An evolutionary and ecological perspectiva of Amazonian Hylaea. species of Hymeneae (Leguminosas: Caesalpinioideae)

JEAN H. LANGENHEIM (*) YIN-TSE LEE (*) SUSAN S. MARTIN (*)

ABSTRACT change in the chemistry of a resin through time, but also to investigate a member of a The resin-producing genus Hymenaea has an family which constitutes one of the mos1 arnphi-Atlantic distribution with 13 species being Neotropical and one occurring along the eastern important elements of tropical rainforest coast of Africa. Present evidence suggests an African ecosystems. Ultimately. it is hoped that an origin for the genus with migration across the analysis can be made regarding the role that Atlantic occurring during the Early Tertiary when tropical environmental conditions play in the the continents were closer and rainforest vegetation synthesis of the resin and the possiblc was considerably more widely distributed than today. The center of its Neotropical distribution ecological significance of resins within tropical is the Amazonian hylaea, although within its ecosystems. In both ecological and evolutionary extensive range from 23° N to 26o S it is found in all contexts Hymenaea is particularly interesting major ecosystem types. The taxonomy, breeding because species of the genus appear to have structure, ecology and resin chemistry of the 9 radiated from a rainforest center into a wide Hymenaea species and 6 varieties occurring in thc variety of ecosystems of drier tropical habitats. Amazonian and historically-related Brazilian Atlantic coastal humid evergreen forests are discussed as The genus has 14 presently-recognized well as the possible evolution of these species in species and 12 varieties (lee, 1973). lts range response to postulated drying trends beguen in Mid is amphi-Atlantic, although 13 species are Tertiary times and during wet.dry oscillations in Neotropical with their distribution centered in the Pleistocene. the Amazon Basin but extending from 23° N in central Mexico, throughout the West lndies

!NTRODUCTION and to 26° S in South America, occurring in ali countries except Chile and Uruguay. The one Evolutionary interest in the leguminous African species is distributed along the coast genus Hymenaea (Caesalpinioideae, Detarieae<'l) of eastern Africa (Kenya, Tanzania and Mozam­ was initiated by the discovery that fossil resin bique) and the offshore islands of Zanzibar, (amber) from Mexico, Colombia and Brazil was Madagascar, Mauritius and the Seychelles, and derived from this source (langenheim, 1966, was only recently returned to Hymenaea from 1967). Subsequently a world-wide investigation the genus Trachylobium (langenheim & l ee. of the botanic source of amber in relation 1973). to present-day resin producers led to the On the basis of inflorescence. flower and conclusion that copious production of resin fruit characters, two taxonomic sections of the was predominantly a tropical phenomenon genus may be recognized (langenheim and lee. (langenheim, 1969). Thus focussing study on 1973; lee, 1973; Fig. 1). Sect. Trachylobium is a tropical leguminous genus, such as Hymenaea, characterized by a Jong-paniculate inflorescence provid~s an opportunity not only to analyze the with small flowers and oval to oboval fruit.

( * ) - Division of Natural Sciences. University of California. Santa Cruz, California. ( 1 ) - In prcvious work (Martin, Langenheim and Zavarin, 1972; Langcnhcim, I 973) wc h ave us~d the trihal de,ig­ nation oí Cynometreae Benth. ( 1840) emend. Léonard (1957) . Recenlly Heywood ( 1971) has presented de Candolle's ( 1825) name Detarieae which h as priority over Bentham's name Cynometreae.

-5

In sect. Hymenaea the inflorescence is densely The closeness of relationship of Hymenaea to corymbose with medium to large flowers and these African genera adds evolutionary and generally large rhomboidal to oblong fruits. phytogeographic interest. On the basis of seedling characters, ENVIRONMENTAL •ANO FLORISTIC PATTERNS Hymenaea may be divided into two other OF THE SOUTH AMERICAN HYLAEA groups; these classes segregate the species ecologically into those occurring in the ever­ General characterization of the environ­ green forest (hylaea) habitats and those in the rnental and floristic variation within the drier forest, savannah and semi-desert thorn hylaea is important to our interpretation shrub habitats (Fig. 1). In species primarily of the ecological relations of Hymenaea found in the evergreen forest habitats, there is species and their evolution. " Hylaea" is a cerm a short primary internode between the cotyledon presented by Humboldt (1807, 1814-1825, 18521 attachment and the two primary leaves; in to reter to the South American equatorial contrast, the extra-evergreen forest habitat raínforest but now is more commonly used by species have a relatively long internode. plant geographers to reter to ali equatorial This paper will cover the 9 species of evergreen forests (Aubreville, 1961; Richards. Hymenaea which occur in evergreen rainforests, 1966). The hylaea of South America covers with emphasis placed upon those presently more than 6 million square kilometers, i. e., occurring within the Amazonian hylaea. Thus most of the Amazon Basin, the basins of the most of the species to be discussed fali into upper Orinoco, and drainages in Guyana, Suri· the ecological group which includes sect. nam and French Guiana as well as the basins Trachylobium and part of sect. Hymenaea (Fig. of the lower Tocantins, including the Rio Pará 1 ). Five species (H. oblongifolia varieties and small rivers ot Atlantic drainage eastward oblongifolia Huber and palustris Lee and Lang., to the Rio Pindaré in Maranhão (Fig. 2). H. parvifolia Ducke, H. reticulata Ducke, H. Evergreen forest extends along the eastern adenotricha Ducke and H. intermedia Ducke) coast of Brazil from Pernambuco to Parana; are restricted to the Amazon Basin. Hymenaea however, the area in southern Bahia and oblongifolia var. davisii Lee and Lang. occurs and northern Espírito Santo is considered truly in central Guyana. Hymenaea courbaril L. has hylaea-type vegetation and has been called essentially the distribution of the entire genus "Hylaea Bahiana" (Andrade-Lima, 1953). but variety subsessilis Ducke is restricted to Thus natural limits of the hylaea are set in the central Amazon Basin. Two species (H. ru­ the east by the Atlantic Ocean and in the west briflora Ducke and H. aurea Lee and Lang.) and by the Andes, except for a Pacific coastal H. oblongifolia var. /atifolia Lee and Lang. are extension in Ecuador and Colombia (Fig. 2). lts distributed in the Brazilian Atlantic coastal northern limits in Colombia and Venezuela and rainforests from Pernambuco to Bahia. Hymenaea its southern boundary in Bolívia and central eriogyne Benth. occurs in forest islands in the Brazil appear to be controlled by climatically caatinga of northeastern Brazil. These Atlantic drier areas and hence a gradual transition coastal and northeastern Brazilian species are occurs to drier forest and savannah-type probably relicts from Early Tertiary times when ecosystems (Soares, 1953). In the Amazon the Amazonian-type forest had a more southerly Basin proper the forest is nearly continuous distribution. in the west but areas of open savannah with a non-hylaea flora are scattered in the central Hymenaea has many characters in common and eastern regions. Although our discussi.on with the abundant resin-producing genus will be centered upon species of Hymenaea Guibourtia which occurs most frequently in within the Amazon Basin, they are distributed West Africa but which has outposts in eastern throughout the hylaea. Additionally, an under­ Africa. lt also has been hypothesized that the standing of the evolution of the genus is very extremely diverse, Pantropical genus Cynometra closely tied to the history of both the Soutr may be the parenta! stock for genera such as American and African equatorial rainforests Hymenaea and Guibourtia (Langenheim, 1973). through Tertiary and Pleistocene times.

-7 600 o o v 15 - w 20 "":::> ~ ""

·····················...... ~;~HYLA~~; ::::.·:::::::::::::::::::

Fig. 2 - Vcgetation map of the South American bylaea with insets sbowing monthJy average rainfall and temp~-raturc cocditions for typical stations in different floristic regions. Climatic data obtained from Ministerio da Agricultura do Brasil, Normais Climatológicas (1968) and U.S. Dept. of Commerce, Monthly Climatic Data for the World (Jan. 1956- Dec. 1969). * lndicatcs the location of the weather stations: Iquitos in Peru: Manaus, Amazonas; Belém, Pará; Re- cife, Pernambuco; Ilhéus, Bahia in BraziJ .

Because of lack prominent physiographic Three major precipitation regimes may be differences and similarity in physiognomy of recognized in Amazonia based on the total the hylaea vegetation. the Amazon and adjacent annual amount and seasonality (Sombroek, basins have often been characterized by 1966; Sioli, 1968). Highest annual rainfalls biologists as a "vast uniform habitat" (Haffer, (over 3000 mm annually) occur in the extreme 1969; Vanzolini and Williams, 1970; Vuilleumier, east (Amazonian Estuary) and west (drainages 1971), an impression that has considerably of the Aio Solimões, Japura and middle and influenced evolutionary interpretations. "Uni­ upper Aio Negro). The lower Amazon (between formity of habitat" is, of course, rei ative to the Aio Trombetas and Xingu) receives only whatever group of organisms is being discussed. 1500-2000 mm annually, whereas the central From the viewpoint of a botanist, the differences Amazon (from the Aio Trombetas westerly to in both annual and seasonal precipitation the eastern Solimões) and an eastern strip in patterns and soil types, which are reflected in Pará and Amapá are intermediate with 2000-2500 significant f.loristic and vegetational variability, mm. The eastern and central areas are dispel the idea of uniformity. In fact, emphasis characterized by relatively heavy Mwinter" rains is shifted toward understanding the degree and (December-June) and a relative dry period in nature of the variability. the "summer" (July-November). In the western

8- part of the basin the precipitation is more well foot of the Andes to the Atlantic Ocean distributed throughout the year, although two (Sombroek, 1966). Although the soils were relatively short dry seasons may occur in derived from relatively uniform lithologic certain areas from February to March and June sources (granites and gneisses) from the to August. Guyana Shield • in north a:1d the central In contrast to precipitation regimes, Brazilian Upland in the south, soils vary temperatures and relative humidity are relatively considerably and are a dominant influence constant in Amazonia. The mean annua! in controlling local floras (Ducke and Black, temperature varies from 23.5 to 26. 9°C. The 1953; Sombroek, 1966). The soil types range relative humidity is high with the annual mean from predominantly sandy to some compact varying from 73-94%. In the central part of clay loams, which tend to be both acid and Amazonia the mean values everywhere above infertile, to small areas of fertile man-created 80%. The constancy of this high humidity, "terra preta" soils. even during dry seasons, means lesser transpi­ The soils of the várzea are quite different ration rates than would otherwise be expected from those of the central and lower Amazonian and may be of considerable importance to the terra firme. At the end of the Pleistocene, the development of the vegetation. rise in ocean levei drowned the river valleys of In addition to the importance of the lower Amazonia and the rivers began to fill patterns of precipitation, temperature and with sediments brought from the headwater humidity in determining moisture availability region, forming new floodplains related to the to plants, the degree of flooding is of major altered river levei (Sioli, 1968, 1973). The soils consequence. In the igapó (swamp forest) the are fertile (in contrast to those of the terra soil never dries out, in the várzea the forest is firme) both because of their being derived from periodically inundated, whereas on the dry diverse bedrock types in the Andes and from the upland terra firme the forests are never annual renewal of sediments on them. Despite flooded. An igapó habitat can also occur on more fertile soils, the várzea forest is not as marshy banks of forest streams in an area ri c h in species as is the terra firme forest. otherwise characterized as várzea or terra Along the Lower Amazon the várzea forest is firme. The lowlands of the coastal belt and restricted to the river banks (with the shallow the Amazonian estuary are influenced by tides; várzea lakes being occupied by grassland or the lowest land (daily flooded) is related to "várzea savannah"). Although a great many igapó, whereas higher land (flooded only by várzea forest species occur throughout the spring tides) resembles the várzea of other Amazon Basin, further west one goes the richer areas. Terra firme and várzea are not so in the species the várzeas become. Hera easily separated in the western part ot the vast várzea grassland and savannahs are Amazonia as in the eastern. The western missing and the forest extends into the water várzea includes many islands ot higher, of the várzea lakes and igapó (Sioli, 1968, 1973). rarely flooded "restingas" where typical várzea Also, many species grow in the várzea here forest is mixed with various species that elsewhere grow only on the upland terra usually growing on terra firme. In addition, firme (Ducke and Black, 1953). terra firme of western Amazonia is not .a The hylaea flora is best known along the continuous upland interrupted only by igapó navigable rivers with most upland areas of some streams, as is usual in the more between the rivers being poorly collected. eastern part of the hylaea. More often it Therefore, it is difficult to establish meaningful consists of an undulating terrain where strips phytogeographic regions of the hylaea based of upland alternate with depressions which are upon our present knowledge of the flora . often marshy and flooded by rain water. However, the preliminary zona! characterization Over 98% of the hylaea occurs on terra by Ducke and Black (1953) is useful for firme, which consists in central Amazonia of ecological and evolutionary considerations. deposits laid down in an enormous Pliocene­ They point out that longitude plays a more Pieistocene inland lake that extended from the important role than latitude in composition of

-9 the flora. There thus are three main longi­ the rivers between the Amazon and the tudinally determined zones with latitudinal Essequibo, and the lower Amazonian tributaries. subdivisions (Fig. 3). Although there is some lt is the most vegetationally heterogeneous general correlation of the floristic zones with part of the hylaea, often with the rainforest the three major climatic areas, the boundaries here being substituted by a summer-dry forest are not strongly coincident, indicating that the of lower trees or shrubland resembling the present floristic patterning is due to complex cerrado of central Brazil. In the areas historical factors and edaphic as well as northeast of Obidos and from Monte Alegre to clímatic conditions. Almeirim there are relatively large areas of The Eastern Zone, subdivided into Northeast grassland with scattered trees or densely and Southeast sections, extend westward from ccvered with shrubs. Some of this grass and the estuary to the Rio Xingu and eastern side savannah area may result from agricultura! of the Trombetas basin. The coastal area from co lonies and the continuing grazing of cattle Maranhão northward to French Guiana and (Sioli, 1973). Ducke and Black (1953), however, Surinam is characterized by open campos have pointed out that islands of seasonally dry (grasslands and savannahs) - the most forest in the "hill campos" of Monte Alegre extensive areas lying in the eastern half of contain endemic species not characteristic ot the island of Marajá and in the Territory of the hylaea flora. The Southeastern hylaea Amapá. The Northeast comprises the hill includes the basin of the Tocantins. with little country of the middle and upper courses of know about the flora of the upper courses ot

FLORISTIC ZONES EASTERN CENTRAL o 200 400 600 . WESTERN Km I

Fig. 3 - Floristic zoncs of lhe Amazonian hylaea as describcd by Ducke nnd Black ( 1959). Map drawn as constructed by Sombroek ( 1968) .

