and small cacao plantations are found throughout the area. In general these plantations are not
well maintained due to continuing low world market prices. The devaluation of the Franc CFA in
1994 was an incentive for the production but still more than 50% of the cacao plantations remain
abandoned. Recently, industrial size oil palm, pineapple and banana plantations have been
created in the TCP area. Villagers who work and live in surrounding towns, appear to be the
initiators of this development.
3.1 LANDSCAPE ECOLOGICAL APPROACH
The Tropenbos Cameroon Programme aims at the development of sustainable forms of land use
in Southwest Cameroon. The specific objective of the first phase of the Lu1 project is to provide
base line information on the biotic and abiotic environment of the TCP area. This information
will be used for land evaluation and planning procedures and for more detailed ecological
research in the area. Based on the size of the area and the need for detail for land evaluation, a
landscape ecological survey at scale 1 : 100,000 (reconnaissance scale) was carried out.
Zonneveld defines landscape (1995; after Schroevers, 1982) as `a complex of relationships
systems, together forming ( . . . ) a recognizable part of the earth's surface, which is formed and
maintained by the mutual action of abiotic and biotic forces as well as human action'. The
landscape can thus be seen, as is done in the present study, as a fully integrated entity that can
and should be studied as a whole (e.g. Breimer et al., 1986; Hommel, 1987; Küchler &
Zonneveld, 1988; Zonneveld, 1995).
Tropenbos has developed a Common Methodology for Land Inventory and Land Evaluation to
contribute to a systematic and interdisciplinary research approach and subsequently sound land
use planning (Touber et al., 1989). This methodology focuses on the integrated description of the
environment and was used for the land unit surveys conducted in the Tropenbos sites in Côte
d'Ivoire (de Rouw et al., 1990; de Rouw, 1991) and Colombia (Duivenvoorden & Lips, 1993;
1995). The holistic nature of the landscape approach requires a multidisciplinary team of
An almost indispensable tool in all landscape oriented studies is the use of remote sensing
materials. Depending on the scale of the study and the objects to be mapped, satellite images and
aerial photographs can be used. For the present reconnaissance survey black & white aerial
photographs have been used. These images reveal the identity of the landscape and their
systematic interpretation enables the delineation of the different land units. A land unit is `a tract
of land that is ecologically relatively homogeneous at the scale level concerned' (Zonneveld,
1995). Aerial photo interpretation provides a basis for stratified sampling in the field.
The aim of the fieldwork is to collect accurate and reliable materials to describe the photo
interpretation units. In addition, the relevance of the boundaries of the photo interpretation units
is studied. Ecological relevant data on soil, vegetation, geomorphology and land use, as well as
geological, hydrological, zoological and other appropriate land attribute information are collected
in selected sample sites. The land attributes are classified and a legend is compiled. The resulting
landscape ecological map is an important tool for land evaluation. In the following sections, the
methodological aspects of the present survey are treated in more detail.
3.2 AERIAL PHOTO INTERPRETATION
Topographical base maps at scale 1 : 50 000 were supplied by the CENADEFOR (1987) and are
in fact enlargements of the 1 : 200 000 topographic maps of the Institute Géographie Nationale
(CGN, 1976). The sheets Edéa NA-32-XXIII 1b, 2a and 2b and Nyabessan NA-32-XVII 3d, 4c
and 4d were joint to make a single topographic map of the TCP area. The topographic map was
updated as for the location of logging roads with the aid of a Global Positioning System
Black & white aerial photographs of the TCP area were taken in 1963-66 at a scale of
1 : 50 000 and in 1983-85 at a scale of 1 : 20 000. The 1963-66 series has insufficient
contrast to allow for stereoscopic vision. The much better quality photographs of the 1983-
85 series have therefore been used for photo interpretation in the present survey. A list of the
photographs is presented in Annex III.
The interpretation of the aerial photographs resulted in the drafting of preliminary maps on
landforms and vegetation at a scale 1 : 50 000 (Touber, 1993a). The maps have hierarchical
legends and discern a total of 49 and 25 legends units, respectively. A five week mission to
the TCP research area was carried out in 1993 and sites were selected for the efficient
sampling of the most important land units in the area (Touber, 1993b).
Observation sites were selected on the basis of the photo interpretation maps. Fieldwork
involved the description of the attributes of the units discerned by photo-interpretation.
