Copyrighted 1987 by The New York Botanical Garden, Bronx, New York 10458-5126. Reprinted with permission from the Memoirs of The New York Botanical Garden 44: 9-29. 1987. Please report comments and corrections to Scott A. Mori (smori@nybg.org).

Abstract

An ecological description of the forest of La Fumée Mountain is provided in order to relate species of Lecythidaceae to the forest in which they grow. The description is based on a sample of 800 trees with minimum diameters of 10 cm and on general observations and botanical collections. The forest is characterized by a relatively high percentage of trees with: buttresses (34%); simple (58.5%), entire (94.1%), mesophyll (88.0%), mostly evergreen leaves; and latex, resin, or distinctive sap (42.7%). Species of Lecythidaceae at La Fumée Mountain are: buttressed (39%) or unbuttressed (61%); with simple (100%), entire (58.5%) or serrate (41.5%), evergreen (93.8%) or deciduous (6.2%) leaves; without latex, resin, or sap. Vascular epiphytes are found on 51% and lianas on 42% of the trunks of all trees. Epiphytes are probably not more abundant because of the well-defined dry season from August through November and because of the relatively dry microclimate where the transects were located. Size class distribution of tree diameters shows the reverse J-shaped distribution typical of most forests which, along with the relatively large average tree size, indicates that this forest has not undergone major disturbance in the recent past. Three tree strata are present (emergent, canopy, and understory), and species of Lecythidaceae are found in all three. Tetragastris altissima (Burseraceae) is the most frequent species occurring in 25 out 40 sampling units. Eschweilera coriacea and Couratari stellata are the most frequent species of Lecythidaceae.

Total tree density (DBH ³ 10 cm) is 619 individuals per hectare, with Tetragastris altissima and Quararibea turbinata (Bombacaceae) the most common species. Eschweilera coriacea, Courtari stellata, and Gustavia hexapetala are the most dense species of Lecythidaceae. Total basal area is 53.0 m² per hectare which yields a relatively high 856.2 cm² per individual. Tetragastris altissima has the greatest basal area of all trees and Couratari stellata is the most dominant Lecythidaceae. Calculation of importance values shows that Tetragastris altissima is the ecologically most important species at La Fumée Mountain (IV = 17.9). The forest is dominated by relatively few species as 8.8% of them account for 42.5% of the importance value of the sample. The five most important families in order of importance are: Burseraceae, Sapotaceae, Lecythidaceae, Mimosaceae, Caesalpiniaceae. La Fumée Mountain is incredibly rich in tree species as is indicated by the appearance of 295 species in the 800 tree sample. Nevertheless, the same sample only included about 53.8% of the total species present in the vicinity of Saül. The authors estimate that 546 species of trees with minimum diameters of 10 cm occur in the 133,600 hectares of the proposed biological reserve.

In order to relate the abundance, dispersion, and structure of Lecythidaceae to that of other families of trees found in the forest of La Fumée Mountain, we made a preliminary ecological survey of the area. The purpose of the survey was to provide information about tree diversity, frequency, density, and dominance as well as to describe the general appearance and structure of the forest.

The forests of the Guianas have been the object of numerous ecological studies. Noteworthy have been those of Rollet (1969) in Venezuelan Guayana; Davis and Richards (1933, 1934) and Fanshawe (1954) in Guyana; Lindeman (1953), Lindeman and Moolenaar (1959), Maas (1971a, 1971b), and Shulz (1960) in Surinam; and Oldeman (1974), de Granville (1978), and the multidisciplinary team of scientists working out of ORSTOM's center in Cayenne (Anonymous, 1981, 1982) in French Guiana.

The transects were placed in the La Fumée Mountain area located 2.5 to 4.5 km NNW of Saül 3° 37'N, 53° 12'W). Here the terrain is relatively dissected. The highest point is 410 m at the junction of the main La Fumée trail with the Antenne Nord de La Fumée trail and the lowest is about 200 m. Much of the La Fumée trail system follows narrow ridge tops between 300 and 400 m (Fig.II-1; Figs.V-1-6).

Details of the climate are elaborated in the discussion of the phenology of Lecythidaceae presented elsewhere in this volume (Chapter XI). While there is an average of 2413 mm of rain, it falls unevenly throughout the year, resulting in a well-defined dry season from August through November.

Disturbance by man has been minimal. Nevertheless, near Saül the forest has been cut numerous times as part of traditional shifting cultivation. The principal crop is Manihot esculenta. Further from the village, selected trees have been removed to obtain lumber for construction and for charcoal production. Eschweilera pedicellata is sometimes selectively cut for the latter. These activities have only penetrated a short distance along the La Fumée trail system — probably because of the difficulty in removing logs from the rugged terrain.

Natural tree falls are common throughout the area and evidence of landslides on steep hillsides is occasionally encountered.

The area is rich in birds and other animals. However, game animals, such as agoutis, wild pigs, and monkeys are heavily hunted. Consequently, studies of game animals, in particular primates, have to be done several days' walk from the village of Saül to avoid the influence of hunters.

Eight hundred trees with diameters at breast height (DBH) equal to or greater than 10 cm were sampled using the point-centered quarter method of Cottam and Curtis (1956). A transect four km long, along which four trees were sampled every 20 m, was employed in the same manner as described by Mori et al. (1983b). The initial starting points of the transects were placed to maximize sampling of trees from as much topographic diversity as possible. Trees 0-200 were sampled from a zigzag transect along the ridge of Antenne Nord. This transect started near our camp marked carbet in Figure V-1. In order to stay on or near the ridgetop, the transect's direction was alternated every 100 m from NE to NW and vice versa. Trees 201-400 were sampled from a W-running transect which started from near the carbet and descended the W-facing slope of Antenne Nord from where it continued over the slightly undulating terrain at its base. Trees 401-584 were sampled from a S-running transect which departed from the end of the preceding transect and made its way up a N-facing slope leading to the ridge of La Fumée Ouest. Finally, trees 585-800 were sampled from a transect which, continuing S from the preceding transect over the ridge of La Fumée Ouest, descended a S-facing slope for 520 meters. The transect then turned to the E for 80 m and proceeded back to the N for 480 meters. The only slope exposure not sampled was E-facing. Stream valleys were not adequately sampled. The transect positions are shown in Figure V-1.

It should be emphasized that this method maximizes sampling for species diversity because a relatively long distance over maximum topographic relief is covered. Most studies of neotropical forests have employed square quadrats 100 x 100 m or linear quadrats 10 x 1000 m. Theoretically, the greatest diversity, if similar forests are sampled, will always be obtained from the point-centered quarter data, the second highest from linear quadrats, and the least from 100 x 100 m square quadrats. Therefore, it is recommended that diversity values obtained in our study be compared only with studies in which the point-centered quarter method is employed. In summary, the point-centered quarter method allows for maximum sampling of species diversity, but the data cannot be extrapolated to give number of species per unit area, and therefore comparison with other sampling methods is not possible.

