19
1993
Renaat S.A.R. van Rompaey
Forest gradients in West Africa
Aspatial gradient analysis
CENTRALE LANDBOUWCATALOGUS
iiiiuiiiiniiiiuiiui i
0000 0545 0396
uo^
Promotor:
Co-promotor:
dr.ir. R.A.A. Oldeman
hoogleraar in debosteelt en bosoecologie
dr.ir. N.R. deGraaf
universitair docent bij devakgroepBosbouw
Renaat S.A.R. van Rompaey
Forest gradients in West Africa
A spatial gradient analysis
Proefschrift
ter verkrijging van de graad van
doctor in de landbouw- en milieuwetenschappen
op gezag van derector magnificus,
dr. H.C. van der Plas
in het openbaar teverdedigen
op dinsdag 29juni 1993
des namiddags te half tweein deAula
van de Landbouwuniversiteit te Wageningen
L^v- ' ^3 l S o o
fcANDBOUWUNIVERSITEU
57AGENINGEN
ABSTRACT
van Rompaey R.S.A.R. (1993). Forest gradients inWest Africa : aspatial gradient analysis.
Doctoral thesis, Wageningen Agricultural University, The Netherlands, xxii + 142pp., 5
tables, 61 figs., 3textboxes, 3apps., 250 refs., 13items in software list, 51terms in
glossary, Eng., French and Dutch summaries. ISBN90-5485-120-1.
Forest gradients were studied at two levels of scale inWest Africa, west of theDahomey interval.
At a regional scale, the forest gradient in SELiberia and SW C6ted'lvoire was analysed using
forest inventory datafromthe pre-loggingera. Intotal 22 000ha of forest were fully inventoried.
53largetree species were*used for theordination. Spatial gradient analysis was applied to the
forest ordination scores and agradient map was produced with isoscorelines.This forest gradient
was related to climate, relief and lithology.
At alocal scale, forest gradients were studied alongthree catenas in Tai" National Park, SW C6te
d'lvoire. The sampleplots of 22to 25ha were subdivided in contour sampleplots, covering 2ha
each, using digital terrain models. Theordination of thetrees above70 cm diameter inthese
subplots allowed to conclude thatthe slopegradients are slidinggradients superimposed onthe
regional gradient and that moisture conditions are likely to control thesegradients.
Forest management should take into account these forest gradients inforest inventory and adapt
silvicultural methods and species choiceto theposition on the gradient. Conservation of biodiversity isneeded over the entire gradient and a 'Green Sickle' isproposed to link National Parks and
Forest Reserves from the savanna down to the Atlantic coast.
CIP-DATA KONINKLUKE BIBLIOTHEEK, DEN HAAG
Rompaey, Renaat S.A.R. van
Forest gradients in West Africa : a spatial gradient analysis / Renaat S.A.R van Rompaey. - [S.l. : s.n.]. 111.
Thesis Wageningen. -With ref. - With summary in Dutch and French.
ISBN 90-5485-120-1
Subject headings: tropical rainforest ; Cote d'lvoire / tropical rainforest ; Liberia / forest ecology.
®Renaat S.A.R. van Rompaey, Wageningen, 1993
All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means, electronic, mechanical, photocopy, recording, or otherwise, without the prior
permission of the author.
Front cover:
Back cover:
Printed by:
Artwork by:
Landsat TM satellite image dd. 14-12-1988, from scene 198-56,bands 4-5-7, piecewize
linear contrast stretching with breakpoints: R (44,0) and (59,255), G (32,0) and (48,255),
B (10,0) and (19,255). Processed by the Laboratory of Remote Sensing and Forest
Management, University of Gent, Belgium
Soilpeels from the soil catena of the Tai study site, units Tsl, Ts3, Ts4 and Tv. From
ISRIC, Wageningen.
Grafisch Service Centrum, Wageningen
P.G.M. Versteeg, tekenlokatie Biotechnion
A>^ Q%2D\
^Se.
Stellingen
1
De grote verscheidenheid aan bossen in West-Aftika rechtvaardigt het vervangen
van de term "het tropisch regenwoud" door de meervoudsvorm "tropische regenbossen".
Dit proefschrift
2
De boomsoortensamenstelling in de Westafrikaanse regenbossen is niet chaotisch
of toevallig bepaald, maar vertoont een duidelijke ordening langs gradienten,
bepaald door klimaat, moedergesteente en positie in het landschap.
Dit proefschrift
3
Steekproefmethoden volgens toeval, al dan niet na stratificering, benutten
onvoldoende de ordening van natuurlijke vegetatie langs gradienten.
ALDER D . & SYNNOTT T.J. (1992).
PERMANENT SAMPLE PLOT TECHNIQUESFOR MIXED TROPICAL FOREST.
TROPICAL FORESTRY PAPERS 25, OXFORD FORESTRY INST.
4
Er bestaan geen homogeneproefvlakken in tropische regenbossen, en ook geen
herhalingen.
Dit proefschrift
5
Het nemen van onafhankelijke steekproeven binnen een en hetzelfde ekosysteem is
een contradictioin terminis.
Dit proefschrift
6
Het uitblijven van overeenstemming over een wereldboskonventie tijdens de
UNCED-konferentie in Rio wordt onder andere veroorzaakt doorverschillen in
kulturele grondhouding van mensen tegenover bossen.
7
Engels is geen bosbouwtaal en vormtdus een hindernis voor het wetenschappelijk
denken over, en het publiceren van bosbouwwetenschap.
8
Rekenbladprogramma's laten toe, in tegenstelling tot databaseprogramma's,
kreatief en soepel gegevens in de computer in te voeren en teverwerken omdatin
elk hokje op het formulier de ruimtepraktisch onbeperkt is.
9
In der Beschrankung wird man niemals Meister.
10
Indien men wetenschap ziet als vissen in open water, dan heeft beleidsondersteunend onderzoek meer weg van vissen naar de goudvis in het aquarium.
11
Een kathedraal met teveel stellingen is ook niet mooi meer.
Stellingen behorende bij het proefschrift:
"Forest gradients in WestAfrica :a spatialgradientanalysis"
Renaat S.A.R. van Rompaey
Wageningen, 29juni 1993
Aan mijn vader,
Prof.dr. Jan Van Rompaey
PREFACE
De natuur heeft me altijd al geboeid enmijn keuzevoor de studievan bosbouwingenieur wasdan
ook ingegeven door het verlangen meer te Ierenover die natuur enhoe mensen ermee omgaan. Ik
volgdehet curriculum van detropischebosbouw omdatdit een wijdere horizon bood enje er
leerdeover de groteverscheidenheid van mensen, dieren enplanten over de gehele wereld. Het
was Prof.dr.ir. J. Schalck die mestimuleerde om een afstudeervak in detropente doen, o.a. met
zijn boeiende colleges over toegepaste tropischeplantensystematiek. Het boek van Marius Jacobs
(1982) over het tropisch regenwoud heb ik toen helemaal uitgelezen. Suriname werd mijn
tropendoop. Ik bestudeerde er met ir. Jan Mallants alsbegeleider enProf.dr.ir. M. Reynders als
promotor de groei van bomen in natuurlijk bos in deuitgestrektebinnenlanden. Van op de
Voltzberg keek ikvoor het eerst in mijn levenuitop nietsdan natuur, tot aandehorizon.
Deze Surinameavonturenbrengen mebij dr.ir. Reitze de Graaf, aan wie ik te danken heb dat ik
voor dit onderzoek werd aangesteld. Zijn enthoesiast vertelde ervaringen uit Suriname kwamenme
later nogvaak vanpas, enzijn begeleiding tijdens mijn AlO-schap, door dik en dun, hieldme
overeind.
Ir. Fred Vooren, directeur van het Centre Neerlandais bij Abidjan, was mijn bosbouwcollega
overzee enbron van kennisvan het Afrikaanse bos. Zijn prima ondersteuning tijdens het veldwerk
in Ivoorkust heb ikzeer gewaardeerd.
Prof.dr.ir. R.A.A. Oldeman ontstak delont van ditonderzoek en naar het eind toe was hij de
motor was omdevele ideeen en ervaringen tot een gestroomlijnd proefschrift omte vormen. De
velegesprekken die webij hem thuisvoerden, zijn voor mij eenblijvende bron van inspiratieen
verdieping.
Je voudrais exprimer ma gratitude aM. le Ministrede la Recherche Scientifique en C6ted'lvoire
qui m'a offert une authorisation derecherche dans sonpays. AussiTheo de la Station Ecologique
a Tai, Angelo Muchetti deBTA, et Claude Frischknecht de EFBA aZagne,je vous remercie pour
votrehospitalityet votre amitie\ Bien sur aussi Daouda Sidibeet Pierre P0I6qui m'ont assisteen
permanence en foret et a Abidjan et tous les Africains dontje garde debons souvenirs de notre
cooperation. Un grand merci sur leplan logistique, resp. mecaniquepour Emmanuel et Abdoulaye
du Centre Neerlandais etpour tous ceux qui ont rendu agreablemon sejour en Afrique.
Zelf eigenlijk nog een beetje student heb ik decompagnievan deWageningse studenten erg
geapprecieerd: Antoinette Staffers, Pans Boddez, Ellen Schmidt, Paul Albers, LucJans, Lourens
Poorter, Maarten deKlerk, Marc Parren, Kees Nooren, Frank Rademacher, Peter Sloot ende
vele andere die steeds maar uit Nederland kwamen aanstromen. Mijn bijzondere dank gaat uit naar
drs. Gerrit-Jan van Herwaarden diede bodemkartering van dedrie onderzoeksgebiedjes verzorgde, samen met de studenten Kees Nooren en Frank Rademacher. Mijn collega's uit andere
disciplinesbinnen het projekt in Ivoorkust (LeonieBonneliin, ir. Henk van Reuler, dr.ir. Anneke
de Rouw, drs. Hans Vellema, ir. Joep Slaats, ir. Jetse Stoorvogel, dr.ir. Wouter Blokhuis, ir.
Ellen Schmidt, ir. Peter Sloot, ir. Gerard Hazeu) maakten het onderzoek nog een stuk interessanter. Ik leerde in team te werken ende middenweg te zoeken tussen gezellig samen en kreatief
alleen bezig zijn. Dememepour mes colleguesau DCGTx, a1'IGCI, au Centre Suisse eta
FIIRSDA etpour mes amis en C6ted'lvoire. I would also liketo thank Peter Weinstabel, Jochen
Weingart and Reinhart Wolf from the German Forestry Mission to Liberia, for their kind
reception in Monrovia in April 1990. They handed meover most of the information and data from
Liberia used in this book.
InWageningen is er tussentijds en vooral na terugkeer ook goed voor megezorgd. Demensenop
Bureau Buitenland, Hans, Eric, Philip en andere, en op devakgroep Bosbouw, o.a. Marthy,
Michael, Bert, Ed, Joke, Job, Annelies waren steeds voor me inde weer. Julliehebben me steeds
vooruitgeholpen en het was meestal nog gezellig ook. Niette vergeten isook de Werkgroep
Ivoorkust die een team vormdeover degrenzen van devakgroepen heen.
Mijn dank ook aan devrienden en collega's op devakgroep Bosbouw en op andere vakgroepen,
aan dr.ir. M.A.J, van Montfoort voor het statistisch advies en aan deteken- en de fotolokatie op
de Biotechnion. Ir. Boudewijn de Roover vanhet Laboratoriumvoor Teledetektie enBosbeheersregelingte Gent maaktehet mooieplaatje op devoorkaft en wist mete interesseren voor
satellietbeelden en landinformatiesystemen. WoutBomer, fotograaf bij het ISRIC in Wageningen,
maakte deopnamen van de lakprofielen op de achterkaft. Het LEB-Fonds ben ik erkentelijk voor
definanciele steun bij de vierkleurendruk van de kaft.
My international colleagues Dr. Mike Swaine and Dr. Denis Alder are thanked for their
enthusiasm. Et Dr. Henri-F61ix Mattre et Dr. Jean Maley pour leur encouragement lors demes
visites aParis et a Montpellier.
De kritische lezers van de manuscripten, Prof.dr.ir. I.S. Zonneveld, Prof.dr. M.J.A. Werger,
dr. Frans Bongers, dr.ir. Rob Peters, ir. Lourens Poorter, ir. Marc Parren, ir. Fred Vooren
worden bedankt voor hun opmerkingen en suggesties. Het Engels indit boek was nooitzomooi
geworden zonder de rode pen van Mevr. Joy Burrough-Boenisch. Deprachtige kaarten zijn vande
hand van Piet Versteeg van detekenlokatie Biotechnion. The following authors are thanked for
their permission to reproduce figures: A.G. Voorhoeve, J.L. Guillaumet, A.P. Vooren, M.
Sachtler, and P. Bosnian. Een aantal foto's werden meter beschikking gesteld door P. Albers, H.
Dop, M.P.E. Parren en A.P. Vooren. De overige foto's heb ik zelf genomen.
Nietsvan dit alles was gerealiseerd zonder denooit aflatendezorg en liefde van Moeke, envan
mijn allerliefste Sandra.
Wageningen, 31 maart 1993
Renaat van Rompaey
CURRICULUM VITAE
Renaat SylvaAngeleRosinevan Rompaey werd geboren op 9juni 1965te Gent, Belgie. HijHep
lagere school op de Gemeentelijke Jongensschool in De Pinte. Nahet behalen van het Humanioradiploma in de richting Latijn-Wetenschappen aan het Sint-Barbaracollegete Gent in 1982,begon
hij zijn studie aan deFakulteit Landbouwwetenschappen van deRijksuniversiteit te Gent met als
studierichting 'Waters enBossen', specialisatie Ontwikkelingslanden. Tijdens het laatstejaar van
zijn studie schreef hij een afstudeerwerk met als titel: "Bosbouw inhet tropisch regenwoud.
Evaluatievan een vrijstellingsexperiment in Suriname". In 1987behaalde hij zijn diploma metde
grootste onderscheiding.
Een maand na afstuderen trad hij in dienst van deLandbouwuniversiteit Wageningen als assistent
in opleiding bij de vakgroep Bosteelt & Bosoecologie. Hij werd voor tweejaar uitgezonden naar
het Steunpunt Ivoorkust van deLandbouwuniversiteit omonderzoek teverrichten naar de
"Bostypologie en groeidynamiek van bos en bomen inhet Tai'Nationaal Park", in zuidwestIvoorkust. Nog denjaar na het aflopen van zijn aanstelling werkte hij verder aan devoltooiingvan
dit proefschrift, samen metzijn promotor Prof. R.A.A. Oldeman. Tussendoor was hij tweemaand
aangesteld voor het schrijven van een hoofdstuk over klimaatsonderzoek inzuidwest-Ivoorkustin
opdracht van de StichtingTropenbos.
CONTENTS
SUMMARY
SAMENVATTING
RESUME
XIV
XVII
XX
INTRODUCTION
Spatial gradient analysis
Forest gradients inWest Africa
History of theproject
Research objectives
Methods of approach
Outlineof thebook
1
2
FORESTS OFWEST AFRICA: SETTING THE SCENE
1.1 The former extent of dense forests in West Africa
1.1.1
Forests on the fringe of acontinent
1.1.2
The forest-savanna boundary and impacts of
1.2 Lithology of SELiberia and SW Cote d'lvoire
1.3 Three man-made forest blocks
1.3.1
Actual forest cover
1.3.2
Timber mining
1.3.3
Forest management and conservation
1
1
3
3
4
5
6
7
7
fire
THE REGIONAL FOREST GRADIENT IN SELIBERIA AND
SWCOTEDTVOIRE
2.1 The SW-oriented rainfall gradient
2.1.1
Sources of spatial rainfall variation: about interpolation
2.1.2
Temporal variation of rainfall: taking the mean?
2.1.3
Description of the rainfall gradient: a cross-section
2.2 The regional forest gradient
2.2.1
Large tree species
2.2.2
Forest inventories from the pre-logging era: a spatial gradient
analysis
2.3 Study sites and methods
2.3.1
Study sites
2.3.2
Data collection methods
2.3.3
Methods for data processing
2.4 Results
2.4.1
Species ranking
2.4.2
Sample ranking
2.4.3
Species richness
2.4.4
Spatial analysis
2.4.5
Cross-section through the gradient
2.4.6
Coverage of the vegetation gradient by the National Parks
2.5 Discussion
2.5.1
Comparison with theforest gradient in southern Ghana
2.5.2
Vegetation studies in SW C6tedTvoire
2.5.3
Forest zonation in Liberia
2.5.4
The exclusion of swamp forests
2.5.5
The issueof "primary" and "secondary" forest
2.6 Conclusions
7
10
11
13
13
14
15
17
17
20
20
22
23
23
27
29
29
29
33
35
35
38
39
40
40
40
41
41
44
47
53
53
54
3
4
FOREST GRADIENTS ALONG SLOPES INTAI NATIONAL PARK
3.1 Local forest gradients
3.1.1
Sequential models
3.1.2
Gradient models
3.1.3
How to sample the large tree species
3.2 Description of thestudy sites
3.2.1
Location
3.2.2
Climate
3.2.3
Lithology
3.2.4
Relief
3.2.5
Hydrology
3.2.6
Soils
3.3 Methods of data collection and analysis
3.3.1
Exclusion of swamp forests
3.3.2
Tree recording in nested plots
3.3.3
Definition of the contour samples
3.3.4
Ordination and comparison of the contour samples
3.4 Results:vegetation response toslope position
3.4.1
Largetree species composition
3.4.2
Tree species richness and tree density
3.4.3
Biomass and basal area
3.4.4
Comparison of the local and regional ordination
3.5 Discussion
3.5.1
Compositional gradients along slopes
3.5.2
The method of contour sampling
3.5.3
Tree density and biomass along the slope
3.5.4
Species richness gradients
3.5.5
Moisture indicator values for the large tree species
3.5.6
Moisture availability as a sitehospitality factor
3.6 Conclusions
71
71
71
73
82
84
84
88
89
93
94
94
95
97
97
101
101
103
IMPLICATIONS OFTHE CONTINUOUS VARIABILITY MODELFOR
FOREST MANAGEMENT AND CONSERVATION OFBIODIVERSITY
4.1 Forest management
4.1.1
Forest inventory
4.1.2
Assessment of forest productivity
4.1.3
Forest sensitivity to climatic change
4.1.4
Forest harvesting systems
4.2 Conservation of biodiversity
4.2.1
Land useplanning: conservation over the entire gradient
4.2.2
Corridor establishment: the Green Sickle
4.2.3
Urgent conservation priorities
4.3 Epilogue
105
105
105
106
106
108
109
109
109
109
111
GLOSSARY
SOFTWARE PACKAGES USED
APPENDICES
I List of largetree species occurring in SELiberia and SW C6ted'lvoire
II Forest inventory data from SELiberia and SW C6ted'lvoire
III Detailed description of thephysiographic units of thethree study sitesZagne\
Tai" and Para
REFERENCES
57
57
58
58
59
60
60
65
66
67
68
69
112
117
118
118
123
127
134
List of figures:
Figure 1 Relief and hydrography of Liberia and west C6ted'lvoire
8
Figure 2 Each year the savanna burns in West Africa
10
Figure 3 Aclose-up of weathered sericite-chlorite schist on the Para study site
11
Figure 4 Lithology of SELiberia and SW CSted'lvoire
12
Figure 5 A Meliaceae trunk istransported to the sawmill in Ghana
14
Figure 6 Remains of aforest elephant found in thePara study plot
15
Figure 7 Rainfall map of SE Liberia and SW C6ted'lvoire
18
Figure 8 Temporal variation in annual rainfall at Daloa (281m asl), central Cdte
d'lvoire
21
Figure 9 The everwet forests of Sapo National Park, looking towards Putu range
(Liberia)
23
Figure 10 The rainfall gradient along a cross-section through SELiberia and SW C6te
d'lvoire
24
Figure 11 Cross-section through relief and regional forest gradient in SELiberia and
SW C6ted'lvoire
25
Figure 12 Alarge Gymnostemon zaizou tree (Simaroubaceae) atthe Para study site
26
Figure 13 Forest complexes in SELiberia and SW C6te d'lvoire used for the ordination
in Chapter 2
30
Figure 14 Spatial gradient analysis of the forests of SELiberia and SW C6ted'lvoire . . . . 36
Figure 15 Number of timber species per forest inventory compartment outof a list of 53 . . 39
Figure 16 Correlation of my species scores with those of Hall & Swaine (1981)
41
Figure 17 Correlation of my sample scores with scores calculated after Hall & Swaine
(1981)
42
Figure 18 My samples placed inthe Ghanaian ordination diagram by 'coordinate
estimation'
43
Figure 19 Vegetation map of SW C6te d'lvoire as drawn by Guillaumet (1967)
45
Figure 20 Distribution of four Caesalpiniaceae tree species, characteristic of thewet
coastal forest in Liberia
48
Figure 21 Distribution of two Caesalpiniaceae tree species typical of themixed
evergreen rain forest inLiberia (from Sachtler 1968)
49
Figure 22 Profile diagram of a singledominant forest of Tetraberlinia tubmaniana
50
Figure23 Evergreen rain forest along Lofa river in west Liberia
51
Figure 24 Semi-deciduous forest in north-west Liberia upstream Lofa river
52
Figure 25 AlargePiptadenisatrum africanum crown in afieldsouth of Ta'i
54
Figure 26 ASacoglottisgabonensis tree atthe Para site
56
Figure 27 Example of a soil and vegetation catena in theTai study area
57
Figure 28 Map of Tai National Park and surrounding forest reserves
61
Figure 29 Topographical map of Zagne"survey area in its landscape
62
Figure 30 Topographical map of Tai'survey area in its landscape
63
Figure 31 Topographical map of Para survey area in its landscape
64
Figure 32 All limits around Tai National Park have been marked by signs
65
Figure 33 Thefieldteam consisting of soil and vegetation scientists crossing theBono
river to reach the Para study site
67
Figure 34 Erosion gully onthe lower slope intheTai study area
69
Figure 35 Ironstoneboulders that broke off an iron hardpan
70
Figure 36 The diameter of large trees was measured with a 2mruler
72
Figure 37 A Canarium schweinfurthiitree of 130cm diameter isbeing measured over
the paint ring
73
Figure 38 Physiographic soil map of the Zagne"study area
74
Figure 39 Contour sampleplots at theZagne"study site
75
Figure 40 Physiographic soil map of the Taistudy area
76
Figure41 Contour sample plots at theTai study site
77
Figure 42 Physiographic soil map of the Para study area
78
Figure 43 Contour sampleplots at the Para study site
79
Figure 44 Distribution of thephysiographicunits over the contour samples in Para, Tai'
and Zagn<5
80
Figure 45 Three soil catenas inTai National Park (Zagng and Tai'on migmatite; Para on
sericiteschist)
83
Figure 46 Relation between altitudeand themain vegetation gradient (DCA1)
86
Figure47 Separateordination of the study sites:DCA1scores against elevation of each
contour sample
87
Figure 48 Tree species richness and tree density per two ha contour plot, plotted against
altitude
88
Figure49 Biomass (tha"1m1) and basal area (m2ha"1)per contour sample, plotted
against altitude
89
Figure 50 A corona of large trees around the crest with ironpan at the Tai site
90
Figure51 Triplochiton scleroxylon, Erythrophleum ivorense and Ceibapentandratrees
growing preferentially on upper and middle slopepositions
91
Figure52 Thedistribution of Sacoglottisgabonensis andMaranthes glabratrees atthe
Para site
91
Figure53 Thebiggest treeof western C6tedTvoire
92
Figure 54 Correlation of the species scores along regional and slopegradients
93
Figure55 Correlation of species scores of the catena ordination withthescores of Hall
& Swaine (1981)
94
Figure56 Thehypothetical positionsof thePleistocene forest refuges in West Africa
according to Guillaumet (1967)
98
Figure57 Mount Kope\ part of thehill ridges around Grabo, ahypothetical glacial
forest refuge
99
Figure 58 Inundation of the Nse" valley bottom near the Zagne"study site
102
Figure59 Excessively logged forest in N'zo fauna reserve north of Tai'National Park . . . 107
Figure 60 The Green Sickle, agreen zone stretching from Mount Sangbe"National Park
in savanna over Ta'iNational Park towards the Atlantic Ocean
110
Figure61 A mother elephant bathingher young inthe Atlantic Ocean
Ill
List of tables:
Table 1
Rainfall stations in SELiberia and SW C6tedTvoire
Table 2
Theforest inventory compartments used for theordination described in
Chapter 2of the present book
Table 3
DCA ordination table using 53 large tree species recorded consistently inthe
forest inventories
Table 4
Geographical position and area (ha) of the soil and tree surveys at the three
study sites
Table 5
DCA ordination table of 95 large (d>70cm) trees species in 35 contour
samples, each covering 2ha
19
List of text boxes:
Text box 1 Uncertainties and scales in natural history research
9
Text box 2 Problems with combining data setsof different origin and inventory method . . . . 34
Text box 3 Facts and figures about Tai National Park
66
31
37
60
85
SUMMARY
Thetropical rain forests of West Africa, westof theDahomey interval,oncecovered some40
million ha. Being onthe western fringe of the African continent, they receive abundant rainfall
from the SW monsoon. Further inland, rainfall gradually decreases and the forests give wayto
savanna and ultimately to the Sahara desert.
This Upper Guinea forest blockused to cover most of Liberia and parts of C6ted'lvoireand
Ghana. Here, deforestation rates are amongthe fastest in the world. Humans have reduced the
forest cover by some80 %. Most of theforest hasbeen converted to agricultural land. Fireand
heavy timber mining have left the remaining forest in apoor state. Sustainable forest management
has notyet been attained. Akey prerequisite for achieving such management ismore and better
knowledgeof the ecology of these complex and highly diverse ecosystems and of their species.
Gradients are gradual changes in space e.g. of species composition inthe ecosystems. In thisbook
forest gradients are studied attwo levels of scale: the regional forest gradient from the coastto the
forest-savanna boundary, and local gradients along slopes inthe landscape. The species compositionof the large forest trees with adiameter exceeding 70 cm was studied; this entailed adapting
the existing methods used invegetation science to cope with these huge subjects. In West African
forest exploitation 70 cmdiameter is a common limit for selective cutting of trees.
At the regional scale, thespatialgradientanalysisconsisted of athree-step approach:
1. ordination of forest areas and species; 2. spatial interpolation of the ordination scores ofthe
forest areas; 3. relating spatial trends in species composition totrends inrainfall and patterns in
lithology and relief.
Tree species composition in aforest area was determined by using forest inventory data from the
pre-logging era in SELiberia and SW C6ted'lvoire. These data were ordinated together with data
from three sample plots of 20ha each established inTai National Park. The old and new sample
plots together covered 21 640ha. Of the largesttree species, 53 were retained for the ordination.
Theforest gradient was mapped by interpolating lines of equal ordination scores and plotting these
onthemap. Theordinationtable allowed theseordination scores to betranslated intothecorresponding species composition. Each tree species has an individual position along the gradient,
givenby itsordination score. The ecological range of the species is indicated by the rangeof
sample scores in which itoccurred. Thegradient map with isoscore lines provides avalid
alternative to mapping by types or classes and overcomes the problem of transitional types.
Over 400km apronounced regional forest gradient was found from the Liberian coast towards the
forest-savanna boundary in C6ted'lvoire. This gradient correlated well withtheSW-oriented
rainfall gradient. Onpart of the map the forest gradient showed an anomaly. The forests ona
band of sericite-chlorite schist from TaTNational Park towards the NE were ranked 'wetter' than
expected from their position onthe rainfall gradient. Apparently, the rainfall effect was compensated by the greater moisture content of the soils derived from sericite-chlorite schist. Furthermore, the rain shadow of Putu range (753mabove sea level) was reflected inthe forest gradient
by azone of fast compositional change. These results are comparable with those of other studies
ontheregional forest gradients in Ghana, C6ted'lvoire and Liberia.
Forest gradients along slopes spanning a20 to 40 melevation interval were studied inTai
National Park in sample plots at three locations: near Zagne"inthe drier north-west (plot size
23ha), near Tai inthe middle (25ha) and near Para inthe wetter south-west of the Park (22ha).
XIV
All trees inthese plots with adiameter larger than 70 cm were recorded and mapped. Swamp
areas were excluded from these plots. Adigital terrain model and aphysiographic soil map were
prepared for each site.
The approach used for the analysis of the slopegradient contained the same elements as for the
regional gradient, but in reverse order: 1. the soil survey revealed that the spatial trend in
environmental variation was related to local elevation; 2. thus, contoursamples of trees growing
within an elevation interval were composed. Aseries of 11or 12consecutive intervals was
calculated at each site in such a way that each sample covered 2ha. 3. Tree species composition
was determined per sample and all samples of thethree sites were ordinated together using
Detrended Correspondence Analysis (DCA). The sample scores onthe first ordination axis were
plotted against elevation to check thehypothesis of elevational gradients. Treedensity, species
richness, basal area and stem biomass per contour sample were analysed in a similar way for
elevational trends.
The contoursampling technique proved to be appropriate for the analysis of floristic slope
gradients inlargetrees. The spatial autocorrelation of the contiguoussamples is assumed to
counterbalance the small number of trees per sample. The species order resulting from this
ordination was similar to the oneobtained inthe regional gradient and hence, could be interpreted
as a wet-dry axis. Onthe three sites species composition changed from "drier" species upslopeto
"wetter" downslope. The lower slopeon adrier sitewas similar in species composition to the
upper slopesof its wetter neighbour. Thus, slopegradients are slidinggradients on the regional
forest gradient. The regional gradient was related to rainfall and lithology. These factors are
,
largely expressed in soil moisture availability. Hence, gradients in moisture availability probably \ j \ / \
explain the forest gradients along slopes.
'
Tree density and species richness of the large trees decreased towards the wet end of the gradient,
both between sites, e.g. from Zagne"to Para, and within theTa'isite, from upper slopeto lower
slope. Thistrend in species richness of the large trees contradicts one of thetenets of the
Pleistocene forest refugia theory, namely that species richness increases towards the core area of a
refugium. APleistocene refugium ishypothesized to have existed inthe Grabohills to the
southwest of Tai" National Park.
Basal area curves showed apeak in middle slopepositions and declined towards both the upper
and lower end of the catena. Ageneral trend of increasing wood density of the large tree species
was found towards the wet end of the gradient.
Some implications of the continuousvariability model are evaluated as ascientific basis for forest
management and conservation of biodiversity. At the regional scale, forest management should be
adapted to the individual position of each forest area alongthe gradient. Thetree species chosen
and the forest land evaluation should both reflect this position. Catena plots are suggested asmore
appropriate for forest inventory than random sampling in areas with pronounced slopegradients.
Any conceivable strategy for conservation of biodiversity in West Africa must aim at protecting
forests alongthe entire gradient, because at any point alongthe gradient different species attain
their optimum. A "Green Sickle" isadvocated which, if adopted and implemented, would link
National Parks and forest reserves from the savanna down to the Atlantic coast.
In western C6te dTvoirethere aretwo promising areas where conservation efforts could still
produce worthwhile results: 1. thehills of Grabo, ahypothetical Pleistocene forest refugium with
XV
a highdegreeof endemism in C6ted'lvoire, which merits the status of national park; 2. the
semi-deciduous forests, which contain many species not found in wetter forests.
Regional and local gradients result in a great diversity of forests in West Africa. The satellite
imageon thefront cover confirms thisbroad variation. Forests arenotjust hectares, trees notjust
cubic metres. Management of the most species-rich ecosystems on earth is a challenge for present
and future generations. It will need international support and the efforts of all thosefascinated by
thispearl of our blue planet.
XVI
SAMENVATTING
Bosgradienten in West-Afrika: een ruimtelijke gradientanalyse
InWest Afrika ten westen vanhet Dahomey interval lagenvroeger 40miljoen hectare tropische
regenwouden. De vochtigezuidwest-moessonzorgt voor overvloedige regenval aan deze rand van
het Afrikaanse continent. Dieper landinwaarts neemt de neerslag snel af engroeit er geen bos
meer, maar savannetotdatuiteindelijk enkel nog woestijn rest.
Dit zogenaamde bosblok van Hoog-Guinea ligtvoornamelijk inLiberia en inzuidelijke delenvan
C6ted'lvoire en Ghana. Deze regio kent Senvan dehoogsteontbossingspercentages ter wereld.
Door toedoen van de mens ishet bosareaal er teruggebracht tot £envijfde van de oorspronkelijke
bedekking. Het meestebos werd tijdelijk of permanent omgezet in landbouwcultures. Het
resterende bos is door vuur en "houtmijnbouw" zwaar toegetakeld. Duurzaam bosbeheer wordt
nog weinigtoegepast in deze landen. Zo'n beheer vergt goede kennis van de ekologie van deze
complexe enhoogbiodiverse ekosystemen envan dedier- enplantesoorten die er leven.
Gradienten zijn geleidelijke veranderingen inde ruimte inb.v. soortensamenstelling in de
ekosystemen. In dit boek worden bosgradienten geanalyseerd op twee schaalniveau's: de regionale
bosgradient vanaf de kust, waar de hoogste neerslag valt, tot waar bos overgaat in savanne, en
plaatselijke bosgradienten langshellingen inhet landschap. De studieconcentreert zich op de grote
boomsoorten die dikker dan 70 cm worden, wat inWest-Afrika een gebruikelijke diameterlimiet is
voor selektieve kap. De bestaande methodieken uitdevegetatiekunde heb ik aangepast aan het
groot formaat en het verspreide voorkomen vandezebomen.
Voor de ruimtelijkegradientanalyse inzuidoost-Liberia en zuidwest-C6te d'lvoire hanteerde ik
een aanpak in drie stappen: 1.ordinatievan bosgebieden enboomsoorten; 2. ruimtelijke
interpolatie van deordinatiescores van de bosgebieden; 3. het relateren van ruimtelijke trends in
soortensamenstelling aan de neerslaggradient en aanpatronen in moedergesteente en relief.
Deboomsoortensamenstellingvan debosgebieden haalde ikuit deresultaten van bosinventarisaties
die degrootschalige "houtmijnbouw" voorafgingen. Deze gegevens werden geordineerd samen
met gegevens van drie eigen proefperken in TaTNationaal Park, elk tussen 22 en 25 ha groot. In
totaal waren gegevens beschikbaar van 21 640 ha oude en nieuwe opnamen. De ordinatie is
gebaseerd op 53boomsoorten die degrootste bomen van het bos leveren.
Debosgradient werd in kaart gebracht door lijnen van gelijke ordinatiescore te interpoleren. Deze
scores langs de eerste ordinatieas werden aan dehand van de ordinatietabel vertaald naar een
bepaalde soortensamenstelling. Elke soort blijkt een individuelepositie langs de gradient in te
nemen. Haar ekologisch optimum wordt gegeven door deordinatiescore en de ekologische
amplitude van de soort is af te lezen aan het bereik aan opnamescores waarin desoort voorkomt.
Degradientkaart met isoscore lijnen is een alternatief voor kartering aan de hand van typesof
klassen en lost het probleem op van de overgangstypes.
Uit de resultaten bleek dat er een uitgesproken bosgradient ligt over 400 km vanaf de Liberiaanse
kust tot waar de savanne begint in C6ted'lvoire. Deze gradientloopt grotendeels parallel met de
ZW-NOgeorienteerde neerslaggradient. Op een deel van de kaart wijkt debosgradient af van de
neerslaggradient. Ditblijkt te zijn op een band met sericiet-chloriet schist als moedergesteente die
vanuit TaTNationaal Park naar het NOloopt. De mindere neerslag wordthier kennelijk gecompenseerd door meer bodemvocht inde kleiige bodems dieuit de schist gevormd zijn. Een zone
van snelleverandering in soortensamenstelling op debosgradientkaart in Liberia isvermoedelijk te
XVII
wijten aan de regenschaduw van dePutu berg (753mboven zeeniveau) enhet omliggend plateau.
De resultaten zijn vergelijkbaar met dievan andere bosgradientstudies in Ghana, Coted'lvoireen
Liberia.
Bosgradienten langshellingen met eenhoogteverschil van 20tot 40 m werden onderzocht op drie
plaatsen inTa'i Nationaal Park: bij Zagne"inhet drogere NW, bij Tai inhet midden en bij Para in
het nattere ZW. Allebomen dikker dan 70 cmwerden opgenomen en gekarteerd binnen een
proefperk van respectievelijk 23,25 en 22 ha. Valleibos werd niet inhet proefperk opgenomen.
Van elkestudiegebied werd een digitaal terreinmodel enfysiografische bodemkaart gemaakt.
De aanpak voor de analyse van dehellinggradienten was analoog met dievan de regionale
gradient, maar de redenering werd in omgekeerde volgordegevoerd: 1. de bodemkaart leerde dat
de ruimtelijke trend in milieuvariatie afhing van deplaatselijke hoogteligging; 2. dusbesloot ik,
achteraf weliswaar, contouropnamen samen testellen langsdehoogtegradient. In elk proefperk
werd een reeks van 11of 12opeenvolgende hoogte'intervallenberekend met als randvoorwaarde
dat elke contouropname 2ha moestbestrijken. 3. Deboomsoortensamenstelling binnen elke
contouropname werd vastgesteld en alleopnamen van dedrie gebieden samen werden geordineerd
met correspondentieanalyse (DCA). De opnamescores langs de eerste ordinatieas werden uitgezet
tegen hoogteliggingomnategaan of er een hellinggradient was. Ookboomdichtheid, soortenrijkdom, grondvlak en stambiomassa werden uitgezet tegen hoogteligging.
De contouropnametechniek bleek geschikt om floristische hellinggradienten te onderzoeken van
degrotebomen. De schijnbare nadelen verbonden aan het klein aantal bomen per opnameworden
ondervangen door de ruimtelijke autocorrelatie tussen aangrenzende opnamen. De soortsvolgorde
bij deze ordinatie leek sterk op dievan de regionale gradient en kon bijgevolg geinterpreteerd
worden als een droog-nat as. Naar beneden toe langs dehelling kwamen in de drie studiegebieden
steeds 'nattere' soorten voor. De soortensamenstelling beneden aan dehelling bij Zagne'slootaan
bij diebovenaan dehelling inTai en idem dito voor Tai en Para. Dit geeft aan dat de hellinggradienten schuivendegradienten zijn over deregionale bosgradient. Neerslag en moedergesteente
bepaalden grotendeels de regionale gradient enbeide faktoren komentot uitdrukking inhet
beschikbaar vocht indebodem. Het isdus aannemelijk dat hellinggradienten in de boomsoortensamenstelling verklaard kunnen worden vanuitbodemvochtgradienten langsdiehelling.
Dedichtheid en soortenrijkdom van grote bomen nam af naar het natte uiteindevan de gradient;
dit washet geval tussen studiegebieden, van Zagne"naar Para, maar ook binnen het proefperk in
Ta'i,van bovenaan dehelling naar beneden toe. Deze trend in soortenrijkdom gaat integen 66n
vande stellingenvan de Pleistocene refugia theorie, namelijk dat de soortenrijkdom zou toenemen
naar het kerngebied van een refugium toe. Er wordt een refugium verondersteld ten zuidenvan
Ta'iNationaal Park bij deheuvels rond Grabo.
Het grondvlak bleek het hoogst op het midden van dehelling en neemt af naar detop en naar de
vallei toe. Degemiddelde dichtheid van het houtvan degrote boomsoorten leek toe te nemen naar
het nattereuiteindevan de gradient.
De continueverandering van debossamenstelling heeft implikaties voor het bosbeheer enhet
natuurbehoud. Langs de regionale gradient moethet bosbeheer aangepast worden aan de
individuelepositievan elk bosgebied. Dit uitzich dan in een aangepasteboomsoortenkeuze en een
individuele evaluatie van de standplaats. In debosinventarisatie sluiten catena-opnamen beter aan
bij de realiteit van degradienten langshellingen dan willekeurig uitgelegde opnamen.
XVIII
Natuurbehoud in West-Afrika moetzich richten op bescherming van bossen over de gehele
gradient. Op elkepositievan dezegradient zijn er weer andere soorten diehun optimum bereiken.
Een 'Groene SikkeF wordt voorgesteld om Nationale Parken en bosgebieden te verbinden vanaf
de savannetot aan de Atlantische kust in Liberia.
Binnen westelijk Cdted'lvoire worden in het bijzonder twee gebieden genoemd waar natuurbescherming nogveel resultaat kan boeken: 1. deheuvelruggen van Grabo, een verondersteld
Pleistoceen regenwoudrefugium met vele soortendiebinnen C6ted'lvoire enkeldaar voorkomen.
Dit gebied verdient de status van Nationaal Park; 2. dehalf-bladverliezende bossen dieook veel
specifieke soorten herbergen.
De regionale en lokale bosgradienten geven aanleiding tot een grote verscheidenheid aanbossen in
West-Afrika. Het satellietbeeld op de kaft van dit boek staaft deze rijkdom. Bossenzijn meer dan
hectares land enbomen meer dan kubieke meters hout. Een duurzaam beheer van de soortenrijkste
ekosystemen op deze aarde vormt eenuitdagende taak voor deze en volgendegeneraties. Diezal
gedragen moeten worden door de hele internationale gemeenschap en door ieder die gefascineerd
raakte door dezeparel van onzeblauwe planeet.
XIX
RESUME
Gradients floristiques dansla foret dense de 1'Afriquedel'Ouest: uneanalyse
spatiale
II yun Steeleet al'Ouest deFintervalledeDahomey, 1'Afrique &aitcouvertede40 millions
hectares deforSt dense. Les pluiesdu mousson arrosaient abondamment cettebordure du continent
africain. Plus a Finterieur, lapluviosite'diminuait et la savane et finalement led&ert succ&leraient
a laveg&ation forestiere.
Ce qui reste maintenant de cebloc forestier, ditede "Haute Guinee", setrouve au Libera etdans
les parties mendionales de la C6ted'lvoire et du Ghana. Cette region connaitun des plus forts
taux dedSboisementdu monde. L'homme ar&luitle couvert forestier jusqu'a un cinquiemede
1'original. La plupart de cesterrains ont &6 converti enterres agricoles. La reste de la foret a6t6
gravement d&erioreepar les feux debrousse etpar l'exploitation 'miniere' du bois. Les forets n'y
sontpas encore g6rees defacon soutenu. Pour developper une gestion soutenue, ilest indispensable de reunir au prealable des bonnes connaissances autant de l'6cologie de ces Scosystemescomplexes et hautement diverses, que des especes veg&ales et animales qui les constituent.
Un gradient constitueunetransition graduelle dans l'espace, par exemple, de la composition
floristique. Dans ce livrej'ai analyst des gradients floristiques dans la forSt dense adeux niveaux
d'integration, respectivement le gradient regional climatique allant de la c6te ala limite foretsavane, etdes gradientstoposequentiels.L'&udeportait sur les grandes especes arborescentes,
pouvant atteindreun diametre de 70 cm, limitehabituelle de la coupe selective en Afrique de
l'Ouest. Les m&hodeshabituelles phytosociologiques ont done 6t6 adaptees a la grande taillede
ces arbres et aleur abondancesouvent faible si Ton comptepar hectare.
Pour Vanalyse spatialedu gradientregional dans le Sud-est du Libera et dans le Sud-ouest dela
Coted'lvoire j'ai proc&le'en trois etapes: 1. en appliquant l'analyse factorielle de correspondance
aux forets et aux especes; 2. en calculant Interpolation spatiale des scores des forets; 3. en
etablissant les correlations entre les tendances spatiales dans la composition d'especes et le
gradient climatique regional, la roche-mere et le relief respectivement.
Afin de caracteriser la composition floristique des forets, j'ai utilise1les donneesprovenant des
inventaires nationaux au Liberia et en C6te d'lvoire. lis d&rivent la situation de ces forets avant
la grande vague d'exploitation forestiere. En complement de ces anciennes donneesj'aiStabli
l'ordination des donnees prelevees dans cebut sur trois parcelles d'&udedans le Pare National de
Ta'i, couvrant entre 22 et 25ha chacune La surface totale des forets inventorizes autrefois etplus
recemment est de 21 640ha. L'analyse de correspondance est bas6sur 53 especes de grands
arbres.
Le gradient regional, allant de la foret sempervirente a la foret semi-decidue, a 6t6 visualiseen
utilisant des courbes interposes de score egal. Ces scores le longdu premier axesont lies ala
composition floristique par letableau de l'analyse de correspondance. Chaque espece occupe une
position individuelle le long du gradient. L'optimum ecologiquede l'espece est donn6par son
score, et l'amplitudede l'espece est proportionnelle al'intervalledes scores des sites ou l'espece
est pr&ente. La carte du gradient avec des courbes d'isoscore fournit une alternative pour les
cartes visualisant des types deforet. Du coup, leproblemedes types transitionnels se trouve
resolu.
XX
Les r&ultats de l'analyse indiquent qu'il existeun gradientfloristiqueprononce"sur les 400km
entrela cdteliberienneet la limiteforet-savane en C6ted'lvoire. Ce gradient suit largement le
gradient pluviom&riqueorientedu SW au NE. Surunepartie dela carte, legradientfloristique
semblait devier du gradient pluviom&rique. Cettedeviation s'expliquait par unebande de schiste
senciteux qui setrouvedans unepartie du Pare National deTaTet qui s'etend vers le nord-est.
Apparemment, il existeune compensation entre unepluviosite"moindre et une capacityplus grande
deretentiond'eau des solsschisteux. Unezonedetransitionfloristiqueplus rapide au Liberiase
trouve al'abri des pluies du au Mont Putu (753 m) et sonplateau. Les resultats sejoignent aceux
d'autres eludesde gradients floristiques au Ghana, en C6te d'lvoire et au Liberia.
Atrois endroits dans le Pare National deTai',j'ai etudie le gradientfloristiquele long d'une
toposSquence. C'etait pres deZagnedans unezone moinspluvieuse au nord-ouest du Pare, pres
deTai au milieu et pres dePara dans le sud-ouest du Pare ou la foret recoit plus depluviosite.
Les pentes couvraient un intervalle de 20 a40 m en altitude. Tout arbre deplus de 70 cm de
diametre a ete inventorie et sapositionportSesur la carte dans des parcelles de23,25 et 22ha
respectivement. Un modelede terrain digital et une cartephysiographique ontdes sols ete
construitspour chaquepenmetrederecherche.
L'approche suiviepour l'analyse des gradients toposequentiels est analogue a celui du gradient
regional, mais le raisonnement a etesuivie en ordre inverse. 1. La carte des sols demontrait que
la tendance spatiale dans les conditions environnementales dependait de l'eievation. 2. Alorsj'ai
delimit^, aposteriori, des sous-parcellesle longdes courbes de niveau. De cette faconj'ai
compost dans chaque site 11ou 12sous-parcelles, chacune couvrant 2hectares. 3. La compositionfloristiquedes grands arbres a etecalcule"pour chaque sous-parcelle et une analysede
correspondance a eteeffectue avec ces releves toposequentiels. La relation entre les scores lelong
du premier axe et l'eievation de chaque releve"a eteetudiee. La densite"d'arbres, larichesseen
especes, la surface terriere et la biomasse des troncs ont ete analyse"le long destoposequences,
eux, aussi.
La techniquedes releves toposequentiels s'est prouve efficace pour l'analyse des gradients
floristiques parmi lesgrands arbres. Ledesavantageapparent du petit nombred'arbres par relevS
a ete corrigepar 1'autocorrelation spatiale entre releves voisins. L'ordre des especes le longdu
premier axe ressemblait beaucoup celui le long du gradient regional. II rut done interpretecomme
un axe indiquant l'humidite du milieu. Sur les trois pgrimetres, des especes "humides" se
trouvaient en bas de pente. La compositionfloristiqueen bas de pente aZagne correspondait a
cellede lahautepente aTai. II en etait de mSmepour Tai et Para. Ceci montreque les gradients
toposequentiels sont des gradients "glissants superposes" sur le gradient regional. La pluviosite" et
la roche-mere d&erminent largement legradient regional. Tous deux s'expriment dans l'humidite
du sol. Les gradientsfloristiquesle longdes toposequences s'expliquent done en majeure partie
par des gradients d'humidite"du sol d£riv£s.
La densited'arbres et la richesse en especes de grands arbres diminuaient tous les deux vers
l'extremitehumidedu gradient; ceci etait le cas en allant deZagne a Para, mais aussi a l'interieur
de la parcelle de Tai, suivant l'aval de la pente. Cettetendance de richessefloristiquedecroissante
parmi les grands arbres contredit Tunedes theses de latheoriedes refuges forestiers dans le
Pieistocfene, quipretend quela richesse en especesdevrait augmenter vers le centreplus humide
du refuge. Je suggere que les collinesde Grabo, au sud du Pare National deTai, faisaient partie
d'un tel refuge parce que les especes les plus "humides" ysontpresentes.
