969) suggested individual variation and res 位iction of the smell to a particular stage stage of the flowering which 1 found to be the case here. Th e stench is generally stated as originating somewhere within the perigone tube. In In R. kerrii ，however ， it definitely comes from the perigone lobes ， as can be easily verified by close smelling.τ 'h e diaphragm seems to release less odour ， if at all ， but this is more difficult difficult to assess since it is so close to the lobes. The smell is not perceptible to the human nose nose in the cavity of the tube when it is sealed from the outside (cf. experiment below). 1 presume the flower st 紅 ts releasing odour at the latest when the lobes approach their their definitive positions ，continuing to emanate it as long as the flower is in full bloom. Th e male flower No. 1 smelled putridly on both days of observation. But ，if the female No. 2 represents 血e norm ， the stench ceases soon after the flower is past full bloom ， while still reddish. reddish. At this stage another ， but much weaker ， smell is perceptible at the perigone lobes and and diaphragm ，reminiscent of rancid butter. In the greyish black male No. 3 the lobes smelled smelled more like fermenting bread and mushrooms ，evidently due to mold. Buds do not stink stink intemally (when cut open). The most exciting finding in R. kerrii is its ‘anticlimax' odour: a pleasant f回. grance grance akin to dried apricots and peaches. is It released from within the perigone tube. It is is less intensive and completely overwhelmed by the 位 onger pu 町id smell except near 出e cen 汀al opening or within the cavity. Th is was assessed in the following way. Th創Ik s to the the size of the flower ，1 could put the whole of my face into it with the diaphragm closely adhering adhering to 血e edge of my face-much like a gas mask-thus excluding stench-contami- nated nated outside air from penetrating into the cavity: the fragrance was distinct. It is is It intriguing that no such fragrant emanation from blooming flowers has been mentioned mentioned for any other Ra jjZ esia species. Although it may be unique to R. kerrii ，1 am in- clined clined to believe that it might have been overlooked in some of the species due to the overpowering overpowering stench. It It is not yet clear exactly where the fragrance comes from ， but a slimy coat covering covering the underside of the disk ，column and adjacent part of the annulus interior of both sexes sexes does have this scen t. 1 am not aware that this wet film has been mentioned for any other other rafflesia.τ 'h e ramentae tested have no such scent. Th e fragrance is already perceptible in large buds which have been gnawed open by animals. Even buds severed from the host have this scent for some time; this was also mentioned mentioned by MEUER & ELLIOTI (1991) who assumed the scent would later undergo chemical chemical change into the offensive smell. The fragrance persists throughout and beyond the the blooming stage but tends to assume a more ferm 巴nted by-aroma. In the grey-black stage stage of male No. 3 it is more fermented and acrid. It is is It known that in certain flowers emanation of scent is concen 甘ated in particu- lar lar floral s佐uctures. However ， R. kerrii seems unique in the extent of the scents' contrast 釦 d the effects they have on the pollinator. 34 HANs BAN Zl G限

Deceptive Deceptive Flower ?

It It is generally assumed 由at rafflesias 紅 e Tauschblumen ， i. e. deceptive flowers. 百1ese a町 act p0 1li nators by faking the presen 回 of resources requ 凶 d by them. Visitors therefore therefore remain unrewarded for services offered. 1 believe 血at R. kerrii may be considered a deceptive flower insofar as 白epu 凶d stench stench deceitfully advertises rotting flesh meals for adult flies and brood sites for their progeny progeny when there are really none. Al so ，no nectar is apparently offered. However ， rafflesias rafflesias produce large amounts of pollen ， part of which is available to visitors entering 白e more secluded ， tenebrous parts below the disk. Moreover ，a slimy secretion is avail- able able here. In In rafflesias pollen comes 泊 form of a yellow mush (Fig. 18) thicker than milk but but more fluid 血an mayonnaise. 官1e mush is much less viscous 血組曲e p0 1li nia of certain orchids ，e.g. Paphiopedilum spp. ，which are so sticky 白紙血ey cannot be sucked up by Diptera Diptera (pers. observ ふIn the moist microhabitat of the perigone tube the pollen mush retains retains its consistency but outside the flower where the air is 合ier ，especially during R. kerrii's kerrii's flowering period ，it thickens substantially.τ 'hi s is impo 此創1t for pollination because it it can stick as a viscous clot fin 叫yωthe dorsum of the 白orax of the p0 1li nator (Fig. 15) and be carried 釘 ound for a long time. 1 found pollen mush to be tasteless to my tongue and with no dete 心table scen t. It It is unlikely 白紙 pollen mush is stolen by calliphorid flies 合'om the 朗自ers as the the cavities in which 血ey 紅 'e set and the channels leading to them 紅 e built in such a way 白紙白e mouthparts of the flies cannot reach the anthers (Figs. 22 ， 23) (cf. morphological section). section). It is a1 so highly unlikely 白紙出e pollen clot on the p0 1li nator's 血orax is used by females females to 泊， vite copulation. 百1e v儲 t majority of blowflies visiting R. kerrii were females and the clot would be in the right position to be sucked 出 a ‘nuptial' reward by mounting males. ‘Wedding gifts' are offered among Empididae (Diptera) and Bittacidae (M ecoptera) (J ACOBS & RENNER ， 1974). However ， in these cases the behaviour is highly ritualized and it it is the male which displays the prize~ In calliphorids it would at best be a casual occur- rence. rence. However ， at least in R. kerrii and in S. himalayana ，血e anthers exude ‘excessive' amounts of pollen mush ， the surplus detaching itself by gravity and dropping onto 白e 佃- nulus nulus interior not fi 紅企om whe 問 the ridges of 由e anthers' channels level off. Fl ies can readily readily imbibe it here. 1 have noticed drops of fallen pollen mush 泊 ma 旬re buds of S. himalayana himalayana before they opened (Fig. 18) (evidenced by surgery). Such drops were even present present in 血e withering R. kerrii flower No. 3，no longer visi 飽 d by p0 1li nating flies. So far 1 have not seen a fly in 曲， e act of sucking such drops (白 ey are not visible from 出e outside) but knowing the flies' pollen-feeding habits (cf. section on their biology) 1 would be astonished if 出ey did not take it up. Th at other authors failed to mention the presence presence of such drops on the annulus does not necessarily mean that other Rafflesia spp. do not produce 血em. 百 e flies indeed might have fed upon them and licked the annulus clean. clean. Besides Besides pollen mush it is likely 白紙 the flies a1 so suck the fragrant ， slimy secre- tion tion coating the p釘 ts below the disk. But ，besides moisture ，1 do not know whether it has any nu 凶 tional value to 白em. It was tasteless to my tongue. STENCH AND FRAGRANCE: RAFFL ES/A KERRII MEIJ ER 35

Figure I 3. The· festering sore' appea rance of R. kerrii (crater shape, lurid and suppuration-like colou rs), and its carrion stench, attract blowflies : two on the peri gone lobe, another at the lower, abrupt end of the band of ramentae. Drosophilas are on the ri m of the diaphragm and tip of the processes (whitish cl ue to molcl) , and 2 nerii cls are on the perigone.