Other parameters, such as frequency, density, dominance, and tree size class distributions, can be compared with studies using any of the three sample methods. For these parameters, Cottam and Curtis (1956) and Gibbs et al. (1980) have shown that the point-centered quarter method yields data as reliable as those obtained with the quadrat methods when the same forests are sampled.

A representative leaf, or leaflet in the case of compound leaves, was selected from each of the species sampled to determine characteristics of structure (simple vs. compound), margin, and size class. The margins were recorded as being either entire or non-entire (lobed leaves were considered non-entire). The size classes used were those of Raunkiaer (1934) as modified by Webb (1959): leptophyll (less than 0.25 cm²), nanophyll (0.26-2.25 cm²), microphyll (2.26-20.25 cm²), notophyll (20.26-45.00 cm²), mesophyll (45.01-182.25 cm²), macrophyll (182.26-1640.25 cm²), and megaphyll (greater than 1640.25 cm²). Cain and Castro's (1959) method was followed to determine leaf blade area (maximum length x maximum width x 2/3). As a measure of coverage of leaves in a particular size class, we used the total trunk basal area of all individuals in that leaf-size class.

Tree diameters and heights were obtained for all individuals of Lecythidaceae. The diameters were recorded with a diameter tape at breast height (1.38 m) or directly above the buttresses in individuals with buttresses at breast height. Heights were obtained by dropping a line of known length from near the top of the trees or by placing a 12 m long aluminum pole against the tree and using it to estimate the overall height.

The presence of absence of buttresses was recorded for the 800 trees of the sample as well as for all species of Lecythidaceae found by us in the vicinity of Saül. Epiphyte and liana occurrence was noted for the first 236 trees of the sample. The percentage of latex- producing trees was calculated by totaling the number of trees in known latex producing families and dividing by the total number of trees. Care was taken to exclude non-latex producing species of the latex producing families. Additional notes were made on the resins and distinctive saps produced by other trees in the sample.

In order to relocate trees during the course of the study, each tree was assigned a numbered aluminum tag, nailed into the trunk. Three and one-half years later, however, many of the tags were difficult to locate, and a few were lost. Some of the tags had been nearly covered by newly formed tree tissue, others had dropped off because the nails had only been driven into continuously sloughing periderm, and others had fallen because the aluminum tags had been corroded by tree exudates. To maintain permanently marked trees with this system, it is necessary to revisit the trees at least every two years to replace problem tags.

All trees on the transect were collected or matched to a collection of the same species from another tree. The transect collections were recorded in the collection number series of Boom and all other plants, when encountered fertile, were collected on Mori's series. A herbarium was prepared and used to resolve species concepts in the field. If individuals were not readily assigned to species using the herbarium specimens, the trees were revisited in order to see if habit and bark allowed their placement. Authorities and voucher collections for all species from the study area are cited in Chapters IV (Lecythidacae), XIV (remaining Angiosperms), and XV (Pteridophytes). Authorities for those species not appearing in the preceding lists are provided in the text, and authorities for the names of bees are given in Table XII-1. The collections are deposited in the herbaria of The New York Botanical Garden (NY) and ORSTOM, Cayenne (CAY).

The family concepts of Cronquist (1981), unless otherwise indicated, were followed even though in some cases they may not follow current concepts of some specialists. For example, we treat the legumes as three separate families (Caesalpiniaceae, Fabaceae, and Mimosaceae) rather than as a single family, as has been espoused by many specialists in the group (Polhill et al., 1981). In the text, all three families may occasionally be treated collectively as legumes. In addition, sterile collections, not assignable to one of the legume families, are lumped as legumes.

A problem in quantitative inventory of tropical forests is the frequent changes in the determinations of species. This is especially true for groups that are under active revision at the time of the study. Consequently, the names presented in this paper may undergo some changes, and the calculated values may be slightly altered as species are either lumped or split. However, we have analyzed the data presented herein in such a way that the general concepts presented will probably not be significantly affected by these taxonomic changes.

Because the large amount of data upon which this manuscript is based is of little use to most readers and because of the high cost of reproduction, it has not been included in this publication. However, a complete data set is available from the authors or from the library of The New York Botanical Garden.

Buttresses

Of the 800 trees sampled, 274 (ca. 34%) possessed buttresses (Fig. II-2). In the Lecythidaceae 107 of the 275 marked trees (ca. 39%) and 8 out of the 24 species (ca. 33%) had buttressed trunks. Only species of Lecythidaceae reaching mature heights of 30 m or more produce buttresses. For example, all of the emergents (Couratari gloriosa, C. guianensis, C. multiflora, C. stellata, Eschweilera sqaumata, and Lecythis zabucaja) have well-developed buttresses, whereas none of the understory trees (Gustavia augusta, G. hexapetala, Eschweilera collina, E. grandiflora, E. pedicellata, and E. parviflora) possess them. Canopy species of Lecythidaceae may lack buttresses (Lecythis confertiflora, L. corrugata, L. idatimon, L. persistens, L. poiteai, Corythophora amapaensis, C. rimosa, Eschweilera apicualta, and E. micrantha) or possess them (E. decolorans and E. laevicarpa). Understory individuals of emergent species such as C. stellata do not develop buttresses until they become larger in stature. Davis and Richards (1933) have also observed that buttressing is common among canopy and emergent species but absent in understory trees.

In the coastal forests of southern Bahia, Mori et al. (1983b) found that only 17% of 360 trees with diameters greater than 10 cm possessed buttresses. The topography of their study site was less dissected than that found at La Fumée Mountain. Another reason for fewer buttresses in the Bahian forest may be the smaller tree size in comparison to the French Guianan forest (519.1 vs. 856.2 cm² average basal area). If the main function of buttresses is support, as has been suggested by Smith (1972) and Henwood (1973), then one would expect a greater number of buttressed trees growing in the more dissected habitats of French Guiana where tree falls resulting from unstable soil are more frequent. Different species composition may also account for some of the difference in number of buttressed trees in the two forests. About 54% of the species of trees of the coastal forests of eastern Brazil are endemic (Mori et al., 1981). The greater prevalence of buttresses among larger trees is to be expected because their crowns are more exposed to high winds, especially at the beginning of the rainy season, and therefore these larger trees require extra support in the form of buttresses.

Leaves

It has been well established that the leaves of tropical rain forest trees are predominantly entire and belong to Raunkiaer's (1934) mesophyll size class (Cain et al., 1956; Gentry, 1969; Grubb et al., 1963; Mori et al., 1983a; Schulz, 1960). Our data from La Fumée Mountain confirm these conclusions (Table II-I). Here, 93.2% of the species and 94.1% of the individuals have entire leaves and 83.6% of the species, 88.0% of the individuals, and 87.0% of the basal area are occupied by species with mesophyllous leaves (the sum of the notophyll and mesophyll classes).

