Vol. 133, No. 5 The American Naturalist May 1989
MONODOMINANT AND SPECIES-RICH FORESTS OF THE HUMID TROPICS: CAUSES FOR THEIR CO-OCCURRENCE
Wildlife Conservation International, New York Zoological Society, Bronx, New York 10460; *Department of Botany and Plant Pathology, Michigan State University, East Lansing, Michigan 48824 Submitted December 9, 1987; Revised August 2, 1988; Accepted November 9, 1988
Tropical forests are among the most diverse plant communities worldwide. This long-standing generalization (Richards 1952) has been reconfirmed by recent reviews (Gentry 1982; Leigh 1982; Whitmore 1984). Nevertheless, the occurrence of large areas of old-growth tropical forest dominated by one or a few species has been noted in several major surveys of tropical vegetation (Richards 1952; Letouzey 1970; Whitmore 1984). Such single-species-dominant forests may cover hundreds of square kilometers and occur adjacent to significantly more-diverse forest types. Single-species-dominant (hereafter, monodominant) forests of the humid trop- ics are often evergreen and are found on a variety of substrates. In this paper we consider only evergreen forests on well-drained soils. Species described as nearly exclusive dominants in their respective forests in equatorial Africa include Gil- bertiodendron dewevrei (De Wild.) Leonard (Lebrun and Gilbert 1954; Gerard 1960), Cynometra alexandri C. H. Wright (Eggeling 1947; Hamilton 1981), and Julbernardia seretii (De Wild.) Troupin (Gerard 1960). Monodominant forests have also been reported in the American tropics (most notably forests dominated by Mora excelsa Benth.; Beard 1946; Richards 1952; Rankin 1978), in Southeast Asia (Zon 1915; Laan 1927; Koopman and Verhoef 1938; Whitmore 1984; Rai and Proctor 1986), and in tropical Australia (Connell and Lowman 1989). In many of these monodominant forests, the principal species accounts for more than 80% of canopy-level trees. The juxtaposition of monodominant forests with adjacent species-rich, mixed forests has led to the suggestion that comparisons of these two systems might be useful (Janzen 1977; Connell 1978). Differences in the structure and composition of the two forests may indicate processes leading to monodominance in the one case and, by extension, mechanisms maintaining species diversity in the other case. We compare a widespread monodominant forest of central Africa, domi- nated by Gilbertiodendron dewevrei, with an adjacent mixed forest that is consid- erably more diverse. Forests dominated by G. dewevrei (hereafter, Gilbertioden- dron forests) are most extensive on the plateau surrounding the central basin of
Am. Nat. 1989. Vol. 133, pp. 613-633. 0 1989 by The University of Chicago. 0003-014718913305-0009S022M)M) All rights reserved. 614 THE AMERICAN NATURALIST
TABLE l
Mechanism Proposed Evidence
Change in substrate Change in forest diversity andlor species composition with change in quality soil quality Succession Monodominance is temporary: dominant canopy species shows poor recruitment in the understory Mortality and herbiv- Monodominance is stable: dominant species' regeneration is abun- ory gradient dant, with reduced seed predation and herbivory relative to other species Disturbance regime Greater abundance of tree falls and shade-intolerant species in mixed forest the Zaire River, but they also occur in smaller patches in forests of the central basin itself (Gauthier et al. 1977) and in Central African Republic, Cameroon, and Gabon (Lebrun and Gilbert 1954; Letouzey 1970). This forest type has been mentioned in earlier studies of central-African vegetation (Louis 1947; Wambeke and Evrard 1954; Devred 1958; Frankart 1960; Jongen et al. 1960; Pecrot and Leonard 1960). It has been the subject of a detailed floristic study (Gerard 1960), but not in comparison with adjacent mixed forests. The study reported here examines the composition, structure, and physical environment of Gilbertiodendron forest and of a co-occurring mixed forest in the Ituri region of Zaire. We also include an overview of the literature on upland monodominant forests of the humid tropics worldwide to enable pantropic com- parisons and to elicit global patterns consistent with our specific central-African case.
PROPOSED MECHANISMS FOR MAINTENANCE OF DIVERSITY Mechanisms proposed as important in maintaining tropical forest diversity, or implicated in cases of low diversity, were used to derive a list of expected patterns in forest composition, structure, and physical environment (table 1). These were verified in the field, where species-rich and species-poor forests co-occur. Con- firmation of an expected pattern was interpreted as support for the importance of one or more of the mechanisms. Substrate The possible importance of substrate in determining species richness in tropical forests has long been recognized (Davis and Richards 1933, 1934; Lemee 1961; Richards 1961; Ashton 1971). Chemical fertility and soil moisture are two factors often considered important. Soils of low nutrient status have been associated with forests of low species diversity or with those dominated by relatively few species (Richards 1952; Ashton 197 1; Brunig 1973; Janzen 1974). Gradients of increasing diversity have been correlated with increasing rainfall (Gentry 1982) or soil mois- ture (Lemee 1961; Hall and Swain 1981), but flooded forests and mangrove forests MONODOMINANT TROPICAL FORESTS 615 are monodominant in a number of tropical regions (Steenis 1958; Chapman 1976; Janzen 1977). We exclude these from consideration here, however, because the conditions for forest regeneration are different and may be inappropriate for direct comparison with our primary example of Gilbertiodendron forest. If the physical environment, specifically substrate quality, determines the diver- sity of forests in a given area, then spatial shifts in floristic diversity should accompany changes in the underlying soils. Furthermore, if a change in forest type is due to a pronounced change in the physical environment, it is likely that some tree species would be characteristic of just one forest type, either because conditions exclude them from the other or because the competitive balance is so altered that they become rare (Goldberg 1985). Succession It has been hypothesized that species richness might result, in part, from the specialization by tree species to variations in available resources (Ashton 1977). For instance, variation in tree-fall openings may offer an opportunity for speciali- zation to different regeneration niches (Hartshorn 1978; Denslow 1980; Orians 1982; Pickett 1983). Accordingly, forests of low diversity might provide only a limited array of such resources (tree falls) or constitute a successional stage in which specialized species had not yet invaded or reached maturity. Connell et al. (1984) have suggested that tropical forest diversity is promoted by "compensatory mechanisms" that favor survival and reproduction of rarer species over more-common ones. For instance, chemical interference or competi- tion among conspecific juveniles might lead to reduced recruitment of common species, thus allowing coexistence with rarer species. If the greater diversity of the mixed forest relative to the monodominant forest is due to compensatory mechanisms, then either such mechanisms are absent from the monodominant forest or, if the monodominant forest is a sera1 community, the effects of these processes are not yet evident. If monodominance is a successional stage, then a more diverse community would be presaged in the understory. The dominant species would be poorly represented in the smaller-size classes, and the invasion of specialized canopy species would be apparent by their presence in the understory. We would expect this pattern to be pronounced in the vicinity of mixed-forest boundaries, where propagules of other species are most available. Predation Interactions across trophic levels-in particular, predation and herbivory- have been shown to promote diversity in producer assemblages. This has been repeatedly demonstrated in rocky-intertidal communities (Connell 1961, 1978; Paine 1980, 1984; Dayton 1984). Janzen (1970) and Connell (1971) have hy- pothesized a similar effect for predation on seeds and seedlings in tropical forests. This has received some confirmation for specific cases (Boucher 1981; Clark and Clark 1984). If predation on juveniles (or any other factor affecting seed survival and estab- lishment) maintains the greater tree-species diversity of mixed forest relative to 616 THE AMERICAN NATURALIST monodominant forest, we would expect higher seed and seedling mortality in mixed forest and poor subcanopy representation of the canopy species. Con- versely, if the importance of a single species in monodominant forest is a result of freedom from seed predation and herbivory, then we would expect abundant regeneration of the dominant in the understory and low seed and seedling mortal- ity, at least relative to that of the mixed-forest canopy species. Disturbance Recently, species richness has been recognized as a possible function of non- equilibrium conditions (Connell 1978; Huston 1979). Wind-throw disturbances have been implicated in the creation of nonequilibrium conditions in tropical- forest zones (Connell 1978; Hubbell 1979). Whitmore (1978) has suggested that wind-throws give rise to a mosaic of successional stages and consequently con- tribute to rain-forest diversity. Hubbell and Foster (1983) have illustrated how disturbance, in combination with chance immigration, can continue to suppress dominance by any single tropical-forest tree species. Species may coexist, not because of their unique adaptations, but rather because of a combination of fortuitous circumstances. If differences in disturbance regime contribute significantly to differences in diversity between two forests, we would expect this to be evident in a comparison of their structure and composition. Disturbances should be less frequent, less extensive, or both in monodominant forest. In a more diverse forest, we would also expect a relatively greater representation of early-successional species in the canopy.
