I

 

.. . ‘ 4.7} :‘ ,1" 2"? ', , A ' 3" ‘ “ A ' m5? W‘-
V . . ' " ’ V ‘3 3 ’9’

. - ~ "-1439; - . J , ~ ‘ y" 340.»

4 ,4 .~ ~ . .J. J, J. 4 , r » ~ . 44:4» ‘4 w» '2'

4 . ~ ' .' . ,' 53:11! 1. ’ ’ ‘ A 95’“ wfi”§4‘;’v
. _ r . .r z , w :__ 'J‘Tz'f‘llf - , ~n‘{~’)(,’fp.{)( ._,
“a ' Wave-m 9' if, wuwx
V r
‘4‘3‘13 ~ M”?>’ ":1" ‘9"

1‘, K3

   

 

mu» "1» d ’1

 

 

(94715:. {33%. ,-.'

 

, _ . . 4. . 4"” ’. ‘- "r » ' H ,f,“,,‘> ‘

. . , ‘ - i J . ”a ..

. . . , . J 4 ~ - ~ ~ , . , , - a 44.. ~.-,:;}.“‘J.
, . , , , ._ , , , 4 I - ' \f i " -. »..'
, ‘ I ‘ . ,1 A, ‘ , . . r-y’ v )4- $5,221. 9‘)

 

H:
v ."

4
l

) 'I

. u ,
«:1
u/

.1“-

4
4.

I. n
"’1’
«A,

If,»
5" I
1

{3'5
an)”:
.5». a

}

v‘v

9,» .
v;
A

71‘ C-‘Ai‘k
1 Air '

 

I , ,
'51} If};
‘ (£5 3'

f » (n. ‘
”firefly:
'r

‘5
:lsi
T"~ 4 ..
«tw-
59.,
:s‘x‘

ms

4"

”x
2*
, ~
Cs
‘4‘»
‘8
s

j; ;
fa.

4,44» 4-:
MM! [7,45

”'I'nu‘m

.
s:

4' .
. “ 95"”:
. Iii-v”

 

"a; ~ 'V'l’_y'1(":' I'.’ 25'"?
‘ I" ilk-1’s; 3,4}; ' '
' £393: rr

r r I
I‘r
”r:

-" 9.9%,»,

) (“11.39; fill-y, .1151]: , . > . If! , ' r v , - , ‘.

:44 Arrival/4"»? ””4" .4 h,- I; J. » - .4 40’4” t ('4’ ' ‘ ’ 3-},

4 ,1 ’1') (tron 'r'f ‘, .',I‘, ”’33; v»- war/3‘94;

4,1,”, , g * ,.r2gear..r;a"

(”17442543
.rt’fi'33»”3x'. .

,4 I g “m , - , I ’ {‘4’ 1”???" '4’ #5:; 7,1,4, m ,4

J ' / I, .. .. a" ,_.. - w - ,. x. o“ In!” '
’9’ Jrr *1/14 , J. x" ,f .qu "‘6 .v' I wring» 7",?!» q’fiafé ' ,7”; "¥#’7‘J’.< 353:545'

.m,,,,H’rx,MJ 4:51;» 9"???” yé/ [£972 “€503 - If - {1' n; n": ' 4,. “W" 4’ , ', .355 . ..-',5(’J;',r.'"~.

111"" ['C/Jin, 5’9 " ”7'9"" "'5. 1:45;! 14%», 9,3,, 13/;- {fi ’1" L432"! ' :‘Eé‘éfiifi-Zfi

37 1"}; “"N‘u'xn." q """' A {/9,;."J"rj.,m. y‘m'fi'fi'w " ' .' '

 

J
' "nil-1'" 12"”! 4 ’34., 'r' ‘ ’ r u; m"; ‘ 9 J ‘ ’ ' . Mi -"
M“: ,, .Ia’, r. 'S ".' ‘ .4 ; . ‘» IJ;;:I')£71" Al?” I r, ‘ ‘ , :1er arr”)..-
3",,1‘? 4,}, rind; , mm! .. 9/": 4:, H [gym I "W "’3: H,’ . r“. . fra‘mafid

my u‘v'l

   

I117

 

 

.9"

 

.
~ 4 mm r '
.4’ ‘f’jé .4

{3’9 ACE/KI;
:9'9» ,
7 €

pm:

   

up” , ' 4 '- 4
l r' 92:69:; i: :‘I I ,; n
, (Ir. .. 5,1» 4: - , .,, $1,11-

71'". {13
J \ ’ I
. ‘ V~ fl m“: » J is???“
f" 1, half} " 4w; w'r 'Lu . #4 ‘ ' w ' » “
3:: r'f'y' fifflg‘ ’ ,Q'I'afli u _ t;- ’2’”: 6'93}??? “SJ-"Is'fAAI-"Y' fp’é’} i531 - 1 ’r . ”'41 an“: 59"".sz
[I]. a, . I: 14'1“" .,, / “A149 gun‘s, 10:3" 3%; n» a 7
, i n .
I

 

A Arr r J . .
4J4» J: »»'» - 4» 44'»
$1". I‘

   

 

x ‘5’??? ‘94”
‘-’-',’.-"d . >’.» r .1’" ,J’i.‘?'
4 ' l! "20 ”A r'

J’

u" ‘4" '
'trfiz’,“4,.,,,;: .
4% £43,141”.
'1‘.¢J" ‘9'”
4,!» 5J9?

