J

Rotanical Journal of the Unnean Soriel)' (1992), 110: 61 - 75, Wilh 6 fi.l(ures

Food bodies and their significance for obligate ant-association in the tree genus Macaranga (Euphorbiaceae)

BRIGITTE FIALA

<oologisches Institut III der Universitiit Wiirzburg, Biozentrum, Am Hubland, D-8700 Wiirzburg, Federal Republic of Germarry

AND

ULRICH MASCHWITZ

Zoologisches Institute, J. W,-Goethe-Universitiit, Siesmayerstr. 70, D-6000 Franlifurt, Federal Republic of Germarry

Received January ' 1991, accepted for publication M arch 1991

FIALA, B. & MASCHWITZ, U., 1992. Food bodies and tbeir significance for obligate ant-association in the tree genus Macaranga (Euphorbiaceae). The production ()r{~xtrafloral nectar and food bodies plays an important role in many tropical ant-plant mutualisms. In Malaysia, a dose association exists between ants and some species of the pioneer tree genus Alacaranga (Euphorbiaccac). Macaranga is a vcry diverse genus which exhibits all stages ofintcraction with ants, from facultative to obligatory associations. The ants nest inside the hollow inlcrnodes and reed mainly on food budies provided by the plants. Food body production had previously been reported only in myrrnecophytic ,\1acaraflga species, where it is usually coneentrated on protected parts or the plants such as re('unTd stipules. \Ve fi)und that tloll-myrm('cophytic Macnrall,f!,a SPl'<"i('s also prodlH,(, food horiics on lean's and stems, where they are collccted by a variety or ants. Levels off(JfJd body prurluniulI difll-r betwt'en facultativdy and obligatorily ant-associated species but also among the various non-myrmecophyt("s. This may he rdated to the degree of interaction with ants. Food body production starts at a younger age in the myrmccophytic species than in the transitional or non-myrmcccophytic ,\1acaranga. Although food bodies of the non-inhabited Macaranga species arc collected by a variety of ants, there is nu evidence of association with specific ant species. Our obsen'ations suggest that food bodies enhanct' tht' evolution of ant-plant interactions. PrOdw.:lion of food bodies alone, howevt'r, does Ilot appear to be the most important factor li)r lhl' development of obligate myrmccopllytism ill .. Wacaran/!,a.

ADDITIONAL KI::Y \\'ORDS: - Am-plant interactions - evolution - food bodies - Macaranga -­Malaysia - myrmrcophytism.

CONTENTS Introduction Material and methods Resuils

i\-Iyrmccophytir . .\Iacaranga species Non-myrmccophytic l\1aCQranga species.

0024- 4074/92/090061 + I j S08.00 /0 61

62 63 63 63 68

© 1992 Tht' Linn('an Society of London

62 B. F1ALA AND U. ~IASC:H\\,ITZ

Dist'w;sioll . 71 COlltelllS (If ((~()d hodies 7 1 Production ralt's Localizatiull FUllction .

Fuod bodies and the n 'olutiull of myrlllc('ophytisrn . Acknowledgcments Rdl'n'llCcs

INTRODUCTION

71 72 72 73 74 74

Food bodies belong to a large category of structures, collectively called pearl bodies, and include a variety of small epidermal and subepidermal structures with diverse shapes. They range from single cells to multicellular bodies (O'Dowd, 1982) and are characterized by a pearl-like lustre . Some authors restrict the term "food bodies" to highly specialized ant-plant associations where they play an essential role as ant food (review e.g. in Buckley, 1982; Beattie, 1985). Keeler (1989) does not differentiate between the terms food bodies and pearl bodies. Earlier, Rickson & Risch (1984) considered the two terms to be different, but in recent transmission electron microscope studies found them to be similar structurally (Rickson, personal communication). We regard the terms as synonymous.

Pearl bodies occur over a wide range of taxa in the Dicotyledonae and have been described from leaves, shoots and stems of 50 genera in 19 families (O'Dowd, 1982), all tropical or subtropical. As with extraOoral nectaries (EFN ), they have long been hypothesized tu attract ants and function as ant food (Penzig, 1982), although collection of pearl bodies by insects has so far been directly observed only in relatively few plants species (e.g . Raciborski, 1898 1900). In highly specialized ant-plant associations, however, food bodies are an essential part of the ants' diet.

Ultrastructure and development of some myrmecophyte food bodies have been described. They contain carbohydrates, lipids and proteins (O 'Dowd, 1982), and can thus offer a nutrient supply complementary to EFN-secretions or honeydew.

In S.E. Asia, several species of the tree genus Macaranga Thou. (Euphorbiaceae) live in close relationship with ants. The genus comprises a full range of species from those not inhabited by ants to obligate myrmecophytes. Non-myrmecophytes are, however, visited by various ant species. Five clear species groups (sections) can be recognized, some further species are individually distinctive (Whitmore, 1973, 1975, 1982). In Peninsular Malaysia, nine of 27 Macaranga species occurring mostly in pioneer habitats are obligate myrmecophytes usually inhabited by Cremalogaster ants, mainly C. borneensis, which is dependent on its host plant for survival (Fiala, 1988; Fiala & Maschwitz, 1990) . The ants protect their host plants against herbivores and competing vegetation, especially vines (Fiala et al., 1989, 1991 ). The ants nest in the hollow stems of the j\1acaranga trees and feed exclusively on food bodies produced by the plants (Beccarian bodies, Rickson , 1980) and honeydew fi'om scale insects cultivated inside the stems (Fiala & Maschwitz, 1990).