XXI
La surface terriere culminait a mi-pente et diminuait aussi bien vers le sommet quevers lebasfond. Par contre, ladensitymoyennedu boisdesgrands arbres augmentaitvers l'extremite'
humidedu gradient.
Le changement continu et non pas discret de la composition floristique des forets Ouest-africaines
a des implications pour leur amenagement et pour la conservation de la biodiversity.Le longdes
gradients, les mesures sylvicoles doivent s'adapter alaposition individuellede chaque partie de
foret. Ceci s'exprime dans le choix d'especes et dans revaluation individuellede chaque site
forestier. Dans lapratiquede l'inventaire forestier, ce sont des parcelles couvrant unetopos6quence entiere qui correspondent au mieux ala realite degradients toposequentiels, et non pas des
parcelles choisies stochastiquement.
La conservation de la biodiversityen Afrique deFOuestdevrait comprendrelaprotection de forets
tout le long du gradient regional. Chaqueposition sur ce gradient indiqueFoptimum ecologique
pour d'autres especes. Un 'Croissant Vert' est propose'qui sefonde sur Fanalyse des gradients et
qui devrait Her entre eux les Pares Nationaux et les Forets Classees de la savane a la c6te
atlantique au Liberia.
Dans la partie ouest de la Coted'lvoire deux regions existent ou la conservation de la naturepeut
encore espeYerbeaucoup de succes. 1. Les collines de Grabo sontun refuge forestier glacial
supposeou poussent beaucoup d'especes trouvees nullepart ailleurs en C6te d'lvoire. Ces forets
mentent le statut dePare National. 2. Les forets semi-d£cidues qui a leur tour hebergent
beaucoup d'especes sp£cifiques, meritent d'etre protegees au memetitre.
Les gradients floristiques regionaux et locaux produisent unegrande diversityde forets dans
l'Afrique de FOuest. L'image satellite sur la couverture de ce livre est unebonne illustrationde
cette richesse. Les forets sont plus quedes hectares deterrain vert, les arbres plus que desmetres
cubes debois. La gestion soutenue de ces ecosystemes les plus divers du mondeforme undeTi
pour les generationsactuelles et futures. Ceci demandera le soutien de la communaute internation a l entiere et de chacun qui setrouve a sontour fascine par cetteperleverte de notreplanete
bleue.
XXII
INTRODUCTION
Agreen ocean from a bird's eyeview, but an ever changing kaleidoscope for the
experienced observer; such are the forests of West Africa. It isall *forest but no two
forests are the same. Atfirstglance noorder can be detected in this "green hell"
(Oldeman 1983), but fairly quickly one learns to distinguish swamp forests from *upland
forests. From the coast inland these upland forests change from evergreen to semideciduous. *Gradients, i.e. gradual changes in space, were the subject of this research.
They were studied in the large forest trees, and therefore the existing methods used in
vegetation science had to be adapted to cope with these huge subjects. These forest giants
are theglory of theWest African rain forests (Rollet 1974).
Spatial gradient analysis
Forest gradients are studied in thepresent book at two levels of scale: the regional forest
gradient from the coast to the forest-savanna boundary, and local gradients along slopes in
the landscape. On a vegetation mapof the world (Walter 1979), West Africa and Russia
are the classic examples of zonal vegetational changeparallel to the equator. On these
large, flat land masses thispattern is not disturbed by mountain ranges as in the
Americas. So, it is not surprising that it was the Russian geobotanist L.G. Ramensky
(1910ex Sobolev and Utekhin 1973; 1924 ex Whittaker 1967) whofirstformulated the
principles of species individuality, vegetation continuity and ecological series, and
practised the method of *ordination of vegetation releves.
Ordination, i.e. the arrangement of samples or species in a uni-or multi-dimensional
order, was the translation by Goodall (1954) of the German "Ordnung" as used by
Ramensky (1930). Curtis &Mcintosh (1951) and Bray &Curtis (1957) applied ordination
to releves from an upland forest *continuum in Wisconsin (USA). They recognized that
theposition of a species on such a continuum is controlled by both thebiotic and the
physical environment. In thebiotic control model, competition or predation between
organisms is considered tobethe primary *factor structuring the continuum, whereas in
the environmental control model, factors from the abiotic environment are seen as
determinants of the variations observed in the occurrence or abundance of organisms
(Borcard et al. 1992). Hence, a species may beabundant in a releve either because abiotic
conditions are optimal for it, or because possible competitors are absent.
From this concept Bray &Curtis (1957) concluded that *patterns and gradients present in
thevegetation itself should be madeclear first, before searching to explain this variation.
The explanation may lie in the vegetation itself and in its environment. They called their
approach a 'direct ordination' of the vegetation data set. Whittaker (1967) termed it
'indirect gradient analysis' in opposition to hisown approach in which compositional
gradients are directly related to environmental gradients.
I agree with Hill (1973) that Whittaker's terminology of 'direct' and 'indirect' gradient
analysis is confusing. Ordination is a method or technique and gradient analysis is the
Each term preceded by an asterisk, is defined in the Glossary.
2
Introduction
field of research in which it is applied. In the same way, *clustering isa method and
vegetation ^classification and mapping is the field of research in which it is applied
(Zonneveld 1988).
What is a *gradient? Literally, it means "stepping", from Latin gradi: to step, to stride.
Whittaker (1967) applied the term to the gradual change in vegetation or environment
when going up a mountain slope or, at the same elevation, when visiting slopesof
varying exposure around a hill. Here the gradients have a spatial connotation and it is
indeed possible to stride along them. In mathematics, the gradient vector, symbol
Vf(x,y,z), at apoint has the direction of thegreatest change at that point along a
continuous line, surface or space and as magnitude the rate of change in this direction
(Gillman &McDowell 1973). Thus, the gradient of a surface indicates the steepest slope
at a point and thecorresponding direction. For non-planar surfaces the gradient changes
from point to point. No confusion should arise between the gradient and the surface itself,
the former being determined by the directional derivatives of the latter. If the surface is
meant, the term ""continuum' is to bepreferred. If thedirection and rate of change in a
point are to be expressed, the term 'gradient' isappropriate.
The term 'gradient' has also been used in an abstract sense (Sobolev &Utekhin 1973;ter
Braak &Prentice 1988), e.g. when samples from non-contiguous areas are compared and
arranged to form an imaginary continuum, e.g. ranging from nutrient-rich to nutrientpoor. Here, the connection with physical space is lost. This leads meto distinguish spatial
and imaginary gradients.
Concerning environmental variables Ramensky (ex Sobolev &Utekhin 1973) warned
against confusing 'ecology' with, what hecalled ""topology' and insisted upon separating
directly acting ecophysiological *factors like heat, light, water, nutrients, etc. from the
condition of the locality described in terms of geomorphology, lithology, hydrology. The
locality or site can be characterized by its elevation, slopeposition, soil and climate.
Ramensky (1930) claimed that the former factors cannot easily be derived from the latter
conditions. Infieldecology the site conditions can be described visually but the factors
acting directly can only be assessed by measurements (see also van Wirdum 1979).
Ter Braak &Looman (1987) analysed the supposed response of species and vegetation to
each directly actingfactor by regression and constrained correspondence analysis. For the
study of patterns and spatial gradients in vegetation and their relation tosite conditions
and for the sake of clarity I propose the term 'spatialgradient analysis', even although
this might bea pleonasm in Whittaker's sense (see also ter Braak 1987b). A three-step
approach isproposed: first, ordination of samples and species using detrended correspondence analysis (DCA), second, spatial analysis of the sampleordination scores
using *kriging and cross-sections and third, relating this spatial compositional surface to
spatial *trends in site conditions (e.g. altitude, distance to coast, rainfall at the regional
level, and slopeposition, soil at the local level).
Forest gradients in West Africa
Since thebeginning of this century botanists and foresters have tried to describe the
regional forest gradient in West Africa (Chevalier 1909, Schnell 1950,Taylor 1952&
1960, Mangenot 1955, Aubreville 1959, Voorhoeve 1965, Guillaumet 1967, Sachtler
1968, Guillaumet &Adjanohoun 1971). They delimited forest types or vegetation zones
by listing characteristic plant and tree species for these types and mapping them. Only in
recent decades have authors based their typologies on numerical clustering and ordination
techniques (Hall &Swaine 1981 for Ghana, deRouw 1991 for the Tairegion).
Thepurpose of vegetation mapping in West Africa was mainly descriptive and sometimes
explanatory. Biologists and ecologists wanted tobetter understand the change in plant and
animal species composition of the forest ecosystems from the coastal areas with heavy
year-round rainfall to the interior with its strongly seasonal rainfall regime (Hall&
Swaine 1981,deRouw 1991). Much attention was alsopaid to special types of vegetation
on inselbergs, floodplains, coastal and montane vegetation types (Guillaumet &
Adjanohoun 1971). Guillaumet (1967) was oneof the first to describe forest variation
along catenas. In hispioneering study of flora and vegetation of SW Cote d'lvoirehe
mentioned some of the issues worked out in thepresent book, but, at that time, adequate
computational techniques were not yet easily available and accessible.
Forest managers, in contrast, need to evaluate theposition of a given forest reserve in a
spatially varying environment, in terms of forest *dynamics, "growth and yield" and
*sustainability per site. At the same time, the forest manager aims to obtain information
on the ecology of large tree species, e.g. exceeding 70 cm diameter, many of which used
tobeconsidered as product carriers. Hence, his management decisions and options
concerned large trees mainly during the long period when silviculture was identified with
'the commercial cultivation of trees for theproduction of timber' (James 1982,Gliick
1987). In this context, foresters used to focus mainly on thecommoner kindsof forests,
as reservoirs of large, accessible quantities of standardized timber. They generally
neglected the rarer forest types because these contained small quantities of wood per
hectare and, furthermore, difficult access alsohindered commercialization. So, swamps,
inselbergs or mountain forests on steep slopes received less attention.
In thepresent book, theabove forestry approach was thebasic concept underlying the
research. Building on this basically timber-oriented concept, improved management
systems close to nature and the knowledge systems necessary for their design and
development were sought and found. On theproduct and service side, the new concept
now being developed is nested multiple use (Bonn^hin 1992, Oldeman 1992), in which
timber does not rank higher than other products.
History of the project
In 1975MAB-Unescoinitiated the "TaiProject" on "Recherche etamenagement en
milieuforestier tropical humide".Tai is the name of a small town along the Cavalla
river which forms theborder between Cote d'lvoire and Liberia (Figure 1, p. 8). The
IET (Institut d'Ecologie Tropicale, Abidjan) research station, 20 km south-east of Tai,
was constructed in 1977as thefieldbase of thisprogramme. To theeast of Tai lies Tai
National Park (446000 ha) which is one of the last vast areas of unlogged forest left in
4
Introduction
West Africa (Text box 3, p. 66). The interdisciplinary research programme focused on
biotic and ecological aspects insidethePark and on the socio-economic setting around the
Park (see Guillaumetet al. 1984 for a list of publications). From 1975to 1983fieldwork
was carried out by ORSTOM, IETand numerous other participants. Wageningen
Agricultural University participated in this programme mainly with agricultural and
forestry studies. Vooren (1985) found considerable local differences in tree species
composition and forest dynamics along a catena near the IET field station. DeRouw
# (1991) compared plant associations in primary forest along a 50 km stretch of the climatic
gradient between the coast and the drier interior (Figure 7) and on several bedrock types
and concluded that climate and bedrock control the floristic composition to a great extent.
In 1985 theUniversity set up theprogramme "Analysis anddesignof land-use systems
inthe Tairegion", co-funded by the Tropenbos Foundation. Conservation of the
National Park and management of its buffer zone were the major objectives theprogramme was to support with research. Forest and soil scientists, agronomists and rural
health specialists did studies on the existing land-use systems in order to improve them
and particularly to make them more *sustainable (WAU 1991).Within the research theme
"Silviculture andsustained timberproduction systems"I was given the task to study the
regional variation of the forests in 1987in order to see to what extent the findings of
Vooren (1985) could beextrapolated over a larger area, i.e. theentire National Parkand
surrounding forest reserves (Figure28).
By the end of 1991, funding of the programme phased out and the University launched a
new programme in Burkina Faso. TheTropenbos Foundation carried out an "Etude de
synthese"in 1992which will result in a complete account of all the research carried out
in SWCote d'lvoire. At the same time GTZ (German Technical Cooperation) is setting
up aPark Management project which aims to implement the research findings. This
project intends to ensure the sustained preservation of thePark by providing solutions to
theland hunger problems caused by an ever-growing population around thePark. Since
1990 the immigration of Liberian war refugees has further intensified thepressure on the
remaining forests.
Research objectives
The objectives of the research setout below were to describe and analyse forest gradients
in West Africa both at a regional and a local scale, soas to contribute to the knowledge
base needed for innovative multiple use forest management and nature conservation.
The following questions were tackled:
• How can sampling techniques be adapted to the large tree species, given their huge
size and often low population density?
• What techniques can be used to map thevegetation as a continuum rather than as
types or zones?
• How can the impact on species composition of both regional and local environmental
gradients be analysed simultaneously at thelevel of a forest sampleplot?
Methods of approach
Scope. Thepresent book covers the forest areas of SELiberia and SWCoted'lvoirein
their pre-logging state. In both countries, these are the regions which are still themost
densely forested (Sayer et al. 1992). The research programme of Wageningen University
focused on the region around Tai, a town on theborder between C6ted'lvoire and
Liberia. None of theearlier studies transgressed national borders. For aproper understanding of the forests around Tai, data from both sides of the border need to beintegrated.
Regional forest gradient. I collected data on three sites (each circa 20 ha) in unlogged
upland forest in Tai National Park and compared the species composition with theresults
of thepre-logging forest inventories in SE Liberia and SW Coted'lvoire covering a total
area of 21 640 ha of sampleplots.
The research strategy consisted of a three-step approach:
1) Ordination: toperform an ordination on releves large enough to contain reliable
information on 53of thelargest tree species;
2) Ordination -»space: the sample scores on the first ordination axis, which describes
most of thevariation, were plotted on the map of the region. For the regional forest
gradient, contours were interpolated using the *kriging method (Stein &Corsten
1991).
3) Space-»environment: a cross-section through this interpolated compositional surface
was compared with a similar cross-section through the rainfall gradient and plotted
against elevation above sea level and distance to the coast. Explanation of deviations
between vegetation and rainfall gradient were sought on the lithological map.
Local catena gradients. Aproblem was immediately apparent: how should the samples
to be used in the ordination be composed? In classical vegetation science, a 'site' or
'quadrat' is thebasic sampling unit in the field to begin with (Mangenot 1955,ter Braak
1986). In contrast, Alder &Synnott (1992) recommended that the coordinates of each tree
should be recorded for permanent sampleplots in tropical rain forest. In the field andon
the record sheets, the tree is thebasic sampling unit. Theproblem of composing samples
for ordination analysis only arises when one starts to process the data.
My approach had the same rationale as for the regional gradient, i.e. ordination «=> space
<=> environment, but now I reasoned backwards in this three-step approach:
1) Environment -»space: I knew from the soil survey that the soil variation was strictly
related to the catena position and that the spatial trend in environmental variation is
therefore related to local elevation.
2) Space-»ordination: I composed samples in such a way that they were constrained to
follow this same spatial trend. Trees were grouped along contour lines into samples
covering 2 ha each, containing about 20 trees exceeding 70cm diameter. I propose
the name *contoursampling for this technique. In this way, at least 10samples were
available per site for the ordination. Ninety-five tree species exceeding thelower
diameter limit of 70cm were used for theordination.
3) Ordination of the contour samples of the three sites together to test twohypotheses:
Is compositional change related to catena position?
- Does the catena gradient correspond with the regional gradient, and if so, might
both be controlled by the same environmental factor or factors?
6
Introduction
Fieldwork. The field work for thepresent study was carried out in Tai National Park
during my sojourn in Coted'lvoirebetween September 1988and April 1991.In 1989I
established two study plots of over 20 ha each in addition to theplot of 10ha set out by
Vooren in 1981 near Tai which was enlarged to 25 ha. All three plots were situated
within Tai National Park and in unlogged forest (see Figure 28). Oneplot was near
Zagne in the northwestern corner of the Park on ferrallitic soilsdeveloped from migmatitic bedrock, and where there is less rainfall than in Tai. The second plot lies near Para in
the southwestern corner of thePark where rainfall ishigher and where the sericite schist
bedrock has weathered to yield soils with greater water retention capacity.
Outline of the book
In the first Chapter the general forest setting in West Africa is described. Here, West
Africa isconsidered tobe the coastal part of Africa west of the Dahomey (B6nin)
interval. Relief and lithology are sketched as the spatial environment considered invariable
with time. Throughout thebook, both form the warp through which the weft of moisture
conditions and tree species composition is woven.
In the second Chapter, I position my three study sites on the regional forest continuum by
ordinating them together with data from pre-logging forest inventories, carried out in the
sixties and seventies in SELiberia and SW Cote d'lvoire. Fifty-three of the largest tree
species are used for this ordination. The regional forest gradient is mapped by plotting
isoscore contours and related to mapsof the SW-oriented rainfall gradient and of
lithology.
In Chapter 3the "catena position" isalso taken into account by ordinating tree species
composition between and within three forested catenas. The three sites are known from
Chapter 2 to receive increasing amounts of rainfall from N to S. Gradients are also
sought in tree species richness and total stem biomass. The *contoursamplingtechnique,
i.e. grouping trees along contour lines, is discussed in detail. Large tree species composition on these contour samples is ordinated and the scores are plotted against elevation and
in this way related to soil characteristics. The correspondence between these local catena
gradients and the regional climatic gradient is discussed.
In Chapter 4 the implications of this model of continuous forest variability are discussed
for forest management and conservation of biodiversity. Recommendations are madeto
develop forestry on a moreecological basis and to extend conservation efforts in the
region by interconnecting the remaining forest areas.
Several text boxes provide additional information on certain topics mentioned in the text.
Conclusions are presented in the text, in italics, as they arise. Chapters 2 and 3are set up
in such a way that they can be read as independent parts of the book. The final Glossary
contains definitions of the terms preceded by an asterisk in the text. In addition to the
literature references, all software used is listed separately. Dutch readers should notethat
the reference list is arranged in international alphabetical order, i.e. names beginning with
"de" or "van" are placed under D and V respectively.
1
FORESTS OFWEST AFRICA: SETTING THE SCENE
In this Chapter thephysiographic setting of SELiberia and SW Cote d'lvoire is sketched
and thedevelopment of the former into theactual forest cover isbriefly recounted. Relief
and lithology are considered as landscape constants, with negligiblevariation in time,and
climatic or soil moisture and tree species composition willberelated asvariables tothis
fixedframe here and in succeeding Chapters.
1.1 The former extent of dense forests inWest Africa
1.1.1 Forests onthe fringe of a continent
West Africa, to thenorth of thecoast from Conakry to Accra, was oncecovered by some
40 million ha of dense forest. ThefirstEuropean explorers might have thought that the
entirecontinent was covered by the hugetrees they saw along thecoast and along the
rivers they used topenetrate into thecountry (Martin 1989). This was not so. Only a
fringe extending somehundreds of kilometres inland consisted of dense *rain forests. The
interior soon became drier and the forests gave way to savanna and steppe vegetation.
Such coastal fringes of forests are also known from theAtlantic coasts of Brazil and
Mexico, from the western coast of India, from north-east Australia and east Madagascar
(Whitmore 1990).
There isplenty of evidence that theextent of dense forests in West Africa has varied
greatly during the last million years (Guillaumet 1967, Fritsch 1980, Maley 1991). The
iron pans capping certain crests in SW Coted'lvoire were formed under a savanna
climate (Ahn 1970, Fritsch 1980), so theentire forest must have expanded recently, i.e.
since circa 10000 years ago, from its Pleistocene refuges. On theother hand thepresent
disjunct distribution of many species between Liberia-SW Cote d'lvoire, SW Ghanaand
sometimes Cameroon-Gabon suggests that in therecent past, i.e. between 9000and
4000 years BP (Maley 1991) those three ranges have sometimes been united in onevast
forest about triple the area of today's (Guillaumet 1967;Hamilton &Taylor 1991).
Phytogeographers and bioclimatologistsreadily associated these areas with a climatic zone
called after the forests themselves: "the tropical rain forest climate" (Koppen 1936,
Walter &Box 1976). As will be shown in thenext chapter, theclimatic conditions
changegradually from constantly humid on the coast toperiodically dry further inland. It
ispossible to draw lines of equal climatic conditions (isolines) on the mapof West
Africa. These lines are not boundaries of climatic types, i.e. they are not markers of
discontinuities in climate.
Thephysiographic mapof West Africa, i.e. a maprepresenting the natural appearanceof
the continent, is the starting point for this ecological study of the forests west of the
Dahomey interval (Figure 1). However, I experienced the mapping of relief as compromising between scale and precision. Text box 1 goes into this matter further.
Setting the scene
z.~>''',DAHO.MJt.Y....A
2f
I9&9-..-A
<
2:V
~
z< "
;.
^
3:
<
a
CI
<
•' - - -
Q:
1
••-,s
UJ
•; 'UJQC
.
1
1-0
c
&
d
-D
*"*»—; •8£
ij
<
*!
?
^ _X
\ <
LU
O
.-.."t.
s
/"
8
-
r?
a
8
'< = , *$s>y/o
S
a oz // >0
. ' " T => :
' 1 U3 )
_,,;»
<rv-.
•- '
13
2?,^P\ /X
Uj'llO /7
LULU/
«-
S
•9
e
0'
Lflp--
•
8
§
8
4)
S
0
0
3 8
<*H
4>
fl
^
>
s^
3*
.8 1
:
>H
a
T3
aj
0
<o •a
U
5
i|
_ fi
eg 00
"C ON
1>
T-I
•a a
O cd
•3
§ .3
a <=
^ S
Relief
In thepresent book "West Africa" does not include Nigeria, Cameroon nor Gabon, even
though these lie on the west sideof Africa. The "Upper Guinea" forest block (Hall&
Swaine 1981) of Liberia, Coted'lvoire and Ghana forms a natural unit, especially apt for
this study on relations between forest vegetation and the abiotic environment. I shall use
thename "Coted'lvoire" instead of "Ivory Coast" as this iswhat the "Ivoiriens" wish
their country to be called.
A mountain range, the Guinea Highland, spreads from theFouta Djalon in Guinea tothe
'Monts des Dans' in west Coted'lvoire with several summitsexceeding 1000mabove
sea level (asl). This mountain range certainly contains its own forest vegetation gradients
and forest boundaries (Schnell 1950, Guillaumet &Adjanohoun 1971,Johansson 1974).
Thepresent book, however, only deals with lowland forest (below 500 mand generally
between 100and 300 masl). Thepresence of these mountains also greatly influences the
rainfall patterns, as will be discussed in Chapter 2. In the lowland forest only thePutu
range in Liberia and the hills of Grabo cross the steadily increasing elevation of the
landscape from the coast into theinterior (see mapsin IFAN 1968).
Theold African *peneplain (Ahn 1970) thatlies under thewet coastal fringe of forests, is
intensely dissected by streams and rivers (see Figure 1). The Cavalla river, called
'Cavally' in Coted'lvoire, with the small town Taion its border will remain in thecentre
of many maps throughout the present book.
Text box 1
Uncertainties and scales in natural history research
When preparing Figure 1and further maps for the present book depicting physiographic features, I noticed
that I almost automatically adjusted precision and detail to the scale of the map. However, the information
content of the figure was not really reduced by these generalizations and smoothing. The following
example taken from Mandelbrot (1977, ex Gleick 1991) brought an explanation of these phenomena.
How long is the coastline of England? On each map the outcome is different and the more detailed the map
and its scale, the longer the coastline becomes. One could also measure it on a 1:1 scale in the field, but
the more points used to pinpoint the coastline, the longer the coastline becomes. When using a measuringrod instead of topographical instruments, the coastline becomes longer in a proportion inverse to the length
of the rod. With an infinitesimally small rod, the coastline becomes infinitely long. The problem is that we
are on the transition between the single (line) and double (surface) dimension, between the derivative and
the integral function. The area of England does not increase to infinity when the fractal of its borderline
increases in length. A fractal is defined as a figure with a broken dimension between 1and 2 (e.g. 9/7,
Gleick 1991).
The fractal nature of coastlines, streamlines and elevation contour lines causes difficulties when drawing
them on a map. Details below drawing precision "fall off and smoothing of lines often improves the
readability of the map. Rivers have the peculiarity of branching quite often. To which branch order do we
map rivers and creeks on a hydrographic map? In the present book I give more detail when the scale
zooms in. Contour lines in Figure 1have been deliberately smoothed in order not to draw attention to the
effect of each river valley, but depict the rise of the landscape from the coast into the interior.
Other scale-detail relations can be seen on isohyet maps over a small creek basin (Audrenisrou, 3775 ha,
Figure 30) near Tai' where Casenave et al. (1984) found a range of 1500 to 1870 mm annual rainfall for
the year 1981, which corresponds to the range of that year (1500 to 1900 mm) over the entire National
Park (400 000 ha) (ANAM 1987).
9
10
Setting the scene
1.1.2 The forest-savanna boundary and impacts offire
The forest-savanna boundary isa striking discontinuity in thevegetation cover (see
Figure 7). As shown by Swaine et al. (1976) and Gautier (1989) theboundary can easily
be seen on satellite imagery. The annual impact of fire isoften held responsible for the
sharpness of theboundary (van Donselaar 1965,Devineau 1976; Figure 2). Less frequent
fires would enableclosed dry forest and dense woodland todevelop and changes would be
moregradual (Spichiger &Lasailly 1981). It comes as a shock to a forester or a
conservationist that theentire savanna between Senegal in the west and Chad lake in the
east isburnt annually, even National Parks as big as one million ha like theComoe
National Park in north-east Cote dTvoire. This frequent fire regime is likely to reduce
biological and structural diversity, biomass reserves and soil fertility (Devineau et al.
1984). But a fire ban would drastically change the ecosystem (de Bie 1991), soapolicy
of what Connell (1978) called "intermediate disturbance level" and thus at present a
reduced fire frequency is advisable in thecase of active nature management.
Figure 2
Each year the savanna burns in West Africa. The fire often penetrates forest reserves and may
cause severe damage (Bertault 1992, Hawthorne 1993).
Lithology
Figure 3
A close-up of weathered sericite-chlorite schist in a gully on the Para study site
1.2 Lithology of SE Liberia and SW Cote d'lvoire
But there is morethan fire and gradually decreasing rainfall (see Chapter 2) to determine
he sharp boundary between forest and savanna. North of Abidjan, it coincides with a
hthologicalboundary (Mangenot 1955, Rougerie 1960, Guillaumet &Adjanohoun 1971)
between schist and granite, the former giving more clayey soils which better retain water
during the dry season (Leneuf 1959). Five major rock types underlie the forests of SE
Liberia and SW Cote d'lvoire (Figure 4): sericite-chlorite schists (Figure 3) mica schists
f ^ f l f d m i S m a t l t e s ' two-mica granites and granites and granodiorites (Papon 1973 '
MPEA Mimstry of Planning and Economic Affairs 1983). In this order, from low tohigh
degree of metamorphism, they generally result in soils with a decreasing soil water
retention capacity. Locally the soil formation processes and lateral and vertical transport
ot soil material may override this general rule (see Chapter 3).
Along the coast near Abidjan sandy Tertiary depositsbore aparticular forest typenow
almost disappeared (Guillaumet &Adjanohoun 1971,de Koning 1983) TheVolta
sandstones in Ghana (Hall &Swaine 1976) may resemble granites in terms of moisture
retention The Nimba mountains owe their higher altitude to more resistant rock types
which include oneof the richest iron ores in the world (MPEA 1983). Figure 4 shows'
12
Setting the scene
1
LEGEND
1
9°
granites and
granodiorites
two-mica
granites
§§§§§§1 s s
sericite-chlorite
schists
Figure 4
gn
gneisses and
migmatites
-_-^--_-_]ms
mica schists
water
fault or shear zone
other rock
types
Lithology ofSE Liberia and SWCote d'lvoire. Main rock types aregneisses, migmatites, twomica granites, granodiorites, mica schists andsericite-chlorite schists. Themajor faults trend SWNE. This lithological map isbased onthemost recent geological maps ofthe region. InLiberia a
geological survey wascarried outbytheUSGeological Survey (USGS) andtheLiberian
Geological Survey intheyears 1965-1972 (Brock andChidester 1977, Force and Beikman 1977,
Tysdal 1977 andothers). The final map waspublished byMPEA (1983) inthePlanningand
Development Atlas ofLiberia. InSWCote d'lvoire a detailed geological survey wasdone by the
Bureau deRecherches Geologiques etMinieres (BRGM, Paris) andtheSociete pour le Developpement Minier (SODEMI, Abidjan). Maps ata scale of 1:50000were prepared byBos(1964),
Jeambrun (1965; 1966), Letalenet (1965a,b) andothers. The final report byPapon (1973) included
a colour map at 1:500000.
Lithology
that most of the Liberian forest are underlain by gneisses, belonging to the Precambrian
basement complex. Someof these gneisses are called migmatites (from the Greek idyfia:
mixture) which isa metamorphic rock with granitic intrusions (Schumann 1979).The
geological mapping of thegneisses doneby the most recent geological surveys in Liberia
(MPEA 1983)and Cote d'lvoire (Papon 1973)is not consistent on both sides of Cavalla
river. Furthermore, deRouw et al. (1990) made nodistinction between gneiss and
migmatiteto differentiate land units in their Land Unit Survey of theTai region. SoI
grouped both rock types into one unit on Figure4.
In the eastern half of Coted'lvoire (Bagarre etTagini 1965) and in most of Ghana (Hall
& Swaine 1981) schist is the dominant bedrock type, soa more northerly extension of
wetter forest may beexpected. Bedrock and rainfall gradient roughly coincide in Ghana
(wetter climate on wettest bedrock, drier climateon drier bedrock), so that Hall &
Swaine (1981) had a sufficient explanation for thevegetation gradient when considering
climate alone. In Liberia bedrock never contradicts climateeither, but in Cote d'lvoire I
confirm the findings of Guillaumet (1967) and deRouw (1991) that a drier climate can be
compensated for by a "wetter" bedrock (see Chapter 2).
1.3 Three man-made forest blocks
1.3.1 Actual forest cover
Only about 8 million ha of West African rain forest remained in the mid-eighties (Martin
1989, Sayer et al. 1992, Parren &de Graaf 1993in prep.). This is some 20 %of the
precolonial area. Enormous quantities of biomass (400 t ha"1 •32 106ha = 13 109tonnes
dry matter) have been burned and their combustion gases emitted into the atmosphere.
Precise estimates of theactual forest cover only exist for Ghana (Ghartey 1989), where
almost no forest remains outside the reserves but where the reserves are well protected
and thus still under dense forest cover. In Cote d'lvoire and Liberia both the forest area
outside reserves and the state of conservation of the reserves are insufficiently documented (Vooren 1992b, Mayers 1992).
These forest relics in West Africa are divided over three major forest blocks, each
centred on theborder between twocountries (Sayer et al. 1992):
1. east Sierra Leone &west Liberia: Gola and Kpelle National Forests (MPEA 1983,
Hammermaster 1985)
2. east Liberia &south-west Cote d'lvoire: Krahn-Bassa, Gbee, Gio and Grebo National
Forests and Sapo and Tai National Park with surrounding forest reserves (see
Figure 13)
3. east Cote d'lvoire and south-west Ghana: Songan-Mabi forest reserve and Bia
National Park, Nini-Suhien National Park and surrounding reserves (Hall &Swaine
1981)
My study focused on the second block, which is the largest one remaining (Sayer etal.
1992)and which contains the longest forest gradient (see Chapter 2). It isan ideal study
area for *spatial gradient analysis.
13
14
Setting the scene
1.3.2 Timber mining
Another difference between Ghana and its neighbouring countries is that the timber stocks
in 1988in the Ghanaian reserves were, on the average, 90 m3ha 1 (above70cm
diameter, all species), and that more than 50 %were traditionally commercial species
(Ghartey 1989;Figure 5). Almost all Ivorian reserves must be considered as depleted of
commercial stock (e.g. Yapo forest with 9 m3ha"1, Mengin-Lecreulx 1990), but no recent
national inventory exists. In Liberia logging was much less intensein thepast, but the
natural stocks of commercial trees in these wet evergreen forests were smaller (Sachtler
1968).
This book only presents data from unlogged forests, either from inventories before the
logging era, or from thelast 200000ha of unlogged forest in Coted'lvoire (Bousquet
1978)which are conserved within Tai National Park (seeText box 3, p. 66). I worked on
a kind of "palaeotypology" of the forests. On the one hand, I considered how the
composition was, not how it will, or should, or can become after effective conservation.
But on the other hand, this study of theprimeval forests of West Africa allows the initial
conditions tobe assessed, before theremaining forest has disappeared.
Figure 5
A Meliaceae trunk is transported to the sawmill in Ghana. The Ghanaian forest service has spared
many of such large trees. In Cote d'lvoire they have all been logged.
Forest history
1.3.3 Forest management and conservation
In the land use plans of the three countries concerned, a number of forests havebeen
designated as reserves, to remain closed forest. Eight of them have received the status of
National Park (Sapo, Tai, Mt Pelco, Maraoue, Azagny, Banco, Biaand Nini-Suhien;
Martin 1989, Sayer et al. 1992). Stock management and natural regeneration silviculture
haveonly occurred on a large scale in Ghana (Parren 1991). Extensive research plots
were laid out by CTFT in Coted'lvoire (Bertault 1986, Mattre 1991), but theresults
have only been applied in a pilotproject in Yapo forest. Recent projects with British
(ODA), French (CTFT) or German (GTZ) cooperation may result in sustainably managed
forests in the future in Ghana, west and east Coted'lvoire respectively. Certain forests,
like the Bosse-Matie"forest in east Cote d'lvoire, need rehabilitation management, sono
timber will become available for the market in the first few decades. The civil war in
Liberia since 1990 has stopped almost all economic activity in that country, and future
management and conservation remain an open question.
However, a forest ecosystem is more than living timber. Wildlife protection isa problem
in all three countries. Even in the large Tai National Park (446000 ha, see Text box 3,
p. 66) the survival of the forest elephant (Loxodonta africana cyclotis Matschie) isnot
guaranteed and hunting impact remains high (Figure 6). Atropical rain forest without
animals is doomed (Alexandre 1978,Jacobs 1988)and forest managers should not content
themselves with saving thetrees.
Figure 6
Remains of a forest elephant found in the Para study plot
15
17
2
THE REGIONAL FOREST GRADIENT IN SE LIBERIA AND
SW COTE D'lVOIRE
In this Chapter, I will analyse the regional forest gradient using ordination and spatial
interpolation techniques. As indicated in Chapter 1, I will concentrate on the middle
forest block in West Africa in itspre-logging state (Figure 13), roughly between Cestos
river (Liberia) in the west and Sassandra river (Cote d'lvoire) in the east and North upto
the forest-savanna boundary. Thebase for interpreting this gradient is the lithological
map, (Chapter 1, Figure 4). However, first the rainfall gradient will be examined.
2.1 The SW-oriented rainfall gradient
These forests of the middle forest block lie on a climatic gradient which is described by
Eldin (1971), DRC (1967a) and ASECNA (1979) for Cote d'lvoire and by Voorhoeve
(1965)and Sachtler (1968) for Liberia. The major components of this gradient are (see
Griffiths 1972,Hayward and Oguntoyinbo 1987):
• thediminishing total amounts of rainfall from the Liberian coast into the interior:
in the study region a NE-SW trend. Important factors here are the permanently
warm waters off theLiberian coast and the upwelling of cold waters from July to
October between Cape Palmas (Tabou) and Lagos;
• the change in the coefficient of variation of annual rainfall, which is high (20to25
%)along the coast, lower (15to 20 %)in the forest zoneand again increasing to
20 %and further even 50 %in the savanna and Sahel zones. This trend is related
to distance inland;
• the change in seasonal distribution from bimodal along thecoast to unimodal in the
interior: a N-Strend resulting from thepath of the zenithal sun;
• greater temperature fluctuations towards the interior, and higher diurnal temperatures towards the North, resulting in more evapotranspiration.
As rainfall totals are best documented for the region and show pronounced differences, I
will concentrate on this component of the climatic gradient. Figure 7 shows the major
rainfall stations in the region and thebest estimate of their long-term average rainfall in
dm y"1or m(10 y)"1 (see Table 1for the measuring period and sources of the data).When
interpreting these rainfall data, spatial and temporal sources of variation need to beconsidered.
18
Forests of SE Liberia and SW Cote d'Ivoire
Tal(123m)
M-C
i870mm
Daloa(281m)
26«o 1420mm Yamoussoukro 26'C
(208 m)
i070n
4°W.L
Greenville (5 m) 2«'C 4420mm
Sassandra (62 m) 26-c
1730mm
rTOOmm
Tabou (20 m) 26*c
Figure 7
2370mm
Rainfall map of SE Liberia and SW Cote d'lvoire. Average precipitation figures per decade in
m (10 y)"1 on the map (sources see Table 1). Shaded line: forest boundaries with savanna (North)
and Ocean (South), dark grey tone mountainous area (elevation above 500 m). The cross-section
A-A' is given in Figure 10. Climate diagrams, originally called ombrothermogramsby Gaussen
(1954), were constructed following the procedure described in Walter (1979). The right-hand axis,
indicating average monthly rainfall, is graded every 20 mm below 100 mm, and every 200 mm
above 100 mm.
19
Tlierainfall gradient
Table 1 Rainfall stations in SE Liberia and SW Cote d'lvoire. Elevation above sea level given for all
stations, distance from the coast for the Liberian ones. Rainfall is expressed as mean rainfall per
decade in m (10 y)"'. This latter dimension is considered more accurate, as the data are based on
10 years of measuring in Liberia and 30 years in Cote d'lvoire. The common practice of
expressing annual rainfall in mm y"' has not been followed, because of the poor precision and the
great variability of the averaged precipitation amounts and because of their limited spatial extrapolation value (see also Text box 1, p.9).
Rainfall
stations
distance
to
Liberian coast
(km)
elevation
above
sea level
(m)
mean
rainfall
per decade
(m (10 y)-')
measuring
period
remark
5
20
100
230
50
160
215
44.2
32.8
28.9
25.5
24.9
20.6
19.3
1951-'66
1960-'66
1965-'66
1952-'66
1928-'81
1952-'61
1951-73
coastal
coastal
3
3
3
2
1
2
1
20
78
270
123
5
62
217
245
134
205
200
281
273
207
351
187
260
208
23.7
21.1
19.5
18.7
18.0
17.3
16.4
16.3
15.9
14.8
14.3
14.2
13.9
13.5
13.4
13.2
12.9
10.7
1951-'80
1951-'80
1951-'80
1951-'80
1967-'80
1951-'80
1951-'80
1956-'80
1951-'80
1951-'80
1951-'80
1951-'80
1966-'80
1951-'80
1951-'80
1951-'80
1953-'80
1964-'80
coastal
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
Liberia
Greenville
Flahntown (AFC)
Juarzon
Pinetown
Firestone Cavalla
Ziatown
Zwedru
0
7
50
110
30
144
160
Cdte d'lvoire
Tabou
Grabo
Toulepleu
Tai'
Fresco
Sassandra
Guiglo
Duekoue
Soubre
Gagnoa
Lakota
Daloa
Sinfra
Oume
Seguela
Bouafle
Vavoua
Yamoussoukro
coastal
coastal
Sources:
1: Meijers and Saye (1983)
2: GFML (1967)
3: Sachtler and Hamer (1967)
4: ANAM (1987); for stations not covering 30 years, the mean has been constructed from neighbouring
stations with complete series of data (see text)
20
Forests ofSE Liberia and SW Cote d'lvoire
2.1.1 Sources of spatial rainfall variation:about interpolation
The standard procedure in climatology for processing rainfall data on a map isby
interpolating *isohyets using all data points. This descriptive procedure isbased on the
assumption that the variation is random and that no factors can bedemonstrated to cause
trends (Stein &Corsten 1991). Atechnique like cokriging (Krajewski 1987, Stein &
Corsten 1991) uses observations of several co-variables in predicting e.g. rainfall.
Gregory (1965,in Griffiths 1972)explained 75 %of thevariation in annual rainfall in
Sierra Leone by four factors: distance from the coast, longitude, latitude and elevation. A
climatic predictive package BIOCLIM was developed for Australia by Busby (1986,ex
Russell-Smith 1991). It isbeyond the scope of this study to develop a similar model for
West Africa but I will speculate on somepossible sources of spatialvariation.
Themountain range from west Cote d'lvoire towards Guinea has a clear impacton
rainfall amounts, distribution and variability (Eldin 1971).Records from rainfall stations
within this range (e.g. city of Man) were omitted on Figure 7. Thefirst50 km from the
coast inland also have high rainfall amounts and, importantly, a greater variability. This
is caused by extremely intense heavy downpours (Sachtler 1968,Brunet-Moret 1976).
Thesepeak volumes contribute littleto the soil moisture content. Becauseof thistypical
coastal effect,caution should be exercised when using data from gauges along the coast to
interpolate isohyets; hence the shading in Figure 7. A third factor, somewhat related to
the first, is relief. The position in the landscape of the rain gauge is a source of known
variation which should betaken intoaccount when interpolating. For instance, gauges
placed in thevalley near the river, as in Ta'i, Soubreand Guiglo, could possibly receive
smaller quantities than gauges near the water divide and thus higher in the landscape like
Pinetown and Ziatown. The maritime air masses mightproduce less rain when descending
from the watershed than when ascending to it (ASECNA 1979).
The limited spatial interpolation value of the rainfall averages was illustrated by Casenave
et al. (1980, 1981, 1984) in the Audrenisrou basin in Ta'iNational Park. Fourteen rain
gauges placed in a small river basin of 38km2 were recorded from 1979 to 1981.Within
a few km, differences in rainfall of more than 0.3 mwere found for the same year, but
without consistent correlation with elevation. Clearly, more studies on small-scale rainfall
variability are needed to obtain reliable interpolation estimates.
2.1.2 Temporal variation of rainfall: taking the mean?
Cyclic trends. Apart from obvious seasonal trends like wet and dry seasons, cyclic trends
havebeen found in Cote d'lvoire (Lhomme 1981;Snoeck 1975) and Nigeria (Adejuwon
et al. 1990). One such cycle showed a period of 2 to 3years, corresponding to theQuasiBiannual Oscillation (QBO) known from meteorology (Lhomme 1981),others displayed
periods of 11and 30 years corresponding to the single and triple sun spot cycle. Asan
example the rainfall series of Daloa (central Coted'lvoire) is given in Figure 8. The
decades 1925-1935 and 1955-1965 contained the maxima of the oscillation and rainfall
was high. The minima were in 1940-1950and in 1975-1985. If thisoscillation, which
corresponds to three times the sun spot cycleof 11years, is really significant, another
wet period may beexpected in the period 1990-2000. Hayward and Oguntoyinbo (1987)
predict a dry period starting about 2003, based on information from Nigeria. An
21
The rainfall gradient
Rainfall in Daloa
cyclic trend (period=32 yr)
annual rainfall (mm)
"false" linear trend -3 mm/yr
•5 yr running mean
2000
1700
1400-
1100
800
1910
Figure 8
— I
1920
1
T
l
1930
1940
1950
T~
1960
year
1970
1980
1990
2000
Temporal variation in annual rainfall, illustrated by the rainfall time series from Daloa (altitude
281 m asl), central Cote d'lvoire (source of data: ORSTOM-CIEH 1973;Laboratoire d'Hydrologie, ORSTOM, Montpellier; Agence Nationale des Aerodromes et de la Meteorologie (ANAM),
Abidjan-Port Bouet). The five-year running mean is given by the bold line. The main cyclic trend
as calculated with a periodogram (using Statgraphics, 1990 STSC), has a period of 32 years and
is depicted by the dotted sine curve. The "false" negative linear trend can be entirely explained by
the cyclic movement. Similar trends have been found for Bouna and Abengourou (Lhomme 1981)
and in Nigeria (Adejuwon et al. 1990).
implication of this cyclic trend is that meanscalculated over different parts over the cycle
are not directly comparable. Most of the rainfall data from Liberia were collected
between 1950and 1960 (Table 1), in a wetter part of the cycle with the five-year running
average being 10to 20 %higher than the mean over the wholecycle.
Linear trends. Adejuwon et al. (1990) reported significant negative trends (-3 to
-8 mmy"1)for annual rainfall at six rainfall stations in Nigeria between 1922 and 1985.
They analysed the time series of 16stations spread over thecountry. Only one station had
a significant positive trend (+5 mm y"').Alinear regression applied to theabovementioned cyclic trend between 1922 and 1985automatically resulted in the "false"
negative trend shown in Figure 8, only because the sine curve starts with high values and
ends with low ones. An appropriate regression model should contain both cyclicand
linear terms (van Montfoort, pers. comm.). The timeperiod considered, is alsoimportant. The curve of the five-year running mean in Figure 8 steeply declines between 1955
and 1982, as can be expected between the maximum and the next minimum of a sine
curve. Given the evidence for a cyclic trend, theextrapolation of this negative trend
beyond 1982 may be far from the truth.
22
Forests ofSE Liberia and SW Cote d'lvoire
If such trends also exist in Liberia and Cote d'lvoire, the mean should lie on thebase line
of the sinecurve. If this middlelineis alsodeclining, nooverall mean can be calculated,
i.e. the mean is time-dependent. The data for the Ivorian stations (ANAM 1987) as given
in Table 1and displayed in Figure 7 havebeen corrected by ANAM in a way which
eliminates at least someproblems of comparability. The mean for each station was
calibrated for theperiod 1951-1980. If only a partial mean could be calculated, then this
mean was corrected using thedata from a nearby station with a complete record:
-
Ri
—
& =— •R
in which:
Rt = adjusted mean for station with incomplete record
Rij. = partial mean over period with available data for station with incomplete record
Rc = mean annual rainfall 1951-1980of station with complete record
It might be advisable to use several neighbouring stations for the calibration and to
analyse the correlation of the stationsfirst.
2.1.3 Description of the rainfall gradient: a cross-section
The forest-savanna boundary represented on Figure 7corresponds to 11to 13m (10y)"1
rainfall and runs parallel to the Liberian coastline. I therefore drew a cross-section
through theregion of interest (Figure 10)from Greenville on theLiberian coast, which is
the only town in West Africa without a dry season, to Vavoua in central-west Cote
d'lvoire, which liesjust in the savanna. Relief is given at thebottom as an independent
and relatively constant factor. Mean 10-year rainfall in m(10y)"1is given in the middle
and its first derivative, the rainfall gradient in m(10 y)"1 (100 km)"1on top. The coastal
effect is clearly visible on the very steep left-hand part of the rainfall curve. Both rainfall
amount and rainfall gradient decrease gradually from the coast towards the savanna,
which suggests an exponential path of thecurve. Note that the rainfall decrease isnot
constant in space, but that rainfall zones become wider. Aresponse of the same kind will
be demonstrated in the vegetation gradient later in this Chapter. At some 150km from
the coast a striking discontinuity is found. It is apparently caused by the higher plateau
around Putu Range (see Figure 1, Figure 9) and its rain shadow. Up to Putu Range the
elevation rises steadily and further NE the relief becomes undulating and dissected by
rivers flowing transversely to thegradient. The forests up to Putu Range directly face the
SW monsoons, and Grebo National Forest and Ta'iNational Park undoubtedly experience
a rain shadow effect. The steep decrease in rainfall between Pinetown and Zwedru
illustrates this rain shadow.
The regionalforest gradient
Figure 9
The everwet forests of Sapo National Park, looking towards Putu range (Liberia; Photo by H
Dop)
2.2 The regional forest gradient
For the analysis of the regional forest gradient I ordinated the tree species composition of
several forests in SE Liberia and SW Coted'lvoire. Inventory data from thepre-logging
era were analysed together with data from my three sample plots in Tai National Park.
First I will explain why I chose the large tree species and why I considered the forest asa
continuum. Both have consequences for the methods of sampling and data processing.
2.2.1 Large tree species
In this research I focused on large tree species that attain a diameter of 70 cm and more
(Figure 12). Out of the total forest flora of at least 2000 species, i.e. the estimate for
vascular plants in Ghana published by Hall &Swaine (1981), I used 53of the largest tree
species to describe thevegetation gradient from the coast to where the savanna begins.
This subsample of the flora isof special interest to foresters and also forms the main
structural element of the forest ecosystem (Albers 1990). The huge size of these trees
when mature, and their often low population density demand an adapted sampling
approach.
