Bntanical Journal qf the Linnean Socie{y (1990), 103: 301-308. With 2 figures Studies in Garcinia, dioecious tropical forest trees: the origin of the :rnangosteen (G. m.angostana L.) A. J. RICHARDS Department of Biology, Ridley Building, University of Newcastle upon Tyne NEJ 7RU Received Octnber 1988, reviled and accepted for publication ]anumy 1989 RICHARDS, A. J., 1990. Studies in Garcinia, dioecious tropical fruit trees: the origin of the mangosteen (G. mangostana L.). Mangos teen does not grow in the wild state. Plants are said to he invariable, and almost all are female. Mangosteen (2n =?88-90), an obligate agamosperm, has only two close relatives, G. hombroniana (2n = 48) and G. malaccensis (2n =?42) which are facultative agamosperms. For the 13 main characters by which they differ, mangosteen takes an intermediate morphological position for five characters, and resembles each of the other species for four of the remainder. It is suggested that mangosteen is an allopolypoid derivative of these species which arose as a female from a single hybridization event in cultivation, and which has since reproduced asexually. To overcome difficulties in propagation and establishment that mangosteen presents, attempts should be made to graft and hybridize mangosteen with its presumptive parents. ADDITIONAL KEY WORDS: Agamospermy - allopolyploidy - Clusiacaeae - Garcinia hombroniana - Garcinia malaccensis. CONTENTS Introduction Morphology Cytology. Discussion. Acknowledgements References. 301 304 306 306 307 308 INTRODUCTION Mangosteen, or manggis, Garcinia mangostana L. (Clusiaceae) is generally considered to be one of the most delicious of all fruits. Despite attempts in Africa and the neotropics, it is only grown successfully in the non-seasonal wet tropical regions of south-east Asia, notably Sri Lanka, Burma, Thailand, peninsular Malaya, Borneo, Java and the Philippines: and recently in northern Australia. The area of origin of mangosteen has been the subject of some debate. King ( 1890) claims that it is native to peninsular Malaya, but Ridley ( 1922) states that is is not known in the wild state. Maheshwari ( 1964) gives it as a "native of the Molucca Islands from where it has been transplanted to Java and Malacca" (Malacca is in peninsular Malaya), but then confusingly states that "the exact 0024-4074/90/08030 I+ 08 $03.00/0 301 ~) 1990 The Linnean Society of London
302 A. J. RICHARDS origin of mangos teen... is believed to be Malay Peninsula or Malaysia". Bur kill ( 1966) says that "it is not found wild in Malaya, and there were no trees on Penang Island until after its colonization, the first trees fruiting about the year 1802". Whitmore ( 1973) states "the true wild original home... is not known. Corner (I 952) thought he had found it wild in S. Trengganu (peninsular Malaya), in the Ulu Kemaman, but confused his trees with G. malaccensis". Corner had claimed this species as native to Malaya. The current view seems to be that although occasional plants of mangosteen can be found in the forests of peninsular Malaysia, they are associated with former cultivation or access points, and there is no clear indication that it is a native wild plant here, or elsewhere (Soepadmo, personal communication). Support for this view seems to come from the breeding system of mangosteen. Although at least ten species of Malaysian Garcinia are thought to be facultative agamosperms (Richards, 1990a), most are also capable of sexual reproduction. In this dioecious genus, obligate agamospermy will yield only females; males must result from sexual reproduction. Males are known in all species except possibly G. scortechinii King, and mangosteen. If males are absent, it must be assumed that obligate agamospermy pertains. Lim ( 1984) and Richards ( 1990a) show that mangosteen is indeed an obligate agamosperm, reproducing by the adventitious budding of proembryos from ovular tissues. There have been widely divergent views as to the occurrence of males, or of male function, in mangos teen. There are a number of detailed descriptions of the male flower, but most seem to rely on the description in Roxburgh ( 1832). As no male plant seems to have been authentically reported between 1832 and 1987, it seems likely that Roxburgh may have been describing the male flowers of a closely related species, perhaps G. malaccensis Hook. fil. which