Biodiversity and Conservation 13: 2399–2417, 2004.
# 2004 Kluwer Academic Publishers. Printed in the Netherlands.
Mesoscale transect sampling of trees in the
Lomako–Yekokora interfluvium, Democratic
Republic of the Congo
JEAN PHILIPPE BOUBLI1,2,*, JONAS ERIKSSON3, SERGE WICH4,
GOTTFRIED HOHMANN3 and BARBARA FRUTH5
1
Estação Biológica de Caratinga, Caixa Postal 82, Ipanema, MG 36950-000, Brazil; 2Zoological
Society of San Diego, P.O. Box 120551, San Diego, CA 92112-0551, USA; 3Max-Planck-Institut für
Evolutionäre Anthropologie, Inselstraße 22, D-04103 Leipzig, Germany; 4Ethology and Socioecology
Group, Utrecht University, P.O. Box 80086, 3508 TB Utrecht, The Netherlands; 5Max-PlanckInstitut für Verhaltensphysiologie, D-82319 Seewiesen, Germany; *Author for correspondence
(e-mail: [email protected])
Received 16 October 2002; accepted in revised form 13 August 2003
Key words: Africa, Democratic Republic of the Congo, Floristics, Transect sampling, Tropical
trees
Abstract. We conducted a mesoscale transect sampling of trees 10 cm DBH in the Lomako–
Yekokora interfluvial forest, Democratic Republic of the Congo. Our objective was to characterize
the forest landscape contained between the Lomako and Yekokora rivers in terms of its floristic
composition and to investigate how representative the Lomako study site, the location of a longterm study of primates, was of the entire forest block. Fifteen transects were laid out at seven
sample stations placed approximately 10 km apart and alongside a 70 km trail running from the
Lomako study site to the margins of the Yekokora river. Three transects totaling 3.65 ha were laid
out at the Lomako study site and two transects totaling 2 ha at each of the remaining six sample
stations, amounting to 15.65 ha in total. Average DBH, tree density, tree species richness and
floristic composition were determined for each transect. There were 5353 trees 10 cm DBH in the
total sample, representing 150 species in 35 families. Caesalpinoideae trees dominated the sample,
followed by Olacaceae and Annonaceae. Four forest types were identified: mixed primary (57% of
the sampled plots), secondary forest (9%), Gilbertiodendron (22%), and swamp (12%). The seven
sample stations differed from each other in average DBH, tree density, tree species richness and
floristic composition. Most of the difference, however, was due to the fact that the four forest types
were not equally represented at each sample station. When forest types were contrasted independently, a marked difference in average DBH, tree density, tree species richness and floristic
composition was recorded. Conversely, when only mixed primary forest was analyzed across the
sample stations, no significant difference was detected except for average DBH. Thus the Lomako
study site is representative of the forest landscape contained between the Lomako and Yekokora
rivers only when the different forest types are treated separately. The sample stations (including
Lomako) differ from each other, however, in the proportional contribution of each forest type.
Introduction
Despite their overall similarity in structure, tropical rain forests may differ
considerably in floristic composition over the landscape. Factors such as soil
quality, rainfall patterns, disturbance regimes, and evolutionary history have
2400
produced forests that differ in the relative abundance of their plant species and
families.
As yet, attempts to characterize tropical rain forests in terms of their plant
composition have been few. As Terborgh and Andressen (1998) point out,
tropical rain forests remain largely undiagnosed in terms of floristic composition. Central African rain forests are no exception. Although considerable
effort has been devoted to produce taxonomic lists of plants from several
African rain forest sites (e.g., Lebrun and Gilbert 1954; Germain and Evrard
1956; Devred 1958; Gerard 1960; Letouzey 1969; White 1983; Newbery and
Gartlan 1996), less attention has been given to the study of patterns of variation in relative abundance of plant species over the landscape.
Here we present the results of a sample of trees 10 cm DBH in a large block
of forest contained between the Lomako and Yekokora rivers, Democratic
Republic of the Congo. Our objective was to characterize the forest contained
between the Lomako and Yekokora rivers in terms of its floristic composition.
This study’s original aim was to investigate the ecology of primates inhabiting
the Lomako–Yekokora interfluvium and more specifically to determine how
representative the Lomako study site, the location of a long-term study of
primates, was of the entire forest block contained between the Lomako and
Yekokora rivers.
In order to attain our objective we set up 15 botanical transects at 7 locations
(henceforth referred to as sample stations) placed at approximately 10 km intervals alongside a 70 km trail running south–north from the Lomako study
site to the Yekokora river, totaling 15.65 ha of area sampled.
Methods
Study site
The study area is located in the interfluvium between the Lomako and
Yekokora rivers, in the Equateur province of the Democratic Republic of
the Congo (208400 –218400 E, 008390 –018120 N) (Figure 1). The Lomako and
Yekokora rivers flow into the Maringa River, a left bank tributary of the
Congo River. The area is flat with altitude averaging 400 m above sea level
(Wiese 1980). Soils are of sedimentary origin and consist largely of sedimentary
rocks and unconsolidated sands and clays containing few weatherable primary
minerals (Juo and Wilding 1996). Soils derived from such sedimentary materials are mostly found to be deep, sandy Oxisols (Orthox) in dissected uplands
and wet Inceptisols (Aquepts) in inland depressions and along rivers and
streams (Juo and Wilding 1996).
Mean annual rainfall for this region is above 2000 mm with approximately
10 wet months a year and a drier period between January and March (Fruth
1995). The forest inventoried here consists of a block of roughly 3800 km2, part
of the large continuous rain forest that covers the Congo basin. The vegetation
2401
Figure 1. Map of the study site containing the locations of all seven sample stations. Lomako is
marked by a large open circle and the remaining six sample stations by black dots. The trail used to
gain access to the sample areas is marked by a dotted line.
consists of evergreen lowland tropical rain forest with at least four distinct
physiognomic types (Fruth 1995):
Mixed primary forest – a mixed-species forest, with a high heterogeneous canopy of 30–40 m with emergent trees >40 m in height. The undergrowth is
sparse with some large-leafed herbs growing under open gaps in the canopy.