#1:? #45,:-

 

 

.‘ .n m hi”

fi' m}, " {3'}!-

 

‘3',"

 

e; u‘fihéifl

 

LIBRARY
LMichigan State
University;

 

 

 

This is to certify that the

dissertation entitled

THE INFLUENCE OF MECHANIZED SELECTIVE LOGGING, FELLING
INTENSITY AND GAP-SIZE ON THE REGENERATION OF A
TROPICAL MOIST FOREST IN THE KIBALE FOREST
RESERVE, UGANDA

presented by

John Massan Kasenene

has been accepted towards fulfillment
of the requirements for

 

Ph.D. degreein Botany & Plant Pathology

Major professor \(

.MSU is an Affirmative Action/Equal Opportunity Institution 0—12771

 

Date December 21, 1987

 

MSU

LlBRARlES

 

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if_book is
returned after the date
stamped below.

 

 

 

i in: £91753?

 

 

 

 

THE INFLUENCE OF MECHANIZED SELECTIVE LOGGING, FELLING
INTENSITY AND GAP-SIZE ON THE REGENERATION OF A
TROPICAL MOIST FOREST IN THE KIBALE FOREST
RESERVE, UGANDA

BY

John Massan Kasenene

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Botany and Plant Pathology
1987

.507. 37.5/

ABSTRACT

THE INFLUENCE OF MECHANIZED SELECTIVE LOGGING, FELLING
INTENSITY AND GAP-SIZE ON THE REGENERATION OF A
TROPICAL MOIST FOREST IN THE KIBALE FOREST
RESERVE, UGANDA

BY

John Massan Kasenene

This study was conducted in the Kibale Forest Reserve
in Western Uganda. The reserve is Equatorial in position
(0° 13' to 0° 41'N and 30° 19' to 30° 32'E) with a medium
altitude tropical moist forest. The effects of a mechanized
polycyclic felling system on the forest was characterized
through study of species composition, structure and the
regeneration of tree species with the uncut mature forest as
the control. Four contiguous forest tracts, including the
control, the lightly cut, the heavily cut, and heavily cut
and arboricide-treated forest, were surveyed.

The results clearly indicate that mechanized selective
timber harvesting in species-rich tropical moist forest was
hard to control and incompatible with minimizing damage and
disturbances or creation of forest gaps characteristic of
natural forest disturbances. The mean gap size for lightly
cut forest was 467m2 (range 75 to 1800m2) whereas natural

tree fall gaps in control forest averaged 256m2 (range 100

ii

to 663m2). The mean gap size in heavily cut forest was very
large, 1307m2 (range 73 to 7100m2) for heavily cut, and
938m2 (range 527 to 3313m2) for heavily cut and treated
forest.

Mechanized selective logging had detrimental effects on
the forests of the reserve which led to drastic decreases in
sapling, pole, and tree species richness, distribution,
density, diversity, and basal areas for all tree species,
including desirable timber or primary forest species. There
was also increased tree mortality through windthrows in
heavily cut areas where annual rates of 1.7, 1.3, and 3.3
trees/ha for uncut, lightly cut and heavily cut forests,
respectively, were observed.

Successful tree seedling, sapling and pole growth was
observed at felling intensities below 14m3/ha and gap sizes
below 650m2, whereas regeneration where 17 to 21m3/ha of
commercial timber had been mechanically removed, and where
gaps above 650m2 were created, was impoverished. Based on
the sapling and pole species richness of unlogged mature
forest (control), the forest regeneration 20 years post-
felling was less by 28.1% in lightly cut, less by 48.9% in
heavily cut and less by 70.8% in heavily cut and treated
forest areas. This reflects the effects of the logging
system, logging intensity and disturbances in the forests 20
years ago. The chemical treatments contributed to further
lowering of the diversity of tree regeneration in heavily

cut and treated areas.

iii

It appears, therefore, that the heavily selectively cut
tracts of the reserve might have to recover for longer
periods than the usual 70 years of harvest cycle for
anything resembling an exploitable forest to reappear.
Timber extraction through controlled systems that could
remove between 2 and 4 canopy trees/ha would cause
relatively minor damage to the forest. Pitsawing and forms
of polycyclic systems that are more labor intensive than
machine based could, if properly controlled, create gaps and
disturbances similar to natural tree falls and be very
compatible with long term sustained yield timber

exploitation of tropical moist forest reserves.

iv

This dissertation is dedicated to
the memory of my brother ISRAEL.

ACKNOWLEDGEMENTS

The author's work alone is by no means the only factor
responsible for the accomplishment of a doctoral
dissertation. This therefore gives me opportunity to
acknowledge the many individuals, organisations and
societies who provided support and assistance during the
course of my study.

I thank the office of the President, The National
Research Council, Uganda and the Uganda Forest Department
for giving me permission to conduct research in Uganda and
the Kibale Forest Reserve in particular.

Special thanks go to Dr. Peter G. Murphy, my major
professor and supervisor whose confidence in me, close
guidance and advice and constructive criticisms throughout
my doctoral course of study were essential for its
realisation. Many thanks to the members of my academic
committee for their contributions to the research proposal,
confidence and freedom accorded to me in the execution of
field research and completion of the study. This
dissertation also benefited greatly from the editorial
comments and constructive criticisms by Dr. Peter G. Murphy

and my Guidance Committee members including Dr. John Beaman,

vi

Dr. George Petrides and Dr. Donald Dickmann. To all of them
I am most grateful.

Dr. Thomas Struhsaker provided input and guidance into
the construction of the proposal, shared and contributed
substantially towards the execution of most phases of field
research. His efforts, encouraging support and advice
throughout the field work will always be remembered and
sincerely appreciated. To all colleagues and my field
assistants at the Kibale Forest Project and the staff of
Makerere University Herbarium, especially Mr. Katende, the
Curator, who generously offered their time, services,
efforts and advice to me during data collection and plant
identification, I wish to register my sentiments of thanks
and appreciation.