Fieldwork was not merely a check to verify the photo interpretation as it largely entailed the
collection of new information.
The land units sampled were considered the most important and widespread in the TCP area.
The actual sample sites were selected on the basis of representativeness and accessibility.
Transects, or transverses, of one to two kilometres with a general orientation perpendicular
to the contour lines, have been laid out. These transects provided access to the land units to
be described and enabled the detection of possible catenas or toposequential processes.
Within the transect three to seven observation points were selected for detailed description.
The locations of the observation points were determined by both the soil surveyor and the
vegetation specialist. At each sampled locality, landform, soil and vegetation characteristics
were described. In Figure 3.1 the distribution of observation points in the TCP research area
During the field work the survey team consisted of a soil surveyor, a vegetation surveyor, a
field botanist (part-time), a local tree spotter and a local soil survey assistant. The transects
were laid out by a `compass man' and two line cutters recruited from the nearest village. The
distance along the transects was measured with a 'Topofyl' and each hundred meter was
(temporarily) marked with a pole. The survey team covered approximately one kilometre per
day for normal sampling procedures. One or two days extra were needed to dig and describe
All transects and observation points have geographical coordinates and can be retraced on
the base map and the aerial photographs. Photo interpretation and fieldwork were not strictly
separated in time; efforts have been made to study the aerial photographs after each period
of fieldwork and to adjust the preliminary legends and maps to the insights gained. The
sampling procedures for landform, soil and vegetation aspects are discussed in the sections
3.3.2 and 3.3.3.
The described land and soil related attributes are based on Touber et al. (1989), FAO (1990)
and Touber et al. (1993), and include: altitude, slope length, slope steepness, slope form,
exposition, landform, stone and rock coverage, erosion and stability characteristics, drainage
characteristics, groundwater level, biological activity and, humus form. Soil characteristics,
described per horizon, are: colour, mottling, texture, structure, consistency, cutans, rooting
characteristics and pores. Additionally, schematic sketches of the transects were made,
showing the variation in slope percentage, slope direction, slope length, and exposition
within each land unit.
At least three auger hole observations per transect were made. The auger holes were made
with a standard type Edelman auger and range in depth from 1.2 to 2.0 m. A total of 207
soil and landform observations were made. In Figure 3.1 the distribution of observation
points in the TCP area is presented.
After a first survey of the transects, about 60 sites representing the different landforms and
soil types of the TCP area were selected, soil pits were dug and soil characteristics were
described in more detail. All soil pits have a minimum depth of 1.50 m. Samples of horizons
were collected for chemical and physical analysis.
Soil samples were chemically and physically analyzed at the IRA soil laboratory in Ekona.
Duplicates of some 40 samples were analyzed as control at ISRIC, Wageningen. Chemical
soil properties determined are organic carbon, total nitrogen, available and total
phosphorous, pH H
O and KCl, exchangeable bases (Na, K, Mg and Ca), aluminum and
hydrogen, and cation exchange capacity (CEC). Physical parameters determined are texture,
water retention characteristics (pF) and bulk density. Additionally, the clay mineralogy of
some twenty-five samples were analyzed by ISRIC. The methods used for chemical and
physical analyses are described in Annex IV.
All field data concerning the site and soil characteristics of the augerings and soil profile
descriptions were compiled in the TROFOLIN Database (Touber et al., 1993). Data of
chemical and physical analysis are stored in QUATTRO PRO.
The distribution of plant species in tropical forests is not random but reflects environmental
conditions (Whitmore, 1984 and Zonneveld, 1995). Species which can successfully compete
with one another within the limits of a particular combination of environmental features can
be considered as a plant community. The mosaic of plant communities forms the vegetation
in an area (Küchler & Zonneveld, 1988). Moreover, the pattern of vegetation types, defined
as plant-communities, may be used to assess the distribution patterns of distinct plant
In the present survey the vegetation was described on the basis of its structure and floristic
composition, in the sense of the French-Swiss school (phytosociological approach; Braun-
Blanquet, 1964). The method implies a classification of vegetation types by tabular
comparison of plot data, which leads to the identification of communities, characterized by
phytosociological groups (Touber et al., 1989).