Simple leaves also dominate this forest where 64.8% of the species and 58.5% of the individuals are of this leaf type (Table II-I). The relatively high percentage of compound leaves (35.2% of the species and 41.5% of the individuals) is due to the predominance of pinnately-leafed species of Burseraceae, Caesalpiniaceae, Fabaceae, and Mimosaceae. The Burseraceae account for 17.8% of the individuals studied but for only 4.4% of the species encountered in the transect. Of the trees sampled, palmately compound leaves are relatively uncommon, appearing only in Didymopanax morototoni, Eriotheca globosa, Caryocar glabrum, C. pallidum, and Hevea guianensis.

Species of Lecythidaceae all are simple-leaved, and 53.8% of the species and 58.5% of the individuals have entire leaves. In regard to leaf size, 92.3% of the species and 97% of the individuals are mesophylls sensu lato. Consequently, in relationship to the overall leaf spectrum of this forest, the Lecythidaceae have more non-entire leaves and a slightly greater representation of mesophyllous leaves.

In the forest the megaphyll size class is absent and the macrophylls are poorly represented, with only 8.9% of the species and 6.9% of the individuals in the size class (Table II-I). The most common species with macrophyllous leaves or leaflets are: Iriartea exorrhiza (Arecaceae), Carapa procera and Guarea grandiflora (Meliaceae), several species of Cecropia and Pourouma (Cecropiaceae), and Ecclinusa lanceolata (Sapotaceae). Similarly, the smaller leptophyll and nanophyll classes are collectively represented by only 3.4% and 2.0% of the species and individuals (Table II-I), respectively. In all cases, these size classes are occupied by leaflets from Caesalpiniaceae (Elizabetha bicolor and E. paraensis) or Mimosaceae (Newtonia suaveolens, Pithecellobium aff. pedicellare, P. basijugum, P. jupunba, P. racemosum, and indets spp. 27 & 28).

Although this forest experiences an annual dry season from August through November, it never displays a high number of trees without leaves at any given time. Nevertheless, scattered leafless individuals are observed from time to time throughout the dry season. One of the more conspicuous deciduous species is Brosimum parinarioides (Moraceae) which drops its leaves, and then flowers and flushes new leaves simultaneously. Its leaves, conspicuous by their ferruginous abaxial pubescence, carpet the forest floor under larger individuals of this species during the dry season. Hevea guianensis, probably all five of the Parkia species present, Martiodendron parviflorum, and Apeiba tibourbou are other deciduous species we have observed. In addition, five species of Lecythidaceae are deciduous. Couratari multiflora, C. guianensis, and C. gloriosa flower when entirely leafless, whereas Lecythis poiteaui and L. zabucaja drop their leaves, remain leafless for a short period, then flush new leaves and flower simultaneously. The remaining species of Lecythidaceae appear to drop and flush new leaves throughout the year, but there are probably peaks for both of these activities at certain times of the year as has been shown for Gustavia superba (Kunth) Berg in Panama (Mori & Kallunki, 1976). We suspect that greater leaf flush occurs at the onset of the rainy season, but we have no data to support this suggestion.

Leaf structures that have coevolved with ants are seen mostly in the 15 species and 34 individuals of Inga. Modified petioles, which serve as domatia, are rare, with only Tachigalia sp. 2 entering the sample. This contrasts with some Amazonian forests in which species of Tachigalia and Sclerolobium may be dominant elements of the tree flora. For example, in a forest near Camaipi, Amapá, Brazil 7.3% of a 1000 tree sample were of these genera (Mori, Rabelo & Daly, unpublished data). The most common plant with leaf domatia for harboring ants at La Fumée Mountain is the small shrub Maieta guianensis.

The leaves of Quiina pteridophylla are noteworthy because of the marked dimorphism between juvenile and adult forms. The leaves of saplings and sprouts from the base of the tree are large and pinnately dissected, whereas those of adults are smaller and not dissected.

Latex, Resins, and Saps

Latex, resins, and saps, which are mixtures of secondary compounds, are ecologically important because they help defend plants against predation. In tropical forests, where animal diversity and predation on plants is high, many species of plants possess defensive secondary compounds. We have identified those species and the number of individuals with latex, resins, and saps in the forest of La Fumée Mountain. Moreover, the presence of latex, saps, and resins provides useful information for the identification of trees in the field.

Prance et al. (1976) recorded 45% of the trees they studied in an Amazonian forest near Manaus as producing either latex, resin, oils, or phenolics. About 12% of these were latex producers. In a study of a southern Bahian moist forest, Mori et al. (1983b) observed that 20% of the individuals exuded latex. Their figures include species of Sapotaceae, Moraceae, Myristicaceae, Apocynaceae, Clusiaceae, Fabaceae (Peltogyne venosa Benth.), Mimosaceae (Stryphnodendron pulcherrimum Hochs.), and Caesalpiniaceae (Swartzia sp.)

The trees of La Fumée Mountain also include numerous individuals with latex, resins, or saps. We class those species producing opaque white or yellow, free-flowing exudates as latex producers; those with colorless or reddish free-flowing exudates as sap producers; and those with colorless or reddish, non free-flowing exudates as resin producers. The latex producers are found mostly in Apocynaceae, Clusiaceae, Euphorbiaceae, Moraceae, and Sapotaceae. These families include 65 species (22.5%) and 133 trees (16.6%) that exude latex from a slash of the trunk. Not all species of these families produce latex. For example, in the Apocynaceae 5 of 9 species and 9 of 13 trees, in the Euphorbiaceae 5 of 10 species and 6 of 16 trees, and in the Moraceae (including Cecropiaceae) 13 of 17 species and 21 of 27 trees are latex producers. Even in families such as the Sapotaceae, in which 100% of the trees are potential latex producers, the latex may not flow from the truck at all times of the year. However, the presence of latex in this family can usually be detected in broken petioles.

Latex is produced in the petioles and inflorescence of species of Thyrsodium as well as in the fruits of Minquartia guianensis, but we did not include them in the above calculations because, as far as we know, it never flows from the trunk.

Resins and saps are also frequently encountered in trees of La Fumée Mountain. The most important resin producing family is the Burseraceae, which produces an exudate, turpentine-like in smell, that is not always immediately apparent. The resin of some species (e.g., Protium pallidum, P. decandrum, P. cf. cuneatum, and P. cf. neglectum) dries into a fine white powder after exposure to the air. Mimosoid legumes of the genera Inga (at least eight species) and Pithecellobium (three species) are also common resin producers, and the caesalpinioid Dicorynia guianensis, Eperua falcata, and Martiodendron parviflorum exude resin when their trunks are cut. The resin from slashes of the latter species accumulates to form a large gelatinous mass after several days. Tapirira guianensis (Anacardiaceae), Simaba cedron (Simaroubaceae), and Sloanea eichleri (Elaeocarpaceae) were also observed to exude resin from cuts in their trunks.