METHODS AND STUDY SITE We conducted our field study between January 1981 and June 1983 near Epulu in the central Ituri Forest of northeastern Zaire at a transition zone between Gil- bertiodendron forest and mixed forest (fig. l). Gilbertiodendron forest occurs in blocks of tens of square kilometers as well as in patches of only a few hectares. (For further description of study area and climate, see Hart 1985.) We collected forest-composition data from six well-drained upland sites, three in mixed forest (1-3) and three in Gilbertiodendron forest (A-C) (fig. 1). We surveyed eight 625-m2 plots (25 m x 25 m) per site, totaling 24 plots (1.5 ha) in each forest type. All trees greater than 10 cm in dbh (diameter at breast height) were identified and marked with an aluminum tag. We measured their diameter at 1.5 m, or just above the buttresses if these were present. Four of the eight plots at each site were divided into smaller nested plots. We identified saplings (2.5 cm
Epulu Study Area
M~xedForest
- Road
FIG.I .-The Epulu study area, Ituri Forest, Zaire. Plot locations: mixed forest (numerals), monodominant Gilbertiodendron forest (letters).
resulting from that year's seed production. Survivors were counted for the final time 5-7 wk later and included only individuals that had produced seedlings with at least one normal leaf. In most cases, the cause of mortality could be observed directly during periodic plot checks. We verified these observations against re- sults from associated plots from which mammals were excluded (see Hart 1985). We measured early seedling survival during years of large seed production of B. laurentii in mixed forest (1980) and of G. dewevrei in Gilbertiodendron forest (1982). In each case seedlings were censused 3-4 mo after seed dispersal. We counted all new B. laurentii seedlings in eight plots of 3 m x 2 m in mixed forest and all newly germinated G. dewevrei seedlings on six plots of 50 m x 6 m in Gil- bertiodendron forest. Two to 4 soil samples were taken at a depth of 20 cm at each of the six sites, for a total of 22 samples. This depth was chosen because it is below the immediate influence of leachates from the litter yet within the root zone of established seedlings. We collected an additional 11 soil samples at a depth of 150 cm. We analyzed soil texture using the soil-hydrometer method of Brower and Zar (1984). Soil was mixed 1:1 with water for pH readings. Chemical analyses were com- pleted according to the procedures of Dahnke (1980). Phosphorus was extracted with hydrochloric acid in ammonium fluoride and the concentration determined with a colorimeter. Exchangeable cations were extracted with ammonium acetate and their concentration determined with the atomic-absorption spectrophotome- ter at the Michigan State University soil laboratory. THE AMERICAN NATURALIST
TABLE 2
Species Ds H' per 0.5 ha --- FOREST R (SD) R (SD) X (SD)
Gilbertiodendron forest 0.37 (0.10) 0.45 (0.12) 18 (1.93) Mixed forest 0.89 (0.05) 1.35 (0.11) 65 (3.98) P <0.01 <0.001 <0.01
NOTE.-Calculated indexes are Simpson's diversity index (Ds), 1 - X[n,(ni - 1)IN(N - I)], and Shannon's diversity index (H'),1 - C (ni/N) log(n,/N), where n, is the number of individuals of species i and N is the total number of individ- uals. Values shown are mean (X) and standard deviation (SD) for three sites in each forest type; the value for each site was calculated for eight combined plots of 25 m x 25 m. Only stems larger than 10 cm in diameter at breast height were included. Differences between forest types analyzed by Stu- dent's r-test, df = 4.
We recorded size, number, and distribution of tree-fall gaps on four 2.5-ha plots (50 m x 500 m) at each of three sites in both mixed and Gilbertiodendron forests (fig. 1: 1, 2, 4 and A, C, D). A total of 30 ha was surveyed in each forest type. A tree fall was counted if the height of the new growth was less than 3 m and the fallen tree trunk was intact. Only disturbances large enough to have caused canopy gaps were censused. We defined the edge of each gap by the stems of unaffected vegetation; we then judged the approximate shape of the gap and took the measurements necessary to calculate area.