Figure 14. Blowfly Chrysomya vifleneuvei . female, gazes into the cavity of male R. kerrii from the rim of the diaphragm befo re flying into the cav ity. 36 1-I ANS Bii. NZ IGER

··~

Figure 15. Female Chrysomya rufifacies with a clot of po llen mush on the back of the thorax (arro w), cleaning its hind legs on the edge of the peri gone lobe of male R. kerrii. Flies are nearl y I 0 mm long.

Figure 16. C. villeneuvei, female, the back and sides dusted with pollen of an unidenti fied flower, is on the perigone lobe of male R. kerrii. STENC H AND FRAGRANCE: RAFFLES IA KERR II M E IJ ER 37

Fi gur e 17. 1-lypop yg iops is IlII llrasv ini ，fema le， on lh e perigon 巴lob e 0 1' 1'， 巴ma le R ker rii

Figllr e 18.C r oss sec ti on throll gh a m ar llr e blld

1' 0 1' raf fl 巴sia (s. l. ) Sap ria himala y一

山 la show ing ye ll ow dr ops (ar - rows) 0 1' sur plu s po ll en mus h fa ll en on the annlllll S int 巴ri or and wa ll (bud was in clin ed too nes id巴 hence th e 'ob li qll e' tr ajec ror y) Leng rh of arr ows is 卜3c m 17

18 38 H ANS Bi\NZIGER

Figure 19. R. kerrii, wilting male, poss ibly a week or more after full bloom. No ca rrion flies are present, on ly a single mu sc id in the cav ity and plenty of saprophagous drosoph ilas.

Figure 20. 13 buds of R. kerrii being so ld ala fruit stall in Perak, Malays ia. A ll bud s have been cut together with part of the host's root s, a des tructive prac ti ce. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEUER 39

百le 仕equency and qu 叩 tity of drops of surplus pollen mush need further study. Nevertheless ， since pollen offers carbohydrates ，lipids and protein potentially for both energy energy supply and egg maturation ， 1 believe R. kerrii is not really a deceptive flower.

Arthropod Activity in and around R. kerrii Flowers In In order not to interfere with fly activity only a limited number of specimens were were caugh t. Unfortunately ，reliable identification of calliphorids requires capture and ex- amination amination under magnification so it is not possible to give exact individual numbers per visiting visiting species. Arthr opod specimens are deposited at DEFACU and the British Museum (Na t. His t.). Male t1 0wer No. 1 Observations Observations on flower in full bloom carried out 140 0- 1630 h and 1045- 1600 1600 h on 26 and 27 January ，respectively. Insects Insects involved in pollination tumed out to be calliphorid flies Ch η Isomya ville- neuvei neuvei Patton ，C.r 叫向cie 白s (Ma 配cq 伊u制)， μL町仰uωcUωiωα IC po 中r， ph 勾yrin 初na (Walk 恥e釘釘の.) r and H 照y， 'Po， 叩'P 沼YY昭gi ωop，戸si 臼S tω um ηlr αsv， 凶lini Kurahash i. Th ere might have been 5-10 仁C .v 凶illeneuvei and C.rufifacies to one L. L. porphyrina; only about 4 H. tumrasvini were seen. Males were very scarce. Up to 6 flies were were seen at the s創 ne time on or in the flower ， and more were perching on the vegetation nearby. nearby. On the first day fly activity was constant ，even 企enetic at times ，some individuals obviously obviously retuming several times. On the second day it was much reduced ，at times hardly any 出ing moving ， probably due to the cloudy weather. During During 2.5 h of observation on the first day 6 C. villeneuvei ，C.r 城向， cies and L. porphyrina porphyrina females were seen with a clot of pollen mush on the back of the thorax (Fig. 15) ，4 of which already had it prior to landing on the flower. During 5h on the second day day the numbers were 8 and 3. Only one H. tumrasvini had a small clot on the anterior right right comer of the thorax back. Th e species is rather larger than Chrysomya and Lucilia spp. spp. and must be hindered in advancing through the channe l. Typically Typically a calliphorid landed onto the perigone lobe ， or less often onto the dia- phragm ， and probed the surface with its labellum for nu 位ients (Figs. 4 ，13). Since these were were wanting ， the fly drew back its labellum ，rested for a while or crawled around and probed probed a new spot. It might then walk or fly onto the diaphragm ， advance towards the rim and and gaze into the cavity (Fig. 14) ， or fly s回 ight into it landing on a process of the disk or or more often on the cavity's wall. The labellum probed again. Th en the fly might crawl downwards toward the annuli and the darker recesses below the disk out of sigh t. It circumambulated circumambulated the disk's column more or less completely as after a minute or so it could be be seen to reemerge from another point at the tube's base. In a number of individuals the back back of 出e thorax was smeared with pollen mush. Th e fly then crawled up the wall ，rested ， or or flew off. Other Other flies observed were one Sarcophaga peregrina Robineau-Desvoidy 佃 da few individuals each of 2-3 species of Muscidae which settled on R. kerrii but did not penetrate penetrate below the disk. Flesh-flies (Sarcophagidae) such 錨 S. peregrina feed and breed in in carrion and other decomposing organic matter ， while the Muscidae have more catholic habits. habits. There is no indication that these flies are involved in pollination but S. peregrina needs needs to be investigated further. 40 HANs BANZIGBR

Drosophila Drosophila spp. ，known to be attracted to fermenting organic m 甜 .er ， were ve:η common ，and a few individuals of Neriidae (Di ptera) were also s田 n. Some of these entered entered the tube but none pene 回.t ed inωthe twilight 釘 'ea below the disk. Th ey are unlikely unlikely to play 佃 y role in pollination because of their small size.