The ofler of nesting space is onc important predisposition for interactions with ants in the genus Macaranga (Fiala & Maschwitz , 1992), supply of food another possible ant-attractant. However, the presence of EFN does not seem to be an

I J

FOOD BODIES AND ANT ASSOCIATION 63

csS<'iltial If-aturc in the evolution of obligate myrmerophytism in that genus (Fiala & Maschwitz, 1991). We, therefore, investigated the production of food bodies as another possible predisposition in the evolution of such symbioses. Here we present data on the distribution of food bodies in the genus Macaranga in Peninsular Malaysia.

MATERIAL AND METHODS

The study was carried out from 1985 to 1990, covering a total of 16 months in various parts of the Malay Peninsula. More than 1000 plants of 23 Macaranga species were studied, mainly in secondary habitats. All plants were sur~eyed along transects about 2 km long (20~50 specimens per species from different loca tions).

In the field we investigated the plant surfaces for the presence of pear! bodies. Whenever possible, we collected material for examination with a dissecting microscope. For comparison we also looked for food bodies in the closely related genus Maltotus Lour. We recorded the presence of ants and gathered information on the degree of association with the plants.

Five myrmecophytic and three non-myrmecophytic Macaranga species were grown in the greenhouse at Frankfurt. For scanning electron microscopy (SEM) wc used fresh leaf material on a Hitachi Type S 500. Preliminary tests on fooG body contents were carried out with Merckotests for lipids (available from MERCK) and the anthrone method of Mokrasch (1954) for carbohydrates. Identification of Macaranga and specific classification followed Whitmore (1967, 1973, 1975). Voucher specimens are held by B. F. and the herbarium of the Forest Research Institute Malaysia, Kepong.

RESULTS

Myrmecophytic Macaranga species

All myrmecophytic Macaranga species develop cpidermal structures of two types: large Beccarian bodies, concentrated in different locations which are specics-spccific, and smaller food bodies scattered over leaf and stem surface. Almost all myrmecophytic members of the section Packystemon sensu stricto (Table 1) produce the Beccarian bodies on the abaxial surface of recurved stipules: M. constricta Whitmore & Airy Shaw, M.ll1Iltettii King ex Hook. li., M . kingii Hook. fi ., M. motleyana subsp. griffithiana (Muell. Arg. ) Whitmore, and M. triloha (Bt.) Muell. Arg. (Fig. I ). An exception is M. kYpoleuca Mucll . Arg., where many Beccarian bodies develop on the abaxial surface of the leaves (Fig. 2). The two West-Malaysian species of the section pruinosa, M. pruino.w (Miq.) Muell. Arg. and M. hosei King ex Hook. f. as well as one species. from Pacll)'stemon Sensu lato, M. caladiiJolia Becc. ( = M. pllllcticulata Gage, Whitlllore, 1975) produce them on the adaxial surface of horizontal or upward turncd stipules (Fig. 3). Many food bodies were also produced on the floral bracts of female inflorescences in M. triloha (Fig. 4) and M . hlllleltii, but not in M. he.vnei I. M. Johnston or M. tanarius (L.) Muell. Arg. (no data are available for other species because these were not flowering during the study period).

So far we have found four different forms of Crematogaster associated with these myrmecophytic Macaranga species, all belonging to the subgenus Decacrema.

64 B. F1ALA AND U. MASCHWITZ

TABLE I. Presence of food bodies in Macaranga species studied in Peninsular Malaysia; relative numbers of food bodies: + (few, < 50 ) to + + + (many, > 500), (+ ) = only rarely

Species

MYRMECOPHYTES Section Pachystemoll sensu stricto AI. constricla M. hulletii M. hypoleuca

M.kingii M. motleyana ssp. grifJithiana M. triloba

'TRANSITIONALS' Section Pachystemon sensu lato M. caladiifolia ( = M. puncticulata)

pruinosa-grou p M . hosei M. pruinosa

NON-MYRMECOPHYTES Section Pachystemon sensu stricto M. quadricornis M. recurvata

Section Pachystemon sensu lato M. curtrsii

M. curtisii var. glabra

Section Stachyella AI. amiHQ M. conifera

Section Pseudo-Rottlera M.lowii

Section Javanime M. heynei

Individually distinctive ,H. dentiw/ala M. diepenhoTStii .11. gigantea .\!. indico AI. (a1lariU5

M. trichocarpa

Abundance of food bodies

Per stipule

++ +++

( + ) Stipules not persistent

++ +++ +++

+

++ +

+

( + ) Stipules not persistent

( + )

( + ) ( + )

Stipules not persistent

Stipules hair-like

Stipules not persistent

Stipules not persistent

+ ( + )

Stipules not persistent ( + )

Per leaf surface

+ +

+++ Lower surface

+ + +

+

+ +

+ +

+

+

( + ) +

+

+

+ + + +

++ +

Species from the C. borneensis group are dominant but these ants are so variable that probably a few more species (subspecies? ) may be involved in what is so far called C. borneensis U. Longino, personal communication). The species taxonomy remains preliminary since Crematogaster is a very large genus which needs taxonomic revision. We have evidence so far for some Crematogaster being restricted to certain Macaranga species, for instance we find a preference for M. hypoleuca in C. decamera. Since C. borneensis was the most abundant ant in the myrmecophytic specimens studied, we will restrict our discussion to this species although we did not recognize behavioural differences in the other as yet undescribed Crematogaster spp. on Macaranga.