Former vegetation studies in SW Coted'lvoire used "complete" plant lists (Guillaumet
1967, de Rouw 1991). In Ghana, Hall &Swaine (1976) recorded 1248vascular plant
species and used 750 of these to describe the forest gradient there. In these data sets the
23
24
Forests ofSE Liberia and SW Cote d'lvoire
decreaseinmeanrainfallperdecadeinm(10y)' (10km)"1
-2 -i
-16.3 a
-1.5-1 -0.5"i
0
1
1
1—
1
100
200
meanrainfallperdecadeinm(10y)
300
400
300
400
-1
4030
201<H
—I
1
100
1
r-
200
elevation(masl)
Puturange
30O
200-
100-
f0
Juarzon 100
200
Flahntown
Pinetown Zwedru
Greenville
300
Guiglo
400
Vavoua
distancetotheLiberiancoast(km)
Figure 10
The rainfall gradient along cross-section AA' (see Figure 7) through SE Liberia and SW Cote
d'lvoire. The average elevation along the cross-section is given at the bottom. Source of rainfall
data: see Table 1; source of elevation data: Sachtler (1968), IFAN 1968, MPEA 1983).
25
The regionalforest gradient
decreaseinDCA1units(10 km)"
-20-15
-10-1
-5
0
100
200
300
400
200
300
400
spatiallyinterpolatedDCA1score
250200
150
100500-
100
elevation(m asl)
Puturange
300-
Labor.
200Sassandrar.
100-
B*
Bo
?0
Juarzon 100
200
'Flahntown
Pinetown Zwedru
Greenville
300
Guiglo
400
Vavoua
distancetotheLiberiancoast(km)
Figure 11
Cross-section BB' through relief, spatially interpolated DCA ordination score (Figure 14) and
regional forest gradient in SE Liberia and SW Cote d'lvoire
26
Forests of SE Liberia and SW Cote d'Ivoire
large trees species are often underrepresented and there are few distribution maps for
Ghana of many of these species in the atlasby Hall &Swaine (1981) due to shortage of
data. Their 155 sampleplots (of 1/16 ha each) covered 10ha in total (Hall &Swaine
1976). Hence, their conclusions werebased on a 0.0005 %inventory, intended tobe
extrapolated over a total forest cover of about 2 million ha in Ghana. Thepower of their
sampling method lay in regular distribution of the samples over southern Ghana and asa
result, they sampled 3/4 of the total forest flora. Thus, the sampling intensity was
sufficient to obtain a representative picture of the species distribution. For large trees, the
scale of the inventory has to be adapted. In Chapter 3I will show that in three areasof
20 ha each I found the major part of the total large tree flora of Tai National Park
(446000ha), at a sampling intensity of 0.01 %. If in addition to the occurrence of a
species, its abundance has tobe estimated, sampling needs to be still moreintense.
Loetsch et al. (1973) advised a sampling intensity of 1 %to obtain reliable estimates of
Figure 12
A Gymnostemon zaizou tree (Simaroubaceae) at the Para study site. This large tree species is
endemic to SW Cote d'lvoire and Liberia. Photo by P. Albers.
The regionalforest gradient
timber volume, which also indicate the abundance of large tree species in forest compartmentscovering thousands of ha.
2.2.2 Forest inventories from the pre-logging era: aspatial gradient analysis
Thebotanical exploration of SELiberia and of SW Cote d'lvoire started in the 19th
century. Aextensive overview isgiven by Voorhoeve (1965) for Liberia and deRouw
(1991) for Coted'lvoire. The vegetation gradient was recognized by early botanist
explorers (Chevalier 1909, Aubreville 1959). Vegetation scientists mapped the forest
continuum by dividing it into forest types (Guillaumet &Adjanohoun 1971;Hall &
Swaine 1976)or vegetation zones (Sachtler 1968,de Rouw 1991). Much valuable
material was collected during timber inventories which were carried out with assistance
from GTZ (German development aid) in the sixties in Liberia (Sachtler 1968) and with
assistance from CTFT (the former Centre Technique Forestier Tropical, now CIRADForet) in the early seventies in Cote d'lvoire (SODEFOR 1976a,b).
The continuous variability of the forest was illustrated by the ordination scatter diagrams
on which someof these classifications are based and on which clustering of samples was
seldom found. The approach of Hall &Swaine (1976) was to search for dividing linesin
the ordination scatter diagram separating geographically contiguous zones on themap.
They did this by trial and error. Afurther spatial analysis of this regional gradient is
suggested in the present book using spatial interpolation techniques.
On the map, Hall &Swaine (1981) isolated a lower montane forest type, which they
called 'upland evergreen forest', on hill ranges at 500 to 700 melevation. The slopeof
these bauxite-capped hills was very steep, so there was no occasion to study the altitudinal
gradient, as was doneby Schnell (1952) and Adam (1983) for Nimba range at thepoint
where Coted'lvoire, Liberia and Guinea touch. These 'upland evergreen forests' arean
example of a cluster of samples that can be split off.
For lowland forests, the ordination scatter diagrams of Hall &Swaine 1976and deRouw
1991 were divided by vertical lines, based on thefirstordination axis, which was indeed
thought to explain the major part of thevariation because it had the highest eigenvalueof
all axes. The regional forest gradient apparently correlated well with thefirstordination
axis. The other axes remained largely unexplained. Hall &Swaine (1976) showed that the
6th axis of their RA ordination isolated theabove-mentioned lower montane forests. So,
gradients related to altitude may find their expression in higher axes. De Rouw (1991)
ordinated "primary" and "secondary" forests together and found that the main floristic
gradient, i.e. thefirstaxis, was related to the degree of, or to the time since disturbance.
Most of the forests I ordinated were unlogged, but disturbance by farming in certain parts
and by large elephant populations was reported, e.g. in Krahn-Bassa forest (Sachtler
1968). These effects might show up in higher ordination axes.
Vegetational changes are gradual, especially when one is studying the large tree species
with an ecological amplitude that allows them to cover a considerable part of the gradient.
Even with a low abundance, such a species may remain present. It is rare to find large
tree species that are unique to an area. Smaller plants more easily attain the status of
uniqueness (Mangenot 1955,Guillaumet 1967,deRouw 1991).This ispartially caused
27
28
Forests of SE Liberia and SW Cote d'lvoire
by the size of theplants, which may be said to havea different "*eigenscale". Divisive
*clustering methods such as Two Way INdicator SPecies ANalysis (TWINSPAN, Hill
1979b) rely greatly on such indicator species. I tried out these methods on my data set
and found that some divisions were placed at the edgeof the range of distribution of rare
species. For analysing a data set consisting of large tree species, DEtrended CORrespondence ANAlysis (DECORANA or DCA, Hill 1979a; Hill &Gauch 1980)provided better
results. This ordination technique ranks the species according to the modeof their
distribution, i.e. where each species attains its greatest abundance (ter Braak 1986).
The large number of "absence values", typical of species-sample matrices, havean
important weight in theanalysis. Samples with a small quadrat size, e.g. 625 m2 (in Hall
& Swaine 1976)and 36 m2plus aplotless sample of trees in the surroundings (inde
Rouw 1991), never contain the complete list of the large tree species present in that
forest, because these tree species have an irregular regeneration in timeand rarely have
seedlings and saplings permanently in the forest undergrowth. When surveying large plots
for trees abovea certain diameter limit, all smaller plants are disregarded and thus the
species list is not complete, but samples in a small quadrat in which allplants are
recorded are also incomplete, because they lack many irregularly regenerating large-sized
species. In fact, this has to do with the question of the minimum area for a vegetation
sample in tropical rain forests (Hommel 1991). On a 625 m2quadrat, Hall &Swaine
(1976) only found 40 %of the vascular plants species present on 1ha. This means that
the other 60 %are marked "absent" in thedata file but are present in the field, although
maybeonly at a density of less than 1individual per 625 m2or with a clustered distribution. For large tree species the percentage of "false absents" may beeven larger.
TWINSPAN, however, relied both on the "presents" and on the "absents" in my data set,
so it wasprobable that Type II errors were made, in the form of artificial distinctions
that did not correspond to forest reality. DCA seemed to be more robust to violation of
theprinciple that all species in thevegetation must be listed (Hill &Gauch 1980, ter
Braak 1986).
Besides its sensitivity to false absents, TWINSPAN splits a gradient even when it is
completely linear. Atheoretical linear spatial gradient on which changeis constant, can
be split into two, three or five equal parts, even if no information isadded. Theonly
information held in the data is that the line is straight and that it covers a certain range
from starting to ending point. In many cases, however, the line is a curve and the
steepness of the gradient changes along the range. Sometimes such curves show a pointof
inflection (Oldeman 1990b).
During thepre-logging forest inventories numerous individual sampleplots or lines were
laid out systematically in each inventory compartment to sample species composition and
timber volume. In this Chapter, I willconsider thecomposition of the sum of all these
sampleplots as representative of the species composition of the entire inventory compartment. However, neither such average compositions nor any singleplot exemplifies how
tree species grow together in these forests. The species composition was obtained from a
mixture of several kinds of forests, valid for analyses at the level of means of species
populations whose architecture is unknown because it has not been surveyed.
The regionalforest gradient
2.3 Study sites and methods
2.3.1 Study sites
The thirty *forest inventory compartments in SELiberia and SW Cote d'lvoirewhose
tree species composition was retrieved from inventory reports (Sachtler 1968, SODEFOR
1976a,b), are shown in Figure 13.Each inventory compartment covered quite a large area
(on average 50 000ha, seeTable 2). They werepart of *forest complexes, such as
National Forests and National Parks. I ordinated myown threepermanent sampleplotsof
about 20ha each within TaiNational Park (see Chapter 3) together with the 30 inventory
compartments using their large (d>70 cm) tree composition.
All the forests are situated in thelowlands, below 400 melevation. Thelithology is
mainly gneiss and migmatite, with small granitic areas in the south-western part of KrahnBassa forest (compartments K167, K69), in Tai National Park and in the lower valley of
Sassandra river (Figure4). Aband of mica schist stretches from Zwedru to Guiglo
(3NW) and a wedgeof sericite-chlorite schist from Para north-eastward (1XV).
Three major rivers drain the area shown in Figure 13:Cestos, Cavalla and Sassandra
rivers. The latter two have their valleys moreor less perpendicular to the SW monsoon,
resulting in the rolling landscape depicted in Figure 10and Figure 11. Each successive
ascent theair masses are forced to make might cause them to drop rain on the windward
side, and rain shadows can be expected on thelee side. Cestos river runs NE-SW along a
major geological fault and the maritime air masses can freely attain its upper stream
basin.
2.3.2 Data collectionmethods
The sampling intensity in the inventories by Sachtler (1968) and SODEFOR (1976a,b)
varied from 0.1%to 5% of the forest area (Table2). Combined with thevarious sizesof
the compartments, this resulted in a fully sampled area of 30 to 2600 haper compartment. In comparison to these, thepermanent sample plotsin TaiNational Park, each
covering about 20 ha, were much smaller in area and moreover, they were contiguous
and not a sum of small systematic sampleplots. The inventory campaigns did not usethe
same lower diameter limit nor the same commercial specieslist. In somecases theinventory was stratified between swamp forest and upland forest. In other cases separate results
were given for closed forest and degraded forest.
I combined data from four different inventories which each used a different method:
1) Cote d'lvoire National Forest Inventory. The results of this inventory campaign were
published in three volumes, covering the north-west, centre-south and centre-east region
respectively. I used all 7 inventory compartments from the report on the North-West
region(SODEFOR 1976b) for the ordination. I selected 2 of the 7 compartments from the
report on the Centre-South region (SODEFOR 1976a).
29
30
8°
N.B.
Forests ofSE Liberia and SW Cote d'lvoire
6UINEA
6°-
6°W.L.
LEGEND
£ 2 2 Nationa'
p rk
°
K61 to K167 Krahn-Bassa National Forest, Liberia
S91 to S93
Sapo National Forest, Liberia
Q51 to G53
Grebo National Forest, Liberia
1NW to 7NW North-West region. Cote d'lvoire
Figure 13
6CSto 7CS
Centre-South region. Cote d'lvoire
1XV to 5XV
Perimetre industriel, XV Cote d'lvoire
Forest inventory compartments in SE Liberia and SW Cote d'lvoire used for the ordination
described in Chapter 2. Note the different inventory methods as evidenced by the codes given to
the compartments. For sources see Table 2. Different grey tones are only used to differentiate the
compartments on the map.
31
The regionalforest gradient
Table 2 The forest inventory compartments used for the ordination described in Chapter 2 of the present book
total area of
the forest
(ha)
sampling
intensity
(%)
1) Cdte d'lvoire National Forest Inventory
North-West region (SODEFOR 1976b)
1NW F.C1. du Scio
2NW Blolekin-Toulepleu
3NW F.C1. du Goin
4NW F.C1. du Haut Cavally
5NW Reserve de faune du N'zo
6NW F.C1. de Duekoue
7NW between Mt Peko and Sass. river
215000
n.a.
145000
but
171000 probably
135000
0.5
140000
to
230000
1%
91000
Centre-South region (SODEFOR 1976a)
6CS F.C1.de Niegre (northern part)
7CS F.C1. de Niegre (southern part)
118000
94000
2) C6te d'lvoire Inventory "PSrimetre industriel XV"
1XV east of Tai' Nat. Park (Clement 1973)
71000
88000
2XV "
19000
3XV
6000
4XV "
34000
5XV east of Sassandra river
area of % swamp forest
sample of total forest
plot(s)
% degraded
forest of total
forest
(ha)
1000
750
850
650
700
1150
450
n.a.
but
incl.
n.a.
n.a.
600
470
n.a. (incl.)
n.a.
n.a. (incl.)
n.a.
0.8 %
0.9 %
0.8 %
0.5 %
0.8 %
570
790
150
30
270
excluded
from
the sampled
forest
excluded
from
the sampled
forest
10 % (incl.)
23 %
12%
5%
11 %
10 %
25 %
3) Liberia National Forest Inventory
Grebo National Forest (250 000 ha) (GFML 1967)
G51
95000
G52
81000
G53
75000
0.5 %
0.5 %
0.5 %
475
405
375
8%
6%
6%
7%
6%
10 %
Krahn-Bassa National Forest (514 000 ha)
(Sachtler & Hamer 1967)
K61
K62
K63
K64
K65
K66
K67
K68
K69
K167
4%
5%
5%
5%
5%
<0.3 %
<0.3 %
<0.3 %
<0.3 %
<0.3 %
1840
2250
2250
2600
2400
80
80
110
90
50
14 %
15%
16 %
15%
27 %
29 %
25 %
26 %
33 %
32 %
1%
33 %
10%
64 %
27 %
16 %
0%
2%
2%
0%
Sapo National Park (150 000 ha) (Sachtler & Hamer 1967)
S91
55000
0.2 %
S92
40000
0.1 %
0.2 %
S93
26000
110
40
52
16 %
n.a.
20 %
19 %
n.a.
32 %
46000
45000
45000
52000
48000
27000
26000
37000
31000
19000
4) Cdte d'lvoire: my three sample plots in Tai'National Park (see Chapter 3)
Zagne
23
excluded
excluded
Tai
20
Para
22
n.a.: not available; incl.: included in the sampling; F.C1.: Foret classee (=forest reserve)
32
Forests ofSE Liberia and SW Cote d'lvoire
• The results were published for two strata:
- forest (closed and degraded): used for theordination
- trees outside forest: not used
• 49 timber species were inventoried above 20 cm diameter; a 'minus' sign in Table 3
means absence of the species. The inventory report used vernacular names. Asa
result, I could specify six of these timber species only to genus level. Other tree
species were inventoried above 60cm diameter, but only the 20 most abundant species
ineachinventory compartment were listed in thereports, as well as the total volume
for all species. Theconsequence of this way of reporting was that thepresence or
absence of the rarer species in a compartment could notbe retrieved.
• To quantify the abundance of a species I selected the tables listing timber volumeper
species (in m3above 70 cm diameter), including all quality classes. These data are
reproduced in Appendix 2. For use in thepresent ordination I rescaled the timber
volume densities to abundance scores from 1to 9 by calculating what percentage of
total volume in thecompartment consists of a given species. Thispercentage was
rounded to the nearest integer and if less than 1, it was given the code 1, if more than
9, it was given thecode 9.
2) C6te d'lvoire Inventory "Perimetre industriel XV". These forests, lying east ofTai"
National Park, were inventoried prior to theNational Forest Inventory (Clement
1973). I used all 5 inventory compartments mentioned in the report.
• Results were specified according to three strata:
- closed forest on well drained soils: used for the ordination
- swamp forest: not used
- degraded forest and shifting cultivation: not used
• 55 timber species (5 of which specified only to genus level) were inventoried above
20cm diameter. No information was given on other species or on the total volumeof
all species.
• Timber volume above 70cm diameter, all quality classes, was used to quantify a
species. Rescaling as for 1).
3) Liberia National Forest Inventory. In cooperation with GTZ, the National Forests in
Liberia were inventoried in the sixties (Sachtler 1968). I selected the three forests in
SELiberia for use in theordination. Grebo National Forest, west of Tai National
Park, was split into 3 inventory compartments (GFML 1967). Sapo National Forest,
in the meanwhile designated a National Park (Mayers 1992), also contained 3
compartments (Sachtler &Hamer 1967) and Krahn-Bassa National Forest, the largest
of all three, contained 10compartments (Sachtler &Hamer 1967).
• In the reports no stratification was madewithin the compartments. Swampsand
degraded forest were included in theresults.
• 63timber species were inventoried above40cm diameter. Smaller trees of those
species, from 12to 40cm diameter, were counted in a subsample. Other species were
grouped into a class "Miscellaneous" and the total volume was reported. 18timber
species could only be specified to genus level (e.g. Uapaca sp., Parinari sp., ...) (see
Appendix 2).
• Timber volumeper km"2above 40cm diameter, good quality, was used to quantify the
species. Rescaling as 1).
The regionalforest gradient
33
4) Cote d'lvoire:my three sample plots in TaiNational Park
• Swamp forest was excluded.
• All species were inventoried above 70cm diameter over 23,20and 22 ha in the
Zagn6, Taiand Para sampleplots. Theentire sampleplot near Tai (seeChapter 3)
covered 25 ha but thelowermost 5 ha were excluded.
• Stem number per 20 ha (all qualities) was used to quantify the species. These figures
were rounded to thenearest integer to be used as abundance score. If more than 9
trees were counted in the sampleplot, the abundance score '9' was given to the
species. These data were extracted from the survey discussed in Chapter 3. See there
for further information on these sites.
2.3.3 Methods for data processing
Ordination. The 33 samples were ordinated using the FORTRAN programme
DECORANA (Hill 1979a) for detrended correspondence analysis (DCA). Two runs were
done, one with theraw data set prepared as described above, a second retaining only the
species common to all four inventory methods. In the first run, DCA was found tobe
sensitive to the incompleteness of the data. Quite a lot of *false absents occur becauseof
thedifferent methods of the inventories. First,the releves were incomplete; they did not
contain allplant species in a given forest, butonly the tree species exceeding the
indicated diameter limit and of actual or potential commercial interest. Certain tree
species may be limited to young ""cohortsand hencepresent below thediameter limit
only. Second, the same species were not recorded consistently in all forests. This was
partly because certain Liberian species do not even occur in Cote d'lvoire and vice versa,
but also because in thefirstmethod the non-commercial species were mentioned for
compartments where they are abundant, but not for other compartments where they were
rare, but may havebeen present. Third, the size of the sampling areas varied greatly, i.e.
from 20 ha to more than 2000 ha. This certainly influenced therecording of the rarer
species. The total number of species recorded for the releves was 123 species. See
Appendix 2 for a full listing.
A second run was made with a more consistent data set. Only those species common to
the species lists of the four inventory methods, were retained. Where necessary, species
were grouped to genus level to improve comparability. This yielded 53 timber species. \
Species and sample scores on the first DCA axis are given in Table 3. The species scores \nie\
indicate the mode of the distribution of the species on the main vegetation gradient (ter
Braak 1987a). The *ecological amplitudeof the species can be read from the rangeof
samples in which the species occurs. Note that this sample score axis, ranging from 0to
243units on axis 1of DCA, henceforth called "u.DCAl", is not identical to the species
score axis, ranging from -120to 346 u.DCAl. Sample scores are weighted mean species
scores (Hill &Gauch 1980).
i
i
34
Forests ofSE Liberia and SW Cote d'lvoire
Text box 2
Problems with combining data sets of different origin and inventory method
In the pioneer years of a science such as vegetation mapping scientists had to build up their own data sets
from scratch. No previously collected data were available, or if they were, they contained far too little
detail. The sampling methods, after fast initial development, were standardized from a certain moment on.
In forest vegetation science many large data sets have been published, both for the tropics and for
temperate countries (e.g. Hall & Swaine (1981) for Ghana, Vanclay (1989) for Queensland, Noirfalise
(1984) for Belgium and van der Werf (1991) for Dutch forests).
Much new knowledge can be generated by combining existing data sets. This was attempted in the present
Chapter. The inventory methods used in the different National Inventory Campaigns, show clear
differences, as indicated in the text. To what extent do these differences in methods influence the results?
(see also Ashton 1977). Each scientist prefers to analyse data that have been obtained by the same method,
but in the present project it was impossible to re-examine the forests. Some forests have changed in the
meantime as a result of logging and some have even disappeared.
Ordination analysis shows structure in data without human scientific bias. In the first DCA run on the
crude data, the first axis (X=0.49) reflected the main wet-dry gradient. The second axis (X=0.21)
separated the inventory methods. The third and fourth axes could not be attributed to any specific
environmental variable. Along the second axis, my own sample plots (Zagne, Tai and Para) with complete
species lists of trees exceeding 70 cm diameter were grouped at one end, followed by method 1, yielding
most information on rarer species, and then methods 2 and 3, using only reduced species lists at the other
end.
The sampling method therefore is seen to be part of the structure of a data set. In this case, DCA showed
a difference in method on its second axis. The main effect, i.e. the wet-dry gradient, explained more
variance in the data as it was related to the first ordination axis which had X=0.49. This led me to
continue the analysis after adjusting the data set by excluding the species not consistently surveyed under
all methods. The second run showed that the methodological effect on the second axis had gone and its
eigenvalue dropped below 0.10 (X:!=0.06; X 3 =0.04; X 4 =0.02). The samples along the first axis were
better ranked as to their position on the climate gradient.
Generally stated, analytic results reflect both the methods used and the reality they are applied to. Making
comparable data sets is essential for the creation of integrated data systems. Sensitive tools must be
developed to separate methodological structure from scientific information structure of data sets.
Spatialanalysis. For spatial analysis, the sample scores (axis 1) were plotted on themap
(Figure 14)and contour lines of the DCA1 surface were interpolated using the *kriging
method, named after Krige (1951 ex Stein &Corsten 1991). This method interpolates
DCA1 scores on afixedgrid, here 25 by 25 km, as a linear combination of up to 10
surrounding data points. The weights of these points depend on the degree of their spatial
correlation (Stein &Corsten 1991). For thecalculations I used the GIS software package
Surfer 4.09 (®1989 Golden Software).
Across-section was made through the DCA1surface in the SW-NE oriented part of the
regional forest gradient, from Greenville to Vavoua (Figure 11). The decrease of the
spatially interpolated DCA1 score was plotted against the distance from the Liberian
coast. Above, the gradient is given, i.e. therate of changeof theDCA1 score with
distance and thusof the tree species composition of the forests.
On the DCA1 surface, a line of maximal change, i.e. a "line" of inflection, can be
drawn. It corresponds to a zone of quick vegetational change over a short distance,
The regionalforest gradient
35
separating two areas with slower change. I analysed theordination results to ascertain
whether or not there were significant spatial trends and if so, whether they were linear or
curved. If curved, I analysed thepoints of inflection, which are lines of inflection on the
map. I looked to see whether there were spatial anomalies that could berelated to the
relief and thus therainfall, and whether thetrend had a consistent geographical orientation, in other words, whether thechange in thevegetation corresponds to the SW-oriented
rainfall gradient.
2.4 Results
The raw data run produced afirstaxis (*eigenvalue X= 0.49) which roughly corresponded to the rainfall gradient, and a second axis (X=0.21) which separated the
samples by inventory method (seeText box 2). It was necessary to eliminate the differences caused by the inventory methods.
In the second run thefirstaxis (X=0.41) indicated the wet-dry gradient more closely and
the second axis lost most of its significance (X=0.06). In accordance with Hall &Swaine
(1976)and deRouw (1991) I continued the analysis with only thefirstaxis. The species
and sample ranking and axis 1scores are given in Table 3, together with the species
abundance in the samples.
2.4.1 Species ranking
In theupper left corner of Table 3, a number of typical wet evergreen tree species from
Liberia formed a group. Someof these species do not occur in Cote d'lvoire. Between
these "everwets" and a number of *ubiquitous species, i.e. present everywhere but with a
variable abundance, sometypical swamp species werepositioned by theordination, which
confirmed the findings of de Rouw (1991) that certain species that occur all over the
catena in a wetter climate are restricted to thevalley bottom in drier climates (e.g.
Heritiera utilis, Gilbertiodendronpreussiiand Sacoglottis gabonensis). Other species like
Mitragyna ciliataand Loesemrakalantha only occur in swamps (Voorhoeve 1965).
In thelowerrightcorner of Table 3, tree species typical of the semi-deciduous forests
were clustered, showing a linear response curve, or perhaps only the left-hand half of the
supposedly unimodal curve of their distribution. Someof them, like Ceibapentandra,
Terminalia superba and to some extent Triplochiton scleroxylon extended quite adistance
into the wet forests, probably because of their pioneer character (Aubreville 1959, Swaine
&Whitmore 1988).
Several tree species showed a uniform distribution over the entire range of forests
considered: Canarium schweinfurthii, Lovoatrichiloides, Daniellia sp.,Anopyxis
klaineana, Afzeliabella,Amphimaspterocarpoides, Terminalia ivorensis, Tieghemella
heckelii.However, this may partly have been an artifact of the rescaling method used.
The abundance score "1" was given to species present with one single tree, as well as to
species present with a timber volume that was equal to 15 %of that of the most abundant
species. This range is still quitelarge.
36
Forests ofSE Liberia and SW Cdte d'lvoire
ATLANTIC OCEAN
100 km
6°W.L.
\000h savanna
Figure 14
Spatial gradient analysis of the forests of SE Liberia and SW C6te d'lvoire. The grey tone
indicates the forest areas on which the ordination is based. The scores on the first DCA ordination
axis are plotted in these forests. The three circled scores indicate my three sample plots in Tai
National Park. The isolines join forest areas with the same DCA axis 1score and thus with similar
large tree species composition. The cross-section BB' is given in Figure 11. For sources of the
data, see Table 2.
The regionalforest gradient
Table 3 DCA ordination table with 53 large tree species recorded consistently in 33 forest inventory
compartments in SE Liberia and SW C6te d'lvoire. Read plot numbers, u.DCAl scores and
number of species vertically. Species which are known to be briefly deciduous (Hall & Swaine
1981) are marked "(b)d".
KKSKSKKSKKKKKPGGG
a
1
a
417T232565gl4367
6696966966666r555NXCaNNXXCNnNXXNN
9 8 2 6 3 7 7 1 5 3 4 2 la213WVS=iWWVVSW6WVVWW
u.DCA1
Largetreespecies
Dldelotla bmvipanlculata
346
Loesenera kalantha
325
Tetraberllnla lubmanlana
3 2O
Brachyslegla leonensls
312
Cynometra ananta
305
Casslpourea spp
296
Monopetalanthus spp
269
Aratopsls soyauxll
268
Herltlera Mills
248
DldelotlaIdae
231
Gllbertlodendronpreussll
2 26
Sacoglottls gabonensls
2 24
Cryptosepalum tetraphyllum
22 3
Mltragyna clllata
223
Lophlra alata
bd
216
Oldfleldla afrtcana
212
Canarium schwelnfurthll
bd
ISO
Lovoa trtchllloldes
163
Danlellla spp
d
156
Anopyxlsklalneana
154
Nauclea dlderrichil
149
Parinari/Maranthesspp
145
Afzellabella
d
137
Amphlmaspterocarpoldes .... d
134
Termlnalla horensls
d
117
Tleghemella heckelll
109
Chlomphora spp
d
95
Turraeanthus alrlcanus
d
94
Distemonanthusbenthamlanus . . .
87
Anthonotha fragrans
77
Berllnla spp
76
Enlandrophragma spp
d
54
Rhodognaphakmbrevlcuspe.. d
51
Khaya anthotheca
bd
48
Plptadenlastrumatrlcanum .... d
44
ErythrophleumIvorense
44
Pycnanthus angolensls
43
Antlaris toxlcaria
d
35
Klalnedoxa gabonensls
30
Peterslanthus macrocarpus .. bd
15
Rlclnodendron heudelotll
d
14
Guanacedrata
7
Celbapentandra
d
-35
Termlnalla superba
d
- 35
Nesogordonlapapaveritera
- 36
Celtlsspp
-61
Alblzleferruglnea
d -111
Mansonla altlsslma
d -112
Anlngeria robusta
-114
Trlplochlton scleroxylon
d - 115
Gulbourtta ehle .-.
d -116
Slercullarhlnopetala
d -119
Pterygota macrocarpa
d -120
Numberofspecies:
22222222211111111
43311110099864320866655433333221
320653390716887168654770988S48390
259 22--41-1
987 959939979-9 9 9 9 9 9 5 9 9 9 7 4 5 - 1-1-12 111-72-53-3 2 2 31612223-1- 211
--2 -3--3111
1-1
- - 1 - 1 - - 1 1 4 1 4 1 - 111
- 1 1 - 1 - - 1 1 1 1 1 1 - 111
9 9 6 974165977945342612122111- 11111
11- 1-111111-1-1413
1 1 2 2 2 2 - 3 2 8 8 9 9 - 5 3 2 1 1 4 - 3 1 - - 2 1 - 1• -1154 142224211192127-1-24--1--1
1111-111
253 1353431111-211211-111113-21111
299 1979564346-642351-791115141211
111 1 1 1 1 1 2 3 1 2 3 - 6 3 1 2 - - 5 1
1
1-1 1 1 2 3 2 1 1 1 1 1 1 1 1 2 1 1 1 - 1 1 1 1 1 1 1 1 1 1 1 1
111 - 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 - 1 1 1 1 1 1 - 1 1 1 1 1
-11 1 1 1 1 2 1 1 1 1 1 - 1 1 1 1 1 1 - 1 1 1 2 1 1 1 1 1 1 1 1
112 11111111111111-11-113311111111
1 2 1 3 1 3 2 1 1 1 1 1 2 2 3 3 2 4 4 2 3 1 25 2 1 1 1 1 1 1 1 1
113 13213111229363734322334111211-11 1 1 1 - 1 1 1 - 1 1 1 1 1 1 1 1 1 - 1 1 1 1 1 1 1 1 1 1 1 1
-11 1 1 1 - 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 1 1 -11 1
1111111111111-11311111-111
1-- 1
11111131111113111111-11111
--1 - 1 1 - 1 1 1 1 1 1 - 1 1 1 1 1 1 - 1 1 1 1 1 1 1 2 3 1 1 1
111111
11-1-11111--11-11
-11-111111-111111-111111311111
121 1 1 1 1 1 1 1 1 1 2 1 2 2 2 3 9 3 3 1 2 4 6 2 3 1 1 1 5 1 111 1 1 1 1 1 1 1 1 1 1 - 1 1 1 1 6 2 - 1 1 2 7 2 2 - 1 4 3 1 1
-1--11111115331119111111Z15111
1
11111111111311113312111111
11111-1111112111111-11111
113 123231154314567799979988189951
112 111212211213223914348715633532
111 1 1 2 1 1 1 1 2 2 1 1 6 6 7 5 5 2 9 5 5 6 7 3 4 1 4 3 8 5 1
11111111111111111111125111
1 1 1 1 1 1 1 1 1 2 1 1 2 1 2 1 2 4 6 5 3 23 8 8 22 1 1 5 72 3
1
1
1
1 1 1 1 1 1 2 4 3 3 2 4 2 4 3 1 552 2 5 5 2 4 1
--1 1
1111
1111111111112221
1-11-111111-111111111111
- 1 - 111-111211-1123333655247293789
1-1-111111-1122412224343421559
1
1111111111111111111
11
111-111111-15211
111-111211111111
11--111111-11111
1111111211112211
1 1 3 23 2 1 6 9 9 6 7 9 4 9 9 9 9 9
11--111211111111
11-111111-11111
111-111111-1-113
223 3 3 2 2 4 4 4 4 4 3 Z 4 3 4 4 4 4 2 4 4 4 4 4 4 3 4 3 3 4 3
3812391033309019220023Z0020028917
37
38
Forests ofSE Liberia and SW Cote d'lvoire
Theunimodal response model (ter Braak 1986) was found for species like Gilbertiodendronpreussii,Anthonothafragrans, Oldfieldia africana and for thegenera Parinari
and/or Maranthes. They were present over theentire range of forests, which was 400km
wide, but each showed a clear optimum.
2.4.2 Sample ranking
It is striking that the forest inventory compartment furthest SW (codeK69 on Figure 13)
showed up in Table 3at one end of the first ordination axis (243u.DCAl) and the forest
compartment furthest NEat the other end (0 u.DCAl), in perfect accordance with what
we might expect from the rainfall gradient. TheKrahn-Bassa and SapoNational Forests
in Liberia are thewettest forests in this *forest block, varying from 243down to 186
u.DCAl. They are characterized by the dominance of Caesalpiniaceae tree species in the
forest canopy and contain several *endemic species (Tetraberlinia tubmaniana, Didelotia
brevipaniculata, Loesenera kalantha, Monopetalanthus compactus). I could not study the
extent of these forests in the extreme SWof Cote d'lvoire (hinterland of Tabou, Grabo
hills, Foret classee de laHauteDodo), as no forest inventory reports were available for
this region, but Aubreville (1959) and Guillaumet (1967) report on thepresenceof
numerous *endemic species in this SWcorner where they located the "wettest" forests of
Cote d'lvoire.
The most north-eastern compartment of Krahn-Bassa forest K61 (168u.DCAl) was
already transitional in composition towards the Grebo National Forest. The wettest forest
compartment studied in Cote d'lvoire was the Foret Classee du Haut-Cavally (code4NW:
88u.DCAl). It was the only inventory compartment in Cote d'lvoire where the everwet
treeDidelotia idae(broutou; Oldeman 1964) was found. This compartment, containing
Mt Trou (Figure 11), lies on thewindward sideof the second major ridge which forms an
obstacle to the maritime air masses drifting inland.
The three sampleplots Zagne, Ta'iand Para had positions of respectively 35, 64and
148u.DCAl on the first ordination axis. Whereas Zagne and Ta'i fitted quite well into the
local gradient as plotted on the map (Figure 14),thePara plot was an outlier, which
according to the ordination could have lain in Liberia. This may have been caused by the
great abundance of Sacoglottis gabonensis on the Para site (Table 3, Figure 26). The
Zagneplot is similar to the neighbouring Reserve deFaune du N'zo (code 5NW). The
three sampleplots were far from internally homogenous in tree species composition. The
intra-samplevariation willbeanalysed in Chapter 3.
39
The regionalforest gradient
Numberoftreespecies(d>70cm)outofalistof53
50-
150
Figure 15
200
250
DCAaxis1scoreoftheforest
Number of timber species per forest inventory compartment out of a list of 53 large tree species,
consistently recorded in all inventories. The x axis gives the DCA axis 1 scores of the samples.
My sample plots near Zagne, Tai and Para each lacked certain species found in other samples on
the same position of the gradient, partly because they covered a smaller area. Towards the wettest
end (right) a drop in species richness was observed. These forests are rich in Caesalpiniaceae
species which locally may dominate the vegetation (Voorhoeve 1965).
2.4.3 Species richness
The rather small plot size of my three sampleplots (20ha each) and the deliberate
exclusion of swamps madethem the samples poorest in species. Figure 15gives the
number of species per compartment out of the list of 53consistently recorded ones.
Compared with other forests on the sameposition of thegradient, they lacked a number
of species that could have occurred when considering Table 3. Someof these "absent"
species are typical swamp species, others are rarer ubiquitous ones. Concerning thePara
site, Guillaumet (1967)already indicated that the forests on the sericite-chlorite schist
band were poorer in evergreen species than the forests around Grabo. Perhaps since the
last Ice Age certain species, e.g. Caesalpiniaceae with relatively heavy seeds, havenot
yet reached all theareas where they could ecologically occur (JJ.F.E. deWilde, pers.
coram.).
Among all forests there was a slight negative trend in species richness at the wetter endof
the gradient. The incompleteness of the records, however, urges caution. In Chapter 3
these aspects of diversity will bediscussed in moredetail.
40
Forests ofSE Liberia and SW Cote d'lvoire
2.4.4 Spatial analysis
Plotted and interpolated on the map, the general direction of the gradient was SW-NE
(Figure 14). A spatial anomaly occurred, from thePara study siteupto Sassandra river.
Three samples (Para, 1XVand 5XV) were ranked "wetter" than was tobe expected from
their geographical position. Comparing Figure 14to the lithology of thearea (Figure4)
revealed that all three of them lieon theband of sericite schist bedrock. The soils derived
from sericite schist bedrock are known to be more clayey and have abetter water
retention capacity (Guillaumet 1967, DRC 1967b, van Herwaarden 1991a). In theextreme
SEcorner theinventory compartment coded 7CS, ranked out at 65u.DCAl, which is
again "wetter" than expected. Here the coastal effect of rainfall may intervene, buta
comparison with neighbouring forests isneeded toverify this.
2.4.5 Cross-section through the gradient
The SW-NEoriented part of the gradient was chosen to makea cross-section. On this
cross-section (Figure 11)and also on the map it became clear that the gradient wasnot
linear, but was sigmoidal with apoint of inflection in the middleat about 140u.DCAl. A
zoneof faster compositional change was found, more or less perpendicular to the SW-NE
oriented gradient. This zone corresponds to a steeper part of therainfall gradient which I
detected in therain shadow from Putu range. In the literature on Liberia (Voorhoeve
1965, Sachtler 1968) this zone has been recognized as the transition between evergreen
and semi-deciduous forest and as the southernmost position of the Intertropical Front (see
Figure 21). The rest of thegradient has a rate of change of about 5 u.DCAl per 10km.
2.4.6 Coverage of the vegetation gradient by the National Parks
Theposition of Tai National Park on thegradient isbetween 40 and 150u.DECl. The
Park does not even cover half the length of the entire gradient, i.e. 240 u.DCAl. Sapo
National Park covers an even less diverse stretch on the wettest side of the gradient (210230 u.DCAl).
41
Tlieregionalforest gradient
2.5
Discussion
2.5.1 Comparison withthe forest gradient in southern Ghana
Hall &Swaine (1976; 1981;Swaine &Hall 1976)analysed and described the forest
vegetation gradient in Ghana, which is largely similar to thegradient in SE Liberia and
SW Cote d'lvoire. They used the ordination technique of reciprocal averaging (RA, Hill
1973), as DCA (Hill 1979a) had not yet been invented. The species scores of the
Ghanaian ordination correlate quite well with the scores of my ordination (R=-0.75,
p<0.01, see Figure 16).The species that are ranked differently in both ordinations often
have a uniform distribution without a clear optimum and the minimal abundance scoreof
1, and include Canarium schweinfurthii, Chlorophora spp., Aningeria robusta, Guibourtia
ehie. Species with the sameabundance haveno influence on theordination and maybe
eliminated from the data set. These rare "wandering" species, however, are typicalof
tropical rain forest, and it is merely therescaling method which is not sensitive enough to
identify their optimal conditions. Hall &Swaine (1981) described a gradient that extended
much further into the drier forests.
speciesscoreinthisChapter (u.DCAl)
400
300
200
100
Did.bre
Loe.kal
Cyn.ana
Bra.leo
o °
TeUub
Mon.sp
Sac.gab Cassip.sp £ , e r u t i
Old.afr
°Lop.ala
R=-0.75
p<0.01
Can.sch
D
° - Dan.sp
J A n ° l a Afz.bel
D
Par.sp~ ^
° Amp.pte
Ant.fra\Zo[;afr[Pis.ben
° Be°r.sP T f c f c ^ B i u p
a
RhoTbre , „ • . ,
r>\^~
Df. Kr
Pip.afivKha.ant
Kla.£abo n D p !P- a , N
L v tri
Tri.scl Alb^fer
- a " - B Man -_ a|t
Pte.mac
drier species
-i
10
Figure 16
Anttox
Ani.rob C e L s p
Gui.ehiaS"—D
Ste.rhi
-100
-200
Chl.sp
•
1
1
20
1
1
30
1
|
i
r
40
50
60
species scoreinGhana(U.HS1)
Correlation of my species scores (u.DCAl) with those found by Hall & Swaine (1981) (u.HSl).
Beside the species common to both data sets, a number of "everwet" species are listed in the upper
left corner. Hall & Swaine's (1981) ordination extended much farther in the dry forests.
42
Forests ofSE Liberia and SW Cote d'lvoire
samplescoreintheGhanaianordination(u.HS1)
30-
25-
6CS
5XV
K 2
^-—,3.
2XV
„
g
>
rsci" >»^
«fi1
nK§?
D D
20-
D
1*5l 3
^
10-
K167
l
0
DK66
|
50
l
|
100
I
|
I
150
|
a
K69
I
200
250
samplescoreinthisChapter(u.DCAl)
Figure 17
Correlation of my sample scores (u.DCAl) with scores calculated after the species scores given by
Hall & Swaine (1981) (u.HSl). On the y axis the Ghanaian forest types are indicated.
I applied the method of identification by 'coordinate estimation' (Swaine &Hall 1976)to
my samples, i.e. I calculated the average species scores of my samples using the
Ghanaian species scores and found that the forests I considered fell within a range of 15
to 28Hall &Swaine (1981) axis 1units (u.HSl; see Figure 18). The correlation with my
own sample scores is good (R=-0.91, p<0.01, seeFigure 17).However, on theHS1
axis my samples formed two clusters: all Ivorian forests except Para at 27 u.HSl and the
Liberian forests plus Para at 20-24 u.HSl, thewettest forest going down to 15u.HSl.
Thepresence of thefirstcluster is an indication that the Ghanaian ordination did not
reveal the differences between the Ivorian forests because only a limited list of large tree
species was considered.
In Ghana only my wettest forests would beclassified as wet evergreen (15-20u.HSl), the
other ones as moist evergreen/moist semi-deciduous (20-33 u.HSl). No dry semideciduous forests were represented among the forests I studied. The distinction between
moist evergreen and moist semi-deciduous forest in Ghana was based on axis 6of Hall&
Swaine's ordination with an eigenvalue of 0.13 and thus explaining littleof the variation.
My 'moist' forests, i.e. with Hall &Swaineaxis 1 values above20u.HSl, haveaxis 6
values between 37and 42 u.HS6, which would indicate that they are transitional between
the moist evergreen (35-40u.HS6) and the moist semi-deciduous (40-44 u.HS6) forests of
these authors (Figure 18).
43
The regionalforest gradient
HS6scoreofmysamples
3U
K69
45-
3
a
1NW
2NW
3NW
5NW
1XV
2XV
3XV
K167
MS
K68
a
D
S92
S93
,
D
•
S91 •
^,
40-
K67
a 1
K66 .
WE
K64
•
K65 „ G52
EPG53
K63
"" « .
G51
a
K61
V 5XN/L -7
\6CS/
6NWT
4XV °°D4NW
a o a
»
_j8s TaT 7NW
K62
D
Zagnd
1*\
i
15
i
i
1
D
•
Para
|
i
1
20
'
* '
ME
i
i
i
i
i
i
i
25
i
30
HS1scoreofmysamples
Figure 18
My samples placed in the Ghanaian ordination diagram by 'coordinate estimation', i.e. the
coordinates of a sample as given by the mean species scores, taken from ordination axis 1and 6 of
Hall & Swaine (1981). The interrupted lines indicate the division in forest types as made by Hall
& Swaine (1981); WE: wet evergreen forests; ME: moist evergreen forests; MS: moist semideciduous forests.
It would be interesting to see theoutcome of a detrended correspondence analysis (DCA)
on the Ghanaian data to complement the reciprocal averaging practised by Hall &Swaine
(1976).
De Rouw (1991) used this technique of coordinate estimation to position therelevesof
Guillaumet (1967) in her ordination diagram. All Guillaumet's releves fell within her
driest forest type, even if they originated from wetter forest. De Rouw (1991) argued that
this wascaused by the sampling method used by Guillaumet (1967) with which he
undersampled theground flora and the understorey tree species. Coordinate estimation
seemed to work well only if the same life forms were recorded and thus the same
inventory criteria were applied both in the original survey and in thenew releves.
At the wet end of mygradient four "everwet" and very abundant tree species were ranked
beyond the wettest Ghanaian species, Cynometra ananta,that both sets have in common.
This means that in Liberia there are wetter forests than in Ghana. Tetraberlinia tubmaniana,Loesenera kalantha and Didelotia brevipaniculata were very abundant in these
44
Forests ofSE Liberia and SW Cote d'lvoire
forests, and theBrachystegia leonensis trees occurring there are among the tallest trees in
Liberian forest. Sachtler (1968)and Voorhoeve (1964) even report on single-dominant
forest with Tetraberlinia tubmaniana in all height levels of the forest canopy (Figure22;
see also Hart (1990) and White (1983) on single-dominant forests).
2.5.2 Vegetation studies in SW Cote d'lvoire
Guillaumet (1967)published a thorough studyonthevegetation andthefloraofSW Cote
d'lvoire. He made releves of all vascular plants on 21 plots, covering from 150to
1000 m2each, and described soil-vegetation catenas on different bedrock types. He
classified theplant and tree species according to three criteria: position in the forest
canopy at maturity (herbs and shrubs, lower, middleand upper canopy speciesand
epiphytes), light intensity needed when young, and water retention capacity of the soilin
which they grow. Based on the last criterion hedistinguished *psammophilous species and
*pelophilous species (the former grow on sandy soil and the latter on clayey soil) plusa
category of indifferent species.
On hisvegetation map of SWCoted'lvoire (Figure 19), heindicated a predominanceof
pelophilous species in all topographical positions in the extreme SWcorner of Cote
d'lvoire: from Guiroutou over Grabo toTabou. On the sericite schist bedrock these
pelophilous species are still dominant but fewer species occur. In therest of SWCote
d'lvoire pelophilous species are restricted to lower slopes. Psammophilous species are
only dominant on a narrow band of Tertiary sands starting east of San Pedro and
extending east upto the Ghanaian border.
In theextreme SW-corner of Cote d'lvoire, "Liberian" high rainfall conditions overruled
the influence of soil texture (Guillaumet 1967). Pelophilous plant species do grow there
not because the soil is soclayey, but because of optimal moisture conditions. Drought
stress is a combined effect of climate and soil. With respect to this it may be better touse
the terms "drought stress tolerant orintolerant species"(see further Chapter 3).
My results confirm the findings of Guillaumet (1967) that "wetter" forests existon
sericite schist bedrock. Adirect comparison is difficult, as his data on absence or
presence of large tree species are much less complete than mine, because he sampled a
much smaller area. On the other hand, most of Guillaumet's and deRouw's (1991)
discriminatory species are lower canopy species, herbs, lianas and shrubs. Both studies
haveonly limited comparability because of thisprofound difference in criteria.
Thepresence of species from the semi-deciduous forests and theabsence of pelophilous
species in a rather large area west of Sassandra river and south of the town Soubre
(forests coded 3XV and 4XV; Figure 13)was also confirmed. Guillaumet (1967) did not
relate thepresence of these drier forests to different bedrock but simply to a drier climate
and a longer dry season. Here again, the problem of interpolation of the climatic data
from the weather stations appeared (see subsection 2.1.1). Rainfall stations are scarce and
therefore neither the existence of a possiblerain shadow of the hills of Grabo, nor of the
effect of the general descent of elevation towards Sassandra river could be proved. The
southern inventory compartment of Grebo National Forest (G53) in Liberia also lay
windward of a large river and in the same way was the driest forest in its region.
The regionalforest gradient
Figure 19
Vegetation map of SW Cote d'lvoire as drawn by Guillaumet (1967)
45
46
Forests ofSE Liberia and SW Cote d'lvoire
Guillaumet &Adjanohoun (1971)published a vegetation mapof Cote d'lvoire at a scale
of 1:500 000. They lumped the wet evergreen forests of Grabo and theimpoverished ones
on theband of sericite schist bedrock together into oneevergreen forest type: "the
Diospyros-Mapania spp. evergreen forest". On this new vegetation mapit was striking to
note that theborder of this wettest type, especially within theTai National Park, was
placed considerably northwards by comparison with its position on the map in Guillaumet
(1967; Figure 19).