was unknown before that early date. Ridley ( 1922) states "I have never seen a male tree, male flowers or a fertile stamen". Maheshwari ( 1964) states that the occurrence of male flowers is doubtful, and quotes Pierre ( 1882) who examined more than 1500 trees without finding a single male flower. Corner ( 1952) and Burkill ( 1966) are amongst experienced authors who seem never to have seen a male mangosteen, while Whitmore (1973) states "males very rare". However, Burkill ( 1966) quotes Chevalier ( 1919) as follows: "female plants bear flowers which produce male organs and one can suppress all the male plants [sic] without untoward results: as the natives of lower Cochinchina do when at the tree's first flowering they cut out those that do not fruit. Thus the natives select good productive hermaphrodite trees". This is a confusing report for, with the exception of Pierre ( 1882) who reports sequential hermaphrodity in Garcinia although not in mangosteen, no-one else has ever suggested that mangosteen or any other Garcinia is hermaphrodite or monoecious. Unlike many female Garcinia, the flowers of mangos teen bear well-formed staminodes (Fig. 1), and these may have led to Chevalier's report. Thus, it was all the more remarkable when Idris and Rukayah ( 1987) reported the occurrence of a single flowering male mangos teen in a row of village mangosteens beside a road in Negeri Sembilan, peninsular Malayasia. This individual is said to be about 70 years old. It must have arisen sexually, and the only question surrounding this careful and beautifully illustrated account concerns the identity of the tree. Although it resembles female mangosteens in
STUDIES IN GARCINIA 303 Figure I. Flower of the rnangostccn (Garcinia mangos/ana) showing the sessile lobed stigma, large red petals and starninodes. Scale bar = 50 rnrn. most respects, the petals are yellow, suffused reddish distally. The flowers of female mangosteens have entirely red petals. Thus, the possibility exists that this plant is a hybrid between female mangosteen, which has an apparently functional embryo-sac (Lim, 1984; Richards, 1990a), and either G. malaccensis or G. hombroniana Pierre. These species are closely related to mangosteen (e.g. Whitmore, 1973): the former has male flowers with pink or yellow flushed pink petals (notes in Hb FRI M); in the latter species the petals are cream-coloured. Certainly, mangosteen is at the least almost always reproductively asexual, and it is not surprising that a number of authors (Horn, 1940; Corner, 1952; Burkill, 1966; Whitmore, 1973) have commented that it is morphologically invariable. However, Bur kill ( 1966) mentions a report that a race of mangos teen occurs in the Sulu Islands with a thicker rind and a more acid flesh, and another of plants of a pyramidal shape. Recent unpublished work by MARDI (M. Khalid Mohamed, personal communication) has detected regional variations in time to fruiting, season of fruiting, growth rate, and thickness of fruit rind in mangosteen from peninsular Malaya, and experiments are currently investigating whether these variations persist in grafted plants growing in the same conditions. The demand of mangosteen fruits usually exceeds the supply. Mangosteen trees are rarely planted in commercial quantities, and most orchard and kampong trees are of a considerable size and age. Seeds arc difficult to germinate (Chin, 1978). and seedlings are very slow growing in early years (Goh, Rao &
304 A. J. RICHARDS Loh, 1988), two-year-old seedlings not exceeding 15 em in height. Roots are particularly slow to establish, perhaps because they lack root hairs (M. Khalid Mohamed, personal communication). Young plants need careful watering and shading from the sun even in optimal conditions, and will only establish when meticulously weeded. Rapid growth usually starts between 5 and 7 years from germination, but fruiting rarely occurs before 12 years (Lim, 1984) and regular heavy fruiting only occurs after 18 to 20 years. As a result, mangosteen is rarely seen as a commercial proposition. Mangosteens do not often self-sow, and it appears that in the absence of careful cultivation, mangosteens would soon become extinct. In contrast, the close relative C. hombroniana shows 100% seed germination within 1 month (personal observation, N = 53), grows rapidly, and fruits heavily at between five and eight