I am highly indebted to the New York Zoological Society
(NYZS), the African Wildlife Foundation (AWF), the Kibale
Forest Project (KFP) and the National Science Foundation
(NSF), Division of International Programs (Grant Number INT-
8411306) who have supported me financially and materially
during course work study at Michigan State University, field
research in Uganda and while completing the doctoral work at

Michigan State University.

Special recognition is also due to three very special

people, who for several years of my education have made

numerous allowances for me and provided ongoing support and
encouragement. Thanks to my beloved parents, John and

Evanice Massa and my dear wife Lydiah Beth.

vii

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES.
NOMENCLATURE OF PLANTS CITED IN THE TEXT

GENERAL INTRODUCTION
Historical Background to the Study. . .
Location of the Study Site and Climate.
Geology and Soils . . . . . . .
Vegetation and Animals. .
The Study Context: Objectives and Hypotheses.

LITERATURE CITED

PART I: THE COMPOSITION, STRUCTURE AND REGENERATION
OF SELECTIVELY CUT AND UNCUT FOREST TRACTS
OF THE RESERVE

INTRODUCTION.
Tropical Forest Disturbances and the Need
for Research
General Management History of the Study
Compartments and Plots . .
The Study Plots

METHODS .
Level of Forest Exploitation
Forest Vegetation Composition and
Structure. .

RESULTS
Level of Forest Exploitation
Forest Vegetation Composition and
Structure.
Tree Species Diversity.
Tree Density and Basal Area

Importance Values of Trees.
Other Measures of Forest Structure

and Composition
Intraplot Vegetation Parameter Comparisons

viii

Page
xi
xvi

xxi

HH
«#ONIO'IHH

20

20

22
27

3O
30

32

35
35

36
36
47
53

56
58

PART

PART

Regeneration in Mature and Secondary
Forest Communities
Tree Size Class Distribution. .
Tree Seedling and Sapling Dynamics.

DISCUSSION.
SUMMARY AND CONCLUSIONS
LITERATURE CITED.

APPENDIX TABLES

II: MODE OF TREE DEATH AND GAP CREATION IN
SELECTIVELY CUT AND UNCUT FOREST TRACTS
OF THE RESERVE

INTRODUCTION.
METHODS

RESULTS .
Comparison of Treefalls in Cut and Uncut
Mature Forests .
Intracompartment Comparisons.
Intercompartment Comparisons.
Diameter Class Distribution of Tree
Falls

DISCUSSION.
Comparison of Treefalls in Out and Uncut
Mature Forests

SUMMARY AND CONCLUSIONS
LITERATURE CITED.

APPENDIX TABLES

III: POST LOGGING PLANT DYNAMICS IN FOREST GAPS

INTRODUCTION
Forest Gaps and Forest Regeneration.

METHODS .

Gap Tree Sapling and Pole Enumerations
Gap Ground Vegetation Cover (gvc).
Herbivory and Seed Predation in Forest

Gaps

Sources of Plant Recruitment into Forest
Gaps .

Mode of Tree Sapling and Pole Regeneration
in Gaps. .

ix

Page
67
67
7O
76
88
92

99

101
105
112
112
112
116
125
132
132
147

152

156

160
165
169
170
171
172

174

Page

RESULTS . . . . . . . . 175
Natural and Artificial Forest Gaps . . . . 175
Forest Gap Sizes and Sapling and Pole

Regeneration . . . . 177
Herbivory and Seed Predation in Forest

Gaps . . . . 190
Sources of Plant Recruitment into Forest

Gaps . . . . 194
Gap Soil Seed. Banks and. Incoming Seed

Dispersal. . . 203
Mode of Tree Sapling and Pole Regeneration

in Gaps. . . . . . . . . . . . . . . . . 206

DISCUSSION. . . . . . . . 213
Natural and Artificial Forest Gaps . . . . 213
Sapling and Pole Regenaration Relative to

Forest Gap Sizes . . . . . . . . . . . . 216
The Ideal Gap Size . . . ‘ . . 223
Herbivory and Seed Predation in Forest

Gaps . . . 225
Sources of Plant Recruitment into. Forest

Gaps . . . . . . . . . . . . . . . . . . 230

SUMMARY . . . . . . . . . . . . . . . . . . . . 238

CONCLUSION. . . . . . . . . . . . . . . . . . . 243

LITERATURE CITED. . . . . . . . . . . . . . . . 247

APPENDIX TABLES . . . . . . . . . . . . . . . . 255

LIST OF TABLES

Table

PART I

1. Volume of commercial timber outtake (ma/ha)
and the estimated density (No/ha) for
selectively cut compartments. A11 estimates
exclude rejected amounts of cut wood.

2. Diversity index values for saplings (>2cm
<5cm dbh) poles and small trees (>5cm <13cm
dbh) and large trees (313cm dbh) in
selectively cut (K-13, K—14, K-15) and uncut
(K-SO) forest tracts of the reserve. N is the
number of plots each 5mx20m.

3. Results of forest vegetation sampling in
uncut mature forest (K—30), lightly cut (K-
14), heavily cut (K-15), and heavily cut and
treated forest (K-13) in the reserve.
Measurements included saplings (>2cm <5cm
dbh), poles and small trees (>5cm <13cm dbh)
and large trees (313cm dbh). N is the
number of plots each 5mx20m . .

4. Percent reduction in density and basal area
of trees, poles and saplings in selectively
cut forest compartments of the reserve. K-30
was uncut and was used as control

5. Importance value (IV) and ranks for desirable
(also forest canopy) and some non—desirable
timber species for uncut (K-30) and
selectively cut (K-14, K-15, K-13) forest
tracts of the reserve . . . . . . . .