The only characteristic treespecies for this type, mentioned by Guillaumet & Adjanohoun
(1971), isHeritiera utilis,which in my data set is ubiquitous with a clear maximumof
abundance in the wettest forests of Liberia. Under its trade name "Niangon" it is themost
important commercial species in Liberia. It represents 27 %of the felled wood and 36%
of the timber valueexported from Liberia in 1989 (FDA 1990).H. utilisalso occurs in
the other evergreen forest types, so it is the association of herbs, lianas and small tree
species that yields characteristic species for this forest type.The forest tree association of
Cynometra ananta,Brachystegia leonensis, Didelotia brevipaniculata, Cassipourea
nialatou and others which is called the mixed evergreen forest in Liberia (Sachtler 1968,
Voorhoeve 1965), does indeed occur on the hills of Grabo, but not on the sericite schist
band in Tai National Park. I suggest separating these forests as distinct types on vegetation maps. The mapby Guillaumet &Adjanohoun (1971) suggests that the same forests
are found around Grabo and in the major part of TaiNational Park. This could be used as
an argument against a special protection of the forests around Grabo.
However,the Font classee delaHauteDodo and the hill ridgefrom Mont Kopi to
Mont Kedio and Mont Bodeledocontainnumeroustreespeciesfound neitherin Tai
NationalPark,noranywhereelsein Cote d'lvoire (Aubriville1959,Guillaumet
1967). Hence theymeritspecialprotection and conservation management (Vooren
1992b).
North of theroad Guiglo-Toulepleu Guillaumet &Adjanohoun (1971) distinguished a
separate climatic forest type, considered tobe transitional between evergreen and semideciduous forest: "the forest with Uapaca corbisieri (syn.: U.esculenta), Uapaca
guineensis and Chidlowia sanguinea". Theinventory compartment 1NW, Foret classeedu
Scio, lay in this forest type and had a u.DCAl score of 34, thus indeed transitional. In
composition it is similar to the Reserve du N'zo (5NW), the Zagne sampleplot and the
inventory compartments 5XV and 6CS (Foret classee deNiegr6,just east of Sassandra
river).
The Foret classee deDuelcoue'(6NW, 19u.DCAl) and the forests east of Mont Peko
(7NW, 0 u.DCAl) are drier than the Scio forest and were classified by Guillaumet&
Adjanohoun (1971) as semi-deciduous forests with Celtis spp. and Triplochiton
scleroxylon.
Further, there is the issue of transitional types and the transition between types which
posed problems for vegetation mapping. It is illustrative to see how Guillaumet&
Adjanohoun (1971) depicted the transition of evergreen to semi-deciduous forest in three
The regionalforest gradient
different ways on their map: 1)by a colour code specific to a transitional type; 2)by
alternating light and dark green bars, i.e. a mixture of their two neighbouring forest
types, and 3) by lightgreen (semi-deciduous) dots in a dark green (evergreen) matrix, i.e.
another kind of mixture.
Faced with these transitional forests, it is convenient topass from mapping types atan
*ordinal scale, to mapping isolines, corresponding toan interval scale. When mapping a
continuous gradient by e.g. three types, both the extreme types are well characterized, but
often the middle oneis not (I.S. Zonneveld, pers. comm.). The clearest proof of thisis
the forest type that Guillaumet &Adjanohoun (1971) called "evergreen forest with
Eremospatha macrocarpa andDiospyros mannii",which is "essentially based on negative
characteristics, namely the absence of species typical of other forest types." (Guillaumet
& Adjanohoun 1971,p. 171). The same weak characterization was used by Hall &
Swaine (1981) for their "Moist semi-deciduous forest type", lying in the middleof the
gradient. Very few species were found to beconfined to this type and, at the same time,
it is the most extensive forest typein Ghana. Hall &Swaine (1981, p. 37), usingan
example ofDiospyros species in Ghana, showed that there is nogeographical coincidence
of ranges of species to support these forest types.
From this, I conclude thatmappinga spatial gradientusingisoscore lines resolved
theproblemof transitional types.
2.5.3 Forest zonation in Liberia
The German Forestry Mission to Liberia, aproject of the German Gesellschaft fur
Technische Zusammenarbeit GTZ, carried out a National Forest Inventory in Liberia
during the sixties (Sachtler 1968). I abstracted climate and vegetation data from their
reports. Sachtler (1968) distinguished two major vegetation zones in Liberia: "the
evergreen forests" and "the moist semi-deciduous forests". He considered the December
position of theIntertropical front as theboundary between the two zones (Figure21).On
Figure 14Sachtler's line followed my 120u.DCAl contour line through Zwedru and
along the Duobe river. This line also followed the northern limit of the distribution of
Didelotia idae(Figure 20). When extrapolated into Cote d'lvoire, the 120u.DCAl line
would coincide with the northern limit near Guiroutou of the wet evergreen forest as
distinguished by Guillaumet (1967;Figure 19).
Throughout Liberia this boundary linewas not congruent with any isohyet. In NW
Liberia it would be at 2800 mm rainfall, in SELiberia at 2000 mm. This boundary line
also corresponded to thelineof inflection, i.e. thegreatest rate of compositional change,
of the forest gradient found in Figure 11. If the forests were to be divided into two
groups, then thisline where the gradient is steepest, would be the most appropriate limit.
47
48
Figure 20
Forests of SE Liberia and SW Cote d'lvoire
Distribution of four Caesalpiniaceae tree species, characteristic of the wet coastal forest in Liberia
(from Sachtler 1968)
The regionalforest gradient
Figure 21
Distribution of two Caesalpiniaceae tree species typical of the mixed evergreen rain forest in
Liberia (from Sachtler 1968). The grey band indicates where Sachtler (1968) positioned the
boundary between evergreen and moist semi-deciduous forest and, at the same time, the southernmost position of the Intertropical Front.
49
50
L
Forests ofSE Liberia and SW Cote d'lvoire
7.6 m
60 m
Large tree species list in Yoma-Gola
Afzelia bella
Amphimas pterocarpoides
Anopyxis klaineana
Anthonotha fragrans
Araliopsis soyauxii
Berlinia confusa
Brachystegia leonensis
Canarium schweinfurthii
Chlorophora regia
Cryptosepalum tetraphyllum
Figure 22
forest (Voorhoeve 1964). Only
Daniellia thurifera
Didelotia idae
Erythrophleum ivorense
Gilbertiodendron preussii
Heritiera utilis
Klainedoxa gabonensis
Lophira alata
Maranthes aubrevittei
Maranthes glabra
Mitragyna ciliata
species from Table 3 are listed.
Monopetalanthus compactus
Nauclea diderrichii
Oldfieldia africana
Parinari excelsa
Piptadeniastrum africanum
Pycnanthus angolensis
Ricinodendron heudelotii
Sacoglottis gabonensis
Tetraberlinia tubmaniana
Profile diagram of a single dominant forest of Tetraberlinia tubmaniana J. Leonard in Yoma-Gola
National Forest near Bomi Hills, Liberia, 1962. The diagram represents a strip of forest 61m long
and 7.6 m wide. Tetraberlinia trees are shaded and marked with a "T" in the ground-plan, a single
Treculia africana tree with a "Tr". Reproduced from Voorhoeve (1964).
The regionalforest gradient
Figure 23
Evergreen rain forest along Lofa river in west Liberia (Photo by H. Dop)
1. Within the "evergreen *forest zone" (Figure 23), Sachtler (1968) distinguished two
sub-zones.
la. The wetcoastal rainforest, ranging along the coast between River Cess and Harper
in abelt of maximally 80km wide (see thedistribution ofDidelotia brevipaniculata on
Figure 20). The wetter compartments of Sapoand Krahn-Bassa Forest fell within this
sub-zone (more than 200 u.DCAl on Figure 14). According to Sachtler (1968), these
forests were characterized by theoccurrence of single-dominant forests of Tetraberlinia
tubmaniana, by the occurrence ofDidelotia brevipaniculata and by a largeproportion of
swamps (20to 30 %)covered with very dense stands ofLoesenera kalantha (see
Figure 20 for distribution mapsof these species). Gilbertiodendronpreussiiand Monopetalanthus spp., typical of "mixed evergreen forest", the second sub-zone, were rare,
and of all Meliaceae only Lovoatrichiloides was found.
The same combination of Tetraberlinia, Didelotia andLoesenera seems to exist in thewet
coastal forests of West Cameroon, in Korup National Park and near Kribi (Letouzey
1968,White 1983). These forests in Korup National Park are associated with soils
displaying littleavailable phosphorus (Gartlan et al. 1986). Their Caesalpiniaceae tree
species live in symbiosis with ectomycorrhizal fungi (Newbery et al. 1988)which,
according to these authors, mayexplain their often gregarious habit.
51
52
Forests of SE Liberia and SW Cote d'lvoire
lb. The mixed evergreen rainforest, its most frequent species being Gilbertiodendron
preussii(seeFigure 21 for a distribution map) and Monopetalanthus compactus. Patches
of Tetraberlinia tubmaniana still exist, but the forest is essentially mixed, there are twice
as many species as in the wet coastal forest and single-dominant stands are rare (Sachtler
1968).
2. In the "moist semi-deciduous forests" Gilbertiodendronpreussii, Monopetalanthus
spp., Sacoglottis gabonensis, Didelotia and Cassipourea spp. become increasingly rare
and are replaced by large numbers of shade intolerant species which shed their leavesin
thedry season (Figure 24):Entandrophragma spp., Khayaanthotheca, Piptadeniastrum
africanum,Triplochiton scleroxylon andothers.
Sachtler (1968) reserved the term "forest types" for the *inventory strata he distinguished
locally in the inventory compartments:
- forests on uplands, in swamps, on river banks and on steep slopes;
- three regeneration stages after shifting cultivation;
- forest damaged by elephants;
- monospecific Tetraberlinia high forest;
- natural low bush or extensive areas of windbroken forest.
This physiognomic nomenclature is essentially different from the two vegetation zones
mentioned above.
, .'^n*Y' !i ^ s
..sjir ;->*..- -,.•'
Figure 24
Semi-deciduous forest in north-west Liberia upstream Lofa river (Photo by H. Dop). Note the
numerous trees that are leafless. This photo was taken on the same day in February as Figure 23.
The regionalforest gradient
53
2.5.4 The exclusion of swamp forests
The forest gradient, described in this Chapter, only applied to non-degraded 'high forest'
on upland soils, to use the terminology of Sachtler (1968).Thevariety of landforms,
however, isgreat and the transition between them isgradual. In most cases, swampscan
be excluded easily, but lower slopes and river banks, which are assumed to be fed by
groundwater as well, are more difficult to distinguish from upland forest. In my data set,
which mainly comprises systematic inventory samples, swamps were included in Liberia
and in the inventories of Northwest and Centre-South Cote d'lvoire (seeTable 2). They
were excluded, however, from the "Perimetre industriel XV" and from my three
permanent sample plots. When verifying theoccurrence of the swamp specialist tree
Mitragyna ciliata in Table 3, I found it was only absent in my sampleplots, although it
grew in swamps near all three of them. In thedata set of the "Perimetre industriel XV"
this species was still recorded. This shows that in large inventories the exclusion of
swamps was not completely "waterproof".
2.5.5 The issue of "primary"and "secondary" forest
Even more complex is the desired exclusion of forest degraded by farming. Sachtler
(1968) considered most of theLiberian forests as 'secondary high forest' and even singledominant forests were said to have originated after farming (Voorhoeve 1965,Hart
1990). Indeed, on all three study sites I sampled within Ta'iNational Park pottery shards
and charcoal were found when digging the soil profile pits. Humans havebeen present in
Africa since time immemorial and they havebeen practising shifting cultivation for
thousands of years. The debate about "primary" and "secondary" forest in West Africa
has always been very lively (Chevalier 1948, Aubreville 1949, Schnell 1950, Mangenot
1955, Aubreville 1959, Voorhoeve 1965,de Namuret Guillaumet 1978,Guillaumet&
Adjanohoun 1971,Hall &Swaine 1981).
For the interpretation of forest gradients it is important to know whether or not tree
species which occur naturally in drier forests are able to colonize former farmland ina
wetter forest zone (e.g. Ceiba pentandra,Terminalia superba,Pycnanthus angolensis,
Canarium schweinfurthii, Triplochiton scleroxylon, Chlorophora spp., Piptadeniastrum
africanum on Figure 25, etc.). This would rank the resulting forests on a drier position
along the gradient and hence mask zonal effects. On the other hand, a species can never
be typical of secondary forest alone, it must havea natural habitat, e.g. steep slopes, rock
outcrops or other sites unfavourable to typical "primary" species. These special sites may
be located within large areas more hospitable to forests, and a systematic sampling design
may include someplots in such special forests. This is in agreement with*Budowski's
rule (1965; amend.Oldeman 1990b) thatanyplant species whichplaysapioneering
rolein hospitableenvironments hasageographical distribution whichincludes less
hospitable environments.
54
Figure 25
Forests ofSE Liberia and SW Cote d'lvoire
A large Piptadeniastrum africanum crown in a field south of Tal. The tree was not felled because
of its huge buttresses and very hard wood. This species may profit from its position and regenerate
in the abandoned fields.
2.6 Conclusions
Theregional forest gradient presented in this Chapter was extracted from large samples
covering 20to 2000ha each. This was necessary to obtain reliable data on the occurrence
and abundance of the large tree species. The resulting species composition was a spatial
average over thearea. Thus theentire gradient also reflected theaverage species
composition.
The compositional trend correlated well with the SW-oriented rainfall gradient. In part of
the map the forest gradient showed an anomaly. The forests on a band of sericite-chlorite
schist from Tal National Park towards theNE were ranked 'wetter' than expected from
their position on the rainfall gradient. Apparently, the rainfall effect was compensated by
the greater moisture content of the soilsderived from sericite-chlorite schist.
Thegradient cannot be divided in an *objective way. Its best representation is that of a
/ DCA1or compositional surface that covers thecontinent like a continuous blanket with
different local patterns and builtby the complete setof species. Its major driving force is
found to be the moisture conditions determined by rainfall and lithology.
In this way the natural vegetation provides a more detailed map of these conditions than
therainfall records or geological surveys can provide. But thevegetation also has a
memory in which its history can be read far beyond the start of scientific research in the
region. Complex systems require powerful approaches that are scale sensitive and that
safeguard thedetail.
The regionalforest gradient
Individual sample plots of e.g. 20ha may show up much wetter or drier on this gradient
than theinventory compartment they liein, because of the effect of spatial averaging that
results from grouping thedata from all sampleplots within an inventory compartment.
When wezoom in and thearea covered by apixel becomes smaller, thegradient will
show moreand more "noise" and theregional trend will become moreand morevague
and overruled by local lithology and catena effects. At a certain pixel size, forest
dynamics and gap formation processes will also influence tree species composition.
In the next Chapter I take up thechallenge and test the robustness of theapproach after
changing the scale and zooming in into moredetail. The once so smooth vegetation surface will become more "hairy" and new techniques will beneeded toadapt thedegreeof
spatial averaging. Lithological mapsare replaced by soil maps, but topography and relief
still hold thekey to understanding vegetation variability.
55
56
Figure 26
A Sacoglottis gabonensis tree at the Para site. Here this species was found all over the toposequence, whereas near Tai it occurred only on lower slopes and in valley bottoms. Near Zagne it
was absent from all slope positions.
57
3
FOREST GRADIENTS ALONG SLOPES
IN TAI NATIONAL PARK
I concluded from Chapter 2 that on a regional scale the tree species composition in forest
areas changed gradually along therainfall gradient in West Africa. Species composition
was averaged over large areas of about 50000 ha, as a result of which local variation due
to landscape diversity was ignored. However, many large tree species not found in upland
forests occur over the landscape in valley bottoms and along rivers. Within the upland
forests, forest gradients are found along slopes. These are the subject of this Chapter.
Forest gradients along slopes havebeen studied in temperate (Whittaker 1956and 1967,
Bormann et al. 1970)and in tropical climates (Schnell 1952,Whitmore 1984, Ohsawa et
al. 1985,Nakashizuka et al. 1991).Most of these studies focused on mountain slopes
spanning altitudinal intervals of hundreds or even thousands of metres. Temperature,
cloudiness and rainfall change considerably along such elevation transects. These
altitudinal gradients are related to a changein climate as was the regional forest gradient
described in Chapter 2.
3.1 Localforest gradients
Slopes in Ta'iNational Park are not mountain slopes. Ta'iNational Park is on a dissected
*peneplain (see Chapter 1).The highest terrain positions, the crests, are remnants of
extensive ironstone sheets that covered theTertiary peneplain and probably developed
during drier periods in thegeological past (Tagini 1972). The valleys that dissected this
peneplain are 20to50 mdeep, and a typical *sequence of soils (*catena) has developed
from thevalley bottom up to the ironstone crest (see Figure27).
Altitude
198 m
Profile
depth
verticallydrainedsoils
Figure 27
superficiallyandlaterallydrainedsoils
waterioofledsoils
Example of a soil and vegetation catena in the Tar study area (modified after Vooren 1985)
58
Forests of Tai National Park
3.1.1 Sequential models
Oneway of describing such a landscape is to makepoint observations at regular intervals
from the summit to the valley bottom. Typically, five physiographic positions are chosen:
summit, upper slope, middleslope, lower slopeand valley bottom (deRouw, Vellema &
Blokhuis 1990).The observations form a sequence related to topography: a *toposequence.
In soil science, such a sequence of soilsis called a *soil catena,if all the soils are
underlain by the samerock type and occur under similar climate (Ahn 1970). InWest
Africa soils frequently occur in such sequences (Nye 1954, DRC 1967b, Ahn 1970,
Lawson et al. 1970,Fritsch 1980, deRouw et al. 1990).
This model has been used as abasis for physiographic soil maps (Lawson et al. 1970,van
Herwaarden 1991a). Most of the lines in these mapsdo not represent ecological boundaries, but connectpoints somewhere half-way between twophysiographic positions. The
extent of alluvial soils in thevalley bottom and of the iron pan on the crest can, in most
cases, be mapped objectively. Along the slope, changes in gravelliness, colour and texture
occur gradually, so subdivisions are drawn arbitrarily. Most authors stress thefact thatin
reality there isa continuum (deRouw et al. 1990).
In vegetation science, thedistinction between riverine, swampand *upland forest isbased
on comparison of similar point observations (Lindeman and Moolenaar 1959, Schulz
1960, Guillaumet 1967, Hubbell &Foster 1983and 1986). Using physiographic soil maps
as abase, Lawson et al (1970), Huttel (1977) and Jonkers (1987) madevegetation relev6s
on each physiographic position. Swamp forest appeared tobe the most distinct, but some
indication of compositional changealong theupper parts of these vegetation catenas was
found (see also Hall &Swaine 1981;Longman and Jenfk 1987,Boddez 1989).Methodological problems were encountered when analysing this directional variation. Unlikethe
soil surveys for which a standard methodology isavailable (FAO 1977, Touber et al.
1989), the vegetation surveys struggle with the huge size that trees can attain, thewide
range in tree sizes and life forms to be sampled, the great species diversity and the forest
dynamics which mean that the species composition in a plot ispermanently changing.
Apparently, the vegetation has to be considered at different scale levels than theone
determined by soil, and, moreover, it is madeup of living components (Oldeman 1990b),
making it moredynamic. It is not as self-evident to makepoint observations, as it isin
soil survey.
3.1.2 Gradient models
Thegradual changein soil and vegetation along slopes can also be studied using a
continuous model of variation. Soil characteristics like gravelliness, clay content or even
available water can be presented as contour maps (Stein &Corsten 1991), or, more
simply, they can beplotted against elevation (seealsoLescure &Boulet 1985). Lieberman et al. (1985) plotted the ordination scores of contiguous sampleplots against
elevation.
Forest gradients along slopes
In a landscape, there isa slight scale-dependent difference between elevation beltsand
slopepositions. Thealtitudeof thevalley bottom decreases gradually downstream because
of the river gradient, e.g. 5 mper km at theTai study site (Casenave et al. 1980). For
instance, at a certain place the valley bottom may be included in the 150-160 melevation
belt, but some kilometres downstream the same swamp forests and soilsare found in the
130-140 melevation belt, whereas at that location middle slope soils and forests are found
in the 150-160melevation belt. If theriver gradient is not too steep and the study siteis
reasonably compact, the elevational gradient modelcan beconsidered toapproximate the
toposequential model.The continuous elevation variable has considerable computational
advantages to thediscontinuous slopeclasses. Theelevation scale is an *interval scale,
i.e. differences can be computed quantitatively between any two sites by taking the
difference in elevation. This is not possible with the ""ordinal scale of slopeposition.
3.1.3 How to sample the large tree species
The soil can be described at a given point by taking a soil core or digging a soil pit. To
sample the tree species composition "at that point", a plot around it must be inventoried
and the tree species listed and their abundance estimated. The size and shape of suchplots
differ in each study, as does the subset of life forms studied.
The size of such a plot depends on the life forms considered. The fewer individuals
available per unit area, the larger theplot needs tobe. Large trees (d>70cm) occur at
densities of 10to 20 trees per ha. With a species pool of 50 tree species that can attain
such large diameters, the plot must beat least a few hectares large to obtain a reliable
species list. For maximal homogeneity, the shape of theplot should be close to circular if
no spatial trend is discernible in species composition. However, such a trend isexpected
along the slope, so theplot must be laid out as much as possible on the same slope
position, or, by approximation, within the same elevation belt.
Here, I propose the *contour sampling technique which fulfils three requirements:
1. the samples are large enough to include numerous big trees;
2. each sample is as close as possible to the samecontour line;
3. the samples occupy equal areas.
Grouping trees according to soil types would havebeen an alternative, but then the
samples would have covered areas very different in size which are not directly comparablein terms of species richness, and both species composition and tree density would
have been known with unequal precision in all samples.
Thevegetation samples obtained with this technique were used to analyse thelocal forest
gradients and to answer the following questions:
1. Can thelocal forest gradient beaccurately described by ordinating these contour
samples?
2. Do these local gradients indeed reflect the same change in species composition as the
regional gradient?
3. Do the densities, or biomass, or species richness of large trees display trends linked to
elevation and if so, how do both trends compare?
59
60
Forests of TaiNational Park
3.2 Description of the study sites
3.2.1 Location
Three study plots of 23, 25and 22 ha respectively were laid out near thevillages Zagn6,
Ta'iand Para (see Figure 28), and within theborder of Tai National Park in southwest
Coted'lvoire. Ta'iNational Park (446000ha) iscovered by *tropical rain forests within
the stream basin of Cavalla river (Text box 3). In these sample plots there was no
evidence of former logging activities. These forests belong to thelast "virgin" forests of
Cote d'lvoire (seeText box 3).
Table 4 Geographical position and area (ha) of the soil and tree surveys at the three study sites
Study site
Geographical
coordinates
Area of soil
survey (ha)
Area (ha) surveyed for trees with d >
70 cm
50 cm
30 cm
Zagne
6°07' N and 7°24* W
145
23
10
5
Tai
5°53' N and 7°20' W
133
25
13
9
Para
5°28' N and 7°10' W
72
22
10
6
350
70
33
20
All sites
Table 4 gives thegeographical position of the three plots, the area of soil surveys and
area of tree surveys. The study site near Ta'i was selected in 1981by Vooren (1985,
1986) for a study of catena effects on tree and branch fall. I selected the other sitesin
1989, one as far as possible from Ta'iin the northern direction but within Tai National
Park, towards a drier climate, and one as far as possible to the south, under a wetter
climate. The part of Ta'iNational Park south of Hana river was logged before 1973, sono
sites were selected there.
The distance between the Tai and the Zagne sampleplots is25 km; between the Taiand
the Para plots 55 km. For the soil surveys the entire toposequence was considered, but
tree surveys were mainly restricted to upper catena positions (see also Figure 38,
Figure 40 and Figure 42). Figure 29 to Figure 31display theposition of the survey areas
in thelandscape. They are centred on water divides and delimited by water courses. The
access road or track is also shown.
61
Forest gradients along slopes
GULF OFGUINEA
p.v.
°
fl^H
Permanent- forest estate
Y///\
TaV National Park
|„-_
Figure 28
7° W.L.
LEGEND
j Industrial plantations
\~~~ surfaced, laterite, local roads
f ^ 2 ^ rivers
[ 5 2 5 9 artificial lakes, sea
Map of TaVNational Park and surrounding forest reserves. Asterisks: three study sites close to
Zagne, TaVand Para villages. Sources: SODEFOR map (1975) of the "Domaine forestier", MPEA
(1983), LANDSAT MSS image from 1986.
62
Forests of Tai National Park
6°05'--
7°30'
7°25'W.L
LEGEND
|
| permanent forest estate
Tai* National Park
agricultural plots near the forest
Figure 2 9
^ ^ . _ . road
-£&? permanent and
temporary water
courses
P
car park and track
to the study plot
elevation point (in m asl)
Topographical map of Zagne survey area in its landscape. Sources: 1:50 0 0 0 topographical map by IGN
( 1 9 6 6 ) , sheet Guiglo 2a; SPOT multispectral satellite image dd. 1 4 / 2 / 8 8 , scene 4 5 - 3 3 8 .
63
Forest gradients along slopes
s
c
a
•3
CD
O
c
3
H
'3 <u
f- e
•'-"= -US.-?:
•s§
8
SS
_,«
..<t-ita^
U\l
2 -a
S? «
f2.e
64
Forests of TaiNational Park
ff
Forest gradients along slopes
Figure 32
All limits around Tai'National Park have been marked by signs. Photo by M.P.E. Parren
3.2.2 Climate
All three sites lie on a rainfall gradient with decreasing rainfall from south-west tonortheast (see Chapter 2). Mean rainfall per decade for the sampleplots Zagne, Tai and Para
can be estimated at 17, 19and 20 m(10 y) 1 respectively. Rainfall distribution goes from
clearly bimodalin the south to unimodalin the north (see Figure 7). Typically, diurnal
temperature variation is greater than seasonal variation (Eldin 1971), and hardly any
difference in average temperature can be found between the climatic stations. But the
difference between theextreme temperatures does increase from south to north. The
lowest absolute temperatures are recorded in theearly mornings at the beginning of the
dry season when the "Harmattan" blows (end December-January). The highest temperatures occur after midday in full dry season (February-March; ASECNA 1979).
65
66
Text box 3
Forests of Tai National Park
Facts and figures about Tai National Park
On the first maps of south-west Cote d'lvoire the Tai Forest is marked as "No man's land" (Boiiys 1933,
ex de Rouw 1991). Chevalier (1909) visited the region in the early 1900s and described the interfluvium
between Sassandra and Cavalla rivers as covered by "vast, primary forests".
1926:
1956:
1972:
1973:
1975:
1977:
1978:
1978:
1981:
1982:
1983:
the colonial administration created the "Pare refuge de la region forestiere" (960 000 ha) with the
status of a Forest reserve and a Fauna reserve (Bousquet 1978)
it was renamed "Reserve de Faune de Tai" and "Foret classee de Tai" and its limits modified
(431 000 ha; DRC 1967a), including the present N'zo Fauna Reserve.
transformation into Tai National Park (350 000 ha; Bousquet 1978; Decret 72-544, 28 August
1972); creation of the "Reserve partielle du N'zo" (73 000 ha) in the north, where timber mining
was allowed.
A 5 km band (20 000 ha) along the northern border of Tai' National Park was transferred to the
Reserve du N'zo and hence to timber mining (Decret no. 73-132, 21 March 1973). Timber
concessions ("chantiers") cover 5 by 5 km in Cote d'lvoire. The remaining Park now covers
330 000 ha.
enlargement of the Foret Classee de la Haute Dodo, south of Tai National Park, so that it extends
up to the Park.
creation of a "Zone peripherique de protection" with the status of "reserve partielle de faune"
around Tai National Park (Decret no. 77-348 of 3 June 1977). This zone covered 66 000 ha. No
protection zone was foreseen in the north where the Reserve du N'zo assured protection, or in the
south-east where a large pulpwood plantation was projected (Perimetre papetier, now Foret
classee de Rapide Grah).
registration of Tai National Park by MAB-UNESCO as Biosphere Reserve, i.e. an inviolate
pristine core surrounded by a buffer zone of forest managed for sustainable production (Whitmore
1990). The protection zone around Tai' National Park has been created in accordance with the
World Conservation Strategy concept of sustainable utilization with the maintenance of full
diversity and species richness: man living in balance with nature (Man And Biosphere).
construction of the "Station ecologique de Tai" (see Figure 30), a research fieldwork station near
Tai village, under the aegis of IET (Institut d'Ecologie Tropicale, office at Abidjan)
start of the inundation of Buyo lake behind a barrage on Sassandra river. The northern part of the
Reserve du N'zo and the southern part of the Foret Classee de Duekoue were transformed into a
periodically flooded grassy plain with dead standing trees.
Tai National Park added to the World Heritage List of UNESCO, i.e. qualified as an area of
acknowledged universal value (Whitmore 1990).
creation of a "Zone de protection" between Tai National Park and this Perimetre papetier, (ArrSte
ministeriel no 09 of 11 May 1983).
The area figures above are taken from the decrees. E. Schmidt (WAU) digitized the map of Tai National
Park prepared by the National Parks Service, Abidjan (1989) and obtained the following area estimates.
Tai National Park consists of a core area of 320 000 ha surrounded by a protection zone of 126 000 ha. In
total, 446 000 ha are well protected. Adjacent to the Park are the Reserve du N'zo (79 500 ha) in the
north, and the Foret Classee de la Haute Dodo in the south (76 000 ha of which are still covered by forest;
Marchesi et al. 1990). Thus, on the interfluvium between Sassandra and Cavalla river a continuous area of
about 600 000 ha of forest is left, which is about one-third of Cote d'lvoire's remaining rain forests. At
the beginning of the 20th century there were still 14 500 000 ha of rain forest in Cote d'lvoire (Gornitz &
NASA 1985). Some 200 000 ha of this elongated area were never logged but almost certainly no
"unpoached" forest remains.
3.2.3 Lithology
The Zagne and Tai sample plots are underlain by migmatites, and the Para plot by sericite
schists (Papon 1973). Figure 4 shows that granite outcrops occur in considerable parts of
the Park. Along the track towards the Para plot such forests on granite were encountered
close to the river Bono. Mount Nienokoue (Figure 28) isalso agranitic inselberg rising
Forest gradients along slopes
Figure 33
The field team consisting of soil and vegetation scientist.1. crowing the Bono river to reach the Para
study site. Photo by M.P.E. Parren
300 mabove the surrounding forests. These forests on granite contained more species
typical of drier forests (e.g. Triplochiton scleroxylon, Terminalia superba, Alstonia
boonei,Chlorophora regia) than sampleplots on schists like "Para" (personal observation). Guillaumet (1967) also found these species in old secondary forests near Tabou ina
wetter climate. This was to beexpected according to *Budowski's rule (1965; see p.53).
In this case, both soil and climate determine site hospitality.
3.2.4 Relief
Thebasement complex of Pre-Cambrian rocks in West Africa has a long erosional history
(van Herwaarden 1991a). The continent has been worn down to an almost flat surface or
*peneplain at least twice (Ahn 1970). The younger of these is associated with iron pans,
remnants of which were found on the hill topsat thePara and Taiplots. At the Zagne site
no continuous pan was found but only someironstone boulders. This could be the reason
why thedifference in altitude between valley bottom and crest is less than 20 min the
Zagne survey area, whereas it is40 to 50 min the Tai'and Para survey areas. The relief
is undulating in Zagne and rolling in theTai and Para plots (van Herwaarden 1991a).
Preparation of thedigitalterrain model. General topographical data could be read from
the 1:50 000 topographical maps (IGN 1965), based on 1:50 000 aerial photocover from
1956-1957. From both sources I selected an interfluvium large enough for mypurposes
(20to 50 ha). A straight access track was cut through the forest and along this track I
assessed traces of logging and searched for the highest point in the landscape. From this
67
68
Forests of Tai'National Park
point a base line was cut down the slope into the valleys on either side of the crest.
Starting from this linea grid was laid out, resulting in twenty to twenty-five square one
haplots, but avoiding lower slopeand valley bottom forest.
Theinclination of the slope at 25 to 50 mintervals along all lines was measured using a
Silva clinometer. These data were converted to altitudinal differences and adjusted for
discrepancies using the spreadsheet package Supercalc (**®1991 Computer
Associates). Irregular terrain characteristics like water courses, ravines and theedgeof
ironstone capping were mapped separately. The hypsometricaldata (x,y and z
coordinates) were transferred to the Surfer package (®1989 Golden software) toproduce
detailed elevation contourmaps.Relative elevation in metres abovethe lowest valley
bottompoint was transformed intoabsolute elevation in metres above sea level by adding
the estimated absolute elevation of thevalley bottom. The latter was deduced from
elevation points on the topographical maps at scale 1:50 000. However, along the road to
theIET research station (the field station of the Institut d'Ecologie Tropicale from
Abidjan; see Introduction) south of Tai, a topographical survey (Casenave et al. 1980)
had shown that the elevation points indicating hill summits on those topographical maps
include the 30 to 40 mhigh forest vegetation, whereas points along roads andrivers
correspond to ground level. This resulted in an exaggerated relief in areas on themap
where both types of measures coincide.
Preliminary interpretation maps showing possible physiographic units for use during the
fieldwork of the soil survey were madeon the basis of these maps. Detailed soil maps
were subsequently drawn at a scale of 1:5000and cross-sections were prepared to
visualize the relation between soil layers and slopeposition, as will be explained in
subsection 3.2.6. The digital terrain models of the sites allowed the calculation of thez
coordinate of theposition of each tree. This z coordinate was used to group trees into
contour samples, see subsection 3.3.3.
3.2.5 Hydrology
The Zagne survey area belongs to the catchment area of Nse river, the Tai area lies in the
Audrenisrou basin and the Para area in the Hana basin. All three rivers are first order
tributaries of Cavalla river, which forms theborder between Cote d'lvoire and Liberia.
On the higher convex geomorphological units of the survey areas (crests, shoulders, upper
and middle slopes) drainage ispredominantly vertical (Fritsch 1980;van Herwaarden
1991a) and soils are well drained in terms of the FAO (1977) classification. On thelower
slopes there isan important lateral drainage component and soils are moderately well
drained. Gullies occur downslope and become larger as they descend. At a certain point,
the water which flows after heavy rains may form a waterfall when it breaks through a
resistant petroferric layer present in the subsoil on lower slopes. Under the waterfall a
permanent pool maybe excavated (Figure 34). From thispoint on a deep ravine is
formed with 2 to 5 mhigh walls. Soils in these concave ravines are imperfectly drained.
The wettest parts are the stream valleys where ponding occurs and the water tableis often
**The full reference can be found in the Software list at the end of this book.
Forest gradients along slopes
69
near the surface. This is morepronounced (very poorly drained soils) downstream than
upstream (poorly drained soils).
3.2.6 Soils
Aphysiographic soil survey was carried out on each siteby van Herwaarden (1991a),
additional work having been doneby Rademacher (1992) and Nooren (1992). The
Tropenbos guidelines for a common methodology for inventory and evaluation of tropical
forest land by Touber et al. (1989) were used for the survey. Theapproach, followed for
the soil survey, relied much on thephysiography of the terrain. For that purpose a
detailed relief map was made first (see subsection 3.2.4; Figure 39, Figure 41and
Figure 43). Mapping units were named after, and mainly follow the physiographic
pattern. More details about the survey procedure and a description of thephysiographic
units on each site are given in Appendix 3.
Pisolithic ironstone gravel is abundant in upper slopepositions and is derived from the
physical disintegration of the ironstone sheets (Figure 35) and subsequent partial transportation of the ironstone fragments (Perraud 1971,deRouw et al. 1990). Its content and the
range in depth in which it occurs, changegradually on upper and middle slopes. In the
subsoil on lower slopepositions, the formation of plinthite ischaracteristic, i.e. an ironrich, humus-poor mixture of clay with quartz commonly occurring as dark red mottlesin
.>>-..
•:wt .. - •£
§>IP'
Figure 34
Erosion gully on the lower slope in the Tai study area. At this point the water breaks though a
resistant petroferric layer. A permanent pool has been excavated under the waterfall. Photo by
A.P. Vooren
.*.,..,.
70
Figure 35
Forests of Tai National Park
Ironstone boulders that broke off an iron hardpan, here on the beach near Fresco. Exactly the
same kind of boulders are found just below the crests with iron pan at the Tai site. They disintegrate into ironstone gravel.
a pale yellow matrix. Both the depth and the consistence of theplinthite layer on lower
slopes vary along theslope.
Drainage isalso clearly related to topography (Lescure &Boulet 1985,Fritsch 1992).
The increasing moisture availability downslope, both from lateral drainage and from finer
texture, may be a major factor influencing tree species composition along a catena.
Rooting conditions in the soil are also linked to catena position. An iron pan occurs at
shallow depth in Tai and Para, and apetroferric layer (FAO 1988), also called petroplinthite, is found at the three sites, typically on lower slope positions. Apermanent
water table in the valley bottom soils may also form a root floor.
Figure 38, Figure 40and Figure 42 show thephysiographic soil maps for each site. The
map units, documented in the legends, correspond tophysiographic positions. The soils
on a given physiographic position may differ from one site to another. The typical crests
with iron pan are only found in Tai and Para, not in Zagne. Lower crests without iron
pan are described in Para. These also exist in the Tailandscape (Fritsch 1980).
Forest gradients along slopes
3.3 Methodsof data collectionand analysis
3.3.1 Exclusion of swamp forests
Swamp forests and upland forests are very different in species composition, physiognomy
and ecology (Guillaumet &Adjanohoun 1971,Hall &Swaine 1981,Lieberman et al.
1985), theformer with a variablebut permanent water table whereas thelatter are rainfed and so are subject to climatic drought stress. The swamp forests in TaliNational Park
contain specialist tree species likeMitragyna ciliata,Uapacapaludosaand Gilbertiodendron splendidum as well as tree species which are often abundant in swamps but can
also occur in higher catena positions, e.g. Heritiera utilis,Sacoglottis gabonensis and
Gilbertiodendronpreussii(Huttel 1977,Bech 1983,deRouw 1991)
Although theinventory reports used in Chapter 2pooled information from swampand
upland forests, I wish to exclude the swamp forest from the analysis in this Chapter, in
order to obtain a clearer picture of local variation of forests along slopes superimposed on
theeffect of the rainfall gradient. I excluded swamp forest when establishing my sample
plots in Zagne and Para. In Tai, thelowermost ha contained some swampand was
excluded when delimiting thecontour sampleplots.
Swamp specialist trees will not be discussed further, but the facultative swamp species
will be taken into account to some extent. Thegreat abundance of these species does
indeed influence the neighbouring lower slope forests on better drained land, and the
swamp-lower slopeboundary is once again a fuzzy forest limit. I thus expect the slope
gradient to be complicated by a "swamp effect", especially on lower slopes.
3.3.2 Tree recording in nested plots
Largetrees. Trees with a diameter exceeding 70 cm were identified in theentire sample
plot (see Table 4), were mapped and had their diameter measured as precisely as
possible. Aglass-fibre tapeand if necessary a ladder were used to measure thegirth of
the tree. If buttresses exceeded the height of the ladder, thediameter was measured with a
2 mruler as described by Cailliez and Alder (1980).The height at which themeasurement was taken, was also noted. Tree numbers were painted in whiteon the bark together
with a ring at the height of measurement.
Smallertrees. These were sampled in half (diameter 50 to 70 cm) and a quarter
(diameter 30to 50cm) of theentire 20ha sampleplot (seenested plotsin Figure38,
Figure 40and Figure 42). At the Para site, the hectare plots to be subsampled for smaller
trees were chosen in such a way that all slopepositions were represented but ina
contiguous layout for later studies of species populations (deKlerk 1991)and forest
dynamics (Jans et al. 1993in press, Poorter et al. 1993in prep). In Tai, existing sample
plots were included and in Zagne, theplot was consecutively enlarged from 5 to 10and
to 23 ha, raising the lower diameter limit by 20cm each time. This layout in Zagne
showed tobe not optimal for theanalysis, as will beexplained in 3.3.4.
71
72
Forests of Tai National Park
I
1
a
% " <%*
*-Iff'
'*
.>
I.. >?w
*: .- - •' - . I f
•%*
Figure 36
The diameter of large trees was measured with a 2 m ruler as described by Cailliez and Alder
(1980). On the photograph a Sacoglottis gabonensis of about 70 cm diameter is being measured in
the Para plot.
Speciesidentification. Trees were identified using the characteristics of theirleaves,
fruits and trunks. Most of the identification work was doneby theIvorian botanists Pierre
Poleand Henri Tehe using Aubreville (1959). The former had collaborated with Vooren
(1985) and de Rouw (1991). DeRouw deposited many specimens in the herbarium at
Wageningen (WAG). The latter had worked for many years for ORSTOM in Abidjan and
contributed largely to the ORSTOM herbarium (ABI) which contains duplicates from
WAG and was recently transferred to the Centre National de Floristique (UCJ) in
Abidjan. At the Tai site I continued working in the sampleplot established by A.P.
Vooren (1985), where trees had been identified by Vooren (Wageningen Agricultural
University) and C. de Namur (ORSTOM).
The nomenclature in thepresent book follows Hall &Swaine (1981), Voorhoeve (1965),
Aubreville (1959) or Hutchinson &Dalziel (1954-1972), in that order (i.e. if a species
was not in the most recent publication, the next older one was consulted). In Appendix I
Forest gradients along slopes
the list of tree species found in the three sampleplots, isgiven, including authors,
sources and synonyms of species according to thefour above references.
3.3.3 Definition of the contoursamples
Samples for *ordination analysis must be homogeneous and complete with regard to
species composition of the life forms considered. I define *contour samples, intentionally
with low precision, as "samples along the same contour line". Such samples should be
sufficiently large to contain enough trees, and that the sampleplots should beof the same
area. Amathematical line of course does not fulfil these conditions. Hence, I proceeded
as follows to obtain belt-like contour sampleplots of 2 ha each. Thedigital terrain model
that resulted from the topographical survey (see subsection 3.2.4) was used to interpolate
theelevation of each 10by 10msquare in the sample plot grid. Trees were assigned the
z coordinate of the square they were standing in. The ZCOORD programme was written
in Turbo Pascal (®1990Borland) by JJ. Stoorvogel toperform thecalculations.
The squares were arranged along descending z coordinate and were then grouped into
2 ha contour sample plots (Figure 39 to Figure 43) by taking.thefirst200cells, the
second 200, and so on. The contour lines which delimited thebelt-like sampleplotscorresponded to the z coordinate of the 200th, 400th, etc. square (see Figure 39 to Figure43).
These levels were used to take the trees together into contour samples. Notice that the
contour levels separating these samples are not equidistant. In Figure 39to Figure 43the
Figure 37
A Canarium schweinfurthii tree of 130 cm diameter is being measured over the white paint ring.
Left, tree spotter Pierre Pole, right Dr N.R. de Graaf (WAU). Photo by M.P.E. Parren
73
74
Forests of Tai National Park
Physiographic units:
D Zc Crest noironpan
Legend:
E3 Zs1
m Zs2
m Zs3
a Zv1
H Zv2
gradualtransition
clear boundary
sample plot boundary
>30 alltreeswith
diameter largerthan
30cm recorded
accesstrack
Upper slope
Middleslope
Lower slope
Gully/Ravine
Valley bottom
0 50 100m
Soil legend of the physiographic units of the Zagne survey area (Rademacher 1992,van Herwaarden 1991a)
Figure 38
Unit
Physiography
Drainage
Depth of
rotten rock/
petroplinthite
Altitude
in m asl
Parent
material
Slope
form
Slope
degree
Colour
Soil texture
FAOUNESCO
classification
Zc
Crest
well drained
> 150 cm
no iron pan
203-205
Migmatite
straight
to
convex
gently
sloping
strong
brown over
red
gravelly sandy
(clay) loam over
sandy clay
Ferric Acrisol
Zsl
Upper slope
well drained
> 150 cm
199-203
Migmatite
convex
gently
sloping
strong
brown
slightly gravelly
sandy (clay)
loam over
gravelly clay
Haplic
Ferralsol
Zs2
Middle slope
moderately
well drained
> 150 cm
190-201
Colluvium/ straight
Migmatite
gently
sloping
yellowish
brown
non-gravelly
sandy (clay)
loam over
gravelly clay
loam
Plinthic
Acrrsol
Zs3
Lower slope
imperfectly
drained
100-130 cm
rotten rock
187-197
Colluvium
concave
gently
sloping
yellowish
brown
non-gravelly
sandy clay loam
Haplic
Acrisol
Zvl
Ravine/gully
poorly
drained
0- 80 cm
petroplinthite
187-199
Alluvium/
Migmatite
concave
gently
sloping
yellowbrown to
flray
non-gravelly to
very gravelly
coarse sand to
sandy clay
Ferrelic
Cembisol
Zv2
Valley
bottom
very poorly
drained
100-150 cm
rotten rock
< 195
Alluvium/
Migmatite
concave
almost
flat
light gray to
greenish
gray
non-gravelly
coarse sand to
sandy clay
Dystric
Ruvisol
Physiographic map of the Zagne study area. Grey tones correspond to physiographic units. The
soil in these units is described in the soil legend. Straight lines delimit the sample plot and the
figures indicate the tree sampling intensity. From van Herwaarden (1991a), Rademacher (1992).
Forest gradients along slopes
Zagne
Legend:
•-> temporaryandpermanent
watercourses
3 contoursample plot
number
lowermostarea,only1 ha
Contour levels:
1
205.1 m
2
204.4
3
203.5
4
202.7
5
201.8
6
200.8
7
199.5
8
198.3
197.0
10I
195.8
11
193.9
12
Figure 39
Contour sample plots at the Zagne study site. Contour levels are not equidistant, but each contour
sample plot covers each 2 ha (except no. 12: 1ha), even when the plot is split up into five parts.
75
Forests of Tat National Park
76
Tai
Legend:
alltreeswithdiameter larger
than 30cm recorded
> 3 0
access track
Physiographic units:
D
Tc
•
Ts1
Crest with iron pan
Upper slope
II
Ts2
Middle slope
•
Ts3
Lower slopewith plinthite
S
Ts4
Lower slopewith petroplinthite
m Tv
Valley bottom
~ gradualtransition
~ clearboundary
— sampleplotboundary
0
50 100 m
Soillegend ofthephysiographic units ofthe Taisurvey area (Nooren 1992, van Herwaarden 1991a)
Unit
Physiography
Drainage
Depth iron pan
/ petroplinthite
Altitude
in m asl
Parent
material
Slope
(orm
Slope
degree
Colour
Soil texture
FAOUNESCO
classification
Tc
Crest
well drained
70 cm
iron pan
185-198
Migmatite
convex
riatto
gently
sloping
red
very gravelly clay
Ferric Acrisol
Tsl
Upper slope
well drained
> 150 cm
180-192
Migmatite
straight
to
convex
sloping
red
very gravelly sandy
clay loam over nongravelly clay
Ferric Acrisol
Ts2
Middle slope
well drained
> 150 cm
170-180
Migmetite
straight
gentty
sloping
strong
brownto
red
very gravelly sandy
clay loam overnongravelly clay
Ferric Acrisol
Ts3
Lower slope
moderately
well drained
> 150 cm
160-175
Colluvium
straight
gently
sloping
yellowish
brown
very gravelly sandy
loam over nongravelly clay
Plinthic
Acrisol
Ts4
Lower slope
moderately
well drained
70-90cm
petroplinthite
160-170
Colluvium
concave
gently
sloping
yellowish
brown
non-gravelly sandy
loam over very
gravelly clay
Xanthic
Ferralsol
Tv
Valley
bottom
poorly
drained
> 150 cm
< 160
Alluvium
concave
almost
flat
light gray
to brown
non-gravelly sandy
loam
Dystric
Gleysol
Figure 40
Physiographic map of the Tai study area. From Herwaarden (1991a), Nooren (1992).
11
Forest gradients along slopes
Tai
Legend:
- - > temporaryandpermanent
watercourses
3 contoursampleplotnumber
R^j excludedlowermostarea
Contourlevels:
1
192.3 m
2
189.6
3
186.9
4
5
184.6
182.5
6
7
8
180.9
179.3
177.2
10
11
174.2
170.6
166.2
12
160.9
Figure 41
Contour sample plots atthe Tai study site. Forexplanation seealso Figure39.