FOOD BODIES AND ANT ASSOCIAT ION 65

Figure I. Ants collecting fiJOd bodies produced under recurved stipules of M. lriloba ( x 7)

Production of food bodies on myrmecophytic Macaranga species was easy to demonstrate. Food bodies were always present under stipules, especiall y ill earlier stages of development since C. borneensis ants collect mainly the larger, fully developed ones. Macaranga of the pruinosa group, however, seem to produce iCwer food bodies that those in Pachyslemon sensu stricto (cL ,\1. llOsei , Tables I and 2). In the field only few ( < 30) food bodies were found on the stipules of' M. pTllinosa. However, even fewer were seen on those of AI. ca/ndiifolin (sen iott Pncliystemon sensu lato ). We regard a ll three species as transitional between nOll­inhabited forms and obligate myrmecophytes since the li-cquency of attt colonization of them is lower than in the latter.

Food body production begins at an early stage of plant development, e.g. itt saplings of M. trilobn (N = 51 ) , M. hllllettii (N = ::ll ) and M. Izypolellm (N = 2R ) from a height of only 9 cm (percentage of plant size wit h food bodies: 8 cm and 9cm: 50% (N= 20), 9- 13cm: 75 % (N= 40) , > 13cm: 100% (N= 50). The earliest food bodies were present at the third set of stipu les (between tile third and fourth internode), a lthough on average, they were not produced until the development of the sixth internode (average height about 13 cm ). This is

66 B. F1ALA AND U. 1\1ASCHWlTZ

Figure 2. SEM or lower lear surrace or M. hypol,uca. Food bodies on lamina (arrow ) and glands at margin or a young lear (scale bar = 500 !Im ).

approximately the same size at which plants start to be colonized by C. borneensis, although rarely queens colonized plants which had not yet produced food bodies (four times in our sample). Among the investigated myrmecophytic specics wc found no differences in plant size at which food body production started. In the

TABLE 2. Comparison of number of food bodies produced on diHcrent Macaran,~a species in the greenhouse (range of, \' = 5 plants each)

Species Cpper leaf surface Lower Iraf surfac(, Slipule

NON· \IYRI\IECOPHYTES .\1. tanarius 70- 200 200- 400 < 10, nut persistent AI. cnnifera 60- 130 30- 100 < 10, not persistent M. g(~an/m 30- 80 50- 200 < 10

"rRANSI 'fIO NA LS' M. hosei 20- 40 20- 50 40- 60 .If. caladiifolia < 30 < 30 < 30 ( = M. pl/nr/icula/a )

~IYRMECOPHYTES

!It . hy/Jolfura 30- 70 300- 800 < 10, not persistent M .hul/fuii 30- 50 50- 65 50- 230 .11. /riloba 40- 60 40- 60 50- 200

FOOD BODIES AND ANT ASSOCIAT ION

Figure 3. A, B, Food bodies on stipules of M. hose; ( x 4); B, with C. horne",s;s ants collcning them ( x 6).

67

transitional species, however, (ood bodies arc produced later, e.g. 011 stipules of M.llOsei on average at a height of about I m (N = 105), a lthough a kw ({HId

bodies are produced earlier, on leaf surfaces from the seventh leaf on. In thi s species, ant colonization as a rule does not take place on plants smaller than I m.

-. 68 B. FIALA AND L'. ~li\SCH\\' ITZ

Figure 4. Food bodies Oil tht hracts or female flowers or j\l. lriloha ( x 7)

I n the greenhouse , myrmccophytic species also produced small food bodies scattered over leaf surface and stem. In the field, this was seen only aftn excl usion of ants, e.g. by means of sticky rings (with tangle foot glue) around the stem.

The occurrence of food bodies is not restricted to intermittent plant growth seasons so ants are permanently supplied with food. Since Macaranga trees grow rapidly, many leaves and stipules, on which food bodies develop, arc produced continuously. In the field , unbranched M. IriLoha saplings produced on average three new leaves per month .

