Sonderbd. naturwiss. Ver. Hamburg 163-186 Hamburg 1983

Dispersal and Distribution of and

By GHILLEANT. PRANCE & SCOTI A. MORI, New York, U.S.A. With 18 figures and 1 table

Sum mar y: The Lecythidaceae and Chrysobalanaceae have both become extremely successful in their colonization of and adaptation to the lowland habitats of the tropics. These families have little variety in their basic type, the Chrysobalanaceae having only a single basic fruit and the Lecythidaceae only two. These have been modified in many ways and have developed a large number of dispersal mechanisms crucial to their adaptability and success. The fruit of the Lecythidaceae (subfamily Lecythidoideae) is either an indehiscent berry­ like structure or it is a dehiscent circumcissile capsule (pyxidium) which has adapted to many different dispersal vectors. The adaptations of the aril have been particularly important, varying from the fleshy-funicle aril of the bat­ dispersed genus to the aril which forms around the seeds of the wind­ dispersed genera and . Other dispersal agents include: wa­ ter, fish, wild pigs, and . The Chrysobalanaceae are tied to a basic fruit type, a drupe, but have also adapted to many dispersal vectors in the different habitats in which they occur. Dispersal varies from marine in the predominantly insular genus Atuna of Indo­ nesia and the Pacific, to dispersal by the South American rhea in Parinari ob­ tusifolia HOOK. f. of Central , to ichthyochory in species of at least three genera. This variety of dispersal types is discussed in relation to habitat adaptabili­ ty, distribution in various habitats and the overall geographic distribution of genera and species in both families. Most genera of Lecythidaceae are characte­ rized by a specific dispersal syndrome but this occurs in many different habi­ tats. The fruit of most genera of Chrysobalanaceae are distinct, but are less correlated with a particular dispersal type. Examples of probable long-distance dispersal in both families are discussed. However, many disjunct species distributions are better explained by historical changes in vegetation cover.

163 A. Introduction

The Lecythidaceae subfamily Lecythidoideae and the Chrysobalana­ ceae are both extremely successful groups in the lowland tropics in terms of abundance and ecological importance. Both groups are found in a wide range of habitats and have some extremely wide­ spread species. As would be expected in such well distributed groups there is a wide range of dispersal mechanisms. Although ma­ ny types of dispersal occur, only three basic types of fruit are involved, two in Lecythidaceae and one in Chrysobalanaceae. These basic fruit types have become modified in many different ways in adaptation to the different vectors present in the lowland tropi­ cal habitats. Thus, all Chrysobalanaceae have a drupaceous fruit and all Neotropical Lecythidaceae a circumscissile capsule or its indehiscent precursor. Although both families are restricted evo­ lutionary to these basic fruit types it does not appear to have impaired their ability to utilize a large number of dispersal vec­ tors.

Ac k now 1 e d gem e n t s : We acknowledge with thanks National Science Founda­ tion grants BMS75-03724 A02; INT75-19282 and INT78-7823341 AOl which have en­ abled extensive field studies by both authors. We thank Frances MARONCELLI for typing the manuscript, Bobbi ANGELL for preparing the illustrations, and Rupert C. BARNEBY and vicki M. FUNK for critical reading of an earlier draft.

B. Fruit types

1. Lecythidaceae

All Neotropical Lecythidaceae belong to the subfamily Lecythidoi­ deae except brasiliensis DESV. which belongs to the predominantly African Napoleonaeoideae (fig. 1A-D). Since our ex­ perience is with the Neotropical subfamily all further references to the family refer to the subfamily Lecythidoideae and not to the other three subfamilies which were defined in PRANCE & MORI (1979).

Three genera of Lecythidaceae have indehiscent , , and . In these three genera the fruit is large, roundish, and often woody, with a softish or a woody pericarp and the seeds are surrounded by a fleshy pulp (see Couroupita in fig. 3). The seven remaining genera have dehiscent woody pyxidia, ex­ cept that of Bertholletia excelsa HUMB. & BONPL~, the , which has become secondarily indehiscent (fig. SA-C). The general characteristics and terminology of the Lecythidaceae pyxidium are given in fig. 2 which illustrates species of Lecythis. The pyxi­ dium is a circumscissile capsule with a dehiscent operculum that encloses compactly organized seeds which are not embedded in a pulp when mature. The variations of the basic fruit are discussed further below in relation to dispersal.

164 I(em.

B :· · . ·.· ·~ . ' 0D: ' O · ··· '· ·· _1 ~:'.':": .:.... i.: . • lem.

Asteranthos

Gustavia K;U ~r,)~' [,i.; , I I" ~h

Grias

Fig. 1: Fruit and seeds of Lecythidaceae. - A-D. Asteranthos brasiliensis (PRANCE & al. 15481): A. Entire fruit with persistent calyx. - B. Lateral view of seed. - C. Longitudinal section of seed showing tubular embryo embedded in copious endosperm. - D. Cross section of seed. - E-H. Gustavia spp.: E. Fruit of G. superba (MORl 837). - F. Fruit of G. brachycarpa (PlTTlER 5269). - G. Seedling of G. augusta (NEE & MORl 4194). - H. Seed of G. augusta, note the fleshy funicle. - l-K. Allantoma lineata (PRANCE 11618): I. Base of fruit. ­ J. Operculum. - K. Seed with attached funicle-aril. - L-M. Grias neuberthii (BOEKE 2211): L. Entire fruit. - M. Apex of fruit. - N. Seed.