10- the smaller tributaries of Pará estuary, both 1953) is distinguished by its great variety ot sides of the Tocantins and that of the woody species representing a genuine, highly headwaters of the Gurupí. Little also is specialized part of the hylaea flora. known of possible affinities of this Tocantins The Southern section of the Central Zone flora with that of the Xingu. Ducke and Black. hylaea includes the whole basin of the Tapajós however, suggest that due to more uniform and the eastern tributaries of the Madeira. This environmental conditions, the flora here may section apparently approaches the Northern not be as rich in species as that of the hylaea in richness of species. A larger number Northeastern section. This Eastern Zone often of species coming from western Amazonia contains the western limits of many species seem to reach their eastern limit here, however. common throughout the coastal region. than in the Northern section. The Central Zone is divided into Northern The Western Zone of the hylaea is an and Southern sections. The Northern hylaea immense plain occupying both sides of the comprises the basin 0f ti.e Rio Negro (except upper Amazon (called Solimões in Brazil, the upper Rio Branco) the basin of the upper Maranón in Peru). The eastern limit of the Orinoco, the eastern part of the Rio Japurá and flora is uncertain, but the flora on both sides the western part of the Trombetas basin. The of the lower Solimões is more allied to flora of southwestern Guyana is related to it. that of the middle than that of the upper The hylaea reaches its maximum diversity in Amazon. The Western hylaea also has caatinga number of genera and species, as well as as does an area of the upper Rio Negro. The endemics, in this area. lt also appears to be Northwestern section, as well as the partly a center of evolution for the Caesalpinioideae, mountainous transition zone to the flora of the with this group being the dominant forest middle Orinoco basin, belongs entirely to element. Ducke and Black (1953) suggest that Colombia. The Southwestern section in Brazi' the floristic diversity in this section is probably is represented by Acre, which includes related to the variety of ecological conditions, numerous southern extra-Amazonian elements greater than any other other part of the hylaea. Ducke and Black ( 1953) point out that the Here one encounters the highest mountains. distribution of many species follows one of such as Serra Curicuriari. The rocky tops of severa I patterns : 1) distributed generally many of these serras are covered with shrubby throughout Amazonia (mainly várzea species), 2) vegetation which unfortunately has not been restricted to the longitudinally defined zones well collected. Also present in the flora of or sections within them, 3) occurring in the the Rio Negro basin are extensive areas of Eastern and Western zones but not in the ·caatinga", shrubs and low trees which occur Central zone. Again these facts tend to on an upland silica soil with acid humus in emphasize the heterogeneity of the environ­ regions with heavy rain throughout the year. mental and resulting floristic and vegetational The variety of species in these caatingas conditions of the hylaea, and again should exceeds that of any other formation in any dispel any concept of a "vast uniform habitat"', part of the hylaea. Almost ali larger trees of at least with regard to the distribution of higher the caatinga also grow in the virgin rainforest plants. of the vicinity. However, the Amazonian The evergreen forest of the Atlantic coastal caatinga has no close affinities with any other plain has much in common climatically and forest type and particularly little with the floristically with that of the Amazonian region. grasslands and open savannahs. "Campinas", lt has been assumed that the Atlantic coastal open spots in the virgin forest on marshy soil forest, extending today from Pernambuco to composed of white sand and black humus, are the region around Rio de Janeiro, was connected also particularly well represented in thi::> with that of the Amazon Basin during at least sectiorf. Floras of clearings are sometimes the early Tertiary. Palentological (Menéndez, called campinas, and large campinas are 1969; Simpson, 1969), floristic (Ducke, 1949; sometimes called campos. However, the flora Andrade-Lima, 1953; Rizzini, 1963, 1967) and of the campina (as defined by Ducke and Black, faunistic (Vanzolini and Williams, 1970) evidence

- 11 supports this view. Floristically, Rizzini (1967) habitats. Hymenaea microcarpa (Huber, 1910) has pointed out that 277 genera occur in both for which no description has been published. Amazonian and Atlantic forests. Usually a is considered synonymous with H. oblongifolia genus which is represented by one or few (Ducke, 1939). species in the Atlantic forest has severa! to Hymenaea oblongifolia var. oblongifo/ia is many in the Amazonian hylaea. Of particular distinguished by its narrowly oblong, gfabrous interest is the region in southern Bahia and leaflets and clawed to subclawed petals. northern Espírito Santo, which has such an This variety has one of the widest ranges of unusually high proportion of Amazonian hylaea the Hymenaea species restricted to the Amazon species that it is called uHylaea Bahiana" and adjacent basins. lts distribution pattern (Andrade-Lima, 1953). Andrade-Lima also felt (Fig. 4) conforms to high annual rainfall areas that the high (2000 mm or more), well­ (more than 2500 mm) in the eastern and distributed rainfall in this area was probably western parts of the hylaea; ít is conspicuously an important factor in maintaining a similarity absent in the central drier areas (annual rainfall in both flora and physiognomy of this hylaea less than 2500 mm). lt is characteristic of with much of the Amazonian hylaea. the várzea alta (its common name being ujutahy de várzea"), but in the Western Zone (Fig. 3) MORPHOLOGIC DISTINCTIONS AND DISTRIBUTION also occurs along streams on terra firme clay OF THE HYLAEA SPECIES OF HYMENAEA soils. As was pointed out previously, however, it is more difficult to distinguish the terra firme We will point out the prominent and várzea habitats in the western Amazon morphologic features that distinguish species than in the eastern. Variety oblongifolia has of Hymenaea and discuss their habitats and been most commonly collected in the eastern distribution as a basis for evolutionary and Amazon where it is relatively frequent on the ecological interpretation. A synoptic key is várzea of the Rio Guamá, the Ilha do Marajá, provided (appendix) to aid in identification of and in ali parts of the Furos de Breves in Pará. these species. lt also has been collected in the Trombetas The earliest work on the Amazonian basin but as yet has not been reported from species of Hymenaea was done by Jacques the southeastern Amazon. lt has been collected Huber (1909), the first director of the Museu over a wide area in the western zone from the Goeldi in Belém, Pará. This beginning was Rio Japurá (Caquetá in Colombia). along the Rio followed by more extensive collection and Solimões, and in the south on the Rio Madeira description by Adolphe Ducke (1935, 1939, and Rio Juruá. Schultes (1953) has pointed 1949), the indefatigable investigator of Brazilian out . that it is associated with Cretaceous legumes who collected both in the Amazon outliers of uthe Venezuela-Guyana land mass". and Atlantic coastal forests. Additional hylaea Hymenaea oblongifolia var. palustris is species have been described from Guyana, the distinguished by the dense, golden-brown caatinga of northeastern Brazil, and the Brazilian tomentose hairs underneath the leaflets, and coastal evergreen forest. by small flowers with almost subsessile petals. lts wood is much harder than that of variety Hymenaea oblongifolia Huber oblongifo/ia. Variety pa/ustris has essentially (Boi. Mus. Paraense Hist. Nat. 5: 386, 1909) the same distribution pattern as variety oblongifolia (Fig. 4), occurring in the same Hymenaea oblongifolia (sect. Trachylobium) high rainfall regions of the extreme eastern is a polytypic species which probably consti­ and western parts of the Amazon Basin, tutes the primitive stock for the evolution of although it is less frequent or at least has Hymenaea in the New World (Langenheim and been less collected. Ecologically, variety Lee, 1973). The four recognized varieties palustris, as its name implies, occurs principally (Lee and Langenheim, 1973) are separated in the swamp forest or igapó (its common morphologically by foliar and floral characters name being "jutahy de igapó"). found along and are isolated ecologically by different the inundated margins of rivers (Fig. 5). lt

12------soe ------9Jo.,------4oo------.~

-+---~-~~~------~~~~~~------+------~------,00

--1------oo

H. oblongifolia ...... _, * VAR . davisi i i • VAR . latifolia " ! o -. VAR. oblongi foi ia .. - ··-·· ·--... ,1' 20 -----20( L e VAR. paI ustris .·· --.., -··--,!,;_ )__"'· 0 ao - 70°- ..,;,_,J_-----"'---"--scf--- ' so0----'---""-··_·_ _..__40° ______J I

Fig . 4 - Distribution of H . oblongifolia and its varieties da visii, /atifolia and palustn s .

occurs particularly in igapós of the western Hymenaea oblongifolia var. latífolia is part of Ilha de Marajá. near Belém, Pará, and distinguished by its broadly-oblong leaflets along rivers in Amapá. In the western zone (having a length : width ratio of about 2 as it has been collected twice along rivers near compared with 3 in the other varieties), and lquitos, Peru. Also it has been reported once by its distinctly clawed petals. Unlike the from the Rio Yurumangui, along the Colombian three varieties just described, variety latifolia coast. This is the only known location for is restricted to the Atlantic coastal evergreen Hymenaea occurring in the hylaea west of the Andes. forests (Fig. 4). Although it is isolated from the Amazonian hylaea, the affinity of variety Hymenaea ob/ongifo/ia var. davisii is latifolía with the other three varieties is distinguished by its narrowly falcate and long­ indicated by similarities in both floral and fruit acuminate leaflets, and its lanceolate and characters. In Pernambuco it has been foound subsessile petals. This variety is known from along the Rio Gurjau, while in southern Bahia only two collections, each from along a river in the drainage of the Essequibo River, Guyana it occurs both along rivers and more commonly (Fig. 4). lt is commonly called "courabarie". in forests on the uplands (Fig. 8). lts common "locust", or "sim i ri", names which may also be names include "jatobá burundanga" and " jatobá applied to H. courbaril. farinheira n.

- 13 (Fig. 7). In the eastern zone it occurs in the coastal areas of Maranhão and Piauí, and in the forests of northeastern Pará. Also it has been collected commonly in campos in the Obidos area near Monte Alegre and Alei rim. lt occurs relatively frequently in the terra firme forests near Santarém and Manaus, along the Rio Madeira, and in the northern part of Ro­ raima. Hymenaea parvifo/ia is most commonly an emergent tree on sandy soil in terra firme forests (Fig. 6); however, it also occurs as small individuais in secondary forests (capo eiros) or in campos.

Hymenaea courbaril, L. (Sp. Pl. 1192, 1753)

Hymenaea courbaril is the most widely distributed species in the genus, having a range almost coincident w ith that of the entire genus. Hymenaea courbaril, morphologically typical of sect. Hymenaea, is thus separated

Fig. 5 - H ymenaea oblongifolia, var. palusrris in riverside igap6, Rio Araguari, Amapá.

Hymenaea parvifolia Huber (Boi. Mus. Paraense Hist. Nat. 5: 385-386, 1909)

Hymenaea parvifolia, the only other Ama­ zonian species belonging to sect. Trachylobium, can be separated from closely related species such as H. oblongifolia by the generally small, short, falcate leaflets; the narrowly oblanceolate and rarely clawed petals which are densely pilose on the upper sides; and the smal l, ovoid, ovoid, often one-seeded fruit. The smallness of the fruit has led to the names • jutaí pororo­ ca· and "jutaí pequeno·. Two names (H. microphylla B. Rod r. and H. pororoca Huber) have appeared in the literature without published descriptions, but are considered synonyms of H. parvifolia (Ducke, 191 5). In contrast to H. ob/ongifolia, H. parvifolia occurs primarily in the drier eastern and central Fig. 6 - Hymenaea parvifolia on sandy terra firme in areas with annual rainfall less than 2500 mm Palhão Reserve near Santarém, Pará.