The localities described in the vegetation survey lie within the most common and
widespread photo interpretation units in the TCP research area (see 3.3.1). The stratification
of sample points is based on the hypothesis that altitude (Fig. 2.1) and disturbance (both
anthropogenic and natural) cause most of the variation in the vegetation in the TCP research area.
The observation points are therefore evenly spread along these two environmental gradients
(Table 3.1). Additionally, in relatively undisturbed forest vegetation efforts have been made to
describe the vegetation along the toposequence, comparable to the study of catena's in the soil
At each selected locality a detailed description of the vegetation, a `relevé', was made. As far as
possible all growth forms, excluding mosses, ferns, epiphytes and small seedlings (due to
difficulties in field identification), have been recorded in a plot of 100 m
. In addition, all trees
with diameter at breast height (dbh)
≥20 cm were recorded in the surrounding 1000 to 2500 m
Though not approaching the minimal area of tropical rain forest communities, this plot size is
sufficiently large for classification purposes if all terrestrial growth forms of flowering plants and
all size classes are included (Hommel, 1987; 1990; de Rouw, 1991).
The vegetation within each sample plot was homogeneous with respect to its structure, and
representative for the surrounding area (Kent & Coker, 1992; Hommel, 1995). The intention was
to describe the characteristic vegetation type(s) within each photo interpretation unit and
therefore a-typical situations were omitted from sampling. A total of 125 vegetation relevés has
been described in the reconnaissance scale survey and is used for the classification of the
vegetation in the TCP research area. Table 3.1 gives the distribution of the vegetation relevés
over the environmental gradients altitude and disturbance.
Table 3.1 Stratification of vegetation relevés along the environmental gradients altitude and disturbance
40-180 m asl 180-340 m as 340-540 m asl
> 540 m asl
Number of observations: - = not found; * = 1 relevé; ** = 2-5 relevés; ***= 6-10 relevés; ****= 11-15 relevés.
The recording of plant species was carried out according to guidelines given by Touber et
al.(1989), i.e. separate records per stratum and with cover/ abundance estimation in 14 classes
(`ITC approach'). However, a complete recording of vegetation characteristics according to the
TROFOLIN procedure (Touber et al., 1993) was considered too time-consuming for a
reconnaissance survey (Hommel, 1995).
Plant identification was done in the field with the help of a field botanist (part time) and a local
tree spotter. Plant material of unknown species and for verification purposes has been
collected in `Quick Herbarium' style (Küchler & Zonneveld, 1988). The `Herbier National
du Cameroun' (HNC), the Limbé Botanic Garden and the Herbarium Vadense were
consulted for identification. Under the prevailing conditions the field identification is not to
be regarded equal to the botanical identification. Each field `species' often consists a cluster
of botanical species with a similar general appearance. Efforts have been made to limit the
size of these groups as much as was deemed appropriate in light of reliability. In the
analyses of the vegetation data these groups have been used as entity.
Next to the detailed description of the sample points, the vegetation structure and land use
along the transects was indicated on drawings. These drawings give insight in the mosaic of
the vegetation in the different mapping units. In the second phase of the Lu1 project, will
study these mosaics are studied in more detail. Also some 20 incomplete or so-called `quick
relevés' were described. These incomplete descriptions do not support the classification of
the vegetation, but help to investigate the relations between vegetation and abiotic factors.
Moreover, they are of cartographic importance.
All vegetation data have been stored in TURBOVEG, a software package for input,
processing and presentation of phytosociological data (Hennekens, 1995). The output of this
package is, in contrast to TROFOLIN (Touber et al., 1993), compatible with vegetation
classification programmes like TWINSPAN (Hill, 1979a), SHAKE (Tongeren, unpubl.),
DECORANA (Hill, 1979b) and CANOCO (ter Braak, 1988).
Based on the variation observed within the TCP research area, preliminary classifications
were drafted for landforms, soils and vegetation. To describe the landscape as a complex of
these attributes, the relations between landforms, soils and vegetation types were studied by
means of cross-tables. The aim of this exercise was to trace the parameters of the abiotic
environment which may best explain the variation in vegetation types. Based on these
analysis, the landforms, soils and vegetation classifications were slightly modified to attain
an optimal `ecological' fit.
The classification of landforms is based on differences in relief characteristics, i.e. slope
length, slope steepness, relief intensity, and the number of interfluves. These characteristics
are important in the light of land utilization.