.l\/(}n-n~vnne(olJ/~vlic !vI araranga .r/Jfcie.r

Non-myrmecophytic species also produced food bodies in the greenhouse (Fig. 5). However, investigation offood bodies in the field was a problem in these

FOO D BODIES AN n ANT ASSO CIATION

Figufe 5. Food bodies on the lIpplT surf;l{'(' o r lean's or llo n-m yrmrco pil Yli (" ,\1 . {Of/a rim (o llc- third or na tural siz(' ).

spec ies because th e)' we re no t co nce lltra ted Oil pro tected pl a llt pa rts: IlIstead . food bodi es were scatte red over lea f surf'aces a long \'(' ins a nd on stems. These a rc no t easil y seen in th e field. Onl y a fte r \ '(' ry tho ro ug h im'estiga tio JlS \\'(' rc sO lll e I(Hlnd in pa rts o f the pl a nts less acccss ible to ant s, e .g. o n th e stl' lll undn \'( '1' \

na rrow sti p ul es w hi ch hindered access o f a ll hilt \T ry slIl a ll a nts. H(l\l'(' \ ·(· r. \I'h t' ll a nt access \I'as excluded (e.g . by m ea ns of stir b ' rin gs \I' ith ta ng lel;)ot ), I(H )d hod\ p rodu ction became o bvio us in the fi eld , with m 3 11 )' IClOd hodi es occ urrill g (lll Ih ,' pl a nt surf:\ce. t-.1 os t o f the no n-m yrm ecoph ytes d o no t possess conspi cuous a l)(l pnsistent stipules but drop them SOO Il a fi cr a leaf' h as full y g r()\\·11. TllOst' \I hi t'h pe rsi st (in M. giganlea (R chb. fil. & Z oll. ) t-.l. A .. M . dirlmlli orsli i ( t-. I iq , I l'.Ju ell. Arg. , M. recurvala G age ) a re usua ll y no t r('c urved as is th e case in tilt' Jll\'rm eco ph yti c speeies of A1acarmll!,a. The a ni), exce ptio ns a re ,If. quor/ricol'll i,\ Ridley (hut he re th e stipul es a rc wid e o pen ) a nd M . illdica \\' ight , in \\ 'hi ch th n a rc na rrow a nd twisted , a nd th e space be tweeli th e m a nd th e ste lll is d iflicult li,r a nt s to reaeh (Fig. 6 ) . In the fi e ld , IlOw('\'('r, we mos tl y fo und lood bocli es a lo n,!!; m a in leaf ve ins on the lower lea f surface (sec T a ble I ) , Altho ugh n{) lI -

70 B. FL\LA ANI> l· . \L\SCII\\"ITZ

Figure 6. Stipules of M. quadricornis (left ) and M. indica (two-thirds of natural size).

myrmecophytic Macaranga spp . developed food bodies, the amount produced differed between them (Table 2).

On M. tanarius saplings many food bodies are produced at the base of the stem. Numbers decrease higher up the stem but rise again at the level of the leaf insertions. Leaf petioles are a lso densely covered with food bodies.

To date, we have not thoroughly investigated food body production in the section Pseudo-Rottlera, which comprises species occurring in primary forest. ObservatioIlS on M. lowii indicate that food body production also occurs in this section. Pseudo-Rottlera is closely related to and morphologically intermediate between Macaranga and the genus Maltotus, which also produces food bodies (Rouppert, 1926), thus indicating a wider distribution of this attribute within the family Euphorbiaceae. On two Maltotus paniculatus culti vated in the greenhouse food bodies were prod uced on both leaf surfaces along the veins (upper leaf surface average of 50 food bodies, lower side 10). They were coll ected from the plant by Plagiolepis sp. , an ant living in the greenhouse. On Maltotus paniClllatlls and Maltotus macrostachys food body production was rarely conspicuous in the field.

In neither the field nor the laboratory did C. borneensis ants accept food bodies of Macaranga tanarius or other non-myrmecophytes offered to them on the leaves of their host plants: they preferred food bodies from myrmecophytic Macaranga species. Food bodies from the neotropical ant-plant Cecropia (Cecropiaceae) were likewise rejected (28: 2).

In the field, food bodies of non-myrmecophytic Macaranga are probably

FOOD BODIES ANI> ANT ASSOCIATION i I

collected by \'arious <liltS: ;\ total of 26 sp('ci es of 11 g('nera were found 011 the lean's of lIoll-nlyrlllf'cophytic J\1acar(lI1/!.a speci('s IFiala & l\Iasch\\'itz, 1991 1, Only 1\\0 of these were not intercsted in the fi)(ld bodies IPolyrachis sp, and Cladotnyrtna sp, ), Many or th ese ants wer(' also fC)lInd on other specics in the \'icinily, There was no e\'idcncc for any spcci('s-sp('cific relationship bet\\'('ell II<JII-nlyrlllecophytic Mawrallga and th(' visiting ants,

Preliminary tests on food body contents wcr(' carried out on AI, {(Il/arilll, At. triloba and ,\1, hOJe£ ( two plan ts each ) rcprescnting respCrI ively non­Illyrmecophyte, myrmecophyte and transitional species. Food bodies containcd about 6- 10° ;) lipids, 6- 12" ;) sugars and 4,0- 60"" water (percentage of\\'et tissue \\Tight ). The average perc(,lltage of carbohydrates appeared to be higher ill the lIoll-myrmceophytic species ( 10 12"0 versus, H",, ) . The amount of protein \'a ried fi'om 5 3% in A1.triloba and A1,llOsei to only about 1,5" " ill M.tal/(///II.\ (H. Crumky, personal (om munication .)