All three species and 22 individuals of Myristicaceae that entered the sample produce a reddish sap from cut trunks. A reddish or sometimes colorless sap issues from the cut trunks of Licania heteromorpha, L. majuscula, L. apetala, and L. caudata, and a reddish sap is found in the cambium of at least three species of Swartzia. A caustic whitish froth exudes from the slash of Hura crepitans.

In all, 105 out of 294 species (35.7%) and 342 out of 800 trees (42.7%) sampled produced latex, resin or sap. Moreover, we probably failed to observe exudates from some trees, and therefore our estimate may be slightly low.

On several occasions we observed bees collecting exudates from the trees we had cut. Trigona dallatoreana, T. lurida, and T. sesquipedalis gathered latex from Brosimum parinarioides, Moronobea coccinea, and Bagassa guianensis, respectively. These bees probably used the latex to line the inside of their nests (Michener, 1974). On one occasion we watched two species of Melipona collect the reddish sap of Licania heteromorpha which they probably used as a nectar substitute as they did not appear to carry it away in their pollen baskets. The senior author, in Amapá, Brazil, has also recorded a species of Eulaema collecting the wintergreen like "odor" from the cut trunk of a species of Cecropia. Naturally occurring exudates are probably used in similar ways by a wide variety of bees.

Vascular Epiphytes

In a sample of 236 trees we found 51.3% and 28.4% to have epiphytes in their crowns and canopies, respectively. Our concept of epiphytes follows the broad definition of Madison (1977) which includes all plants, except parasites, that are not connected to the ground at some stage in their life cycle.

The epiphytic flora of La Fumée Mountain appears to be somewhat depauperate. Although 51.3% of the trees had epiphytes on their trunks, most of the trees had relatively few plants. In contrast, Mori et al. (1983b) found 80% of the trees in a southern Bahian forest to possess epiphytes. The Orchidaceae, which comprises nearly two-thirds of the world's epiphytic flora (Madison, 1977), is particularly poorly represented at La Fumée Mountain. We found only Elleanthus linifolius, Pleurothallis sp., Rodriguezia lanceolata, Scaphyglottis cuneata, Sobralia yauaperyensis, Stanhopea sp., and Stelis sp. In addition, only one of the numerous euglossine bees we captured at artificial scents or at the flowers of Lecythidaceae was found with an orchid pollinarium on its body. Another common epiphytic family, the Bromeliaceae, was represented in our collections by only Aechmea mertensii, Araeococcus micranthus, Guzmania lingulata, Tillandsia anceps, and T. bulbosa. We observed but forwent collecting one very large tank bromeliad, but, in general the larger members of the family, such as Aechmea conifera L.B. Smith and Hohenbergia spp. observed by Mori et al. (1983b) in eastern Brazil, are absent in this forest.

The most conspicuous epiphytes at La Fumée Mountain are Araceae. Although represented in our collections by only ten species (Monstera adansonii, M. spruceana, Philodendron acutatum, P. deflexum, P. goeldii, P. guttiferum, P. linnaei, P. pedatum, P. scandens, and Syngonium podophyllum), it is the dominant epiphytic family in terms of biomass.

The Cyclanthaceae and Gesneriaceae, common in the epiphytic flora of some forests, are represented by few species at La Fumée Mountain. The most common epiphytes of the former family are Thoracocarpus bissectus and Ludovia lancifolia. The showiest gesneriad is Drymonia coccinea.

Less conspicuous in biomass but more diverse in species are the ferns, ranging from delicate Hymenophyllum and Trichomanes to the robust "bird's nest," Niphidium crassifolium. We recorded the 57 species of epiphytic ferns marked with an asterisk in Chapter XV.

Epiphytic shrubs are represented by Oreopanax capitatus, Coussapoa latifolia, Markea coccinea, M. porphyrobaphes, Hillia illustris, and various, as yet unidentified, species of Clusiaceae. Strangler figs of several species, including Ficus nymphaeifolia, are relatively common. Species of Clusia and Ficus are important food sources for birds and arboreal mammals.

Phoradendron crassifolium and Loranthaceae 1 are the only hemiparasitic epiphytes we have collected.

The epiphytic flora of La Fumée Mountain is probably not more diverse, especially in numbers of Orchidaceae and Bromeliaceae, because of the relatively severe yearly dry season (August to November). Although the average annual rainfall of 2413 mm is greater than the 1847.5 mm recorded for the eastern Brazilian coastal forest studied by Mori et al. (1982, 1983b), the latter area appears to be richer in epiphytes because there is no annual dry season. Benzing (1983) and Pires (1984) have pointed out that year-round high atmospheric humidity, rather than high total rainfall, is most conducive to epiphytic abundance in at least some South American locations. In addition, much of the study site at La Fumée Mountain is dominated by ridges which have local microclimates drier than those of the surrounding valleys or the cloud forest of nearby Galbao Mountain (de Granville, pers. comm.).

Lianas

In a sample of 236 trees, we found 41.5% and 45.8% of them to have lianas on their trunks or in their crowns, respectively. This compares to 37.5% of the trees with lianas found on their trunks in a southern Bahian moist forest (Mori et al., 1983b). Families represented or to be expected are: Aocynaceae (several species including Odontadenia perottetii, which dropped abundant light yellow corollas in November and December), Bignoniaceae (Arrabidaea trailii, Callichlamys latifolia, Cydista aequinoctialis, Distictella elongata, Memora moringifolia, M. racemosa, and Pleonotoma albiflora), Caesalpiniaceae (flowers of Bauhinia were seen abundantly on forest floor but not collected), Combretaceae (to be expected), Connaraceae (Cnestidium guianense, Connarus perrottetii, and Rourea surinamensis), Convolvulaceae (Dicranostyles guianensis, Maripa glabra, and M. scandens), Cucurbitaceae (Gurania spinulosa), Dilleniaceae (Doliocarpus guianensis and Davilla kunthii), Euphoribaceae (Dalechampia tiliifolia, Omphalea diandra), Fabaceae (various unidentified species; Dalbergia, Lonchocarpus, and Machaerium to be expected), Gnetaceae (Gnetum urens), Hippocrateaceae (Cheiloclinium hippocrateoides, Hippocratea volubilis, Prionostemma aspera, Salacia impressifolia, S. insignis, S. multiflora, Tontelea attenuata, T. cylindrocarpa, and T. laxiflora), Loganiaceae (Strychnos cogens, S. peckii, and at least one other Strychnos), Malpighiaceae (Banisteriopsis wurdackii, Heteropterys siderosa, and at least four other unidentified species), Marcgraviaceae (Margravia coriacea and Norantea guiansis), Menispermaceae (Anomospermum chloranthum subsp. confusum, Abuta bullata, A. grandifolia, A. imene, A. rufescens, A. solimoesensis, Caryomene glaucescens, Curarea candicans, Elephantome eburnea, Sciadotenia cayennensis, and S. eichleriana), Mimosaceae (Acacia adhaerens), Passifloraceae (various unidentified species), Polygalaceae (Moutabea guianensis, Securidaca uniflora), Polygonaceae (Coccoloba aff. parimensis), Rhamnaceae (Gouania frangulaefolia), Rubiaceae (Melanea hypoleuca), Sapindaceae (to be expected), Trigoniaceae (to be expected), Ulmaceae (Celtis iguanea), and Verbenaceae (Petrea to be expected).