RESULTS Forest Composition Mixed forest is significantly more diverse than Gilbertiodendron forest as mea- sured by Simpson's and Shannon's diversity indexes (table 2). These indexes take into account both the total species richness and the proportion of the total accounted for by each species. Mixed-forest plots (25 m x 25 m) averaged 18 species per plot (>I0 cm in dbh). Gilbertiodendron forest, in contrast, had an average of only 6 species per plot. In Gilbertiodendron forest, a mean of 88% of the total basal area was accounted for by the single species G. dewevrei. In mixed forest, in contrast, the most abundant species, B. laurentii, accounted for a mean of only 32% of the basal area (Student's t = 8.3, df = 4, P < 0.01). Averaging over the three sites for each forest type shows that G. dewevrei in monodominant forest had a mean basal area almost three times that of B. laurentii in mixed forest (table 3). MONODOMINANT TROPICAL FORESTS 619
TABLE 3
SIZE Cleistanrh~ls Cynomerra Klainedoxa Brachysregia Gilbertiodendron CLASS* rnichelsonii alexandri gabonensis laurentii dewevrei
GILBERTIODENDRON FOREST Sapling 0.0 0.0 Pole 0.0 0.0 Tree 1.3 0.7 Large tree 0.0 2.0 Basal areat 0.14 0.64 Standard deviation (0.24) (0.49) MIXED FOREST Sapling 18.5 7.4 Pole 15.3 0.0 Tree 8.0 0.0 Large tree 11.0 2.7 Basal area? 4.99 0.90 Standard deviation
* Size classes: sapling, >2.5 cm in dbh and 510 cm in dbh; pole, 10 cm < dbh 5 25 cm; tree, 25 cm < dbh 5 50 cm; large tree, >50 cm in dbh. ? Basal area (mz/ha)shown is the mean for three sites; the value for each site was calculated for eight combined 625-m2 plots.
Despite their relatively small contribution to total basal area in the monodomi- nant forest, the tree species occurring with G, dewevrei are the same species associated with B. laurentii in mixed forest (table 3). The species with the greatest mean basal area in mixed forest, after B. laurentii, were Cynometra alexandri, Cleistanthus michelsonii Leonard, and Klainedoxa gabonensis Pierre. All three of these species, along with B. laurentii, were also present in Gilbertiodendron forest and contributed the most significant addition to basal area after G, dewevrei. Immature individuals of various sizes of all the important canopy species were found in the subcanopy of both forest types (table 3), suggesting that these species are relatively tolerant of shade. Shade-intolerant pioneer tree species (as defined in Brokaw 1985) are rare in both mixed and monodominant forests. In tree-fall gaps of both forest types 18 species are commonly reported as seedlings (Hart 1985). These species, however, accounted for only a small proportion of all trees greater than 10 cm in dbh, 2.8% in mixed forest and 0.6% in Gilbertiodendron forest. Forest Structure The transition from mixed forest to Gilbertiodendron forest is abrupt and occurs without altitudinal or other apparent change in topography. The canopy of both mixed and monodominant forests attains a height of 30-40 m. The two forests, however, present strikingly different appearances. The Gilbertiodendron forest canopy is homogeneous, formed of the contiguous THE AMERICAN NATURALIST
TABLE 4
STEMDENSITY IN MONODOMINANTGILBERTIODENDRONFOREST AND MIXEDFOREST
MEANNUMBER OF STEMS PLOTS PER HECTARE PLOT PER SIZE SIZE FOREST Gilberrrodendron M~xed CLASS* (m2) TYPE Forest Forest Pt Seedling 25 12 28,000 21,833 NS Sapling 225 12 1,911 2,700 <0.001 Pole 625 24 191 344 <0.001 Tree 625 24 77 96 NS Large tree 625 24 55 37 <0.05
* Size classes: seedling, >0.5 m in height and 5 2.5 cm in dbh; sapling, 2.5 cm < dbh 5 10 cm; pole, 10 cm
TABLE 5
OCCURRENCEAND AREAOF TREE-FALLGAPS ON PLOTSIN MONOWMINANTGILBERTIODENDRON FOREST AND MIXEDFOREST
No. OF GAP PLOT GAPSPER AREA AREA 2.5 ha (mZ) OPENED
n (SD) n (SD) x (SD) Monodominant-forest sites A C D Mixed-forest sites 1 3 4 Comparisons (P)* Within forest type Monodominant Mixed Among forest types
NOTE.-Twelve plots sampled in each forest type (three sites with four plots of 2.5 ha each), for a total of 30 ha in each forest type. Location of sample sites shown in figure I. * Comparisons analyzed by one-way ANOVA with preplanned comparisons; probabilities less than 0.1 are shown; probabilities larger than 0.1 are not significant (NS).
species in mixed forest, decreased in percentage of abundance in the largest-size class (fig. 2). Both species are members of the same tribe of the Caesalpiniaceae, Amher- stiae, and share a number of life-history traits. Their seeds are self-dispersed by the explosive force of the dehiscing legumes. The seeds of both species have thin papery seed coats, and germination generally occurs within 5 days after the seed falls to the ground. The seeds of B, laurentii are regularly dispersed farther from the parent tree than are those of G. dewevrei. Dispersal by propulsion is com- monly more than 30 m from the crown edge for B, laurentii and rarely beyond 6 m for G. dewevrei. Seeds of G, dewevrei are more than twice the diameter of B. laurentii seeds (average 5.5 cm vs. 2.5 cm) and more than six times the dry weight (average 18.2 g vs. 2.8 g) (Hart 1985). Vertebrate dispersal of these seeds is unlikely. The seeds are neither arillate nor enclosed in a fleshy fruit. We recorded mammalian predation by antelope, wild pigs, elephants, and various rodents. Since the seeds either germinate or rot within a week of expulsion from the pod, they are not good candidates for scatter hoarding; nor was any evidence found of seed removal and storage by rodents (Hart 1985). Once the first true leaves are fully expanded, seedlings of both G. dewevrei and B. laurentii are relatively free of herbivory. Seed Predation and Seedling Survival There were high levels of predation on seeds of both G. dewevrei and B. laurentii;however, mortality was significantly greater for G. dewevrei seeds in the THE AMERICAN NATURALIST
(I-,w. ;"01
Class: sapling pole tree large tree
FIG. 2.-Percentage of total stems by size class (by diameter at breast height in cm) for Gilberriodendron dewevrei in monodominant forest (open bars) and Brachysregia laurentii in mixed forest (crosshatched bars). n, Number of sample plots.