The tree species Scorodophloeus zenkeri (Caesalpinoideae), locally known as
‘bofili’, is usually a dominant species.
Gilbertiodendron dewevrei dominated primary forest (henceforth ‘Gilbertiodendron forest) – a G. dewevrei (Caesalpinoideae) dominated forest with a high
homogeneous canopy of 30–40 m. Also with little undergrowth and largeleafed herbs growing under gaps of open canopy.
Secondary forest – former human agricultural fields and settlements or tree fall
gaps, characterized by an absence of large trees, with dense undergrowth, often
with presence of large-leafed herbs, lianas and smaller trees with air roots and/
or thorny tree trunks.
2402
Swamp forest – wet or temporarily inundated forest with a high density of small
trees with low crowns (5–10 m) and no undergrowth.
Data collection and analyses
The botanical survey was conducted in two parts. The first part during April–
December of 1996 was conducted at the Lomako study site and the second part
during October–November of 1997 and May of 1998 between the Lomako and
Yekokora rivers alongside a preexisting but rarely used human trail (70 km)
running north from Lomako to the Yekokora river (Figure 1).
At the Lomako study site, three transects of 3 km in length were laid out,
parallel to each other, along three preexisting east–west trails 1.5 km apart.
These formed one sample station. Between the Lomako and the Yekokora, six
additional sampling stations were established, 8–12 km apart with each containing two transect lines: one of 3 km running north–south and one of 1 km
running east–west. These transects were placed perpendicular to one another in
a ‘T’ shape position. Each sampling station was positioned haphazardly
without previous knowledge of the area and a few hundred meters away from
the trail. All transects were cut by two field assistants using machetes and the
direction in which to proceed was established by the researcher using a compass bearing.
Along the transects, 50 m 10 m (0.05 ha) plots were positioned longitudinally and in sequence, and were separated from each other by a 50 m gap.
In total, 313 plots were established corresponding to a sampled area of
15.65 ha. Seventy-three plots (3.65 ha) were located in the Lomako study site
and comprised the Lomako sample station, whereas the remaining 240 plots
were located along the Lomako–Yekokora trail, distributed in the remaining
six sample stations (Stations 1–6) with 40 plots (2 ha) each.
Within the plots, trees 10 cm DBH were measured and identified by their local
names with the help of four experienced workers from the Lomako study site.
Determination of their scientific name was achieved by using available species lists
from Projet Pan (Fruth 1995) and former researchers of the Centre de Recherches
en Science Naturelles (CRSN) at Lwiro. Literature on Central African tree species
(Letouzey 1969, 1972; Gauthier et al. 1977; Vivien and Faure 1985, Flore Rwanda
Urundi) were also used. Given that transect work was done by foot and supplies
including food were carried by the survey team, it was not possible logistically
to spend time collecting; neither was it feasible to transport a large number
of voucher specimens. Therefore, no collections were attempted at the time.
For the combined sample, number of individuals, occurrences, density and
dominance of each tree species were calculated. A diversity index was obtained
by relating the total number of tree species to the total number of individuals in
the sample. The Shannon–Wiener diversity index H = S(pi log pi) was used
for comparative purposes (Pielou 1974), where pi is the proportion of each
species in the sampled area.
2403
To assess the total contribution of each habitat type to the total sample, each
plot was classified according to the dominating forest type using the fourcategory system described above, that is, mixed primary, secondary, Gilbertiodendron and swamp forest.
Measures of forest structure (tree density, basal area, average DBH) were
calculated for each sample station. One-way analysis of variance (ANOVA)
was used to test for differences in these variables among the sample stations
and also among different forest types. Statistical levels of significance were set
at p < 0.01.
The coefficient of Jaccard (Krebs 1999), a binary similarity coefficient, was
used to quantify the overlap in species composition among the sample stations
and forest types. It is defined as: J = a/(a+b+c), where a is the number of
species found in the two samples being compared, b the number of species in
Sample 1 but not in Sample 2, c the number of species in Sample 2 but not in
Sample 1. The coefficient varies from 0 when no species overlap between the
compared samples is observed, to 1 when there is a complete overlap in species
and the plots are identical.
Since the coefficient of Jaccard does not take into account the abundance of
the tree species in the samples, Euclidean distance (Krebs 1999), a quantitative
similarity coefficient, was calculated between the sample stations and between
the different forest types to assess the difference in floristic composition taking
into account the relative abundance of the tree species in the analysis. Given
that the Lomako station had a larger sampled area (3.65 ha), for the purpose of
the calculation of Jaccard’s coefficient and Euclidean distances, a total of 33
plots (1.65 ha) were discarded from the Lomako sample in order to have equal
area sampled at all the seven stations. The discarded plots were randomly
selected.
Finally, an ordination method, principal component analyses (Ludwig and
Reynolds 1988), was used to assess the distribution of species among the 313
0.05 ha plots. This analysis was used to observe similarities between the plots
in relation to habitat type. The ordination was performed with SPSS for
Windows with all default options.
Results
Forest structure, tree species richness and floristic composition
of the forest landscape
The overall results of the botanical inventory are summarized in Table 1 and
Appendices 1 and 2. A total of 5353 trees 10 cm DBH, comprising a total
basal area of 572.4 m2, were present in the 15.65 ha sample. The calculated stem
density was 342 trees ha1 corresponding to a basal area of 36.57 m2 ha1. The
average DBH for all trees was 28.23 ± 23.77 cm (mean ± SD). Most trees
(51%) had DBHs between 10 and 20 cm. Only 12% had DBHs 50 cm.
2
2
2
2
2
3
4
5
6
Total
2
2
15.65 36.6
33.9
35.9
43.5
41.8
43.8
29.0
3.65 32.0
1
S. zenkeri, C. griseiflora, D. zenkeri,
P. suaveolens
D. zenkeri, C. mildbraedii, P. suaveolens,
Diospyrus sp.