The foregoing list is based on collections made in the dry and early wet seasons of 1982-1983 and mid wet season of 1986 as well as collections encountered in the herbaria of CAY and NY during the routine determination of the collected lianas. No attempt was made to search these herbaria systematically for collections made in the vicinity of Saül. Consequently, our list is incomplete, especially for some families. For example, Gentry (1976) states that about 20 species of Bignoniaceae occur at any suitable Central American site. In contrast, we have collected and observed (including several species of trees) only nine species of Bignonaceae in the vicinity of Saül. Likewise, we collected no species of liana in the family Sapindaceae although Donselaar (1970) records five species of Paullinia and Urvillea ulmacea H.B.K. from a similar forest in Surinam. Legumes are also very poorly represented in our collections. On the other hand, lists for some of the families may be relatively complete. For example, the Menispermaceae of the area have been well collected because of the encouragement B. A. Krukoff gave to de Granville, Fournet, and other French botanists to collect this family. The Hippocrateaceae are also well represented in our list and may approximate the actual number of species of this family in the area. Donselaar (1970), based on a much larger sample, also found nine species of Hippocrateaceae at his Surinam site.

Tree Size and Stratification

Carvalho (1981), Hartshorn (1978), Richards (1952), Schulz (1960), and many others have shown that the size-class distribution of tree diameters of tropical forests show the reverse J-shape or negative exponential distribution characteristic of most forests. Data from La Fumée Mountain, Alto Ivon (Beni, Bolivia; Boom unpubl. data), Añangu (Ecuador; Balslev et al., in press), Bahia (eastern Brasil; Mori et al., 1983b), Camaipi (Amapá, Brasil; Mori, Rabelo & Daly, unpubl. data), and the Rio Falsino (Amapá, Brasil; Campbell et al., unpubl. data) display the same pattern (Table II-II). Deviations from the reverse J-shape size-class distribution probably indicate major disturbance sometime in the past (Shulz, 1960).

Hartshorn (1978) has pointed out that the frequent occurrence of tree falls, which, in the lowland Costa Rican forest he studied, produced a turnover time of 118 Ý 27 years, may account for the fact that trees of tropical forests seldom reach enormous diameters. Similar turnover rates are to be expected at La Fumée Mountain. Riera (1982) recorded the appearance of 17 gaps in 20 hectares in less than a year in another comparable French Guianan forest. This value is of the same magnitude as the 0.7 to 1.3 gaps per hectare per year found by Hartshorn (1978). Our observations also support the contention of Hartshorn (1978) and Riera (1982) that gap formation is much more frequent during the rainy season than during the dry season. Tree falls in 1982-1983 at La Fumée Mountain were frequent during the onset of the wet season, whereas they were nearly absent throughout the dry season. It is noteworthy that peak fruit fall and seed germination also occur at the onset of the rainy season at La Fumée Mountain. This agrees with what has been observed in other neotropical forests (Foster, 1982; Smythe, 1970). Therefore, germinating seeds which often depend upon gaps for establishment, have a greater chance of encountering conditions favorable for their survival (i.e., increased water, nutrients, and light) during the early rainy season.

The largest trees of La Fumée Mountain, i.e., those with diameters over 80 cm, are found in the Burseraceae (Protium cf. cuneatum, Tetragastris altissima), Clusiaceae (Moronobea coccinea), Combretaceae (Terminalia guyanensis), Euphorbiaceae (Hura crepitans), Fabaceae (Monopteryx inpae), Humiriraceae (Humiriastrum subcrenatum), Lecythidaceae (Couratari stellata, Lecythis zabucaja), Mimosaceae (Inga alba, Parkia nitida, Pithecellobium aff. pedicellare), Moraceae (Ficus nymphaefolia), Myristicaceae (Virola michelii), Rubiaceae (Chimarrhis microcarpa), and Sapotaceae (Pouteria cayennensis, P. engleri, and P. hispida).

A comparison of the percent of trees in the larger size-classes and average tree basal area in the forests represented in Tables II-II and II-III shows that the La Fumée Mountain forest has the highest number of large trees. Here, 16.7% and 3.1% of the trees are over 40 and 80 cm DBH, respectively. At the other end of the spectrum, the Alto Ivon (Bolivia) forest, with only 3.5% and 0.5% of the trees in these size-classes, has the fewest number of large trees. In their study of an Amazonian forest, Cain et al. (1956) found only one tree over 150 cm, six trees over 100 cm, and 69 trees over 50 cm DBH in a 1170 tree sample. Carvalho (1981) has also demonstrated the relatively low numbers of trees in the larger size classes in another Amazonian forest. From this comparison, it is apparent that the forest at La Fumée Mountain is as rich or richer in larger-sized trees as other lowland neotropical forests that have been studied. This may indicate that the La Fumée Mountain forest has not undergone major disturbance, other than normal gap formation, in the recent past.

The occurrence of strata in tropical forests is still under debate. Richards, in his classic "Tropical Rain Forest" (1952), stated that there are five strata of erect plants. However, more recently Richards (1983) admitted that the boundaries between strata are more or less arbitrary and can not be objectively defined. Nevertheless, many students of neotropical lowland forests (e.g., Davis & Richards, 1933; Grubb et al., 1963; Oldeman, 1974; Terborgh, 1985) provide some discussion and evidence that strata do exist. Grubb et al. (1963) point out that there are two types of stratification, that of species and that of individuals. It is clear that species do occupy different strata in the forest, as has been demonstrated with diameter versus height graphs by Schulz (1960, p. 164). Nevertheless, it has not been convincingly demonstrated that, when all of the species of a neotropical forest are taken as a whole, distinct strata of species groups exist. There is some evidence, however, for stratification of individuals in neotropical forests. Grubb et al. (1963) found one of the forests they studied to have three distinct strata. In addition, Davis and Richards (1933) make the case for the presence of three tree strata (their understory, canopy, and outstanding (=emergent) trees) and two ill-defined layers of undergrowth (a layer of small shrubs and tall herbs and a layer of tree seedlings and herbs). Oldeman's profile (1974, p.82) at Plateau de la Douane on La Fumée Mountain clearly shows three tree strata, which he concludes occur at 15, 40, and 55 m, respectively.