The classification of soils in the TCP research area is primarily based on soil drainage and
soil texture. Soil depth and stoniness are other differentiating criteria but could not be used
at this scale. Soil drainage and texture are found to have a correlation with vegetation in the
TCP research area. As soil texture and soil drainage are indicators for moisture availability,
they are two of the functional soil parameters for land evaluation as well.
The classification of the vegetation of the area was done by means of `tabular comparison' of
the relevé data using the computer programme TWINSPAN (Hill, 1979a). The programme
performs a multi-variate analysis of vegetation data and produces a hierarchical
classification of both sample points and species. Manual refinement of this classification
included the reallotment of a limited number of borderline relevés. Also some 11 relevés
were excluded from analysis because they proved to represent either intermediate situations on
the highest level of classification, or represented ecologically aberrant situations that were not
considered relevant to the reconnaissance survey.
The cover values of the three structural layers, i.e. tree, shrub and herb layer, have been
combined into one value for each species. Four coverage classes were used, 0-5%, > 5-25%,
> 25-50% and > 50-100% cover, regardless of the abundance of the species. For classification
of vegetation on a reconnaissance scale, this is considered sufficiently detailed.
The classification by TWINSPAN is primarily based on the floristic composition of the
sample plots. Next, a synecological interpretation of the vegetation types discerned has been
carried out on the basis of the environmental data collected at the different plot sites.
3.5 LEGEND AND MAP COMPILATION
The final legend of the landscape ecological map is based on the classification systems for the
individual attributes (landform, soil and vegetation), the study of correlations between these
attributes and the annotated preliminary photo interpretation maps (Hommel, 1987; Küchler &
The landscape ecological map (1 : 100 000) is thus based on 256 landform and soil observations
and 125 vegetation relevés, and covers a total surface of 167,350 ha. Consequently, the
observation density for landform and soil attributes is one per 650 ha and one per 1150 ha for the
vegetation attributes. These observation densities are theoretically just sufficient (Landon, 1991).
The representation of the identified vegetation, soil and landform types on a map depends largely
on mapping scale. In the present reconnaissance survey (scale 1 : 100 000) most of the units of
the present landscape ecological map contain mosaics of soil and vegetation types.
Map compilation has been facilitated by the GIS package ArcInfo (ESRI, 1990; 1994). Digitized
topographic base maps have been made compatible with the land attribute classification systems,
enabling spatial analysis and the compilation of land attribute and evaluation maps (Bakkum,
4.1 LITERATURE REVIEW
The geomorphology of Southwest Cameroon is described by Martin & Segalen (1966) and by
Franqueville (1973) at a scale of 1 : 1 000 000. The western part of the TCP area belongs to the
coastal lowland of late Tertiary-Quaternary age, whereas the eastern part belongs to the interior
plateau of Eocene age. At the boundary of these two zones a transitional complex is found (Sega-
len, 1967). At this transition different planation levels are present. The planation levels at 200 to
300 m and 600 to 800 m correspond with the erosion surfaces Africa II and I (Martin & Segalen,
1966; Franqueville, 1973). The two erosion surfaces have a well pronounced relief. They are
characterized by hills and dissected uplands with numerous small streams. The slopes of the hills
and uplands have convex upper parts and concave lower parts. Laterite banks are of minor
importance and the tropical clay soils are generally deep.We have subdivided the area into four
altitude classes which party coincide with the mentioned erosion surfaces. The altitude classes are
< 350 m, 350-500 m, 500-700 m and > 700 m.
Important processes leading to the formation of today's landforms have been tectonic movements
of parts of the Precambrian shield, e.g. block faulting along NE - SW lines, climatic changes and
erosion processes related to rapid changes of the erosion basis. These processes resulted in
different planation levels or erosion surfaces (Martin & Segalen, 1966; Buckle, 1978; Embleton
& Thornes, 1979). The physiognomy of the landscape is still being `reformed'. The most
important process is water erosion resulting in the dissection of the area. The intensity of this
dissection is determined by the relative differences in erodibility of both soil cover and
underlying rock, and by the amount of rainfall. Differences in erodibility of the rock are caused
by differences in density of fractures and/or its mineral composition.
Table 4.1 Relief characteristics of the different landforms