DISCUSSION

Contents o}Jood bodies

Food offered by plants to potential ant visitors is likely 10 he a predispositi()n filr the evolution of myrmecophytism . ExtraOoral nec tarics have beell cited as a feature of myrmecophytes Ua nzen , 1966; Wilson, 1971 ) . Production of (lad bodies in addition to nectar from EFN provides a more complete die t for ants than nectar alone. The larger amounts of lipids and /or proteins in food bodies can offer a dietary complement to nectar or honeydew obtained fi'o\11 homopterans (Beattie, 1985 ).

Food bodies of Macarallga contain large amounts of lipids (Rickson, 1980) and amino acids. Of the ten amino acids consid('!'cd to be essential fl)r insect s b\' Hagen et al. ( 1984) only two have not becn found so br in Mawl"{lll f,a food bodil'~

(Rickson, personal communication). In addition, food bodies prO\"idc carbohydrates and are therefore an energy-rich and easily accessible f(lc'lCl source, which, being solid, has one more advantage: contrary to the open\\­secreted nectar they are aseptically packed and therefore a longer-l ast ing fCJOd source.

Little is known about diflerences in nutrient content of food bodies £i'01ll

myrmecophytic and non-myrmecophytic species. Comparison of our data and those of Rickson's analysis for M. triloba (personal communication ), sho\\Td rather high variation, probably due to differing methods, However, the resu lt s d('monstrated that all important nutrients occ ur. The preliminary resu ll of' a higher protein and lower carbohydrate content in a myrmerophytic spccic's compared to the non-myrmecophytic M, tal/a rillS is especially int eres ting. Fond bodies of M. (ana rills were collected by various ant species, but not all anls \'isiting fl,f. tanarius leaves accepted them . Da\idson et al. ( 1991 ) also rqJPrt rejection of food bodies by generalized foragers on Cecropia plants, so there nla \ ex ist a dietary adaptation of ants to plant food.

Productiol/ rates

Wc found that the non-myrmecophytic Mawralll'.a specics studied also produced food bodies. Contrary to the observations of Baker ( 1934) Ong ( 197B)

72 B. F1Al.A AND U. ~IASCH\\TrZ

and Rickson ( 1980), we have noted scattered food hodies o("nIITill~ Oil 1l01l­

myrmecophytic I\[acaranga spccics. Rccordin~ food body production ill the field is difficult because they are continuously harvestcd by a wide range of ant species. It is likely, thnel(lre, that substantial I(lod hody productioll lIlay occur ill all non-myrmecophytic j\1acaranga species.

The numbers of food bodies produced differed between 1I0n-myrmeccophytic and myrmecophytic species and may indicate degrees of specialization . Further differences in production rates were found even among myrmecophytic species, with species of the /nuinosa-group producing fewer food bodies than those of section Paclzystemon sensu stricto. Plants with enhanced food body production may support larger ant colonies which means a better protection for the host plants. This may be one reason why on equally tall plants of M. l1iloba and M. hosei relatively smaller colonies are found on the latter (personal observation).

Localization

In some myrmecophytic Macaranga species food body production takes place in the protected space under recurved stipules, in others ('transitionals': M. hosei, M. pruinosa) food bodies develop openly on the stipule surface. The latter situation has two consequences: food bodies are not sheltered against unfavourable weather and are easily accessible to all ant species. This could lead to stronger competition between inhabiting ants and visiting ant species.

Non-myrmecophytic Macaranga species usually lack sheltered food body production sites. Macarangus tanarius produces the greatest number orfood bodies of all non-myrmecophytes investigated. In this species the pattern of food body distribution (many food bodies produced at the base of the stem) gives the impression that they serve to attract ground foraging ants to climb up the plants.

Function

EFN in non-myrmecophytic Macaranga species are visited by many different ant species, which probably also collect food bodies (Fiala, 1988; Fiala & Maschwi tz, 1991 ) . The specifici ty of the in teraction is very low, and the presence of ants on the plants is only of short-duration. A possible protective role of the visiting ants has bccn tested for three non-myrmecophytic Macaranga in New Guinea but such a function was not clearly supported by the evidence (Whalen & Mackay, 1988) .

In many studies the protective role of EFN through ant attraction has been investigated (review e.g. Bealtie, 1985; H uxley, 1986; Keeler, (989), but almost no information exists on. the distinct eITects of food bodies. Despite the production of EFN and food bodies some uninhabited Macaranga species had a higher amount of herbivore damage and vine cover than myrmecophytic species (Fiala et al., 1989, 1991 ) . However, after removal of EFN and food bodies the non-myrmecophytes were subject to even stronger herbivore damage. There appeared to be less herbivore damage on plants which had only food bodies removed than those in which EFN were selectively extirpated. This may indicate that these food rewards are differentially used by several ant species with different protective value for the plants.

.. '

FOOD BODIES AN D ANT ASSOCIATION 7~

\ \ ' (' lillllld dells!' lilOd hod y protcrtioll also on the floral bracts of female flowers or M, tri/alia (mynnecophyte), These bracts were visited by the inhabiting C'. bnrneellsis ants which may protect the flowers from herbivorous insec ts, M alr flo wns ormyrnw('oph yti c "'1. IlIIllfltii pl a nts, lor exam ple, which did not prod ucc li)od bodies, were fi'equentl y damaged by geomctrid ca trrpill ars (personal observa tion ).