6. Summary of the results of forest canopy
cover, percent gap light inside the forest,
and forest ground vegetation cover in cut and
uncut forest tracts of the reserve.

7a. Pearson's correlations between forest

vegetation parameters of uncut forest (K-30)
in the reserve. .

xi

Page

37

38

48

51

54

57

59

Table i Page

7b. Pearson's correlations between forest
vegetation parameters of lightly cut forest
(K-14) in the reserve . . . . . . . . . . . . . 62

7c. Pearson's correlations between forest
vegetation parameters of heavily selectively
cut forest (K-15) in the reserve. . . . . . . . 65

7d. Pearson's correlations between forest
vegetation parameters of heavily selectively
cut and treated forest (K-13) in the reserve. . 66

8. Size class distribution of tree species in
the study compartments of the reserve and
potential contribution to the possible
maximum standing stock since harvest with K-
30 as a control (=100%) . . . . . . . . . . . . 68

9. Saplings and pole distribution in cm dbh size
classes for the four study areas. . . . . . . . 72

10. Correlative comparisons of the density of
desirable and some important non-desirable
tree saplings and poles of the study plots. . . 75

A1. Summary of Analysis of variance and Duncan's
Multiple Range Test on some forest vegetation
enumeration parameters. . . . . . . . . . . . . 99

A2 Estimates of density of desirable species
sampling (Sa) and Poles (Po) for the study
plots . . . . . . . . . . . . . . . . 100

PART II

1. Parameters of tree death rates for the
lightly cut forest (K-14), heavily
selectively cut (K-15) and uncut mature
forest (K—30) of the reserve. . . . . . . . . . 117

2. Density (No/ha) of desirable and important
non-desirable timber or canopy tree species
involved in tree falls, and dead standing
trees in cut and uncut forest tracts of the

reserve . . . . . . . . . . . . . . . . . . . . 123
3. Tree mortality records from Barro Colorado
Island, Malaysia and Uganda, East Africa. . . . 140

A1. Monthly distribution and density (No/ha) of
tree and branch falls in lightly selectively
cut forest (K-14) . . . . . . . . . . . . . . . 156

xii

Table Page

 

A2. Monthly distribution and density (No/ha) of
tree and branch falls in heavily selectively
cut forest (K-15) . . . . . . . . . . . . . . . 157

A3. Monthly distribution and density (No/ha) of
tree and branch falls in uncut mature forest

(K-30). . . . . . . . . . . . . . . . . . . . . 158

A4. Size-class distribution of trees in terms of
density (No/ha) of stem snaps, tree uproots,
and dead standing trees in selectively cut

(K-14, K-15) and uncut mature forest (K-30) . . 159
PART III
1. Average sapling and pole (21.5cm <10cm dbh)

species richness/m and density (No/ha) for

the 20 randomly selected gaps (G) and

adjacent forest plots (F) for each of the

selectively cut (K-14 and K-15) and uncut (K-

30) forest tracts of the reserve. . . . . . . . 185

2. rThe density (No/ha) of saplings and poles
(range >1.5cm 510cm dbh) of the common
desirable timber or canopy tree species in
central forest reserve regenerating in forest
gaps (G) and forest plots (F) adjacent to the
gaps in selectively cut (K-13, K—14, K-15)
and uncut (K-SO) forest tracts. . . . 189

3. The frequency of elephant incursions and
percentage of marked forest gaps used by the
elephants as browsing sites in lightly cut
(K-14), heavily selectively cut (K-15) and
uncut forest (K—30) in the reserve. . . . . . . 191

4. Rates of seed predation in forest gaps of
lightly cut (K-14), uncut (K-30):and heavily
selectively cut (K-15) forest tracts of the
reserve . . . . . . . . . . . . . . . . . . . 193

5. Mean plant species richness and density
(No/m ) germinated in 1m plots of the top
100m of forest soil (F) and gap soils (G)
from K—14, K—15 and K-30 and grown in
screened nursery. . . . . . . . . . . . . . . . 195

xiii

 

Table

6.

A1.

A2.

A3.

A4.

A5.

Mean density (No/m2) and frequency of
occurence (100%=1) of tree seedlings
germinated from 1m plots of the top 10cm of
forest soil and forest gap sail tended in
screened nursery. No. of 1m plots for K- 14
K- 15 and K— 30 were equal, n=14.

Mean canopy species sapling and pole (>1.5cm
510cm dbh) density (No/ha) per species and
percent species population (in brackets)
regenerating with normal stems (N), as
coppices (c) or as stem sprouts (SS) in the
forest gaps of selectively cut (K—13, K—14,
K— 15) and uncut (K—SO) forest in the

reserve . . . . . . . . . . .
Comparison of soil seed germination results
for some tropical rain forests. . . .

Size-class distribution of 40 randomly
selected gaps and the distribution of
saplings (Sa) and poles (P) (>1. 5cm <10cm
dbh) in the gaps of heavily selectively cut
and treated forest (K- 13)

Size—class distribution of 40 randomly
selected gaps and the distribution of sapling
(Sa) and poles (P) (31.5cm <10cm dbh) in the
gaps of lightly selectively cut forest (K-14)

Size—class distribution of 40 randomly
selected gaps and the distribution of
saplings (Sa) and poles (P) (31.5cm <10cm
dbh) in the gaps of heavily selectively
cut forest (K-15) . . . . . . . .

Size—class distribution of 40 randomly
selected gaps and the distribution of
saplings (Sa) and poles (P) (>1. 5cm <10cm
dbh) in naturally formed tree fall gaps of
uncut mature forest (K— 30).

Mean species richness and density (No/m2) of
plants newly germinated in manipulated gap

plots of K—14, K—15 and K—SO. n=20 gaps for
K-14, K—15 and K-30 . . . . . . . . .