Our measurements of tree diameters against tree height for the Lecythidaceae of La Fumée Mountain indicate the presence of three strata among species of this family. Gustavia augusta, G. hexapetala, Eschweilera grandiflora, E. pedicellata, and E. parviflora are understory species. Common understory trees of other families are Siparuna decipiens (Monimiaceae), Lacistema polystachyum (Flacourtiaceae), Oxandra asbecki, Fusaea longifolia (Annonaceae), Quararibea turbinata (Bombacaceae), Ambelania acida (Apocynaceae), Duroia aquatica, and Coussarea racemosa (Rubiaceae).

The canopy is partially composed of species of Lecythidaceae reaching 20 to 35 m which include Corythophora amapaensis, C. rimosa, Eschweilera apiculata, E. collina, E. coriacea, E. decolorans, E. laevicarpa, E. micrantha, E. sagotiana, E. squamata, Lecythis confertiflora, L. corrugata, L. idatimon, L. persistens, and L. poiteaui.

The outstanding, or emergent, stratum includes many of the species listed previously as possessing large diameters. Their crowns are raised above the canopy layer and overall heights of up to 56 m are attained (Oldeman, 1974, fig. 42 showing Terminalia sp.). The Lecythidaceae are represented in this stratum by Couratari gloriosa, C. guianensis, C. multiflora, C. stellata, and Lecythis zabucaja.

All emergent and most canopy species of Lecythidaceae have zygomorphic flowers. Species with zygomorphic flowers are also found in the understory, whereas actinomorphic flowered species are mostly limited to the understory.

In some areas, such as near the junction of Antenne Nord and La Fumée Nordest, the undergrowth is dominated by the spiny palms Astrocaryum paramaca and A. sciophilum. These species are especially conspicuous because their arching leaves capture organic matter which accumulates among the leaf bases and builds up into a mound at the base of the plant (de Granville, 1977). Another acaulescent palm, Attalea attaleoides, has a similar growth form but is found only as scattered individuals throughout the forest. Heliconia bihai and H. pendula, both reaching heights of 2 to 3 meters, sometimes dominate the undergrowth, especially along ridges where sufficient light is available. A large stand of these species of Heliconia is found along La Fumée Ouest beyond the point where our transect crosses the trail (Fig. V-1). Here, hummingbirds are frequently seen visiting their flowers. Heliconia acuminata, reaching 1.5 m in height, grows in isolated scattered clumps, but, because of its bright red bracts, forms a conspicuous part of the undergrowth. Common woody shrubs or treelets of the undergrowth are: Tapura guianensis (Dichapetalaceae), Talisia carinata (Sapindaceae), Carpotroche crispidentata (Flacourtiaceae), Clidemia conglomerata, Maieta guianensis, Miconia tillettii (Melastomataceae), and various species of Psychotria (Rubiaceae), especially P. carapichea. In gaps, some of the more common species to appear are: Siparuna amazonica (Monimiaceae), Leandra rufescens, Loreya mespiloides (Melastomataceae), and several species of Piper (Piperaceae).

The herbaceous layer of the La Fumée Mountain forest is dominated by many species of ferns (Chapter XV); by scattered individuals of Calathea cyclophora in undisturbed areas and large clumps of Calathea propinqua and Maranta friedrichstahliana in disturbed areas, especially along trails; Poaceae, especially Ichanthus pallens, I. panicoides, Olyra latifolia, Orthoclada laxa, Paspalum conjugatum, Pharus parvifolius, P. virescens, and Streptogyne americana along trails and in gaps; and Cyperaceae such as Bisboeckelera microcephala and Hypolytrum jenmanii in undisturbed forest and Scleria secans along trails and in gaps. Saprophytes, represented by Triuridaceae (Sciaphila albescens), Burmaniaceae (Dictyostega orobanchioides subsp. parviflora and the rare Thismia), Balanophoraceae (Helosis cayennensis var. cayennensis), and Gentianaceae (Voyria corymbosa subsp. corymbosa, V. rosea, and Voyriella parviflora) are conspicuous because of their habit, but unimportant in terms of biomass.

Frequency

In order to calculate frequency, five consecutive points were combined to make a single sampling unit, i.e., points 1-5 = sampling unit 1, points 6-10 = sampling unit 2, and so forth. In this way our 200 points yielded 40 sampling units. The presence of each species was recorded each time it appeared in a sampling unit. Points were aggregated into larger sampling units because, for most species, if the points alone are used, frequency and density are nearly equal. The 295 species found in the 800 tree sample gave 630 occurrences, i.e., each species appeared in an average of 2.1 (ca. 5.3%) sampling units. Absolute and relative frequencies were calculated in the manner described by Mori et al. (1983b).

Eight species (Tetragastris altissima (25 sampling units), Protium apiculatum (19), Quararibea turbinata (15), Eschweilera coriacea (13), Virola michelii (10), Protium cf. cuneatum (10), P. opacum (10), and Dicorynia guianensis (10) were found in over 25% of the sampling units. On the other hand, 184 species were found in only one of the sampling units.

Tetragastis altissima is the most frequent species encountered in the La Fumée Mountain forest. Moreover, the Burseraceae, although only accounting for 13 of the 295 species, is by far the most frequently encountered family of trees in this forest, with 95 occurrences. Legumes (89), Sapotaceae (67), and Lecythidaceae (48) are other frequent families. Additional frequent species are listed in Table II-IV.

The most frequent species of Lecythidaceae is Eschweilera coriacea (13 occurrences), followed by Couratari stellata (9) and Gustavia hexapetala (7). The remaining species entering into the sample had three or fewer occurrences (Table II-V). Species of Lecythidaceae appeared in 31 of the 40 sample units and the family has a relative frequency of 7.6%. The average occurrence per species of Lecythidaceae is 3.7, whereas that of the Burseraceae is 7.7, the Sapotaceae is 1.86, and the legumes 1.34. From these data it is apparent that Lecythidaceae are found throughout the forest of La Fumée Mountain.

Density

Based on our 800 tree sample, there are 619 trees with a minimum DBH of 10 cm per hectare. Using the same sampling methods and tree size, we have found a range of 492 to 891 trees/hectare in five other comparable lowland South American forests (Table II-III). Gentry's summary of ecological studies in neotropical forests (1982, table II) indicates an even greater variation in tree density, ranging from 167 to 1947 trees/hectare with a minimum DBH of 10 cm. However, most of the studies cited by Gentry report 300 to 700 trees/hectare.