TABLE 6
ORIGINAL SEEDSSURVIVING TO NUMBER SEEDLINGSTAGE, 92 OF SEEDS SPECIES PLOTS PER PLOT* x (SD)
Gilbertiodendron dewevrei 25 15 9.07 (17.51) Brachysregia laurenrii 10 50 47.20 (20.00) * Seeds of G. denlevrei placed in monodominant forest; seeds of B. laurentii placed in mixed forest. monodominant forest than for B. laurentii seeds in the mixed forest (Mann- Whitney U-test, z = 4.01, P < 0.001; table 6). Insects were the primary cause of seed mortality for both species; two curculionid species attacked G. dewevrei, and at least one bruchid species attacked B. laurentii. Mammals were a secondary cause of mortality. Fungi usually invaded only seeds that had been damaged by insects. The seeds of both G. dewevrei and B. laurentii germinate and become estab- lished in the shade of the parent tree. Surveys in the understory of Gilbertioden- dron forest show high seed mortality for G. dewevrei in 1982 (table 7). The average density of surviving 1982 seedlings in March 1983, 4 mo after seedfall, was 0.02 seedlings per square meter. The density of older G. dewevrei seedlings that were, nevertheless, still less than 0.5 m in height was 0.8 seedlings per square meter. This was 40 times that of first-year G. dewevrei seedlings, and larger seedlings (0.5 m in height to 2.5 cm in dbh) were even more dense (1.07 seedlings1m2). The density of first-year B. laurentii seedlings (13.71m2) for a single year's MONODOMINANT TROPICAL FORESTS 623
TABLE 7
GILBERTIODENDRON DEWEVREI IN MONODOMINANTFOREST
No. of No. of TREESIZE x (SD) Plots x (SD) Plots
First year 0.02 (0.02) 6 13.74 (11.03) 8 2 2 yrand 5 0.5 m high 0.80 (0.43) 12 1.55 (1.20) 12 > 0.5 m high and 5 2.5 cm dbh* 1.07 (0.86) 12 0.15 (0.23) 12
* dbh, Diameter at breast height
seedfall (1980) was significantly greater than that of G. dewevrei seedlings (table 7). The density of B, laurentii seedlings decreased with age and larger-size classes. Small seedlings (more than 1 yr old but still less than 0.5 m in height) occurred at only 11% the density of first-year seedlings (1.55 seedlings/m2). Larger seedlings (>0.5 m in height and 52.5 cm in dbh) were even sparser (0.15 seedlings/m2). Thus, G. dewevrei and B. laurentii exhibited opposite trends in density with increasing seedling size and age (table 7). Larger seedlings of G. dewevrei had the greatest density of all seedling size classes, whereas larger seedlings of B. laurentii had the lowest density. Soils The soils under plots of both mixed and Gilbertiodetzdron forests ranged in color from reddish brown through ocher to yellowish brown. In all sample pits, soils were deep, uniform in texture, and lacking distinct horizons from approxi- mately 3 cm to at least 100 cm in depth. Such uniformity is characteristic of highly weathered tropical oxisols (Sanchez 1976). Of 12 1.5-m pits, bedrock was encoun- tered only once, at 135 cm in Gilbertiodetzdron forest. There was no significant difference between mixed forest and Gilbertiodetzdron forest for most soil factors tested (table 8). Soil texture did not differ significantly between forest types. Soils in Gilbertiodendron forest ranged from loamy sand to sandy clay, with the majority of samples being either sandy loam or sandy clay loam. Mixed-forest soils ranged from sandy clay loam to sandy clay with most falling in the former category. Both forests occurred on acid soils. The mean soil pH at a depth of 20 cm was lower in mixed forest than in Gilbertiodendron forest (table 8). This difference may indicate an effect of leachates from the distinctive and relatively thick leaf litter characteristic of Gilbertiodendron forest. At 150 cm in depth, the difference in soil pH between the forest types was no longer significant. Overall, soil nutrient levels are low in both forest types. Phosphorus and magnesium were not detectable in either forest type by the methods used. The significant difference in potassium content of soils at 150 cm in depth is unex- plained. The four sampling points for mixed forest at this depth are all within a few kilometers of one another. The difference between mean values of all measured soil variables for the two THE AMERICAN NATURALIST
TABLE 8 SOILPROPERTIES
Sample Size and Soil Gilbertiodendron Mixed Variables Forest Forest P*
20-Cm DEPTH Sample pits 12 pH 4.17 % sand 71.7 K mdg 0.09 Ca mdg 0.13 Mg mdg nd P mdg nd
Sample pits pH % sand K mgk Ca mdg Mg mdg P mdg NOTE.--Mean values are shown. nd, Levels that were not detected. The Bray-1 technique that was used does not detect iron-bonded phosphates. * Comparison was by Student's I-test. NS, Not significant (P > 0.05). forest types at 20 cm in depth was tested simultaneously (Hotelling's T*). NO significant difference between the soils was found (F = 1.443, df = 4,17, P = 0.263). Observations of earlier researchers indicate that mycorrhizal relations may be pertinent to the mineral nutrition of the major canopy species of both forest types, although the impact of mycorrhizae on dominance is not certain. Bonnier (1957) found that neither G. dewevrei nor B, laurentii formed nitrogen-fixing nodules in undisturbed forest, although at least the former species did so in some altered habitats. Gilbertiodendron dewevrei, however, did form extensive ectomycor- rhizal root sheaves (Bonnier 1957). No specific study of B, laurentii has been made, but four other species of the same genus have been reported to support ectomycorrhizal fungi (Hogberg and Piearce 1986).
DISCUSSION Forest Dominated by 'Gilbertiodendron dewevrei' A comparative study of composition, structure, and substrates of monodomi- nant and adjacent mixed forests in the Ituri region of Zaire did not support any of the suggested mechanisms for the maintenance of species diversity in the mixed- forest type. If the significant difference in species diversity between mixed and Gilber- MONODOMINANT TROPICAL FORESTS 625
tiodendron forests is mediated by a change in substrate quality, then the soil characteristics of the two communities should differ. The textural measures and most chemical measures of soil samples from Gilbertiodendron and mixed forests in the Ituri region reveal neither abrupt changes nor a more gradual gradient between the two forest types. The differences noted (pH and potassium) should be viewed in the light of results from extensive soil and vegetation studies of Gilber- tiodendron and mixed forests outside the Ituri region. Gilbertiodendron forest has been recorded on a wide variety of soils. In the Uele region, it is found on red clay soils and yellow gravelly soils with a sandy- clay matrix (Frankart 1960). On the Kivu plateau, soils supporting Gilbertioden- dron forest range from heavy clay to yellow, sandy alluvial soils (Pecrot and Leonard 1960). In the Ubangi, this forest is found on red, sandy, clay soil as well as on soils containing numerous chips of dismantling lateritic shield (Jongen et al. 1960). At Yangambi (on the Zaire River), Gilbertiodendron forest occurs on white sands and on yellow, ocher, to brown sandy soils with variable clay content of up to 40% (Louis 1947; Wambeke and Evrard 1954). The parent materials of soils underlying Gilbertiodendron forests are also di- verse. Over the central and northern range of Gilbertiodendron forests (Ituri, Uele, Ubangi), the parent material is Precambrian eruptive and metamorphic rock. Decomposing lateritic shield is an important component at the northern limits of the forest type (in Ubangi and Uele). At the western limits of the species' range, the underlying substrate is aeolian sand deposits of Pleistocene age; at its southern limit, soils were formed from sedimentary rocks of Carboniferous and Permian ages (Cahen and Lepersonne 1948; Heinzelin 1952; Wambeke and Evrard 1954; Gerard 1960; Jongen et al. 1960; FAO/UNESCO1976). In the preceding cartographic studies, where both vegetation and soils were mapped, the borders of