Page

202

208

232

255

256

257

259

Table

A6.

The density (No/ha) and percent species
population (on the right) of canopy tree
species saplings and poles (range 31.5cm
<10cm dbh) regenerating with normal stems
(N), as coppices (C) or as stem sprouts (SS)
in 25 forest gaps of uncut mature forest (K—
30) . . . . . . . . . . . . .

XV

Page

260

 

LIST OF FIGURES
Figure Page
GENERAL INTRODUCTION

1. Approximate locations of major forest

reserves and geographical features of Uganda.

The forest reserves include 1-Zoka, 2- '

Budongo, 3—Bugoma, 4-Itwara, 5-Kibale (where

the research was conducted), 6-Ruwenzori, 7—
Kasyoha-Kitomi, 8-Kalinzu, 9-Maramagambo, 10—
Impenetrable, 11—Mgahinga, 12-Malabigambo,

13-Mabira, and 14-Mount Elgon . . . . . . . . . 2

2. The Kibale Forest Reserve: showing
approximate locations of areas protected as
Nature Reserves (NR) and the Research Plot
(RP-703) where part of this study was
conducted . . . . . . . . . . . . . . . . . . . 3

3. Map of Kibale Forest Reserve showing the
position of the three Forest Blocks, major
river systems and the extent of forest and
grasslands. This study was conducted in the
Central Block of the reserve near Kanyawara

Forest Station. . . . . . . . . . . . . . . . . 6
4. Monthly Rainfall at Kanyawara during the
study period. Data was obtained from Dr.
Thomas Struhsaker's weather records . . . . . . 8
PART I
1. Map of Uganda Showing the position of Kibale

Forest Reserve. The Reserve is

administratively divided into the North

Block, Central Block and South Block. The

whole reserve is a mosaic of 7 forest types . . 23

2. Map of Kibale Forest Reserve showing the
approximate position of the study area and

relative location of timber compartments
included in the study. The Butanzi, Kyakato

and Nyamusika hills have been planted with

exotic species of Pines and Cypress. The

shading indicates the distributional limits

of the Parinari forest type . . . . . . . . . . 25

xvi

 

 

Figure Page

3. Culmulative increase in the number of species
of saplings (>2cm <5cm dbh), poles and small
trees (>5 <13cm dbh) and large trees (213cm
dbh) with corresponding increase in forest area
sampled in lightly cut (K- 14) and uncut (K-
30) forest tracts of the reserve. . . . . . . 41

4. Cumulative increase in the number of species
of saplings (>2cm <5cm dbh) poles and small
trees (>5cm <13¢m dbh) and large trees (313cm
dbh) with corresponding increase in forest
area sampled in heavily selectively cut and
treated forest (K-13) and the heavily cut but
untreated forest (K-15) of the reserve. . . . . 42

5. Mean species richness for sample plots in
uncut (K-30), lightly cut (K-14), heavily cut
(K-15) and heavily cut and treated (K—13)
forest tracts of the reserve. . . . . . . . . . 45

6. Correlations between seedling (<2cm dbh) and
sapling (>2cm <5cm dbh) species richness and
the level of forest ground vegetation cover

of the study plots in the reserve . . . . . . . 61
7. Correlation between sapling density (No/ha

X10 ) and percent forest ground vegetation

cover of the study plots in the reserve . . . . 63

8. Stand curves for uncut forest (K-30) lightly
cut forest (K-14), heavily selectively cut
forest (K-15), and the heavily selectively
cut and treated forest tract (K-13) of the
reserve . . . . . . . . . . . . . . . . . . . . 69

PART II

1. Sectors of study compartments 14 (K-14) and
30 (K-30) showing tree fall census routes
along which tree falls were enumerated for 21
consecutive months. Only a few numbered or
lettered gridlines or trails have been
included in order to enhance clarity of the
census route (thick line) . . . . . . . . . . . 106

2. Sectors of study compartment 15 (K-15)
showing the tree fall census route (thick
line) along which tree fall enumerations were
conducted for 21 consecutive months . . . . . . 107

xvii

Figure

3.

Frequency distribution of tree falls with
respect to their perpendicular distance of
location in the census strip from the census
trail. The 2% or lower cut off points of K-
14 and K-15 tree falls was estimated at 10m
and 12m for K—30. . . . . . . . . .

Monthly density of total tree falls (Snaps +
Uproots) for the lightly cut (K-14), heavily
selectively cut (K— 15) and uncut mature
forest (K- 30) compared with monthly rainfall
(e——-—o) . . . . . . . . . . . . . .

Rate of tree and branch fall in terms of the
density of tree uproots, stem snaps, and big
branch falls (i.e. No/ha/year) expressed as
the percentage of the density (No/ha) of
susceptible trees in lightly cut (K-14),
heavily selectively cut (K-15) and uncut (K-
30) forest tracts of the reserve. . . .

Rates of live and dead tree and branch falls
(No/ha/yr) and monthly estimates of the
density of dead standing trees (No/ha/month)
expressed as precentage of the density of
susceptible trees in lightly cut (K-14),
heavily selectively cut (K—15) and uncut (K-
30) forest tracts of the reserve. . . .