A few individuals of 由ew 儲 pPo か'bioides ， probably gracilis v. d. Vecht ，a genus new to Th ailand ，persistently flew around the flower ， at times settling on the lobes but not advancing 泊to 由e 加be. Th ey certainly are not involved in pollination. Th ey occasionally 1阻 ded on my skin and gnawed at it. At least 4 individuals were together wi 白 blowflies c. c. megacephala (Fabricius) and C. rz 抗向 cies (Macqu 副) on the severed head of a macaque where where the w 錨 ps were biting off flesh ，a behaviour not previously known in these wasps. Aspec 泊施n was also seen visiting the fragrant inflorescence of Tetrastigma sp. 7. No 出泊g is is otherwise known about the biology of these wasps but according to Dr. B. Petersen (in litt.) litt.) they ‘are rather fierce when n伺 r 出.e nest with about 2000 spec 泊施ns'. A spider of the Salticidae ，possibly Jot ωsp. ，was lurking on a plant adjacent to 血 e rafflesia; it grabbed and ove 叩owered a C. villeneuvei. Two more calliphorid spp. were present in the study area but ， so far ， were not caught caught from R. kerrii ，viz. C. pinguis (Walker) 血 dH. u拘mata Bigo t. It c佃 be expected 血創出ese ， and C. megacephala also frequent 出e flower. Female flower No. 2 Observations Observations of flower somewhat past full bloom made during 092 0- 1700 h， 1115-1700 1115-1700 h and 103 0- 1300 h on 7，8 ，佃 d 10 April ，respectively. On the frrst day calliphorids arrived as late 拙 1020 h and the last ones left about 1515 1515 h. An estimated 5 C. villeneuvei ，20 L. porphyr 初a and 6 H. tumrasvini (Fig. 17) visited visited the flower. Some individuals ，especially H. tumrasvini ，made repeated attempts to enter enter the tube. None had pollen on the back. 百le behaviour of the calliphorids on 白is female rafflesia ， which discontinued to emanate emanate carrion stench ，w 部 in st 紅'k con 位'ast with 白紙 observed on the male No. 1 with its its pu 凶d smell. Of 血e 31 individuals only 12 (2 ，7， 3 sp 回 imens of the species mentioned above above in this sequence) flew into 出e perigone tU be; of these only 6 (2 ，3 ， 1) did actu 剖ly settle settle (on disk ，processes ， wall) while the other 6 flew inωand out again without landing. Most importantly ， none proceeded down into the obscure recesses below the disk. As discussed discussed in the following section ， these blowflies appe 釘 to need preconditioning by carrion-like carrion-like stench to advance into dark cavities. No calliphorid was detected visiting the flower the following days. Other Other insects 企equen 曲19 the flower were as mentioned for the male ，including a s. s. peregrina on 出 e first day. New visitors were 4 Therates kraazi Hom (Cicindelidae). Tiger Tiger beetles 釘 'e known to be aggressive predators of other insects and T. kraazi probably were were lurking for prey a位 acted by the flower's scents. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEUER 41

Male flower No. 3 Observations Observations on this wilting ，grey-black flower still retaining much of its original shape shape were carried out during 10 00- 1300 h on 6 March ，093 0- 1000 and again 1530 ー1600 h on 7M 紅 ch. No calliphorid fly was seen on the flower although 1 did see them visiting f詑sh human faeces in 由e area. On the first days a S. peregrina and one neerid fly settled briefly on the lobe; a muscid fly (Fig. 19) did so on the inner wall where there were plenty of mites. mites. At least two species of drosophila ，viz. D. su 抑rigaster Duda and D. albomicans Duda ， were flying around the rafflesia. On the second day no insects but great numbers of of drosophila were around ，especially on the disk. A dry and shrivelled bud of colony No. 1 which must have died while half open ， had plenty of ants inside. One bud of colony No. 4， which died shortly before blooming due due to a hole gnawed by some rodent ， had a blattid inside. Other shriveled flowers had termites termites in their bases.

Biological Biological Notes on the F1 y Pollinators For For clarification of the mechanism of pollination in R. kerrii it is necessary to understand understand certain aspects of the flies' physiology and behaviour. Only data pertinent to fly fly pollination are mentioned. Calliphoridae (blowflies) such as Calliphora (b luebottles) ， Luc i/i a (greenbottles) and Ch ηsomya are fairly well researched because of their impor- tance tance as health hazards. Adults are attracted to meat ，carrion ， faeces and other decaying animal animal matter for egg laying and/or feeding. 百le larvae of Chrysomya bezziana V i1l eneuve and and Lucilia sericata Meigen are parasites in wounds ，abscesses and body cavities of man and and animal (SM 汀 H， 1973) A number of workers on pollination ecology have overlooked the that fact flower visiting visiting and/or sucking of sug 紅 y plant sap is essential for adult carrion flies (cf. Fig. 16). They frequent especially Umbelliferae (STE 町 ER ，1948) ，Rosaceae ， Euphorbiaceae and Compositae Compositae (KUGLER ，1951) ，Salicaceae and Araliaceae (SCHREMMER ， 1963). Except for the the Compositae ， where they presumably feed mainly on pollen (KUGLER ，1951) ，on the other other families mainly nectar is taken. Calliphoridae play a role as pollinators in 仕uit orchards orchards such as mango (Mangifera indica L.) in Thailand (SUVARNAYA 叩 IPAT ， 1984). Another Another food calliphorids eagerly feed upon is sap of fruit ，especially if fermenting (STE 別ER ， 1948). Sugars Sugars are the main energy suppliers for both males and females ， while protein is is needed for egg maturation. Females of L. sericata. L. cuprina Wiedemann and C. rポifa- cies cies require at least one meal of protein to lay eggs (MACKERRAS ， 1933). But meat alone is is a poor source of nourishment for both males and females. While L. sericata die within 2-3 days if given water but no sugar (WIGGLESWOR 叩， 1972) ， they lose weight and also die die relatively soon if given meat and water alone. They increase in weight and live 1-2 months months if given sugar and water but no meat; addition of meat will increase weight and age age only slightly in males though in females the weight rises significantly due to egg maturation maturation (not available for energy supply) (Ev ANS ， 1935 ). is It clear that a diet of sug 訂 S and and meat is best for both sexes but especially for the females. 42 HANSBλNZIGER

Pollen Pollen offers carbohydrates ，proteins ，lipids and other nutrients which could cover all all the adult flies' nu 位itional needs; however ，1 have found no reference analyzing the actual actual value of pollen for calliphorids. Th ere is little doubt that despite the chemically resistant resistant walls of the pollen ，its nutrient content can be digested by the flies' enzymes pene 町ating through the germination pores of the pollen ， as is the case with bees (STANLEY &L 町 S阻 NS ， 1974). The average age of Lucilia in captivity was 7 weeks but they can live over 3 months months (SALT ， 1932; MACKERRAS ， 1933). Calliphorids are energetic f1 yers and recapture of of marked individuals has proved that they can f1 y distances of 22 km within a few days (BISHOPP (BISHOPP & LAAKE ， 1921). Because of these two features ，1 think these flies should be considered considered as potential long distance pollinators ，comparable ，on a lesser scale ， to the euglossine euglossine bees (D 阻 SSLER ， 1968; JANZEN ，1971) ， an important factor for rare plants such as as raf f1 esias. According According to KUGLER (1970) in Lucilia and most other insects ，attraction to f1 ow- ers ers from a distance is optically guided while at close range olfaction is more important. Exceptions Exceptions are at 位action to f1 0wers emanating carrion ， excrement and other smells of de- caying caying animal matter which work as long distance lures. In experiments (KUGLER ， 1956) showed that carrion flies are drawn to brownish red colours in the presence of carrion or faeces faeces smells but not when it is absen t. AUTRUM & STU 悶 'F (1 953) proved experimentally that ，unlike bees and m 叩 y other insects ，Calliphora erythrocephala Meigen is able to see red red as a colour. In the absence of fetid smells ， yellow and white 釘 e the most attractive colours colours for calliphorids (KUGLER ，1951). 官ley 紅 'e likely to act at medium and close r加 ge in in raf f1 esia. At medium distance they may help make the otherwise essentially reddish R. R. kerrii more conspicuous. Th e whitβblotches on the underside of the diaphragm are visible visible at low level flight ， the moldy yellowish white patches (i f present) on the disk' s processes processes at higher leve l. At close range ，from 出e edge of the diaphragm ， the recurring white white blotches on the dark underside of it could act like a checker mode l. Th e black-and- white white checker design has been shown to be more at 釘active than a pale grey or dark one (S 百町田， 1948).