S. zenkeri, P. suaveolens, D. pachyphylum,
C. griseiflora
S. zenkeri, D. zenkeri, C. mildbraedii,
Diospyrus sp.
G. dewevrei, S. zenkeri, C. griseiflora,
Strombosia glaucesens.
S. zenkeri, C. griseiflora, Garcinia punktata,
D. zenkeri
G. dewevrei, D. zenkeri, C. griseiflora,
S. zenkeri
P. suaveolens, C. griseiflora, D. pachyphylum,
S. zenkeri
Total Basal
Most abundant tree species
area area
2
1
(ha) (m ha )
Lomako
Sample
station
Caes, Olac, Ann, Sterc
Caes, Ann, Olac, Sterc
Caes, Olac, Ann, Clus
Caes, Olac, Clus, Sterc
Caes, Olac, Ann, Sterc
Caes, Olac, Euph, Eben
Caes, Ann, Olac, Euph
Caes, Olac, Ann, Euph
342
335
273
348
322
341
356
384
63
72.5
46.5
59.5
69.5
53.0
69.5
68.0
1.7
1.6
1.4
1.5
1.6
1.5
1.6
1.6
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
56
60
0
100
60
70
50
60
9
40
0
0
0
0
25
3
22
0
100
0
35
10
8
7
12
0
0
0
0
20
17
30
Most abundant tree family Tree
Number Diversity Evenness Forest types (% plots)
density
of
(H0 )
(J0 )
1
Mixed primary Secondary Gilbertiodendron Swamp
(trees ha ) species
(ha1)
Table 1. Summary of survey of trees 10 cm DBH found in seven sample stations placed alongside a 70 km trail running north from the Lomako to
the Yekokora river. Ann – Annonaceae, Caes – Caesalpinoideae, Clus – Clusiaceae, Eben – Ebenaceae, Euph – Euphorbiaceae, Olac – Olacaceae, Sterc –
Sterculiaceae.
2404
2405
A total of 5088 trees were identified to genus or species level (Appendix 1).
Of these, 4597 were identified to species (122 species), 491 to genus (16 genera),
261 to local names (25 unidentified local names) and there were four unknown
individuals. At least 32 families were represented in the sample. The Shannon–
Wiener diversity index for the total sample was H0 = 1.7 and the evenness
component, V = H0 /Hmax = 0.8 (Hmax = 2.2).
Four tree species were abundant in the botanical plots, comprising 28% of
all stems in the sample: S. zenkeri (Caesalpinoideae, 455 individuals),
Cola griseflora (Sterculiaceae, 375 individuals), Diogoa zenkeri (Olacaceae,
349 individuals) and Polyalthia suaveolens (Annonaceae, 325 individuals).
Sixteen percent of the species identified (28 species) were represented by
only one individual and 42% (73 species) had 10 or more individuals in the
sample.
The plant families Leguminosae sensu latu and Olacaceae had the greatest
number of individuals in the sampled plots (1586 and 651, respectively), corresponding to 42% of all the trees in the plots. Other abundant plant families
were Annonaceae, Euphorbiaceae, Sterculiaceae, and Clusiaceae (Appendix 2).
Leguminosae had the greatest basal area in the total sample, followed by
Olacaceae, Moraceae and Annonaceae. In terms of number of species, Leguminosae was the most species-rich family, followed by Meliaceae and Euphorbiaceae. Although Leguminosae sensu latu was by far the most important
family in this study, it should be pointed out that most of its trees (95%)
belonged to the Caesalpinoideae sub-family. Faboideae and Mimosoideae trees
had a relatively very low abundance.
Mixed primary forest was the dominant habitat corresponding to 57% of the
0.05 ha plots, followed by Gilbertiodendron forest (22%), swamp forest (12%),
and secondary forest (9%). Mixed primary forest dominated across all sample
stations, except for Station 5 which consisted of 100% of Gilbertiodendron
forest (Table 1). Gilbertiodendron forest was found on sloping ground adjacent
to streams (sample stations 1–4 and Lomako) and on flat ground in the central
part of the surveyed area (sample stations 3 and 5) without association to
streams. Secondary forest was found on the three sample stations closest to the
Lomako and Yekokora rivers (Lomako, 1 and 6). Swamp or temporarily inundated forest was mostly restricted to the first three sample stations, that is,
Lomako, 1 and 2.
Comparisons between the sample stations
The seven sample stations differed significantly in DBH (F = 10.64, p < 0.001,
df = 6). Sample station 1 had the smallest average DBH (25.24 ± 19.77) and
sample station 3 the highest (31.65 ± 25.59). Considering the 313, 50 m 10 m
plots independently, average DBH, density of trees and species richness differed. Highest tree density per plot was observed in the Lomako sample station
and the lowest at sample station 5 (Table 2). Species richness per 0.05 ha plot
2406
Table 2. Comparison between Lomako and sample stations 1–6 in terms of average DBH,
average number of individuals and average number of species for trees 10 cm DBH per 0.05 ha
(50 m 10 m) plots (numbers are mean ±SD).
Sample
station
Lomako
1
2
3
4
5
6
Total
Average DBH
per plota
Average number
of trees per plotb
Average number of
species per plotc
73
40
40
40
40
40
40
25.8 ± 20.6
25.2 ± 19.8
30.4 ± 26.6
31.6 ± 25.6
30.3 ± 25.9
30.6 ± 27.2
27.9 ± 22.6
19.07 ± 5.12
17.83 ± 4.77
17.13 ± 3.51
16.13 ± 4.24
17.40 ± 4.48
13.65 ± 3.06
16.75 ± 4.02
10.68 ± 2.72
10.50 ± 2.28
9.75 ± 2.17
10.85 ± 3.13
11.13 ± 2.42
9.08 ± 1.97
11.1 ± 2.53
313
28.2 ± 23.8
17.08 ± 4.58
10.46 ± 2.57
Number
of
plots
One-way ANOVA:
df = 6.
a
F = 10.6, p < 0.01, df = 6;
b
F = 7.3, p < 0.01, df = 6; cF = 3.7, p < 0.01,
Table 3. Results of pair-wise comparison between Lomako and sample stations 1–6 in terms of
floristic composition as measured by Jaccard coefficients and Euclidean distances. Jaccard
coefficients are presented above the ‘–’ diagonal and Euclidean distances below. Highest and lowest
values are in bold face.