78
Forests ofTai National Park
Para ^ >
Legend:
> 30 alltreeswithdiameter larger
than30 cmrecorded
accesstrack
Physiographicunits:
D
Pel Highcrestwithironpan
D
Pc2 Lowcrestwithoutironpan
Q
Pc3 Shoulder withrottenrock
Q
Ps1 Upper slope
El Ps2 Middleslope
B
Ps3 Lower slope
B
Pv1 Gully/Ravine
H
Pv2 Highervalley bottom
•
Pv3 Lowervalley bottom
~
gradualtransition
~
clearboundary
— sampleplotboundary
0
50 100m
Soil legend of the physiographic units of the Para survey area (van Herwaarden 1 9 9 1 a )
Unit
Physiography
Drainage
Depth of
rotten rock
/iron pan
/petroplinthite
Altitude
in m asl
Parent
material
Slope
form
Slope
degree
Colour
Soil texture
FAOUNESCO
classification
Pel
High crest
well drained
125-150 cm
iron pan
143-155
Schist
convex
gently
sloping
red
(very) gravelly
clay loam over
clay
Haptic
Ferralsol
Pc2
Low crest
well drained
110-150cm
rotten rock
128-133
Schist
convex
gently
sloping
strong
brown
gravelly tovery
gravelly clay
loamtoclay
Haplic Acrisol
Pc3
Shoulder
well drained
10-100 cm
rotten rock
131-148
Schist
convex
gently
sloping
reddish
yellow
non-gravelly
clay loemto
clay
Ferraltc
Cambisol
Psl
Upper slope
well drained
125-150 cm
rotten rock
133-153
Colluvium/
Schist
straight
moderately
steep
yellowish
red
gravelly loam
over clay loam
Haplic
Ferralsol
Ps2
Middle slope
welt drained
> 150 cm
119-141
Colluvium
straight
to
convex
sloping
yellowish
brown
slightly gravelly
clay loam
Xanthic
Ferralsol
Ps3
Lower slope
moderately
well drained
70-150cm
(petro)plinthite
115-135
Colluvium
concave
sloping
brownish
yellow
slightly gravelly
clay loam over
clay
Plinthio
Ferralsol
Pvl
Gully/Ravine
imperfectly
drained
0-50 cm
rotten rock
115-127
Schist
concave
gently
sloping
variegated
slightly gravelly
clay loam
Farralic
Cambisol
Pv2
Higher valley
bottom
poorly
drained
80 cm
rotten rock
115-123
Alluvium/
Schist
concave
almost flat
gray over
variegated
nongravellyto
gravelly coarse
sandtoclay
Dystric
Fluvisol
Pv3
Lower valley
bottom
very poorly
drained
> 150 cm
< 115
Alluvium
concave
almost flat
gray
non gravelly
loamy sand
Dystric
Gleysol
Figure 42
Physiographic map ofthe Para study area. From van Herwaarden (1991a).
79
Forest gradients along slopes
Para
Legend:
->temporaryandpermanentwatar
courses
3numberofcontoursampleplot
Contour levels:
... i
-SI
Figure 43
V-
•* t>
Contour sample plots at the Para study site. For explanation see also Figure 39.
Forests of Tea National Park
80
stepped lines separating the squares of thecontour samples, havebeen smoothed into
curvilinear lines.
This method of aposteriori resampling of the trees has three advantages:
1. within a sample all trees grow in the same slopeposition;
2. the sampleplots all have the same size (here2 ha), so species richness and density
figures such asbasal area are comparable;
3. for ordination purposes 11or 12samples are available per site, which is morethan
the 4 soil types within the sampleplot.
Distribution of thephysiographic units overthe contoursamples. The contour samples
with a fixed area of two ha always span several physiographic units (Figure44).
However, soils along the slope form a continuum from onephysiographic unit to thenext
(see Figure 45). In Para, only the lower crest (Pc2) and the ravine (Pvl) unit havea wide
elevation range, because of their geomorphology. Lower crests do not have a protective
iron pan and are thus steadily lowered by erosion. Ravines pass through all elevations
starting from their origin on the middle slope. The digital terrain model with a 10by 10
mprecision does not take into account thelocally lower relief in the ravines, which is
permissible because these account for less than 2 %of the area of the sampleplots. For
Tai the lower slope soils (Ts4) show a disjunct distribution, by thehigher elevation of the
northern valley which lies upstream from the southern valley. Both are tributaries of
Audrenisrou river, which flows southwards. Thecontour lineof the source of a stream
Zagne contour samples
Physiographic unitsZagne (withtotal area)
lllllll
Illllll!
z1 -rTTTTTTI 111111111
z2 l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l :
| Zc (4.0ha) crestwithoutironpan
I Zsl (7.9ha) upperslope
zatmnnnmin
I Zsz(10.7ha) mWdteslope
IE
24
I Zs3 (0.2ha) lowerslope
25
I Zv1 (0.3ha) gully/ravine
z6
TotalJ
c 23.0ha
z7
z8
m
z9
z10|
MB
z11
z12|
ssiiilllll
~o!2 65 o!e oi i i^2 T5 i ^ Ts
wm
i
cumulative areaofthe physiographic unitswithinthecontour samples (ha)
Figure 44
Distribution of the physiographic units over the contour samples in Zagne, Tai and Para. In the
legend the total cover (in ha) of each physiographic unit within the entire sample plot is given. For
explanation of the codes of the units see the legends in Figure 38, Figure 40 and Figure 42.
81
Forest gradients along slopes
Taicontoursamples
PhysiographicunitsTai(withtotalarea)
tl
••••••••—'•••••••I |
t2
••'•'
| Tc (2.2ha) crestwithironpan
'•,', .1 £ ~ j Ttl (10.1ha) upperslops
t3
1
14
[""] Ts2 (4.1 ha) middleslope
' P g j Ts3 (5.0ha) lowerslopewith pBnthte
ts
H i T * * l 2 - 8 "*) lowerslopewithpetropinlriita
' i"ij ESSE
H
Tv (1.0ha) valleybottom(0.75haexcluded)
JB'l.DOOOjXJ
™™
1
"" , I V W Y > " Totalarea:25.0ha
•JHfflrgKaflgfl
"™
t6
t7 :
IS
WSSSt**:
t9 -L
no
til
m-wm
6.2 OA o!6 55
1 i5 T5 Te i5
2
cumulativeareaofthephysiographicunitswithinthecontoursamples(ha)
Paracontour samples
PhysiographicunitsPara(withtotalarea)
P1
El
P2
P3
nrn
P4
IIIIII
H i
p5
[ ~ J Pel (1.1ha)
highcrestwithiron pan
ITT] Pc2 (1.0ha)
lowcrestwithoutpan
|
[
a
[""J Psl (3.8ha)
•
[ " J Ps2(14.2ha)
_
J H
11
HUB
p9-{
si»l)lllll
p10
P11
OA 6S o5
i
rS iS i l i i 5
2
cumulative areaofthephysiographic unitswithinthecontour samples (ha)
Figure 44 (continued)
f | Pvl (0.3ha)
Totalarea:22.0ha
P8
^2
shoulderwithrottenrock
upperslope
middleslope
H J Ps3 (l.3ha) towerslope
P6
P7
1 Pc3 (0.4ha)
guty/ravine
82
Forests of Tax National Park
cuts through all slopepositions when it is followed downstream, therefore the fall of the
water courses is adrawback to *contour sampling.
The middle slope unit isby far the most common physiographic unit in thePara plot
(65 %)and the Zagne plot (50 %). In theTai plot the upper slope unit is the most
common (40 %). Three cross-sections are displayed in Figure 45 representing the typical
sequence of soil characteristics at thethree sites.
3.3.4 Ordination andcomparison of the contoursamples
An indirect approach was followed to relate species composition and other forest characteristics to soil. Both vegetation and soil were first related to topography, which is
invariable at the time scale used and can beprecisely measured. Therelation between soil
and elevation is shown in Figure 45. To analyse the response of thevegetation I plotted
the compositional gradient and other forest characteristics against elevation.
Species composition was determined for each contour sample. In a first run I only
considered the large trees (d>70 cm). Theabundance of a species was expressed on a
scale of 1to 9. Itcorresponds to thebasal area of the species in the2 ha contour plot,
rounded to the nearest integer. Species with basal area less than 1m2per 2 ha were given
thecode 1, those with basal area more than 9 m2per 2 ha the code 9.
The correspondence between these samples was analysed with Detrended Correspondence
Analysis (DCA, Hill &Gauch 1980). AsI found out that after thefirstrun thevegetation
table was rather incomplete with regard to thepresence or absence of certain species, I
decided to search the data set of the smaller trees (30<d<70cm) for trees belonging to
the 95 large tree species found in thefirstrun. Ifine-tunedthevegetation table, especially
for the species that only rarely attain 70cm diameter, but that are moreabundant in the
smaller diameters. If present below but not above 70 cm diameter, the abundance code 1
was given to the species, regardless of its basal area below 70cm. Thevegetation table
resulting from this second run is shown in Table5. Fine-tuning was not possible in all
plots, however. In the crest sampleplots zl to z3in Zagne smaller trees were not
recorded because recording was donebefore contour sampling was conceived, as can be
seen by superimposing Figure 39on Figure 38 (see3.3.2). Hence, these samples are not
fine-tuned and caution is needed when comparing them with the other samples in Zagne.
For the same reason the area of forest surveyed for smaller trees in the other contour
samples was variable, as contour sampleplots were designed for the entire 22 to 24ha
and not for the 10or 5 ha surveyed in more detail. In thepresent Chapter these aspects
are always taken into account.
The sample scores on thefirstDCA axis were plotted against theelevation midpointsof
the contour intervals of thesamples.
Tree species richness and population density of all trees over 70 cm diameter were
calculated per contour sample and plotted against elevation. Basal area and biomasswere
analysed in the same way. Biomass per contour sample was calculated by multiplying
basal area per species by the wood density of each species (see Appendix I for densities;
from Bolza &Keating 1972, Dudek et al. 1981,Durand 1985,Vivien et Faure 1985).
83
Forest gradients along slopes
Zagne
* » - ironpan
++*- - rottenrock
Tai
o.'c - plinthite
sst^.- petro-plinthite
HUH - colluvium
= - alluvium
40m
Para
500m
Figure 45
Three soil catenas in Tai'National Park (Zagne and Tai on migmatite; Para on sericite schist; from
van Herwaarden 1991a) indicating the origin of boundary lines (iron pans, alluvium) and other soil
characteristics whose importance changes gradually along the slope. For a description of the
physiographic units see Appendix III.
84
Forests of Tat National Park
3.4 Results:vegetation responseto slope position
3.4.1 Largetree species composition
In 35contour sampling plots covering a total of 69ha of upland forest, 894 trees were
found to havea diameter exceeding 70cm. They belonged to95 species. The lowermost
contour samplein Zagne (zl2) covered only 1ha, soit was only accessorily considered
when comparing species composition. The ordination results from the second run are
given in Table5. For this run the abundance scores of the species were fine-tuned by
including occurrence information from 1474 smaller trees, i.e. with a diameter between
30and 70cm. Downweightingof rare species was applied as an option within DCA. The
first axis had an eigenvalue X of 0.55 and separated the three study sites. The other axes
had eigenvalues of 0.13, 0.10 and 0.07, mixed up the sites and, at first sight, could not
be related to any spatial trend.
Figure 46 shows the relation between theordination of the samples along DCA1and the
altitude of each sample in mabove sea level. From Zagne towards Para altitude decreases
and rainfall increases. Para is closer to theLiberian coast than Zagne. The catenas in
Zagne and Tai overlap in absolute altitude. Tai'and Para are disjunct with respect to
altitude. Soilconditions show pronounced differences within each catena (see Figure45),
but in general, more moisture isavailable in the lower slopepositions.
Sample ranking. The first DCA axis ranked the study sites in the same order as in
Chapter 2, along the regional gradient. In Chapter 2 it was shown that the changein
species composition between the sites correlates with an increase in rainfall. There is
considerable variation within the sites. The Tai samples show the clearest trend within the
site.
The compositional changein the downslope directionis similartothe changeof
forest composition which occursin directions towards areaswithwetter climate.
In other words, *compensation occurs between theecological factors "climate" and
"catena position" (see also de Rouw 1991).This compensation isalso found at Zagne
from z5 to zl2 and in Para from pi to plO. The samples zl to z4 and pll do not follow
this trend in Figure46.
Separate ordination oneach site. Theoutlying position of zl, thecrest forest at Zagne,
becameeven morepronounced (Figure 47). As mentioned in the methods section (3.3.4)
the samples zl to z3 were notfine-tuned,so their composition is lessprecisely known
than that of the other samples. When considering all samples in Zagne, no correlation was
found with altitude (R= 0.06, p=0.85), when zl was omitted the correlation became
significant (R= -0.77, p=0.006). Statistically significant correlations were also found for
Tai (R= -0.81, p=0.002) and Para (R= -0.82, p=0.002). Notice how DCA inverted the
gradient's direction for thePara site, wet having a low score, dry a high one.
85
Forest gradients along slopes
Table 5 DCA ordination table of95 large (d>70 cm) trees species in35 contour samples, each covering
2 ha. Read plot numbers, DCA1 scores and number ofspecies vertically.
Large tree species
Sacoglottfs gaboneasls . .
Protormgabariastaptlana
Anthocletstanobltis
Maranthes glabra
Pentadesmabutyracea ..
AfzeHa belta
Uapaca gulneemls
Uapaca coro/sferf
Antlarlstoxlcarfawel. wet.
Scytopetalumtteghemll ..
HermerauttHs
Dacryodes klalneana . . . .
CoulaeduSs
Canarlumschwelnturthll..
DtaSum aubrevlllel
Nauclea dldorrlchl
Tteghemella heckelll
Amphlntaspterocarpolaes
Trlchoscyphaarborea
trvlnglagabononsls
Rhodognaphalonbrevhuspa
Padnarlexcelsa
Parkla tricolor
Pentactethramacrophylla
LovoatrlchlHoldes
Datarlumsenegaktnse
Samanea dlnklagel
Entandrophragma angolense
Strombosla glauoascens
Newtontaduparquetiana
Klalnedoxagabonansls
Pachypodanthlum staudtll
Copalterasallkounda
Spathodea campanulata
Scottelllaklalneana
Maranthes aubravlltel
Reus sop
Placodlscus boya
Bellschmledlamannll
Pycnanthus angolensls
TrlchlUa splendtda
Chrysophyllum tatense
Plptadentastrum afrlcanum
Rlclnodendron heudatotll
Discogtypremna cakjneura
TermlnaUa Ivorensis
Oldlieldia atricana
Anopyxlsklaineana
Calpocalyxaubrevlllel
Stereospermumacuminatlsslmum .
Corynanthepachyceras
Lannaawetwttscnli
Lophlraalata
Nesogordoniapapaverlfara
Gymnostemon zalzou
Aubragrlnlatalensls.
Entandrophragma candotlal
Aubmvllloa
a platycarpa
piaty
Anthonotha tragrans
Khayaanthotheca
Erythroxylum mannll
Antlarlstoxlcarta wel. air
Erythrophleum Ivorense
Petersfanthus macrocarpus .. .
Entandrophragmautile
Flcuselastlcotdes
Anlngerlarobusta
Entandrophragma cyHndrlcum
Guarea cedrata
Sterculta oblonga
Celbapentandra
Dubosclavlrldltlora
Canthlum arnotdlanum
Balaniteswltsonlana
Bombaxbuonopozense
Keayodendron brldetloides
Zanthoxylum gllletll
TermlnaUa superba
Alblzla terruglnea
Chldlowlasangulnea
Xyllaevansll
Alblzla zygla
Danlelllathurltera
Trlptochlton scteroxylon
Alstonlaboonet
Chtorophora regla
Dlstemonanthus benthamlanus.
Bllghlawelwltschli
Chtorophora excelsa
Saplumaubrevlllel
Gulbourtla ehle
Brldellagrandis
Alblzla glaberrima
Aubrevmeakerstlngll
Numberofspecies
Contour
sample
plots:
DCA1
score
38S
369
356
348
3Z5
316
308
305
295
294
287
285
273
272
271
267
256
253
248
247
245
243
242
241
201
194
191
177
173
169
166
165
156
154
149
148
143
137
128
126
1Z1
11 7
104
IOZ
lOl
98
98
97
91
91
82
78
77
76
73
67
59
58
54
51
51
47
36
35
26
23
22
20
-1
-1
-5
-15
-19
-21
-36
-38
-39
-40
-57
-60
-64
-69
-74
-74
-82
-87
-91
-92
-93
-97
-l O O
-l O l
-1 0 2
-1 3 3
-1 3 3
PPPPPPPPPP pttttttttttttzzzzzzzzzzzz
1
1111
1 1
1
8065792431
1120895427613223049711658
3332222222 2111111111
1 1 0 9 9 7 7 6 6 5 4 9 7 6 6 4 2 2 1 0 9 9 9 7 6 5 43 3 3 2 1 1
8227684731 1413053059941419264249840
99999 85451
11111 11111
1 1 1 1 1 1 1--1
2 - -12 11114 2 13 1 - 1 1 1-- - 1 - 1 1 -1
-1--12-112- 12111
1--1-1-11I1--1
11111
11-11 -1311
-1111
11-11 11111
--111111-1
1 -1111
11-1-2
11-12
- - 1 1 111
I I- - 1
13 21
1111
__ __
1111-11
- m u ll
1-1-111-1
1111-1]
112
11 - 1 1 1 - 111
-111-121
- -]
1 -- 1 --1
[111l i i m -
- 1 1 - 1 1 - 1 1- 1 1 1 - - --11 -111 3 1 - 1 1 1 1 - 1 1-- 1
-- 1
-111-1--111
2 --11-
111-1
1 2 - 1 - 1 1 -1-- 1 1 1 1 - 1 1 1 1 L - 1 1 11 1 1 1 1 1 - 1 1 1 1 1 1 1 1 1 1 1 1 1
- 1 - 1 1 - 2 1 1 - 11-1--1
1 1 1 1 - 11-
i u m1__-
-1-1--il-
-11111111111111
l-i
!_
11-1 1 - - -1- 1121 191-
li-2
1122
11
3 1211 1111
- 1 2 1 2 - 3 3 4 11 1 2 1 1
-111-2211-11114-1
-111
1
1
1
1--1--1
11-11-1
11-12
S
-1-111-11-
1--11
•11111111-11
-111 2 3 - 3 - 1-1
1 - 1 1 - - 11
11111 13211111 11111111111
1-111111--1
-1-12
-1-1
-1-
-2111-11- 1111-11--11
1-221-1-1-1111 --11111-1-1
1
1-1
121133 1515242-56-1111-1
1111 1-11-11- 21121112-11
Ill
13--2241
-11-1111-1
1 1-1
-1-2- --217-
1-- 1 1-1 1-1-1
2 1-
-1 1 - 4 1 - 1 1
1-1
-11
-1111
-2111 111234
61698
2_ 7
31-111
21574
11111
_1
-352
-11
1
11--1 - 21-153
-11-1-1-
1 1 1 1 1 1 2 1 2 2 1 2 2 3 2 2 3 2 2 2 2 2 3 1 12 3 2 2 2 2 1 2 2 2
29875744 43186 646135
5 4 4 8 6 359 7 0 8 2 9 1 3 2
86
Forests of Tai National Park
elevation (m)
200zl
Z6
af„«
v
^
7
180-
^k.
160-
t12
-
\
140-
120
-50
Figure 46
i
I
I
I
0
50
100
150
I
200
250
Axis 1 score
^P2
i
i
300
350
Relation between altitude and the DCA1 scores of the contour samples. The point label indicates
the site (z=Zagne, t=Tai, p=Para) and position of the contour sample plot, " 1 " is the highest,
"11" or "12" the lowest plot. Elevation in m above sea level.
By comparing theoverall ordination (Figure 46) with the separate ones (Figure 47)an
"attractive" or "repulsive" effect of the presence of other sites on the within-site-ranking
maybeobserved; zl is skewed left by theTai samples, tl2 by thePara samples; pll
seems to be "attracted" by theTai samples. All this results from the fact that sample
scores are derived from the same species ordination shown in Figure 46. In the separate
ordination the information of the other sites is not used to rank the species, so certain
species may end up in a central, undetermined position, although they can beprecisely
positioned with thehelpof information from theother sites. When using correspondence
analysis the composition of thedata set seems to influence the results of the ordination.
Nevertheless, I conclude that thecatena, although a distinct cause, has a similar effect on
tree species composition as climatic variation. The length of a catena gradient (e.g. Tai:
100u.DCA1-Chapter 3) corresponds to apiece of thegradient induced by climateabout
46 u.DCAl-Chapter 2 long. The latter figure isan estimate derived from theDCA1Chapter 2 scores of the three sites: Zagne 35,Tai 64and Para 148.From Figure 46we
read that thedriest contour samples lie at two-fifths of the distance from the *centroid of
Tai (2/5 of (64-35=29) is 12u.DCAl-Chapter 2) towards the centroid of Zagn<£,and the
wettest samples lieat two-fifths from the centroid of Tai (2/5 of (148-64=84) is34
u.DCAl-Chapter 2) towards thecentroid of Para, bringing the length of thegradient to
46 u.DCAl-Chapter 2. On the regional gradient the rate of compositional change is5
u.DCAl-Chapter 2per 10km, sothisdistance (seeFigure 11)is some90 km longand
87
Forest gradients along slopes
Zagne
elevation (m)
B ri.qra
205-
21
24
Z5fc]
131
130
125
125
125
118
118
78
78
78
64
50
41
41
33
29
26
23
23
18
9
9
6
2
-4
-12
-14
451
421
392
388
388
324
324
306
293
287
281
280
263
247
243
216
213
199
199
196
191
Her Uti
Dia aub
Dub vir
Pen but
Cou edu
Scy t i e
Tri a r b
Ant t o a
Tri scl
FicUK E D
Chr tai
Nau did
Lov t r i
Kla gab
Pet mac
Old afr
Cor pac
Ery ivo
Str gla
Pip afr
Pyc ang
Ent ang
Bal wil
Pac sta
Hes
Bel man
189
186
181
181
177
175
172
165
153
145
139
135
108
105
103
102
90
79
75
68
64
58
39
35
20
17
Die
Ste
Ant
Cop
Kea
Fie
Ter
Chi
Pla
Ery
Kha
Lan
Ani
Ent
Amp
Bom
Gym
Aub
Zan
Aub
Cei
Tie
Ent
Ent
Gua
cal
acu
fra
sal
bri
ela
sup
san
boy
man
ant
wel
rob
uti
pte
buo
zai
pla
gil
tai
pen
hec
can
cyl
ced
17
15
13
6
-9
-io
-11
-21
-31
-45
-54
-56
-67
-70
-91
-95
-99
-108
-108
-109
-112
-175
-191
-191
-191
445
369
331
290
286
277
273
261
261
261
251
243
Tri.arb
Dap.cor
Ent.ang
Pyc.ang
Sco.kla
Her.uti
Mar.gla
Par.bic
Tie.hec
Amp.pte
Can.sch
Str.gla
Dia.aub
Uap.gui
Cou.edu
Scy.tie
Pro.sta
233
223
216
215
198
187
182
177
142
141
141
140
123
122
108
lOl
95
Rho.bre
Dis.cal
Afz.bel
Ant.nob
Ant.mac
Sac.gab
Nau.did
Irv.gab
Cal.aub
Ter.ivo
Ery.man
Par.exc
E r y . iv o
Pip.afr
Chr.tai
Ant.fra
-47
-54
-112
-127
-175
-205
-303
459
459
459
328
324
270
C h i .r e g
27 """•»»
G y m .zai
.•""iS
195-
Cor.pae
str.gla
Pyc.anq
\
112
i
I
50
AJb.gla
I
I
-14
-17
Her
gil
ivo
sup
wil
wel
sen
vir
tri
aub
arn
obi
toa
san
pte
ela
afr
cal
ala
.bri
.afr
.utl
.gab
.fra
.nan
235
200-
X Y 1 sva
Ale boo
Tri eel
Nau
Tri epl
Uap gui
Pac
Par exc
Ery i v o
Can sen
Lan wel
Alb zyg
Par
Ric heu
Rho bre
Ani rob
E n t .can
B o m ,t>uo
Dan
K l a .gab
Afz
A l b .fer
C o u .edu
Tor
Ter.
Bal
Bli
Det
Dub
LOV
Cal
Can
Ste
Ant
Chi
Anp
pic
Pip
Die
Lop
Kea
Old
Ent
Irv
Ant
Ery
Zagne
227
197
197
193
170
164
164
I
100
150
DCA Axis 1 score
TaT
200
elevation (m)
Tai
Al=0.28
A2=0.14
72 species
Lop ala
Ste obi
Ant tow
Mar gla
Uap gui
Ant nob
Pen mac
Ric heu
New dup
Trv g a b
Hap cor
Ant mac
Sam din
Rho bre
Can arn
Spa c a m
Mar aub
Par exc
SCO K3a
Par bic
Dae kla
50
-50
100
150
DCA Axis 1 score
200
250
Para
160
elevation (m)
<ra
1=0,26
t2=0.15
45 s p e c i
Lan . w e 1
Kla.gab
Cor.pac
Ano.kla
Pet.mac
Pac.sta
Pen.but
Ant.tow
Det.sen
Tri.spl
Lov.tri
Dae.kla
-50
Figure 47
50
100
150
DCA Axis 1score
200
250
Separate ordination of the study sites: DCA axis 1scores of the contour samples against the
middle contour line of each contour sample, respectively for Zagne, Tai and Para.
88
Forests of TaiNational Park
reflects a rainfall increase of 3.5 m(10y)' (350 mm y 1 ). In other words, the wettest and
thedriestforests onthe catena with an elevation interval ofabout30mdifferasmuch
in composition asforests lying90kmapartontherainfallgradient, when comparing a
spatially averaged composition of the latter forests with the composition of contour
samples.
3.4.2 Tree *species richness andtree density
Ageneral trend of decrease of large tree species richness is found from Zagn6towards
Para, i.e. with increasing rainfall (Figure 48). Among the sampleplots covering a fixed
area of 2 ha, most species are found in Zagne and least in Para. Within the catenasa
decrease is found in Tai down the slopeand an optimum in Para in upper slope, shoulder
and lower crest positions. Tree density shows even more variation: decrease from crest to
lower slopein Tai, and a maximum in upper and middlepositions in Zagne and Para.
This mightbe related to the steepness of the terrain. On thelatter sites crests and lower
slopes are less densely populated by trees. The number of species is a treepopulation
characteristic rather than a.forest characteristic, it should be compared for equal numbers
210
Zagne
09
Cd
190
o
CL
o.
170-
c
o
o
150Para
CO
>
130
110
— l
|
H
[
I
|
—I
— ;
10
20
30
40
3— number of species —V— number of trees
per contour sample (2 ha)
Figure 48
1
—
50
~1
10
1
12
1
14
1
16
- number of species
per 20 trees
Tree species richness (bold line) and tree density (thin line) per two ha contour plot, plotted
against altitude. On the right the quotient of both variables is given, expressed as the number of
species per 20 trees.
89
Forest gradients along slopes
of trees, andnotforequal areas offorest. So,I calculated the numberofspeciesper 20
treesasameasureofdiversity ofthe treepopulation. High and low diversity now occur
both inZagne and Para. Variation inlargetreespecies richnesswithinsitesis greater
than variation betweensites.Atthe Ta'i siteamaximum diversity isfound onupperand
middle slopes. The samples with low diversity areoften dominated byabundant species
like Chidlowia sanguinea inZagne z5, Pycnanthus angolensis inTa'i tl, and Sacoglottis
gabonensis inPara p5,p7andp8.
3.4.3 Biomass andbasal area
Basal area peaks inmiddle slopepositions and issmall atboth endsofthe catena. Here
theinfluence ofthe very bigtrees isfelt (Figure 53).InTai hugeEntandrophragma spp.
(wood density d=0.6) andTieghemella heckelii (d=0.7) trees occur asacorona around ;
the ironstone cap (see Figure 50 and also Oldeman 1974, Vooren 1985).InZagn6
Triplochiton scleroxylon (d=0.4), Erythrophleum ivorense (d=0.9) and Ceibapentandra
(d=0.3) trees grow very large, upto240 cm diameter, with apreference forthe middle
and upper slopeposition (Figure 51). InPara numerous very bigbut relatively short-boled
Sacoglottis gabonensis (d=0.9) trees were present (Figure52)
210
o
a.
£
a.
E
ID
C
o
o
i
1
r
0
5
10
15
biomass (t/ha.m) and basal area (m2/ha)
per contour sample (2 ha)
Figure 49
0.6
0.8
0.9
wood density (t/m3)
Biomass int h a ' m"1andbasal area inm2ha"' percontour sample, plotted against altitude.
Biomass is calculated asthesumofthe basal area perspecies weighted bythespecies' wood
density. Ontheright theaverage wood density int m"3ofthe large trees within a contour sample
is given.
90
Forests of Tai National Park
7
?0.
"€rr_ -qe—e620
50
100m
_cr----"o.
—WQo.o
.O-r O
C)
0
25-il.J?^ O ^ - " ; °*
O
Figure 50
o o
c\
A corona of large trees around the crest at the Tai' site. The bold line indicates the extent of the
iron pan. All large trees (d>70 cm) are plotted as open circles. The solid circles indicate the
position of 20 Entandrophragnia spp. trees, 11Erythrophleum ivorense trees, 5 Khaya anthotheca
trees and 2 Tieghemella heckelii trees. The diameter of the circles is square-root proportional to
the diameter of the tree, the smallest circles are trees of 70 cm d, the largest circle a tree of 240
cm d. Elevation contours, in m above sea level, are given every 5 metres.
Biomass was calculated as the sum of the basal area per species weighted by the species'
wood density. The high basal areas in Zagne consisted mainly of wood of relatively low
density, although heavy species also occurred. In Para the dominant large tree species had
very dense wood. The quotient of biomass and basal area is given on the right in
Figure 49, being the average wood density of the large trees. An increase in average
wood densityfrom 0.7inZagneand Taito0.8 inParaisfound withincreasing rainfall.
A similar trend can be discerned in the data published by Whitmore and Silva (1990) for
Amazonian forests.
91
Forest gradients along slopes
r©-<&
S^T^
^ ^ ^ o 0 p y o ^ 8 P o o ..04,
°CQ_ & o® °
"y?
-QO
°-!^^o^gcag^-^nP%jg^i: _®.
,P °
'9f\° ••& °
o
-•eP.
0
Figure 51
50 100 m
Triplochiton scleroxylon (16 ex.), Erythrophleum ivorense (35 ex.) and Ceibapentandra (6 ex.)
trees growing preferentially on upper and middle slope positions at the Zagne site. Drawing
conventions as in Figure 50.
—crfr--
'is.
-§r%
^
\° b
G„
,®0 -,©8
0
50 100m
.-«*
'.-' .^'
' • * ,
0 /••''.->
Figure 52
The distribution of Sacoglottis gabonensis (74 ex.) and Maranthes glabra (23 ex.) trees at the Para
site. S. gabonensis seems to avoid the high crest with iron pan, where M. glabra prefers the steep
and gravelly upper slopes.
92
Figure 53
Forests of Tai National Park
The biggest tree of western C6te d'lvoire. An Entandrophragma candollei tree in the ForSt classee
du Haut-Sassandra with a swollen base, 6 metres diameter at the height of the ruler, held by
Sormongar S. Zwuen, a Liberian forest researcher. The ruler itself measures 2 m.
93
Forest gradients along slopes
3.4.4 Comparison of the local and regional ordination
Correlation analysis of the species scores of the ordinations of Chapters 2 and 3 shows
the *similarity between the regional and the local forest gradient. 38 tree species were
represented in both ordination tables (Table 3and Table5). Their correlation isplotted'in
Figure 54 (R= 0.59, p<0.001). 'Wet' species are grouped in the upper right corner,
'dry' species in the lower left. The further away the species lay from the regression line,
the more their position differed in both ordinations. Both ordinations are subject to error
resulting from the varying sample size per species, and thus the varying precision of the
species' score. The ordination from Chapter 2has a wider scope in terms of the wet-dry
gradient and isbased on a larger data set containing many more trees, and thus its species
ranking isjudged morereliable.
D
aacgab
c
o
to
c
o
afz.bal
300-
C
-2
o
Q
D
at
c
0)
° 1 > " *
amp.pte
200-
a
lov.tri
^ ent.ang
kla.gab
®
"O
CD
opycang
a.
richeu
D
na>.pap
0-
D
anirob
plp.afr^^*^
•
old.ajr
lop.ala
^""Oanlutl
D
entcyl
a
ter.sup
chLreg
dis.beno Q
trgcl
a
dan.thu
chl.axc
gui.ahi
Figure 54
a
cei.pen
alb.ter
I
-100
lerivo
^aa^^Qentcan
^ ^ a n .l!»3ha.ant D«nllf»
petjnac
O
gua.ced
•*^
-100-
^ ^ ^
•
100-
8
(A
<
o
a
IksJiec
rho.bre
her.utj
D
canacfc
D
1
1
1
1
1
100
200
DCA1 scoreofthespecies (regionalordination)
'
Correlation of the species scores from the slope gradient ordination with those of the regional
ordination. R= 0.59, p<0.001. There were 38 large tree species common to both ordinations.
Correlation of the slope ordination with the regional gradient described for Ghana (Hall&
Swaine 1981) is of the same statistical significance. This time 67 species could be
compared and R= -0.64, p<0.001. Someof the outliers are rather rare but ubiquitous
species likeAfzeliabella, Canarium schweinfurthii, Irvingia gabonensis and Daniellia
thurifera. DCA1 species scores estimate where the species attain their maximum abundance along the gradient (ter Braak 1987a). The confidence limits of this estimate are
larger for species with a rather flat response curve than for species with a narrower
ecological range and apeaked response curve.
94
Forests of Tai National Park
DCA1scoreofthespecies (catenaordination)
400-
300
200
pen.but
uao.Qui mfir.gla
D
scy.tie Q
B
uap.cor dia.aubcP ,zb< "
can.sch
her.utl
O nr-""
amp.pte3
daeJila °
_..
edu n ,
• irv.gab
D""~
anUnec a
^tn.Srb
par.exo
tle.hoc p8r.bic
D
100
pyc.an
plpa
Q
CD
cor.paCi^
aub.plarfnt.can
entuj
petn^c < * > * [
gua.ced ste.obl
0O
dan.thu
-100-
gui.ehi
-200
10
Figure 55
D
lan.wel
Dg
rlc.heuD
nes.pap ate.acu
D
.toa
20
bli.wel
—\
30
1
I
40
50
60
H&Saxis1 scoreofthespecies
Correlation of the species scores of the catena ordination with the scores of Hall & Swaine (1981).
R = -0.65, p < 0.001. There were 67 tree species common to both ordinations. The further a
species is situated from the regression line, the more differently it was ranked in both ordinations.
For species codes, see Appendix I.
3.5 Discussion
3.5.1 Compositional gradients alongslopes
Slope gradients in tropical forest were described by Davis &Richards (1933) in Guyana,
by Lawson et al. (1970) in Ghana, by Lieberman et al. (1985) in Costa Rica and by
Jonkers (1987) in Suriname. In each study, a different lower diameter limit and a
different plot size were used to study thegradient in species composition. Asin tropical
rain forests many large tree species do not have an exponential but rather a bell-shaped
diameter distribution curve (Rollet 1974, Bongers et al. 1988)and each have a different
maximal size (Poker 1992), different species are abundant in different diameter intervals.
Thus, the species pool considered to describe the slope gradient depends on thediameter
interval chosen for study. Thelist of 95 species found at the three sites contains mostof
the large tree species known from the region between Zagne and Para (Guillaumet 1967,
Huttel 1977, Alexandre 1980, deRouw 1991.Not found but present at low density in the
region are Kantou guereensis, Pericopsis elata,Hannoa klaineana and Okoubaka
aubrevillei. Gilbertiodendronpreussiiwas found in the valley bottom at the Tai site(Bech
1983); de Rouw (1991) found it in association with Heritiera utilisover the entire catena
in forests half-way between Taiand Para.
Forest gradients along slopes
In forest close to my Tai site, Huttel (1977) studied the slope gradient considering mostly
smaller tree species, likeDiospyros mannii,most abundant in thediameter class 3.5 to
13cm, and Corynanthepachyceras,most abundant above 13cm diameter (seealso
Vooren 1979, Bech 1983). I considered trees above 70cm diameter and foundPycnanthus angolensis mostprominent on thecrest and Piptadeniastrum africanum on the
slope. Depending on the size class studied, different tree species are prominent, but the
changing abundance of any of these along the slope can express the slope gradient.
Large tree species often cover a broad interval along the slope gradient, so the gradient
analysis relies on differences in abundance of the species, rather than on presence or
'
absence data alone. A moreintensive sampling is necessary to obtain relevant quantitative
data per species. I admit that the abundance scores and absence data in Table 5 can be
interpreted neither with high precision nor in an absolute sense because of the limited
number of trees per sample (13to50 trees above70 cm diameter per 2 ha). Hardly any
individuals of Piptadeniastrum africanum, Sacoglottis gabonensis or Triplochiton
scleroxylon were found in the data set on smaller trees (30<d<70 cm), so a reduction of
thediameter limit provided noadditional information on their distribution.
Contour samples are spatially contiguous, e.g. t7 touches both t6 and t8, and species
composition is a spatially *autocorrelated attribute (Hubbell &Foster 1983). Hence, t7 is
supposed to have a composition intermediate to that of t6 and t8. As a result, it is
permissible to spatially interpolate the ordination scores of the contour samples, so t6and
t8 also provide indirect information on the species composition of t7. The weight of the
neighbouring samples is inversely related to the distance (elevation interval) to these
samples. The spatial correlation of the sample with its neighbours compensates to some
extent the limited precision of the species composition of the sampleitself.
Besides theanalysis of individual slopegradients, theresults prove the *similarity
!•
between the local forest gradients determined by slope and the regional forest gradient !
induced by climate. Slopegradients are slidinggradientssuperimposed on the regional j
forest gradients, as shown by Figure 46. When analysing four soil-vegetation catenas in
SW Cote d'lvoire, Guillaumet (1967) already found this compensation effect of rainfall by
slopeposition. De Rouw (1991) found that valley bottom species near Tai, like Gilbertiodendronpreussiiand Heritiera utilis, "climbed up" the slopeunder wetter climate moreto
the south. My results substantiate thesefindingsand yield a beginning of explanation.
3.5.2 Themethod of contoursampling
A strong point of the method is the constant area of each sample. Jonkers (1987) for
instance compared samples with areas varying from 740 to 1600 m2and sointroduced
differences between samples which indeed reflect the sampling method rather than forest
reality (see also Text box 2, p. 34).
On the contrary, a small methodological discrepancy between absolute elevation contour
lines and relative lines of equal slope position does not affect the results. However, within
larger plots, e.g. 50ha instead of 20 ha, this will probably present a problem. Its solution
lies in correcting the digital terrain model for theriver gradient. In this way, a continuous
95
96
Forests of Tat National Park
variable would be created, indicating the relative catena position for each point in the
forest. In their study in Costa Rica Lieberman et al. (1985)alsodirectly compared plots
with the same absolute elevation, although oneplot was close to the main river and the
other two were situated further upstream in the tributary water basins.
When comparing contour samples it isadvisable to check thealtitude levelsofthe
valley bottoms where theyoccurwithin aplot, andto examine the needfor a
correction.Therelative altitude abovethe valley bottom levelshouldbeassessedfor
anypermanent sample plot intropical rainforest (seealso Alder &Synnott 1992).
Species composition on a slopeposition could beassessed moreprecisely by increasing
the size of thecontour samples to e.g. 3or 4 ha, and thus the number of trees in the
sample. Within the same sampleplot, this would enlarge the elevation intervals and so
increase thelikelihood of including different catena positions in one sample.
Thecontour levels separating the samples are not equidistant. This is a consequence of
thechoiceof constant area of the samples. My method automatically adjusted the
elevation interval to the steepness of the terrain. In Para, for instance, contour sample2
on steep upper slopes covered an elevation interval of 6.5 mwhereas samples 7 and 9on
gently sloping lower slopes only covered 0.9 m(Figure 43). Soil variation may be greater
in samples with a wider elevation interval.
Constant area and equidistance of thecontour lines are not necessarily conflicting restrictions. If I had had a much larger plot, I could have taken 2 ha subsamples from fixed
elevation intervals, e.g. 2 m. Inevitably, less of the information contained in the data set
would be used in this way. However, certain intervals would be solarge that two or more
subsamples would be available. Then also the within-sample variance for the same catena
position could beassessed. In Figure 46 and Figure 47 this unexplained variance can only
beroughly guessed from theoscillation of thecurve around thelinear trend, although this
residual variance is interesting because it may reflect patterns resulting from forest
dynamics and seed dispersal that are independent of environmental conditions.
Apoint of debate might be the homogeneity within the contour samples. When designing
them, I assumed that the within-sample variation would be smaller than variation in rigid
square subplots of the same size. The shape of some, however, becamevery elongated.
Someother contour plots on lower positions were composed of subplots dispersed on the
outskirts of the plot. Given the spatial *autocorrelation character of species composition,
this design therefore may have lumped heterogeneous parts of the forest into the same
contour plot.
Moreover, from a theoretical point of view, species richness may beexpected tobe
greater in an elongated plot or in a setof distant subplots than in a square or circular plot
(Begon et al. 1986). However, thepresence of a strong slope gradient means that samples
of the same *dissimilarity are found much farther along the contour line than down the
slope, so the 'ideal' plot shape would amount to a very elongated ellipse, of which my
contour samples are an acceptable approximation.
Forest gradients along slopes
3.5.3 Tree density and biomass alongthe slope
Large tree density was found tobe greatest in Zagne\ This probably is caused by the
abundant occurrence of Chidlowia sanguinea with many individualsjust exceeding the
70 cm diameter limit. The great height of theemergent trees and their long boles also
leave more space for thecanopy trees. In Para, Sacoglottis gabonensis trees had broad
crowns and short boles and sooccupied large crown volumes, filling most of the canopy.
In Tai, Bech (1983) found that more emergent trees (h> 45 m)grew on the upper slope
than in other slopepositions. This was confirmed by my results. In the same forest Huttel
(1977) found more trees above 13cm diameter but fewer large trees (d>40 cm)on
summit sites with iron pans. Rollet (1974) stated that theabsence of larger trees generally
coincides with a greater density of smaller trees. The scarcity of large trees on crests
mightbe related to the iron pan, impeding deep rooting. Another aspect is that summit
sites are moreexposed to strong winds (Williamset al. 1969, Vooren 1985), sothat
uprooting and gap formation maybe more frequent (see also Bonnis 1980, Poorter etal.
1993in prep.).
3.5.4 Species richness gradients
*Species diversity, which isa characteristic of a population of trees, not of an area of
forest, was expressed as the number of species per 20 trees. The essential difference
between *species richness, a plot-related characteristic, and species diversity, apopulation-related one, clearly emerged from Figure 48. In grassland vegetation such a
distinction is uncommon, because it is not individual plants but species that are recorded.
The fact that perennial phanerophytes constitute most of thelife form spectrum in tropical
rain forests (Guillaumet 1967), requires that the researcher makes much more detailed
inventories by recording the complete spatial setting (x, yand z coordinates) of each
individual.
This greater precision of rain forestrelevesinspired Oldeman (1974; 1990b) in developing his forest architecture analysis. He noted that by discarding the spatial setting of the
data one ascended to a level of abstraction at which much essential information on the
ecosystem and its functioning was lost, e.g. by extracting average species listsand
diameter distributions, by summarizing tree data in averaged basal area or stem volume
figures and by averaging diameter increment figures or tree mortality data. Abstraction is
part of the knowledge acquisition process, but should be done in prudent steps, as
suggested by Koop (1989) or Leersnijder &Boeijink (1990). Oldeman (1974)plotted total
tree height against trunk diameter, neither in order to calculate theaverage h/d rationor
to fit an average h/d curve, but in order to rank the individuals and to distinguish their
affinity to a reference value. Deviation is then not considered as statistical noise butas
information (also cf. Hoekman 1985, for signals obtained by radar remote sensing).
Hamilton (1982) suggested that species richness increased towards Pleistocene forest
refugia. Species richness gradients might thusbe expected tobe oriented towards the core
areas of Pleistocene refugia. This orientation isoften parallel to theactual rainfall
gradient. On a 100by 100 mplot, Faber-Langendoen &Gentry (1991) found over 250
tree species above 10cm diameter in thevery wet Chocd region in Colombia, receiving
97
98
Forests of TaiNational Park
over 7 mrainfall per year. They ranked these forests among the most species-rich forests
in the world and concluded that great species richness is correlated with high rainfall and
with low nutrient levels (see also Bongers et al. (1988) and Hall &Swaine (1976) for
further discussion of the relation between diversity and soil fertility).
Capepalmas
CapeThreePoints
Mountainrange
*•*;<•«;* Pleistocene
*S&' forestrefugia
Actualforest-savannaboundary
EvergreenforestzoneboundaryinLiberia
Figure 56
The hypothetical positions of the Pleistocene forest refuges in West Africa according to Guillaumet
(1967). The western extent of the refuge on the hills of Grabo remains unknown (question mark on
the map).
However, Maley (1987, 1991)pointed out that in West Africa refugia were not located
near thecoast, where nowadays the highest rainfall is recorded, but in mountainous areas.
He believed that the stratiform cloud cover, generated by the upwelling of cold sea water
in the Gulf of Guinea at glacial times, only produced much rain whenrisingover
mountain ranges. Hetherefore doubted the existence of Pleistocene forest refugia near
CapePalmas in C6te d'lvoire and near CapeThree Points in Ghana as suggested by
Aubreville (1962) and Guillaumet (1967) and admitted only one refuge towards the
Guinea Highlands. However, the region of Cape Palmas contains hills up to 750 min
Liberia (Putu range) and upto 475 min Coted'lvoire and thuscannot beconsidered as
being flat (Figure 1, Figure 56). Moreover, the sea level was up to 110mlower than
today at the timeof glacial maximum (Martin 1972), socoastal relief was much more
pronounced than it is now.
Based on bird distribution maps of Hall &Moreau (1970), Hamilton &Taylor (1991)
supported the location of a core refugial area in SECote d'lvoire and SW Ghana (Cape
Three Points) but not the one near Cape Palmas in SELiberia and SWCote d'lvoire.
They situated another refugium core area along thecoast in Sierra Leone and western
Forest gradients along slopes
Liberia (see also Endler 1982, Mayr and O'Hara 1986).However, the sampling intensity
in Hall &Moreau (1970) was much higher in Sierra Leone and western Liberia and in
Ghana than it was in east Liberia and Cote d'lvoire, so the gradients of declining species
diversity pointing towards Cote d'lvoire, as interpreted by Hamilton (1982) and Hamilton
& Taylor (1991), may be a result of this sampling pattern.
Within Cote d'lvoire plant species richness was found to be greatest in the SWand SE
corners (Aubreville 1959, Guillaumet 1967)and decreases towards the centre region
where the savanna comes close to the coast, called the V-Baoule. Within Liberia, the
wettest forests are centred on Greenville with endemic large tree species likeDidelotia
brevipaniculata andLoesenera kalantha (Figure 20). New evidence concerning narrowly
endemicBegoniaspp. again pointed at refugia in SW Coted'lvoire and SW Ghana (Sosef
1993in press; see also Hall &Swaine (1981) for a discussion of the SWGhana
refugium). It ispossible that theentire evergreen forest zone, sensu Sachtler (1968; see
Figure 21) constituted a lowland forest refugium in glacial times, including the extreme
SWcorner of Cote d'lvoire and the SE corner of Sierra Leone.
Thehill ridges around Grabo withsummitsattaining475m atsome50kmfrom the
coastshould beinvestigatedfurther, preferablyincludingpalynological studies,to
confirm theposition of the hypothetical Pleistocene refugia. Special attention should
bepaid tothe westward extentof this refugium, namelyto seewhethertheGrabo
forests have eithera biogeographically uniquepositionoraremerelythe easternmost
part offorestsfound under similarclimatic conditions throughoutLiberia.
Figure 57
Mount Kope (412 m asl) is one of the summits of the hill ridges around Grabo, a hypothetical
glacial forest refuge. At present, it does not yet have a protected status and is threatened by
conversion to cocoa plantations and by logging.
99
100
Forests of TatNational Park
Thecontinuous model of rain forest variability, as presented here, brings a new dimension to therefuge theory, namely that of the "compressing" and "stretching" of such
gradients under long-term oscillation of climatic conditions. Maps of a reconstruction of
thevegetation cover at glacial maximum (18000 years BP), should not only distinguish
savanna from dense forest, but also indicate the gradient from evergreen to semideciduous forest within thewhole forested zone.
I suggest that the everwet Caesalpiniaceae species from the wettest forests in Liberia
mighthavea evolutionary migration history differing from the "normal" wet speciesalso
found in Ghana. The everwet coastal forests of Liberia may havebeen separated from
similar forests in Cameroon for much longer than the wet evergreen forests, sensu Hall &
Swaine (1981), in Coted'lvoire, Ghana and Nigeria. Among theoriginal species of the
everwet forests in Liberia and Cameroon, I therefore expect much more speciation to
haveoccurred than among those of the wet evergreen forests in Cote d'lvoire, Ghanaand
Nigeria.