14- from H. obfongifolia and H. parvifolia by Although they are similar in their faleate inflorescence, floral and fruit characters . leaflets, they may be distinguished from one Within this section H. courbari/ may be another by the length of the stipe and the separated from H. eriogyne and H. aurea shape of their fruits. The fruits of H. courbaril by its glabrous ovary; from H. intermedia and vary (often in the same individual) in size, H. adenotricha by its more faleate leaflets in degree of eompression, and in eolor. In and the large, oblong and eompressed fruits; fact, the fruits from one of our colleetions from H. reticu/ata by its relatively small and from the Palhão Reserve, near Santarém, non-reticulate leaflets; from H. rubrif/ora by Pará, are as large as 21 em x 11 em x 4 .5 em. its glabrous leaflets and the white and ovate the largest known. Generally, fruits of variety petals . Ecologieally H . courbaril displays courbaril are longer and more eompressed remarkable adaptabllity and oeeupies habitats while in variety subsessilis they are shorter ranging from caatinga (annual rainfall less and more eylindrieal. Ecologieal differentiation than 500 mm) to the Amazonian hylaea, is also evident between these two varieties . although it is more frequent in a variety of Hymenaea courbaril var. courbaril is commonly drier eeosystems north and south of the found on terra firme in the Amazonian estuary, Amazon Basin proper (Fig. 9) . the littoral zone in Pará, and oecasionally in Two varieties of H. courbari/ are presently the Central Zone (Fig. 9). In these terra firme reeognized : var. courbari/ and var. subsessilis. sites it is an emergent tree, attaining sueh a

-----60° I r --t--

• H. odenotricho  H. intermedio H. reticu lo to IJ H. aureo • H. rubrifloro 20°- ~ H. eriogyne *~ H. porvifolio L~80° 70° 6t:l ---- Fig. 7 - Distribution o( H . adenotric!la, H . aurea, H . eric gyne, H . intemredia, Tf . parvifolia, H . reticulata , and fi . mbriflora .

- 15 Hymenaea courbaril var. obtusifolia, which was described by Ducke from a specimen collected from a cultivated tree (with seed source probably from Ilha de Marajó) in the Museu Goeldi, Belém, is morphologically so similar to the typical form that we do not now consider it a valid variety.

Hymenaea intermedia Ducke (Arch. Jard. Bot. Rio de Janeiro 3: 92, 1922)

Hymenaea intermedia is distinguished by its intermediate position in leaf and fruit characters between H. courbaril and H. oblongifolia. lt differs from H. oblongifolia by its glabrous ovary and the densely corymbose inflorescence. lt differs from H. courbaril in having much smaller flowers, more oblong leaflets, and a relatively small few-seeded fruit, rhomboidal when fully developed. Hymenaea intermedia is an emergent tree in the primary terra firme forest in the eastern and central Amazon Basin (from the Ilha do Marajó to Manaus), thus occupying the drier habitats characteristic of H. courbaril rather than the moist ones of H. oblongifolia (Fig. 7).

Fig. 8 - Hymenaea oblongifolia var. latifolia in the Bahian Hymenaea adenotricila Ducke uplands near Una. (Buli. Mus. Hist. Nat., Paris, Ser. 2, 4: 727, 1932)

Hymenaea adenotricha has flowers and large size that it is locally called " jutaí grande" fruits similar to those of H. intermedia and can o r "jutaí assu ". lt also occurs, reduced in only be distinguished by the leaflets, which in size, in capoeiras and occasionally in campos. H. adenotricha are broadly expãnded on one The variety subsessilis is more frequent in side and densely pale-tomentose underneath. both the eastern and particularly the central Therefore, this species probably does not Amazon Basin, and occupies a somewhat deserve more than a varietal status of H . wetter habitat, such as sandy beaches along intermedia. This species has been collected rivers (Fig. 9 and 10). Specimens collected only twice, in the Western Zone. lt was from campos near Almeirim appear mor­ described from a single emergent tree from phologically intermediate between the two "mattas das terras altas" near the border of varieties. Amazonas and Peru (Fig. 7). Here Ducke and Black (1953) listed it as a possible endemic Hymenaea multiflora Klein h. has been species. Another collection, however, has reported from only one area in western Surinam subsequently been made along the Rio Curu­ close to Guyana. Amschoff (1939) considered quetê, Amazonas. 1-t. multiflora to be synonymous with H. courbaril. We agree with her viewpoint, Hymenaea reticulata Ducke because tlíe former species may be dis­ (Buli. Mus. Hist. Nat., Paris, Ser. 2, 4: 726, 1932) tinguished from H. courbaril only in having a great density of small flowers in the in­ In his original description, Ducke con­ florescence. sidered Hymenaea reticu/ata as morphologically

16------, 50° 40° I o I [; I I 10° I

,. .· ...... ;~=·:.. ""· ... •· ~._...... I \. \ ) ...... _. j ...... __ .../ I \ .. \ H. courbari I • , ---,. VAR. courbari I • , • V AR. subsessi I is ~ • J 20°- \ /.... .-- ·r·--t: ) .., í I ) \ I \ .. - ··- ·· i--.! '··-··-. L -eo0 ·------700--l..L 600 .

Fig. 9 - Distributíon of H . courburil and its variety subsessilis. similar to H. courbaril, but indicated that it Hymenaea rubriflora Ducke could be distinguished by large leaves with (Mem. lnst. Oswaldo Cruz 51: 457, 1953) densely reticulate venation on both sides. much smaller flowers, narrowly elliptic and Hymenaea rubriflora, the first species we distinctly clawed petals, and the relatively will discuss whose distribution is completely large, oblong-rhomboidal fruits with a smooth, isolated from the Amazon Basin, clearly felt-like surface. On the other hand, H. belongs in sect. Hymenaea. lt does not show reticu/ata appears to have some affinities with close relationships with any particular species. H. rubiflora on the basis of floral characters, although there are some affinities with H . particularly the nature of petals. Until recently reticulata as indicated above. lts uniqueness ff. reticulata had been reported only from a probably results from an early isolation in the restricted area around Manaus, where a few Atlantic coastal forest and subsequent inde­ big trees occur in the terra firme forest on pendent differentiation. lt is distinguished by sandy soil or along margins of streams (Fig. 7). reddish-brown tomentum on the under-surface One collection, however. previously determined of the leaflets, and the compressed fruit with as H. oblongifolia, from near lquitos. Peru , is few seeds. The most distinctive feature, from consid~red by us to be H. reticulata on the which the name is derived, is the red color of basis of both floral and foliar characters. the petals and filaments. This coloration is Thus H. reticulata may occupy a wider range unique in the genus, as ali other species have than h as been thought. white or occasionally pale rose-colored petals

- 17 and filaments. Until recently t his species was Hymenaea species by the dense, golden-yellow known only from the evergreen forest near wooly tomentum on the ovary; the large, flat Recife, Pernambuco. lt has now been collected fruits; the obovate-oblong leaflets which are from Rio Grande do Norte and Espírito Santo densely golden-brown tomentose underneath (Fig. 7). lt thus has the widest distribution of and the relatively large flowers with distinctly Hymenaea species restricted to the Atlantic clawed petals. H. aurea, known locally as "ja­ coastal forest. tobá peloso" and "jatobá verdadeiro", is restricted to upland sites in coastal evergreen forests of southern Bahia (Fig. 7) .

Hymenaea eriogyne Benth. (Mart. Fl. Bras. 15(2): 237, 1870)

Hymenaea eriogyne, found in tree thickets in the caatinga of northeastern Brazil, clearly belongs by inflorescence. flower and fruH characters to sect. Hymenaea. lts Amazonian relationships, evident from its seedling type and pattern of ovary vesture, suggest it is relict from a once expanded hylaea. lt occurs today as a low tree in the caatinga of northern Bahia. southern Piauí, and Ceará.

REPRODUC TIVE BEHAVIOR OF THE HYLAEA SPECIES OF HYMENAEA

Reproductive behavior, including the size of the interbreeding population, the seasons of flowering and leaf drop and their correlation with pollinating agents, and the development of fruit and dispersai of seeds, is closely tied to ecolog ica l conditions and knowledge about it is important in providing background for interpretation of speciation.

THE INTERBREEDING POPULATION

The generally low density of individuais and species within t he hylaea would appear to limit the genetically effective population size. Fig. 10 - llym~naea courbaril var subsessifis along sandy i. e .. the interbreeding population, unless this beacbcs ncar Curuá-Una. Pará dispersion can be overcome by efficient pollinating mechanisms. Hymenaea aurea Lee and Langenheim Species of Hymenaea, and individuais (J. Arnold Arbor. 54(1): 96-98, 1973) within species, are highly dispersed in stands which we have observed in the Amazonian The Atlantic coastal Hymenaea aurea, like and Atlantic coastal forest regions. In drier H. rubriflor8, has no immediate relatives in the ecosystems, such as semideciduous forests genus, perhaps also t he result of early isolation and savannahs, the individuais may be more and subsequent independent differentiation. abundant, even to attaining a clearcut domi­ Hymenaea aurea may be separated from other nance (Langenheim, 1967, 1973). Ouantified

18- sociologic descriptions of local Amazonian the other place are known to be widely communities are rare (Biack, Dobzhansky and d•stributed in the terra firme forests of Pará Pavan, 1950; Pires, and Black, 1953; Takeuchi, and thus the authors of these phyto-sociological 1961; Rodrigues, 1967), but Hymenaea species papers assumed their absence to be acci­ have been mentioned in two studies and dental. Only a few species occurred in generalizations from this work are pertinen~ common to the terra firme and igapó forests to ou r considerations. where the plots were only two kilometers Species of Hymenaea occurred in two of apart; however, the environmental conditions the four Amazonian hylaea p!ots studied by are strikingly different in these two habitats. Black, Dobzhansky and Pavan (1950) and Pires, Thus data both from general observations Dobzhansky and Black (1953). Two individuais during collection of Hymenaea species and a of an unknown species of Hymenaea wer~ few quantified phytosociologic descriptions present in a 3. 5 hectare terra firme plot 120 suggest that efficient pollinating and dispersai Km from Belém (Tres de Outubro), but agents would probably be necessary for the Hymenaea was not found in a one hectare success of such highly dispersed species in was not found in a one hectare terra firme the Amazonian hylaea . plot near Belém. Black, Dobzhansky and Pavan , however, indicated that they had encountered FLOWERING, FRUITING AND LEAF DROP no more than half of the species in the communities they had analyzed with one Knowledge of the flowering period is hectare plots and that most of the rare, and essential to an understanding of pollinating even some of the moderately common species, mechanisms. Often there is an apparent were missing from their plots. Thus the size relationship between flowering and vegetative. of the Belém plot may explain the absence of activity, with flowering generally occurring Hymenaea in an area and habitat where it is when the trees are bare of leaves or at least kr.own to occur. On the other hand, three after some leaf shed (Aivim, 1964). Beard inaíviduals of Hymenaea occurred in a one (i946), however, indicates that among the hectare igapó plot along the Rio Guamá near deciduous trees of Trinidad one half flower Belém. Three individuais of any species was when in leaf and fruit when bare. The othe1 the maximum number found in the one hectare half flower when bare, but one quarter fruit in plots; in fact, one-third of the species occurring the following dry season with the others fruiting in the one hectare plots were represented by in the following wet season. a single individual. Two of the individuais of Unfortunately, few phenological records Hymenaea were identified as H. parvifolia and have been kept for hylaea species. The the third was unidentified. We suspect a information for Hymenaea species (Table 1 J misidentification of H. parvifolia, because in has been compiled from flowering dates given our experience it does not occur in the igapó, on herbarium sheets, general observations, one but instead in terra firme habitats. Nonetheless, phenological study made at the Ducke Reserve these studies point out the high degree of near Manaus from 1962 to 1965 (Araujo, 1970), dispersion of Hymenaea species and of and from unpublished Ducke Reserve records individuais within species in the communities from 1966 to 1971 (Volpato and Schmidt, pers. ot these terra firme and igapó habitats in the comm.). Only general observations are availa­ Eastern Zone . ble for leaf drop, since this is primarily a Also, comparison of data from these 1950 relative phenomenon, with Hymenaea generally and 1953 studies showed considerable differ being considered "evergreen". lhus only in ence in composition between the two Pará certain species and under certain conditions is terra firme plots less than 120 Km distant from leaf drop noteworthy. each ~ther. Of the 87 species found in the The primary flowering time for Hymenaea Belém plot and the 179 species in the Tres de species in the Amazon and adjacent basins Outubre plot, only 54 were in common to both. appears to be from September to November, Some of the speci es recorded in one but not although in some species it either starts a

- 19 1\:) o

Specles Months

• Jaa Feb Mar Apr May June July Aug Sept Od No'Y Dee H. oblongifo/ia F F F F var. oblongifo/ia var. palustris I F F F v ar. latifolia F F F?