Food bodies and the evolution 0/ n!),TlllfWpl!l'tism

O 'Dowd ( 1982) suggested that intrgration of' two food resources (EFN a nd fi)()cl bodies ) should increase the dependence or associa ted ants on the plants. Schupp & Feener ( 1991 ) found no correlation between food body production and prrsence of EFN in their ex tensive survey of these ant-associated fCaturcs among Panamanian forest plants. In A1ararallga a nd the related genus Alal/otlls EFN and food bodies occur together at on the gc neri c lel'CL Therr is, howe\,('I' . no indication of prevalence of food bodies in ,Hararallga species which in add ition have EFN. On the contrary, food bodies are best developed in myrmecoph yti c species which have reduced EFN, but offer nesting space. This is in agreemcnt with O'Dowd's finding that all three resources- - food bodies, EFN and domatia (structures providing nesting space)-rarely occur together, but food bodies either with EFN or domati a do.

Both EFN and food bodies a rc widespread in A1acarmlga a nd were importa nt prerequisites for the evolution of association with ants. However, since both still occur simultaneously on non-myrmecophytes, although EFN have been reduccd and food body production was increased in myrlllecoph ytcs, thr prcsence of' thcse two kinds of food rewards a lone does not ex plain Ill yrlllccoph ytism . The availability of protec ted nest sites on the host plant Illay ha\'e b('en the clecisi\'( ' additional factor which led to obligatory associations.

SC\'eral well known obligate myrmecophytes ex ist in other gene ra which oflc r: not EFN but food bodies, for insta nce Cecropia, Piper (review e.g. Huxley, 1986), II"here ants may tend homoptera ns. The dominating ant pa rtner C'. borneensis or the myrmccophytic A1acarallga species seems to depend ra th er strongl y on food bodies but also lives in close relationship with scale insec ts through which the ants obtain plant sap . Although a nts may not neccssa rily require additional food li'olll the host plant, food bodies may be an importa nt I'cature a llrac ting the ants to patrol the leafsurlaces, thus increasing ant defence against herbivory. For the pla nts it is important that ants do not simply visil thcm tempora ril y fi)r f()ra gin g but stay permanently. Therefore structures increasing ant fidelity , such as nes ting space, shou ld be selected for. Dom at ia (or suitable preada ptations fi ll" nest ing space) should be a n importa nt a ttribute enhancing the fidelit y or the ant assoc iation and faci lita ting the cultiva tion of scale insects.

McKey (1989) discussed the significance 0 (' various preadapl a tions fi)r symbiotic ant-plant systems and em phasized the possible role of domatia. \\'c believe tha t evolution of myrmecophytic species in the genus A1acarallga was a lso based on morphological host plant traits which could sen'e as nest ing space fi)!' ants. Studies of stem morphology showed tha t hollow structures or pred ispositions for cavities seem to be a very important fea ture for obligate interactions with a nts in this genus (Fiala & Maschwitz, 1992).

74 11. FL\LA ANIl 11 ,\L\SCI-I\\TrZ

Anl-atlracting food bodies have certai nly played a rolc in evolving a closer relationship wilh ants. Despite this, many J\1acarnllga spec ies ha\T not evolved more than a facultative interaction with ants. The localization and increased prod uction of foud bodics on spcci lic structu rcs is fllLIIld only in AI awrnllga species with domalia. vVe sugges t that only Macarnnga species with predispositions for domatia developed into obligate myrmecophytes . Ant inhabitation certainly offers advantages for the Macaranga plants, especia ll y concern ing plant competition, e.g. vines, which arc abundant in habitats where Macarmlga grows (Fiala et al., 1989). Species without ants have developed other survival strategies, and their competitive abi lit y can be achieved by different habitat requirements as well as by different types of growth (Fiala et al., 1991 ). However, myrmecophytic Alacaranga species belong to those who have most successfu ll y made the transition from primary forest to secondary habi tats. The association with ants is an important factor allowing Macaranga plants to grow at sites of strongest competition .

ACKNOWLEDG EM ENTS

The director of the Forest Research Institute Malaysia (FR IM ) Dato Dr Salleh Mohd. Nor generously gave permission to work at FRIM and its field stations. We thank Mr K. M. Kochummen and Mr L. G. Saw for their support at the herbarium of FRIM. The Department of Zoology, Universiti Malaya granted permission to use their field station in Ulu Gombak. We are indebted to W. Killmann (GTZ) for logistic support. Dr.J. Longino kindly checked the Crematogaster specimens. M. Ruppel helped with the SEM studies. We thank Prof. F. Rickson for information on unpublished work, Prof. D.O'Dowd, Dr P. E. Brandham, Or A. J. Helbig, and an anonymous referee for valuable commen ts on the manuscript. This study was financially supported by the Deutsche Forschungsgemeinschaft (DFG) and by research grants of the DAAD (German Academic Exhange Service) to B. F.