years from germination (Richards, l990a). In each of one montane and one maritime localities studied, numerous self-sown seedlings were evident, often to the extent of forming thickets around the parents. Mangosteen does not seem to occur as a wild plant, is asexual, producing only females, is apparently very invariable, and is difficult and slow to grow when young. These features all suggest that mangosteen arose in cultivation, probably as a hybrid between two related species. It seems likely that all mangosteens are the asexual progeny of a single hybrid female, and essentially represent a single seed-reproducing clone. Such genetic variation as does exist is likely to have arisen from somatic mutation, or possibly from occasional backcrossing to hybrids with males from related species. In view of the difficulty experienced in the propagation and cultivation of mangosteen, it is important that the wild relatives of mangos teen are identified. Significant advances in the cultivation of mangosteen might be obtained by grafting or hybridization with wild relatives. This has already been suggested by Maheshwari (1964) and Whitmore (1973). MORPHOLOGY If, as seems likely, mangosteen arose in peninsular Malaya, the presumptive parents should be sought there. Whitmore ( 1973) lists 39 species of Carcinia from this region, with another ten unnamed. He follows Ridley ( 1922), Corner ( 1952), and Bur kill ( 1966) in suggesting that G. hombroniana and C. malaccensis form with C. mangostana a natural morphological grouping with no other close relatives. Whitmore states "these species form a very close-knit group, virtually indistinguishable when sterile... the exact taxonomic status of members of this group will need critical reappraisal... ". Amongst many characters which distinguish this group are large leathery leaves which are green when dry and which have almost invisible lateral venation; large flowers which are usually borne singly on the female and bear a large sessile or subsessile discoid stigma (Fig. l); male flowers with four lobed but contiguous stamen masses, and wellmarked pistillodes; and four fleshy petals. The fruit is large, globose or subpyriform, fleshy within, and red or purple when ripe. Within this group, the three species are differentiated by 13 characters (Table 1). For four characters, mangosteen resembles C. malaccensis (latex colour, petal colour, the sessile stigma and the fruit colour). For four, it resembles
STUDIES IN GARCI.NIA 305 Figure 2. Flower (right) and young fruit of Garcinia hombroniana showing the weakly lobed, stipitatc stigma and cream coloured petals. Scale bar = 20 mm. G. hombroniana (Fig. 2) (smooth stigma surface, stamen/staminode mass clearly lobed, fruit globose, surface smooth), and for five it falls between the other two species (flowering time, depth of stigma lobes, stigma diameter, presence of staminodes in the female flower, and fruit taste). There is no character for which mangosteen falls outside the range of variability of the other two species. Morphologically, mangosteen is a very good candidate for a presumptive hybrid between G.lwmbroniana and G. malaccensis. TABLE I. Morphological cha1 acters of C. hombroniana, G. malaccensis and C. mangos/ana (from Ridley, 1922; Corner, 1952; Maheshwari, 1964; Whitmore, 1973 and personal observations) G. hombroniana G. malaccensis G. mangos/ana Character (Fig. 2) (Fig. l) Flowering Jan-Mar (-June) April-July March-April, July-Sept. Latex White Yellow Yellow Petal colour Cream White to pink Red Stigma Stipitate Sessile Sessile Stigma surface Smooth Corrugated Smooth Stigma lobes 20'/' 0 diameter 80';{, diameter 50% diameter Stigma diameter 5-6 mm l0-20 mm 8-12 mm Stamen mass Lobed Nearly unlobed Staminode mass lobed Female flower No staminodes Staminodcs Small staminodcs Fruit shape Globose Depressed globose Globose Fruit surface Smooth Wrinkled Smooth Fruit colour Red Maroon Purple Fruit flavour Astringent Sweet, insipid Swectjsour