var. davissii F*

H. parvifolia F F F F

H. reticulata F F F

H. intermedia L L F F F

H. adenotricha F* F•

H. courbaril F F F F var. courbaril var. subsessí/is I L L I L F F F H. aurea I F I F J I I I I L I L I I I I F H. rubriflora I F? I F I F I F I F l I F I L I L I L I I F H. eriogyne I I F I F I I I I I I L I L I L Table 1 - Time of flowering and leaf drop for hylae11 speeies of Hymenaea. F = flowering. L = leaf drop. * lnformation derived from a single specimen. little earlier or lasts a little longer (Table 1). Jimited data available, it appears that the várzea Hymenaea courbaril var. courbaril, H. adeno­ and igapó varieties of H. oblongifolia flower tricha and H. intermedia seem to begin their during the driest season (October and No­ flowering period earlier than H. parvifo/ia, 11. vember) at Jeast in the Belém area (Fig. 2). In oblongifolia var. oblongifolia and var. pa/ustris, the Western Zone, there are two distinct but H. reticulata and even H. courbaril var. subses­ shorter dry seasons in February and June to si/is. The June flowering time of H. oblongifo­ August (Fig. 2) and thus here these varieties lia var. davisii, known from only one collection of H. oblongifolia flower following the August in Guyana, is aberrant from that of the Ama­ dry season and in the transition to the wet zonian species. season. In the Atlantic coastal forests, H. oblongi­ In the Atlantic coastal forests, H. aurea folia var. /atifo/ia and H. aurea flower in January and H. ob/ongifolia var. latifolia flower in a and February, although the latter species may relatively dry period from December through begin flowering slightly earlier in December . February before the heavy wet period from Hymenaea rubriflora apparently has a much March to July; they do not, however, flower longer period of flowering, beginning in De­ in the driest period from August to September cember and continuing until May in Pernambu­ (Fig. 2). Hymenaea rubriflora in Pernambuco co where most of the collections have been ~tarts toward the end of the dry season in made. Flowering is also recorded in July from December but continues well into the wet one collection of this species from Rio Grande season in May. do Norte. However, in the caatinga region of We have observed fruit production gener­ Piauí, Hymenaea eriogyne flowers during Febru­ ally to be heavy where we have collected. ary and March, just before the onset of the Available fruit collections in herbaria are very short wet season . poor, but we have no way to evaluate whether fruit were not often present or whether the Ducke and Black (1953) indicate from their collector was not able to obtain them. Fruit collecting and general observations in the remain on the tree for at least six to eight Amazon Basin that flowering of trees in the months, but when they fali to the ground, they terra firme forests, where there is a long, often are quickly attacked by microorganisms, heavy rainy season and a relatively dry seasor. rodents or larger mammals such as wild pigs. is restricted to the beginning or end of the dry The Ducke Reserve phenological studies season. In those parts of the hylaea where the indicate that individuais of H. intermedia two set:~sons of the year are not sharply defined studied there flower and fruit on a two-year and often irregular, flowering can be found in cycle. Flowering occurs from August to

every month; in M normal" years. however, October of alternate years with fruit maturing flowers are more abundant during the dry in the following April and May (end of the season, although, tlhe particular schedule varies rainy season). Whether other species follow from species to species. Some do, however, this regime in the Amazon Basin is not known. flower in the rainy season. Phenological Ducke and Black (1953), however, have noted records in the Ducke Reserve study were kept that many trees of the "interior of the virgin only for H. intermedia, although general obser­ forest" do not f lower yearly, but only at vations indicate somewhat similar patterns for intervals of severa I to many years. H. parvifolia and H. reticulata (Volpato and Although Hymenaea is usually considered Schmidt, pers. comm.). These species flowered to be "evergreen ", most species seem to be here from August to November, in the middle facultatively deciduous. Richards (1966) con­ of the driest season and into the transition to siders any tree which is mostly bare, even for the wettest period, January to April (Fig. 2). a few days, to be classified as "deciduous". The ~rlier flowering of H. courbaril var. He further points out that deciduous trees by courbaril and !ater flowering of H. courbaril var. this definition are numerous in ali tropical subsessilis at present cannot be explained in rainforests. These types of deciduous trees terms of climatic patterns. Also from the Gllso almost always occur in the top storey or

-21 are emergent trees in the rainforest. Young POLLINATING MECHANISMS AND FRUIT DISPERSAL plants generally do not Jose their leaves, but leaf drop appears to become more periodic as With the high degree of dispersion of the tree attains maturity. Certainly, if one individuais within the forest, pollinating mecha­ follows Richards' definition of "rainforest nisms become criticaJ. lt has therefore been deciduousness", many species of Hymenae& assumed that many tropical trees may be self would be considered to be deciduous rather fertile but, may also outcross when pollinators than evergreen. The trees may go through a are available. Wind pollination appears to b~ heavy le: f fali during dry periods and may a rare phenomenon in the rainforest, thus the even Jose ali of their leaves under severe activity of insects and bats is likely to be of moisture stress (Wilson, pers. comm.). Again, considerable significance. Araujo' s ( 1970) phenological data for H. inter­ Like most other phenomena, pollinating media (with general observations also for H. mechanisms have been little studied In parvifolia and H. reticulata) from the DuckA Hymenaea. Lee (1973) has pointed out that Reserve give us important leads to behavioral the inflorescence and floral structure of activity. Apparently leaf fali varies with the species within the two sections may suggest flowering-fruiting cycle. In years of flowering. different major pollinating agents. The flowers leaf fali immediately precedes flowering; in of members of sect. Trachylobium are small, years of fruiting, leaf fali occurs when flowering and are located at the edges of a conical would normally. In both cases, however, the panicle with long, flexuous branches; these leaves fali during the dry period from June characters suggest pollination by small-sized through September. Although many leaves social insects which exhibit "opportunist"-type fali each year, these trees are rarely com· behavior, as described by Janzen (1971b) and pletely denuded because of periodic flushing. Baker (1973). In contrast, the flowers ot The possible exception is H. courbaril var. species in sect. Hymenaea are larger and subsessi/is, essentially restricted to the sandy fewer, reach about the same levei in anthesis, beaches along the central Amazon, which and are borne on short and strongly flexuous apparently does become almost entirely de­ branches. These characteristics indicate the possibility of pollination predominantly by bats foliated from August to September. Also, and larger, solitary insects with "trap-lining" Hymenaea eriogyne in the northeastern Brazi· behavior (Janzen, 1971b; Baker, 1973). lian caatinga was completely denuded when observed during November. lnformation on floral behavior and polli­ nation is presently only available from several We have not observed leaf drop in the populations of H. courbaril var. courbaril. Amazonian varieties of H. ob/ongifo/ia. In Anthesis is nocturnal and Vogel (1968) has addition, in the Atlantic coastal forest, H. indicated that only one to five flowers in an rubriflora and H. aurea clearly Jose a great inflorescence open at any particular time. Vogel many of their leaves during the dry season, but (1968) and Frankie (pers. comm.) observed that during these periods H. oblongifolia var. latifolia flowers usually open at or just after sunset, still possesses essentially a full complement between 5 and 9 p.m., and Jose their calyx of leaves. corolla and stamens by the following noon. Although the individual flowers are short-lived, lt is unfortunate that so little data are the flowering period of a particular tree is available on leaf drop, for it appears that some continuous over several months. The flowers species of Hymenaea behave differently from of H. courbaril have a strong odor and a others. Leaf fali data might not only be useful massive nectary. This floral behavior also in interpretating present moisture stress provides characteristics which favor bat conditions but possibly in understanding pollination. In fact, bat visitation has been spec1es which had evolved mechanisms fo• observed in H. courbaril by Carvalho (1960, adjusting to dry oscillations during the 1961), Vogel (1968) and Frankie (pers. comm.J. Pleistocene. Carvalho observed bat visitation (Giossophaga

22- soricina) of H. courbarif var. courbaril near would presently assume that small insect Belém, whereas Vogel studied their visitation pollination would predominate only in such (Phyllostomus disco!or) shortly after sunset species as H. ob/ongifo/ia and H. parvifolia with to H. courbaril var. subsessilis near Ma· bat pollination being more important in the naus. Although Vogel also observed that species within séct. Hymenaea. However, too hummingbirds and bees visited Hymenaeél little in-i'ormation on pollinating agents is pre­ flowers before sunset, and sphinx and night sently available to allow thinking in terms of moths after sunset, he thought that bats were the potentialities of phenological synchro­ probably the most efficient transmitters of nization with insect pollination as suggested pollen. Frankie has recorded visitation of by Janzen (1969). Likewise, Alvim (1964) has unidentified bats to H. courbaril var. courbaril clearly indicated the need for further physio­ in the seasonally dry forests of Costa Rica. logical studies of the effects of total radiation, Bawa (pers. comm.) has also found H. courbaril changes in radiation quality, thermoperiod, and var. courbaril to be self-incompatible in the moisture stress on the growth periodicities of same Costa Rican ecosystem in which Frankie tropical trees. lt seems possible that both has studied bat pollination. At present no other physical environmental factors as well as information is available regarding compatibility pollinators may be involved in this synchro- in H. courbari! in other ecosystem types nor nization. _L for any other species, and it seems improbable Dispersai of the indehiscent pods occurs that the data for this one species within a commonly by water. The pods float easily single kind of ecosystem should be extrapolated and apparently in salt water, at least, remain for other species and ecosystems. intact for long periods of time before being Early-morning visitation of insects to attacked by microorganisms. Water dispersai flowers of H. oblongifolia var. oblongifolia ir. probably explains the relatively frequent distri· the Guamá Reserve near Belém has been bution of Hymenaea along streams. Also recorded from observation towers built in the occurrence of H. courbaril on islands in the forest. Many bees and wasps have been West lndies and of H. verrucosa on African collected but as yet are unidentified. Lee offshore islands is most likely due to water (1973) has also noted visitation of bees and transport of the fruits, although Man may have butterflies to the flowers of H. ob/ongifolia var. aided in its distribution in some cases. On palustris and var. /atifolia. land, rodents and wild pigs have been observed Janzen (1967) has pointed out that the to disperse the pods. coincidence of leaf shed with the peak of Seed predation is a serious problen in tro­ flowering of many trees and maximum activity pical lowland ecosystems. Janzen (1970) has of pollinating insects during the dry season in even suggested that the low densities of many the forests of Central Amercia leads to an tropical tree species, with long distances extremely efficient pollinating system. He between conspecific adults (particularly com­ further hypothesizes that this relationship pared with temperate-zone forests), result from between loss of leaves and peak flowering the action of predators on seds and seedlings. results from selection for the occurrence of A considerable amount of work has been pollination during the dry season at the most dane regarding seed predation on leguminous opportune time of the year for the activity of species, among which has been H. courbaril insects such as bees, rather than from physi­ (Janzen, 1969, 1971a). Janzen has indicated ological processes that could only occur at that populations from Puerto Rico and Costa that time of year. Once a tree species has Rica are free from bruchid beetle attack evolved such a periodicity, it would be difficult (common to many leguminous trees) but for it to break out of the pattern because of its severa! species of weevils of the genus leaving its pollinator behind and thus disrupting Rhinochenus oviposit on the pods of Costa the reproductive process. Although leaf fali Rican populations. Severa! mechanisms may generally precedes flowering in the hylaea, be used by the trees to avoid such predation. where records of leaf fali are available, we First, trees in relatively dense, seasonally dry

-23 forest sites in Costa Rica appear to have a mental conditions, with H. courbaril, however different reproductive pattern from those always displaying the most rapid growth. Mos1 occurring in open sites. Those in the dense species grow relatively rapidly immediately forest habitats have a severa! year (3-5) fruiting following germinati~n. then go into a slow period whereas those in the open habitats may growth stage which lasts differing periods fruit in consecutive years. He suggests that dependent upon species, before continuing the adaptive significance of sexual dormancy more rapid activity. Rapid germination and an is in strongly reducing seed predation by ability of seedlings to remain alive for a long Rhinochenus beetles (Janzen 1970, 1971a) time in slow-growing condition have been This behavior of the tree means to the weevil pointed out as providing survival value to that seeds are three to five times as sparse species having to cope with abundant seed as would be indicated by the total density of predators of ali types (fungi. bacteria and the adult trees. Also, where this curculenoid various kinds of animais) under tropical warm weevil is absent (EI Salvador north through and humid conditions (Richards, 1952). southern Mexico and in Puerto Rico). the tree usually fruits every year. The second mechanism of protection against weevil attack RESIN PRODUCTION OF THE HYLAEA SPECIES probably is associated with the abundant resin OF HYMENAEA produced in the walls of the pods. Rehr, Bell, Copious production of resin by the hylaea Janzen and Feeny ( 1973) h ave shown that th e species of Hymenaea permitted its being used seeds of H. courbaril from Costa Rican and commercially, especially for hard varnishes, Puerto Rican populations do not contain either and to a limited extent for medicinal purposes. uncommon amino acids or alkaloids that Commercial collectors valued the resin, referred provide protection for some tropical leguminous to in the trade literature as "Brazilian copa I", sceds. Therefore, they conclude that the resins and collected it particularly in the eastern in the pod walls can provide an alternativa Amazon Basin vvhere certaln local populations strategy to seed toxins by physically forcing vvere noted for exceptional resin yields. Tht: out the weevil as it attempts oviposition . factors determining resin yield have as yet not From our work on the chemistry of the resins, been fully determined, but it appears both it would seem theoretically possible that a genetic and environmental tactors are involveo. number of either chemical or physical factors Although the resin can probably be collected relating to changes in resin composition could from many areas in the Amazon Basin, the deter the activities of these beetles. great collection center seems to have been ...! ~ From our greenhouse experiments we around the Ilha do Marajá near Belém and along have discovered that seeds can remain viabiP tributaries such as the Xingu and Tapajós. for at least eight years and probably longer. Hymenaea courbaril has generally been con· For the first severai years, the germination sidered to be the most abundant resin producet percentage is high in ali of the species with of the genus, but it now seems probable that which we have worked. When the seeds are resin was collected and utilized as • Braztl older, the percentage of those which germinate copa!" from species such as H. ob/ongifolia or decreases slightly. Scarified seeds usually 1-1. parvifolia which today at least are more germinate decreases slightly. Scarified seeds common in those areas than H. courbaril. usually germinate within 10 days under Resin pieces varying in size from that ot a pea favorable !11Pisture and temperature conditions to that of a man's toot were often recovered in the greenhouse or in growth chambers. from streams (Howes, 1949). Record and Hess Growth rates vary considerably for differem (1943) indicate that "gatherers dig around the species when grown under uniform environ· roots and sometimes obtain a barrelful of copal