165 II. Chrysobalanaceae

The fruit of all Chrysobalanaceae is basically a unilocular or bi­ locular drupe with an outer mesocarp surrounding a thick or thin endocarp that encloses one or two seeds. The mesocarp is most fre­ quently soft and fleshy and of variable thickness. In many cases

proximal end

infracalycine [ zane

supra calycine { zane distal end line of aperucular dehiscence B

calycine ring

line of opercular apercu I or opening dehiscence opercular ring

c D ~ E ~"m.", a~.il ~ ~ ...... funicle

Fig. 2: Diagramatic representation of fruit characteristics and terminology in Lecythis, a genus with dehiscent fruits. - A. Lateral view of the fruit of L. ampla (operculum removed, MORl 316). - B. Distal view of fruit of L. ampla (operculum removed). - C. Operculum of L. ampla. - D. Operculum of L. pisonis (MORl 399). - E. Seed of L. minor (MORl 423). it is sweet, and consequently edible. In some, for example Atuna (fig. 7A), the mesocarp is hard and woody. The endocarp is usually indehiscent and on germination gradually breaks up. The endocarp has become dehiscent in some genera, either by longitudinal lines of dehiscence as in Chrysobalanus (fig. 8) and most species of Hirtella, or by lateral plates as in Grangeria, Maranthes and Para­ stemon (fig. 17), or by basal stoppers or obturamenta as in Neo­ carya and Parinari.

166 C. Dispersal mechanisms

I. Lecythidaceae

1. Indehiscent fruit

As pointed out above, the fruits of only three genera of Lecythi­ doideae are indehiscent. These are all genera which we consider to be relatively primitive. It is interesting to note that all Old World Lecythidaceae of the other three subfamilies have indehis­ cent fruits, indicating that, since indehiscence is a primitive character, the family originated in the Old World. This conclusion is correlated with many floral characters and is not based on fruit dehiscence type alone.

The two genera, Gustavia and Grias, which we consider most primiti­ ve in both floral and fruit structure, have indehiscent fruit (fig. 1: E-H, L-M). Those of both genera mature on the and drop directly to the ground or water as the case may be. Those of Gusta­ via and Grias differ from those of all other Neotropical Lecythi­ daceae in that they are softish. The mesocarp is easily bitten in­ to by animals. The entire fruit together with the numerous seeds is the initial unit of dispersal as it falls from the (baro­ chory). However, in each species where observations have been made, secondary dispersal occurs, most frequently by animals. For exam­ ple, the orange mesocarps of (KUNTH) BERG and G. grandibracteata CROAT & MaRl are eaten by mammals. Once the exo­ carp has broken open, the orange mesocarp attracts animals and seeds are dispersed as the fruits are carried to eating places and the seeds themselves are often scatter hoarded. Gustavia augusta L. is a riverine species whose fruits usually fall into the water. Not only are they rafted by water, but they are eaten by fish and have been found in their digestive tract (HONDA, 1974°; PRANCE, pers. obs.). This is one of many Amazon sp~cies in which ichthyo­ chory is important (see GOULDING, 1980).

Although most species of Gustavia and Grias have a soft fruit, a few species such as L. have a harqer exocarp. Grias cauliflora is a riverine species and as early as 1917 GUPPY reported that the entire fruits are buoyant, although the seeds alone sink. He observed that the fruits f Loa't; readily beside the banks of rivers on which it grows. They can float for months and germinate in river drifts. They are, however, killed by sea water, although the seeds are frequently found amongst beach drift. G~ias cauliflora is common in from Belize to , but also occurs in Jamaica, which may indicate that long distance dis­ persal over sea water occurred successfully at least once. However, the fruits of G. cauliflora, do not differ SUbstantially from those of the other species of Grias and Gustavia. The exocarp of G. cauliflora does not differ from that of its animal dispersed congeners. This species is riverine and also primarily animal-dis­ persed. Its secondary dispersal by water has drawn more attention from previous workers because of its occurrence on Jamaica. The fruits of Grias have a soft fleshy mesocarp that is eaten also by native people throughout its range and it is possible that it was transported to Jamaica by early humans.

167 The genus Couroupita (fig. 3E-H), the cannon-ball tree, also has an indehiscent fruit, but this is much larger and the pericarp much more woody than in Gustavia and Grias. The large, round, hea­ vy fruit fall from the trees of all three species and can be found in quantities under most mature trees. We have frequently found rotting fruit under the parent trees and seedlings beginning to germinate without secondary dispersal having ocurred (fig. 4C-E). However, there is on record at least one observation (H.-W. KOEP­ CKE, pers. corom.; see PRANCE & MORI, 1979) of a tree of Couroupi­ ta guianensis AUBL. whereafter three weeks of observation a herd of wild pigs foraging under the tree ate all available fruit. It is possible that the pigs, who made only occasiona~ visits to the Couroupita during their wide-ranging foraging tre~s' through the forest, are dispersal agents. All species of Couroupita have their seeds embedded in a sweet smelling, but unpleasant-tasting pulp which is used widely in Amazonia as a food for chickens and dome­ stic pigs.