Wrinkled Wrinkled surfaces ，warts ，filaments ，processes also appe 訂 to exert attraction (KUGLER ， 1956). VOGEL (1961 ， 1963) pointed 0脱出at in most cases processes like those on the disk ，besides being optically attractive ，also release scents from special glands ，出 e osmophores osmophores of Ar cangeli. 百le rows of bristles on the crests of the processes may be 泊， dicative dicative of such osmophores. The nodules and ramentae might also draw attention opti- cally. cally. As mentioned by VOGEL (1 961) Diptera generally are known to react with posi- tive tive phototaxis when in dangerous or unfamiliar surroundings -they move toward the light. light. However ， an observation 1 have made on carcasses of monkeys in the study area is of of importance. Chrysomya megacephala 叩 d C. rufifacies readily enter dark recesses ， cavities cavities and holes when on carrion. Its smell seems to condition them not to be scared of darkness. darkness. The observations on rafflesia No. 2 (previous section) indicate 白紙 without previous previous conditioning by stench such calliphorids do not advance into darkness. STENCH AND FRAGRANCE: RAFFLES/A KERR/I MEIJER 43

cd

a

23

Figure 2 1. Obl ique view of cross secti on of di sk and column (part) of male R. kerrii depi cting the general situation of the pollination sequence shown in Fi gs. 22 - 28 (Figs. 27 - 28 are of fema le which lacks the chan­ nels). Schematic; proportions in all draw ings not to scale.

Fi gure 22. Cross section of disk and column showing detail s of hairs lining the channel leading to the anther.

Figure 23. View of a disk sector from below, showing the position of anther and hairs which protect its access from three sides.

Figure 24. Calli phorid tl y entering a channel leading up to an anther. Oblique view; disk omitted. a = anther; c = column; ca =channel leadi ng to anther; cd = coll ar of disk; d =disk; p = process; s = sulcus. 44 H ANS B iiNZIGER

ss

Figures 25-28. Pollination of R. kerrii. Above: male (hairs omitted); be low: female. 25: Calliphorid nearing the end of the channel just before the back of the thorax touches a drop of po llen mush hanging from the anther. 26: After a short advance the fl y retreats taking along a clot of pollen. 27 : Calliphorid with clot of pollen climbs the wedge below the disk bringing the clot near the stigmati c surface (ss). Further ad­ vancing will cause the clot to be brushed off the sti gmatic surface. 28: The fl y retreats leaving behind most of the pollen. STENCH AND FRAGRANCE: RAFFLESIA KERRII 恥ffiUER 45

The Pollination Mechanism (Figs. 21-28) From a distance of probably many dozens of meters a po IIi nator c aI liphorid en- counters counters a whiff of the pu 回d smeII from a R. kerrii. Following the gradient it fIi es toward the the source of the smell and perceives the lurid reddish colour. Olfaction is now assisted by vision including the white of the blotches on black background and/or white patches (if (if any present) on the processes. Th e fly locates the flower and settles on a perigone lobe. As neither tarsi nor labellum sense any rotting food ，it crawls around to find a more promising promising spo t. Upon climbing or flying onto the diaphragm ，when near its m 訂 gin it can see see the checker design and the processes ，bo 出 attracting the fly towards the cavity of the tube ，especially if the processes 訂 e moldy ('festering sore' appear. 叩 ce). 官 le insect flies into into the tube ， landing first on a process or directly on the wal l. Here the pu 凶 d smell is weak or absent wh iI e the s仕onger fruity fragrance announces plant food; the fly crawls down below the disk. (Movements below the disk were not seen; the following sequence is is a deduction based on morphologic aI and behavioural evidence ，combined with an aI o- gous gous mechanisms in other flowers.) Th e fly crawls around the column ， presumably probing the the sIi my film and sucking poIIen mush (if any left by ear Ii er visitors) dropped down near where the channel to the anther starts. Searching for more it climbs up the channel (Figs.

24 ， 25) until ne 紅 the the end the back of 出e thorax comes into contact with the pollen mush hanging from the 組 ther. Further advance is obstructed by the cavity's w aI I，and turning turning is prevented by the n創 Towness of the channe l. Wh ile retreating ，it w iIl take aI ong a clot of poIIen mush on the thorax (Fig. 26). It may search for more food or come out 合om below the disk ，w aI k around ， or fly off. In In the female flower the whole process is presumably the same up to the moment the the fly crawls under the disk. SIi me is possibly sucked. Given the lack of channels on the column ， the fly can proceed in any direction but w iIl tend to cIi mb the narrowing wedge (Fig. (Fig. 27) between the annulus/column and the stigmatic surface. Sooner or later the back of of the thorax is brought into contact with it ， smearing or rubbing the pollen clot (if any is is present) off onto the velvety mat of the stigmatic surface (Fig. 28).