Sample station
Lomako
1
2
3
4
5
6
Lomako
1
2
3
4
5
6
–
24.93
24.67
29.13
30.95
44.14
27.65
0.44
–
24.54
26.78
25.78
46.37
22.39
0.42
0.5
–
24.93
27.66
40.40
32.61
0.45
0.54
0.5
–
25.89
27.80
28.67
0.43
0.50
0.49
0.57
–
44.02
25.11
0.35
0.43
0.52
0.43
0.50
–
46.76
0.38
0.48
0.38
0.47
0.59
0.43
–
also differed slightly between sample stations, with Lomako supporting more
species per plot and sample station 5 the least (Table 2).
The mean coefficient of Jaccard calculated was 0.46 ± 0.06, indicating that
there was, on average, 46% overlap in floristic composition between sample
stations. The least overlap was observed between Lomako and Station 5 (35%)
and the greatest between Stations 4 and 6 (59%) (Table 3). Much of the floristic
difference between sample stations, however, was due to rare species with <10
individuals in the total sample. When only abundant species (that is, with 10
individuals) were considered, the mean coefficient of Jaccard went up to
0.85 ± 0.09, revealing a great degree of homogeneity among the sample stations.
Euclidean distance analysis, the method that takes species relative abundances into account, revealed that the stations most similar to one another
2407
Figure 2. Diameter size class distribution of trees 10 cm in DBH from mixed primary, secondary, Gilbertiodendron and swamp forests. See text for details. The X-axis indicates the diameter
size class intervals in cm and the Y-axis the number of trees contained in each size class.
were Stations 1–3 and, as already shown by the coefficient of Jaccard, Stations
5 and 7. The average Euclidean distance calculated was 31.00 ± 8.04 and
the greatest distances were found between Station 6 and Stations 1–3 and 5
(Table 3).
Comparisons between forest types
For this analysis, the 313, 0.05 ha plots were pooled together according to
forest type and irrespective of location. The four forest types differed significantly in terms of DBH distribution (Figure 2) and average, tree density and
tree species richness (Table 4). Swamp forest had the highest density of trees
and the lowest average DBH and number of species. The highest numbers of
species were found in mixed primary and secondary forests.
Because the four forest types were unevenly sampled in this study (i.e.,
samples were not stratified), only a sub-set of the 0.05 ha plots were used to
24.15 ± 19.39
30.22 ± 25.33
30
69
37
313
Secondary
Gilbertiodendron
Swamp
Total
17.08 ± 4.58
22.19 ± 5.17
0.96 ± 4.1
16.37 ± 4.15
17 ± 3.96
Number
of trees
10.46 ± 2.57
9.49 ± 2.64
9.64 ± 2.49
10.67 ± 2.53
10.96 ± 2.47
Number
of
species
S. zenkeri, D. zenkeri,
C. griseiflora, P. suaveolens
P. suaveolens, S. zenkeri,
D. pachyphylum,
Anthonota macrophylla
G. dewevrei, S. zenkeri,
C. griseiflora, D. zenkeri
C. mildbraedii, Polyalthia sp.,
Diospyros sp., Euphorb sp. 1
Most important
tree species
One-way ANOVA, F = 24.9, p < 0.01, df = 3F = 24.0, p < 0.01, df = 3F = 6.79, p < 0.01, df = 3.
28.2 ± 23.8
23.05 ± 14.62
29.64 ± 25.54
177
Mixed primary
DBH
Number of
0.05 ha plots
Forest type
Euph, Caes, Ann, Eben
Caes, Olac, Ann, Clus
Caes, Ann, Clus, Euph
Caes, Olac, Ann, Sterc
Most important
tree families
Table 4. Contrast of mixed primary, secondary, Gilbertiodendron and swamp forests in terms of five different ecological variables based on 313, 0.05 ha
(50 m 10 m) plots. Numbers are averages ± SD per 0.05 ha plot. Ann – Annonaceae, Caes – Caesalpinoideae, Clus – Clusiaceae, Eben – Ebenaceae, Euph –
Euphorbiaceae, Olac – Olacaceae, Sterc – Sterculiaceae.
2408
2409
Table 5. Results of pair-wise comparison between forest types in terms of floristic composition as
measured by Jaccard coefficients and Euclidean distances. Jaccard coefficients are presented above
the ‘–’ diagonal and Euclidean distances below. Highest and lowest values are in bold face.
Forest type
Primary
Secondary
Gilbertiodendron
Swamp
Primary
Secondary
Gilbertiodendron
Swamp
–
22.57
24.90
42.12
0.56
–
32.92
43.37
0.51
0.37
–
45.20
0.41
0.30
0.38
–
calculate Jaccard’s coefficients for the pair-wise comparison of forest types.
Thirty plots (1.5 ha) per forest type were randomly selected. This number (30)
was the maximum possible because only 30 swamp plots were present in the
total sample. There was an average of 42% overlap in tree species composition
(coefficient of Jaccard = 0.42 ± 0.09). The lowest overlap was found to be
between secondary and swamp forests (30%) and the highest between mixed
primary and secondary forests (56%) (Table 5). Swamp forest stood out as the
most distinct habitat floristically, showing the lowest average coefficient of
Jaccard (0.36 ± 0.06) when compared with the other three habitats. The
average species overlap between Gilbertiodendron plots and the other habitats
was 42% (coefficient of Jaccard = 0.42 ± 0.08).