2. Dehiscent fruit

In species with dehiscent woody fruits without a fleshy pulp, the development and variations of the aril structure are most impor­ tant for dispersal. The large-fruited species of Lecythis have a fleshy aril surrounding the funicle (figs. 3J-K, 4B) which is ad­ apted to bat dispersal. GREENHALL (1965) first observed bat dis­ persal in subsp. usitata (MIERS) PRANCE & MORI and we now have additional evidence of this. Upon dehiscence of the operculum the large woody pyxidium remains on the tree. After the tack-like operculum falls to the ground the seeds remain han­ ging attached by the large funicle-aril (fig. 4B), which is sweet­ tasting. Bats enter the pyxidium and remove the aril and seed to­ gether. After eating the aril they drop the untouched seeds either in flight or at their roosts. Other animals, especially monkeys, eat the seeds of Lecythis. They probably disperse some seeds, but are mainly predators since they destroy many. Lecythis' appears to be primarily adapted to bat dispersal.

The closely related genus , has smaller woody pyxidia (usually containing fewer seeds) with only a small funicle-aril appressed laterally to them (figs. 4F, 5M-R). The pyxidia remain on the tree and the seeds are exposed by the dehiscence of the operculum. The fruit of several species of Eschweilera are opened by parrots and macaws which are quite destructive, but drop many seeds to the ground, usually around the parent tree. Monkeys also eat the seeds of some species of Eschweilera. BUCHANAN & al. (1981) reported that seed of E. corrugata (POIT.) MIERS and E. I, subglandulosa (STEUD.) MIERS are eaten by Saki monkeys in Surina­ me, and ROOSMALEN & al. (1981) report that Chiropotes monkeys use very different techniques to open fruits and remove seeds of se­ veral species of Lecythidaceae. They list five species of Esch­ weilera, E. chartacea (BERG) EYMA, E. congestiflora (R. BEN.) EY­ MA, E. corrugata (POIT.) MIERS, E. poiteauii (BERG) NDZ. andE. simiorum (R. BEN.) EYMA as well as other Lecythids, Couratari ob­ longifolia DUCKE & KNUTH, Couratari stellata A.C. SMITH, Gustavia hexapetala URB. and Lecythis davisii SANDW. Parrots and some mon­ keys certainly eat the seeds of some species of Eschweilera, but only the lateral aril is eaten by other animals and the seed is left intact. Many species of Eschweilera have extremely bitter seeds. The seeds can often be found on the ground in large quan­

168 tities under trees of Eschweilera. They are secondarily dispersed by agoutis which scatter-hoard and thereby carry the seeds further away from the parent tree.

At least one species of Eschweilera, E. obtecta (MIERS) NDZ. has an aril that completely surrounds the seed and hardens, making the seeds bouyant (fig. 5H-M). This species is a common riverside tree that is water dispersed. When the pyxidium dehisces, the unusually numerous seeds fall into the water. They can remain afloat up to two months (PRANCE, pers. obs.). The monotypic genus Allantoma has unique flowers and fruit (fig. 1I-K). The elongate cylindrical fruit of the riverine A. lineata (BERG) MIERS dehisce during the flood season. The numerous seeds are showered onto the water below and float because of their high oil content, first noted by DUCKE (1948). The seeds remain afloat and viable for up to six months (PRANCE, pers. obs.) but are usually stranded on the forest floor and riverbank as the water level recedes. They are woody and are unlikely to be eaten by fish; consequently, A. lineata is probably entirely dispersed by water. Its distribution (fig. 6) limited to the main courses of the Rio Negro and Amazonas indicates its dis­ persal limitations.

3. The case of Bertholletia

The best known fruit of Lecythidaceae is that of the monotypic ge­ nus Bertholletia (fig. SA-C) which is the Brazil nut of commerce. The large round fruits have an exceptionally hard woody endocarp enclosing the strong woody 3-angled seeds which have a thick, in­ durate rugose coat. There is no funicle or aril in the mature seed. There is a small operculum, but unlike that of other genera it falls inward and the small opening is much narrower than the seeds. In contrast to the large pyxidia of Lecythis which remain on the tree, the entire pyxidium of Bertholletia is dropped to the fo­ rest floor. HUBER (1910) was the first to solve the mystery of the dispersal of Bertholletia when he discovered that the pyxidia are gnawed open by agoutis. The agoutis eat a few seeds and then carry and bury the rest by their well-known scatter-hoarding techniques, often as far as 50 m from the tree. SMYTHE (1978), from observa­ tions of dispersal in other families, pointed out the poten­ tial for further dispersal by secondary dispersal when another invades the hoard of the first one.

The reduction of operculum size in Bertholletia and subsequent loss of effective dehiscence is probably a secondary characteris­ tic which has developed from a dehiscent fruited ancestor. The closest living relative of Bertholletia is Lecythis, in which the pyxidium of all species except L. lucida (MIERS) Mori is dehis­ cent. The secondary indehiscence in Bertholletia would be selected for as a protection from predators. The mature fruit is so hard and woody that the seeds are effectively protected from any other mammal or bird predator.