As to the large tree species, my data showed a decrease in species richness from Zagne
towards Para, this is in the direction of increasing rainfall. I also showed that this trend is
related to a decrease in tree density from Zagne to Para. During avisit to the forests on
Mount Kop6near Grabo several large tree species belonging to everwet environments
were found which were absent from the Para study site, e.g. Brachystegia leonensis,
Cynometra ananta,Didelotia spp.. It shouldbe investigated whether these forests are
richer in species than the Para forests. From Figure 15in Chapter 2 no trend in species
richness on a regional scale could be detected, except from a drop in species richness in
thewettest forests in Liberia. Hall &Swaine (1976; 1981) found a clear species richness
gradient in Ghana's forests when considering all vascular plants: high moisture was
correlated with high species richness. When considering only tree species (largeand
small), they found no trend below 35 u.HSl (Hall &Swaine 1976), i.e. for the wetand
moist evergreen forests. All the forests I studied were wetter than this limit (see
Figure 17). Gentry (1982) also found apositive correlation between species richness and
rainfall for trees with a diameter above2.5 cm, but no correlation for trees above 10cm
diameter. This means that on wetter sites the diversity of biological plant types other than
trees increases markedly, which was also found by Guillaumet (1967) and de Rouw et al.
(1990).
The species richnessof the largetreeswithinthe West African tropicalrainforests
doesnot increasewithhigherrainfall. Onthe contrary,in verywetconditions with
rainfallabove 30mper tenyearsthe diversity in largetreespecies is often seento
decrease. As discussed in Chapter2, certainCaesalpiniaceae species thenbecome
single-dominant in theseforests.
With some caution I conclude that the regional maximum of tree species richness is
achieved at sites with intermediate rainfall conditions, where the tree species areas of the
evergreen and the semi-deciduous forests overlap. The local slope gradient found at the
Tai site also provides an example of such a combination: tree species typical of semideciduous forests grow on the upper slope and on the crest, whereas the tree species on
the lower slope are typical of evergreen forests. The entireTaliplot istherefore richer in
species, but given the trend in species composition along the slope, this is 8 rather than
Forest gradients along slopes
a diversity, i.e. diversity related to kinds of forests occurring besides each other, rather
than species diversity within one single kind of forest.
The a-diversity, i.e. diversity per slopeposition, was greatest on the middleto upper
slopes near Tal (Figure48). This was also found by Huttel (1977). It mightbecaused by
the fact that the two species groups, evergreen and deciduous, overlap in this intermediate
slopeposition.
3.5.5 Moisture indicator values for the largetree species
DCA proved successful in extracting one singleaxis pinpointing vegetation composition
out of a multivariate data set of up to 100tree species. Soit finally becamepossible to
describe moreprecisely the ecological optimaand ranges of large trees in theWest
African tropical rain forests, without being forced to use rather vague descriptive terms
like wet or moist evergreen, semi-deciduous, transitional, etc. Theecological optimumof
a species can be read from its position on thefirstordination axis (seeTable 3and
Table 5). Variation along this axis is apparently related to moisture, so the species scores
can be used as moisture indicator values. The ecological range of the species can beread
from the range of samples in which the species occurred. Adding moreobservations to
the data set would increase the reliability of these estimates and might result in a setof
more or less absolute environmental parameters of each species, similar to the indicator
values Ellenberg (1979) gave to theplants of Central Europe.
To make this possible, a West African data base would have to be created including as
many methodologically comparable or translatable sampleplots as possible all over the
Upper Guinea forest block: data collected by GTZ in Liberia (Sachtler 1968,W611 1981,
Poker 1989and 1992), and many herbarium specimens available in theWageningen
herbarium (Voorhoeve 1965), existing data sets from Cote d'lvoire (Aubreville 1959,
SODEFOR 1976a and 1976b, Bertault 1986, Vooren 1979and 1985,Huttel 1977, Maitre
1991, de Rouw 1991,Ak6Assi &Pfeffer 1975, AkeAssi 1984 and new inventory data
which are being collected at the moment by SODEFOR); and last but not least therecent
Ghanaian forest inventory (Ghartey 1989, Alder 1990, Hawthorne &Juam Musah 1993)
as well as the Hall &Swaine (1976, 1981)data set and the Kade data (Swaineet al.
1987). Such a project with both European and African partners would fit well into the
objectives of the development programmes of the European Community.
3.5.6 Moisture availability asa site hospitality factor
I showed that vegetation composition changes along soil catenas in a similar way as along
the rainfall gradient (see also deRouw 1991). I therefore linked both gradients to the
moisture available in the ecosystem. Van Herwaarden (1991b) found nodifference in
available moisture contents between gravelly crest and upper slope soilsand non-gravelly
lower slope soils (Figure 45). He found that the water contents at saturation of the
ironstone nodules was up to 35 vol %, whereas quartz gravel only contained 3vol.%.
How can wethen explain thewetter forests on lower slopes? Lateral drainage from uphill
may be one factor (Lescure &Boulet 1985, Bruijnzeel 1990, Fritsch 1992), thegroundwater level in thevalley bottom may be another. All slope positions receive the same
amount of rainfall, but lateral distribution and also availability of water are different.
101
102
Figure 58
Forests of Tal'National Park
Inundation of the Nse valley bottom near the Zagne study site in September 1990. Most of the
year this road can be used without any problem. These peak flow also have a great impact on the
forest vegetation.
Sothe ""hierarchy of forest gradients can be formulated as follows:
1. amajor gradient correlated with the rainfall gradient;
2. compensation or enhancing of the rainfall effect by bedrock at a meso-scale level.
On the one hand, under less rainfall, wetter forests are still found on sericite schist
bedrock, weathered to more clayey soils, rich in ironstone gravel in the upper catena
positions and with high water capacity. On the other hand, under higher rainfall, drier
forests are found on granite bedrock, because the soilsare sandier, the gravel is quartz
and often the soil is shallow with low water capacity and frequent rock outcrops.
3. small-scale slope gradients sliding onthe regional gradient. Summits carry rather
dry forests because an iron pan in the subsoil hampers the roots from reaching the
underlying clayey rotten rock. Lower down the slope, the rainfall water is augmented
by lateral seepage and flow.
The two sources of rainfall compensation, i.e. bedrock and slopeposition, are strong
enough to induce differences in forest composition which would otherwise be found 50 or
100km further away on the main rainfall gradient. This means that the species composition of a random sample of forest for which twoof the three sources of forest variation
are unknown, remains highly unpredictable. Careful assessment of all three sources
allows the moisture in a forest's environment to be predicted and the corresponding tree
species composition derived.
Forest gradients along slopes
The calculation of the compensation factors aspart of an algebraic explanatory model was
beyond the scope of this study. However, I indicated that schist bedrock resulted in a rise
of about 30 u.DCAl of Chapter 2in Figure 13. On granite bedrock near Para I found
forests containing numerous tree species also found in Zagne, which is some 70 u.DCAl
of Chapter 2 drier on theregional gradient. Along theTaicatena the forests on the crest
are some 50 u.DCAl of Chapter 2away from the lower slope forests. The axes, provided
by DCA, compress species information into a single dimension which can be further used
in algebraic modelling.
As shown in Chapter 2, it is difficult to quantify climateand bedrock. Furthermore, along
the soil catena rooting depth as well as theprocesses involved in lateral internal drainage
remain largely unknown. The interdisciplinary terminology developed by Beaudou et al.
(1978) and Richard (1989) can be used to describe and analyse these "layered" abioticand
bioticphenomena.
3.6 Conclusions
Aseries of two hectare *contour samples allowed compositional gradients of large trees
to be analysed along slopes. Theordination of thecontour samples clearly demonstrated
that tree species composition was strongly determined by slopeposition.
The simultaneous ordination of the three sites showed that these local slope gradients
were sliding gradients superimposed on the regional gradient. The regional gradient was
related to rainfall and lithology. These factors were largely expressed in soil moisture
availability. Gradients in moisture availability hence are expected to explain the slope
gradients too.
So, I conclude that moisture correlates well with tree species composition in West African
rain forests. For agiven forest, moisture conditions are a function of climate (rainfall,
temperature, air humidity), lithology (determining essential characteristics of the soil
catena) and slope position (local hydrology, presence of a hardpan).
Asto large tree speciesrichness,nopositive correlation was found with moisture
conditions. On thecontrary, a decrease in large tree density and species richness was
found towards the wettest study site. This supports the hypothesis that maximal tree
speciesrichnessis related tointermediate rainfall conditions.
Having studied the great variability of tropical rain forests at several scale levels, I warn
against the artless use of averaging and mixing techniques in rain forest studies, which are
based on over-simplified models of forest reality. The patterns in the three-dimensional
settings in natural forest ecosystems must be respected. I created the contour sampling
technique toprovide a sensitive tool for thispurpose.
103
105
4
IMPLICATIONS OF THE CONTINUOUS VARIABILITY MODEL
FOR FOREST MANAGEMENT AND CONSERVATION OF
BIODIVERSITY
In this Chapter, some implications of the continuous variability model are evaluated.
What can forest managers and conservationists do with theresults from thepresent study?
This Chapter suggests ways of using this model, after confronting it with thepracticeof
forest management and nature conservation in West Africa.
4.1 Forest management
Sustainable forest management has not yet been attained in West Africa (Parren &de
Graaf 1993in prep.). Timber mining companies have been profitably managed, not the
forests. Future management should focus first on conservation of remaining forest areas
and of nature and timber and other "values" in these forests. Another important task for
forestry will be therehabilitation of overexploited and depleted forests and of degraded
agricultural land (WRR 1992). Thepresent book depicts the forest wealth before
exploitation, but additional research must bedone on the numerous problems encountered
in such rehabilitation.
4.1.1 Forest inventory
Theclassic methods used for timber inventory Cote d'lvoireand Ghana werebased on
systematic line-sampling with a high sampling intensity in theline and a large distance
between lines (DRC 1967a, SODEFOR 1976a and 1976b, Ghartey 1989). For the
Liberian inventory, the area under investigation was divided into blocks of 8km2each,
within which the location of tracks (lines or squares) was determined at random (Sachtler
1968). The use of blocks introduced a systematic element into theinventory method.
An alternative method that involuntarily profited by the slope gradients present in the
forests was used in the soil survey of SWCoted'lvoire (DRC 1967b, seealso Gillison &
Bremer 1985). Large rectangular blocks of 50ha each which included all topographical
positions were delimited. Both soil and trees were studied within these blocks.
Unfortunately, I did not manage to retrieve theoriginal treedata but the three-dimensional representations of these landscapes are included in the report of DRC (1967b). Someof
these have been reproduced by de Rouw et al. (1990).
Aparallel can be drawn with two existing approaches to soil survey (Blokhuis 1993).
Either a regular grid is laid over the landscape and at each grid nodethe soil is described,
or the surveyor first studies intensively a number of representative catenas and tries to
understand the pedogenetic processes that lead to thepresent situation (cf. Fritsch 1980).
Special attention ispaid to lateral relations between slopepositions as induced by
drainage, transport of soil material and erosion.
I believe that the latter approach can be much more informative in matters of forest
ecology, forest architecture analysis and the study of forest dynamics. Narrow strips of
forest maybe sufficient for the timber miner tocalculate therichnessof his wooden ore,
106
Implications of the continuous variability model
but for the forest manager who wants to intervene in the forest ecosystem in an ecologically sound way, the relation between forest composition and geomorphology of the
terrain is of major importance (van Miegroet 1976, Mayer 1981,Fanta 1985,Oldeman
1991).
The species DCA scores presented by Hall &Swaine (1981) and in thepresent study
providea helpful framework toprecisely determine theposition of any newly sampled
West African forest area on the regional gradient. The impact of lithology can be assessed
on the forest gradient mapby comparing the score of this sample with nearby samples
underlain by different bedrock. In addition, thedegree of compensation on a particular
site as a result of catena position can be evaluated from the correlation of local gradients
with the regional gradient. To increase the reliability of species and sample scores, all
existing and newly gathered information should becompiled into an integrated interactive
data system.
4.1.2 Assessment of forest productivity
Besides an inventory of the existing stock, a forest manager needs information on forest
dynamics and productivity (de Graaf 1986). As the present forest is the result of its
growth in thepast, theevaluation of theproductive quality of a site will rely heavily on
our knowledge about the vegetation catenas. The method proposed by Alder &Synnott
(1992) to install permanent sample plots of 1ha, randomly distributed over the forests,
again disregards the landscape coherence of these forests. Therefore, information on
mortality, regeneration and increment of the 100largest tree species in West African
forests, should becollected in much larger plots, covering at least one complete catena.
Forest dynamics are driven by the mortality of thevery big trees (Vooren 1992c)and
valueincrement isrealized by their diameter growth (Staffers 1989). Only 5 to 10such
trees occur on a 1ha plot. So, catenaplots of 50ha as used by DRC (1967b) are
definitely moreappropriate in these specific forests than 1ha permanent sampleplots.
Regeneration surveys should be oriented to entire populations of a species, instead of
applying 'standard' systematic subsampling of the smaller diameter or height classes
(Swaine &Hall 1988;deKlerk 1991). Phenology, seed production, seedling establishment andjuvenile growth should be studied species by species. Otherwise there is a risk
of gathering vast amounts of largely useless data. Here also the knowledge of the
existence of catena gradients should be used.
4.1.3 Forest sensitivity to climatic change
The strong dependence of tree species on moisture conditions, as demonstrated in
Chapters 2 and 3, implies that they are sensitive to climatic change. Climatic changemay
or may not be induced by large-scale deforestation. Butclimate is known to be dynamic
and forests migrated over considerable distance to adapt to such changes in Pleistocene
times (Hamilton 1992).There was, however, great stress and many species became
extinct; this accounts for the relative actual poverty of the West African forest flora and
fauna (Hamilton 1976, Hall &Swaine 1981). Forest managers wish to arm the forests
against future modifications by controlling the hydrological cycles (Monteny & Casenave
1989)and by ensuring that the moisture does not leave theecosystem too quickly. In the
Forest management
Figure 59
Excessively logged forest in N'zo fauna reserve north of Tal National Park. The canopy has been
opened up completely and the control of the forest ecosystem on moisture conditions severely
weakened.
event of climatic change, species composition will change too. Stability, if it ever existed, \
is lost. This may explain thelack of regeneration of certain emergent tree speciesas
found by Aubreville (1938) and deKlerk (1991) and as discussed by Swaine &Hall
(1988).
Excessive opening upof the forest canopy by timber exploitation weakens the forest
i
ecosystem's ability to control moisture conditions (Figure 59) and so induces a shift in
1
species composition towards drier conditions. It is tobe expected that timber production '
forests under sustained management will occupy a drier position on thevegetation
gradient than thepristine forest that once existed on that spot. As mentioned in Chapter 3,
*Budowski's (1965) rule says that "secondary" tree species in wetter forests which are
likely to profit from canopy openings, are also characteristic of drier forests (Guillaumet
1967). Forest exploitation may therefore result in a "drier" forest composition.
y^XUM^
107
108
Implications of the continuous variability model
4.1.4 Forest harvesting systems
Several cycles of logging most often preceded forest management. The statein which the
forest is left by the loggers largely determines which silvicultural techniques can be
applied afterwards. However, measures will be more effective if the forest manager does
not wait until the logging dragon has appeased his hunger but starts right away with the
domestication of theloggers (de Graaf 1986) and tries to get their harvesting habits under
his control (Vooren 1992c). Ultimately, tree spotting, i.e. the selection of the trees tobe
harvested, is theresponsibility of the Forest Service. This has been practised in Ghana for
decades (Parren &de Graaf 1993in prep.).
Hendrison (1990) studied actual and improved logging systems in Suriname. He stressed
the importance of a careful design of the skidding pattern to reduce damage to the
remaining stand. In Coted'lvoire, forest roads are preferentially constructed on the water
divides on crest soils. The forests richest in timber are often located besides the crestson
upper and middle slopes (Vooren 1985). Excessive felling and damage in these most
valuable forests is likepulling the heart out of the future timber production forest. If these
parts of the forests crucial for timber production are severely damaged, future production
will be late and poor, even if the rest of the forest landscape remains untouched. In other
words, the yield, calculated over an entire landscape but taken from the richest parts of
the forest, is not a sustainable yield.Average damage mightbe slight, but local devastationof the most productive parts is complete.
It isa common mistake to consider the entire forested area in a forest reserve asproductive forest (see also Vanclay 1989). Real value production can only beassured in that part
of the landscape where soil conditions are optimal for forest growth. Swamps, ravines,
rocky outcrops and iron pans reduce theproportion of really useful forest land. In
industrial plantations of rubber trees (Hevea brasiliensis), the SAPH (Societe Africaine
pour les Plantations d'Hevea, west of Zagne) only planted trees on this more suitable half
of their concession. I advise forest managers to apply this also to their forests, as was the
practice in Queensland (Vanclay 1989). Productivity figures of 2 m3ha"1 y1 obtained from
sampleplots on fertile soil (Maitre 1991) should not be expected from the entire forest
estate (Ministere des Eaux et Forets 1988), but maximally from half of it, and then only
if not all highly productive large trees in these forests have been logged or destroyed.
Conservation of biodiversity
4.2 Conservation of biodiversity
4.2.1 Land use planning:conservation overthe entire gradient
Chapter 2 showed a continuous change of tree species composition from theLiberian
coast towards the forest-savanna boundary. Each part of the gradient has its proper
species that attain their optimum there. As a consequence, species conservation areas
should extend along theentire gradient. But how can we find a compromise between
conservation along the entire gradient, and the designation of reserves large enough to
attain the minimum area needed by large mammals to survive? In any case, there are no
forests left in West Africa covering areas so large that poachers are unable to affect the
mammalpopulations. Large mammals likeelephants or buffaloes manage to survivein
secondary forests as well (Hoppe-Dominik 1989), in even larger densities than in pristine
forests.
4.2.2 Corridor establishment: the Green Sickle
The problem of isolated populations could bepartially solved by installing corridors.
Would it be feasible to makea linked natural infrastructure from the semi-deciduous
forests in Foret Classee du Haut-Sassandra and the Mount Peko National Park, over the
Reserve de faune du N'zo, Ta'iNational Park and Grebo National Forest towards Sapo
National Park, Krahn-Bassa National Forest and ending near the mouth of Cestos River
(Figure 60)? Such a "Green Sickle" would not necessitate the removal of all human
settlements in the corridors, but these regions could begiven a "greener" appearance,
with permanently protected patches and belts of forests along water courses. It mustbe
explained to local people that the 100 %exploitation of landscape and wildlife will not
bring salvation to their children. In western Europe, private forestry is subsidized to
provide recreation, timber, wildlife habitats and landscape scenery. To some extent these
forest functions are also valid in the villages and towns in West Africa (Vooren 1987,
1992a).
4.2.3 Urgent conservation priorities
From abiodiversity point of view the attention of conservation agencies and Ministries
should be drawn to two promising areas within Cote d'lvoire where many results could
still be obtained:
• thesemi-deciduousforests, which contain their proper biodiversity. To date, conservation efforts in these formerly timber-rich forests have lagged behind, both in Ghanaand
in Coted'lvoire. I suggest improved protection plans for Marahoue National Park in
central Cote d'lvoire and e.g. the Foret Classee du Haut-Sassandra.
• thewettestforests of Cote d'lvoire. The hypothetical Pleistocene forest refuge on the
hills of Grabo merits the status of National Park. No foreign aid programme has yet
focused on these forests, which Guillaumet (1967, p. 72) praised as "C'est du Sud-Ouest,
la foret la plus riche et la plus belle que nous connaissions".
109
110
Implications of the continuous variability model
9°
LEGEND
V//A
8
Ten National Park
savanna
6°W.L.
other protected areas
Green Sickle
lake, sea
Figure 60
The Green Sickle, a green zone stretching from Mount Sangbe National Park in savanna over Tai
National Park towards the Atlantic Ocean. The installation of corridors between existing reserves
can guarantee the conservation of biodiversity over the entire gradient.
Other core areas in Liberia should receive all necessary protection. Further research could
clarify the issue of diversity centres in the region. Near the mouth of Cestos river,
between River Cess and Bafu Bay, the rain forest still borders the Atlantic Ocean for long
stretches, an ideal place topreserve all natural ecosystems from thebeach and the lagoons
near theriver's mouth up totheeverwet rain forests inland (Figure61).
Conservation of biodiversity
4.3 Epilogue
Forest gradients on regional and local scale explain a considerable part of the forest
variability in West Africa. A forest gradient is essentially an ecosystem gradient. The
large tree species I studied are only one set of forest organisms, albeit impressive ones
and thepillars of forest architecture. Gradient studies on other sets of organisms and at
other levels of scale will provide additional insight into the functioning of tropical forest
ecosystems. The faunal aspect has been neglected in this book, but is of paramount
importance in rain forest ecology. Koptur (1985), for instance, studied plant-animal
relations along an elevational gradient in Costa Rica and demonstrated alternative
defences of trees against herbivores along the gradient.
On a more general level, I hope that gradient approaches may find further acceptance
among ecologists and other scientists. Since the 18th century people have been at great
pains to classify all kinds of phenomena in this world. Now apromising path is to focus
on transitions and fuzzy limits and to redefine and reappreciate thevariability of nature.
At thebeginning of this study I abandoned existing classifications and this seemed to bea
stepback into uncertainty, but in the end it yielded a cleared and quantified picture of the
ever changing rain forest kaleidoscope.
Forests are notjust hectares, trees notjust cubic metres. The management of the most
species-rich ecosystems on earth is a challenge for present and future generations. Itwill
need international support and theefforts of all those fascinated by thispearl of our blue
planet.
Figure 61
A mother elephant bathing her young in the Atlantic Ocean. Based on sightings in Gabon,
reproduced from Bosman & Hall-Martin (1986). An image for the future in West Africa?
111
112
Glossary
GLOSSARY
autocorrelation: (spatial - ) an attribute is said tobe spatially autocorrelated when sample
points close together tend to be more similar than points further apart, but often without
easy, direct relation between sample sitelocation and the value of the attribute (Burrough
1987)
Budowski'srule: Budowski (1965) stated that pioneer or secondary woody species of the
humid tropics generally havea geographical area that includes drier regions. Oldeman
(1990b) amended its formulation, so that it may be applied world-wide: any plant species
which plays a pioneering role in hospitable environments has a geographical distribution
which includes less hospitable environments.
catena: (< Latin, chain) a connected series of related things; e.g. a *soil catena (see
there)
centroid: centre of mass, e.g. from a cloud of points in two-or three-dimensional space
classification: (vegetation - ) : the modelling in discrete units of vegetation by identifying
sociological or ecological species groups after first arranging species and samples and
then clustering them. Combinations of these groups define community types, each with
characteristic species combination (after Zonneveld 1988).
clustering: methods for grouping species and samples. There are agglomerative and
divisive clustering methods (van Tongeren 1987).
compensation: (of ecological factors) interaction between factors in such a way that they
may counterbalance each other's effects (e.g. soil moisture and climatic moisture)
continuum: something that is continuous, e.g. a forest continuum. The term 'continuum'
applies to the spatial variable itself and not to the direction or rate of change of the
variable (cf. *gradient).
contour sampling: combining avegetation sample for gradient analysis by selecting trees
growing between two elevation contour lines. Coordinates of the trees must be known and
a digital terrain model of the plot available.
cohort: (in forest ecosystem analysis) contains all individuals of one species within one
ecosystem that are of the sameage (Oldeman 1990b)
dissimilarity: thedegree of ecological difference between vegetation samples or species,
sometimes expressed as a distance function (see van Tongeren 1987)
diversity: see species diversity
dynamics (forest -): all processes of growth, death and species reproduction taking place
during *silvigenesis
113
ecological amplitude: (of a species) indicates therange of theenvironmental variables
within which the species occurs, e.g. temperature range, range in rainfall conditions, etc.
eigenscale: (by analogy with *eigenvalue) term indicating that objects (such as trees) of
different size that are studied together, are tobeconsidered each at their proper (in
German and Dutch: eigen) scale in matters of area of distribution and of dispersion. See
also Walter (1974) who discussed different scales for different plant sizes.
eigenvalue (of an *ordination axis): a value, proper to the *eigenvector of theaxis,
indicating thedispersion of the species scores on the *ordination axis. In this way it isa
measure of the importance of theaxis (ter Braak 1987a).
eigenvector: thevector characterizing an *ordination axis. If an extra iteration cycle is
carried out, the scores remain the same, so the vector is transformed into itself (ter Braak
1987a).
endemic (species): restricted to a particular geographical area or country (Webster's
Dictionary 1976)
factor (environmental - ; < Latin, maker): something that actively contributes to the
production of a result, e.g. plant growth (Webster's 1976)
false absent: a species not found in a certain sample and thus falsely considered tobe
absent if it does occur but is not detected because of the inventory method used, e.g.
because sample plots were too small, because only trees above a certain diameter limit
were considered, or because only commercial trees species were recorded.
forest: 1. (uncountable noun) land including a tree cover; 2. (countable noun) aplant and
animal community in a given abiotic environment
forest architecture: spatio-temporal structure linked to a well-defined hierarchical level
(Oldeman 1990b), here the forest ecosystem levels
forest complex: a large forested area, like a National Forest or a National Park, which
for an inventory needs to be split up into inventory compartments
forest block: a geographically more or less contiguous group of forests, covering e.g.
500 000 to 1000 000 ha. Typically, theWest African forests can be grouped into three
such blocks separated by deforested areas.
forest dynamics: seedynamics
forest inventory compartment: area of forest for which theinventory results are given
separately in the reports. Such an area generally covers 20000 to 50000 ha.
forest sample plot: a contiguous area of forest for which the data are recorded in oneset
offilesduring the fieldwork. Such plots can be 1ha, 20 ha as in Chapter 2, or 2 ha asin
Chapter 3.
114
Glossary
forest structure: the mathematical expression of structure of the trees in a forest sample
plot, e.g. as expressed by size (diameter or height) class distributions (cf. Rollet 1974)
forest zone: a living, forested vegetation zone, corresponding to a certain abiotic climatic
zone. The concept of zonation implies large geographical areas and a more gradual
changebetween zones than between e.g. types (Sachtler 1968).
gradient: 1. change in thevalue of a quantity per unit of distance in the direction of
greatest change, mathematically expressed as avector (Webster's Dictionary 1976,
Gillman &McDowell 1973), e.g. a river gradient: the degree of slope of theriver inm
per km. For non-linear or non-planar continua the gradient vector differs from point to
point. 2. a regularly increasing or decreasing change in a factor, e.g. temperature
(Lincoln et al. 1982) 3. a character gradient, also called 'cline', e.g. when samples from
areas not in contact, are compared and arranged according to this character to form an
imaginary continuum (Sobolev &Utekhin 1973)
hierarchy: a system organized in levels from lowest to highest in which one higher-level
system adds coherence to several subsystems one level lower. In ecology: a series of
levels in which each lower level consists of smaller subsystems of the higher level (e.g.
forest-organisms-organs-cells, Oldeman 1990b).
interval scale: a scale which possesses a constant unit of measurement. The differences
between values can be compared with each other, e.g. temperature in degrees centigrade
(Jager &Looman 1987).
inventory strata: distinguished in the inventory method "Stratified sampling". Strata (like
swamp forest, rocky outcrops, natural low bush, etc.) are often distinguished by aerial
photo-interpretation.
isohyet: line on a mapjoining places receiving the same amount of rainfall
kriging: a spatial interpolation technique, developed by Krige (1951), that calculates the
interpolated value of a spatial variable as a linear combination of the values from several
nearby observations. The weight of these observations depends on the degree of spatial
*autocorrelation of thevariable (Stein &Corsten 1991)
objective: adj. 1. dealing with outward things, actual facts, etc. uninfluenced by personal
feelings or opinions (Hornby et al. 1974);2. (of a choice or a method) accompanied bya
user's guide so that it can be checked or repeated in the same way, i.e. independently of
the researcher (subject) (Oldeman 1990b)
ordinal scale: a scale of measurement that implies a rank order between thevaluesof
classes, e.g. the Braun-Blanquet scale for measuring abundances of plants (Jager &
Looman 1987).
ordination:a technique used in vegetation scienceby which vegetation samplesand
species are arranged in a uni-or multidimensional order (Goodall 1954)
115
pattern: (originally: visible organization of a woven tissue) spatial configuration
(Webster's Dictionary 1976); see also Grace (1989)
pelophilousspecies: species growing on clayey soils with a high water retention capacity
(Mangenot 1955,Guillaumet 1967)
peneplain: ('pene'(Latin)=almost) a flattish plain resulting from erosion (ingeomorphology; Ahn 1970)
psammophilousspecies: species growing on sandy soil with a small water retention
capacity (Mangenot 1955, Guillaumet 1967)
rain forest: see tropical rain forest
sequence: a continuous or connected series of discrete units or events
scale: see interval, ordinal scale
silvigenesis: theprocess containing and organising all processes making a forest, e.g.
forest dynamics, succession, pollination, dispersal, etc. (Oldeman 1990b)
similarity: expresses their ecological resemblance of two sites or species. Several indices
of similarity exist (seevan Tongeren 1987; seealso dissimilarity).
soil catena: a sequence of soils underlain by similar bedrock and occurring under similar
climatic conditions, but having different characteristics due tovariation in relief, in
natural drainage and in position in relation to other soils (Ahn 1970)
spatial gradient analysis: an analysis of the gradual change of e.g. the species composition of theforest vegetation, but taking full account of the spatial setting of thesamples.
After ordination of the samples, contours of thegradient are drawn on the mapusing
spatial interpolation techniques (seealso ter Braak 1987b).
species diversity: thediversity of species within a given number of organisms. Several
indices exist, each taking into account the number of individuals per species (see e.g.
Krebs 1985).
species richness: the number of species on a sampleplot (Krebs 1985, Begon et al. 1986)
structure: see forest structure
subjective: (of a method or choice) the grounds for the choice are not or cannot be fully
explained and so cannot be divorced from theperson who made thechoice.
sustainability: the capability to supply services and products at oneand the same fixed
level and/or rate over an undefined but long timespan
116
Glossary
topology: 1. topographical study of a particular place; in this sense used byRamensky
(ex Sobolev &Utekhin 1973). 2. a branch of mathematics that investigates the properties
of a geometric configuration (asa point set) that are unaltered if theconfiguration is
subjected to a one-to-one continuous transformation (Webster's 1976)
toposequence: a sequence of topographical positions; e.g. from the lowest point (valley)
to the highest point (hill summit) in a landscape
trend: general direction of change, in space or time (Webster's 1976)
tropical rain forest: in thepresent book, I use this term in its broadest sense to indicate
theclosed-canopy forests south of the savanna region in West Africa.
TypeII error: is madewhen a null hypothesis is falsely accepted (Sokal&Rohlf 1969)
ubiquitous: omnipresent
upland: in thepresent book: referring to the upper part of the *catena; oppos.: swamp
and lower slope forests at the lower part of thecatena
vegetation classification: see classification
117
SOFTWARE PACKAGES USED
Cardbox 4.1
literature references database package
° 1989Business Simulations Limited
DECORANA
ordination program for vegetation tables
° 1979M.O. Hill
DrawPerfect 1.1
drawing package
° 1990WordPerfect Corporation, Orem Utah, USA
LotusFreelance 3.01 drawing package
0
1990Lotus Development Corporation
Opname
data entry facility for vegetation data
® Roelf Pot
• Department of Vegetation Science, Plant Ecology and Weed Science, Wageningen
Agricultural University
ScreenExtender™ for WordPerfect 5.1® 2.0
®1991 Stairway Software Inc.
Statgraphics 4.0
statistical package with graphics facilities
®1989STSC and Statistical Graphics Corporation
Supercalc5.1
spreadsheet package
®1991Computer Associates International, Inc.
Surfer 4.13
spatial interpolation package (2Dand 3D)
° 1989 Golden Software
Turbo Pascal 6.0
programming language
° 1990Borland
TWINSPAN
clustering program for vegetation table
° 1979M.O. Hill
Wolters'WordDwc Dutch-English-Dutch dictionary
° 1991Textware A/S, Kopenhagen
• Wolters-Noordhoff bv., Groningen
WordPerfect 5.1
word processing package
©1991WordPerfect Corporation, Orem Utah, USA
118
Appendices
APPENDICES
Appendix I
List of large tree species occurring in SELiberia and SW Cote d'lvoire,
including synonyms
All species from theforest inventories in Chapter 2 are listed, as well asthosefound in my three
sampleplots inTai National Park (Chapter 3). Nomenclature follows Hall & Swaine (1981, H),
Voorhoeve (1965, V), Aubreville (1959, A) or Hutchinson &Dalziel (1954-1972) D, inthat
order. Synonyms have no code (colomn 1)before the scientific name (2)and in column (3) there
is referred to the nameused inthisbook. The source of the scientific name isgiven in column
(4); "-" means: not mentioned inthis reference; "." means: mentioned with another name.
Column (5)gives a reference that mentiones the species for SE Liberia or SW C6te d'lvoire
(Guillaumet 1967, G; Sachtler 1968, S; deRouw et al. 1990, R; de Rouw 1991,R91;).
Columns (6)to (11) givethe number of trees inthe size classes >70 cm diameter (Z,T,P) and
30<d<70 cm (z,t,p) in the sample plots Zagne"(23ha for >70 cm, 10ha 50-70 cm, 5ha 30-50
cm), Tai (24ha for >70cm, 12ha 50-70 cm, 8hafor 30-50 cm) and Para (22ha for >70 cm,
10ha 50-70 cm, 6ha 30-50 cm). The maximal diameter found isgiven in column (12); the airdry wood density ingper dm3in column (13), sources: Bolza&Keating 1972, Dudek et al.
1981, Durand 1985, Vivien et Faure 1985).
code
(1)
scientific name
(2)
Aci.bar
Acioa barteri (Hook.f. ex Oliv.) Engl.
Afrormosia elata Harms
Afrosersalisia afzelii (Engl.) A . C h e v .
Afzelia bella Harms var.gracilior Keay
Afzelia bracteata T.Vogel ex Benth.
Albizia adianthifolia (Schumach.) W.F.Wight
Albizia ferruginea (Guill. & Perr.l Benth.
Albizia glaberrima (Schumach. & Thonn.) Benth.
Albizia zygia (DC.) J.F.Macbr.
Alstonia boonei De W i l d .
Alstonia congensis Engl.
A m p h i m a s pterocarpoides Harms
Aningeria robusta (A.Chev.) Aubrev. & Pellegr.
A n o p y x i s klaineana (Pierre) Engl.
Anthocleista nobilis G. Don
A n t h o n o t h a crassifolia (Baill.) J.Leonard
A n t h o n o t h a fragrans (Bak.f.) Exell & Hillcoat
A n t h o n o t h a macrophylla P.Beauv.
Antiaris africana Engl.
Antiaris toxicaria Leschenault subsp.
w e l w i t s c h i i (Engl.) C.C.Berg, var. africana
Antiaris t. subsp. w . , var. welwitschii
Antiaris w e l w i t s c h i i Engl.
A n t r o c a r y o n micraster A . C h e v . et Guillaum.
Araliopsis soyauxii Engl.
Araliopsis tabouensis Aubrfiv. & Pellegr.
Aubregrinia taiensis (AubreV. & Pellegr.) Heine
Aubrevillea kerstingii (Harms) Pellegr.
Aubrevillea platycarpa Pellegr.
Balanites wilsoniana Dawe & Sprague
Beilschmiedia bitehi A u b r e v .
Beilschmiedia mannii (Meisn.) Benth. & Hook.f.
Berlinia confusa Hoyle
Berlinia occidentalis Keay
Blighia w e l w i t s c h i i (Hiern) Radlk.
Bombax brevicuspe Sprague
Bombax buonopozense P.Beauv.
Bosquiea phoberos Baill.
Brachystegia leonensis H u t c h . & B.Davy
Afr.afz
Afz.bel
Afz.bra
Alb.adi
Alb.fer
Alb.gia
Alb.zyg
Ais.boo
Amp.pte
Ani.rob
Ano.kla
Ant.nob
Ant.era
Ant.fra
Ant.mac
Ant.toa
Ant.tow
Ant.mic
Ara.soy
Aub.tai
Aub.ker
Aub.pla
Bai.wii
Bei.bit
Bei.man
Ber.con
Ber.occ
Bli.wel
Bom.buo
Bra.leo
family
(3)
source
(4) (5)
chrysobat . HVA
=Per.eia
..A
Sapotacea . HVA
caesaipin . HVA
caesaipin . -VA G,V
Mimosacea . HVA R
Mimosacea . HVA
Mimosacea . HVA
Mimosacea . HVA
Apocynace . HV.
=Als.boo
..A
caesaipin . HVA
sapotacea. HVA
Rhizophor . HVA
loganiace . H-A
Caesaipin . - - A
caesaipin. HVA
caesaipin . HVA
=Ant.toa
.VA
Moraceae . H..
Moraceae . H..
=Ant.tou
..A
Anacardia . HVA R
Rutaceae . H..
=Ara.soy
.VA
sapotacea . HV.
Mimosacea . HVA
Mimosacea . HVA
Batanitac . HVA
tauraceae . -VAR
lauraceae . HVA
Caesaipin . HVA
Caesaipin . HVA R
sapindace . HVA
=Rho.bre
H.A
Bombacace . HVA
-iri.mad
.-A
caesaipin. -VAG
occurrence
Zagne
tai
Z
z
T
(6) (7)
(8)
(10) (11)
38
1
1
1
3
3
1
1
3
3
4
a 3
wood
dmaxdensity
(12)
(13)
2
4
'8
16
50
83
980
700
141
100
105
120
600
670
500
360
100
140
100
94
44
130
115
750
500
880
550
100
83
150
140
600
850
850
500
83
710
750
910
81
900
120
380
119
code
CD
scientific name
(2)
Bre.lep
Breviea leptosperma (Baehni) Heine
Breviea sericea A u b r e V et Pellegr.
Bridelia aubrevillei Pellegr.
Bridelia grandis Pierre ex H u t c h .
Brieya fasciculata De W i l d .
Buchholzia coriacea Engl.
Bussea occidentalis H u t c h .
Caloncoba brevipes (Stapf) Gilg.
Calpocalyx aubrevillei Pellegr.
Calpocalyx brevibracteatus Harms
Canarium schweinfurthii Engl.
Canthium arnoldianum (De W i l d . & Th.Dur.) Hepper
Canthium manense A u b r g v . & Pellegr.
Canthium tekbe A u b r e V & Pellegr.
Carapa procera DC.
Cassia aubrevillei Pellegr.
Cassia fikifiki A u b r g v . & Pellegr.
Cassipourea nialatou AubnSv. & Pellegr.
Ceiba pentandra (Linn.) Gaertn.
Celtis adolfi-friderici Engl.
Celtis mildbraedii Engl.
Chidlowia sanguinea Hoyle
Chlorophora excelsa (Welw.) Benth.
Chlorophora regia A . C h e v .
Chrysophyllum africanum DC.
Chrysophyllum albidum G.Don
Chrysophyllum delevoyi De W i l d .
Chrysophyllum perpulchrum Mildbr. ex H u t c h . &
Dalz.
Chrysophyllum pruniforme Pierre ex Engl.
Chrysophyllum taiense Aubr§v. & Pellegr.
Coelocaryon o x y c a r p u m Stapf
Cola lateritia K. S c h u m
Cola nitida (Vent.) S c h o t t & Endl.
Combretodendron africanum (Welw. ex Benth. &
Hook.f.) Exell
Combretodendron macrocarpum (P.Beauv.) Keay
Copaifera salikounda Heckel
Cordia platythyrsa Bak.
Corynanthe pachyceras K.Schum.
Coula edulis Baill.
Crudia gabonensis Pierre ex Harms
Crudia klainei Pierre ex De W i l d .
Cryptosepalum m i n u t i f o l i u m H u t c h . & Dalz.
Cryptosepalum tetraphyllum (Hook.f.) Benth.
Cynometra ananta H u t c h . & Dalz.
Cynometra megalophylla Harms
Dacryodes klaineana (Pierre) H.J.Lam
Daniellia ogea (Harms) Rolfe ex Holl.
Daniellia thurifera Benn.
Detarium senegalense J.F.Gmel.
Dialium aubrevillei Pellegr.
Dialium dinklagei Harms
Dialium guineense Willd.
Didelotia brevipaniculata J.Leonard
Didelotia idae Oldeman, De W i t & Leonard
Didelotia unifoliolata Oldeman, De W i t & Leonard
Diospyros ivorensis A u b r e v . & Pellegr.
Diospyros mannii Hiern
Diospyros sanza-minika A.Chev.
Diospyros soubreana F.White
Discoglypremna caloneura (Pax) Prain
Bn.gra
Succor
Bus.occ
cac.bre
cai.aub
Cal.bre
Can.sch
Can.arn
can.man
Car.pro
cas.aub
cas.fik
Cas.nia
cei.pen
cei.ado
Cel.mil
chi.san
chi.exc
chi.reg
chr.alb
chr.del
chr.per
chr.pru
chr.tai
coe.oxy
col.cor
col.nit
cop.sal
cor.pla
cor.pac
Cou.edu
cru.gab
cru.kia
Cry.min
cry.tet
Cyn.ana
Cyn.meg
oac.kla
Oan.oge
Dan.thu
Det.sen
Dia.aub
Dia.din
Dia.gui
Did.bre
Did.ida
Dio.man
oio.san
Dio.sou
ois.cal
family
(3)
source
(4) (5)
sapotacea .
=Bre.iep
=Bri.gra
Euphorbia.
=pip.fas
capparace .
caesaipin .
Fiacourti .
Mimosacea .
Mimosacea .
Burserace .
Rubiaceae .
Rubiaceae .
=can.arn
Heliaceae .
Caesaipin .
caesaipin .
Rhizophor .
Bombacace .
uimaceae .
Uimaceae .
Caesaipin.
Horaceae .
Horaceae .
=chr.del
sapotacea .
sapotacea .
sapotacea .
HV.
..A
..A
HV.
.-A
sapotacea .
sapotacea .
Hyristica .
sterculia .
sterculia.
=pet.mac
HVA
--A
HVA
HVA
HVA
..A
-Pet.mac
caesaipin .
Boraginac .
Rubiaceae .
Olacaceae .
caesaipin .
caesaipin .
caesaipin .
caesaipin .
Caesaipin .
Caesaipin .
Burserace .
Caesaipin .
caesaipin .
Caesaipin.
Caesaipin .
Caesaipin .
caesaipin .
Caesaipin .
Caesaipin .
-did.ida
=Dio.man
Ebenaceae .
Ebenaceae .
Ebenaceae .
Euphorbia .
-V.
HVA
H-A
H-A
HVA
Z
I
z
t
P
P dmaxdensity
(6) (7) (8) (9) (10)(11) (12)
(13)
1
1
H-A R
HVA
--A
-VA
HVA
HVA
H-.
--A
.-A
10
24
2
2
1
2
2
50
62
7
46
2
4
1
2
1
2
2
4
6
3
1
40
130
65
130
HVA R
--A
--A
64
47
•VA V
HVA
HVA
6
4
1
230
64
320
740
117
95
120
950
640
640
HVA R91
HVA
HVA
HVA
..A
109 174
2
4
12
15
3
1
60
HVA V
HV.
HVA V
3
5
38
1
84
740
34
610
650
2
1
4
19
1
1
32
1
00
4 119
27
3
3 67
96
80
780
500
760
990
143
123
119
550
730
970
36
53
32
85
960
860
HVA V,G
-VA R
-VA G
HVA V,S
HVA R
HVA R
HVA
2
1
3
4
HVA V
HVA
HVA
HVA
2
1
1
1
1
2
17
4
1
3
1
3
1
5 13
HVA R
HVA
-V. G
-V. G
-.A
.-A
H-A
H-A
H-HVA
3
30
14
1
6
400
120
code
CD
Ois.ben
Ory.aub
Dry.ayl
Dry.Jcla
Dry.pel
Dub.vir
Ehr.tra
Eia.gui
Ena.pol
Ent.ang
Ent.can
Ent.cyl
Ent.uti
Ery.ivo
Ery.man
Fag.tes
Fic.ela
Fun.afr
Fun.eta
Gil.bil
Gil.ivo
Gil.pre
Gil.spl
Glu.ivo
Gua.ced
Gua.tho
Gui.ehi
Gym.zai
Han.kla
Hap.mon
Her.uti
Hol.gra
Horn.let
Hun.ebu
Irv.gab
Kan.gue
Kea.bri
Kha.ant
Kig.afr
Kla.gab
Lan.wel
Loe.kal
Lop.aLa
Lov.tri
Hac.bar
Hag.but
Mam.afr
Man.obo
Man.alt
Mar.aub
Har.chr
Appendices
s c i e n t i f i c name
<2)
Distemonanthus benthamianus Baill.
Drypetes aubrevillei L6andri
Drypetes aylmeri Hutch. & Dalz.
Drypetes klainei Pierre ex Pax.
Drypetes pellegrinii Leandri
Duboscia viridiflora (K.Schum.) Mildbr.
Dumoria heckelii A.Chev.
Ehretia trachyphylla C.H.Wright
Elaies guineensis Jacq.
Enantia polycarpa (DC.) Engl. & Diels
Endotricha taiensis Aubr6v. & Pellegr.
Entandrophragma angolense (Welw.) DC.
Entandrophragma candollei Harms
Entandrophragma cylindricum {Sprague) Sprague
Entandrophragma utile (Dawe & Sprague) Sprague
Erythrophleum ivorense A.Chev.
Erythroxylum mannii Oliv.
Fagara macrophylla (Oliv.) Engl.
Fagara tessmannii Engl.
Ficus elasticoides De Wild.
Funtumia africana (Benth.) Stapf
Funtumia elastica (Preuss) Stapf
Gilbertiodendron bilineatum (Hutch. & Dalz.)
J.Leonard
Gilbertiodendron ivorense (A.Chev.) J.Leonard
Gilbertiodendron preussii (Harms) J. Leonard
Gilbertiodendron splendidum (A.Chev. ex Hutch. &
Dalz.) J.Leonard
Gilbertiodendron taiense AubrSv.
Gluema ivorensis Aubr6v. & Pellegr.
Guarea cedrata (A.Chev.) Pellegr.
Guarea thompsonii Sprague et Hutch.
Guibourtia ehie (A.Chev.) J.Leonard
Gymnostemon zaizou Aubr6v. & Pellegr.
Hannoa klaineana Pierre & Engl.
Haplormosia monophylla (Harms) Harms
Heritiera utilis (Sprague) Sprague
Hirtella butayei (De Wild.) Brenan.
Holoptelea grandis (Hutch.) Mildbr.
Homalium letestui Pellegr.
Hunteria eburnea Pichon
Irvingia gabonensis (Aubry-Lecomte ex O'Rorke) Baill
Kantou guereensis Aubre"v. & Pellegr.
Kaoue stapfiana (A.Chev.) Pellegr.
Keayodendron bridelioides (Mildbr. ex Hutch, et
Dalz.) Le'andri
Khaya anthotheca (Welw.) CDC.
Kigelia africana (Lam.) Benth.
Klainedoxa gabonensis Pierre ex Engl.
Lannea welwttschii (Hiern) Engl.
Loesenera kalantha Harms
Lophira alata Banks ex Geartn.f.
Lovoa trichilioides Harms
Macaranga barteri Muell. Arg.
Magnistipula butayei subsp. sargosii De Wild.
Mammea africana Sabine
Manilkara multinervis (Bak.) Dubard
Manilkara obovata (Sabine & G.Don) J.H.Hemsley
Manilkara sylvestris AubreV. & Pellegr.
Mansonia altissima (A.Chev.) A.Chev.
Maranthes aubrevillei (Pellegr.) Prance
Maranthes chrysophylla (Oliv.) Prance
family
(3)
source
(4) (5)
Caesalpin.
HVA
HVA
H-A
-VA
H-A
HVA
..A
H-A
H-A
HVA
..A
HVA
HVA
HVA
HVA
HVA
HVA
.VA
Euphorbia.
Euphorbia.
Euphorbia.
Euphorbia.
Tiliaceae.
=Tie.hec
Boraginac.
Palmae
Annonacea.
=Aub.tai
Metiaceae.
Meliaceae.
Meliaceae.
Metiaceae.
Caesalpin.
Erythroxy.
=Zan.git
Rutaceae .
Moraceae .
Apocynace.
Apocynace.
Caesalpin.
Z
z
T
P
P dmaxdensity
(6) (7) (8) (9) (10)(11) (12)
(13)
7
16
1
1
1
1
1
2
1
1
2
35
4
2
4
6
1
8
1
1
10
11
3
1
1
7
9
2
4
105
50
35
33
41
91
1
540
130
165
125
150
182
540
690
620
660
950
1
2
120
4
56
11
56
HVA G
Caesalpin.
-VA G
Caesalpin.
HV.
HVA
Caesalpin.
=GiI.pre
..A
Sapotacea.
HVA R
Heliaceae.
HVA
Heliaceae.
HVA A
Caesalpin.
Simarouba.