DISCUSSION It It is perhaps typic aI for rafflesias that the elucidation of the po IIi nation mecha- nism had to wait for more 由加 150 years although there was Ii ttle doubt about it ever since DELPINO'S DELPINO'S (1873) fundamental work on poIIination. DELPINO recognized eight types of insect insect poIIination systems (and 45 groups of flower scents!) and characterized the sapro- miofile miofile (sapromyoph iI ous) flowers thus: display lurid colours (reddish) ，have cadaveric stench stench and are poIIinated by carrion fIi es-all sa Ii ent attributes of rafflesias. Of course many researchers have proposed fIi es as po IIi nators. The reason for the interregnum are the difficulties difficulties connected with the study of blooming flowers and the other peculiarities of the parasite parasite mentioned in the introduction. 百 e location of the stigmatic surface had Ii kewise remained uncertain (Ku 町， 1969). 1969). But there are compeIIing mo 中hologic aI and functional reasons for assuming that the the annular band on 曲e underside of the disk is the stigmatic SUrface. Th e whitish mat of hairs hairs or pap iII ae with which it is covered is typic aI of the stigmatic area of some other flowers flowers (KUGLER ， 1970). Considering the po IIi nation syndrome from a functional aspect ， 46 46 HANS BANZIGER

the the annular band seems to be the only ar 回 where the pollen clot can possibly be smeared off off the back of the calliphorid fly. It It has been mentioned 白紙 only smells of decaying organic matter act as long dis- tance tance flower attractants in pollinating insects. Other scents work at close range and flowers generally generally have to rely on visuallures for long distance attraction. If this is so ， then in the dense dense vegetation of tropical forests stenchy flowers have a selective advantage over fra- grant grant ones ，however optically conspicuous they may be ， since visibility is reduced there. This This may help explain why smell-producing flowers like Rafflesia ，Aristolochia ，Amor- phophallus ，Sauromatum ， Masdevallia and others 釘 'e exclusively or mainly found in the tropics. tropics. The Stapelia of the S. African semi-deserts would seem to contradict this but because because of their specific habitats a different explanation may apply in their case. Stenchy plants plants in the temperate zone seem to grow mainly where there is thick vegetation ， as is the case case with the stinkhorn (Phallus impudicus L. ex Pers.) (SCHREMMER ， 1963). 1 follow FAEGRI & PUL (1 979) in viewing sapromyophily as a more recent deve- lopment lopment than other pollination systems. Stench ， as an evolution 紅 y ‘new decoy' is a more effective effective lure 由組合uity fragrance ， yet the latter has not become obsolete in R. kerrii. Its Its possible function may be in continuing to adve 同 se carbohydrate food to prevent calliphorid calliphorid flies from being mislead by the stench to lay eggs in an unsuitable brood site. It It is not in 出e flower's interest to kill off prospective pollinators or to endanger the development development of its own fruits and seeds by inviting oviposition by flies the larvae of which could could be destructive. In fact ， while the larvae of the mentioned Calliphoridae normally develop develop in decaying animal flesh they probably can survive in other decomposing organic organic matter as many of their relatives are known to (SMITH ， 1973). Larval C. bez- ziana ziana infest mammalian sores but will readily invade and devour healthy ， live tissue (SMITH ， 1973) and L. sericata is so voracious in this that it eventually causes the death of of sheep (DONGES ， 1980). L. porphyrina has apparently been reported to cause fatal myia- sis sis in toads (WY AτT ， in litt.) which were healthy initially. Moreover ，little-known flies such such as H. tumrasvini or other ， not yet observed ，rafflesia-pollinating calliphorids may have have larvae with more catholic tastes than the ones so f; 釘 seen on R. kerrii ，and accept decaying decaying plant tissue. Difference in the pollinating fauna ，e.g. presence of more steno- phagous larvae ，may explain why rafflesias in Malesia are without fruity fragrance. However ， unless females are under ‘egg-laying' stress due ， for instance ， to long-delayed opportunity opportunity to lay ，oviposition is often induced only by a combination of stimuli perceived by the antennae ，labellum ， legs and/or ovipositor. It may therefore be questionable whether blowflies blowflies would be mislead into laying in rafflesias ， whatever the scents. Nevertheless ，出 ere are a number of documented cases of ‘misled' oviposition especially especially in stapelias (FAEG 則&PUL ， 1979; KUGLER ，1970) ， the orchid Paphiopedilum rothschildianum rothschildianum (Reichb. f.) Stein (ATWOOD ，1985) ， with subsequent larval death reported in in Aristolochia grandiflora Swartz and other Aristolochia spp. (CAMME 乱 O 田 R， 1923). τbe opposite is also known where the brood of the pollinating fly developed normally in the the male Alocasia pubera (PIJL ， 1953). Even in R. kerrii maggots have been mentioned (KERR ， in MEUER & ELLIOπ ， 1991) in old flowers. Moreover ，1 have found tiny larvae in rotting rotting males. 1 suspect they were drosophilids or other saprophagous flies living upon decomposing plant matter. A rich saprophagous fauna and flora is only to be expected in such such large rotting flowers. STENCH AND FRAGRANCE: RAFFLESIA KER Rl I MEUER 47

Perhaps 白e presence of the fragrance in R. kerrii but lack in other raf f1 esias can be be explained from an evolutionary perspective. R. kerrii could be a link between the more primitive primitive pollination system based on scents advertising nec 町 an d/ or pollen ，and 白e derived derived sapromyophily of more advanced raf f1 esias of Malesia ， where speciation has been more rapid. R. kerrii would have developed stench production in the perigone lobes but not in in the cavity where it retained the fragrance; for pollination an olfactory lure is nec 四 sary to to at 回 .ct flies into the innermost ， dark part of the f1 ower. Malesian rafflesias might have advanced advanced further ，replac 泊g the 企agrance with stench. Fragrance 部組‘oviposition deter- rent' rent' may be an advantage ， but not vi ta1， at least in Malesia. Detailed future research may prove 白紙 minor amounts of 仕agrance continue to be emanated 合om the cavity of some Malesian Malesian rafflesias. My view 由at R. kerrii is not really a deceptive f1 0wer would receive further impetus impetus if the interpretation that 出e 企uity 仕agr 叩 ce keeps flies from laying eggs should prove prove co 町 ec 1. In the light of the above ， the function of such as 仕組ge s加 C旬 re as 血e diaphragm diaphragm may be to keep the stench off from the tube and prevent the fragrance 企omωo fast fast diluting out of i1. It also works somewhat as a roofωreduce rainwater falling into 白e interior ，assisted to some extent by the disk ー-c entered exactly below the diaphragm's opening-the high collar keeping the water on the disk. Waterlogging is a lesser problem for for R.kerrii and S. himalayana which elegantly solved it by f1 0wering when rains 釘 e spo ・ radic. radic. It is not yet understood how Malesian rafflesias cope with this problem (cf. Jusτ 芭SEN ， 1922 and W 町 KLER ，1927). Finally ， the white blotches on 出e dark background of of the underside of the diaphragm probably work as an a町 ac 旬n1. In my view the blotches should should not be inte 中reted as ‘light windows' (as in KUUT ， 1969). L 町DNER'S (1 928) light windows of Aristolochia lindneri Berger ，CAMMERLOHER'S (1923) of A. grandiflora ， VOOEL'S (1 961) of various Ceropegia spp. ， and TROLL'S (1951) of yet other species ， are diaphanous diaphanous structures with a specific function ，viz. to illuminate an otherwise dark recess of of the f1 ower. 百lI s is necessary to draw phototactically positive flies to 白紙 p釘 ticul 紅 section. section. Th e blotches in rafflesias 釘 e wrongly located for that function which would be ， in in any event ，useless for the calliphorids since 白ey readily enter dark areas ， at least when conditioned conditioned by stench. Also ， while indeed diaphanous to some extent ， the blotches 釘 e more e釘ective in re f1 ec 血19 the ligh 1. KUI 汀(1 969) commented that ranging animals from ants to elephants have been suggested suggested to act as dispersal agents of rafflesia seeds. In my study area 由e mouse deer may be be involved besides squirrels. 百 le latter have been reported to eat the fruit (MEUER ，1958). MEUER also suggested 白紙 they c紅 ry the seeds on their feet and effect inoculation when their their claws penetrate the bark while the squirrel climbs a new Tetrastigma liana. Inocula- tion tion may be more likely to occur during rummaging by wild boar (MEUER ， 1958) but they were were not present at my study sites. 1 出ink 曲at the soil fauna ， mainly arthropods and nematodes ，may be more effective. Myriads of them populate the upper humus-rich soil s凶凪 Many of them gnaw ，scratch and suck from rootlets ，leaving wounds ， often micro- scopic ，出 rough which the tiny root filaments of the seedling could penetrate Tetrastigma roots. roots. It is not impossible 由at the earliest s阻ge of the rafflesia seedling may enter into some kind of beneficial relationship with soil fungi until its rootlets can infect a Tetras- tigma. tigma. Th e relatively modest size of infected T. quadrangulum may be an indication 也at infection infection by seedlings' rootlets is more common than suspected (cf. also other explanation 48 HANs BANZIGER