Fig. 4: Fruit of Lecythidaceae. - A-B. Lecythis pisonis ssp. usitata (PRANCE & ale 24363): A. operculum of uppermost fruit has fallen. - B. fruit with opercu­ lum removed, showing seeds attached to funicle-aril. - C-E. Couroupita subses­ silis (PRANCE & ale 24446): C-D. fruit found germinating under parent tree; ­ E. showing seedling. - F. Eschweilera sp. (MORl & al. 8727): showing the late­

170 Couroupita

",,,,,,;;;.~d'....

Fig. 3: Fruit seed and seedling characteristics of Lecythidaceae. - A-C. Cari­ niana micrantha (KRUKOFF 5095): A. base of fruit; - B. operculum; - C. winged seed, the wing is a flattened aril with the funicle passing through it; - D. (from DUKE, 1969): apex of seedling showing foliaceous co­ tyledons. - E-H. couioupita guianensis (E-G, FROES 1705; H. MORl & KALLUNKl 3263):E. entire fruit; - F. cross section of fruit showing six pulpy segments; - G. longitudinal section showing seeds embedded in pulp; - H. apex of seed­ ling showing foliaceous cotyledons; - l-K. Lecythis usitata (l-J, PRANCE & al. 20210; K, PRANCE & al. 24363): I. base of fruit with attached pedicel; - J. operculum; - K. seed, not funicle surrounded by fleshy aril; - L. Lecythis mi­ nor (PRANCE 23171): seedling. - M. rimosa (PRANCE & al. 23518): base of fruit; - N-P. Corythophora alta PRANCE & RAMOS 23312): N. base of fruit; - O. operculum; - P. seed.

169 ral funicle-aril. - G. (MORl & KALLUNKl 3263): showing the foliaceous, phanerocotylar cotyledons.

171 G

D

ctoN .,.] M p R 11": A" : ~"v.«

I ')"':;)::1:1 1 K 5cm. ~.J. "1.-, Q

JO OL Eschweilera Eschweilera

Fig. 5: Fruit, seeds and seedling characteristics of Lecythidaceae. - A-C. Bertholletia excelsa (A-B: PRANCE 16599; C: PRANCE blastogeny col. 4): A. enti­ re fruit; - B. seed; - C. seedling. - D-F. (MAGUIRE & CO­ WAN 33918): D. base of fruit; - E. operculum; - F. seed. - G. Couratari stella­ ta (MORI & al. 9160), seedling. - H-L. Eschweilera obtecta (H-L, PRANCE & al. 24357): H. entire fruit; - I. lateral view of seed; - J. cross section of I; ­ K. lateral view of seed; - L. cross section of K. - M-N: E. odora (M, PRANCE & al. 8728; N, MORI & al. 8728): M. seedling showing lateral germination; - N. two lateral views of seeds. - O. Eschweilera sp. (MORI & al. 9098), lateral view of seed. - P-R. Eschweilera pittieri (MORI & KALLUNKI 2830): P. basal view of fruit; - Q. lateral view of fruit; - R. seedling.

172 4. Winged seeds

Two genera of Lecythidaceae, Cariniana (fig. 3A-D) and Couratari (fig. SD-G), with dehiscent pyxidia, have winged seeds. The pyxi­ dia in both genera are long and cylindrical like that of Allanto­ ma, or are conical. The aril, which is normally thick and fleshy in other genera, is flattened to form a wing. In Cariniana it is

Fig. 6: Distribution of Allantoma lineata.

unilateral, whereas in Couratari it surrounds the seed. The diffe­ rence in the wing is not surprising as floral morphology would place Cariniana among the more primitive actinomorphic Lecythida­ ceae whereas Couratari with its most complicated androecial struc­ ture is the most advanced of all Neotropical Lecythidaceae. Wind dispersal appears to have developed twice in Neotropical Lecythi­ daceae. The unilateral seed wings of Cariniana, probably evolved from an ancestor with a basal funicle-aril. Cariniana is closely related to Allantoma which has this aril type. Couratari, with the wing surrounding the seed, probably evolved from an ancestor with a lateral funicle such as that found in the closely related Esch­ weilera.

The long narrow, pendulous pyxidium occurs only in Allantoma, Ca­ riniana and Couratari all which shed their seeds mechanically from the attached pyxidium and have no beneficial animal association (However, ROOSMALEN & al., 1981, reported monkey predation of seeds of two species of Couratari). The shape is probably asso­ ciated with the efficient ejection of the seeds.

173 The above review shows the wide variety of dispersal types that have developed from two basic fruit types. The modification of the aril, the dehiscence or indehiscence of the fruit, whether the en­ tire fruit or just the operculum is shed from the tree, and the angle at which the fruit occur on the branches are all important ways in which dispersal variability has been achieved in Lecythi­ daceae.