REFERENCES

BAKER, J. A., 1934. Notes 011 the biology of Macarall.~a spp. Gardens Hul/rtill Straits Sruiemmts, 8: 63 - 68. BEATTIE, A . .1. , 1985. The l::volutiollary Ecology of Ant-plant mutualisms. Cambridge: Cambridge Uni"ersit y Press. BCCKLEY, R. C. (Ed. l. 1982 . ..IlIt-plallt illterartionJ ill ..Iustralia. The Ha"uc: IV. Junk Publishers. D.-\\·ll)SON , D . ,,' .. FOSTER. R.ll .. SNELLI NG. R . R . & LOZ.-\DA. 1'. \I'., 1!l91. VariabI.- compositioll of

SUIllC lropkai alll-plalll symbios('s. In P. \\'. Prin: et al. l Eds ), Plant-.·llI;mal lnier{1(tions: Evolutioll{17J! I~'(olo,~v ill Tropiral and Trlllp"atf R'.~ions: 1{5- 162. Nl'w York : .I. \\'iky & SOIlS.

FL\LA, B .. 198H. Hiolo,!!,if. I:llnk ,ioll lI1ul El'vlutinll eint'.\" lI/aln.rsij(lim . \~ )'rmekfll)kvti(~rslemJ: Die ..t sso.:ialiof1 l'fln

Crem{1t()lf,mter bomemsiJ ( Form.: .\~)'rmi(illaf) mil Hiiumfl1 der Gatllmg J\1araranga (EuphorhiaCfoe). Cl1publishcd PhD. thesis of thr Cnivcrsity of Frankfurt.

FIALA, B. & \ ·IASCH\\TrZ, lJ., 1990. Stud ies 011 the South East Asian ant-pla nt asso{' iatioll Crnualo/!,aJtff bOT11UlIsis/J\I{1(arnnga : Adaptatiolls of the allt partncr. 11Is(cles Soc/aux. 37(3): 212- 231.

FIALA, B. & \.-IASCH\\TrZ, tj .. 1991. EXlraOoral lllTtaries ill the gcnus l\J{1Wrall,!!,a (Euphurhiact'at' ) ill \I a laysia: ("urnparat i\T studies ur tlll'ir possihk significalHT as prl'disposili(lIls Icx myrrnccuphytislll. Iiinlngical .7ollrnal 0/ the I.innran Sorir!v, 44: 2n7 307 .

FIALA, B. & MASCH\VITZ. C ., 1992. Dumalia as most important adaptations in Ihe (" 'u lu lioll or m yrm('("ophyll 's in the pa\cotl"opicai tree grllus .\lararnlll?a (Euphorhiacrac.: ). Plant 5lvslfmatiCJ and Evolul;oll, ~~M. -

FtALrI, B. , ~IASCH\\TrZ, c. & THO, Y. P., 1991. The associat ion between Mararangn and ants in SOllth East Asia. In D. F. Cutler & C. R . Huxky (Eds i, Interar/ions Iietwall !lnts alld Plallts: 263- 270. Oxford: Oxfi)rd C ni\"{TsilY Press.

FOOD BODIES ,\ N D ANT ASSO(:I ,\TIO N

ITII..I . B .. \I.\SCII\I·IT/ .. t · THO. Y. 1'. &. IIEI.BI( ; .. \ . ./. . I!lfl'l. Sludi," .. I' a Soulh East .\ , iall ,,,<1 -1'1.,,<1 .t""Ol ' j ;ltillll: prOll'llill1l or . \I(/((IIt11I.~1I 11'1'1'''; h y (."I'IIU1II/.~I/I'O b OIlII'''''I/I. Oflll/" ,t:ia ( Il , din ) . i!J: -H,:\ ·I i tl.

11 .\(;1-::\.1\. . S .. D.II)I) . R . 11 .. &. REESE . .1 .. 1'1!I4. TI ... Ii,od or ill ' ''' 1<. III C . B. lIulbk,.,. &. R . I.. R.d.!, ! Eel , , ,.:,."llIp,i,II/ 1:'lIfIlIl/IJIIJ,l!.r: 7q 11 2. ~t·\, · York : \\ ' ik~

1Il ' ;\I.EY. C . It .. IqHIi . 1':\0111114111 OrlWlW\'IIIt ' 1l1 anl -pblJ[ relatiollships. III B . .Jllllipn & R . SIlIIIII\\"(HJd Ld". 1,,1/,(/\ (II/(llh,' 1"(11/1 IIIr/ofl' : '2 :-, 7 :W'2. I.olldoll : E. .\I"II()ld .

. 1:\:'\!ZE 0!. D. H .. 1966. (:o(' \'o IUl ioll of mu ilia lis III ht' ! W(T Il a nts a lld ac;tc ias ill C(, 111ral Anu'rica . 1~·l'olllliflrJ. 'l(j : H'1 2i5

KEELER , K. H ., 19H9. Anl-planl intera ctions. 111\\', C. AhrahaIllS() II . P/r11l1-flUimlll l uln(1(limB: 207 2-1-2. ;\n, York: ~lacG raw-Hill.