Size—class (10cm dbh interval) distribution

of tree falls (No/ha) attributed to stem snaps
and tree uproots in lightly selectively cut
(K-14), heavily selectively cut (K-15) and
uncut (K-30) forest tracts of the reserve

Frequency distribution of diameter at breast
hight (10 cm abh interval) of treefalls (snaps
+ uproots) expressed as the percentage of the
density (No/ha) of living stems in the same
diameter class in lightly cut (K-14), heavily
cut (K-15) and uncut (K-30) forest tracts of
the reserve . . . . . . . . . . . . .
Frequency distribution of the diameter at
breast height (10 cm dbh interval) of dead
standing trees expressed as the percentage of
the density (No/ha) of living trees in the same
diameter class in lightly cut (K-14), heavily
selectively cut (K-15) and uncut (K-30) forest
tracts of the reserve . . . . . . . . .

xviii

Page

110

114

119

121

127

128

130

.LuMLJ

Figure Page
PART III

1. Approximate location of the forty randomly
selected forest gaps per forest compartment
used for gap vegetation studies in lightly
cut (K-14) and uncut (K-SO) forest tracts of
the reserve . . . . . . . . . . . . . . . . 167

2. Approximate location of forty randomly
selected forest gaps per compartment used for
gap vegetation studies in the heavily
selectively cut forest (K—15) and heavily
selectively cut and treated forest (K—13) in
the reserve . . . . . . . . . . . . . . . . . . 168

3. Size-class distribution of 40 randomly
selected gaps in uncut mature forest (K—30),
lightly selectively cut forest (K-14),
heavily selectively cut forest (K—15) and the
heavily selectively cut and treated forest
(K—13) tracts of the reserve. . . . . . . . . . 176

4. The density (No/ha) of saplings and poles of
all tree species combined and desirable tree
species (31.5cm to 10 cm dbh) per species in
the forest gaps of lightly selectively cut
(K-14), uncut mature forest (K—30) and the
heavily selectively cut (K—15) and the
heavily selectively cut and treated (K—13)
forest tracts of the reserve. . . . . . . . . . 178

5. Percent contribution of the four plant life
forms to plant pecies richness and density
of plants (No/m ) germinated in 1m plots of
the top 10cm of forest soil and gap soil
collected from the forest gaps and forest
plots adjacent to the gaps, respectively, in
lightly cut (K-14), heavily selectively cut
(K-15) and uncut mature forest (K-30) and
grown in screened nursery. T=tree seedlings,
S=shrub species seedlings, H=herbaceous
species seedlings and V=vines and climber
seedlings . . . . . . . . . . . . . . . . . . . 197

6. Comparison of tree seedling specieszrichness
and density of trSe seedlings (No/m ) newly
germinated in 1 m manipulated gap plots
in forest gaps of K—14, K-15 and K—30 after
6 months and 20 months of experimentation . . . 205

Figure

7.

Page

Mean percent canopy tree species population
of sapling and poles regenerating with normal
stems (N), as coppices (C) or as stem sprouts
(SS) in the forest gaps of uncut (K-30),
lightly cut (K-14), heavily selectively cut
(K-15), and heavily selectively cut and

treated forest (K-13) in the reserve. 209

XX

 

NOMENCLATURE OF PLANTS CITED IN THE TEXT

TREE SPECIES

Albizzia gummifera (Gmel.) C. A. Sm.
Aningeria altissima (A. Chev) Aubrev. & Pellegr.
Antiaris toxicaria (Rumph. ex. Pers.) Lesch.
Aphania senegalensis (Juss. ex Bernh.) Radlk.
Blighia unijugata Bak.

Bosquiea phoberos Baill.

Carapa grandifolia Sprague

Celtis africana Burn. f.

Celtis durandii Engl.

Chaetacmea aristata Planch.
Chrysophyllum_gorungosanum G. Don.

Clausena anisata (Willd.) Oliv.

Cordia millenii Bak.

Cynometra alexandri C. H. Wright

Diospyros abyssinica (Hiern) F. White
Entandrophragma utile (Dawe & Sprague) Sprague
Fagara macrophylla (Oliv.) Engl.

Fagaropsis angolensis (Engl.) Dale

Funtumia latifolia (Stapf) Stapf ex Schltr.
Lovoa swynnertonii Bak. f.

Maesa lanceolata Forrsk.

Markhamia platycalyx (Bak.) Sprague

Mimusops bagshawei S. Moore

Neoboutonia macrocalyx Pax

Newtonia buchananii (Baker) Gilb. & Boutique
Olea welwitschii (Knob.) Gilg. & Schellenb.
Parinari excelsa Sabine

Piptadeniastrum africanum (Hook. f.) Brenan
Pterygota mildbraedii Schott. & Endl.

Pygeum africana Hook. f.

Strombosia scheffleri Engl.

Symphonia globulifera L. f.

Teclea nobilis Del.

Trema guineensis Ficalho

Trichilia splendida A. Chev.

Uvariopsis congensis Nov.

Wurbagia ugandensis Sprague

xxi

NON-TREE SPECIES

Aspilia africana (Pers.) C. D. Adams
Bidens pilosa L.

Brillantaisia nitens Lindau
Cymbopogon afronardus Schult.
Fleurya urophylla Mildbr.
Imperata cylindrica Beav.
Mimulopsis solmsii Schweinf.
Monechma subsessile Hochst.
Palisota schweinfurthii C. B. Cl.
Pennisetum purpureum Schum.
Pollia condensate C. B. Cl.

Nomenclature after Daydon (1895), Eggeling and Dale

and Hamilton (1981).

xxii

(1951)

GENERAL INTRODUCTION

Historical Backgroggd to the Study

The study was conducted in Uganda, a land—locked
equatorial country bordered by the Sudan to the North, Kenya
to the East, Tanzania and Rwanda to the South and Zaire to
the West (Figure 1). Uganda is blessed with a diversity of
freshwater lakes, rivers, swamps and great mountain ranges.
Consequently, a diversity of habitats occur in the country.
Most habitats are transitional between the lowland tropical
rain forests of Zaire to the West and the dry savanna of
Kenya to the East.