II. Chrysobalanaceae

Although there is much less apparent morphological variation in the single fruit type of Chrysobalanaceae (fig. 7), there is also a wide range of dispersal types. The predominant fruit type is a drupe with a fleshy mesocarp surrounding a hard woody endocarp that protects the embryo. As would be expected with this fruit ty­ pe, many Chrysobalanaceae are distributed by mammals and birds. For example, RIDLEY (1930) recorded that the fruits of Parinari insularum A. GRAY of are eaten by fruit pigeons and he repor­ ted on experimental feeding of Chrysobalanus icaco L. (fig. 8) to captive hornbills and cassowaries that retained the unharmed seeds for over ten hours before excretion; PIJL (1972) reports that Chry­ sobalanaceae fruits are food for bats; and GOULDING (1980) repor­ ted three occurrences of Licania longipetala PRANCE (fig. 7G) in the stomach contents of Amazonian fish. One of us (PRANCE) has ob­ served Chrysobalanaceae fruit being eaten by birds, bats, agoutis, monkeys, pacas, and fish. In all cases, the mesocarp is eaten by the animals and the endocarp remains intact and is weakly transpor­ ted some distance.

Some specific examples from recent field work which remain unre­ ported include the following:

Observations made on a small blackwater tributary of the Rio Soli­ mOes of Licania apetala (E. MEY.) FRITSCH, a common riverside spe­ cies of Amazonia, showed that the soft fleshy fruit fell from the trees in great quantity during the flood season. As the fruit fell into the river they were rapidly swallowed by fish. This supports the observations of GOULDING (1980) in L. longipetala. Ichthyocho­ ry is apparently quite common in riverside and inundated forest species of Chrysobalanaceae, especially in the genera Licania (fig. 9), Parinari (fig. 10) (P. sprucei HOOK. f.), and Couepia (fig. 11) (c. maguirei PRANCE).

The geoxylic suffrutex, Parinari obtusifolia HOOK f., which is com­ mon in the cerrado of Central Brazil, has the local name fruta de ema for a good reason. It is a favorite fruit of the South Ameri­ can rhea (= ema in Portuguese) that inhabits the same habitat (PRANCE, field data). This flightless bird, which runs fast, can effectively disperse the seeds over a considerable distance. The genus Hirtella has a smaller fruit than that of most Chrysobalana­ ceae (fig. 12). It has a red or black exocarp and a fleshy meso­ carp, and that of many species are eaten by birds (observed in H. racemosa LAM. by MACEDO & PRANCE, 1978 and in H. glandulosa SPRENG. by PRANCE) .

174 I I 2em 2em I 1 B A

) 2em \ 2em I I

D c

I \ I @) T T T F

I CD I T I 2em J 2r

I K

Fig. 7: Fruit of Chrysobalanaceae. - A. (MERILL 270). - B. Coue­ pia racemosa (PRANCE & al. 23971). - C. (PRANCE 4706). - D. (PRANCE 4368). - E. Licania brittenii (PRANCE 22843). ­ F. Licania alba (BUZA 305). - G. Licania longipetala (RIMACHI 2866). - H. Lica­ nia heteromorpha (RAMOS 378). - I. Exellodendron barbatum (PRANCE 25574). - K. Acioa gillettii (LOUIS 16122).

175 Various Chrysobalanaceae do not have a soft mesocarp and are dis­ persed by other methods. For example some species of Couepia have a hard pericarp (e. g. C. longipendula PILG.). When they fall to the ground they are scatter-hoarded by agoutis like the Brazil nut. In the case of C. longipendula (fig. 7D), it is the entire fruit

.J~OP!E_oJ £a!:!ser ' q ------~ --!fs:fJ;- ~- ----­

I~op~c_oJ_9p~I£0!~ _ (j .------­---. ----- c:)---. --~.

ChrYsobalanus (7 d

Fig. 8: Distribution of the genus Chrysobalanus and fruit of C. icaco (PRANCE 2102).

~U!!O~ _ ~'~~~'*".m:'fl!:!r~~------J

J~op!..C_O! _~a.P~I£O!~ , .-u-~.-. -~-~ ------~ .. -{j------~.

Licania (7 d

Fig. 9: Distribution of Licania (Chrysobalanaceae).

176 that is buried and not just the seeds or 'nuts' as in Bertholletia. Couepia racemosa FRITSCH (fig. 5B) is also eaten and dispersed by agoutis. Hydrochory is common in Chrysobalanaceae. There is obvious poten­

.J~OD!f~o.!S_"!:!£.e!. "

------Equator

!!.op~r;_o.!_c;.ap~l£o!,:! I

Pari n ar!

Fig. 10: Distribution of Parinari (Chrysobalanaceae).

~U!,!O~ _

!!OJI!.c_~ _~'P!.I£CI!Il _

J~~~;~~~; Couep I a

... Magnlstipula

mm! HUMga

Fig. 11: Distribution of genera of Chrysobalanaceae: Couepia, Magnistipula and Hunga. - Fruit of Hunga rhamnoides (GUILLAUM.) PRANCE (MacKEE 28267).

177 tial for water dispersal in the species that are fish dispersed. However, there are other waterside species whose fruit are unlike­ ly to be eaten by fish. The genus Atuna of the Archipelagos of the

.J[OD!LqLCjl~e.!. " q

~U!!o!... ~ _ t!Jr- ~-..---­ -­

.!~op!.c_oJ _~ap~r£~~ M _ .~------_ . .: ----- L---.J p .