:\lrKEY, D ., 1989. Interact iolls between a nts a nd leJ.{lIln inous plallt s. III C . H . Slirtoll & J. 1... Zarucchi ( Ed ~ , .-Jdl.'(J1/CfJ ill I .f/!,ume lli% ,f!,J: 673- 718 . I ,\1011o,!!,ralJ/u in Srslf1l1alirnl n(llm~r. 29. SI. I.ouis: l\lissfluri Botal1ica l G<lrd('11.

~IOKRASCH, L. C:., 195 1k Analysis of hexose phosphat(,s alld Sll .gars \\"ilh tilt' alllhrOIIt · r('agcllt. ] 1Jf/nw/ 0/ Iliolv,l!,i(fl/ Chfmil·lr]. 208: 55 59.

( )' 1)( )\VD, D . .J ., 19B'2. Pearl bodies as allt fi )od: all ccological rl1lc 14.r SOIlIC' kaf eIlHT.l{t' IHTS of t rlJp iral pia Ills. liio!mpica , /4: 40- 49.

ONC. S . L. , 1978. Ec%!{yq/ lhfal1t-aJsQcialioTl ill l\taca rall~a Irilol );I . l ' npllhlishco Ph .D. Ih ('Sisofl hl' l -ni \"tT ... il\ or \Ialaya, Kuala Lumpur.

PENZI<; , 0 ., IR92. U I)('r die P('r1driis(, 1I dt ·s \\'( 'illsl()('ks IlIHI ;1I1(lnn Pflilll i'.C Il . . 1111 dt! (,'01/.er(\ \{) /l1l/(l1lfHI

Intt'ma:;..iollalr (Gel/tHla ): 237 -- 245. R .. \CIBORSI\.I , ~I., 1898. Biulogisdlc ~1i1ll'ilulIgl'1I au, ,I;,,·a. I ·j"",. 8';: :\2'i :Hi7 . RACIBORSKl, M., 1900. Uber m yrmcko philc I'lIaIlZl·II. N om . Ri: 38 ·15. R ICKSON , F. R., 1980. Develupmental a n a tolllY and ultrastructure (lr tht' allt-limd hodies {13ctTariall hod it · ... '

or .\InrnrnTl,l!,a Irilobn and NI. ~)I!JlJII'lJ({/ (I':up li ori>i acc·'l(· l . . 11II0;U/1/ }l!lIma{ (!/ n(llm~r. 67: 2B:> 2()2 . RICKSON I F. R ., & RISCH , S.J. , 19tH. Anatomical and IIltr;\slrnrtur;d a spect s oftl!c ;IIJt~li)tld cd1 of I'I/If'

{flfllOcladum C.DC. (Piperarrac ). AlIIfrira ll J ournal oJ 80tf1!~r. 71: 12{iH 1'27 .. \-' ROCPPERT, C., 1926. Ohscr\'atiolls sur !cs prrlulcs de dinTs('s ('spi'("('s r'" phalll'rogallH's. RIII/f l inr dll .\flllllutl

,/'IIiJ!oirr. .Va!/lTflle PariJ, 32: 102- 107. SCHliPP, E. IV. & fEENER , D. H ., 199 1. Ph ylogl"lly. lild()J"(JI 'lIId hibil"l dl'pl'l1d"II(T ura l1l-dl"klld .. d 1''''"1 '

ill a Pa namanian forest. In D. F. Cu tler & C. R . Hu xll'y (Eds l . I IIII'r(l(lill1l .1 Ht'lwfI'1I..I1I15 and P lrml.l: 17:') I!I '. O xford : Oxu)rd U ni versit y Press.

\\'HALEN, ~1. A., & ~1ACKAY, D . A ., 1988. PatttTlls or a llt a nd llt'rhi \·t1 IT act i\· it y till fi \-T under .... t,. ~ ('uphorhiac('u1I5 saplings in 5ubmolllallc Papua Ne\,· (;UilH·;I . 1I llllmpim. 20(.J ): '2~)4 30n.

\\ ' HIT~1 0RE , T . C., 1967. Studies ill A,ln{arflll,l!,(I. 1111 (' il SY gt'llllS or ~L1b Y;1I 1 \""ysi<ic trccs . . \ffllr~I'fl" . \"flllI't

J,urnal , 20: 89- 99. \1· HIT~10RE . T. C. , 1973. Trer jora 0/ ,\lalara. I\."a l" LUlllpu r. I.olldoll: 1.0 11 "Ill OJ 11 . \\"HIT~10RE . T . C .. 1975. "laCaral1,l!.a. III H. K .. \iry~S ha \\". 7"1' F II!,""rhif//fOfll/ /I f/nI l 'lI. JI ·/'if ' /lu lll'lllI. Jddil iolltll

Saie.<.4: 140- 159. \\"HIT~I()RE , T. C: .. 1982. Macarall.~fI. III H . 1\.. ,\i n ·-SIt 'I\'·. t hr Ic'upluohifl(Nu' '!! Suma!ra. k m' /lullr!ill. :~;:

312 :In \\' II.SON . E. 0. , 1971 . 7"f lusert SI/cif /ifs. Camhridge. \lass.: Iklkll ;lJ ) l' ress or I-lan·ard L'lIinTsily Prc .... s.