G. dewevrei-dominated forest failed to follow the outlines of soil types. Nevertheless, in all of the mapped areas, blocks of Gilbertiodendron forest coexisted with blocks of more-diverse mixed forest. This earlier work thus concurs with our study in failing to find any correlation between Gilbertiodendron forest and either specific soil types or other substrate factors. Furthermore, a difference in substrate quality extreme enough to exclude the dominant species of one forest type from an adjacent forest type might be ex- pected to affect more than a single species (Goldberg 1985). For instance, a poor substrate might have a characteristic flora in which several species found on a nearby richer substrate are not represented. This is not the case for the mono- dominant and mixed forests of the Ituri. Except for the near absence of G. dewevrei from mixed forest, the same tree species were represented in both mixed and monodominant forest types, although at different densities. A detailed survey of monodominant Gilbertiodendron forest in the Uele region (Gerard 1960), at a site 450 km northwest of the Ituri study site, reveals a similar pattern. In the Uele, the canopy species that co-occur with G. dewevrei in the monodominant forest are also the best-represented species in the adjacent mixed forest. Two of the most important of these species, Julbernardia seretii (Caesal- piniaceae) and Staudtia stipitata Warb. (Myristicaceae), are absent or extremely rare in the central Ituri forest. Likewise, two prominent tree species in the mixed 626 THE AMERICAN NATURALIST forest of the Ituri study site, Cleistanthus michelsonii (Euphorbiaceae) and Brachystegia laurentii (Caesalpiniaceae), are not represented in the Uele forest. The monodominant Gilbertiodendron forests of both the Uele and the Ituri regions are remarkably similar in the percentage of dominance by G. dewevrei and in overall forest structure, despite differences in the identity of subdominant species between the two study sites. Both the total basal area and the predomi- nance of G. dewevrei across all size classes in the Uele forest correspond closely to our findings in the Ituri Gilbertiodendron forest. In fact, any references in the literature to the composition of forests containing G. dewevrei suggest that, where the species occurs, it is dominant in the understory as well as in the canopy (Louis 1947; Jongen et al. 1960; Letouzey 1970; Gauthier et al. 1977). In summary, although G. dewevrei's dominance is similar for all forests where it occurs, total species composition is not. Species occurring in a given Gilbert- iodendron forest are more likely to be the same as those in a nearby mixed forest, where G, dewevrei does not occur, than those in another Gilbertiodendron forest at a distance of several hundred kilometers. Tropical sera1 communities, such as the storm forest in Kelantan (Wyatt-Smith 1954), may be dominated by a single tree species. If the Gilbertiodendron forest is a successional community, we would expect G. dewevrei to have characteristics of a pioneer species and to be replaced in the understory by juveniles of mixed- forest species. Gilbertiodendron dewevrei, however, is tolerant of shade and well represented in all understory size classes. Indeed, all tree species contributing appreciably to Gilbertiodendron-forest basal area are well represented as ju- veniles, with the most abundant adults having the most abundant regeneration. Thus, there is no evidence for an effect of compensatory mechanisms or for the establishment of more-specialized species at the expense of the dominant. There is no reason to suspect that G. dewevrei's dominance will not continue through subsequent generations. Abundant regeneration of the dominant species is equally prevalent in smaller islands of Gilbertiodendron forest surrounded by mixed forest, where seed rain from mixed forest is greatest (Hart 1985). We suggest that species-specific predation on the juveniles of canopy species could contribute to the greater diversity in mixed forest than in Gilbertiodendron forest. If predation, or other causes of early juvenile mortality, determine the diversity of canopy species in mixed forest, we would expect mortality of seeds and young seedlings in the vicinity of parent trees to be greater in mixed forest than in Gilbertiodendron forest. This was not the case. Predators destroyed a greater percentage of G. dewevrei seeds in monodominant forest than of B. laurentii seeds in mixed forest. Similarly, first-year seedlings of B. laurentii in mixed forest occurred at higher densities than seedlings of G. dewevrei of the same age in monodominant forest. Gilbertiodendron dewevrei is nevertheless well represented in the understory of the monodominant forest. Its shade-tolerant seedlings accumulate from one year to the next. The greater density of the larger- seedling size classes supports the observation that seedling persistence, rather than escape from predators, determines the abundance of G. dewevrei in the monodominant-forest understory. Our method of measuring the relative abundance of tree-fall gaps allowed us to MONODOMINANT TROPICAL FORESTS 627 adequately sample only relatively small tree falls involving one to several trees. These were more frequent in mixed forest than in the monodominant Gilbert- iodendron forest; however, the low proportion of gap-regenerating pioneer species that reached maturity in either forest indicates that the difference in gap frequency alone is not sufficient to explain the difference in the forests' diversity and composition. Nevertheless, the more numerous gaps in mixed forest add to the more broken aspect of the mixed-forest canopy and may allow more light to penetrate to the forest floor. Thus, seedlings of lower shade tolerance may survive better in mixed forest than in Gilbertiodendrotz forest. This would permit a greater diversity of persistent seedlings awaiting openings in the forest canopy. The frequency and impact of large disturbances could not be directly measured by our methods. During the 4 yr we were in the central Ituri Forest, however, two storms caused a series of blowdowns of several hectares each in the vicinity of the study area. An older blowdown, found within 10 km of our base camp, was over 100 ha in size. Disturbances on this scale may control forest composition at the transition from mixed to monodominant forest. A sample from a buried charcoal layer, found under areas of the Ituri Forest that do not burn today, has been given a preliminary age of 2290 yr * 90 (Hart and Hart 1986). More information about major disturbances of past millennia may be essential for developing an explana- tion for the current distribution of forest types. Although this study did not identify a single cause responsible for the coexis- tence of monodominant Gilbertiodendrotz forest and diverse mixed forest, it did establish that monodominance in the Ituri Forest depends on the presence of a particular species, G. dewevrei. The borders between the monodominant and mixed forest are, in fact, very nearly determined by the presence or absence of this single species. All observations of G. dewevrei indicate that where this species becomes established it dominates all size classes. Other associated tree species become, consequentially, less abundant and have lowered potential for regeneration. In comparison with common mixed-forest species, G. dewevrei is notable for its large seeds of very limited dispersibility. Thus, even without any intervening factor, the advance of Gilbertiodendron forest on mixed forest would be very slow, as measured in human or even tree lifetimes. The widespread establishment of monodominant forest would be favored only by long periods of low distur- bance. Large-scale disturbance (as promoted by a drying climatic trend) would favor less shade-tolerant, more vagile species. By this line of reasoning, areas of central Africa that lost their forest cover during Pleistocene dry periods (Liv- ingstone 1980a,b; Hamilton 1982) may have been colonized by and maintained mixed-forest types long after their climate ameliorated. The phenomenon of single-species dominance, at least by G. dewevrei, may thus depend on certain life-history characteristics of this species and on the long-term absence of major disturbance. Other Motzodominaizt Tropical Forests In other tropical areas, the principal species of monodominant forests of well- drained soils include members of several families (table 9). Caesalpiniaceae are TABLE 9