in in s即 tion on hosts). It It has been assumed (e.g. ME 田 R ，1958) 白紙 for successful po 11in ation of raffle- si 邸， m a1 e and fem a1 e flowers must be present at 白e same time ，and obviously not too far 合'om each 0 血er ，a serious drawback for such short-lived ， rare flowers. In白 e light of the new results above ，it is clear 白紙 pollination may occ 町 wi 白 flowers blooming many days or or weeks ap 制，a1由 ough 由e e: 旺ectiveness must diminish with time. Th e pollinating ca1・ liphorids 紅 e long-living ， strong fliers ， and the pollen clot firmly sticks to the back since the the pollen mush solidifies to some extent 泊 the outside air. Th ey could act 邸 long distance pollinators. pollinators. It would be conceivable ，a1 beit ex 回 mely unlikely ， that pollen 企'om a flower 泊 Prachuab Kh討 ikhan in Janu 釘 Y could be delivered onto a fem a1 e in Surat τ'h ani ，sever a1 hundred hundred km to 白e sou 也，由民 e months later in Apri l. Th us ， even isolated ，single rafflesias can can play a role in helping conserve a species 剖 d 由ey should not be written off. Yet the present present study indicates 白紙 for long-term surviv a1 rafflesias are dependent on an unusually diverse 釘 ray of factors: a forest rich 泊出.e rl 叫 u国 d Tetrastigma spp.; sui 旬.b le frugivorous bats ，appropriate 臨時S凶a1 and arbore a1 mamm a1 s for dispers a1 of Tetrastigma seeds; sufficient sufficient medium to large sized mamm a1 s (e.g. primates ，ungulates ，etc.) to provide ， once dead ， brood sites for larvae of calliphorid pollinators (car 加 hydrate food for the ad 叫ts should should be no problem for gener a1 ists like calliphorids); possibly sql 凶rrels and sm a1 1 terres-

凶a1 mamm a1 s as dispersing agents of rafflesia s問 ds; 佃 da he a1 thy layer of soil and humus rich rich in soil fauna which 1 suspect is the main agent allowing new host infection. An y conservation conservation program for rafflesia has to consider these conditions in toto ωbe effective 泊出e long term.

A C K N 0 W c- E D G M E N T S

Sincerest 血創 lks 釘 e due to Dr. S. Elliott ， Biology Dep t.， for criticizing the manu- script script and for identifying bud materi a1企 om Malaysia; to Mr. J. Maxwell ，Pharmacy Fac. ， for for help in iden 的 ing Tetrastigma spp.;ω 地 U. Ari吋 at for Th ai references ，附. P. Sukum a1 anan and Dr. M. Titayawan ，部 well as the other colleagues of the Enωmology Dep t. for their continuous support of my research; to Dr. P. Smitamana ， Plant Pathology Dep t.， for clarification about plant infection mechanisms; a1 1 of Chiang Mai University. 1 創 n particularly indebted to Dr. N. Wyatt ，British Museum ，whom 1 pestered like a blowfly with with blowflies for identification ， his patience being matched only by 白紙 of Mr. G. Linder ， ETH - Zurich Library ， with reference trac 泊g. 1 am grateful to Prof. Dr. W. Brockelm 佃， Mahidol University ，Bangkok ， for improving the manuscript. Dr. B. Petersen ，Zo ologic a1 Museum ，Copenhagen ，identified Hymenoptera and Dr. R. Naviaux ，Domerat (France) ， Cicindelidae; Cicindelidae; Dr. P. Schwendinger ， Innsbruck University ， did so with spiders and Dr. G. Bachli ，Zo ologic a1 Museum ，Zurich ， with Drosophilidae. Dr. T. Smitinand ，Bangkok For- est est Herbarium ，and Pr of. Dr. A. Nilsson ，Upps a1 a University ， gave va1 uable suggestions. h 加ce Bhisadej Rajanee helped wi 出 official ma 悦ぉ while Mr. J. op de La ak ， Coffee Pr oject Chiang Mai ， supplied obscure references. 官le work described in 白is paper was p制 ly funded by British Ec ological Society Small Ec ologic a1 Pr oj 回 t Gr ant No. 805. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEIJER 49

REFERENCES

ATW ∞D， J. T. 1985. Pollination of Paphiopedilum rothschildianum: Brood-site deception. Nation a1 Geogr. Research ， Spring 1985: 247 -254.

AUTRUM ，H. 組 d H. STUMPF. 1953. Elek 回 physiologische Untersuchungen uber das Farbensehen von Calliphora. Zeitschr. Zeitschr. f. vergl. Physiol. 35: 71 - 104. BAIN ，J. R.組 dS. R. HUM 聞隠Y. 1980. A Projile ofthe Endangered Species ofThailand. Rep. No. 4，0 丘Ec o1. Serv. ，Fl orida State Museum ，367 pp. BANZIG 殴， H. 1988. How wildlife is helping to save Doi Suthep: Buddhist Sanctuary and Nation a1 Park of

百 ail 組 d.ln 1. Hedberg (ed.) ， Systematic Botany -a key science for 回 pic a1 research and d∞umen ・ 削 on. Symb. Bot. 均s. 28: 255 - 267. BANZIG 限， H. 1989. Lard izab a1 aceae: New plant fam i1 y for 百lailand ‘predicted' by rar 官 moth on Do i Suthep. Nat. Nat. Bull. Hist. Siam Soc. 37: 187 - 208.