Hirtella \7 d B

Fig. 12: Distribution and fruit of Hirtella (Chrysobalanaceae). - A. H. guate­ malensis STANDL. (JONES & al. 3518). - B. Hirtella racemosa LAM. (LISBOA 348).

Far East and Pacific is the best example. The fruits of Atuna (fig. 7A) have a hard pericarp which are light and consequently gives the fruit buoyancy. RIDLEY (1930) observed the fruits of Parinari laurina (= Atuna racemosa RAFIN.) floating in estuaries in Fiji and the and noted that the seeds float for months. He also recorded that the fruit of Chrysobalanus icaco L., in ad­ dition to being eaten by birds is dispersed by the sea. GUPPY (1917) recorded that seeds of C. icaco floated for two months and that the seed of at least one remained intact.

D. Dispersal and habitat

The correlation and modification of dispersal type with habitat is interesting since species of both Lecythidaceae and Chrysobalana­ ceae occur in a large variety of habitats. The winged seeds of Ca­ riniana and Couratari, are the only ones adapted to wind dispersal, a mechanism which is not common in neotropical forest species. An analysis of the species of both genera shows that either they grow in open habitats such as riverside or savanna or that they are un­ usually large trees emergent above the forest canopy (table 1). Winged seeds are obviously a more efficient means of dispersal for an emergent than for a canopy species. Throughout their range the correlation is close, for example Cariniana pyriformis MIERS is an

178 emergent of the forests of northern and c. legalis (MART.) KUNTZE is an equally tall tree of the forest of Atlantic coastal Brazil. The adaption to wind dispersal has also committed the fo­ rest species of the genus to the emergent layer of the forest. Si­ milarly the genus Atuna (fig. 13) is adapted to water dispersal by

Tab. 1: Species of the winged fruited Cariniana and Couratari and their habitat, showing the correlation between habitat and effective wind dispersal.

Cariniana Couratari

decandra Forest emergent asterotricha Forest emergent domestica Forest emergent atrovinosa Forest canopy 50 m tall calycina ? estrellensis Forest emergent gloriosa Forest emergent 50 m tall guianensis Forest emergent ianierensis Forest emergent longipedicellata Forest emergent integrifolia Forest canopy macrosperma Forest emergent kuhlmannii ? multiflora Forest emergent legalis Forest emergent oblongifolia Forest emergent micrantha Forest emergent oligantha Riverside 65 m tall prancei Forest emergent multiflora Forest emergent pyramidata Forest canopy pachyantha Forest emergent riparia Riverside pauciramosa Forest canopy sandwithii Riverside and penduliflora ? savanna pyriformis Forest emergent scottmorii Forest emergent 50 m tall stellata Forest emergent rubra Cerrado tauari Forest emergent uaupensis ? tenuicarpa Riverside

.J~.op'~..:..oJ_Cjl!!£.e.!. '