306 A. J. RICHARDS CYTOLOGY Garcinia have very small chromosomes. Most cytological studies have been based on male meiosis. Chromosomes stain poorly, but metaphase plates spread well. The cytology of the genus is further discussed by Ha ( 1978) and Richards (1990a, b). For the male of the maritime race of C. hombroniana growing at Rimba Ilmu Botanic Garden, Kuala Lumpur, I found n = 24 (2n = 48). This seems to be the only recorded count for this species. There appears to be only a single "unconfirmed" count of 2n = 42~43 for C. malaccensis (Ha, 1978). This seems to have been taken from a seedling root-tip mitosis. The cytology of mangosteen has a chequered history. There being no male meiosis, counts have been taken from seedling root-tips, and this, coupled with the very small, numerous chromosomes has probably led to inaccurate reports. The first report by Krishnaswamy & Raman ( 1949) is 2n = c. 76, while Tixicr (1960) reports 2n = 96. Ha (1978: 172) mentions a count of2n = 110~120, but there is no authority quoted, and this count is not listed on p. 129. Ha ( 1978) himself reports 2n = 88~90, which is stated to be "unconfirmed". The only firm conclusion that can be reached with respect to the cytology of G. mangostana is that is is polyploid, and probably tetraploid, with respect to its only close relatives, C. hombroniana and C. malaccensis. However, it is worth noting that if Ha's counts for C. malaccensis and C. mangos lana are given greater credence than he is willing to accept himself, it would appear that C. man,t;ostana (2n =?90) may be the allotetraploid derivative of C. malaccensis (2n =?42) and G. hombroniana (2n = 48). DISCUSSION The only close relatives of mangosteen are wild species of peninsular Malaya. Garcinia hombroniana is found in both montane and maritime forest, and is often associated with secondary forest near kampongs. It is also apparently native to the Nicobar Islands and south Thailand (Tenarissim), and is occasionally cultivated elsewhere, for instance southern India and Cochinchina. In peninsular Malaya it is cultivated in kampongs for its wood (which is used for fencing and house-posts), for its astringent fruits which are dried and used as flavourings, and for the medicinal properties of the fruit rind, leaves and roots. Garcinia malaccensis is confined to the lowland forests of peninsular Malaya, where it sems to be scarce and scattered in its distribution. Formerly, it seems to have been confused with mangosteen, and before Whitmore ( 1973) it was only known from the type material from Malacca, described in 1866. It is now known to occur in six states, and it has been studied in the field by Ha ( 1978). Although the fruits are stated to be delicious, it seems never to have been cultivated, and no cultivated tree seems to be known today. Both C. hombroniana and C. malaccensis are facultative agamosperms; for G. hombroniana Richards ( l990a) demonstrates adventitious embryony similar to that found in mangosteen. Males occur in both C. hombroniana and C. malaccensis, and Richards ( l990a, b) shows that the former species readily undertakes sexual reproduction in the presence of males, sexual embryos inhibiting the later formation of proembryos.
STUDIES IN GARCINIA 307 Garcinia malaccensis must also be capable of sexual reproduction. Garcinia mangostana also develops embryo-sacs, and these are apparently capable of some parthenogenetic (Lim, 1984) and sexual hybrid (Richards, l990a) embryogenesis, although there is as yet no evidence that embryos ever mature from an egg in this species. Garcinia mangostana is a polyploid which is morphologically intermediate between the diploids G.lwmbroniana and G. malaccensis. The limited cytological evidence suggests that mangosteen may well be the allotetraploid derivative of these two wild species. The mother of the original female mangosteen may have been a cultivated G. hombroniana and the father a wild G. malaccensis occurring nearby. The reproductively isolated original hybrid would only have been able to reproduce apomictically to give female offspring, having inherited the capability for adventitious embryony from one or both parents. Although it is posible that the cross occurred on more than one occasion, present day mangosteens may all be offspring of the one asexual line. As both G. hombroniana and G. malaccensis are native to peninsular Malaya, and G. malaccensis appears to grow nowhere else, it must be assumed that the mangosteen first arose in that area. After the qualities of the new hybrid were first appreciated, it would soon have been transported as fruits to other areas of south-east Asia where it is cultivated today. Mangosteen has a very considerable economic potential both for local consumption and