PANTROPICCOMPARIIONOF MONODOMINANTFORESTS OF WELL-DRAINEDSOILS - -- Dommant Spec~es Ramfall S~ze Substrate Assoc~ated Shade Locat~on (Fam~ly) (cmlyr) of Stands Dom~nance Spec~fic~ty Spec~es D~spersal Tolerance Source
Malaya Shoreu currisii >200 groves on excludes hill crests similar to poor dis- persistent Whitmore (Dipterocarpaceae) hills other lowland persal seedlings 1984 domi- dipterocarp Burgess 1969 nants forest Malaya, Dryobu1anop.s >250 several 60%-90% wide range similar to poor dis- shade- Whitmore 1984 Sumatra urornatica thousand of timber of soils adjacent persal tolerant Lee 1967 (Dipterocarpaceae) kmz trees and topog- dipterocarp raphy forest Borneo, Eusideroxylon zwa- >200 several pure wide range ? large, shade- Koopman & Sumatra geri (Lauraceae) thousand stands of soils heavy tolerant Verhoef 1938 km2 seeds Trinidad Mora excelsa >200 largest 85%-95% wide range similar to falls be- shade- Beard 1946 (Caesalpiniaceae) stand canopy of soils adjacent neath tolerant >210 and topog- mixed for- parent kmz raphy est India Poeciloceuron >500 2-4 km2 almost similar to ad- similar to 9 persistent Kadambi 1942 puuc@orurn (Gut- pure jacent for- adjacent seedlings tiferae) est types forest types East Cynornetra ulexandri about >75% variable soils shared with poor shade-toler- Eggeling 1947 Africa (Caesalpiniaceae) 180 canopy mixed for- dispersal ant as est saplings and poles West Talboriella gentii about pure variable soils few associ- shade- Swaine & Africa (Caesalpiniaceae) 100 stands and parent ated can- tolerant Hall 1981 material opy species Central Gilbertiodendron about largest >90% wide range shared with very poor shade- Gtrard 1960 Africa dewevrei 180 stand > canopy of soils adjacent dispersal tolerant Louis 1947 (Caesalpiniaceae) 100 km2 and topog- mixed for- raphy est Central Julbernardia seretii about 3 >90% shallow soils shared with shade- Gtrard 1960 Africa (Caesalpiniaceae) 180 canopy adjacent tolerant mixed for- est MONODOMINANT TROPICAL FORESTS
especially well represented in Africa but also occur in the Neotropics (e.g., Mora excelsa). Members of the Dipterocarpaceae, the Lauraceae, and the Guttiferae are dominant species in various monodominant Asian forests. The climatic information available indicate that monodominant evergreen for- ests occur under a variety of rainfall regimes in both aseasonal humid areas (e.g., Eusideroxylon zwageri T. & B.) and seasonal areas (e.g., Poeciloceuron paucifio- rum). These forests occur as small stands of several hectares (as with Talbotiella gentii Hutch. & Greenway) or cover thousands of hectares (as does Mora ex- cels~).In all cases the dominant species was reported to be tolerant of shade and well represented in the understory. This is strong evidence for the maturity of these forests, all of which were considered self-replacing climax types by the authors of the respective studies. Monodominant forests generally are not restricted to poor substrates. Edaphic factors may limit the distribution of some of these forests (Shorea curtisii Dyer ex King and Julbernardia seretii); however, others span a diversity of substrate conditions, such as forests dominated by E. zwageri, M. excelsa, T. gentii, and C. alexandri, as well as G. dewevrei. In some cases, restriction of monodominant forests to specific soil types had been assumed but not borne out by detailed study (e.g., forests dominated by Dryobalanops aromatica Gaertn. in Malaysia; Lee 1967). The published descriptions of monodominant forests do not support the view that these forests are distinct associations with characteristic species not found in other forest types. As was the case for the Gilbertiodendron forests, the species co-occurring with the canopy dominant are usually also found in adjacent, more diverse, mixed forest. Beard (1946), for instance, found that the species associ- ated with Mora excelsa in Trinidad varied with location but were always the same as in the adjacent mixed forest. It is tempting to suggest that all these tropical, monodominant forests of well- drained upland sites constitute a cohesive forest type. This would imply that in all cases similar mechanisms led to the dominance by a single species. Most of the species listed in table 9 share several key life-history characteristics with G. dewevrei. In particular, all species for which information was found produce large, poorly dispersed seeds, giving rise to shade-tolerant, persistent seedlings. A number of the studies reported that the boundaries of the monodominant forest were not stable and that they appeared to be steadily encroaching on adjacent mixed forest (Beard 1946; Eggeling 1947; Louis 1947). This supports the proposition that the current boundaries of monodominant forests, in some areas, reflect no more than the slow rate of dispersal of the dominant species following reduction in its range by ancient, large-scale disturbance or climatic change. Our review of the literature has shown that the dominant species of many monodominant forests throughout the tropics share a suite of characteristics, suggesting that the communities they form have been free of major disturbance over relatively long periods of time. The large seeds and pronounced shade tolerance of the juveniles have apparently evolved in conjunction with reduced dispersal capabilities and slow growth. The traits that allow these tree species to THE AMERICAN NATURALIST persist in the shade of their own canopies are associated with traits that preclude these species' rapid recovery and spread after major disturbance. The scenario presented here supports nonequilibrium explanations for the high species diversity of many tropical humid forests of well-drained sites. The lower diversity that results from the high representation by a single species requires the absence of large-scale disturbance for long periods of time, along with the pres- ence of a species having the appropriate suite of characteristics to allow establish- ment and eventual dominance.
SUMMARY A study of the structure and floristics at a transition zone from a monodominant to a more diverse forest in the African humid tropics was conducted to elucidate the mechanisms maintaining floristic diversity and the discontinuity between mixed forests and forests dominated by a single tree species (monodominant). The mixed forest's greater diversity could not be explained by substrate differences, greater maturity, or greater predation on seeds or juveniles. The dominant species of the monodominant forest was shade-tolerant and had poorly dispersed seeds. Tree species associated with the dominant were also found in the mixed forest. Monodominant and mixed forests occur side by side in the Asian and American tropics as well. As in the African example, many of these monodominant forests share most species with neighboring mixed forests. Characteristically, the domi- nant species have large seeds and shade-tolerant seedlings. Monodominant trop- ical forests are widespread and may indicate areas that have not experienced large-scale disturbance for long periods. Subsequent to major disturbance, it is likely that such forests regenerate and spread more slowly than mixed-forest associations.