~u!.!~ _

I!Op!.C_o! _~ap~I£C!,! • _ mm Exellodendron

• Neocarya

~~~~~;~!~: At una

Fig, 13: Distribution of genera of Chrysobalanaceae: Exellodendron, Neocarya and Atuna.

179 its large fruits with a hard light pericarp which float. It is a genus distributed throughout and into the Pacific as far as Fiji. The habitat and distribution are logical for water dis­ persal.

The two monotypic genera Allantoma and Bertholletia which exhibit extremely marked fruit adaptations are also restricted in habitat. Allantoma is found only in riverside and flooded forest of the Rio Negro and the Rio Amazonas below its confluence with the Rio Negro (fig. 6). This linear distribution is more probable in a solely water dispersal plant. Many other riverine which have been discussed are also fish-dispersed, a method which enables upstream dispersal of a riverside plant. For example the ichthyochorous Gu­ stavia augusta L. is much more widespread than Allantoma. Berthol­ letia, with its hard woody fruit that can only be opened by agou­ tis, is confined to the forest on terra firme rather than the ri­ verside.

The other genera of both families, although they have a distinct fruit type, are not so confined to one habitat by their dispersal vector. For example species of Eschweilera are found in many diffe­ rent habitats. Their more general dispersal by ~irds and animals is viable in forest, savanna, mountains and elsewhere. Licania fruits are generally fleshy and are eaten by animals, but there are possible dispersers for the savanna species L. incana AUBL., the ichthyochorous L. longipetala PRANCE of riversides, the forest species of which the fruits are eaten by birds and bats or for L. tomentosa (BENTH.) FRITSCH which is confined to the restingas of Atlantic coastal Brazil.

Another example can be found in Hirtella, a rather small fruited Chrysobalanaceae, that is also represented in many habitats within . It is predominantly dispersed by birds which ingest the seeds. It is interesting that there are several widely distri­ buted savanna species (H. ciliata MART. & ZUCC., H. glandulosa SPRENG., fig. 14) occurring on many isolated savanna islands with­ in the Amazon forest region. As is characteristic of many savanna species these are adapted to long distance dispersal through the endoornithochory.

E. Dispersal and distribution

There are many interesting distribution patterns in both Lecythi­ daceae and Chrysobalanaceae. It is almost certain that trans-ocea­ nic dispersal has occurred several times. For example Chrysobala­ nus icaco L., whose fruit is known to remain viable in sea water (GUPPY, 1917), is distributed along the coasts of West Africa and eastern South America (fig. 8). It is widespread in coastal re­ gions, but extends inland on both continents, a pattern compatible with transport of fruits by both water and birds. Transport from island to island in the Caribbean and across the Atlantic are li­ kely to have been by water, but its fleshy mesocarp eaten by birds has enabled it to extend inland. Parinari (fig. 10) is a genus of extremely closely related species which are little differentiated morphologically even from continent to continent. One species,

180 \ \ ------\--­

\ \ \ I

09 " P. excelsa SABINE, occurs in Africa and South America and is like­ ly to have crossed the Atlantic. Although the fleshy fruit of Pari­ nari is eaten by animals and is predominantly dispersed by bats, the endocarp of many species are bouyant. The same is true of Ma­ ranthes (fig. 15), another genus of Chrysobalanaceae that occurs

.1!.oO!f_o.U;!I~e.!. '

------Equator

I!.oP!"C_oJ _~aP!.I~O!,! _

Maranthes I @ P T \7 d

Fig. 15: Distribution of Maranthes (Chrysobalanaceae). - Note its occurrence in the Neotropics in Panama, fruit of M. corymbosa (ELMER 11208). on three continents. The single Asiatic species, M. corymbosa MIQ., has successfully spread throughout the Malesian archipelago and its close relative M. panamensis (STANDL.) PRANCE & WHITE must ha­ ve reached Panama by long distance dispersal, not by human intro­ duction as originally stated (PRANCE, 1968).

Grangeria (fig. 16), a genus of Madagascar and Mauritius is most closely related to the Malesian Parastemon (fig. 17). They have identical fruit structure with the unusual feature of lateral de­ hiscence plates on the endocarp. It is probable that an oceanic dispersal allowed an ancestor to become established and later dif­ ferentiate generically. Similarly the occurrence of a single spe­ cies of Licania (fig. 9) in Africa and another in Asia indicates long distance dispersal. The other 175 species are all neotropical.

The Lecythidaceae distribution most difficult to explain is that of Asteranthos brasiliensis DESV. (fig. 18) which is an isolated genus of the African subfamily Napoleonaeoidae. However, it is in­ tersting to note that Asteranthos is a species of the flooded caa­ tingas of the upper Rio Negro. DUCKE (1948) observed that the per­ sistent calyx serves as a flotation device for the water dispersed fruits. The present inland distribution of Asteranthos is hard to explain by aquatic dispersal from the African Napoleonaeoideae.

There are disjunct species distributions in Neotropical members of

182 both families, such as Licania affinis FRITSCH in Panama and the Guianas, Licania guianensis (AUBL.) GRISEB. in the Guianas and Ama­ zonian , and Licania cymosa FRITSCH in Amazonia and Atlantic coastal Brazil. This has been examined elsewhere (for' example in

~C>

.!!.OP~_olJl'~~e! .'

~...~o!.... ,

.!~op~c_oJ _c;.ap~II:.t!~ _ ---a- HHlH Acios I II Grangeria T db ~;~;~~;;; Kostermanthus

Fig. 16: Distribution of genera of Chrysobalanaceae: Acioa, Grangeria and Ko­ stermanthus. - Fruit of Kostermanthus heteropetala (SCORTECH. ex KING) PRANCE ( FOR. DEPT. 21848) .

.!~OP!LoLC_a!!SeL '

------Equator

!!.op!..c.oJ _9p!.,~.o!~ _

;;;~;;;;~\ Bafodeya

II Parastemon