BEAMAN ， R. S. ， P. J. DECKER 佃 d J. H. BEAM 刷. 1988. Pollination of rafflesia (Rafflesiaceae). Amer. J. Bot. 75: 1148-1162. BISHOPP ， F. C. and E. W. LAA 阻. 192 1. Di spersion of flies by fligh t.よ Agric. Res. 21: 729 -766. BROWN ， R. 182 1. An ac ∞ unt of a new genus of plan 岱， named Rafflesia. Trans. μ inn. Soc. Lo nd. 13: 201 - 234. CA 剛 ERLOHER ，H. 1923. Zur Biologie der Blute von Aristolochia grandiflora Sw 釘 tz. Osterr. Bot. Ze itschr. 72: 180-198. CRAIB ，W. G. 1926. Florae Siamensis Enumeratio. Siam Soc. ，Bangkok ，809 pp. DELPINO ， F. 1873. Ul teriori osservazioni e considerazioni sulla dicogamia nel Regno Vegetale. IV. De lle piante zoidiofile. zoidiofile. A 凶 Soc. Ital. Sci. Nat. 16: 151 - 349. DONGES ，J. 1980. Parasitologie. G. 百 ieme ，Stuttg 副， 325 pp. DR 図 SLER ， R. L. 1968. Pollination byeugloss 泊e bees. Evolution 22: 202 - 210. Ewσπ ，S. ，J. F. MAXWELL and O. PRAKoBVrrAYAKIT-BEAVER. 1989. A tr 創 ISect survey of monsoon forest in Doi Su 曲ep- Pu iNation a1 Park. Nat. Bull. Hist. Siam Soc. 37: 137 -17 1. EVANS ，A. C. 1935. Some no 胞s on the biology and physiology of the sheep blowfly ，Lucilia sericatal Meig. Bull. Ent. Ent. Res. 26: 115 - 122. FA 回 RI ， K. and L. VAN DER P凪 1979. The Principles of Pollination Ecology. Pergamon Pre ss ，Oxford ，225 pp. GAG 阻 PA 町， F. 1912. Fl ore gener a1 e de l' In d← Chine. M 邸 son ，Paris ， 1070 pp. GAGNEPA 則， F. 1930. Fl ore gener a1 e do l' In d仔 ζhine. Supplemen t. Masson ，Paris ， 1027 pp. HANSEN ， B. 1972. pp. 182-184 泊 Flora of Thailand. 11. Rafflesiaceae. App 1. Sc i. Res. Corp. Th ailand ，Bang- kok ，93 -196. HOSSEUS ， C. C. 1907. Eine neue Rafflesiaceengattung aus Siam. Bot. Jb. 41:55-6 1.

JACOBS ，W. 加 dM. R 町 NER. 1974. Taschenlexikon zur Biologie der Insekten. G. Fischer ，Stuttgart ，635 pp. JANZEN ，D. H. 197 1. Euglossine bees as long -

Jus 官 SEN ， P. T. 1922. Morphologic a1 and biologic a1 notes on Rafflesia flowers ，observed 泊由e highlands of Mid ・ Sumatra Sumatra (p a伽ngsche Bovenlanden). Ann. Jard. bot. Buitenz. 32: 64 - 87. KUGLER ，H. 195 1. Blutenokologische Untersuchungen mit Goldfliegen (Lucilien). Ber. Dtsch. Bot. Ges. 64: 327 - 34 1. KUG 回 R，H. 1956. Uber die optische W 廿kung von Fl iegenblumen auf Fl iegen. Ber. Dtsch. Bot. Ges. 69: 387- 398. 398. KUGLER ，H. 1970. Blutenokologie. G. Fischer ，Stuttgart ， 345 pp. Kmπ ，J. 1969. The Biology of Parasitic Flowering Plants. Univ. C a1 if. Press ， Berkeley. LA 百 FF ，A. 1984. Studies in Ma lesian Vitaceae. VII. 百le genus Tetrastigma 泊 the Malay Peninsula. Gard. Bull. Sing. Sing. 36: 213 - 228. LINDNER ， E. 1928. Aristolochia Lindneri Berger und ihre Bestaubung durch Fl iegen. Biol. Ze ntral- BI. 48: 93 -10 1. MACKERRAS ，M. J. 1933. Observations on 曲e lif e-ゐ istories ，nu 肘tio na1 requ 回 men 包 and fecundity of blowflies. Bull. Bull. Ent. Res. 24: 353 - 362. MAT SALL 回， K. and A. LA 慨 1989. A new speciωof R相抑制no 蜘 on 0 血er species from Trus Madi Range ，Sabah (Bomω). Blumea 34: 111 - 116. MElJER ，W. 1958. A contribution to 血e 蹴 onomy and biology of Rafflesia arnoldi in West Sumatra. Ann. Bogor. 33-44. 50 HANs BAN Zl GER