Uganda has several forest reserves, mostly concentrated
in the western part of the country (Figure 1). They are
government controlled and protected reserves (Kingston,
1967), established to serve several functions, such as
protection reserves (e.g. the Ruwenzori and Mount Elgon) or
reserves for limited and controlled exploitation. The first
legal status of the Kibale Forest Reserve was through
gazetting as a crown forest in 1932 and being made a Central
Forest Reserve in 1948. It was not until 1963 that
consolidated gazetting was effected with the reserve
officially covering an area of 560 sq km (Kingston 1967).

Two areas within the reserve, one 59.8 km2 and other only

1A

Figure 1. Approximate locations of major Forest Reserves and
geographical features of Uganda. The Forest Reserves
include: 1-Zoka, 2-Budongo, 3-Bugoma, 4-Itwara, 5-Kibale
(where the research was conducted), 6-Ruwenzori, 7-Kasyoha-
Kitomi, 8-Kalinzu, 9-Maramagambo, 10-Impenetrable, 11-
Mgahinga, 12-Malabigambo, 13-Mabira and 14-Mount Elgon.

 

 

 

 

 

 

 

 

 

  
    
  
 

f;
9
S SUDAN
53"
S
‘ I
g I
v UGANDA
Ictoria 4’.
N ZAIRE ’49
I
/’ L.Kvoga
I
I 14
I.
\4
1L S’Kihale 13
5 Forest KAMPALA o KENYA
ALGeorge A 4 A 0
LEdward , 7 P, O
' I
4‘8 ‘
g LAKE VICTORIA I
‘10 -------- l \
RWANDA TANZANIA
fl

  

 

 

Scale in Kilometers
F I

fi

160

2A

Figure 2. The Kibale Forest Reserve: Showing approximate
location of areas protected as Nature Reserves (NR) and the
research plot (RP-703) where part of this study was
conducted.

 

 

 

, FORT room (:\.\

\ ,.. T0 KAMPALA
.— . ./

 

[4
IR.
II
I

LI-
2

1‘4 '- 3—_ %:

Tamra—witty: ".
_$__TATIou__s -.- ....., _ _

 
 
  

 

If, Roads "

Rivers 8. I; "—
7 Streams ‘

...—-\°.:../ _ ., '
3 Forest b-s 3.. :’.q

 

i331".- Grasslands. :33. a"
.. ti ->To mmwmos
o 5 g " u -

 

 

L—1—_I I: "‘ .
“I“ I “ti—03TH

i ELF!;¥§

Fig... -.—1

:x;::. ququ
_' J41: :1 /rom roam

 

 

 

 

 

 

 

 

2.0 km2 were declared Nature Reserves (Figure 2) where any
form of human interference with nature is strictly
prohibited. ~Another area, 1210 ha, also within the reserve.
was designated as Research Plot 703 to promote research.

In Uganda, like elsewhere in the tropics, the major
cause of forest destruction centers on the conversion of
forests to agricultural land, uncontrolled cutting of trees
for fuelwood, timber and building poles, and poaching of
game (Goodland 1980: Committee on Res. Priorities in Trop.
Biol. 1980). With the ever increasing human population as
the ultimate cause of forest destruction, the situation in
Uganda is expected to be complicated or worsened by changing
demographic and social circumstances, economic uncertainty.
lack of modern forest laws, and ineffective enforcement of
these laws.

The area of Uganda's land now under closed natural
forest is well below the 4.6% estimated in 1958 (Langdale-
Brown 1960, cited by Hamilton 1984). Currently, the Uganda
Forest Department administers 7720 km2 of natural forest and
this represents only 3.3% of Uganda's total area (FAO-UNEP
1981). The Forest Department accepted a system adopted by
the sawmillers during the first half of this decade which
was a modified form of monocyclic system (Whitmore 1975).
apparently destructive to species-rich tropical forests.

Trees were harvested regardless of size or rarity and no

enrichment planting was undertaken. Extensive damage was

done to the remaining trees, poles. seedlings and saplings

by the practice of dragging logs over long distances to
extraction roads. Sawmill operations of this kind are now
underway in Budongo, Mabira and Itwara Forest Reserves. The
Forest Department authorities are reluctant to abandon such
forest felling practices as there is a lack ofstrong
evidence to challenge the appropriateness of the system. To
resolve the predicament, the Kibale Forest Project proposed
a study of the current status of regeneration under a range
of forest disturbance regimes incuding mechanized logging.
Location ogithe Stggy7Site and Climate

The Kibale Forest Reserve is a government controlled
reserve in Kaborole District, Western Uganda. It is
situated approximately 24 km. east of the foothills of the
Mountains of the Moon (Ruwenzori) and 10 km from the town of
Fort Portal (Figure 3). It is equatorial in position (0°
13' to 0° 41' N and 30° 19' to 30 ° 32' a) and lies on an
undulating plateau slightly tilted to the south (Kingston
1967, Wing and Buss 1970). The altitude varies between 1590
m in the northern extreme of the forest to 1110 m in the
south. The Fort Portal-Kampala road cuts the reserve in the
northern part while the Fort Portal-Kamwenge road cuts it in
the South. The three forest blocks formed by the road cuts
include the North Block (13.55 kmz) the Central Block
(361.51 kmz) and the South Block (184.49 km2). The reserve
forms a long strip running north and south for 56 km on its
longest axis (Kingston 1967) and covers an area of 560 sq.

km. Numerous hills, valleys, swamps and streams cause

5A

 

Figure 3. Map of Kibale Forest Reserve showing the position
of the three Forest Blocks, major river systems and the
extent of forest and grasslands. This study was conducted
in the Central Block of the reserve near Kanyawara Forest
Station.