Fig. 17: Distribution of genera of Chrysobalanaceae: Bafodeya and Parastemon.­ Fruit of Parastemon urophyllus (A. DC.) A. DC. (TOROES 3315).

183 PRANCE, 1973, 1981, 1982) and is not discussed in detail here. However, the diaspore type of th~se species and of a number with heavy diaspores in other families make them unlikely candidates for long distance dispersal. These disjunctions have certainly been caused by a variety of vicariant events in the region such as the end of continuous forest contact between Atlantic coastal Bra­ zil and Amazonia in the late Tertiary, and the reduction of forest cover during the drier periods of the Pleistocene. It is important

Fig. 18: Distribution of Asteranthos (Lecythidaceae). to distinguish between this type of disjunction and those more li­ kely to have been caused by long distance dispersal. The compara­ tive study of the diaspore and dispersal in conjunction with the evolutionary relationships among the taxa is essential for this conclusion.

F. Conclusions

Although Lecythidaceae and Chrysobalanaceae each have a single ba­ sic fruit type, there is a greater variety of fruit morphology in Lecythidaceae. In the Lecythidaceae the evolution of the distinc­ tive generic fruit type parallels the evolution in most genera of extremely distinct floral characters. In Chrysobalanaceae there is a wide variety of floral variation and the characters that define

184 genera are closely correlated with a syndrome, but not with a dispersal type.

Both families have large diaspores that are mostly not well adap­ ted for long distance dispersal. Many of the genera have heavy fruit or seeds which fall to the forest floor. However, as has been shown above, in many cases there are secondary dispersal agents that move the fruits once they reach the forest floor. Al­ though at first sight the rain forests of America and Africa may appear to have many barochorous species, this is misleading becau­ se of the amount of animal activity on the forest floor. More dis­ tant dispersal in the heavy fruited seed types may be selected by the fluctuations of rain forest caused by climate changes in the Pleistocene and earlier eras. In times of forest coalescence rapid colonization is necessary, and rapid transport of diaspores is ad­ vantageous. Many estimates of the possibilities for dispersal in the rain forest are too low. For example, GLEASON & COOK's (1926) estimate that many gravity dispersed trees of the rain forest can disperse only one kilometer in 500 years, assuming good establish­ ment, does not seem likely in view of the amount of animal disper­ sal taking place on the rain forest floor. The heavy fruit of Cou­ roupita guianensis AUBL. observed for three weeks under the parent tree were not touched by rodents or any other foraging animal. On­ ly after three weeks did wild pigs make one of their occasional foraging visits to the trees and ate all the fruits under the tree. Occasional and consequently usually unrecorded events such as this must have enabled the heavy fruited rain forest species to move around.

G. References

BUCHANAN, ~.D., MITTERMEIER, R.A. & VAN ROOSMALEN, M.G.M. (1981): The saki mon­ . keys, Genus Pithecia. - In: COIMBRA-FILHO, A.F. & MITTERMEIER, R.A. (Eds.), Ecology and behavior of Neotropical Primates, : 391-417. (Acad. Bras. de Ciencias). DUCKE, A. (1948): Arvores Amazonicas e sua propagacao. - Bol. Mus. Paraense Hist. natur., 10: 81-92. ' DUKE, J.A. (1969): On tropical tree seedlings I. Seeds, seedlings, systems, sy­ stematics. - Ann. Missouri bot. Gard., 56(2): 125-161. GLEASON, R. & COOK, O.F. (1926): Plant ecology of Puerto Rico. Sci. Surv. Puer­ to Rico Virgin Islands. 6. GOULDING, M. (1980): The fishes and the forest. Los Angeles (Univ. California Press, Berkeley). GREENHALL, A.M. (1965): Sapucaia nut dispersal by greater spear-nosed bats in Trinidad. - Caribbean J. Sci., 5: 167-171. GUPPY, R.B. (1917): Plant, Seeds and Currents in the and Azores. London (Williams & Norgate). HONDA, E.M.S. (1974): Contriburao ao conhecimento de biologia de peixes do Ama­ zonas II. - Alimentacao de tambaqui, Colossoma bidens (SPIX). - Acta Ama­ zonica, 4: 47-53. ' HUBER, J. (1910): Mattas e madeiras amazonicas. - Bol. Mus. Paraense Rist. na­ tur., 6: 91-225. MACEDO, M. &PRANCE, G.T. (1978): Notes on the vegetation of Amazonia II. The dispersal of plants in Amazonia white sand campinas. The campinas as func­ tional islands. - Brittonia, 30: 203-215.

185 PIJL, L. van der (1972): Principles if Dispersal in Higher Plants. 2nd Ed. Ber­ lin & New York (Springer-Verlag). PRANCE, G.T. (1968): Maranthes (Chrysobalanaceae), a new generic record for America. - Brittonia, 20: 203-204. (1973): Phytogeographic support for the theory of Pleistocene forest refu­ ges in the Amazon Basin, based on evidence from distribution patterns in , Chrysobalanaceae, Dichapetalaceae and lecythidaceae. - Acta Amazonica, 3: 5-28. (1981): Comment on: J. HAFFER: Aspects of Neotropical bird speciation. In: NELSON, G. & ROSEN, D. (Eds.), Vicariance Biogeography: A Critique, 395-405. New York (Columbia Univ. Press). (1982): Forest Refuges: Evidence from Woody Angicsperms. - In: PRANCE, G.T. (Ed.), Biological Diversification in the Tropics, : 137-147. New York (Columbia Univ. Press). -- & MaRl, S.A. (1979): Monograph of Lecythidaceae - Part I. - Flora Neotropica, 21: 1-270. New York (The New York Botanical Garden). RIDLEY, H.N. (1930): The Dispersal of Plants Throughout the World. Kent (L. Ree­ ve & Co., Ashford). ROOSMALEN, M.G.M. van, MITTERMEIER, R.A. & MILTON, K. (1981): The bearded sakis, genus Chiropotes. - In: COIMBRA-FILHO, A.F. & MITTERMEIER, R.A. (Eds.), Ecology and behavior of Neotropical Primates, : 419-441. Rio de Janeiro (Acad. Bras. de Ciencias). SMYTHE, N. (1978): The natural history of the Central American Agouti (Dasyproc­ ta punctata). - Smithsonian Contr. Zool., 257: 1-51.

Authors' addresses: Dr. Ghillean T. PRANCE & Dr. Scott A. MaRl, The New York Bo­ tanical Garden, Bronx, New York, 10458, New York.

"\ I 1 I f \

,I

186 \