HVA
HVA
Simarouba.
H.A G,V
PapiIiona.
-VA V
Sterculia.
Flacourti.
HV.
.-A
H-A
H-A
Apocynace.
H-A R
Irvingiac.
HVA
=Hag.but
Ulmaeeae .
Sapotacea.
=Sta.sta
Euphorbia.
-.A
H-A
HVA
Bignoniac.
H-A R
Irvingiac.
HVA
HVA
Caesalpin.
-V- V,S
Ochnaceae.
HVA
HVA
HVA
H..
HVA
.-A
HV.
-.A
HVA
H..
Heliaceae.
Euphorbia.
Chrysobal.
Guttifera.
=Man.obo
Sapotacea.
=Man.obo
Sterculia.
Chrysobal.
Chrysobal.
1
2
1
80
6
2
109
8
18
1
1
15 23
2
-VA R
Meliaceae.
Anacardia.
1
H.. G
9
20
6
5
2
2
3
2
7
6
5
3
2
3
2
3
1
1
1
160
2
2
1
92
1
2
35
38
53
1
3
65
2
2
700
550
37
42
33
-V- V
H-A
HVA
HVA
670
79
121
code
(1)
s c i e n t i f i c name
(2)
family
source
(3)
(4) (5)
Mar.gU
Mar.rob
Har.raic
Mem.iat
Mit.cii
Mit.sti
Mon.com
Hor.sen
Mor.mes
Nau.afv
Nau.did
Nau.pob
Maranthes glabra (Oliv.) Prance
Maranthes robusta (Oliv.) Prance
Mareya micrantha (Benth.) Muell. A r g .
Memecylon lateriflorum (G.Don) Bremek.
Mitragyna ciliata Aubrev. & Pellegr.
Mitragyna stipulosa (DC.) O.Ktze
Monopetalanthus compactus H u t c h . & Dalz.
Morelia senegalensis A.Rich.
Morus mesozygia Stapf
Nauclea aff. vanderguchtii (De Wild.) Petit
Nauclea diderrichii (De Wild. & Th.Dur.) Merrill
Nauclea pobeguinii (Pob^guin ex Pellegr.) Petit
Nauclea trillesii (Pierre) Merrill
Nauclea xanthoxylon (A.Chev.) AubnSv.
Nesogordonia papaverifera (A.Chev.) R.Capuron
Newtonia aubrevillei (Pellegr.) Keay
Newtonia duparquetiana (Baill.) Keay
Ochthocosmus africanus Hook.f.
Octoknema borealis H u t c h . & Dalz.
Okoubaka aubrevillei Pellegr. & Normand
Oldfieldia africana Benth. & Hook.f.
Omphalocarpum ahia A.Chev.
Ongokea gore (Hua) Pierre
Pachypodanthium staudtii Engl. & Diels
Pachystela brevipes (Bak.l Baill. ex Engl.
Panda oleosa Pierre
Chrysobal.
H..
H-.
H-A
H-A
HVA
HVA
-V--A
HVA
-VHV.
HVA
..A
-VA
HVA
HVA
HVA
.VA
HVA
HVA
•VA
HVA
HVA
HVA
H-A
HVA
-VA
.VA
HVA
HVA
.VA
.-A
HVA
H--
Nau.xan
Nes.pap
Neu.aub
Neu.dup
oct.bor
oko.aub
oid.afr
Omp.ahi
Ong.gor
Pac.sta
Pac.bre
Pan.ole
par.con
Par.exc
Par.bic
Pav.cor
Pen.mac
Pen.but
Per.eia
Pet.mac
Phy.afr
pip.afr
Pip.fas
Pia.boy
.pia.pse
Pla.ema
Parinari aubrevillei Pellegr.
Parinari chrysophylla Oliv.
Parinari congensis F. Didr.
Parinari excelsa Sabine
Parinari glabra Oliv.
Parinari robusta Oliv.
Parkia bicolor A.Chev.
Pavetta corymbosa (D.C.) F.N.Williams
Pavetta nitida (Schum.& Thon.) Hutch. & Dalz.
Pentaclethra macrophylla Benth.
Pentadesma butyracea Sabine
Pericopsis elata (Harms) van Meeuwen
Petersianthus macrocarpus (P.Beauv.) Liben
Phyllocosmus africanus (Hook.f.) Klotzsch
Piptadeniastrum africanum (Hook.f.) Brenan
Piptostigma fasciculatum (De Wild.) Paiva
Placodiscus boya A u b r g v . & Pellegr.
Placodiscus pseudostipularis Radlk.
Plagiosyphon emarginatus (Hutch. & Dalz.)
J.Leonard
Pro.sta Protomegabaria stapfiana IBeille) Hutch,
pte.hyi Pteleopsis hylodendron Mildbr.
pte.san Pterocarpus santaloides L'Her. ex DC.
pte.beq Pterygota bequaertii De W i l d ,
pte.mac Pterygota macrocarpa K.Schum.
Pyc.ang Pycnanthus angolensis (Welw.) Warb.
Pycnanthus kombo W a r b .
Quassia undulata (Guill. & Perr.) D.Dietr.
Rho.bre Rhodognaphalon brevicuspe (Sprague) Roberty
Ric.heu Ricinodendron heudelotii (Baill.) Pierre ex Pax
Sac.gab Sacoglottis gabonensis (Baill.) Urb.
Sam.din Samanea dinklagei (Harms) Keay
sap.aub Sapium aubrevillei Leandri
Sch.arb Schrebera arborea A . C h e v .
Scotellia chevalieri Chipp.
Scotellia coriacea A . C h e v . ex H u t c h . & Dalz.
Chrysobal.
Euphorbia.
Metastoma.
Rubiaceae.
Rubiaceae.
Caesaipin,
Rubiaceae.
Horaceae .
Rubiaceae.
Rubiaceae.
Rubiaceae.
=Nau.did
Rubiaceae.
Sterculia.
Mimosacea.
Himosacea.
=Phy.afr
Olacaceae.
Santalace.
Euphorbia.
Sapotacea.
Olacaceae.
Annonacea.
Sapotacea.
Pandaceae.
=Mar.aub
=Mar.chr
Chrysobal.
Chrysobal.
=Mar.gla
=Har.rob
Mimosacea.
Rubiaceae.
1
Mimosacea.
Annonacea.
Sapindace.
Sapindace.
Caesaipin.
Euphorbia.
Combretac.
Papiliona.
Sterculia.
Sterculia.
Myristica.
=Pyc.ang
=Han.kla
Bombacace.
Euphorbia.
Humiriace.
Mimosacea.
Euphorbia
Oleaceae
=Sco.kla
=Sco.kla
140
5
v.s
1
31
125
2
3
9
2
1
1
9
100
2
1
90
7
2
R
R
9
12
740
730
730
43
A
1
13
2
135
970
63
90
810
720
140
37
470
90
90
950
850
150
60
148
59
76
69
810
R
4
3
3
6
3
5
1
R
R
.--Rouw91
Ixonantha.
23
D
HVA
HVA
HV. G
H..
H..
HVA
H-.
H-A
H-A
-VA R
Lecythida.
3
dmaxdensity
(12)
(13)
32
=Pav.cor
Papiliona.
T
t
P
P
(8) (9) (10) (11)
0,G
Mimosacea.
Guttifera.
Z
(6)
HVA
HVA
HVA R
H-A A,R )1
HVA
HV.
..A
.V.
.V.
HVA
HVA
HVA
HVA
H-A
.-A
.VA
2
7
5
7
3
8
4
7
1
1
2
3
6
1
14
7
9
1
3
7
21
3
1
2
3
1
2
1
1
690
1 36
3
34
3
3
1
4
5
137
610
470
4
2
74
3
110
111
240
108
95
460
350
880
900
650
1
5
4
1
1
Appendices
122
code
(1)
scientificname
(2)
sco.Ida
Scy.tie
spa.cam
Spo.pre
sta.sta
Scottellia klaineana Pierre
Scytopetalum tieghemii (A.Chev.) Hutch. & Dalz.
Spathodea campanulata P.Beauv.
Spondianthus preussii Engl.
Stachyothyrsus stapfiana (A.Chev.) J.Leonard &
Voorhoeve
Sterculia oblonga Mast.
Sterculia rhinopetala K.Schum.
Sterculia tragacantha Lindl.
Stereospermum acuminatissimum K.Schum.
Strephonema pseudocola A.Chev.
Strombosia glaucescens Engl.
Symphonia globulifera L.f.
Tarrietia utilis ISpraguel Sprague
Terminalia ivorensis A.Chev.
Terminalia superba Engl. & Diels
Tetraberlinia tubmaniana J.Leonard
Tetrapleura tetraptera (Schum. & Thonn.) Taub.
Tieghemella heckelii Pierre ex A.Chev.
Toubaouate brevipaniculata (J.Leonard! Aubr. et
Pellegr.
Treculia africana Decne
Trichilia heudelotii Oliv.
Trichilia lanata A.Chev.
Trichilia megalantha Harms
Trichilia monadelpha (Thonn.) J.J. De Wilde
Trichilia splendida A.Chev.
Trichilia tesmannii Harms
Trichoscypha arborea (A.Chev.) A.Chev.
Trilepisium madagascariense DC.
Triplochiton scleroxylon K.Schum.
Turraeanthus africanus (Welw. ex C D C . ) Pellegr.
Uapaca corbisieri De Wild.
Uapaca esculenta AubreV. & Leandri.
Uapaca guineensis Muell. Arg.
Uapaca heudelotii Baill.
Uapaca paludosa Aubrev. & Leandri
Vitex ferruginea Schum. & Thonn.
Vitex fosteri C.H. Wright
Vitex micrantha GOrke
Vitex rivularis Gurke
Xylia evansii Hutch.
Xylopia aethiopica (Dunal) A.Rich.
Xylopia elliotii Engl. & Diels
Xylopia parviflora (A.Rich.) Benth.
Xylopia quintasii Engl. & Diels
Xylopia Staudtii Engl. & Diels
Xylopia villosa Chipp
Xylopiastrum villosum (Chipp) Aubr.
Zanthoxylum gilletii (De Wild.) Waterman
ste.obi
Ste.rhi
ste.tra
ste.acu
str.pse
str.gia
sym.glo
Ter.ivo
Ter.sup
Tet.tub
ret.tet
Tie.hec
Tre.afr
Tri.meg
Tri.mon
Tri.spl
Tn'.tes
rri.arb
rri.mad
Tri.sci
Tur.afr
uap.cor
uap.gui
uap.heu
uap.pal
vit.fer
Vit.mic
vit.riv
Xyl.eva
xyl.aet
Xyl.etl
xyl.par
Xyl.qui
Xyl.sta
Xyl.vil
Zan.gil
Total number of trees:
family
(3)
source
(4) (5)
Fiacourti .
scytopeta .
Bignoniac .
Euphorbia .
Caesalpin .
H..
HVA
HVA
HVA
sterculia .
sterculia.
sterculia .
Bignoniac .
combretac .
oiacaceae .
Guttifera .
=Her.uti
Combretac.
combretac .
caesalpin .
Himosacea .
sapotacea .
=Did.bre
HVA
H-A
HVA
HVA
Moraceae .
-Tri.mon
=Tri.tes
Heliaceae.
Meiiaceae.
Heliaceae.
Meiiaceae.
Anacardia .
Horaceae .
sterculia.
Heliaceae .
Euphorbia .
=uap.cor
Euphorbia .
Euphorbia .
Euphorbia .
verbenace .
=vit.fer
Verbenace .
Verbenace.
Himosacea.
Annonacea .
Annonacea.
Annonacea.
Annonacea.
Annonacea .
Annonacea.
=xyi.vil
Rutaceae .
HVA
.VA
.VA
H-A
Z z
(6)(7)
T t
P P
(8)(9) (10)(11)
5 26
10
9
1
17
1
28
1
1
dmaxdensity
(13)
(12)
90
72
77
30
660
700
300
90
69
36
106
780
840
380
120
95
540
550
65
247
610
670
39
51
83
31
76
600
550
780
205
390
-V. G,V
2
8
1
3
2
1
1
8
HVA I!
HVA
26
1 41
2 26
HVA G
..A
HVA
HVA
1
2
29 12
7
1
-V- V,S
HVA
HV.
-.A
«..
--A
1
1
2
1
2
1
1
H..
H-A
5
1 13
H-. D
HVA
1
16
HVA R
HV.
..A
HVA
HVAR
HVA
H-.
.-A
HVA
H-A
HVA
H-A
H-A
H-A
H-A
HVA
H-.
.-A
H..
3
1 1
3
6
90
720
1 1
11
9
125
710
750
40
1
21 46
2
2
2
2
1
1
401 511
1
277592
213368
55
38
110
47
42
43
52
40
530
800
610
850
420
123
Appendix II Forest inventory data from SELiberia and SW Cdted'lvoire
From the SODEFOR (1976a,b) inventory reports Iused the figures of total volumeper inventory
compartment of trees exceeding 70 cmdiameter. Thefirsttable givesthe first class species, and
the second tablethe non-commercial species. The latter were not exhaustively inventoried.
Commercial species
SODEFOR (1976a,b)National Forest Inventory: timber volumes above70cmdinm3perinventory compartment
Centre-South region
North-West region
species
vern. name
1NW
2NU
3NW
4NW
5NW
6NW
7NW
6CS
7CS
afz.bel
alb.fer
als.boo
ani.rob
ant.tox
ber.con
ber.occ
can.sch
cei.pen
eel.ado
cel.mil
chlorsp
chr.gig
dan.thu
dis.ben
ent.ang
ent.can
ent.cyl
ent.uti
ery.ivo
fun.afr
gil.pre
gua.ced
gui.ehi
her.uti
hol.gra
kha.ant
lop.ala
lov.tri
man.alt
mit.cil
mor.mes
nau.did
nes.pap
per.ela
pip.afr
pte.mac
pyc.ang
rho.bre
ric.heu
sco.kla
ste.obl
ste.rhi
ter.ivo
ter.sup
tie.hec
tri.scl
tur.afr
Azodau-Lingue
Iatandza
Emien
Aniegre blanc
47
170
273
227
301
2
23
224
25
45
151
36
43
16
1
102
505
18
17
84
23
106
29
28
10
21
40
261
15
239
7
14
104
1
36
599
48
1
122
3
116
14
2
25
118
217
66
55
8
17
161
695
34
15
104
47
106
52
71
43
40
67
562
12
45
28
38
216
5
81
33
35
106
19
51
50
38
123
289
1
5
24
1
97
33
38
37
45
54
310
4
40
11
18
156
4
38
293
96
1
142
3
340
18
0
653
1
479
106
60
5
23
0
28
176
33
140
1
46
103
104
54
59
2
182
74
622
20
44
73
1
69
63
61
52
57
103
448
10
1
25
15
75
9
77
497
50
59
291
1
109
65
0
773
49
416
223
113
20
45
3
57
278
15
35
198
186
99
229
10
94
81
24
14
237
11
42
1
34
3
772
22
30
24
1
22
12
6
22
34
32
93
1
0
8
41
8
79
19
5
10
17
35
2
49
44
0
25
135
63
39
55
74
15
71
21
734
9
19
40
154
117
67
64
158
82
367
37
32
55
22
53
38
56
46
24
26
56
4
177
33
1
63
7
43
13
50
6
34
2
84
49
2
692
6
247
57
61
28
21
45
33
294
45
565
80
7
27
81
44
30
72
125
46
196
11
14
15
43
40
15
58
17
15
16
80
3
261
26
Ako
Melegba
Pocouli
Aiele
Fromager
Lohonfe
Ba
Iroko
Aniegre rouge
Faro
Movingui
Tiama
Kosipo
Aboudikro
Sipo
Tali
Pouo
Vaa, (Limbali)
Bosse
Amazakoue
Niangon
Kekele
Acajou
Azobe
Dibetou
Bete
Bahia
Difou
Badi, (Bilinga)
Kotibe
Asamela
Dabema
Koto
Ilomba
Kondroti
Eho
Akossika
Bi
Lotofa
Framire
Frake
Makore
Samba, (Obeche)
Avodire
1693
132
114
299
52
223
120
78
96
55
134
596
12
22
40
59
113
20
81
798
44
63
348
2
111
95
92
1546
70
693
140
266
14
52
1
160
413
59
3735
2
1061
15
377
55
103
6
21
1
59
173
28
521
1
1232
81
15
182
4
254
41
0
942
12
614
122
154
15
34
1
102
277
35
1214
1
1394
94
81
239
12
195
93
78
74
92
153
457
56
7
50
72
74
26
140
181
59
42
258
8
156
123
1
963
33
892
212
380
37
63
80
186
938
33
1227
4053
0
1
1061
1
44
2
27
11
31
43
1
143
10
0
533
4
143
32
61
24
6
15
7
80
17
129
66
124
Appendices
Other species
North-West region
3NU
4NW
SNU
species
vern. name
1NU
2NU
alb.adi
alb.zyg
amp.pte
ano.kla
ant.fra
bei.man
bom.buo
bos.pha
bus.occ
caI.aub
cel.zen
chi.san
chryssp
cor.pac
cor.pU
det.sen
dia.aub
did.ida
ery.man
gil.spl
gym.zai
irv.gab
kea.bri
kla.gab
lan.wel
man.obo
mar.gla
old.afr
ong.gor
par.bic
par.exc
pen.mac
pet.mac
phy.afr
pte.hyl
sac.gab
sap.aub
sch.arb
scy.tie
ste.acu
tre.afr
uapacsp
xyl.eva
Bangbaye
Ouochi
Lati
Bodioa
Adomonteu
Atiokouo
Kapokier-Oba
Daocou
Nomotcho
Guepizou
Asan
Bala
Longui
Gaouo
82
152
121
191
119
68
40
86
68
74
138
275
220
73
157
56
74
297
266
105
215
79
246
160
84
Fou
Aramon
Dantoue
Kouero
Lo
Sougue
OvaIa
Abale
Abrahassa
Koframire
Akouapo
Cocoti
OuaIio
Moussangoue
Fara
Bleblendou
Rikio
Tchebuessain
7NW
51
60
128
262
171
127
98
220
19
107
189
86
68
87
134
27
36
7
17
34
151
65
Bon
Bodo
Kropio
Broutou
Dabe
Medjilagbaagros
Zaizou
Boborou
Kohaingue
Kroma
Loloti
6NW
270
428
30
141
20
6
211
51
315
Centre-South region
6CS
7CS
106
76
82
205
42
83
47
47
70
182
114
152
129
45
150
34
49
19
39
fruits
40
41
75
68
111
93
167
413
129
93
415
169
188
856
57
48
61
244
196
152
328
132
188
117
157
73
123
322
233
32
356
61
45
66
246
249
206
344
162
193
62
250
444
107
256
205
98
90
213
57
274
114
95
726
64
567
611
38
75
52
10
139
13
180
85
277
52
96
86
639
331
398
85
430
228
45
213
56
123
78
37
28
47
240
300
78
11
20
34
15
11
205
142
200
83
155
118
390
115
104
137
166
332
136
primary species
other species
13910
4644
5303
2849
8260
4837
4258
4773
6740
2761
13018
4791
4057
1150
4255
4138
2671
3349
all species
area
% degraded forest
18554
215038
9031
8152
17809
9501
13097
144737 171417 134658 139950 229601
5207
90632
8393
118208
6020
94475
vol/ha70+
10%
23%
12%
5%
11%
10%
25%
N/A
N/A
86
56
76
67
68
78
57
71
64
125
SODEFOR InventairePerimetreXV (Clement 1973): timber volume above 70cmd per inventory compartment
foret densesur sol ferme (marecages et degrades exclus)
species vernacular name
afz.bel
alb.fer
amp.pte
ani.rob
ano.kla
ant.fra
ant.tox
aub.ker
ber.con
ber.occ
can.sch
cei.pen
eel.ado
cel.mil
cel.zen
chlor sp
cop.sal
dan.thu
det.sen
dis.ben
ent.ang
ent.can
ent.cyl
ent.uti
ery.ivo
giI.pre
gua.ced
gui.ehi
gym.zai
her.uti
kan.gue
kha.ant
kla.gab
lop.ala
lov.tri
mam.afr
man.alt
mit.cil
nau.did
nes.pap
par.exc
pet.mac
pip.afr
pte.mac
pyc.ang
rho.bre
ric.heu
sco.kla
ste.rhi
ter.ivo
ter.sup
tie.hec
tri.scl
tur.afr
zan.gil
Azodau-Lingue
Iatandza
Lati
Aniegre blanc
Bodioa
Adomonteu
Ako
Kodabema
Melegba
Pocouli
Aiele
Fromager
Lohonfe
Ba
Asan
Iroko
Etimoe
Faro
Bodo
Movingui
Tiama
Kosipo
Aboudikro
Sipo
Tali
Vaa, (Limbali)
Bosse
Amazakoue
Zaizou
Niangon
Kantou
Acajou blanc
Kroma
Azobe
Dibetou
Djimbo
Bete
Bahia
Badi, (Bilinga)
Kotibe
Sougue
Abale
Dabema
Koto
Ilomba
<ondroti
Eho
Akossika
Lotofa
Framire
Frake
Makore
Samba, (Obeche)
Avodire
Bahe, (Olon-dur)
Commercial species:
surface tot: foret dense sol ferme
surface echantillonne
% marec. foret dense du secteur
% degradee/cultiveedu secteur
1XV
2XV
3XV
4XV
5XV
31
29
34
14
25
375
20
17
21
184
10
92
3
2
31
42
51
25
108
150
37
12
75
50
35
196
15
33
6
13
8
24
12
67
15
12
6
46
8
98
19
24
1
2
11
7
4
4
19
2
6
2
2
11
6
5
32
17
37
60
104
13
18
43
96
3
30
1
10
1
1
1
9
21
43
11
21
9
34
17
28
299
1
5
12
74
178
7
32
201
156
42
1
2
5
117
13
108
54
228
1
148
81
14
35
7
25
133
18
93
30
5
42
6
37
25
19
30
29
320
0
33
1
138
88
2
27
338
4
54
9
4
40
212
21
144
28
427
2
272
130
63
19
8
109
176
17
11
1
17
0
12
12
3
5
16
79
0
9
1
53
17
0
17
105
23
9
1
5
2
22
6
20
32
213
1
116
22
28
6
21
14
73
1
12
0
2
6
1
1
6
4
10
10
0
3
1
12
2
0
1
18
5
3
1
1
1
2
1
8
18
71
0
11
5
7
1
1
0
1
1
9
13
32
5
17
7
19
10
20
117
0
3
33
12
12
1
4
140
23
7
6
1
5
42
19
50
85
212
3
126
46
19
9
24
10
59
7
438
1
143
0
58
1
106
1
22
23
6
2
17
3162
4231
1480
376
1792
70963 87519 18511
569
12
18
755
7
21
147
7
67
5817 34138
27
5
85
278
13
53
Appendices
126
From the Liberian inventory (Sachtler 1968)Iused timber volumes per 100ha of trees exceeding
40 cmdiameter.
Liberia National Forest Inventory (Sachtler 1968): timber volumes above40cmdinm3/100ha
Krahn-Bassa
Crebo
Sapo
species
code
G51
G52
G53
K61
K62
K63
K64
K65
K66
K67
K68
K69
K167
S91
S92
S93
afz.bel
aIs.boo
amp.pte
ano.kla
ant.fra
ant.tow
ara.soy
aub.pla
bei.man
berlisp
bra.leo
calposp
can.sen
cassisp
cei.pen
celtisp
chlorsp
chryssp
cry.tet
cyn.ana
daniesp
dia.aub
did.bre
did.ida
dis.ben
entansp
ery.ivo
fagarsp
funtusp
gil.pre
gua.ced
gui.ehi
hap.mon
her.uti
khayasp
kIa.gab
loe.kal
lop.ala
lov.tri
mam.afr
mit.ciI
monopsp
nau.did
nes.pap
newtosp
old.afr
ong.gor
pac.sta
par.bic
parinsp
pet.mac
pip.afr
pyc.ang
qua.und
rho.bre
ric.heu
sac.gab
ter.ivo
ter.sup
tet.tub
tie.hec
tri.scl
tur.afr
uapacsp
AFZ
ALS
AHP
ANO
ANT
ANA
ARA
AUB
13
77
65
80
201
90
36
30
53
74
0
192
111
0
86
47
40
50
95
200
60
22
12
31
62
0
126
122
1
98
15
32
55
108
187
66
10
4
22
16
0
135
143
1
195
389
138
21
130
8
265
42
7
256
79
121
190
521
332
451
521
2
104
548
116
29
153
7
254
4
35
557
73
141
112
299
155
377
522
0
109
420
779
52
26
303
35
37
0
75
185
8
38
194
155
41
163
184
16
146
13
100
221
2
465
98
0
7
13
61
138
28
4591
45
0
0
104
35
220
0
198
11
98
3
43
47
0
245
85
35
13
4
463
49
34
1739
2
0
1
293
85
120
0
110
7
134
0
106
46
0
448
76
57
24
4
260
18
61
1660
67
0
14
333
82
93
0
41
134
157
0
351
202
0
0
0
220
0
39
1615
0
0
0
205
588
0
86
268
531
96
17
370
71
60
0
40
118
14
14
228
275
63
284
49
147
6
481
0
37
508
953
25
3
235
3
49
0
28
82
25
0
162
204
71
254
514
10
91
152
160
18
123
84
175
15
18
28
39
102
268
265
271
529
626
42
113
0
14
0
125
25
0
0
0
52
34
0
1570
286
0
0
0
0
0
0
58
78
0
0
8
0
0
157
0
0
0
0
0
0
8
0
36
1324
1186
63
0
240
0
132
0
108
91
0
48
135
34
0
185
68
0
0
0
141
0
0
406
0
0
0
67
48
24
113
143
0
5
11
76
9
33
181
0
3
325
37
197
0
0
0
94
83
0
0
0
4
14
0
635
5
0
0
0
0
0
0
366
0
9
230
60
0
0
84
10
0
75
0
0
0
1011
0
59
1488
185
68
21
172
0
55
0
38
82
0
0
71
129
0
50
8
0
0
0
76
0
0
4688
23
0
0
22
171
33
62
748
0
0
0
188
28
55
163
0
1
423
50
180
25
48
11
142
187
0
5
0
7
5
116
874
0
0
17
0
0
31
0
259
95
45
516
7
0
0
108
0
0
137
0
0
0
991
0
69
859
1194
71
68
516
0
214
0
119
165
0
2
141
92
0
51
5
11
0
0
594
9
0
1548
0
0
0
144
184
216
113
444
76
5
7
92
36
94
276
0
3
285
60
104
2
24
10
139
140
0
0
0
139
27
130
724
181
0
148
0
32
5
0
654
26
4
0
118
10
0
148
8
8
206
0
0
0
451
0
54
1039
721
11
42
11
16
3
27
260
174
22
100
178
7
21
0
71
123
103
0
0
0
14
9
40
829
62
0
70
0
0
9
0
402
62
23
268
66
0
0
55
45
20
342
0
0
0
1272
0
158
2104
57
0
124
118
0
364
0
95
161
16
2
122
98
23
194
23
50
37
26
426
277
13
19
429
12
4
28
29
84
109
10
6
7
17
28
235
765
94
34
82
14
26
2
0
284
66
86
159
10
14
37
186
6
4
207
0
0
0
613
2
113
1587
675
57
36
40
121
331
17
306
144
20
17
288
10
0
1
58
35
135
11
18
0
45
97
560
456
67
95
249
0
35
11
8
353
50
181
6
9
1
85
153
103
126
869
0
0
0
754
3
91
797
296
67
18
140
113
37
6
68
107
38
20
242
137
138
512
202
151
16
27
37
162
102
782
35
0
3
484
12
0
46
105
44
0
70
0
24
8
75
307
77
222
9
16
63
7
12
43
75
10
14
21
33
176
47
2
3
23
42
N/A
688
123
10
27
30
10
30
5
290
23
162
0
1
4
185
201
6
0
964
3
0
0
1167
14
254
1281
478
22
89
29
6
26
133
7
0
35
0
41
42
156
218
94
157
0
9
26
10
200
56
82
9
2
32
45
135
42
10
3
29
93
241
418
89
0
121
0
11
40
1
0
27
143
0
0
21
93
111
8
10
773
6
0
0
665
9
132
881
318
92
18
34
385
68
0
24
193
80
23
254
160
53
322
164
43
18
14
38
9
39
328
29
0
3
199
13
24
81
49
84
2
38
2
142
50
680
238
185
305
24
9
29
1
107
51
76
7
16
15
15
140
18
22
16
52
24
0
759
99
0
54
0
13
36
6
28
54
142
0
4
4
106
154
12
0
857
17
0
0
911
12
187
N/A
548
93
20
128
124
154
0
31
257
58
37
161
200
56
222
83
44
23
0
90
2
25
446
29
0
9
151
218
53
52
1096
47
0
1
230
288
60
0
705
0
0
0
169
329
0
0
810
0
0
0
190
miscellaneous
1813
1659
1722
1926
1916
2221
1696
1956
1216
1263
1578
1398
1695
1443
548
1698
Totalvol ume
9207 9021
11076 11313
8374
BEI
BER
BRA
CAL
CAN
CAS
CEI
CEL
CHL
CHR
CRY
CYN
DAN
DIA
TOU
DID
DIS
ENT
ERY
FAG
FUN
GIL
GUA
GUI
HAP
TAR
KHA
KLA
LOE
LOP
LOV
MAM
MIT
HON
NAU
NES
NEW
OLD
ONG
PAC
PAK
PAR
COM
PIP
PYC
HAN
BOM
RIC
SAC
TEI
TES
TET
DUM
TRI
TUR
UAP
8435
8697
9024
81
480
36
0
65
327
59
38
221
114
7
191
12852
11098 11624
13843
502
0
345
0
10450
251
5
61
0
20
65
12
0
196
250
41
197
10086 7560 7957
127
Appendix III Detailed description of the physiographic units of the three study sites Zagne,
Tai and Para by G.J. vanHerwaarden (1991a), Rademacher (1992)and
Nooren (1992)
The distribution of thedifferent soil characteristics is shown in thephysiographic soil maps
together with thelegends. Adetailed description of thefieldwork and of the mappingunits is
givenhere.
Soilprofilepits andsoilaugerings (van Herwaarden 1991a).Soil profiles weredescribed along
soil transects according to the FAO (1977) guidelines for soil profile description. Colours were
defined with the use of the Munsell (1975) colour charts. Apreliminary classification of the soils
was made according to the revised FAO-UNESCO system (van Kekem 1984, FAO 1988).
Altogether 10soil profile pits were described inthe Zagne'survey area, 8intheTai survey area
and 11in the survey area near Para. In all, 186augerings were described in theZagne"survey
area, 261 in the Tai survey area and 115 inthe Para survey area. The density of observation
points is 1.4 per ha or 0.3 per cm2 onthe Zagne"area map, 2.0 per ha or 0.5 per cm2 on theTai
area map and 1.7 per ha or 0.4 per cm2 onthe Para area map. Line cutting, digging of thepits,
augering and most of the slope measurements weredone outby local personnel.
Soilanalyses.In the survey area of Tai the studies of Fritsch (1980) and Fraters (1986) wereused
asthebasis and source for soil descriptions and soil analytical data. For the other survey areas,
eight additional soil profile pits were sampled and analysed by the soil laboratory of theDirection
et Contrdle des Grands Traveaux in Bouake\ C6te d'lvoire. Analyses concerned texture, acidity,
organic matter content, total nitrogen content, cation exchange capacity and exchangeable bases
per horizon (for more details seevan Herwaarden 1991a).
1 The legends
To distinguish the mapping units four major differentiating characteristics are used: physiography,
drainage, presence of hardened plinthite and depth of rotten rock (saprolite). Bymeans of these
easily recognizable characteristics the boundaries havebeen drawn. The description of the units is
refined on the basis of diagnostic characteristics. These aretexture, colour, slopeform and
degree, parent material and altitude.
In each mapping unit representative soilshave been classified according to the FAO-UNESCO
system (FAO, 1988),the Soil Taxonomy (Soil Survey Staff, 1987) and the french system of soil
classification (CPCS, 1967). Classifications arepartly based on analytical data of some soil
profiles (see appendix 1in van Herwaarden 1991a). These data aretaken as representative for
other soil types as well. Therefore it is assumed thatbase saturations in the soils are far lessthan
50% and that the cation exchange capacities of the non-alluvial soils are lessthan 24 cmol (+) per
kg clay in at least somepart of the B-horizon within 125cmof the surface. This assumption is
also based on results from prior investigations intheTai region by de Rouw et al. (1990), Fritsch
(1980), Fraters (1986) and Development and Resources Corporation (1967).
In Figures 38, 40 and 42 the legends of the four survey areas are shown with the mapping units
and thedifferentiating and diagnostic characteristics in ahierarchical order (as far aspossible). It
should be noted that all these characteristics interact and cannot be seen in isolation. To distinguish the units of one survey area from thoseof another theprefixes of the codes are different: Z
for the survey area Zagne" 1,T for the Tai survey area and P for the Para survey area.
128
Appendices
2Description ofthemappingunits
For each of the four maps the units are described by means of the characteristics as mentioned
above. Thetotal extent of each unit isgiven inhectares and as apercentage of thetotal area of
the corresponding survey (see Table 4). For Tai and Zagn6this area islarger than the sampleplot
and its surroundings which are displayed in Figures 38to 42. For most of the units atypical soil
profile description canbe found invan Herwaarden (1991a).
2.1 Mapping units of thesurvey area Zagne
ZC: Crest
Total extent: 8ha (5%)
Thehighest part of the survey area Zagne"does not reveal the features of a real crest and its
physiography is notvery pronounced. Boulders of ironstone are found at the surface, possibly
remnants of an ancient ironstone crust. Cemented ironstone is absent inthe soil profiles of this
unit. The red, well drained soils are situated ongently slopingslopes They aregravelly (ironstone) throughout theprofile. Clay ratios clearly increase with depth within the profile and an
argic kandic B-horizon ispresent. The soils are classified as Ferric Acrisol (FAO/UNESCO), as
Kanhaplic Haplustult (Soil Taxonomy) and as Sol ferralitique fortement desaturd, remanie"modal,
sur migmatites (CPCS).
ZS1: Upper slope
Total extent: 36ha (25%)
This unit covers the largest part of the survey area Zagne\ It isfound around the crest. The soils
are strong brown and well drained. Their texture isgravelly (ironstone) clay loam. The soils are
situated onvery gently sloping convex slopes. They are classified as Haplic Ferralsol
(FAO/UNESCO), asKanhaplic Haplustult (Soil Taxonomy) and as Sol ferralitique fortement
desature\ remanie\ modal, sur migmatites (CPCS).
ZS2: Middle slope
Total extent: 29 ha (20%)
Thisunit concerns the southern middle slopeof the survey area Zagne\ The yellowish brown soils
are moderately well drained and consist mainly of colluvium. Gravel and clay contents seem to
increase with depth within the profile. In the lower part of the soil profile red mottles are present.
They are associated with plinthite. The soils are classified as Plinthic Acrisol (FAO/UNESCO), as
Plinthic Haplustult (Soil Taxonomy) and as Sol ferralitique fortement desaturg, remanig, modal sur
colluvions recouvrant les alterationsde migmatite (CPCS).
ZS3: Lower slope
Total extent: 22 ha (15%)
Thisunit borders thevalley bottoms in the North and in the South of the survey area Zagne\ The
whiteto light yellowish brown soils are imperfectly drained. The slopes are concave and gently
sloping. Rotten rock (saprolite) ispresent below adepth of 100cm. The soils are non-gravelly.
They are classified as Haplic Acrisol (FAO/UNESCO), asTypic Kandiustult (Soil Taxonomy) and
as Sol ferralitique fortement desarur6, rajeuni, hydromorphe, sur colluvions recouvrant les
alterations de migmatite (CPCS).
129
ZV1: Ravine/gully
Total extent: 4 ha (3%)
Theupper parts of the ravines/gullies are gently slopingwhereas the lower are almost flat.
Locally a cascade ispresent onthe border of these two. In the steep, 1 to 5 mhigh walls rotten
rock below hardened plinthite isobserved. Thetextureof the soils varies from very gravelly
(ironstone) clay loam intheupper parts to non-gravelly coarse sand to sandy clay inthe lower
parts of this unit. The soils are classified as Ferralic Cambisol (FAO/UNESCO), asTypic
Kandiustult (Soil Taxonomy) and as Sol ferralitique fortement desaturg, rajeuni, avec erosion, sur
migmatites (CPCS).
ZV2: Valley bottom
Total extent: 20ha (14%)
In the survey area Zagngtwo valley bottomshavebeen mapped. They are almost flat and
sedimentation occurs here. Thethickness of the alluvium however israrely more than 100cm.
Rotten rock ispresent below the alluvial deposits. Thegrey soils arepoorly drained. They are
classified as Dystric Fluvisols (FAO/UNESCO), asTropaquent (Soil Taxonomy) and as Solpeu
evolue'd'origine non climatique d'apport hydromorphe sur alluvions (CPCS).
2.2 Mapping units of theTai survey area
TC: Crest
Total extent: 0.4 ha (0.3%)
This unit is confined to the highest topographic positions inthe northern part of the TaTsurvey
area. It concerns small, nearly flat tops, isolated from the surrounding landscape by relatively
steep short slopes. The red soils arewell drained and very gravelly. Thegravel consistsof
reddish, hard, pea shaped, ferruginous nodules (ironstone). Thetextureof thefine earth fraction
changes from sandy loam inthe humus rich topsoil to clay in the subsoil. At adepth of about 70
cm a layer of continuously indurated ironstone ispresent, in which individual concretions canbe
recognised. The crust can only be broken with ahammer. Stones and boulders of ironstone are
found on the surface. The soils are classified asFerric Acrisol (FAO/UNESCO), as Orthoxic
Palehumult (Soil Taxonomy) and as Sol ferralitique fortement desature\ indur^remaniSsur
migmatites (CPCS).
TS1: Upper slope
Total extent: 5ha (4%)
Thisunit islocated on relatively steep slopes adjacent to the highest crests. The red soils arewell
drained and very gravelly (ironstone). Gravel ratios decrease with depth within the profile. An
argic/kandic B-horizon has been recognised and thetextureof thefineearth fraction is sandy loam
to sandy clay loam inthetopsoil and clay in the subsoil. The soils are classified as Ferric Acrisol
(FAO/UNESCO), as Orthoxic Palehumult (Soil Taxonomy) and as Sol ferralitique fortement
d6sature\ remaniSmodal sur migmatites (CPCS).
TS2: Middle slope
Total extent: 23 ha (17%)
This mapping unit has some similarities with theprevious (TS1) but islocated on lower topographic positions (outsidethe survey area thisunit locally changes into a sloping crest remnant).
The soils are therefore more yellowish. They arewell drained. Thetexture consists of very
gravelly sandy clay loam to clay over non-gravelly clay. The soils are classified asFerric Acrisol
(FAO/UNESCO), as Orthoxic Palehumult (Soil Taxonomy) and as Sol ferralitique fortement
d6sature\ faiblement appauvri sur migmatites (CPCS).
130
TS3: Lower slope
Appendices
Total extent: 56ha (42%)
This isthe most extensivemapping unit of theTaisurvey area. The yellowish brown soils are
moderately well drained. The soil material consist of colluvium, thetopsoil beingvery gravelly
sandy clay to clay and the subsoil non-gravelly sandy clay loam to clay. An argic/kandicBhorizon ispresent. Inthe subsoil red mottles and soft nodules (both associated with plinthite) are
observed inthe subsoil. Thered mottling can be so abundant that it dominates the matrix colour.
The soils are classified asPlinthic Acrisol (FAO/UNESCO), asPlinthustult (Soil Taxonomy) and
as Sol ferralitique fortement desature'remante modal sur colluvions recouvrant les alterations de
migmatite(CPCS).
TS4: Lower slope
Total extent: 16ha (12%)
Thisunit concerns the lower slopesparts as well, but is confined to the areas adjacent to the
valley bottom. The yellowish brown soils are moderately well drained. The soil material consists
of colluvium. In the subsoil avery gravelly layer consisting of hardening plinthite (petroplinthite)
isfound. Clay ratios do not clearly increase with depth within the profile. An oxic/ferralic Bhorizon ispresent. The soils are classified as Xanthic Ferralsol (FAO/UNESCO), as Tropeptic
Haplortox (Soil Taxonomy) and as Sol ferralitique fortement desaturg, indur6, appauvri,
hydromorphesur colluvions recouvrant les alterationsde migmatite (CPCS).
TV: Valley bottom
Total extent: 33ha (25%)
This mapping unit coversthe lowest parts of theTai survey area. Because of thisposition the soils
are poorly drained and white and greyish colours dominate. Mottling is frequently present. The
textureof the soils isnon-gravelly sandy loam to loamy sand. At adepth of approximately 80cm
a stone lineof angular quartz fragments ispresent. The soils are classified asDystric Gleysol
(FAO/UNESCO), asTropaquent (Soil Taxonomy) and as Sol hydromorphe peu humifere, a
amphigley, anappe phr£atiqueprofond sur alluvions(CPCS).
Gully/ravine
Gullies/ravines have not been separately mapped inthe Tai survey area but arepresent from the
middle slopesto thevalley bottom. They are deeply incised with locally 2to 5mhigh vertical
walls. In these wallshardened plinthite ispresent, often over rotten rock (saprolite).
2.3 Mapping units of thePara survey area
PCI: High crest
Total extent: 2ha (3%)
These crests form thehighest parts of thePara survey area. They are convex and gently sloping
(2-6%). The well drained soils found here aregravelly (tovery gravelly) with avery clayey
matrix. The gravel consists of pisolitic ironstone. At a depth of 125-150 cm ahorizon consisting
of cemented ironstone ispresent. This hardened horizon ismore distinct on the western than on
the eastern high crest. Rotten rock of schist (assumed to be theparent material) is not found
within 150cm depth. In comparison with other units organic matter contents arehigh (see
appendix 1invan Herwaarden 1991a). Tree growth islimited by the extremely hostile rooting
conditions (caused by the high gravel content and the presence of the ironstone crust). Therefore
small short trees are dominant. Larger trees topple over easily. Fallen trees create gravel holes
and mounds resulting in aspecial micro topography. With respect to the FAO/UNESCO
131
classification system, aferralic B-horizon ispresent. With respect to the soil classification
according to the Soil Taxonomy, thepresence of a kandicB-horizon isnoted. Thishorizon meets
the weatherablemineral requirements of an oxic horizon. The soils are classified as Haplic
Ferralsol (FAO\UNESCO), asTypic Kandiustox (Soil Taxonomy) and as Sol ferralitique
fortement d£satur6, remaniSfaiblement indur6sur schistes (CPCS).
PC2: Low crest
Total extent: 3ha (4%)
This unit consists of small and low summits. They are gently sloping (2-6%) and convex. Their
strongbrown, well drained soils contain large amounts of ironstonegravel. Thegravel is absent at
a certain depth (approximately 1m) where rotten rock of schist isfound within 150cm depth.
Thepresenceof aferralic or oxic horizon is notnoted, but an argic/kandic B-horizon ispresent.
The soils are classified as Ferric Acrisol (FAO\UNESCO), asTypicKandiustult (Soil Taxonomy)
and as Sol ferralitique fortement desatur£, rajeuni remanie' sur schistes (CPCS).
PC3: Shoulder
Total extent: 2ha (3%)
This unit has suffered erosion to such an extent that ironstone gravel and/or colluvium havebeen
removed and rotten rock (originating from schist) ispresent at shallow depth. Its physiography is
notvery clear and theunits appear asgently sloping (2-6%) protuberant parts, appearing like
shoulders. They are all located on barriers of two different watersheds. The soils are considered to
be rejuvenated. Nevertheless analytical data of rotten rock in aprofile on a low crest show very
low values of the base saturation and CEC (see appendix 1).The colours of these well drained
soils aredominated by the weathered schist and show ahigh variety (from red to yellow). They
are non-gravelly. The soils are classified asFerralic Cambisol, (FAO/UNESCO), as Typic
Kandiustult (Soil Taxonomy) and as Sol ferralitique fortement desature\ rajeuni avec erosion sur
schistes (CPCS).
PS1: Upper slope
Total extent: 7ha (10%)
This unit is clearly associated with the high crests. The moderately steep (13-25%) upper slopes
are located directly beneath thehigh crests. Ironstonegravel isabundantly present inthesoil
profile. In thisunit as inPCI relatively many fallen trees havebeen observed. This isprobably
caused by the difficult rooting conditions (because of thehigh contents of ironstonegravel) in
combination with the relative steep slopes. Charcoal has been found in someprofiles. Thewell
drained soils areto some extent similar to the soils onthe high crests butthey are considered to
be truncated here.
Furthermore they are not red but yellowish red. Rotten rock originating from schist has been
found at adepth of 125cm. The soils are classified as Haplic Ferralsol (FAO/UNESCO) asTypic
Kandiustox (Soil Taxonomy) and as Sol ferralitique fortement d6sature\ remanie rajeuni sur
schistes (CPCS).
PS2: Middle slope
Total extent: 35ha (49%)
This unit isthe largest of the Para survey area covering almost half of it. The topographical
position isbetween low crests and upper slopes onthe oneside and lower slopesonthe other.
The middle slopes are straight to slightly convex and sloping (6-13%). The soils consistof
colluvium (slightly gravelly clay loams) and are yellowish brown coloured. Rotten rock has not
been found within 150cm depth on middle slopes. Rooting conditions are good and in accordance
with observations by Vooren (1985) only few fallen trees are present inthisphysiographicunit. In
oneof the profiles faint orange brown mottleshavebeen observed, which possibly indicatethe
132
Appendices
beginning of plinthiteformation. In these well drained soils aferralic B-horizon (FAO/UNESCO)
ispresent. According to the Soil Taxonomy thishorizon isa kandichorizon but it meets the
weatherable mineral requirements of an oxichorizon. The soils are classified as Xanthic Ferralsol
(FAO/UNESCO), as Xanthic Kandiustox (Soil Taxonomy) and as Sol ferralitique fortement
d6sature\ typiquejaune recouvrement sur colluvions recouvrant les alterations de schiste (CPCS).
PS3: Lower slope
Total extent: 18ha (25%)
The lower slopes are situated between the middle slopes and the stream valleys. They are sloping
(6-13%) and concave. Because of their positionthey are moderately well drained and the
formation ofplinthiteappears within 125cmdepth. Plinthite ispresent inmostbut not all of the
soil profiles of the lower slopes. It is assumed that thisplinthite, which is easily recognisableby
its red colour isrelated with afluctuating water table. Probably it isformed by absolute accumulation of iron coming from thehigher situated ironstone crust. Thetexture inthe subsoil of the
profiles isvery clayey. Possibly some illuviationhas taken place as well asthe formation of
kaolinite. In some augerings in lower slopes
inthe western part of the Para survey area kaolinitehas been clearly recognised by its distinct
white colour. Parent material of the soils is colluvium. The soils are slightly gravelly and
brownish yellow coloured. The gravel consists mainly of quartz and little ironstone (transported
from higher parts of the landscape). Rooting islimited where plinthite isnear the surface and
hardened irreversibly. Thephenomenon of hardened plinthiteismore common inthe western part
of thePara survey areathan inthe central and eastern parts.Thesoils inthisunit are classified as
PlinthicFerralsol (FAO /UNESCO), as Plinthic Kandiustox (Soil Taxonomy) and as Sol
ferralitique fortement d6sature\ remanidjaune recouvrement sur colluvions recouvrant les
alterations de schistes (CPCS). In someprofiles plinthite is absent and soils canbe classified as
Xanthic Ferralsol (FAO/UNESCO) and as Xanthic Kandiustox (Soil Taxonomy). According to
CPCS the classification unit remains thesame.
PV1: Gully/Ravine
Total extent: 1 ha (1%)
This unit isthe smallest inthe Para survey area, covering approximately 1ha. Its reflection onthe
map is slightly exaggerated. Butthe soil map would notbe appropriate without indicating the
location of the strikinggullies/ravines. Notonlytheirphysiography withgently sloping (2-6%)
bottoms and very steep (>55%) walls (being 1-3 mhigh) but also their soils arevery different
from other units. No distinction has been made between the smaller gullies and the larger and
deeper ravines because thetransition between the two isvery gradual. Furthermore it wouldbe
impossibleto map them separately because of their small size. Because erosion isdominant inthis
unit, rotten rock originating from schist isat or near the surface, sometimes covered by avery
thin layer of deposits. The colours of the rotten rock arevariegated ranging from yellowish red to
oliveyellow. The texture of the soils is slightly gravelly (quartz, schist, and ironstonegravel) clay
loam. In the wallshardened plinthitecanbe found lying immediately on very slightly weathered
schist. Drainage is imperfect, but thegullies/ravines only contain water after heavy rainfall. The
soils are classified as Ferralic Cambisol (FAO/UNESCO), as Typic Kandiustult (Soil Taxonomy)
and as Sol ferralitique fortement d£sature\ rajeuni avec Erosionsur schiste(CPCS).