When the distribution of the most abundant tree species in this sub-sample
was analyzed by calculating the Euclidean distances, the average distance was
35.2 ± 9.8, slightly higher than the average obtained when sample stations
were compared (see above). Again, swamp was the most distinct habitat since,
when removed from the analysis, the average Euclidean distance dropped to
26.80 ± 5.4. Mixed primary and secondary forests were the most similar floristically, followed by Gilbertiodendron forest (Table 5).
Several abundant tree species were restricted to swamp forest, explaining
why, floristically, this habitat was the most distinct. Sixty-two percent of
Diospyros sp., 84% of Cleistanthus mildbraedii, 92% of Lasiodiscus mannii,
95% of Trichoscypha ferruginea, 96% of Guibourtia demeusi, and 99% of
Polyalthia sp. trees were located in swamp plots. Other habitat-specific tree
species were G. dewevrei with 82% of the individuals found, as expected, in
Gilbertiodendron forest plots, D. zenkeri with 92% of the individuals found in
both primary and Gilbertiodendron forest, Dialium pachyphylum with 88% of
the individuals in primary and secondary forest and both Trichilia rubescens
and Caloncoba welwitschii restricted to secondary forest with 93 and 92% of
their trees, respectively, restricted, to this habitat.
Analysis of mixed primary forest
The results obtained from the above comparisons among sample stations and
forest types suggest that much of the difference found between sample stations
2410
Figure 3. Principal Component Analysis of 313, 0.05 ha plots located in Lomako and sample
stations 1–6. See text for details. A – swamp forest plots, B – Gilbertiodendron forest plots, C –
secondary forest plots, D – mixed primary forest plots. Circles indicate the areas of high concentration of plots from the corresponding forest type.
was due to differences in the total contribution of each forest type to each
sample station. We decided then, to determine how variable a single habitat
was over the landscape. Only mixed primary forest was used since this was the
only relatively well-sampled habitat across all but one sample station (Station
5). For all variables considered, that is, average DBH, tree density and species
richness, primary forest plots in all sample stations differed significantly only in
DBH (F = 6, p < 0.01) and this difference was relatively small.
As in the previous section, a sub-set of plots was taken in order to calculate
Jaccard’s coefficients and Euclidean distances. The analysis was restricted to 20
randomly selected plots (1 ha) per station, the maximum number of primary
forest plots present in sample station 2. Station 5 was left out since it contained
no primary forest. The floristic comparison as assessed by the coefficient of
Jaccard (mean = 0.45 ± 0.05) and Euclidean distances (mean = 17.27 ± 2.03)
showed that mixed primary forest habitat was quite similar across sample
stations.
2411
Ordination analysis
Given the relative homogeneity observed in floristic composition among all
sample stations, PCA analysis was employed to look for an ordering of samples in reduced space. In the PCA analysis, despite the observed scatter, the
313, 0.05 ha plots tended to separate themselves by habitat types (Figure 3). In
Figure 3, circles are drawn around areas of greater concentrations of plots
belonging to each forest type. Not all plots fell within the perimeter of their
corresponding forest type boundaries, but these represented a small fraction of
the total sample. Component 1 accounted for 26% of the total variance and
Component 2 accounted for 8%. Component 1 discriminated mixed primary
and secondary forest plots from swamp and Gilbertiodendron plots. Component 2 discriminated Gilbertiodendron transects from all others.
Discussion
Forest structure
With 352 stems ha1 the forest in the Lomako–Yekokora interfluvium was
found to have a relatively low stem density for trees 10 cm in DBH. According
to Richards (1996), the density of trees in tropical rain forests worldwide ranges
from 300 to 700 for trees 10 cm DBH per hectare in well-drained soils. To date,
the reason for such variation in density is still obscure (Richards 1996). Despite
such low densities, total basal area per hectare was comparable to other African
tropical rain forests indicating a larger number of large trees (e.g., Hart et al.
1989; Maisels 1996; Newbery and Gartlan 1996).
When forest types were considered separately, swamp forest stood out with
a high density of trees 10 cm DBH (444 trees ha1). This finding was surprising considering that forests growing on poorly drained soils are known to
support lower densities of trees when compared to forests growing on welldrained soils (Hartshorn 1980; Richards 1996). Contrarily to swamp, Gilbertiodendron forest had a low density of trees (299 trees ha1); G. dewevrei
dominated forests have been considered to be the only real climax forests in the
Congo (Devred 1958), occurring in areas not naturally disturbed for centuries
(see below).
As expected from the number of trees per hectare, DBH distributions were
distinct between the four forest types. Swamp and secondary forest had more
small trees than mixed primary and Gilbertiodendron forests.
Species richness
With an average of 63 species per hectare, the Lomako–Yekokora forest
conformed to the pattern of low tree species diversity of African rain forests
2412
described by Richards (1973). Richards (1973) referred to African rain forest as
the ‘odd man out’ when compared to the American and Indo-Malayan forests
because of the poverty of its flora and because its plant species were widely
distributed, resulting in low alpha and beta tree species diversity.
Although significant, the difference in species richness was small between the
sample stations and between the forest types. On average, mixed primary and
secondary forests had more species than swamp and Gilbertiodendron forests.
In the latter case, low species richness was a result of the high dominance
attained by G. dewevrei trees.
Floristic composition
The Lomako–Yekokora sample was dominated by trees of the Caesalpinoideae
sub-family. In particular, very abundant were two types of Caesalpinoideae
trees, S. zenkeri and G. dewevrei. S. zenkeri is a characteristic tree of the Congo
Basin (White 1983). G. dewevrei can occur outside the Congo Basin but then it
is usually confined to river banks or swamps (White 1983). In the Basin itself,
G. dewevrei forms islands of monodominant stands surrounded by mixed
primary forest. The transition between these two forest types is usually
abrupt, however. The factors involved in such physiognomic change are not
well understood (Connell and Lowman 1989; Hart et al. 1989).