The most common tree is Tetragastris altissima with a relative density of 7.5% and an absolute density of 46.4 trees/hectare. Other common trees in the La Fumée Mountain forest are Quaribea turbinata (38.7 tree/hectare), Protium apiculatum (17.8 trees/hectare), Eperua falcata (14.8 trees/hectare), and those listed in Table II-IV. Other studies of neotropical forests have given maximum values of 41.9 (Myrtaceae indet. 17; Mori et al., 1983b), 67 (Eschweilera coriacea, reported as E. odora (Pöppig ex Berg) Miers; Cain et al., 1956), and 87.4 (Mabea brasiliensis Müll.-Arg.; Silva, 1980) for trees equal to or greater than 10 cm/hectare.

Of the 295 species found in the 800 tree sample, 178 (60.8%) are represented by only one individual. The average number of trees per species is 2.7 and the 26 most common species account for 43.3% of all the individuals. On the average, these 26 species are each represented by 13.3 individuals. Consequently, many individuals are represented by just a few species and many species are represented by only one individual. Black et al. (1950) and Pires et al. (1953) report the same individuals to species relationship in the eastern Amazonian forest they studied.

Eschweilera coriacea, with a relative density of 2.3% and an absolute density of 13.9 trees/hectare, is the most common Lecythidaceae. This species has been reported (as E. odora, a synonym of E. coriacea) to have high densities in other neotropical lowland forests (Cain et al., 1956; Dantas et al., 1980) but these values may be slightly inflated because, due to the difficulty of field identification, they may have grouped several species into their concept of E. coriacea. Couratari stellata (10.8 trees/hectare) and Gustavia hexapetala (6.2 trees/hectare) are other relatively common species at La Fumée Mountain. Density values for the remaining species of Lecythidaceae can be calculated from the data given in Table II-V.

The Lecythidaceae are represented by 65 of the 800 trees sampled or 50.3 trees of this family per hectare. The only families with higher densities are the Burseraceae (109.9 trees/hectare), legumes (84.3 trees/hectare) and Sapotaceae (54.9 trees/hectare). The Burseraceae, and to a lesser extent the Lecythidaceae, maintain high densities by having relatively high numbers of individuals/species, whereas the Sapotaceae and legumes reach high densities by having relatively high numbers of species with few individuals per species.

Dominance

The average basal area per tree at La Fumée Mountain is 856.2 cm². This figure, multiplied by the number of trees/hectare (619) gives a total basal area of 53.0 m²/hectare. This is a greater basal area than any of the other forests studied by us (Table II-III) and at the top of the range of basal areas reported for other tropical forests (Herwitz, 1981). The high average basal area per tree indicates that the La Fumée Mountain forest has not undergone major disturbance for some time.

Tetragastris altissima, with a basal area of 33,913.8 cm² /hectare (6.4% relative dominance) is the most dominant species followed by Virola michelii (22,580.4 cm²/hectare), Couratari stellata (19,489.7 cm²/hectare), and the other species listed in Table II-IV. The dominance of T. altissima is due to the presence of many relatively small individuals, whereas those of V. michelli and C. stellata are the result of fewer, but larger, trees.

The relative dominance of T. altissima is comparable to that of Eriotheca macrophylla A. Robyns (6.4%), the leading dominant in an eastern Brazilian forest (Mori et al., 1983b), but low in comparison to Vochysia guianensis (15.1%) (Cain et al., 1956) and Mabea brasiliensis (13.3%) (Silva, 1980) of an eastern Amazonian and eastern Brazilian forest, respectively.

The Lecythidaceae is the fourth most dominant family in the La Fumée forest. Trees of this family have an average basal area of 54,661.0 cm²/hectare and a relative dominance of 10.3%. Only the legumes (101,278.0 cm²/hectare, 19.1% relative dominance), Burseraceae (70,599.8 cm²/hectare, 13.3% relative dominance), and Sapotaceae (54,674.9 cm²/hectare, 10.3% relative dominance) account for greater basal areas. Among these four most dominant families, average tree size is largest in the legumes (1201.4 cm²/tree), followed by Lecythidaceae (1086.7 cm²/tree), Sapotaceae (995.9 cm²/tree), and Burseraceae (642.4 cm²/tree). The Combretaceae (4300.9 cm²/tree), Euphorbiaceae (1935.2 cm²/tree), Myristacaceae (1492.9 cm²/tree), and Moraceae (including Cecropiaceae) (1316.1 cm²/tree) are other families with large trees. On the other hand, trees of families predominantly found in the understory, such as Bombacacaea (171.3 cm²/tree), Arecaceae (171.7 cm²/tree), and Annonaceae (232.2 cm²/tree), have much smaller average basal areas.

Importance Values (IV)

Importance values for each species were calculated by adding the relative frequency, relative density, and relative dominance (Curtis & Cottam, 1962). This index has been criticized by Spurr (1964), and Silva (1980) has pointed out that in tropical forests, relative frequency and relative density are often nearly equal, and therefore for the index does not give enough weight to tree size. However, Schulz (1960, pp.160 - 161) points out the danger in placing too much value on basal area (=tree size) as an indicator of importance. For example, he shows that if basal area alone is used to calculate importance, a species that is represented by a single tree of 100 cm DBH would rank as high as a species in which 25 trees of 20 cm DBH occurred in the same unit area. Nevertheless, importance value does give an indication of 1) how frequently a species is encountered throughout the forest (relative frequency), 2) its abundance (relative density), and 3) how large its individuals are (relative dominance) in relation to all of the other species encountered in the sample. The sum of these values is an indication of the overall importance of a given species. However, as Pires et al. (1953) and Shulz (1960) have emphasized, commonness of a species at one site may go with rarity or even absence of the same species at another site.

The most important species at La Fumée Mountain is Tetragastris altissima with an IV of 17.9. It is by far the most frequent , the most abundant, and the most dominant of all species present. It is a canopy species that may reach 20-35 m x 61 cm DBH. Howe (1982) has demonstrated the importance of the sugar-rich arils of T. panamensis (Engl.) O. Kuntze as food for mammals in Panama, and presumably T. altissima plays a similar role at La Fumée Mountain. The 26 most important species of trees at La Fumée Mountain are listed in Table II-IV.

The most important species in other comparable neotropical studies have yielded IV's of 12.5 (Myrtaceae indet. 17; Mori et al., 1983b), 23.4 (Vochysia guianensis; Cain et al., 1956), and 28.66 (Mabea brasiliensis; Silva, 1980).

At La Fumée Mountain, 26 of the 295 species (Table II-IV) account for 127.4 of the 300 point IV index. In other words, 8.8% of the species account for 42.5% of the forest's importance value. Mori et al. (1983b) found that the leading 25 species (14% of the species) of an eastern Brazilian forest made up 132.3 IV points (44.1% of the points). These findings support Schulz's (1960) and Pires' (1984) claims that, although there is never a single dominant in mesic neotropical forest without soil limitations, a small group of species dominates the forest.