PV2: Higher valley bottom
Total extent: 2 ha (3%)
In the Para survey area one major valley bottom (also termed bas fond) ispresent, which is
supposed to contain flowing water throughout the year. Thevalley bottom is almost flat (0-2%).
This unit covers theupstream part of it. Sedimentation takes place here butthe valley bottom is
covered only by arather thin layer of alluvium. Rotten rock asfound inthe gullies/ravines is
133
present beneath it at a depth of 80to 120cm. Thetexture of the soilsvaries from clay to loamy
sand, with the latter predominating. Thedomination of sand indicates arather high stream
velocity which allowsonly the coarser particlesto sink. Fresh material isdeposited regularly.
Because of the poor drainage tree roots penetratethe soil onlyto shallow depths. Topplingof trees
istherefore notuncommon. Certain tree species however are adapted to these circumstances. The
alluviumhas a grey colour. Thesoils are classified asDystricFluvisol (FAO/UNESCO), as
Tropic Fluvaquent (Soil Taxonomy) and as Sol peu evolue" d'origine non climatique d'apport
hydromorphe sur alluvions recouvrant les alterationsde schistes (CPCS).
PV3: Lower valley bottom
Total extent: 1 ha (2%)
Thisunit is almost similar to the previous one, butbecause it covers thedownstream part of the
valley bottom (where the drainage basin is larger) conditions arewetter here. As aresult ofthis,
water is at or near the surface most of thetime and drainage isvery poor. Thesoils are reduced
and iron ismobile. Thedominant colour of the soil istherefore grey. In the upper part of thesoil
yellowish red mottles occur indicating where the air has penetrated indry seasonsthus oxidizing
the iron compounds. Thetextureof these soils is non gravelly loamy sand. Although the
morphological characteristics of the soils do not completely match the criteria of gleyic properties
the soils are classified as Dystric Gleysol (FAO/UNESCO), asTypicHydraquent (Soil Taxonomy)
and as Sol hydromorphe peu humifere agley ensemble sur alluvions (CPCS).
134
REFERENCES
For the Dutch reader: references are arranged in international alphabetical order. Names
beginning with "de" or "van" should be searched for under D and V respectively.
Adam J.G. (1983). Flore descriptive des Monts Nimba (Cote d'lvoire, Guinee, Liberia). Editions du Centre
National de la Recherche Scientifique, 2181 pp.
Adejuwon J.O., Balogun E.E. & Adejuwon S.A. (1990). On the annual and seasonal patterns of rainfall
fluctuations in Sub-Saharan West Africa. Int. J. of Climatology 10: 839-848.
Ahn P.M. (1970). West African soils. Oxford Univ. Press, 332 pp.
Ake Assi L. (1984). Flore de la C6te d'lvoire: etude descriptive et biogeographique, avec quelques notes
ethnobotaniques. These Doct., Universite d'Abidjan, Fac. des Sciences, Dept. Phys. Vegetale, 1206 pp.
Ake Assi L. & Pfeffer P. (1975). Etude d'amenagement touristique du Pare National de Tai. Tome 2: Inventaire
de la flore et de la faune. BDPA, Paris, 58 pp.
Albers P. (1990). Contribution to a diagnostic key for the high forest trees of the Tai National Pare (Ivory
Coast). Stageverslag, Dept. Plant Taxonomy, Univ. Wageningen.
Alder D. (1990). Ghafosim: a projection system for natural forest growth and yield in Ghana. Manas Systems
Ltd., Oxford, 114 pp.
Alder D. & Synnott T.J. (1992). Permanent sample plot techniques for mixed tropical forest. Tropical forestry
papers 25, Oxford Forestry Institute, Oxford, 124 pp.
Alexandre D.-Y. (1978). Le rdle disseminateur des elephants en foret de Tai, Cote d'lvoire. Rev. Ecol., La
Terre et la Vie 32: 47-71.
Alexandre D.-Y. (1980). Caractere saisonnier de la fructification dans une foret hygrophile de Cote d'lvoire.
Rev. Ecol. (Terre et Vie) 34: 335-350.
ANAM (1987). Les normales pluviometriques 1951-1980. Agence Nationale des Aerodromes et de la
Meteorologie, Abidjan-Port Bouet, 37 pp.
ASECNA (1979). Le climat de la Cote d'lvoire. ASECNA, Abidjan, 74 pp.
Ashton P.S. (1977). Variation of tropical moist forest with site: its relevance, and the methodological problems
that its study presents in tropical forest systems. In: Briinig E.F. (ed.). Joint MAB-IUFRO Rainforest
Ecosystem Workshop, Hamburg-Reinbek, May 1977, Chair of World Forestry, Hamburg-Reinbek.
Aubreville A. (1938). La foret coloniale. Les forets de l'Afrique occidentale francaise. Annales de l'Academie
des sciences coloniales, tome IX, Soc. d'editions geographiques, maritimes et coloniales, Paris, 243 pp.
Aubreville A. (1949). Contribution a la paleohistoire des forets de l'Afrique tropicale. Soc. d'editions geograph.,
mar. et colon., Paris, 99 pp.
Aubreville A. (1959). La flore forestiere de la Cote d'lvoire. Publ. CTFT no 15 (3 tomes), Nogent-sur-Marne,
1031 pp.
Aubreville A. (1962). Savanisation tropicale et glaciations quaternaires. Adansonia 2(1): 16-84.
Bagarre E. et Tagini B. (1965). Carte geologique de la Cote d'lvoire au 1/1.000.000°. SODEMI, Abidjan.
Bech N.J. (1983). La duree du cycle sylvigenetique en foret de Tai, C6te d'lvoire. These MSc, Univ.
Wageningen, 79 pp.
Begon M., Harper J.L. & Townsend C.R. (1986). Ecology. Individuals, populations and communities.
Blackwell Scientific Publ., Oxford, 876 pp.
Bertault J.-G. (1986). Etude de l'effet d'interventions sylvicoles sur la regeneration naturelle au sein d'un
perimetre experimental d'amenagement en foret dense humide de Cote d'lvoire. These de doctorat,
Universite de Nancy, Faculte des sciences, 254 pp.
Bertault J.-G. (1992). Comparaison d'ecosystemes forestiers naturels et modifies apres incendie en Cote d'lvoire.
Dans: Puig H. et Maitre H.F. (eds.). Actes de l'Atelier sur l'amenagement et la conservation de l'ecosysteme forestier tropical humide, 12-16 mars 1990, Cayenne. MAB/Unesco, MAB/France, IUFRO, FAO,
Paris, pp. 1-25.
Blokhuis W.A. (1993). Vertisols in the Central Clay Plain of the Sudan. Doctoral thesis, Agricultural
University, Wageningen, 418 pp.
Boddez P. (1989). Comparaison des deux types de foret: la structure, la composition et le sol dans le Pare
National de Tai, Cote d'lvoire. Rapport de stage I.A.H.L Velp.
Bolza E. and Keating W.G. (1972). African timbers. The properties, uses and characteristics fo 700 species.
Division of building research, CSIRO, Melbourne, 760 pp.
135
Bongers F., Popma J., Meave del Castillo J. & Carabias J. (1988). Structure and floristic composition of the
lowland rain forest of Los Tuxtlas, Mexico. Vegetatio 74: 55-80.
Bonnehin L. (1992). Importance des produits forestiers non ligneux pour la participation des populations locales
a l'amenagement de la foret dans la region de Tai, Cote d'lvoire. Tropenbos, in prep.
Bonnis G. (1980). Etude des chablis en foret dense humide sempervirente naturelle de Tai (Cote d'lvoire).
Rapport ORSTOM, Adiopodoume, 28 pp.
Borcard D., Legendre P. & Drapeau P. (1992). Partialling out the spatial component of ecological variation.
Ecology 73(3): 1045-1055.
Bormann F.H., Siccama T.G., Likens G.E. and Whittaker R.H. (1970). The Hubbard Brook ecosystem study:
composition and dynamics of the tree stratum. Ecological monographs 40: 377-388.
Bos P. (1964). Rapport de fin de leve des coupures Tai 4a-4c au 1/50.000. Rapport no. 127, Societe pour le
Developpement Minier de la Cote d'lvoire, Abidjan, 58 pp.
Bosman P. and Hall-Martin A. (1986). Elephants of Africa. Struik, Cape town, RSA, 120 pp.
Bousquet B. (1978). Un pare de foret dense en Afrique. Le Pare National de Tai (C6te d'lvoire). Bois et Forets
des Tropiques 179: 27-46 & 180: 23-37.
Bouys Ph. (1933). Le bas Cavally (Afrique Occidentale Francaise) et son avenir. Etude sur les confins francoliberiens de l'ancienne Cote du Mauvais Peuple. Imprimerie Mari-Lavit, Montpellier, 183 pp.
Bray J.R. & Curtis J.T. (1957). An ordination of the upland forest communities of southern Wisconsin.
Ecological Monographs 27(4): 325-349.
Brock M.R. and Chidester A.H. (1977). Geologic map of the Harper quadrangle, Liberia. United States
Geological Survey Misc. Inv. Ser. Map I-780-D, Reston, USA.
Bruijnzeel L.A. (1990). Hydrology of moist tropical forests and effects of conversion: a state of knowledge
review. Unesco, Paris, 224 pp.
Brunet-Moret Y. (1976). Etablissement d'un fichier pluviometrique operationnel et etude des averses
exceptionnelles. Application a la Cote d'lvoire. ORSTOM, CIEH, Montpellier, 14 pp.
Budowski G. (1965). Forest species in successional process. Turrialba 15(1): 40-42.
Burrough P.A. (1987). Spatial aspects of ecological data. In: Jongman R.H.G., ter Braak C.J.F. and van
Tongeren O.F.R. (eds). Data analysis in community and landscape ecology, pp. 213-252. Pudoc,
Wageningen.
Busby J. (1986). BIOCLIM user's manual version 2.0. Bureau of Flora and Fauna, Canberra.
Cailliez F. and Alder D. (1980). Forest volume estimation and yield prediction. FAO Forestry Paper 22/1 and
22/2, FAO, Rome.
Casenave A., Flory J., Guiguin N., Ranc N., Simon J.M., Toilliez J. et Tourne M. (1980). Etude hydrologique
des bassins de Tai. Campagnes 1978-1979. ORSTOM, Adiopodoume, 47 pp.
Casenave A., Flory J., Mathieux A. et Simon J.M. (1984). Etude hydrologique des bassins de Tai. Campagne
1981. ORSTOM, Adiopodoume, 85 pp.
Casenave A., Flory J., Ranc N. et Simon J.M. (1981). Etude hydrologique des bassins de Tai. Campagne 1980.
ORSTOM, Adiopodoume, 83 pp.
Chevalier A. (1909). Rapport sur une mission scientifique en Afrique occidentale. Recherches de 1906-1907 a la
Cote d'lvoire. Nouvelles Archives des Missions Scientifiques et Litteraires 18 (3): 73-82.
Chevalier A. (1948). Biogeographie et ecologie de la foret dense ombrophile de la C6te d'lvoire. Rev. int. de
Bot. appl. et d'Agr. trop. 28: 101-115.
Clement J. (1973). Inventaire forestier du Perimetre Industriel XV. CTFT, Nogent-sur-Marne, 68 pp.
Connell J.H. (1978). Diversity in tropical rain forest and coral reefs. High diversity of trees and corals is
maintained only in a nonequilibrium state. Science 199: 1302-1310.
Curtis J.T. & Mcintosh R.P. (1951). An upland forest continuum in the prairie-forest border region of
Wisconsin. Ecology 32: 476-496.
Davis T.W.A. & Richards P.W. (1933-34). The vegetation of Moraballi Creek, British Guyana: an ecological
survey of a limited area of tropical rain forest. J.Ecol. 21: 350-384; 22: 106-155.
de Bie S. (1991). Wildlife resources of the West African savanna. PhD thesis, Agricultural University,
Wageningen, 266 pp.
de Graaf N.R. (1986). A silvicultural system for natural regeneration of tropical rain forest in Suriname.
Doctoral thesis, Wageningen Agricultural University, 250 pp. Also published in the series: Ecology and
Management of Tropical Rain Forest in Suriname no. 1, Wageningen Agricultural University,
de Klerk M. (1991). Regeneration strategies of some emergent tree species in Cote d'lvoire. MSc thesis AV
91/29, Department of Forestry, Agricultural University, Wageningen, 60 pp.
de Koning J. (1983). La foret du Banco. PhD thesis, Section de Taxinomie et de Geographie botaniques, Univ.
Agronomique de Wageningen, 921 pp.
136
References
de Namur Ch. et Guillaumet J.L. (1978). Grands traits de la reconstitution dans le Sud-Ouest ivoirien. Cahiers
ORSTOM, ser. Biologie 13 (3): 197-201.
de Rouw A. (1991). Rice, weeds and shifting cultivation in a tropical rain forest. A study of vegetation
dynamics. PhD thesis, Agricultural University, Wageningen, 263 pp.
de Rouw A., Vellema H.C. & Blokhuis W.A. (1990). Land unit survey of the Tai region, south-west C6te
d'lvoire. Tropenbos Technical series 7, Tropenbos, Ede, 222 pp.
Devineau J.-L. (1976). Principales caracteristiques physionomiques et floristiques des formations forestieres de
Lamto (moyenne Cote d'lvoire). Ann. Univ. Abidjan, serie E. (Ecologie), 9: 274-303.
Devineau J.-L., Lecordier C. & Vuattoux R. (1984). Evolution de la diversite specifique du peuplement ligneux
dans une succession preforestiere de colonisation d'une savane protegee des feux (Lamto, C6te d'lvoire).
Candollea 39: 103-134.
DRC (1967a). Forestry resources of the Southwest region (Ivory Coast). Report to the government of the
Republic of the Ivory Coast, Development and Resources Corporation, New-York, 60 pp.
DRC (1967b). Soil survey of the Southwest Region (Ivory Coast). Development and Resources Corporation,
New York, 2 volumes.
Dudek S., Forster B., Klissenbauer K. (1981). Lesser known Liberian timber species. Desciption of physical and
mechanical properties, natual durability, treatability, wordability and suggested uses. GTZ, Eschborn.
Durand Y. (1985). Nomenclature des essences ivoiriennes. Noms vernaculaires, noms commerciaux, noms
scientifiques. (Mise ajour 1985). CTFT-CI, Abidjan.
Eldin M. (1971). Le climat. Dans: Avenard et al. Le milieu naturel de la Cote d'lvoire, Memoires ORSTOM
no. 50, pp. 73-108.
Ellenberg H. (1979). Zeigerwerte des GefaBpflanzen Mitteleuropas. Scripta Geobotanica 9, 2. Aufl., Gottingen,
122 pp.
Endler J.A. (1982). Pleistocene forest refuges: fact or fancy. In: Prance G.T. (ed.). Biological diversification in
the tropics: 641-657. Columbia Univ. Press, New York.
Faber-Langendoen D. and Gentry A.H. (1991). The structure and diversity of rain forests at Bajo Calima,
Chocd region, western Colombia. Biotropica 23(1): 2-11.
Fanta J. (1985). Groeiplaats: onderzoek, classificatie en betekenis voor de bosbouw. Nederlands Bosbouwtijdschrift 57 (10-11): 333-347.
FAO (1977). Guidelines for soil profile description. FAO, Rome, 66 pp.
FAO (1988). FAO/Unesco Soil map of the World, revised legend. World resources report 60, FAO, Rome,
138 pp.
FDA (1990). Annual Report 1989. Forestry Development Authority, Monrovia, 32 pp.
Force E.R. and Beikman H.M. (1977). Geologic map of the Zwedru quadrangle, Liberia. United States
Geological Survey Misc. Inv. Ser. Map I-777-D, Reston, USA.
Fraters D. (1986). A study of a catena in the Tai forest, Ivory Coast. MSc thesis, Agricultural University,
Wageningen, 64 pp.
Fritsch E. (1980). Etude pedologique et representation cartographique a 1/15.000 eme d'une zone de 1.600 ha
representative de la region forestiere du Sud-Ouest ivoirien. Rapport ORSTOM, Abidjan, 137 pp.
Fritsch J.M. (1992). Les effets du defrichement de la foret amazonienne et de la mise en culture sur l'hydrologie
de petits bassins versants. Operation ECEREX en Guyane francaise. Collection Etudes et Theses,
ORSTOM, Paris, 392 pp.
Gartlan J.S., Newbery D.McC, Thomas D.W. & Waterman P.G. (1986). The influence of topography and soil
phosphorus on the vegetation of Korup Forest Reserve, Cameroun. Vegetatio 65: 131-148.
Gaussen H. (1954). Theorie et classification des climats et microclimats. 8 eme Congres International Botanique
Paris, Sect. 7 et 3, pp. 125-130.
Gautier L. (1989). Le contact foret-savane a Lamto. Bull. Soc. bot. Fr. 136, Actual, bot. 3/4: 85-95.
Gentry A.H. (1982). Patterns of neotropical plant species diversity. Evolutionary Biology 15: 1-84.
GFML (1967). Inventory of Grebo National Forest. German Forestry Mission to Liberia, Technical report no. 5,
Monrovia, 54 pp.
Ghartey K.K.F. (1989). Results of the inventory. In: Wong J.L.G. (ed.). Ghana forest inventory project.
Seminar proceedings 29-30 March 1989, Accra, pp. 32-46.
Gillison A.N. & Brewer K.R.W. (1985). The use of gradient directed transects or gradsects in natural resource
surveys. Journal of Env. Manag. 20: 103-127.
Gillman L. & McDowell R.H. (1973). Calculus. Norton & Cie, New York, 674 pp.
Gleick J. (1991). Chaos : de derde wetenschappelijke revolutie. Contact, Amsterdam, 314 pp.
Gliick P. (1987). Das Wertsystem der Forstleute. Cbl. ges. Forstwesen 104 (1): 44-51.
137
Goodall D.W. (1954). Vegetational classification and vegetational continua. Angew. Pflanzensoziologie, Wien.
Festschrift Aichinger 1: 168-182.
Gornitz V. & NASA (1985). A survey of anthropogenic vegetation changes in West Africa during the last
century - Climatic implications. Climatic Change 7: 285-325.
Grace J. (1989). Pattern analysis for forest ecology. In: Schmidt P., Oldeman R.A.A. & Teller A. (eds.).
Unification of European forest pattern research. Proceedings of a ESF-FERN Workshop, Strasbourgh,
France, 24-26 April 1989. Pudoc, Wageningen.
Gregory S. (1965). Rainfall over Sierra Leone. Dept. Geography, Univ. Liverpool, Res. Papers pp. 2-58.
Griffiths J.F. (ed.) (1972). Climates of Africa. World Survey of Climatology, vol. 10, Elsevier, Amsterdam,
604 pp.
Guillaumet J.L. (1967). Recherches sur la vegetation et la flore de la region du Bas-Cavally (Cote d'lvoire).
Memoires ORSTOM no. 20, ORSTOM, Paris, 247 pp.
Guillaumet J.L. & Adjanohoun E. (1971). La vegetation de la Cote d'lvoire. Dans: Le milieu naturel de la C6te
d'lvoire, Memoires ORSTOM no. 50, pp. 156-263.
Guillaumet J.L., Couturier G. et Dosso H. (1984). Recherche et amenagement en milieu forestier tropical
humide: le Projet Tai de C6te d'lvoire. Notes techniques du MAB no. 15, UNESCO, Paris.
Hall B.P. & Moreau R.E. (1970). An atlas of speciation in African passerine birds. Trustees of the British
Museum, London, 423 pp.
Hall J.B. & Swaine M.D. (1976). Classification and ecology of closed-canopy forest in Ghana. J. of Ecol. 64:
913-951.
Hall J.B. & Swaine M.D. (1981). Distribution and ecology of vascular plants in a tropical rain forest: forest
vegetation in Ghana. Geobotany 1, Dr W. Junk Publishers, The Hague, 383 pp.
Hamilton A.C. (1976). The significance of patterns of distribution shown by forest plants and animals in tropical
Africa for the reconstruction of upper Pleistocene palaeoenvironments: a review. In: van Zinderen Bakker
E.M. (ed.). Palaeoecology of Africa 9: 63-97.
Hamilton A.C. (1982). Environmental history of East Africa: a study of the quaternary. Academic Press,
London,311 pp.
Hamilton A.C. (1992). History of forests and climate. In: Sayer J.A., Harcourt C.S. & Collins N.M. (eds.).
The conservation atlas of tropical forests: Africa. MacMillan, Basingstoke, pp. 17-25.
Hamilton A.C. & Taylor D. (1991). History of climate and forests in tropical Africa during the last 8 million
years. Climatic Change 19: 65-78.
Hammermaster E.T. (1985). Forest resource mapping in Liberia. FAO Tropical Forest Project, Rome, 29 pp.
Hart T.B. (1990). Monospecific dominance in tropical rain forests. Tree 5(1): 6-11.
Hawthorne W. (1993). Fire damage and forest regeneration in Ghana. NRI-ODA, Chatham.
Hawthorne W. & Juam Musah A. (1993). Forest protection in Ghana. Forestry Dept., Kumasi, Ghana, 186 pp.
Hayward D.F. and Oguntoyinbo J.S. (1987). The climatology of West Africa. Hutchinson, London, 271 pp.
Hendrison J. (1990). Damage-controlled logging in managed tropical rain forest in Suriname. Doctoral thesis,
Wageningen Agricultural University, 204 pp.
Hill M.O. (1973). Reciprocal averaging: an eigenvector method of ordination. Journal of Ecology 61: 237-249.
Hill M.O. (1979a). Decorana. A Fortran programme for detrended correspondence analysis and reciprocal
averaging. Ecology and systematics, Cornell University, Ithaca, New York, 52 pp.
Hill M.O. (1979b). Twinspan. A Fortran programme for arranging multivariate data in an ordered two-way table
by classification of the individuals and the attributes. Ecology and systematics, Cornell University, Ithaca,
New York.
Hill M.O. & Gauch H.G. (1980). Detrended correspondence analysis: an improved ordination technique.
Vegetatio 42: 47-58.
Hoekman D.H. (1985). Radar backscattering of forest stands. Intern. Journal of Remote Sensing 6(2): 325-343.
Hommel P.W.F.M. (1990). A phyto-sociological study of a forest area in the humid tropics (Ujung Kulon, West
Java, Indonesia). Vegetatio 89: 39-54.
Hoppe-Dominik B. (1989). Habitatpraferenz und Nahrungsanpriiche des Waldbiiffels, Syncerus caffer nanus im
Regenwald der Elfenbeinkiiste. Ph.D. thesis, Univ. Braunschweig, 187 pp.
Hornby A.S., Cowie A.P. and Gimson A.C. (eds.) (1974). Oxford advanced learner's dictionary of current
English. Oxford University Press, Oxford, 1037 pp.
Hubbell S.P. and Foster R.B. (1983). Diversity of canopy trees in a neotropical forest and implications for the
conservation of tropical trees. In: Sutton et al. (eds.). Tropical rain forest: ecology and management. 41
Brit. Ecol. Soc. Spec. Publ. 2, Blackwell, Oxford, pp. 25-41.
138
References
Hubbell S.P. and Foster R.B. (1986). Commonness and rarity in a neotropical forest: implications for tropical
tree conservation. In: Soule M. (ed.). Conservation biology: science of scarcity and diversity. Sinauer Ass.,
Sunderland, Massachusetts, pp. 205-231.
Hutchinson J. & Dalziel J.M. (1954-72). Flora of West Tropical Africa. 3 volumes, (Second edition by Keay
R.W.J. & Hepper F.N.) Crown Agents, London.
Huttel Ch. (1977). Etude de quelques caracteristiques structurales de la vegetation du bassin versant de
l'Audrenisrou. Rapport ORSTOM, Adiopodoume, 33 pp.
IFAN (1968). International atlas of West Africa = Atlas international de l'ouest africain. Organisation of
African Unity, Scientific, Technical and Research Commission, Dakar, 44 plates + text.
IGN (1965). Cartes topographiques au 1:50.000 eme. Feuilles Tai 4c, Guiglo 2b et Tai 2d. Institut Geographique
National, Paris.
Jacobs M. (1988). The tropical rain forest, a first encounter. Springer, Heidelberg, 295 pp.
Jager J.C. & Looman C.W.N. (1987). Data collection. In: Jongman R.H.G., ter Braak C.J.F. and van Tongeren
O.F.R. (eds). Data analysis in community and landscape ecology, pp. 10-28. Pudoc, Wageningen.
James N.D.G. (1982). The forester's companion. Basil Blackwell, Oxford, 381 pp.
Jans L., Poorter L., van Rompaey R.S.A.R. and Bongers F. (1993). Gaps and forest zones in tropical moist
forest in Ivory Coast. Biotropica 25(2), accepted.
Jeambrun M. (1965). Rapport de fin de leve de la coupure Tai lb-2a-2c au 1/50.000. Rapport no 115, SociSte'
pour le Developpement Minier de la Cote d'lvoire, Abidjan, 53 pp.
Jeambrun M. (1966). Rapport de fin de leve des coupures Tai 2d-4b-4d et Guiglo 2b au 1/50.000. Rapport no.
153, Society pour le Developpement Minier de la Cote d'lvoire, Abidjan, 89 pp.
Johansson D. (1974). Ecology of vascular epiphytes in West African rain forest. Acta Phytogeographica Suecica
59, 136 pp.
Jonkers W.B.J. (1987). Vegetation structure, logging damage and silviculture in a tropical rain forest in
Suriname. PhD thesis, Agricultural University, Wageningen, 172 pp.
Koop H. (1989). Forest dynamics. SILVI-STAR: a comprehensive monitoring system. Springer-Verlag,
Heidelberg, 229 pp.
Koppen W. (1936). Das geographische System der Klimate. In: Koppen W. & Geiger R. (eds.). Handbuch der
Klimatologie. Band I, Teil C, 44 pp., Verlag von Gebriider Borntraeger, Berlin.
Koptur S. (1985). Alternative defenses against herbivores in Inga (Fabaceae: Mimosoideae) over an elevational
gradient. Ecology 66(5): 1639-1650.
Krajewski W.F. (1987). Cokriging radar-rainfall and rain gauge data. Journal of Geophysical Research 92:
9571-9580.
Krebs C.J. (1985). Ecology. The experimental analysis of distribution and abundance. Harper & Row, New
York, 678 pp.
Krige D.G. (1951). A statistical approach to some basic mine evalution problems on the Witwatersrand. Journal
of the Chemical, Metallurgical and Mining Society of South Africa 52: 119-138.
Lawson G.W., Hall J.B. and Armstrong-Mensah K.O. (1970). A catena in tropical moist semi-deciduous forest
near Kade, Ghana. Journal of Ecology 58: 371-398.
Leersnijder R.P & Boeijink D.E. (1990). Modal transect construction for silvicultural design. In: Oldeman
R.A.A., Schmidt P. & Arnolds E.J.M. (eds.). Forest components. Wageningen Agric. Univ. Papers 90-6,
pp. 13-26.
Leneuf N. (1959). L'alteration des granites calco-alcalines et des grano-diorites en Cote d'lvoire forestiere et les
sols qui en sont derives. Memoires ORSTOM, 191 pp.
Lescure J.P. & Boulet R. (1985). Relationships between soil and vegetation in a tropical rain forest in French
Guiana. Biotropica 17(2): 155-164.
Letalenet J. (1965a). Rapport de fin de leve de la coupure Guiglo 2a au 1/50.000. Rapport no. 101,Societe pour
le Developpement Minier de la Cote d'lvoire, Abidjan, 40 pp.
Letalenet J. (1965b). Rapport de fin de leve des coupures Tai 2b- Soubre la Tabou 4d,4b- Sassandra 3c, 3a.
Rapport no 149, Societe pour le Developpement Minier de la Cote d'lvoire, Abidjan, 99 pp.
Letouzey R. (1968). Etude phytogeographique du Cameroun. Lechevalier, Paris.
Lhomme J.P. (1981). L'evolution de la pluviosite annuelle en Cote d'lvoire au cours des soixante dernieres
annees. La Meteorologie Vie Serie 25: 135-140.
Lieberman M., D. Lieberman, G.S. Hartshorn & R. Peralta (1985). Small scale altitudinal variation in lowland
wet tropical forest vegetation. Journal of Ecology 73: 505-516.
Lincoln R.J., Boxshall G.A. & Clark P.F. (1982). A dictionary of ecology, evolution and systematics.
Cambridge University Press, Cambridge, 298 pp.
139
Lindeman J.C. and Moolenaar S.P. (1959). Preliminary survey of the vegetation types of Northern Suriname.
The vegetation of Suriname, Vol. I part 2, Van Eedenfonds, Amsterdam.
Loetsch F., Zohrer F. and Haller K.E. (1973). Forest inventory. Volume 2, B.L.V., Miinchen, 417 pp.
Longman K.A. & Jenik J. (1987). Tropical forest and its environment. Longman Sc. and Techn., Essex.
Maitre H.F. (1991). Silvicultural interventions and their effects on forest dynamics and production in some rain
forests of in Cote d'lvoire. In: Gomez-Pompa A., Whitmore T.C. and Hadley M. (eds.). Rain forest
regeneration and management. Man and the Biosphere Series 6, pp. 383-392, Unesco, Paris.
Maley J. (1987). Fragmentation de la foret dense humide africaine et extension des biotopes montagnards au
quaternaire recent: nouvelles donnees polliniques et chronologiques. Implications paleoclimatiques et
biogeographiques. Palaeoecology of Africa 18: 307-334, Balkema, Rotterdam.
Maley J. (1991). The African rain forest vegetation and palaeoenvironments during late Quaternary. Climatic
Change 19: 79-98.
Mandelbrot B.B. (1983). The fractal geometry of nature. Freeman, New York.
Mangenot G. (1955). Etude sur les forets des plaines et plateaux de la Cote d'lvoire. Etudes Eburneennes IV,
pp. 5-61, Inst. Fr. d'Afr. Noire, Centre de Cote d'lvoire.
Marchesi P., Marchesi N. et Boesch Ch. (1990). Estimation des surfaces forestieres de Cote d'lvoire d'apres des
images satellites. Unpublished manuscript, 9 pp.
Martin C. (1989). Die Regenwalder Westafrikas: Okologie-Bedrohung-Schutz. Birkhauser Verlag, Basel,
235 pp.
Martin L. (1972). Variations du niveau de la mer et du climat en Cote d'lvoire depuis 25 000 ans. Cah.
ORSTOM, ser. Geologie 4 (2): 93-103.
Mayer H. (1980). Waldbau auf soziologisch-okologischer Grundlage. Gustav Fischer Verlag, Stuttgart, 483 pp.
Mayers J. (1992). Liberia. In: Sayer J.A., Harcourt C.S. & Collins N.M. (eds.). The conservation atlas of
tropical forests: Africa. MacMillan, Basingstoke, pp. 214-220.
Mayr E. & O'Hara R.J. (1986). The biogeographic evidence supporting the Pleistocene forest refuge hypothesis.
Evolution 40: 55-67.
Meijers G.J. and Saye J. (1983). Rainfall data book of Liberia 1981. Liberian Hydrological Service, Monrovia,
37Pp.
Mengin-Lecreulx P. (1990). Simulation de la croissance d'un peuplement de foret dense. Le cas de la for&t de
Yapo. SODEFOR-CTFT, Abidjan, Nogent-sur-Marne.
Ministere des Eaux et Forets (1988). Plan directeur forestier 1988-2015. Min. des Eaux et Forets, Abidjan.
Monteny B.A. & Casenave A. (1989). The forest contribution to the hydrological budget in Tropical West
Africa. Annales Geophysicae 7(4): 427-436.
MPEA (1983). Republic of Liberia planning and development atlas. Ministry of Planning and Economic Affairs,
Monrovia, Liberia, 67 pp.
Munsell (1975). Soil color charts. Kollmorgen corporation, Baltimore, Maryland, USA.
Nakashizuka T., Zulkifli Yusop & Abdul Rahim Nik (1992). Altitudinal zonation of forest communities in
Selangor, Peninsular Malaysia. Journal of Tropical Forest Science 4(3): 233-244.
Newbery D.M., Alexander I.J., Thomas D.W. & Gartlan J.S. (1988). Ectomycorrhizal rain-forest legumes and
soil phosphorus in Korup Forest Reserve, Cameroon. New Phytol. 109: 433-450.
Noirfalise A. (1984). ForSt et stations forestieres en Belgique. Les Presses Agronomiques de Gembloux, 234 pp.
Nooren C.A.M. (1992). Detailed soil survey of a watershed & Study on the role of earthworms in the formation
of sandy surface soils in Tai National Park (Cote d'lvoire). Practical period & MSc. thesis, Department of
Soil Science and Geology, Agricultural University, Wageningen, 68 pp.
Nye P.H. (1954). Some soil-forming processes in the humid tropics. I. A field study of a catena in the West
African forest. Journal of Soil Science 5: 7-21.
Ohsawa M., Nainggolan P.H.J., Tanaka N. & Anwar C. (1985). Altitudinal zonation of forest vegetation on
Mount Kerinci, Sumatra: with comparisons to zonation in the temperate region of east Asia. J. Trop.
Ecology 1: 193-216.
Oldeman R.A.A. (1964). Revision of Didelotia Baill. (Caesalpiniaceae). Primitiaea Africanae IV, Blumea 12:
209-239.
Oldeman R.A.A. (1974). L'architecture de la foret guyanaise. Memoires ORSTOM no. 73, ORSTOM, Paris,
204 pp.
Oldeman R.A.A. (1983). Dood hout in tropische regenbossen. Nederlands Bosbouw Tijdschrift 55 (1983):
112-118.
Oldeman R.A.A. (1990a). Forest ecosystems and their components: an introduction. In: Oldeman R.A.A.,
Schmidt P. & Arnolds E.J.M. (eds). Forest components. Wageningen Agricultural University Papers 90-6,
pp. 3-12.
140
References
Oldeman R.A.A. (1990b). Forests, elements of silvology. Springer Verlag, Heidelberg, 624 pp.
Oldeman R.A.A. (1991). The paradox of forest management. Proceedings Xth World Forestry Congress, 1991,
Paris, 4: 153-182.
Oldeman R.A.A. (1992). Forest resource utilization. In: Hummel J.A. & Parren M.P.E. (eds.) Forests, a
growing concern. Proceedings XlXth International Forestry Students Symposium, September 1991,
Wageningen. IUCN, Gland, pp. 27-32.
ORSTOM-CIEH (1973). Precipitationsjournalieres de l'origine des stations a 1965. Comite Interafricain
d'Etudes Hydrauliques CIEH, ORSTOM, Paris.
Papon A. (1973). Geologie et mineralisations du sud-ouest de la Cote d'lvoire. Synthese des travaux de
l'operation Sacsa 1962-1968. SODEMI, Abidjan.
Parren M.P.E. (1991). Silviculture with natural regeneration: a comparison between Ghana, Cote d'lvoire and
Liberia. MSc thesis AV. 90/50, Department of Forestry, Agricultural University, Wageningen, 82 pp.
Parren M.P.E. and de Graaf N.R. (1993 in prep.). Forestry in West Africa : the quest for natural forest
management: lessons to be learned and examples to be followed. Tropenbos series, The Tropenbos
Foundation, Wageningen, 202 pp.
Perraud A. (1971). Les sols. Dans: Avenard et al. Le milieu naturel de la Cote d'lvoire, Memoire ORSTOM
50, pp. 265-391.
Poker J. (1989). Struktur und Wachstum in selektiv genutzten Bestanden im Grebo National Forest von Liberia.
AbschluBbericht zum GTZ-Projekt Nr. 87.2050.0-01.100, Institut fur Weltforstwirtschaft und Okologie,
Hamburg, 274 pp.
Poker J. (1992). Struktur und Dynamik des Bestandesmosaiks tropischer Regenwalder - Entwicklung eines
Modellansatzes zur Simulation nariirlicher Mischbestande. Dissertation, Universitat Hamburg, 221 pp.
Poorter L., Jans L., van Rompaey R.S.A.R. and Bongers F. (1993 in prep.). Spatial distribution of gaps along a
catena gradient in Tai, Ivory Coast. Journal of Tropical Ecology, submitted.
Rademacher F.E.P. (1992). A detailed soil survey in the northern part of the Tai National Park, southwest Cote
d'lvoire. MSc. thesis, Department of Soil Science and Geology, Agricultural University, Wageningen, 58
pp.
Ramensky L.G. (1910). On the comparative method for ecological study of plant communities (in Russian).
Dnevnik S'ezda russk. Estestvoisp Vrach. 2(9): 389-390.
Ramensky L.G. (1930). Zur Methodik des vergleichenden Bearbeitung und Ordnung von Pflanzenlisten und
andere Objekten, die durch mehrere, verschiedartigen wirkende Faktoren bestimmt werden. Beitr. Biol. Pfl.
18: 269-304.
Richard J.F. (1989). Le paysage, un nouveau langage pour l'etude des milieux tropicaux. Initiations - Documentations techniques no 72, ORSTOM, Paris, 210 pp.
Rollet B. (1974). L'architecrure des forets denses sempervirentes de plaine. CTFT, Nogent sur Marne.
Rougerie G. (1960). Le faconnement actuel des modeles en Cote d'lvoire forestiere. Memoires de PInstitut
Francais d'Afrique Noire (TFAN) 58, Dakar, 493 pp.
Russell-Smith J. (1991). Classification, species richness, and environmental relations of monsoon rain forest in
northern Australia. Journal of Vegetation Science 2: 259-278.
Sachtler M. (1968). General report on National Forest Inventory in Liberia. Technical report no. 1, German
Forestry Mission to Liberia, Monrovia, 149 pp.
Sachtler M. and Hamer K. (1967). Inventory of Krahn-Bassa and Sapo National Forest. Technical report no. 7,
German Forestry Mission to Liberia, Monrovia, 92 pp.
Sayer J.A., Harcourt C.S. & Collins N.M. (1992). The conservation atlas of tropical forests: Africa.
MacMillan, Basingstoke, 288 pp.
Schmidt E. (1990). Inventarisation of the bufferzone of the Park Tai by remote sensing. MSc thesis, Depart, of
Silviculture and Forest Ecology, Wageningen Agr. Univ.
Schnell R. (1950). La foret dense. Introduction a l'etude botanique de la region forestiere d'Afrique occidentale.
Manuels Ouest-africains 1, P. Lechevalier, Paris, 330 pp.
Schnell R. (1952). Vegetation et flore de la region montagneuse du Nimba (A.O.F.). Mem. IFAN 22, 604 pp.
Schulz J.P. (1960). Ecological studies on rain forest in Northern Suriname. The vegetation of Suriname Vol. II,
Van Eedenfonds, Amsterdam, 267 pp.
Schumann W. (1979). Elsevier gids voor stenen & mineralen. Elsevier, Amsterdam, 232 pp.
Snoeck J. (1975). Variations de la pluviosite en zone forestiere ivoirienne. Cafe-Cacao-The 19(3): 165-176.
Sobolev L.N. & Utekhin V.D. (1973). Russian (Ramensky) approaches to community systematization. In:
Whittaker R.H. (ed.). Ordination and classification of communities. Handbook of vegetation science 5:
77-103. Junk Publishers, The Hague.
SODEFOR (1976a). Inventaire forestier national. Resultats de la region centre-sud. SODEFOR, Abidjan.
141
SODEFOR (1976b). Inventaire forestier national. Resultats de la region nord-ouest. SODEFOR, Abidjan.
Sokal R.R. & Rohlf F.J. (1969). Biometry. The principles and practice of statistics in biological research.
Freeman, San Francisco, 776 pp.
Sosef M.S.M. (1993 in press). Glacial rain forest refuges in relation to speciation in Begonia sect. Loasibegonia
and sect. Scutobegonia. Proceedings of the 13th AETFAT congress, Zomba, Malawi 1991.
Spichiger R. & Lasailly V. (1981). Recherche sur le contact foret-savanne en Cote-d'Ivoire: note sur 1'evolution
de la vegetation dans la region de Beoumi (Cote d'lvoire centrale). Candollea 36: 145-153.
Stein A. and Corsten L.C.A. (1991). Universal kriging and cokriging as a regression procedure. Biometrics 47:
575-587.
Staffers, A. (1989). Groeimetingen in het Pare National de Tai. Stageverslag, Larenstein Internat. Hogeschool,
Velp.
Swaine M.D. & Hall J.B (1976). An application of ordination to the identification of forest types. Vegetatio 32:
83-86.
Swaine M.D. & Hall J.B. (1988). The mosaic theory of forest regeneration and the determination of forest
composition in Ghana. Journal of Tropical Ecology 4(3): 253-269.
Swaine M.D., Hall J.B. & Alexander I.J. (1987). Tree population dynamics at Kade, Ghana (1968-1982).
Journal of Tropical Ecology 3(4): 331-345.
Swaine M.D., Hall J.B & Lock J.M. (1976). The forest-savanna boundary in west-central Ghana. Ghana J. Sci.
16(1): 35-52.
Swaine M.D. & Whitmore T.C. (1988). On the definition of ecological species groups in tropical rain forests.
Vegetatio 75:81-86.
Tagini B. (1972). Notice explicative a la carte geologique de Cote d'lvoire a 1/2.000.000. Rapport SODEMI
No. 279, Abidjan, 19 pp.
Taylor C.J. (1952). The vegetation zones of the Gold Coast. Bull. Gold Coast For. Dep. 4: 1-12.
Taylor C.J. (1960). Synecology and silviculture in Ghana. Nelson, Edinburgh, 418 pp.
ter Braak C.J.F. (1986). Canonical correspondence analysis: a new eigenvector technique for multivariate direct
gradient analysis. Ecology 67(5): 1167-1179.
ter Braak C.J.F. (1987a). Ordination. In: Jongman R.H.G., ter Braak C.J.F. and van Tongeren O.F.R. (eds).
Data analysis in community and landscape ecology, pp. 91-173. Pudoc, Wageningen.
ter Braak C.J.F. (1987b). The analysis of vegetation-environmental relationships by canonical correspondence
analysis. Vegetatio 69(1/3): 69-77.
ter Braak C.J.F. & Looman C.W.N. (1987). Regression. In: Jongman R.H.G., ter Braak C.J.F. and van
Tongeren O.F.R. (eds). Data analysis in community and landscape ecology, pp. 29-77. Pudoc, Wageningen.
ter Braak C.J.F. & Prentice I.C. (1988). A theory of gradient analysis. Advances in Ecological Research 18:
271-317.
Touber L., Smaling E.M.A., Andriesse W. & Hakkeling R.T.A. (1989). Inventory and evaluation of tropical
forest land. Guidelines for a common methodology. Tropenbos Technical Series 4, Ede, 170 pp.
Tysdal R.G. (1977). Geologic map of the Juazohn quadrangle, Liberia. United States Geological Survey Misc.
Inv. Ser. Map I-779-D, Reston, USA.
van der Werf S. (1991). Bosgemeenschappen. Pudoc, Wageningen, 375 pp.
van Donselaar J. (1965). An ecological and phytogeographic study of northern Surinam savannas. North-Holland
Publ. Co., Amsterdam,
van Herwaarden G.J. (1991a). Compound report on three soil surveys in the Tai forest (Cote d'lvoire).
UNESCO/ Dept. of Soil science and geology, Agricultural University, Wageningen, 59 pp.
van Herwaarden G.J. (1991b). Some physical soil properties on a catena in the Tai region (south-west Cote
d'lvoire). Unesco, Dept. of Soil Science & Geology, Agricultural University, Wageningen, 58 pp.
van Kekem A.J. (1984). Legende pour la carte des sols du sud-ouest de la Cote d'lvoire. Application de la
methode utilisee dans le programme pedologique du M.A.B. Projet Tai, UNESCO.
Van Miegroet M. (1976). Van bomen en bossen. Story-Scientia, Gent, 2 vol., 1166 pp.
van Tongeren O.F.R. (1987). Cluster analysis. In: Jongman R.H.G., ter Braak C.J.F. and van Tongeren O.F.R.
(eds). Data analysis in community and landscape ecology, pp. 174-212. Pudoc, Wageningen.
Vanclay J.K. (1989). A growth model for North Queensland rainforests. Forest Ecology and Management 27:
245-271.
Vivien J. et Faure J.J. (1985). Arbres des forets denses d'Afrique centrale. Min. de Cooperation, Paris, 565 pp.
Vooren A.P. (1979). La voute forestiere et sa regeneration. Analyse structurelle et numerique d'une
toposequence en foret de Tai, Cote d'lvoire. MSc thesis, Agricultural University, Wageningen, 89 pp.
Vooren A.P. (1985). Patterns in tree and branch-fall in a West African rain forest. Report D85-05, Dept. of
Silviculture, Agricultural University, Wageningen, 33 pp.
142
References
Vooren A.P. (1986). Nature and origin of tree and branch fall in the Tai Forest (Ivory Coast). Neth. J. Agricult.
Sci. 34: 112-115.
Vooren A.P. (1987). Development versus conservation: avoiding a conflict in the Tai region (Ivory Coast). In:
Beusekom, C.F. van; Goor, C.P. van; Schmidt, P. (eds.) Wise utilization of tropical rain forest lands,
Tropenbos Scientific Series 1: 130-137, Tropenbos/ MAB Unesco, Ede.
Vooren A.P. (1992a). Appropriate buffer zone management strategies for Tai' National Park. In: Puig H. et
Maitre H.F. (eds.). Actes de l'Atelier sur l'amenagement et la conservation de l'ecosysteme forestier
tropical humide, mars 1990, Cayenne, pp. 26-39.
Vooren A.P. (1992b). Cote d'lvoire. In: Sayer J.A., Harcourt C.S. & Collins N.M. (eds.). The conservation
atlas of tropical forests, Africa. MacMillan, Basingstoke, pp. 133-142.
Vooren A.P. (1992c). Harvest criteria for tropical forest trees. In: Cleaver K. et al. (eds.). Conservation of West
and Central African rainforests. World Bank environment paper 1: 134-140, The World Bank, Washington,
D.C.
Voorhoeve A.G. (1964). Some notes on the tropical rainforest of the Yoma-Gola National Forest near Bomi
Hills, Liberia. Commonwealth Forestry Revue 43(1): 17-24.
Voorhoeve A.G. (1965). Liberian high forest trees. A systematic botanical study of the 75 most important or
frequent high forest trees, with reference to numerous related species. PhD thesis, Agricultural University,
Wageningen, 416 pp.
Walter H. (1979). Vegetation of the earth and ecological systems of the geo-biosphere. Springer Verlag,
Heidelberg.
Walter H. & Box E. (1976). Global classification of natural terrestrial ecosystems. Vegetatio 32: 75-81.
Walter J.-M.N. (1974). Arbres et forets alluviales du Rhin. Bull. Soc. Hist. Nat. Colmar 55: 37-88.
WAU (1991). Annual report 1990. Centre Neerlandais d'Adiopodoume, C6te d'lvoire & Research programme:
"Analysis and design of land-use systems in the Tai region". Wageningen Agricultural University,
Wageningen, 28 pp.
Webster's (1976). Webster's seventh new Collegiate dictionary. Merriam Co., Springfield, Mass., 1224 pp.
White F. (1983). The vegetation of Africa, a descriptive memoir to accompany the UNESCO/ AETFAT/ UNSO
vegetation map of Africa. Natural resources research XX, UNESCO, Paris, 384 pp.
Whitmore T.C. (1984). Tropical rain forests of the Far East. Oxford Science Publ., Oxford, 352 pp.
Whitmore T.C. (1990). An introduction to tropical rain forests. Clarendon Press, Oxford, 226 pp.
Whitmore T.C. and Silva J.N.M. (1990). Brazil rain forest timbers are mostly very dense. Commonwealth
Forestry Review 69(1): 87-90.
Whittaker R.H. (1956). Vegetation of the Smoky Mountains. Ecological Monographs 26: 1-80.
Whittaker R.H. (1967). Gradient analysis of vegetation. Biological Reviews 42: 207-264.
Williams W.T., Lance G.N., Webb L.J., Tracey J.G. & Connell J.H. (1969). Studies in the numerical analysis
of complex rain forest communities IV. A method for the evaluation of small-scale forest pattern. J.Ecol.
57: 635-654.
W611H.J. (1981). Silvicultural evaluation. Diagnostic sampling 1978-1980. Internal report, German Forestry
Mission to Liberia, Monrovia, 65 pp.
WRR (1992). Grond voor keuzen. WRR rapport 42, WRR, Den Haag, 149 pp.
Zonneveld I.S. (1988). Establishing a floristic classification. In: Kiichler A.W. & Zonneveld I.S. (eds).
Vegetation mapping, pp. 81-88. Handbook of Vegetation Science 10, Kluwer, Dordrecht.
View publication stats