24- lumps in a place, while the former site of a population. A list of compounds with an tree, long since decayed, may yield severa! identification number for each sesquiterpene, barreis of so-called fossil copa!". There is no to be used in the figures and discussion to indication that the trees were tapped for the follow, is given in Table 2 . copal, nor were the legumnious Congo and Zanzibar copals obtained by tapping. Table 2. Leaf pocket resin sesquiterpenes in Hymenaea. Hymenaea secretes resin by two anatomical ------mechanisms, either into schizogenously pro­ Compound number Sesquiterpene duced pockets surrounded by an epitheliai layer of secretory cells. or into lysigenously 1 a - cubebene produced cavities resulting from the breakdown 2 a - copaene of the walls of the secretory cells (Langenheim, 3 unidentified 1967). The schizogenous pockets form in 4 13 - copaene parenchyma tissue in leaves, young stems, 5 caryophyllene floral parts and fruits. In these pockets, 7 13 - humulene primarily sesquiterpene hydrocarbons and a 8 Y - muurolene few oxygenated sesquiterpenes are present. 9 a + 13 - selinene Resin does not appear to be synthesized in 10 unidentified young roots, but as the root and stem develop 11 b - cadinene secondary tissue, resin COJTlpOsed primariJy of 12 Y - cadinene diterpenoids with some sesquiterpenes is 13 unidentified produced. The cavities into which the viscous 14 uni dentifi ed resin is secreted may be continually increased by the lysigenous breakdown of the secretory Table 2 - Lcaf pockci r ~sin sesquiterpenes in Hymenaea cells, allowing the accumulation of Iarge amounts. When a natural or manmade break The remarkable consistency in leaf pocket in the bark occurs, the material may be exuded resin chemistry among the Hymenaea species to the exterior of the tree. In pods, resin is in contrast to both qualitative and quanü­ consisting mainly of diterpenoids and some tative differences found between Hymenaea sesquiterpenoids is produced, although its and such closely related genera as Daniel/ia composition is quite different from that ot and Guibourtia (Martin and Langenheim, unpub. trunk resin. The pod resin is secreted initially data). into schizogenous pockets, but these can be Within this set of sesquiterpenes common enlarged by lysigeny when inj ury occurs to the to the species of Hymenaea, quantitative fruit. Thus the pods trequently are covered by composition of their leaf pocket resins shows masses of - rE:?sin, often filling and exuded specific differences (Figure 11). The experi­ through chann~Is resulting from attempts by mental plants for this study were germinated Rhinochenus beetles to penetrate the fruit. and grown in our greenhouses under essentially identical environmental conditions. Ouantitative The leaf pocket resin of ali Hymenaea ::nalysis were made by gas chromatography of species we have examined is comprised o1 ether extracts of fresh leaves, as described essentially the same set of sesquiterpenes. elsewhere (Martin, Langenheim and Zavarin, about a dozen of significance (1% or more 1973). We have analyzed these quantitative of the total). Caryophyllene and a - and leaf pocket resin composition of Amazonian 13 - selinene are usually the predominant Hymenaea (Figure 11) for the relationships compounds. We report the latter two com­ they suggest. We wish to emphasize. however. pound~ together, as a + @- selinene, because that these data are based on seedlings grown there appears to be no significance to the from seed originating from a single geographk: particular isomer present, wide variations in locality. The leaf pocket resin compositior. their proportions occurring within a given data should, therefore, be considered repre·

- 25 H· RUBRIFLCRA H· RETICLLAT A PERNAMBUCO , OJCKE RESERVE AM · , BRAZIL

H· PARVIFOLIA H· 08l.ONGIFOLIA OJCKE RESERVE GUAMÁ RESERVE , MANAUS, AM ·, BRAZIL BELÉM , PARÁ, BRAZIL

3 4 5 6 7 B 9 10 1.1121.3 14 1S SES

H· COURBARIL H· COJRBARIL BELÉM, BRAZIL VAR · SUBSESSILIS MANAUS , AM · , 8RAZIL

~3 ~2

.. ~ 1~::::1=jLLJL.dLLJLLib::.~

Figo li - Leaf pockel rcsin compositions of Amazonian spccies of Hymenaea o Each graph shows thc m ~an of eigr,t samples exccpt H o reticu/ata (5 samples) and H o rubri/lora (3 samples) o Onco standard deviation above and below the mean is showo for caryophyll~ne and oc+ 13 selineoeo sentative of the local population sampled rather five and three differences respectively belng than of an entire species. significant. Full details of the statistical analysis of leaf Further analysis of quantitative leaf pocket pocket resin composition among Amazonian resin composition of these Amazonian popu· and other Hymenaea species have been lations of Hymenaea has been made by described elsewhere (Martin, 1973). To summa­ treating the data by the methods of numerical rize briefly, an analysis of variance was per­ taxonomy to obtain coefficients of association formed for each sesquiterpene, excluding those or similarity (Sokal and Sneath, 1963). Analysis present in trace amounts in ali species. Th ~ of variance weighted similarity coefficients ten compounds showing highly significant were obtained and clustered by the single (p = O. 01) F tests in the analysis of variance linkage method in arder to more readi!y were further examined by Duncan's new visualize the chemical relationship among the multiple range test to determine which species Amazonian Hymenaea species and populations or population means were highly significantly (Martin, 1973). In this way, the three ge­ different from one another (Steel and Torrie, ographic populations of H. courbaril cluster 1960). Th is analysis indicated that no single together. There appears to be little difference species of the Amazonian Hymenaea is com­ in the grouping of the other species, although pletely quantitatively different from any othet H. rubriffora . and H. parvifolia form a less in every compound; the underlying qualitative well-defined group. H. oblongifo/ia and H . similarity has already been emphasized. reticulata stand apart in this analysis. Hymenaea reticulata is chemically the most We have previously suggested that Hyme· unusual, differing from H. oblongifolia [(and menaea oblongifolia may be the primitive stock from the two African geographic populations from which many species have evolved . of Hymenaea verrucosa examined (Martin, Because the leaf pocket resin chemistry of the Langenheim and Zavarin, 1973)] in quantitative Amazonian species of Hymenaea is qualitatively contents of six o f the then compounds. In alike, one cannot discern evolutionary lines; contrast, it differs significantly from the Ma­ clearly, however, the chemical data do not naus populations of H. parvifolia and H . disallow the possibility that Hymenaea oblongi­ courbaril var. subsessilis in content of only folia could be the parenta! stock. Alterations three of the ten sesquiterpenes. in quantitative resin compositlon might have The two geographic populations of H. accompanied processes of differentiation into courbaril (Belém and Curuá-Una) differ signifi­ specific habitats, leading to the populational cantly only in their content of the minar and specific differences found today. compound 3. The Belém population differs Trunk resins, usually found as solidified from H. courbaril var. subsessí/ís in content of masses on the exterior of the tree, were the minar compounds 3 and 10, while Curuá-Una point of our initial interest in Hymenaea. lt H. courbaril differs from variety subsessilis only was through infrared spectral comparison of in compound 10 content. The close relationship Hymenaea trunk resins with amber from Me­ among these three populations of H. courbaril xico, Colombia and Brazi l that their amazing is therefore clearly demonstrated in their leaf similarity was recognized, and Hymenaea es­ pocket resln chemistry. tablished as the source of the fossilized resin H. oblongifolia differs significantly from (Langenheim, 1966, 1967, 1969). Ali of the H. courbaril var. subsessilis only in content of Hymenaea trunk resins consist primarly of compound 13. Traces of this compound are diterpenoids, with some sesquiterpenoids an C: present in the other species, but in H . small amounts of non-terpenoid material. The oblongifolia constitutes 3. 6% of the total diterpenoids include a series of compounds sesquiterpenes. More differences are show by having basically similar skeletons, but repre H obltJngifolia it constitutes 3. 6% of the total senting various oxidation stages : hydrocarbons, sesquiterpenes. More differences are shown aldehydes, alcohols and acids. We have in­ by H. oblongifolia in comparison with the Belém vestigated primarily the resin acids, major and Curuá-Una populations of H. courbaril, with components of the resin (20 to 38 % in our

-27 samples) which are of special interest because lhrough long periods of time is clearly reflected they are potentially useful for chemosystematic by the underlying similarity in resin components purposes (Harborne, 1971). among species of the genus as we know Prior to our beginning work on Hymenaea it today. trunk resins, two other groups had investigated resins of specles w ithin the genus. Nakano EVOLUTIONARY HISTORY OF THE HYLAEA and Djerassi ( 1961) described from a com­ SPECIES OF HYMENAEA marcial "Brazil copal" sample the bicyclic Evidence from severa( sources supports diterpenoid copalic acid and severa( double an Africa origin for the genus Hymenaea. bond isomers. Another group identified eight despite its primary distribution today being resin acids and severa( neutra( fraction com­ tJeotropical. First, its closest relatives are the pounds from the trunk resin of Hymenaea African genus Guibourtia and the Pantropical verrucosa, then known as Trachylobium verru­ Cynometra, which has its center of diversity cosum (Hugel, Oehlschalager and Ourisson. in Africa. Hymenaea has no evident elos e 1966). Excluding one sesquiterpenoid, ali of atfinities with Neotropical genera; a close these compounds were based on the bicyclic relationship to Peltogyne has been previously enantio-labdane skeleton. Subsequently, we suggested, but Lee (1973) shows this relation­ have characterized from other species of ship to be relatively distant. Secondly, Brenan Hymenaea seven enantio-labdane type resiP ( 1965) has pointed out that 67% of the Cae­ acids (Cunningham, Martin and Langenheim salpinioideae are endemic to Africa. the most 1973a and unpub. data), and one resin prominent components belonging within the c:cid having the normal labdane skeleton Amherstieae-Cynometreae (Detarieae) trib!!!l (Cunningham, Martin and Langenheim. 1973b) complex. Ducke and Black (1953) have further Through gas chromatography of resin acid stated that t he Caesalpinioideae have their methyl esters, we have found that most of secondary center of diversity in the Amazon t hese resin acids are common to severa! Hyme­ Basin. Unfortunately, there is no accredited r.aea species, although proportions of com­ evidence for this tribal complex in the geologi< pounds vary enormously. Unfortunately, quanti­ record until the Paleocene Epoch (Germeraad, tativa comparisons of resin acids in trunk Hopping and Muller, 1968; Muller, 1970) . About samples is not justified for the following reason. 75% of the African Caesalpinioideae are re­ Trunk resin samples as normally obtained stricted to evergreen rainforest habitats. and consist of solidified masses collected from the the sub-family likewise reaches its optimum exterior of the tree, where they have been development in the Neotropics in the Amazo· exposed to the atmosphere for unknown lengths nian hylaea. During the Eocene Epoch, tropical of time. lt is known that upon exudation, trunk rainforest vegetation reached its maximum de­ resin undergoes progressiva oxidation and poly­ ve lopment, probably attaining almost double merization; the exact nature and rates of these the area of its present distribution (Wolfe. processes, their importance under varying 1971). Thus an Early Tertiary speciation (if environmental conditions, and the consequent not origin) of the Caesalpinioideae seems likely potential compositional change in trunk resins and conditions during the Eocene were probably with time are not known. This difficulty excellent for dispersai. With increased popu­ effective prohibits any detailed quantitativa lation size of the rainforest species, a greater comparison of the trunk resins of the various number ot d1sseminules would be suppl1ed fo1 Hymenaea species, although such difference:> dispersai, while at the same time the area are almost certainly present. suit9ble for colonization was larger. Our increasing knowledge of trunk resins 1-Jymenaea verrucosa in eastern Africa of Hymenaea species has the interesting appears to be a relict of a once more con­ potential of 6oon allowing chemical comparison tinuous distnbution in evergreen rainforests with the fossil resins known to onginate frcw 2cross Afrlca, and the clear lies between the Hymenaea. The stabllity ot 3 tunae~mental African and Amazonlan spec1es (members of structural pattern In the resin chemistry sect. Trachyfobium) is one of the most im-