Other Caesalpinoideae species are also known to form monodominant
stands in the Congo Basin: Brachystegia laurentii (De Wild.) Louis ex Hoyle,
Cynometra alexandri C.H. Wright, Julbernardia seretii (De Wild.) Troupin,
Michelsonia microphylla (Troupin) Hauman (White 1983). All these are ectomycorrhizal species (Newbery et al. 1988). These monodominant forests occur
throughout the Congo Basin but are most extensive on the plateau surrounding
the central basin of the Congo River (Gauthier et al. 1977; White 1983).
In an attempt to investigate the factors involved in the formation of
monodominant stands of G. dewevrei in the Ituri forest, DR Congo, Hart et al.
(1989) examined the effect of substrate quality, herbivory gradient and disturbance effects, which potentially could explain the transition from one forest
type to the other. However, these authors were not able to establish a correlation between the examined factors and the monodominant forest. For Pierlot
(1966), the G. dewevrei forest as well as the B. laurentii forest (the latter not
present in the Lomako–Yekokora area) are the only real primary forest types
in the Congo Basin. Kortlandt (1995) believes that these two species slowly
invade and overwhelm the Scorodophloeus forest (in this study referred to as
mixed primary forest) which he considers a ‘century old’ secondary forest. The
observed transition from one forest type to the other in this study would
consequently be caused by previous disturbance by ‘‘sparse human population
and/or nomadic elephant foraging’’ (Kortland 1995).
Floristically, our Lomako–Yekokora sample stations were quite similar to
one another and most of the observed differences were due to the uneven
2413
sampling of forest types at each sample station. In this study, as expected, rare
species were found to be unevenly distributed across sample stations and forest
types. Abundant species were relatively common in all sample stations, although some species tended to clump in different habitats, particularly in the
swamp forest. Thus, although relative abundances of tree species tended to
vary to a greater or lesser degree between sample stations, species composition
per se remained relatively constant (hence high Jaccard’s coefficients obtained
in most pair-wise comparisons), conforming to the low beta-diversity of
African forests reported by Richards (1973).
Low beta-diversity does not appear to be an African-only phenomenon,
however. Pitman et al. (1999) report ‘an unexpected’ low beta-diversity in their
western Amazonia tree samples. Most of their sampled tree species were not
restricted to any given habitat, although their relative abundances showed
marked spatial variance. They concluded that . . . ‘‘the habitat preferences of
Amazonian plants are a matter of degree and not as strict as suggested by earlier
researchers’’ (Pitman et al. 1999). For Pitman et al. (1999), high beta-diversity
reported by earlier researchers was an artifact of insufficient sampling.
In conclusion, at the landscape scale, the Lomako–Yekokora interfluvial
forest was found to be relatively homogeneous in terms of the ecological
variables measured here. The forest types, however, differed markedly, justifying the separation into the four physiognomic types, that is, mixed primary,
secondary, Gilbertiodendron and swamp forests. The results of this study indicate that the Lomako study site is representative of the forest landscape
contained between the Lomako and Yekokora rivers only when the different
forest types are treated in separate. The sample stations (including Lomako)
differ from each other, however, in the proportional contribution of each forest
type.
In terms of the floristic characterization of the forest landscape contained
between the Lomako and Yekokora rivers, this study showed that Caesalpinoideae trees, in particular S. zenkeri and G. dewevrei, dominated the
sample. Trees of this family were followed in abundance by trees of the
Olacaceae and Annonaceae. According to Newbery and Gartlan (1996),
Caesalpinoideae trees are abundant throughout the Guinea–Congo region
becoming particularly dominant in areas of sandy, low fertility soils, as is the
case for Salonga National Park, DRC (Gautier-Hion et al. 1993; Maisels
et al. 1994), and Tiwai forest, Sierra Leone (Oates et al. 1990; Dasilva 1994).
Such species become less abundant on richer soils, as in the case of Lopé,
Gabon (Tutin et al. 1997), Odzala National Park, Congo (Maisels 1996) and
Kibale, Uganda (Struhsaker 1975). In these forests, one or more of the
tree families Annonaceae, Burseraceae, Euphorbiaceae, Meliaceae, Olacaceae
and Rubiaceae attain higher abundances than Caesalpinoideae. It remains to
be investigated what factors (e.g., soil quality, rainfall patterns, disturbance
regimes, and evolutionary history among others) are responsible for the observed differences in floristic composition at these different African rain forest
sites.
2414
Acknowledgements
Our most sincere gratitude to all field assistants in Ndele for their invaluable
help in setting up and conducting the long-term study of primates in Lomako;
in particular Papa Maurice, Papa Adula, Banana, and Mira. Support for
fieldwork was obtained from the Max Planck Institute, Seewiesen, Germany.
This manuscript was written while visiting the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, as an Alexander von Humboldt
Research Fellow.
Appendix 1
List of tree species found in a 15.65 ha sample in the Lomako–Yekokora interfluvium, Democratic
Republic of the Congo.
Scientific name
Family
Number
of
individuals
Occurrences
(number
of plots)
Density
(trees ha1)
Dominance
(m2 ha1)
Anonidium mannii (D. Oliver) Engl. & Diels
Monodora angolensis Welw
Polyalthia sp.
P. suaveolens Engl. & Diels
Tabernaemontana durissima Stapf
Dacryodes edulis (G. Don) H.J. Lam
Dacryodes sp.
Anthonota fragrans
Anthonota macrophylla
Crudia harmsiana De Wild.
Dialium corbisieri Staner
Dialium pachyphyllum Harms
Dialium sp 2
Dialium sp.
Dialium zenkeri Harms
Erythrophleum suaveolens
(Guillemin & Perrottet) Brenan
G. dewevrei (De Wild.) J. Léonard
G. demeusi (Harms) J. Léonard
Monopetalanthus microphyllus Harms
Oxystigma oxyphyllum (Harms) Léonard
S. zenkeri Harms
Tessmannia anonala (Micheli) Harms
Parinari excelsa Sabine
‘njete manza’
Garcinia kola Heckel
Garcinia ovalifolia Oliv.