ACKNOWLEDGMENTS We thank the Institut Zairois pour la Conservation de la Nature for authorizing our fieldwork in the Ituri Forest. Field research was supported by two grants from the U.S. Department of Agriculture Forest Service through the U.S. Man and the Biosphere Program (4789-4 and 4789-6). Further support was provided by Wildlife Conservation International, a division of the New York Zoological Society. We appreciate the helpful comments provided by J. Connell, G. Hartshorn, D. Jan- Zen, D. Livingstone, T. Miller, A. Wynn, and three anonymous reviewers.
LITERATURE CITED Ashton, P. S. 1971. The plants and vegetation of Bako National Park. Malay. Nat. J. 24:151-162. -. 1977. A contribution of rain forest research to evolutionary theory. Ann. Mo. Bot. Gard. 64:694-705. Beard, J. S. 1946. The Mora forest of Trinidad, British West Indies. J. Ecol. 33:173-192. Bonnier, C. 1957. Symbiose Rlzizobi~on-ICgumineuses en region equatoriale. INEAC (Inst. Natl. Etude Agron. Congo) Ser. Sci. no. 72, Brussels. Boucher, D. H. 1981. Seed predation by mammals and forest dominance by Q~rerc,~c~olr.oides. a tropical lowland oak. Oecologia (Berl.) 49:409-414. MONODOMINANT TROPICAL FORESTS 631
Brokaw, N. V. L. 1985. Treefalls, regrowth, and community structure in tropical forests. Pages 53-69 in S. T. A. Pickett and P. S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, New York. Brower, J. E.. and J. H. Zar. 1984. Field and laboratory methods for general ecology. 2d ed. W. C. Brown, Dubuque, Iowa. Brunig, E. F. 1973. Species richness and stand diversity in relation to site and succession of forests in Sarawak and Brunei (Borneo). Amazoniana 4:293-320. Burgess, P. F. 1969. Preliminary observations on the autecology of Shorru crorisii Dyer ex King in the Malay Peninsula. Malays. For. 32:438. Cahen, L., and J. Lepersonne. 1948. Atlas general du Congo et du Ruanda-Urundi. no. 31. Notice de la carte geologique du Congo Belge et du Ruanda-Urundi. Institut Royal Colonial Belge, Brussels. Chapman, V. J. 1976. Mangrove vegetation. J. Cramer, Vaduz, Liechtenstein. Clark. D. A,, and D. B. Clark. 1984. Spacing dynamics of a tropical rain-forest tree: evaluation of the Janzen-Connell model. Am. Nat. 124:769-'788. Connell, J. H. 1961. The influence of interspecific competition and other factors on the distribution of the barnacle Chthatnulrrs src,llarus. Ecology 42:710-723. -. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. Pages 298-312 in P. J. den Boer and G. R. Gradwell, eds. Dynamics of populations. Proceedings of the Advanced Study Institute on Dynamics of numbers in populations, Oosterbeek, The Netherlands, September 7-18, 1970. Centre for Agricultural Publishing and Documentation, Wageningen, The Netherlands. -. 1978. Diversity in tropical rain forests and coral reefs. Science (Wash., D.C.) 199: 1302-1309. Connell, J. H.. and M. D. Lowman. 1989. Low-diversity tropical rain forests: some possible mecha- nisms for their existence. Am. Nat. 134 (in press). Connell, J. H., J. G. Tracey, and L. J. Webb. 1984. Compensatory recruitment, growth, and mortality as factors maintaining rain forest tree diversity. Ecol. Monogr. 54: 141-164. Dahnke, W. C., ed. 1980. Recommended chemical soil test procedures for the North Central Region. Bull. 499. North Dakota Agricultural Experiment Station. Fargo. Davis. T. A. W., and P. W. Richards. 1933. The vegetation of Moraballi Creek. British Guiana: an ecological study of a limited area of tropical rain forest. Part I. J. Ecol. 21:350-384. -. 1934. The vegetation of Moraballi Creek. British Guiana: an ecological study of a limited area of tropical rain forest. Part 2. J. Ecol. 2?:106-133. Dayton, P. K. 1984. Processes structuring some marine communities: are they general'! Pages 181-197 in D. A. Strong, Jr., D. Simberloff. L. G. Abele, and A. B. Thistle, eds. Ecological com- munities: conceptual issues and the evidence. Princeton University Press, Princeton, N.J. Denslow, J. S. 1980. Gap partitioning among tropical rain forest trees. Biotropica 12 (Suppl.):47-55. Devred, R. 1958. La vegetation forestiere du Congo Belge et du Rwanda-Burundi. Bull. Soc. R. For. Belg. 6:408-468. Eggeling, W. J. 1947. Observations on the ecology of Budongo rain forest, Uganda. J. Ecol. 34:20-87. FAOIUNESCO.1976. Carte mondiale des sols. Vol. 6. Afrique. UNESCO.Paris. Frankart. R. 1960. Carte des sols et de la vegetation du Congo Belge et du Ruanda-Urundi. Vol. 14. Uele. INEAC (Inst. Natl. Etude Agron. Congo). Brussels. Gauthier, Poulin, and Theriault. Ltd. 1977. Manuel de dendrologie. Prepared for the government of the Republic of Zaire in accord with the program of cooperation of the Canadian Agency for International Development. Quebec. Gentry. A. H. 1982. Patterns of Neotropical plant species diversity. Evol. Biol. 15: 1-84. Gerard. P. 1960. Etude ecologique de la foret dense a Gilherriodendron deu.e~.reidans la region de I'Uele. INEAC (Inst. Natl. Etude Agron. Congo) Ser. Sci. no. 87. Brussels. Germain, R.. and C. Evrard. 1956. Etude ecologique et phytosociologique de la foret a Bruc,h?stegirr laurenrii. INEAC (Inst. Natl. Etude Agron. Congo) Ser. Sci. no. 67, Brussels. Goldberg. D. E. 1985. Effects of soil pH, competition. and seed predation on the distribution of two tree species. Ecology 66:503-511. Hall, J. B., and M. D. Swaine. 1981. Distribution and ecology of vascular plants in a tropical rain forest: forest vegetation in Ghana. Junk, Boston 632 THE AMERICAN NATURALIST