28 - portant factors in considerations of the ongm time and during dry oscillations of the Pleis­ o f the genus Hymenaea. A possible explanation ~ocene. for both the origin of the genus and the amphi­ Paleontologic data suggest that rainforest Atlantic distribution pattern is an assumed flora (Archangelsky, 1968) and associated common West Africa rainforest ancestral stock fc:unas durlng the • Eocene were as far soulh as (Langenheim and Lee, 1973) . Two explanations central Patagonia on the South American cont!· may then be presented for the Neotropic distri­ nent. Menéndez ( 1969) and Simpson ( 1969) bution : 1) un1on of the South American and present evidence from both fossil plants and African continents, or 2) oceanic dispersai mammals which indicates that strong climatic Although it is assumed that Africa and change occu:·red in South America during the South America were united, geologic evidence Cenozoic, similar in mirrar image to that better indicates that drifting of the two continents established in North America and other northern took place too early (around the Jurassic­ continents. Early in the Cenozoic Era, climatic Cretaceous boundary) to explain the disjunct zones were less marked than today. Typicaliy distribution of the New World and Afrícan tropical fauna, such as crocodilians and ce(tain species of Hymenaea or their ímmediate large snakes, occurred in areas which now &re ancestors. Even if drifting occurred during the temperate. In Late Oligocene there were also arboreal mammals of subtropical to tropical middle of the Cretaceous, as suggesteó by kinds, monkeys, and somewhat later anteaters Veevers, Jones and Talent (1971J and Raven and tree sloths far south of their present and Axelrod (1972), it is highly problematic distribution. By the Miocene Epoch, however, whether any living genera of angiosperms the climate was on the way to drier and cooler existed at that lime. Thorne (1973) also points conditions and that change, with Pleistocene out that the "tenous links" between tropical fluctuations, has continued to the present. Africa and trO!JlCal America and the stronger Although the presence of rainforest vege­ floristic relationships of tropical Africa and tation in southern South America in the Early South America to other continents rufe out Tertiary Period and its retreat northward !ater continental drift as a valid explanat10n for in the Tertiary is generally accepted, it also has amphi-Attantic disjunctions. In a careful analy­ commonly been assumed that the Amazonian sis of the taxa occurring on the two continents, hylaea probably has remained relatively unaf­ he believes that the distribution patterns indi fected by climatic change. In fact, the cate occasional long-distance oceanic dispersai enormous diversity of animal species in equa­ rather than retention of ancient wide ranges torial rainforests has often been explained on split by continental drift. However. migration the basis of climatic stability through time was probably easier between the continents (Darlington, 1957; Schwabe, 1969). In other during late Cretaceous and early Tertiary time words, the assumption is that communities will because long .after the two continents were tend to diversify with passage of time and thus separated there was fairly direct access across older communities will contain more species relatively nar~ow seas and possibly along than younger ones. lt also h as been assumed volcanic islands on mid-ocean ridges. Thus it that stable climates will permit the evolution seems more probsble that oceanic dispersai of finer adaptations, with finer specialization could have occurred from the African ancestral allow:!lg more different kinds of organisms to stock during the Early Tertiary to rainforests of be accommodated, i. e., greater niche differ­ the New World - one sucessful establishmen1 rentiation (McArthur, 1969; Baker ,1970). leading to the Cuban species H. torrei and Relatively recent evidence from a variety another to the differentiation of the hylaea of sources, however, has suggested the pos­ species centering around H. obongifolia . sibility that the Amazonian hylaea may also Subseqtfently the rainforest species in the Nt::w have been disrupted by the drying tends !nitia­ World have provided the stock for species ted during the Miocene Epoch and by subse­ which radiated into the drier ecosystems in quent wet-dry oseillations during the Pleistoce­ drying trends during Middle to Late Tertiary ne. Sombroek ( 1966) has summarized geo-

- 29 morphological and soils data from Barbosa and for birds and lizards that the areas of refuge Ramos (1950), Wilhelmy (1952) and Ab'Saber are peripheral to Amazon Basin and determined ( 1959, 1967) which indicate that semi-arid by topographic features. In other words, both climates occurred during the Miocene and Plio­ Vanzolini and Haffer emphasize the repeated, cene, probably resulting in "open vegetation relatively rapid fluctuation of hyl~ea between types" in the area now occupied by hylaea. occupancy of vast areas to reduction to a few Also the hylaea was greatly restricted during isolated peripheral patches. In fact, this the Plio-Pieistocene, when a great lake occupied peripheral location and rapidity of climatic most of the central area of the basin from the changes appear to them to explain the great Andes to the Atlantic mouth of the river. diversity of the Amazonian biota. Each new Sombroek states that, during the deposition of c:oalescence of expanding refugia would provide the Pleistocene terrace materiais, the Amazon a complex mixing of populations that had been valley must have had a long and pronounced dry differentiated during isolatíon . season which limited the growth of dense vege­ The correlations in time of the humid-arid tation, and a short and intense rainy season cycles with South American glacial events which created strong erosional capabilities . In remain controversial. Presently, few logical the present West African rainforests, convincing data confirm this hypothesis, although Vuilleu­ geologic evidence has been found for the: mier (1971) suggests that future work may well presence of Kalihari sands, and hence a rela­ document this synchronization. Paleobotanícal tively arid climate, both in Middle Pleistocene evidence also is not available from Amazonia; (75,000-55,000 years ago) and later (22,000 years however, palynological data from northern ago) . This indication of aridity in the African South America (van der Hammen, 1961) indi­ equatorial zone during the Pleistocene (Moreau, cate repeated changes over large areas in the 1966) adds to the credibility of such a phe­ Pleistocene that do correlata with glacial nomenon occurring in the Amazonian area. events. Recent studies on bird, anolid lizard and Only animal populations have been con­ lepidopteran speciation in the hylaea have sidered in these recent discussions of Pleito­ resulted in an hypothesis that climatic oscll· cene climatic changes affecting the Amazonian lations during the Pleistocene had dramatic hylaea. However, wet-dry Pleistocene oscil­ effects upon vegetational habitats of these lations and their influence on vegetational organisms (Haffer, 1969; Vanzolini and Williams, changes in equatorial Africa have been di s­ 1970; Vuilleumier, 1971; Brown, 1972; Brown cussed by Aubreville (1962). Also, Stebbins and Mielke, 1972; Vanozlini, 1973). The ( 1952) has presented a speciation model for speciation model assumes at least three alter­ higher plants which is applícable to the Ama­ nations of wet and dry periods during the last zonian situation. H e has hypothesized that the 100,000 years. During the dry phases it is most rapid evolution of higher plants should hypothesized that dry-type ecosystems dis­ take place in habitats which are changing and rupted the hylaea, reducing it to isolated those in which some factor is limiting . He patches where conditions were relatively more thus thinks that many species originate under favorable than over the major portion of the limitation of moisture and then may become basin. For the forest fauna these areas acted readapted to moist conditions under influence as refuges in which populations were isolated of some climatic changes. In regions isolated and consequently became differentiated. In from each other, the xerophytes continue their subsequent wet phases, the formerly isolated divergent evolution. In this way, the flora of animal populations followed the spread of the a mesic region could cont.::)in members of the forest. W h e r e the populations met in same genus or family, ali adapted to essentia! ly reforested areas, complex character variation similar conditions but adapted in different reflected the degree of genetic divergence ways, and hence distinguished from each other reached during the period of isolation. Vanzo­ because of different evolutionary histories. lini (1973) points out that it is essential to the These ideas provide a speciation model for arguments presented for the specíation model higher plant5 similar to those presented for the

30- animais under wet-dry oscillations, and conse­ H. verrucosa, and we assume that the two quently expansion and contraction of rainforest species arose from a common West African and drier ecosystem types. The present floristic stock (Langenheim and Lee, 1973). We also diversity in the Amazonian hylaea, wjth re la­ have hypothesized that this species probably tively large patches of campos and caatingas became established during the Early Tertiary comprised of natural, non-hylaea type vege­ when 1) rainforest vegetation was most tation, campinas of specialized hylaea flora, widespread and the probabilities for dispersai areas of seasonally dry forests, and shrubby were greatest, and 2) the continents were closer vegetation capping low ·serras", suggest a together than today. In other words, we are complex history of dry climates in the past as assuming that H. ob/ongifolia was an early well as the diversity of climatic and edaphic inhabitant of the Amazonian-type hylaea . lt conditions today. In addition, Ducke and seems reasonable that disseminules of the Black (1953) have emphasized that the presence present stock H. oblongifolia arrived from West of the same species of plants in severa! or Africa either in the region of the Amazon Basin many campos or campinas, whích are separated via the South Equatorial Current or to a more by extensive areas of hylaea, implies the southerly site via the Brazil Current_ In either "remate origin" of these vegetation types . case, the primary distribution of the species In the evolution of Hymenaea species there today is represented by variety oblongilolia in appears to have been adaptive radiation from the várzeas of the eastern and western zones humid evergreen forest to a variety of drier of the Amazon Basin. The variety pa/ustris ecosystem types over a wide geographic range. has a similar distribution to variety ob/ongifolia This differentiation was probably initiated but has different lated into the igapó habitat. during the Mid Tertiary drying trend and perhaps The occurrence of H. oblongifolia in these continued during periods of oscillation between habitats might lend additional support to the wet and dry conditions during the Pleistocene. hypothesis of its early appearance in the Ama­ In the South American hylaea, Hymeanea zon, as the várzea and igapó are likely to species also are ecologically differentiated into r.; present the oldest habitat types, existing as severa! of the drier local vegetation types. long as the river itself. The plants inhabiting One can then speculate whether or not their them would also not be forced to tolerate the distribution and morphological variability give effects of drying conditions as much as those evidence for past climatic changes in the on the terra firme. Consequently, the oc­ hylaea. rence of H. oblongifolia In the complex network Species in the "Hylaea Bahia na", such as of igapó, várzea and terra firme In the Western H . aurea, were probably initially isolated by Zone could have provlded an ideal situation for Mid Tertiary drying trends ,although therf' may speciatlon from wet to progressively drier have b3en additional modifications of this habitats. In addition, the record of var. coastal habitat during marine transgressions in palustris occurring along a river near the coast the Plio-Pieistocene. Elements of the older of Colombia could be interpreted as indicative tropical flora appear to have found a refugium of a more: extensive past distrlbution, a rellct 1n the uplands where H. aurea occurs (Lee and of hylaea conditions prevlous to the rise of the Langenheim, 1973) . Hymenaea eriogyne. oc­ Andes in Middle Tertiary time. curring in forest patches of the caatinga of Further differentiation of the H. oblongifolia northeastern Brazil, appears to be a relict of complex is represented by variety davisii in rainforest stock isolated in extremely arid the hylaea of Guyana, where it occurs in moist conditions. lt would likewise be assumed sites along rivers. On the other hand, variety that this differentiation probably took place latifolia occurs prlncipally in upland sites, during the Tertiary. although it also is found along streams in this Vl1ith respect to ecological differentiation habitat. lt seems clear, therefore, from the within the hylaea, the distribution of the range of habitats In which the varieties of polytypic H. oblongifolia is especially inter­ H. oblongifolia occur, that it ls a plastic stock esting. lt is the species closest to the African in which some morphological and probably