Garcinia punctata Oliv.
Garcinia sp.
Symphonia globulifera L. f.
Diospyros sp. 2
Ann
Ann
Ann
Ann
Apoc
Burs
Burs
Caes
Caes
Caes
Caes
Caes
Caes
Caes
Caes
Caes
122
19
102
326
12
12
25
43
122
120
38
221
11
13
18
25
88
9
32
173
10
12
19
36
64
76
34
131
9
10
17
22
7.8
1.21
6.52
20.83
0.77
0.77
1.6
2.75
7.8
7.67
2.43
14.12
0.7
0.83
1.15
1.6
1.1
0.0
0.1
1.0
0.0
0.0
0.1
0.6
0.3
2.4
0.8
0.3
0.1
0.1
0.1
0.4
Caes
Caes
Caes
Caes
Caes
Caes
Chrys
Clus
Clus
Clus
Clus
Clus
Clus
Eben
245
51
12
60
455
37
10
13
30
31
270
42
15
37
77
20
7
51
190
34
10
11
29
25
152
37
12
33
15.65
3.26
0.77
3.83
29.07
2.36
0.64
0.83
1.92
1.98
17.25
2.68
0.96
2.36
4.1
0.2
0.0
2.0
3.3
0.5
0.4
0.1
0.1
0.0
0.5
0.1
0.1
0.0
2415
Appendix 1 (continued)
Scientific name
Family
Number
of
individuals
Occurrences
(number
of plots)
Density
(trees ha1)
Dominance
(m2 ha1)
Diospyros sp.
‘Winzinzi’
C. mildbraedii Jabl.
Croton haumanianus J. Léonard
Croton sp.
Macaranga laurentii
Macaranga sp.
Uapaca guineensis Müll. Arg.
Millettia drastica Welw. ex Baker
Pterocarpus soyauxii Taub.
Barteria fistulosa Mast
C. welwitschii (Oliv.) Gilg
‘Iondjo’
Irvingia wombolu Vermoesen
Klainedoxa gabonensis Pierre ex Engl.
Guarea cedrata (A. Chev.) Pellegr.
Guarea laurentii De Wild.
T. rubescens Oliv.
Pentaclethra macrophylla Benth.
Chlorophora excelsa (Welw.) Benth.
Ficus sp.
Treculia africana Decne.
Stadtia stipitata
Ouratea arnoldiana De Wild. & T. Durand
D. zenkeri (Engl.) Exell & Mendonça
Ongokea gore (Hua) Pierre
Strombosia glaucescens Engl
Strombosia grandifolia Hook. f.
Strombosiopsis tetrandra Engl.
Panda oleosa Pierre
L. mannii Hook. f.
‘Litsitsi’
Massularia acuminata (G. Don)
Bullock ex Hoyle
Allophylus africanus P. Beauv.
Blighia welwitschii (Hiern) Radlk.
Pancovia laurentii (De Wild.)
Gilg ex De Wild.
Gambeya lacourtiana (De Wild.)
Aubrév. & Pellegr.
Synsepalum subcordatum De Wild.
Cola chlamydantha K. Schum.
C. griseflora De Willd.
Pterygota bequaertii De Wild.
Desplatsia dewevrei (De Wild.
& T. Durand) Burret
Vitex wilmsii
Other 109 taxa
Eben
Euph
Euph
Euph
Euph
Euph
Euph
Euph
Fab
Fab
Flac
Flac
‘Iondjo’
Irv
Irv
Meli
Meli
Meli
Mimo
Mor
Mor
Mor
Myri
Ochn
Olac
Olac
Olac
Olac
Olac
Panda
Rham
Rub
Rub
195
109
225
10
18
14
18
12
21
33
15
35
36
34
23
32
18
39
18
68
12
10
40
38
349
30
172
73
27
51
79
49
23
116
53
164
8
17
7
14
11
15
25
14
22
32
30
22
31
15
26
16
54
11
10
36
36
156
26
121
52
24
42
30
40
21
12.46
6.96
14.38
0.64
1.15
0.89
1.15
0.77
1.34
2.11
0.96
2.24
2.3
2.17
1.47
2.04
1.15
2.49
1.15
4.35
0.77
0.64
2.56
2.43
22.3
1.92
10.99
4.66
1.73
3.26
5.05
3.13
1.47
0.3
0.4
0.7
0.1
0.0
0.1
0.1
0.2
0.1
0.5
0.0
0.1
0.1
0.1
0.4
0.2
0.1
0.0
0.6
0.4
1.3
0.1
0.8
0.1
0.9
0.5
0.7
0.4
0.3
0.4
0.3
0.1
0.0
Sapi
Sapi
Sapi
18
15
50
17
13
42
1.15
0.96
3.19
0.1
0.2
0.2
Sapot
14
14
0.89
0.2
Sapot
Sterc
Sterc
Sterc
Tili
21
11
378
17
20
20
8
190
17
15
1.34
0.7
24.153
1.09
1.28
0.1
0.0
0.5
0.4
0.1
Verb
27
319
26
297
1.73
20.28
0.1
5.7
2416
Appendix 2
Ten most abundant plant families from a 15.65 ha sample of trees 10 cm DBH in the Lomako–
Yekokora interfluvium, Democratic Republic of the Congo.
Family
No. of
species
Total no.
of trees
Density
(trees ha1)
Dominance
(m2 ha1)
Leguminosae
Caesalpinoideae
Faboideae
Mimosoideae
Olacaceae
Annonnaeae
Euphorbiaceae
Sterculiaceae
Clusiaceae
Ebenaceae
Meliaceae
Moraceae
Rubiaceae
31
24
3
4
6
8
13
5
8
2
12
6
9
1588
1503
62
23
651
587
426
407
403
232
113
100
100
101.47
96.04
3.96
1.47
41.60
37.51
27.22
26.01
25.75
14.82
7.22
6.39
6.39
17.45
15.56
9.83
0.90
2.86
2.34
1.78
0.98
0.75
0.34
1.30
2.41
0.22
References
Connell J. and Lowman M.D. 1989. Low-diversity tropical rain forests: some possible mechanisms
for their existence. American Naturalist 134: 88–119.