Hamilton, A. C. 1981. A field guide to Uganda forest trees. Makerere University Press, Kampala, Uganda. -. 1982. Environmental history of East Africa: a study of the Quaternary. Academic Press, New York. Hart, T. 1985. The ecology of a single-species-dominant forest and of a mixed forest in Zaire, Africa. Ph.D, diss. Michigan State University, East Lansing. Hart, T., and J. Hart. 1986. Ecological basis of hunter gatherer subsistence in the African rain forest: the Mbuti of eastern Zaire. Hum. Ecol. 14:29-55. Hartshorn, G. S. 1978. Treefalls and tropical forest dynamics. Pages 617-638 in P. B. Tomlinson and M. H. Zimmermann, eds. Tropical trees as living systems. Cambridge University Press, Cambridge. Heinzelin, J, de. 1952. Sols, palCosols et dCsertifications anciennes dans le secteur nord-oriental du bassin du Congo. INEAC (Inst. Natl. Etude Agron. Congo), Brussels. Hogberg, P., and G. D. Piearce. 1986. Mycorrhizae in Zambian trees in relation to host taxonomy, vegetation type and successional patterns. J. Ecol. 74:775-785. Hubbell, S. P. 1979. Tree dispersion, abundance, and diversity in a tropical dry forest. Science (Wash., D.C.) 203:1299-1309. Hubbell, S. P., and R. B. Foster. 1983. Diversity of canopy trees in a Neotropical forest and implications for conservation. Pages 24-41 in S. L. Sutton, T. C. Whitmore, and A. C. Chadwick, eds. Tropical rain forest: ecology and management. Blackwell, Oxford. Huston, M. 1979. A general hypothesis of species diversity. Am. Nat. 113:81-101. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. Am. Nat. 104: 501-528. -. 1974. Tropical blackwater rivers, animals and mast fruiting by Dipterocarpaceae. Biotropica 6:69-103. -. 1977. Promising directions of study in tropical animal-plant interactions. Ann. Mo. Bot. Gard. 64:706-736. Jongen, P., M. van Oosten, C. Evrard, and J.-M. Berce. 1960. Carte des sols et de la vBgCtation du Congo Belge et du Ruanda-Urundi. Vol. 11. Ubangi. INEAC (Inst. Natl. Etude Agron. Congo), Brussels. Kadambi, K. 1942. The evergreen Ghat rain forest of the Tunga and Bhadra river sources, Kadur District, Mysore State. Parts 1, 2. Indian For. 68:233-240, 305-312. Koopman, M. J. F., and L. Verhoef. 1938. Eusideroxylon zwageri T. & B., het ijzerhout van Borneo en Sumatra. Tectona 31:381-399. Laan, E. van der. 1927. De analyse der bosschen van de eilanden poeloe laoet en seboekoe der residentie zuider en oosterafdeeling van Borneo. Tectona 20: 19-36. Lebrun, J., and G. Gilbert. 1954. Une classification Ccologique des forCts du Congo. INEAC (Inst. Natl. Etude Agron. Congo) Ser. Sci, no. 63, Brussels. Lee, P. C. 1967. Ecological studies on Dryobalanops aromatica Gaertn. Ph.D. diss. University of Malaysia, Kuala Lumpur. Leigh, E. G., Jr. 1982. Introduction: why are there so many kinds of tropical trees? Pages 63-66 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest, seasonal rhythms and long-term changes. Smithsonian Institution Press, Washington, D.C. LemCe, G. 1961. Effets des caracteres du sol sur la localisation de la vCgCtation en zones Cquatoriale et tropicale humide. Pages 25-39 in Proceedings of the symposium on tropical soils and vegeta- tion, Abidjan. UNESCO,Paris. Letouzey, R. 1970. Manuel de botanique forestiere: Afrique tropicale. Vol. 2A. Centre Technique Forestier Tropical, Nogent-sur-Marne, France. Livingstone, D. A. 1980a. Environmental changes in the Nile headwaters. Pages 339-359 in M. A. J. Williams and H. Faure, eds. The Sahara and the Nile: Quaternary environments and prehis- toric occupation in northern Africa. Balkema, Rotterdam, The Netherlands. -. 1980b. History of the tropical rain forest. Paleobiology 6:243-244. Louis, J. 1947. Contribution A I'ttude des forCts Cquatoriales congolaises. Pages 902-915 in Comptes rendus de la Semaine agricole de Yangambi. INEAC (Inst. Natl. Etude Agron. Congo), Brussels. MONODOMINANT TROPICAL FORESTS
Orians. G. 1982. The influence of tree-falls in tropical forests on species richness. Trop. Ecol. 23: 255-279. Paine, R. T. 1980. Food webs: linkage, interaction strength, and community infrastructure. J. Anim. Ecol. 49:667-685. -. 1984. Ecological determinism in the competition for space. Ecology 65: 1339-1348. Pecrot, A.. and A. Leonard. 1960. Carte des sols et de la vCgCtation du Congo Belge et du Ruanda- Urundi. Vol. 16. Dorsale du Kivu. INEAC (Inst. Natl. Etude Agron. Congo), Brussels. Pickett, S. T. A. 1983. Differential adaptation of tropical tree species to canopy gaps and its role in community dynamics. Trop. Ecol. 24:68-84. Rai, S. N., and J. Proctor. 1986. Ecological studies on four rain forests in Karnataka, India. I. Environment, structure, floristics and biomass. J. Ecol. 74:439-454. Rankin, J. M. 1978. The influence of seed predation and plant competition on tree species abundances in two adjacent tropical rain forests. Ph.D. diss. University of Michigan, Ann Arbor. Richards, P. W. 1952. The tropical rain forest, an ecological study. Cambridge University Press, Cambridge. -. 1961. The types of vegetation of the humid tropics in relation to the soil. Pages 15-23 in Proceedings of the symposium on tropical soils and vegetation, Abidjan. UNESCO,Paris. Sanchez, P. A. 1976. Properties and management of soils in the tropics. Wiley, New York. Steenis, C. J. van. 1958. Ecology of mangroves. Flora Malesiana Ser. 1, Spermatophyta 5:429-448. Swaine, M. D., and J. B. Hall. 1981. The monospecific tropical forest of the Ghanaian endemic tree, Talbotiella gentii. Pages 355-363 in H. Synge, ed. The biological aspects of rare plant conservation: proceedings of an international conference. King's College, Cambridge, July 14-19, 1980. BSBI (Bot. Soc. Br. Isles) Conf. Rep. 17. Wiley, Chichester. Wambeke, A. van, and C. Evrard. 1954. Carte des sols et de la vCgCtation du Congo Belge et du Ruanda-Urundi. Vol. 6. Yangambi. INEAC (Inst. Natl. Etude Agron. Congo), Brussels. Whitmore, T. C. 1978. Gaps in the forest canopy. Pages 639-655 in P. B. Tomlinson and M. H. Zimmermann, eds. Tropical trees as living systems. Cambridge University Press, Cambridge. -. 1984. Tropical rain forests of the Far East. 2d ed. Clarendon, Oxford. Wyatt-Smith, J. 1954. Storm forest in Kelantan. Malay. For. 17511. Zon, P. van. 1915. Mededeelingen omtrent den kamferboom (Dryobalanops aromatics). Tectona 8:220-224.
(EDITOR'SNOTE.-A related article by J. H. Connell and M. D. Lowman will be published in the July