MEuER， MEuER， W. 1984. New species of RajJl esia (Rafflesiaceae). Blumea 30: 209 - 215. 地皿R，W. and S. E 比 101T. 1991 (1 990). Taxonomy ， ecology and conservation of Rafflesia kerrii Meijer 副 southemτ 'h ailand. Nat. Hist. Bull. Siam Soc. 38: 117 -133. NIYOMDHAM ， C. and R. KUBAT. 1987. B a1 anophoraceae and Raffl esiaceae in Th ailand. In H. Weber and W. Fors 田 uter (eds.) ，Parasitic Flowering Plants ，Pro c. 4也 ISPFP ，M 紅 burg. R 孔， L. VAN DER. 1953. On the flower biology of some pl 飢餓合。m Java. Ann. Bogor. 1: 77 - 99. SALT ， G. 1932. Th e natural con 住'01 of the sheep blowfly ，Lucilia sericata Meigen. Bull. Ent. Res. 23: 235 - 245. SCHRElI品4ER， F. 1963. Wechselbeziehungen zwischen Pi 1z en und In sekten. Beobachtungen an der Stinktnorchel ， Phallus Phallus impudic ωL. ex Pers. Osterr. Bot. Ze itschr. 110: 380 -4 ∞. SMITII ，K. G. V. 1973. Cal liphoridae and Sarcophagidae. In K. G. V. Smith (ed.) ，Insects and Other Arthroptx お ザMedicallmportance. British Museum (N at. Hist.) ，Lo ndon ， 561 pp. S附百NAND ， T. 1980. Thai Plant Names. Fl 凶 my Pu blishing ，B 銅 gkok ，379 pp. ST 釧 LBY ， R. G. and H. F. L lN s臨 NS. 1974. Pollen: Biology ，Biochemistry ，Management. Springer ，Be r1 in ，307 pp. 'BINBR， Sτ 'BINBR， G. 1948. Fa1 1enversuche z町 Kennzeic 伽IUng des Ve 巾a1飽 ns von Schmeissfliegen gegenuber verschie- denen denen Merkm a1 en ihrer Umgebung. Ze itschr. f vergl. Physiol. 31: 1- 37. SUVARNAYA 叩 IPAT ，W. 1984. Mango Cultivation. Kasetsart U 凶versity ，Bangkok ，64 pp (in 百凶). 百 OLL ，W. 195 1. Botanische Notizen n (6. Diaphane S回 kturen). Akad. d. Wiss. u. d. Lit. Mainz ， Abh. math.- nat. nat. Kl. 2: 48 - 62. (Not seen). V ∞乱， S. 196 1. Die Bestaubung der Kesself a1 1en - Bluten von Ceropegia. Beitr. Biol. Pflanzen 36: 159 - 237. V ∞乱， S. 1963 (1 962). Duftdrusen 凶 Diens 飽 der Bestaubung. Ube r Bau und Funktion der Osmophoren. Akad. Wiss. Wiss. (Mainz) ， Abh. Math.-nat. KI. 10: 6∞-763. WHITMO 阻， T. C. 1975. Tropical Rain Forests o[ the Far East. Clarendon Pre ss ，Oxford ，282 pp. WIGGLBSWORTH ， V. B. 1972. The Principles o[ Insect Physiology. Ch apman & H a1 1，Lo ndon ，827 pp. W 別乱凪 H.1927. むber eine Raffl esia aus Ze n位置 lbomω .Planta 4: 1- 97. STENCH AND FRAGRANCE: RAFFLESIA KERRII MEI 厄 R 51

Appendix 1. R. kerrii in Ranong Province ， S. Th ailand

On 13 April ，1991 ，2 buds 7 cm across and a rotten male flower which did not open open fully were found growing on a large individual (1 5 cm across stem) of Tetrastigma sp. sp. 12 (same sp. as Banziger coll. No. 941 of that neighbourhood). This is another new host for for R. kerrii. The place is at 150 m elevation ， in an evergreen forest. Following c a1 liphorids were p陀 sent: Catapicephala sinica Fan ， Hemipyrellia ligurriens (Wiedemann) ，Hy- popygiopsis popygiopsis iゆtmata ，H. tumrasvini ，Lucilia papuensis Macquart and L. porphyrina. Lo cal gum (Dipterocarpus sp.) tappers were aware of the rafflesias in the area. 百ley collect them for for sale to a nearby temple for up to 100 Baht depending on the size. Th e resident monk prepares prepares concoctions for various purposes.

Appendix 2. R. kerrii in Upper Perak State ，W. Malaysia

In In Baling a total of some 25 buds were on sale at two street stalls selling fruit ， for for 10 M$ each (Fig. 20). 1 was told they come from the deep forest in the 紅 ea. In In Grik a dozen buds were on sale in a traditional medicine pharmacy. According to to the seller ，Orang Asli from the forest bring them in by the hundreds during the ‘ dry season' season' months of October to December. Concoctions are given to pregn 佃 t women but they 紅 e also invigorating to men. AIso on sale were two medicine preparations ，a syrup (‘ Faizol tonic ') and pills (‘ pill buasir') ，containing among other ingredients ，rafflesia ex- tracts. tracts. Both Both B a1 ing and Grik 訂 e about 20 km from the nearest frontier with Th ailand. Although Although it is most likely that the plants came from within Malaysia ，it cannot be excluded that that the forest-roaming Orang Asli have taken them from Th ailand. Nevertheless the above records records are the first from Malaysia.

Appendix 3. R. kerrii as a tourist attraction

MEIJER & ELLIOTI (1 991) proposed R. kerrii as a new Thai tourist attraction in order order to save it from extinction. Th ey support the idea with a number of convincing arguments. ‘Ecotourism' with rafflesia has met with some success in Suma 住'a (though there there are cases where colonies have been badly trampled by too many visitors). It seems feasible feasible that with appropriate financial investment and expertise a sensible ‘ecotouris t' de- velopment velopment project could be implemented with rafflesias at sites not too far off roads ， where the the environment is already degraded.

Nevertheless ，1 wonder about the wisdom of exposing such r訂 e and sensitive flowers flowers to the whimsical tourist business. We are only just starting to understand raffle- sia's sia's ecology ，still know nothing about its propagation ，加 d cannot yet cultivate it. Nature education education deserves all --o ut support but why put endangered species at risk? Th ere is so much else ，less vulnerable ， to show and educate abou t. Whatlfe 訂 most is 出at ‘ecotouris t' promotion of rafflesias in the m 訂 g加 al habitats will will lead to an increased threat to 出 e populations deep in the forest ， the most pristine and only only viable ones in the long term. 1 doubt these can - or actually should - be saved by ecotourism'. ‘ecotourism'. It would be an ordeal for the average tourist to be brought to a population 52 HANs BANZIGER

like like the one 1 studied ， and hence it is likely that it is the rafflesia which will be brought to to the touris t. Th ailand is different from Sumatra. Tourism is f:紅 more advanced and is spilling into into every comer of the kingdom. Th e communication network is capillary and for the mostp 訂 tgood to excellent. 百lai people 釘 ef ，訂 more enterprising (a veηpositive trait on most counts) but 出is is coupled with an unusu aI ly liber aI laissez-faire attitude towards what what is permissible and lax law enforcemen t. On top of this comes 白 e devastating success with with which the Tourist Authority of Th ailand has been able to promote m 出 s to 町 ism. Rafflesias Rafflesias aIr eady have a market value due to 出eir purported medicinal properties and they do not need another from tourism. Th us the best populations could be in serious danger of of being wiped out by local collectors of forest produce ，many of whom understandably need need additional income. As mentioned by MEUER & ELLIOTT (1 99 1) there is already a small small souvenir business with rafflesias. Collectors rarely are aware of the dependency of the the parasite on the liana and will cut its stem to make passage easier and sever roots (Fig. 20) 20) when taking the buds or flowers. (Shortly (Shortly after 1 had written the above lines ， marauders wiped out the only popu- lation lation 1 knew of 血e rare slipper orchid Paphiopedilum parishii (Reichb. f.) Stein - object of of a long term study which 1 had been carrying out for a year already. The site was in a wildlife wildlife sanctu ぽ y and much less accessible than the R. kerrii population.) In In the case of Th ailand 1 would like to propose a moratorium until scientists have mastered mastered a technique to cultivate rafflesias. Then perhaps 出ey can be grown in the land- scape scape g釘 dens of luxury hotels and resorts-under appropriate glass cover lest the stench and carrion flies scare their customers away.