Dasilva G.L. 1994. Diet of Colubus Polykomos on Tiwai Island: selection of food in relation to
its seasonal abundance and nutritional quality. International Journal of Primatology 15:
655–678.
Devred R. 1958. La Vegetation Forestiere du Congo Belge et du Ruanda-Urundi. Bulletin de la
Société Royale Forestiere de Belgique 65: 409–468.
Fruth B. 1995. Nests and nest groups in wild bonobos (Pan paniscus): ecological and behavioural
correlates. In: Biologie, Ludwig-Maximilians-Universität, München, Germany, 187 pp.
Gauthier Poulin Thériault and Ltée 1977. Manuel de Dendrologie. ACDICI, Québec, Canada.
Gautier-Hion A., Gautier J.P. and Maisels F. 1993. Seed dispersal versus seed predation: an
intersite comparison of two related African monkeys. Vegetatio 107/108: 237–244.
Gerard P. 1960. Etude Écologique de la Forêt Dense à Gilbertiodendron dewevrei dans la Région de
l’Uele. No. 87 Publication de l’Institute National pour l’Étude Agronomique du Congo Belge
(INÉAC), Série Scientifique No. 87, 159 pp.
Germain R. and Evrard C. 1956. Etude Écologique et Phytosociologique de la Forêt à Brachystegia
laurentii. Vol. 67. Publications de l’Institute National pour l’Étude Agronomique du Congo
Belge, pp. 1–105.
Hart T.B., Hart J.A. and Murphy P.G. 1989. Monodominant and species-rich forests of the humid
tropics: causes for their co-occurrence. American Naturalist 133(5): 613–633.
Hartshorn G.S. 1980. Neotropical forest dynamics. Biotropica 12: 23–30.
Juo A.S.R. and Wilding L.P. 1996. Soils of the lowlands of west and central Africa. Proceedings of
the Royal Society of Edinburgh 104B: 5–29.
Kortlandt A. 1995. A survey of the geographical range, habitats and conservation of the Pygmy
chimpanzee (Pan paniscus); An ecological perspective. Primate Conservation 16: 21–36.
Krebs C.J. 1999. Ecological methodology. Benjamin/Cummings, Menlo Park, California.
2417
Lebrun J. and Gilbert G. 1954. Une Classification Écologique des Forêts du Congo. Publication de
l’Institute National pour l’Étude Agronomique du Congo Belge (INÉAC), Série Scientifique,
pp. 63–89.
Letouzey R. 1969. Manuel de Botanique Forestière, Afrique Tropicale. Centre Technique Forestier
Tropical, Marne, France.
Letouzey R. 1972. Manuel de Botanique Forestière, Afrique Tropicale, Centre Technique Forestier
Tropical, Marne, France.
Ludwig J.A. and Reynolds J.F. 1988. Statistical Ecology. John Wiley and Sons, New York.
Maisels F. 1996. Synthesis of information concerning the Park National D’Odzala, Congo. Projet
Ecofac-Composante Congo. Unpublished report.
Maisels F., Gautier-Hion A. and Gautier J.P. 1994. Diets of two sympatric colobines in Zaire: more
evidence on seed-eating in forests on poor soils. International Journal of Primatology 15:
681–701.
Newbery D.M. and Gartland J.S. 1996. A structural analysis of rain forest at Korup and DoualaEdea, Cameroon. Proceedings of the Royal Society of Edinburgh 104B: 177–224.
Newbery D.M., Alexander I.J., Thomas D.W. and Gartlan J.S. 1988. Ectomycorrhizal rain-forest
legumes and soil-phosphorus in Korup-National-Park, Cameroon. New Phytologist 109:
433–450.
Oates J.F., Whitesides G.H., Davies A.G., Waterman P.G., Green S.M., Dasilva G.L. and Mole S.
1990. Determinants of variation in tropical forest primate biomass: new evidence from west
Africa. Ecology 71: 328–343.
Pielou E.C. 1974. Population and Community Ecology. Gordon & Breach, New York.
Pitman N.C.A., Terborgh J., Silman M.R. and Nunez P. 1999. Tree species distributions in an
upper Amazonian forest. Ecology 80: 2651–2661.
Richards P.W. 1973. The tropical rainforest. Africa, the ‘Odd man out’. In: Meggers B.J., Ayensu
E.S. and Duckworth W.D. (eds). Tropical Forest Ecosystems in Africa and South America: A
Comparative Review. Smithsonian Institute Press, Washington, DC.
Richards P.W. 1996. The Tropical Rain Forest. Cambridge University Press, Cambridge, UK.
Struhsaker T.T. 1975. The Red Colobus Monkey. University of Chicago Press, Chicago, Illinois.
Terborgh J. and Andresen E. 1998. The composition of Amazonian forests: patterns at local and
regional scales. Journal of Tropical Ecology 14: 645–664.
Tutin C.E.G., Ham R.M., White L.J.T. and Harrison M.J.S. 1997. The primate community of the
Lope Reserve, Gabon: diets, responses to fruit scarcity and effects on biomass. American Journal
of Primatology 42: 1–24.
Vivien J. and Faure J.J. 1985. Arbres des Forêts Denses d’Afrique Centrale. Agence de
Coopération Culturelle et Technique, Paris, France.
White F. 1983. The Vegetation of Africa. UNESCO, Paris, France.
Wiese B. 1980. Zaire: Landesnatur, Bevolverung, Wirtschaff. Wissenschaftliche Buchgesellschaft,
Darmstadt, Germany.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.