export. However, unless the cultivation of seedlings becomes less labour intensive, and until early fruiting characteristics are developed, farmers will rarely consider the development of large mangosteen orchards an economically viable proposition. Only a few as yet unsuccessful attempts have been made to hybridize mangosteen with G. hombroniana. The cross with G. malaccensis has yet to be attempted, although it is possible that the 'male mangosteen' (Idris & Rukayah, 1987) is this hybrid. As mangosteen has apparently remained substantially unhybridized, it may be that such crosses will prove difficult to achieve, despite the presence of embryo-sacs in mangosteen. Hybrids will presumably be triploid, and may prove sterile. Garcinia hombroniana is considerably easier to germinate and cultivate than mangosteen, and flowers in five to seven years from seed. Although unsuccessful attempts to graft mangosteen onto Garcinia hombroniana seedlings have been reported in the past (Bunting & Milsum ex Bur kill, 1966), techniques for selfgrafts of mangosteen have improved markedly in recent years, and this procedure is promising. No attempts have yet been made to graft mangosteen onto G. malaccensis, and this also deserves attention. ACKNOWLEDGEMENTS I should like to thank the Socioeconomic Unit of the Prime Minister's Department of the Government of Malaysia for permission to conduct scientific research in Malaysia, and the U niversiti Malaya for granting temporary research associate status. Professor Nawami of the Jabatan Botani kindly provided research facilities, and I am grateful to several members of that Department for help and collaboration, notably Dr Halijah Ibrahim, and Mustapha Mohamed of the Rimba Ilmu Botanic Garden. I am also very
308 A. j. RICHARDS grateful to members of MARDI for their helpful collaboration, notably Dr Mohamed Khalid Mohamed Zin. Research funding was provided by the Royal Society of London, the British Council, and the University of Newcastle upon Tyne. REFERENCES BURKILL, I. H., 1966. A dictionary of the economic products of the Malay Peninsula, I. Second edition. Kuala Lumpur: Ministry of Agriculture and Cooperatives. CHIN, H. F., 1978. Production and storage of recalcitrant seeds in the tropics. Acta Horticulturae, 83: 17-21. CORNER, E. J. H., 1952. Wayside Trees of Malaya, I. Second edition. Singapore: Government Printing Office. GOH, H. K. L., RAO, A. N. & LOH, C. S., 1988. In vitro plantlet formation in mangosteen (Garcinia mangos/ana L.). Annals of Botany, 62: 8 7-93. HA, C. 0., 1978. Embryological and cytological aspects of the reproductive biology of some understorey rainforest trees. Unpublished Ph.D. thesis of the University of Malaya. HORN, C. L., 1940. Existence of only one varit"ty of cultivated mangosteen explained by asexually formed 'seed'. Science, 92: 237-238. IDRIS, S. & RUKAYAH, A., 1987. Description of the male mangosteen (Garcinia mangos/anal.) discovered in peninsular Malaysia. MARDI Research Bulletin, I5: 63-66. KING, G., 1890. Materials for a flora of the Malayan Peninsula. Journal of the Asiatic Society of Bengal, 59:!56. KRISHNASWAMY, N. & RAMAN, V. S.,!949. A note on the chromosome numbers of some economic plants of India. Current Science, 18 ( 10): 376-378. LIM, A. L., 1984. The embryology of Garcinia mangostana L. (Clusicaeae). Gardens' Bulletin, Singapore, 37: 93-103. MAHESHWARI, J. K., 1964. Taxonomic studies on Indian Guttiferae. III. The genus Garcinia L. s.l. Bulletin of the Botanical Survey of India, 6:!07-135. PIERRE, L., 1882. Hare forestiere de Ia Cochinchine, 4: 54-64. RICHARDS, A. J.,!990a. Studies in Garcinia, dioecious tropical forest trees: agamospermy. Botanical Journal of the Linnean Society, I03: 233-250. RICHARDS, A. j., 1990b. Studies in Garcinia, dioecious tropical forest trees: the phenology, pollination biology and fertilization of G. hombroniana Pierre. Botanical Journal of the Unnean Society, 103: 251-261 RIDLEY, H. N., 1922. Flora of the Malay Peninsula, I. London: Reeve. ROXBURGH, W., 1832. Flora Indica, 2: 618-633. Serampore. TIXIER, P., 1960. Donnees cytologiques sur quelques Guttiferales recoltees au Laos. Revue cytologique et de Biologique vigitale, 22: 65-70. WHITMORE, T. C., 1973. Garcinia L. In T. C. Whitmore (Ed.), Tree Flora of Malaya, 2: 196-225. Kuala Lumpur: Forest Departm<"nt, Ministry of Primary Industries, Malaysia.