- 31 physiological differentiation has taken place in drier forest E:cosystems. lt generally seems with local ecological isolation. lt has the best adapted to seasonally dry and savannah widest distribution of any hylaea species of forests than to continually moist hylaea . Hymenaea. The restriction of the varieties However, specimens have been collected in oblongifolia and palustris to the moist eastern several isolated pockets in the extreme and western zones of the Amazon Basin ap· northwestern and southwestern portions of the pears curious, since várzea and igapó plants hylaea (Fig. 9). Although most of these are usually water dispersed and tend to occur specimens are incompleta, they appear to along the length of the Amazon. Ducke and belong to H. courbaril var. courbaril. The Black (1953), however, point out that some habitat data also are poor, but several speci· plants are distributed in the Eastern and Western mens clearly come from the hylaea, whereas Zones, but missing in the Central Zone. some in Bolívia are from relatively high ele­ Whether this distributional phenomenon could vations (900-1800 m) on the western side of be related to both lesser annual precipitation the Andes. This distribution of H. courbaril and a more pronounced dry season in the Cen· is particularly anomalous, since no other trai Zone is not known. specimens have b e e n collected, to our The only other member of sect. Trachylo­ knowledge, eitfler south or west of Manaus bium occuring in the hylaea, H. parvifolia, is lfig. 9). Jt does have a dislribution south of relatively closely related to the H. ob/ongifolia the Amazon in central and eastern Brazil where complex, but apparently represents a drier considerable variation suggests new varieties habitat facies. lt occurs primarily on sandy may be recognized in the future. Also it appears terra firme sites within the Eastern and Central to be the parenta! stock for radiation of closely Zones of Amazonia, although a few collections related cerrado species such as H. stilbocarpa, have been made in the Southwestern section H. martiana and H. stigonocarpa (Lee, 1973). along the upper Rio Madeira and in Acre . In the Amazon Basin, H . courbaril seems best Considerable morphological variability charac­ represented by its variety subsessi/is, which terizes the populations of H. parvifolia &nd this most generally occurs along sandy beaches m variability is also reflected in the quantitativa trioutaries in the central Amazonian region . sesquiterpene composition of the leaf pocket Here it is a tree much reduced in size which resin. Hymenaea parvifolia occurs in the cam· essentially becomes denuded of its leaves po and savannah areas more than any other during the dry season and flowers on a different hylaea species of Hymenaea than possibly H. schedule from variety courbaril. lt coul possioly courbaril var. courbaril. lts form ranges from be a form which developed during a dry intervai shrub-like in the center of the campo to of the Pleistocene. emergent tree 1n the nei ghbori ng rai nforest. Apparently non-hylaea species characterize Hymenaea intermedia morphologically is most campos; however, H. parvifolia occurs in intermediate between H. oblongifo/ia and H. both habitats. lt is in these campo-hylaea courbaril. Hymenaea intermedia g e n e r a I I Y areas where H. parvifolia displays its greatest occurs on terra firme in a relatively restricted variability. This situation may possibly fit the area of the drier parts of the Central Zone and model suggested by Stebbins (1952). western part of the Eastern Zone. Ecologically H. intermedia is closer to H. courbaril than to Hymenaea courbaril has the widest distri­ H. oblongifolia. However, the closely related bution of any species of Hymenaea, occurring essentially over the range of the entire genus species H. adenotricha, probably best con­ in every major ecosystem type, thus giving sidered only a variety of H. intermedia, occurs evidence of amazing ability to adjust to a on terra firme sites of the moist Western Zone. ~tadient of moisture stress conditions. Hymc· Jt seems possible that H. intermedia represents naea courb<~ril var. courbaril seems to be less the stock most closely related to H. oblongifolia common in the Amazon Basin than in northern from which the physiologically plastic H. South America and Central America where it is courbaril cold have evolved. From its present very common and may even attain dominance distribution we might surmise that H. courbaril

32- (and its closely related species in the dry of forest species lnto these habitats. In ecosystem types) evolved during the drying other words, as suggested by Gomez·Pompa, trends of the Mid Tertiary. We have evidence Vasquez-Pompa, Vasquez-Yanes and Guevara, for H. courbaril, or possibly H. intermedia, the reproductive behavior of species which occurring as remains in Oligo-Miocene amber have evolved in the hylaea habitat makes it in Chiapas, Mexico (Langenheim, 1966) thus difficult for them to colonize open areas. Thev indicating a wide northern distribution at this also hypothesize that as a consequence of the time. low population densities of rainforest species, Hymenaea reticulata is thought to be the gene pool of these species will be greatly related to H. courbaril and possibly H. rubriflora, restricted unless large tracts of rainforests although it is a distinctive species. lt is also remain intact. Thus it appears from information distinctive from ali other Amazonian Hymenaea about the ecology of rainforest tree species in the quantitativa composition of its leaf that speciation might be increased in dry-wet pocket resin. Until recently, it appeared to contact zones but the maintenance of the have a very restricted distribution in relatively integrity of the hylaea ecosystem might dry terra firme sites in the central Amazon near demand relatively large areas of hylaea re­ Manaus. An additional collection from near maining even during dry Pleistocene climatic lquitos, Peru, extends its range westward and periods. possibly indicates that it was distributed more More research needs to be dane regarding widely in the past. the role of dry environmental conditions in the Although our present evidence regarding evolution of hylaea species and vegetation. speciation within Amazonian Hymenaea does Hymenaea is an excellent example of a genus not clearly support the hypothesis of dry which probably had its origin and early history oscillations during the Pleistocene which in moist rainforest conditions in Africa and greatly restricted the Amazonian hylaea, it South America but then differentiated in does not negate the possibility. Certainly response to dry conditions within the hylaea as there is evidence that evolution within the well as over a wide range of drier habitats both genus has responded to dry environmental north and south of the hylaea. The evolution conditions. The distribution of species is of some of the dry-adapted species of obviously related to moisture gradients and Hymenaea probably began in the early to emphasizes the heterogeneity of the present middle part of the Tertiary, continued through environmental conditions and flora within t h6 the drying trends during that Period, and finally hylaea. The Amazonian hylaea today is defi­ could have responded to dry oscillations during nitely not a "vast uniform habitat"; in fact, as t he Pleistocene. noted previously, the area of heterogeneous seasonally dry forest, savannah, campo, cam­ ACKNOWLEDGMENTS pina and caatinga is considerable. Even though the influence of drier climates during the We wish to express our sincere appreci­ Pleistocene seems likely, we would question ation to the numerous individuais who have whether the hylaea vegetation was repeatedly enable this work to be carried out : Dr. Paulo and rapidly restricted to small peripheral de A lmeida Machado, Director and Dr. William islands as suggested by Haffer (1969), Vanzolinl Rodrigues of INPA. Manaus, Amazonas; Dr and Williams (1970), and Vanzolini (197'3). João Murca Pires of IPEAN, Dr. Paulo Cavalcan­ Present data from study of regeneration o;: te of the Museu Goeldi, and personnel of tropical rainforests indicate that primary tree ICOMI, Belém, Pará; and Mr. Heliomar Magnago species are incapable of recolonizing large of SUDAM, Santarém, Pará, without whose areas opened at least to agric11lture l Gome7· assistance the collection of specimens in the Pompa, Vazquez-Yanes and Guevara, 1972). Amazonian region would no thave been Also it appears that the present floras of the possible. We owe similar thanks to Prof. Dar· Amazonian campos are characterized by non­ dano Andrade-Lima, IPA, Recife, Pernambuco hylaea species, with little evidence ot invasion for aid in collecting in t he northeastern area

-33 (Pernambuco, Paraíba, Piauí, Ceará and Bahia), ANDRADE-LIMA, 0. and to Dr. Paulo de T. Alvim, Director, CEPEC 1953 - Notas sôbre ~ dispersão de algunas espécies ve­ ltabuna, Bahia for assistance in collecting in getais no Brasil. A 11 . Soe. Biol . Pernambuco, J1 (I ) : 25-49. southern Bahia. We also are grateful for help ARI\UJO, v. c. DE provided by Dr. Luis Emygdio de Mello Fi­ 1970 - Genolog!a de esscncias florestais Amazonicas I . lho, Museu Nacional and Dona Graziela M Boi. INPA; Pesquisas Florestais. 4 : 1-25. Barroso, Jardim Botânico, Aio de Janeiro. ARCIIANGELSKY, S Grateful acknowledgment is made to Dr. Alva 1968 - Paleobolany and palynology in Soulh America: Whittingham and Mr. William Stubblebine for A historica: rcview . Rev. Paleobot. Palynol., various aspects of preparation of the manuscript 7 : 249-266. and to the Maria Moors Cabot Foundation of AUBRÉVILLE, A. Harvard University and the National Science 1961 - Ewde éculogique des principa/es formatiOIIS Foundation (grants GB-5816, GB-13659 and GB- végétales du Brésil et contributio11 a la connais· sance des furêts de I' A mazonian bdsilienne. 29278) for funds to support this study. Nogent·sur-Manne. Centr. Tech . Forest. Trop. (Sienc- Francc) . 1962 - Savanisation tropicale et glaciations Quater· RESUMO mures. Ada/ISU/Iia . N s '2(1): 16-84. O gênero Hymenaea, produtor de resina, tem AXELROO, D. l. uma distribuição anfi.atlãntica com 13 espécies neo. 1970 - Mesozoic paleogeography and early angiosperm trópicas e 1 africana de ocorrência ao longo de sua bistory. Bot . Re1•., 36: 277-319. costa oriental. A evidência presente sugere umu BAKhR, H. G. origem africana do gênero com migração através do 1970 - Evolution in lhe lropics. Biutropica, 1(2) : Atlântico, ocorrida durante o começo do Terciário, 101-111. quando os continentes estavam mais próximos e a 1973 - Evolutionary relationships t>ctween flowering vegetação de mata pluvial tinha uma dispersão mui­ plan.s and aoimals in Amcrican and African to maior do que atualmente. O centro de sua distn­ tropical forests. In : Meggcr~. B. J ~t ali i . buição neotrópica é a hiléia amazônica, apesar de o Tropical Furesr Ecmysums in A/rtca mui Soutll gênero em questão se encontrar em todos os maio. America. A Cumpt.. rtllil•e Re1•ie•• . Wal>hmgl<.n, res tipos de ecossistema, dentro de sua extensa faixa D. C., Smithsonian Press, pp. 115-160. que vai de 23oN a 28oS. A taxonomia, ecologia, quí• bAI{,;OsA, O. & RAM<>S, J . A. R. mica da resina e estrutura de melhoramento gené­ tico das 9 espécies e 8 variedades de Hymenaeae das 1959 - Território do Rio Branco (Aspectos principais matas pluviais da Amazônia e da historicamente da geomorfologia, da geologia e das possibilida­ relacionada costa atlântica brasileira são discuti. des minerais de sua zonr. setentrional). Boi. Di1• . Geo/. Mineral Dep. Nac. Prod . Minero/, das, como também a evolução possível destas espé­ cies em resposta às postuladas tendências de seca Rio de Janeiro, 196 : iniciada no Terciário Médio e durante as oscilações BEARO, J . S. de clima seco e úmido do Pleistoceno 1946 - Tbc natural vcgetation of Trinidad. Oxford For. Mem., 20. BENTIIAM, G. LITERATURE CITED 1840 - Hooker's 1. Bot. Kew Gard. Misc. 2: 38-146. BLACK. G. A. ET ALU Aa'SABER, A. 1950 - Some att!mpts lo eslimate species divcrsily and 1967 - Problemas geomorfo16gicos da Amazônia brasi­ leira. Atas do Simpósio sôbre a Biora Amazo population density of trees in Amazoniar. nica. 1 : (Geociencias). Rio de Janeiro: Con­ forests. Bot. Gaz., 3 : 413-425. selho Nacional de Pesquisas, pp. 35-62. BRENAN, J. p . M. 1968 - Pavimentos dctríticos atuais e subatuaas das 1956 - The geographical relationships of Lhe gcnera of caatingas brasileiras. Notícia Geomorfo/6gica. the Leguminosae in tropical Africa. Webbia Pontif . Uni v. C atol. Campinas, 3 : 52-54. 19 (2) : 545-578.

ALVIM, P. DE T. BROWN, K. S. JR. 1964 - Tree growlh periodicity in tropical climates. 1972 - The Heliconians of Brazil ,Lepidptera: Nyn• · ln: Zimmerman, M. Wood Formation . New phalldae). Pt. lll : Ecology and biology of York, Academic Press, pp . 479-495. He/iconius nallercri, a key primitive species near AMSHOI'F, G. J . H. extinction and comments on tbe evolulionary 1939 - Caesalpinioideae. In: A. Pullc (Ed. ), Flora development of He/iconius and Eucides. Zoolo of Suriname (NetiJerlands Guiana), 2(2) : 23. gica (New York Zoologicul Society), Spring.