Couratari stellata (6.9) and Eschweilera coriacea (6.6) are among the 26 most important species at La Fumée Mountain (Table II-IV). Other studies (Cain et al., 1956; Dantas et al., 1980; Davis & Richards, 1934; Maas, 1971a, 1971b; Prance & Mori, 1979; Schulz, 1960) have also shown species of Lecythidaceae to be among the most important in lowland neotropical forests. This is especially true in the vast area to the north and east of the Amazon, Negro, and Orinoco Rivers where the Lecythidaceae have reached their greatest ecological importance.

Family Importance Values (FIV)

Ecological studies of the trees of neotropical lowland moist forests demonstrate that the families Bombacaceae, Burseraceae, Caesalpiniaceae, Chrysobalanceae, Euphorbiaceae, Fabaceae, Lauraceae, Lecythidaceae, Meliaceae, Mimosaceae, Moraceae, Myristicaceae, Myrtaceae, Sapotaceae, and Vochysiaceae usually rank among the leading dominants (Cain et al., 1956; Dansereau, 1947; Davis & Richards, 1934; Fanshawe, 1954; Gentry, 1982; Grubb et al., 1963; Maas, 1971a, 1971b; Mori et al., 1983a, 1983b, Prance et al., 1976; Sabatier, 1985; Schulz, 1960; Silva, 1980; Takeuchi, 1962; Veloso, 1946). Preliminary indications are that samples of a hectare of mesic forest with no soil limitations will contain 41 to 44 families of trees using Cronquist's (1981) system of classification and a minimum tree diameter of 10 cm as standard (Mori et al. 1983b; Prance et al., 1976). Pires et al., (1953) found 50 families (reported as 48 but expanded by us to reflect Cronquist's recognition of the subfamilies of Leguminosae as families) represented by trees 10 cm or more in diameter in a 3.5 hectare plot in eastern Amazonia. It is our feeling that the same families will occur throughout Amazonia and the Guianas but that their rankings will change, depending on the individual sample and its geographic location. For example, while the Myrtaceae is the leading family in one study of an eastern Brazilian forest, it is relatively unimportant in other lowland, moist forests (Mori et al., 1983a). Likewise, the Myristicaceae reaches its greatest importance in western Amazonia (Black et al., 1950; Boom, pers. obs.; Rodrigues, pers. comm.), even though the family is found throughout lowland neotropical America.

We have used the Family Importance Value (FIV), which is the sum of the relative diversity, the relative density, and the relative dominance of all individuals of the family in the sample (Mori et al., 1983a, 1983b), to rank the importance of tree families of La Fumée Mountain. The FIV is a modification of the Importance Value (IV) frequently used by ecologists to indicate the importance of species of plants in a community (Curtis & Cottam, 1962). The two indices differ in that families instead of species are analyzed in the FIV, and in the substitution of relative frequency (number of occurrence of one species as a percentage of the total number of occurrences of all species) in the IV by relative diversity (number of species of a family in the sample as a percentage of the total number of species sampled) in the FIV. The Family Importance Values for the 15 most important of the 43 families sampled are given in Table II-VI. These families are those to be expected throughout the Guianas and eastern Amazonia in similar kinds of forests (Black et al., 1950; Cain et al., 1956; Maas, 1971a, 1971b; Prance et al., 1976; Schulz, 1960).

The Lecythidaceae rank as the third most important family of trees at La Fumée Mountain, the first or second most important at Belém, Brazil (Black et al., 1950; Cain et al., 1956), and among the important families in central Amazonia (Prance et al., 1976). This family is less important in western Amazonia (Mori, pers. obs.) and eastern Brazil (Mori et al., 1983a). Nevertheless, it is well represented ecologically in at least one forest in the Chocó of Colombia (Gentry, 1982). Although the family is represented in Central America by numerous species, it never attains the ecological importance it holds in the Guianas and eastern Amazonia (Gentry, 1982).

Species Diversity

The point-centered quarter method used in this study was laid out on a transect that covered four kilometers in length. Therefore, the species diversity obtained from our sample will be greater than that obtained from a 10 x 1000 meter quadrat, which in turn will be greater than that of a 100 x 100 meter quadrat. The reason for this is that tree species at La Fumée Mountain tend to be clumped (Chapter X), and therefore the more spread out a sampling unit in a linear sense, the greater the chance of encountering different aggregations of clumped species. In addition, our transect included most microhabitats of La Fumée Mountain; i.e., it crossed small streams, where Euterpe oleracea entered the sample, and ridge tops, where such species as Coussarea racemosa and Lecythis idatimon are restricted, as well as including all intermediate habitats. Therefore, the data reported here can only be used in comparison with those obtained by similar sampling methods.

We found 295 species in the 800 tree sample (an average of 2.7 individuals for every species). But even these high values do not give an approximation of the total numbers of tree species present in the vicinity of Saül. A species area curve, elaborated using the method described by Mori et al. (1983b), demonstrates that a considerable number of additional species were encountered in all of the subsamples, including the last ( Fig. II-3).

We obtained an approximation of the total number of species of trees (DBH ³ 10 cm) in the vicinity of Saül by assuming that the 27 species of Lecythidaceae we have collected there represent all of the species of that family present. The ratio 27/15 represents the total number of species of Lecythidaceae in the area to the number of species of Lecythidaceae in the 800 tree sample. This indicates that only 53.8% of the species with minimum diameters of 10 cm was sampled. Multiplying this ratio by the total number of species sampled (295) suggests that 531 species of trees of this size class are present in the 133,600 hectare proposed reserve surrounding Saül (de Granville, 1975). This is much higher than might be expected based on Thiel's (1983) estimation of 500 species for all of French Guiana.

Gentry (1982) summarized studies of neotropical diversity, but none of these studies is comparable to our study because of the different sampling methods used. For Amazonian and Guianan forests, values as low as 65 (Klinge & Rodrigues, 1968) and as high as 179 (Prance et al., 1976) species/hectare have been reported. Using the same methods employed in this study, Mori et al. (1983a) found 178 species in a 600 tree sample in an eastern Brazilian forest and 208 species in a 1000 tree sample in an eastern Amazonian forest (unpubl. data). Data from another study located on the Guayana Shield in Amapá (and only 500 km SE of Saül) (Campbell et al., unpubl.), indicates that this area is nearly as rich as that of Saül in species. Further study is needed to determine if there is greater species diversity in the forests of the older Guayanan Shield in comparison to those of the younger Amazonian Basin.

Acknowledgments

We are grateful to Jef Boeke, Sue Keller, John and Beth Mitchell, and Ghillean Prance for their help in collecting data. We thank Jean-Jacques de Granville, Carol Gracie, D. Hammond, João Murça Pires, G.T. Prance, and Jan Reitsma for reviewing various drafts of the manuscript. We are grateful to Jeanne Goode for secretarial and editorial help in the preparation of the manuscript.