34- BROWN, K. S. JR. & MlELitE, o. F. F. GOMEZ-PoMPA, A. F.T AUI 1972 - The Heliconians oí Braz.il (Lepidoptera: Nym­ 1972 - The tropical raio forest: A nonrenewablc phalidadc. Part. n. Iotroduction and general resource. Science, 177 : 762-765. comments, with a supplemeotary revision of thc GUNN, R. tribe. Zoo/ogical (New York, Zoological So­ 1968 - Stranded seeds and fruits from the southeastern ciety) . Spring. shore o f Florida. Garden Jo umai ( March/ CANDOLLE, A. P. DE April) : 43-45. 1825 - Prodromus systematis naturalis regri vegetobilis HAFFER, J. 2. Paris, p. 94. 1969 - Spcciation in Amazoniao forest birds. Science, CARVALHO, C. T. DE 165(3889) : 131-136. 1960 - Das vistas de Morcegos às Flores. (Mamma­ H ARBORNC:, J . B. lis, Chiroptera). Anais Acad. Brasil Ci., 32: 1971 - Terpcnoid and other 1ow molecular weigbt 359-377. substances of systematic interest in the Legumi· 1961 - Sôbre os hábitos alimentaires de Phillostomideos nos a c. Jn : Harborne, J. B . eí alii ( Eds. ) - (Mammalia, Chiroptera). Rev . . Bio. Trop .. Cllemotaxonomy of the Leguminosae. London, 9(1) : 53-60. Academic Press, pp. 257-283. CUNNINGiiAM, A . , ET ALI! HEYWOOO, V. H. l973a- Resin acids from two Amazooian species of 197 1 - The Leguminosae - a systematic purview . Hymenaea . Pllytochem., 12 : 633-635. ln : Harborne, J. B. et alii (Eds.). Chemo­ 1973b- Labd-13-en-8-ol-15-oic acid in the trunk resin taxonomy of the Leguminosae. London, Aca­ of Amaz.onian Hymenoea courbaril. Pllito­ demic Prcss, pp. 1-29. chem.: (in press). HowES, F. N. DARLINGTON, P. J. J R . 1949 - Vegetable gums and resins. Waltham, Mass.: 1957 - Zoogeography : The Geographica/ Distribution Chron. Bot. of Animais. New York, John Wiley and Sons. HUBER, J. DIETZ, R. A. & HOLOEN, J. C. 1909- Materiais para Flora amazônica. 1970 - Reconstruction of Pangaea: Breakup and a Boi. do Museu Paraense Hist. Nat., 5: 386. dispersion of continents, Permian to Present . Jour . Geophysical Research, 75(26) :4939-4956. 1910 - Maltas c madeiras amazonicas. Boi. do Museu 0UCKE, A. Paraense Hist. Nat., 6: 181. 1915- Arch . lt1rd. Bot., Rio de Janeiro, 1: 23. HUGEL, 0. F.T ALU 1966 - Thc structure and stereochemistry of diterpenes 1925 - As Leguminosas do Estado do Pará. Arcl1. from Trachylobium verrucosum Oliv. Tetrahe­ Jard . Bot ., Rio de Janeiro, 4 : 283. dron Suppl., 8, Pt. I : 203-216. 1930 - As Leguminosas do Estado do Pará. Supplcmen­ to Arch. Jard. Bot., V. Rio de Janeiro. H UMBOLDT, A. 1807 - Paris, 1935 - As espécies Brasileiras do jatahy, jutahy ou ja­ Essai sur ia geographie des plantes. F. tobá (gênero Hymenaea L., Leguminosas Cae­ Schoell. salpiniaceas). Anais Acad. Brasil Sei. 7(3) : 1814-1825- Voyage aux regions equinoxa/es du Nouveau 203-211. Continent . Paris, F . Schoell. 3 vols. 1949 - Notas sôbre a Flora Neotropica. ll. As Legu­ 1852 - Personal narrative of travels to the equinoctial minosas da Amazônia Brasileira. Boi. Tecn . regions o/ A merica. Transl. London, T. Ross. lnst. Agron. N ., Belém, Pará, 18. JANZEN, D. H. OUCKE, A . & B~.t-CK, 0 . A . 1967 - Synchronization of sexual reproduction of trees 1953 - Phytogcograpbical notes on the Brazilian Ama­ within the dry season in Central America. zon. Anais Acad. Brasil Sei., 24(1) : 1-46. Evo/ution, 21 : 620-637. EJrEN, G. 1969 - Seed-eaters versus seed size, number, toxicity 1972 - The cerrado vegetation of Braz.il. Bot. Re1· ., and dispersai. Evolution, 23: 1-27. 38(2) : 201-341. 1970 - H erbivores and lhe number of tree species in F ASBr-NDt:R, M . V. tropical forests. Amer. Nalllralist, 104:501-528. 1959 - Pollen grain morphology and its taxonomic 1917a- Scllee/ea, Hymenaea and Sickingia and their seed significnnce in thc Amhcrsitieae, Cynometreac predators. OrganiUltion for Tropical Swdies and Sclcrobieae (Caesalpiniaceae) with special News, 11(5): 3-4. refercnce to lhe American genera . Lloydia, 1971b- Euglossine bees as long distance pollinators of 22(2) : 107-162. tropical plants. Science, 171 : 203-205. GERMERAAO, J. H. ll.T ALlf lANG~HE J M, J . H. 1968 - Palynology of Te:rtiary sediments from tropical 1966 - Botanical source of ambe'I' from Chiapas, Me­ arcas. Rev. Paleobot. Palynol., 6 : 189-348. xico. Ciencia, México, 24 : 20 1-211.

- 35 . 1967 - Prelimir.ary investigations of Hymcnr.ea cour­ M ENÉNDEZ, c. A. baril as a resin producer. J. A mold A rbor., 1969 - Die Fossilcn Floras Sudamerikas. In: Fittkau, 48 : 203-230. E. F., et ali i (Eds.) - Biogeography and 1969 - Amber · A botanical inquiry. Science, 1633 : Ecology in South America 2. The Hague, Dr. 1157-1169. W. Junk Publ . , p.: 519-561. 1973 - Legumir.ous resin·produciog trees in Africa and MINISTÉRIO DA AGRICULTURA DO BRASIL South America. (n: Meggers, B. J. et ali1 (ESCRITÓRIO DE METEREOLOGIA) (Eds.) - Tropical Forest Ecosystems in A f ri­ L968 - Normas climatológicas, I: 1-39. Rio de Ja­ ca and Soutll America: a Comparative Review. neiro. Washington, D. C., Smithsonian Press, p : MOREAV, R. E. 89-104. 1966- The Bird Faunas of Africa and irs lslands. LANGENH.ElM, J. H., ET ALU New York, Academic Press. 1970 - Amazonian species of the resin·prouucing MULLER, J . genus H yme11aea. Amer. J. Bot., 57 : 754. 1970 - Palynological evidence on early differentiat10n LANGENHEIM, J. H . & LEE, y. T. of angi0spe-rms. Biol. Rev. Cambridge Philo. 1973 - Rcinstatement of the genus Hymenaea (Le Soe., 45: 417-450. gumino:;ae.: Caesalpinioidea) in Africa. Britto­ NAKANO, T. & DJERASSI, c. nia, 25 : in prcss. 1961 - Terpenoids. XLVI. Copalic acid. J. Org. LF.E, y. T. Chem., 26: 167-173. 1973 - A sistematic study oj t/Je ge11us Iiymcnaca L. PiRES, J . M ., ET AJ...ll (Lcguminosae, Caesalpinioideae, Dctarieac). Ph. 1953 - An estimatc of the number of species in an D. dissertation, Univ. of California, Santa Amazonian forest community. Bot. Gaz., 6 : Cruz. 467-477. LEE, Y. T. & LANGENHEIM, J . H . R,weN, P. H . & AxELROD, D. I. 1973 - New taxa of the genus Hymenaea (Legumino­ l972 - Plate tectonics and Australasian paleobiogeo­ sae : Caesalpinioideae) from Bra1.il and Guya­ graphy. Science, 1763 : 1379-1386. na. J. Arnold Arbor., 54(1): 94-104. Rt,CORD, S. J . & HEss, R. W. Lt-

36 - SCHWABE, G. H. A/rica and Souch America: A Comparative 1969 - Toward an ecological cbaracterization of the ReviQw. Washington, D. C., Smithsonian Press, Soulh Amcrican continent. In : Fittkau, E. F. p.: 27-48. et a/. (Eds. ) . Biogeography and Ecology 111 u. s. DEPT. OF CoMMERCE (U. s. ENviRONMENTAL South America 1. The Hague, Dr. W. Junk DATA SERVICE) Publ., p.: li J.l34. Jan. 1959 - Dec. 1969 - Monthly climatic datâ for the SIMPSON, G . G . world. Washington, D . C. 1969 - Soutb AmeTican mammals. lo: Fitkau, E. F., V ANZOLINI, p. E. et al. (Eds.) . Biogqogrophy and Ecology in 1973 - Paleoclimates, relief and species mulliplication South A merica 2. The Hague; Dr. W. Junk Publ., p. : 879·909 . in equatorial forests. lo.: Meggers, B. J. et a/ii (Eds. ), Tropical Forest Ecosystems in Africa SIOLI, H. and South America : A Comparative Revil?w. 1968 - Zur õkolog:ie des Amazonas-gebietes. In.: Fittkau, E. F. et ai. (Eds.) . Biogeography Washington, D . C., Smilhsonian Press, and Ecology in South America 1. Tbe Hague, p.: 255-258. Dr. W. Junk Publ., p.: 137-170. VANZJOLINl, P. E. & WILLIAMS, E. E. 1973 - Recent humao activities in the Brazilian Amazon 1970 - South American Anoles. The geographic differ­ Region. Jn : Me-ggers, B. J. et alii (Eds.), entiation and evolution of the Anolis cryso/epsis Tropical Forest Ecosystems in A/rica and South species group (Sauria, Iguaoidae). Arq. Zool. America : A Comparative Re1•iew. Washington, São Paulo, 19~1-2): 1-124. Smithsonian Press, p.: 321-334. VAN D ER HAMMEN SoARES, L. DE C. T. 1953 - Limites meridionale e orientais da área de ocor­ 1961 - The Quarternary climatic changes of norlhem rência da floresta amazônica em território bra­ South America. Ann. New York Academy of sileiro. Revista Brasil. (Jeogr., 15: 3-122. Sciences, 95 : 676-683. SOKAL, R. R. & SNEATH, P. H. A. VEEVE.RS, H. J . ET "LII 1963 - Principies of Numerical Taxonomy. San Fran­ 1971 - 1ndo-Australian stratigraphy and lhe configu­ cisco, W. H. Freeman Co. ration and dispersai of Gondwanaland. Noture, SOMBROEK, w. G. 229.: 383-388. 1966 - Amazon soils, a reconnaissance oj the soils oj V

..

-37 APPENDIX

Synoptic Key to the Hylaea Species of Hymenaea

I. lnfloreseenee long-panleulate, its branehes long. slender, only sllghtly frexuous; flowers small; ovary mostly pu­ beseent throughout. densely hlrsute at the base; fruit ovoid to obovoid. sub-eompressed, 1· to few-seeded ...... seet. Trachylobium . 11 . Petals densely pilose above; leaflets generally small. 5-& (12) x 2 .5·3.5 (4.5) em, broadly faleate ...... H . parvifolia . 11 . Petals glabrous; leaflets large, 11-15 x 4-6 em, oblong to faleate ...... • . . H . oblongifolia. 111 . leatlets glabrous on both sides. IV . leaflets narrowly faleate; petals oblaneel)fate and subsessile; hypanthium with a stalk-like base, ca . 9 mm long ...... var. davisií. IV. i.eaflets narrowly or hroadly oblong; petals subclawed to distinctly clawed ; hyanpthium with a stalk­ like base, ca. 3 mm long.

V Leatlets narrowly oblong (l/ W ca . 2); petals sub-clawed to clawed, the claws to 2 mm lon9 ...... var . oblongifolia. V. leatlets broadly oblong (l/ W ca . 3); petals distinctly clawed, the claws to 8 mm long ...... var. latifolia. 111 . leaflets densely golden-brow tomentose beneath . . .. var. palustris.

Young lnflorescence ~hort-paniculate, densely corymbose when mature, its branches short. thick, strongly tlexuous; flowers large; ovary densely woolly-tomentose to glabrous; fruit obovoid. rhomboidal but mostly cyllndrieal to oblong, compressed or not, usually manyseeded ...... sect. Hymenaea. 11 . Ovary densely pubescent throughout; petals spatu late, distlnctly clawed. the claws more than 6 mm long 111 . leaflets obovate. pubescent beneath or occasionally glabrate; ovary moderately hirsute at base; fruit narrowly oblong, subcompressed ...... H . eriogyne 111. Leaflets obovate-oblong. densely golden · brown tomentose beneath and sparsely tomentose above; ovary densely woolly-tomentose throughout; fruit broad and flat ...... H . aurea . 11 . Ovary glabrous throuç)hout or when young with trace ot pubescence at the base; petals ovate. obovate or oblanceolate. not spatulate, sessile to subclawed, the claws less than 2 mm long . 111. Ovary borne on a slender stipe, rather small (2 x 2 mm), rhomboidal; fruit ovoid or obovoid, usually rhomboidal at maturity, subcompressed; seeds 1 or 2, rarely more . IV. Leaflets oblong. glabrous o r occaslonally sparsely puberulent beneath ; base stralght, not rounded- protracted on the outer side ...... H. intermedia. IV. ...eatlets obovate, lightly to densely pubescent beneath; base rounded-protracted on the outer side ...... H. adenotricha. 111 · Qvary rigidly stipitate, oecasionally subsessile, oblong-oblique (6 x 2 .5 mm); fruil oblong to cylindrlcal. terete to strongly compressed; seeds 3-6 or more. IV . Petals narrowy oblanceolate. acute, shortly clawed. the claws ca . 2 mm long; leaflets slightly falcate to oblong, large, 7-18 x 3.5-7 em . V. leaflets falcate, glabrous, dull, the veinle ts elevated, prominently retieulate; fruit large . rhomboidai-Qbfong (10·20 x 6·10 x 2.5·3.5 em), broad, smooth (without evident resinous pocKets and lentieels) felt-like ...... H . retículada. V Leaflets oblong. sparsely pubeseent, shining above, densely golden-brown tomentose bcneath. the veinlets obscure; frult smali (5-6 x 3 x 1 S-2 em). oblong. with evident resinous pockets and lenticels, not telt-like ...... H . rubriflora . IV. Petals ovate to obovate, sessile to subsessile. rarely clawed; leaflets f a I cate, small. 4-10 x 2 .5 · 3 .5 em ...... H . courbaril . V. Ovarv with stipe 4-6 ( + ) mm long; leaflets broadly falcate, short-acuminate to obtuse, the base nearly straight to rounded on outer side; fruit large and oblong ...... var. courbaril . V Ovary subsessíle or with stipe ca. 2 mm long; leaflets narrowly talcate short-to long acuminate. the base straight on the outer side; fruit smaller; cylindrleal ...... var . subsessilis.

38-