 

FORT PORTAL

   
   
 

NORTH BLOCK
('\
. /"s

‘ .. '. .. ....'I:.. I‘
a . f0 .' 0 o . . e o
' x. ~'..-: ~--.: "\
:::-.-:..-.=....:-:.::-- CENTRAL -

BLOCK

.V . ’ 0-0”.

KAMWENGE
ROAD

TO KAMPALA

I

./
)

so”

altitudinal differences in the relief. The streams feed
into the two major rivers, Mpanga and Dura, which flow
southwards into Lake George. The South Block is 6 km north
of the Queen Elizabeth (formerly Ruwenzori) National Park.
The rainfall is low to moderate (mean annual
preciptation 1700 mm) and distributed in two nearly equal
maxi-a and minima (Kingston 1967, Struhsaker 1975). The
typical two wet seasons are March to May and September to
November and the two dry seasons are in December to February
and June to August (see Figure 4). However, the onset of
both wet and dry seasons varies from year to year as does
their intensity. Locally, there is more rain in the north
than in the south of the reserve. Wind is generally mild.
being strongest immediately prior to and during the
convectional rainstorms. Because of the elevation, the
temperature of the reserve is constantly low compared to
those of most equatorial regions (Wing and Buss 1970) and is
stable over the years with an annual average of 20.5 °C
(Struhsaker's Weather Records 1977-1979). The reserve
forest is distinguished from lowland tropical rainforest on
account of greater altitude, lower temperature and lower

rainfall but is physiognomically similar (Langdale-Brown et
al. 1964).

Geology and Soils
The geology and soils of the Kibale Forest have been

extensively described by Osmaston, 1959; Harrop, 1962, Lang-

Brown and Harrop, 1962: and Langdale-Brown 55 al. 1964. Two

 

7A

Figure 4. Monthly rainfall at Kanyawara during the study
period. Data were obtained from Dr. Thomas Struhsaker's
weather records.

mIPZOE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0mm _. w mwm _. .vwm _.

om<ss§<susozom<sss<sasnzo
ll 1 _r Tm.

i— L

l4. ll

1 .. l.

i. .. II

”Mu I .J

 

 

 

 

 

 

 

 

 

 

IT
IHIH

 

 

 

 

 

 

 

 

 

 

 

H

 

 

 

( ww) TIV:IN|VH A1H.LNO|N

 

types of rocks formed in the Precambrian period and
underlain by undifferentiated acid gneisses constitute the
major substrate of the reserve. The Toro System overlying
these rocks forms prominent ridges mainly of quartzite and
sometimes schists and phyllites which are intruded by
amphibiolites, gneisses and granites (Lang-Brown and Harrop
1962 and Harrop 1962). Some hills in the reserve have
layers of hard laterite exposed on them.

About 90% of the reserve contains red ferrallitic
soils. The northern 2/3 consists mainly of ferrallitic
sandy clay loams while part of the remaining south has sandy
loams (Harrop 1962, Van Orsdol, 1983). These soils are
deeply weathered, shallow and stony and show little
differentiation into distinct horizons. Such lateritic
soils, therefore, have very few weatherable minerals and are
consequently infertile. Any sustained agricultural
production on these ferrallitic soils will definitely
require heavy investment in fertilization. The remaining
10% of the reserve has eutrophic soils overlying a base of
volcanic ash. These are characterized by high contents of
weatherable minerals and so are excellent for agriculture.
Such soils are limited to the western edge of the reserve in
Mpokya and Isunga areas where the problems of agricultural
encroachment on the reserve and forest destruction are the

highest.

10

Vegetation and Animals
The Xibale Forest is a medium altitude transitional,

moist forest falling between dry tropical and wet tropical
rainforest (Struhsaker-1975, Van Orsdol 1983, 1986). The
major habitat types of the forest reserve include swamp
forest and grasslands which intersperse with the forest to
form a mosaic of vegetation (Figure 3). 0f the 560 km2 of
the reserve. only 350 km2 are composed of forest and the
remainder is swamp and grasslands (Lockwood Consultants
1974). The grassland communities are essentially similar to
those found in savanna-woodland. i.e. tall grassland
communities in high rainfall areas with Pennisetum
purpureum. Imperata cylindrica and Cymbopogon afronardus
species dominating. The grasslands occupy the hilly summits
and are historically a result of past human settlement and
hunter's fires which maintain their integrity by rendering
natural colonization by tree species impossible.

In general. the northern, central and southern blocks
of the reserve vary in dominant tree species and
consociations. Osmaston (i959), Langdale-Brown et al.
(1964) and Wing and Buss (1970) found Parinari excelsa the
dominant species in the mature types north of the reserve
(circa 1500 m altitude) but with Strombosia scheffleri,
Aningeria altissima, Newtonia buchananii, 9353 gglgiggghii

and Carapa grandifolia as typical associates. These form
the richest timber types in the forest. In the Central

Block (1200-1500 m altitude) the forest is of a mixed type

ll

dominated by species of Chrysophyllum gorungosanum, Celtis
africana, Markhamia platycalyx and Diospyros abyssinica with
g; excelsa occuring in low numbers. The South Block ( 1200
m altitude) is dominated by species of glgg welwitschii ,
Pterygota mildbraedii and Piptadeniastrum africanum and a
mixture of ngoa swynnertonii, Diospyros abyssinica and
Markamia platycalyx forests. However, the lower altitudes
and river valleys in the South Block are occupied by pure
stands of Ironwood, Cynometra alexandri, with Celtis
durandii, Celtis africana, Diospyros abyssinica and
Wurbagia ugandensis sparsely present.

In the Central Forest Block, the high forest is a
mixture of deciduous and evergreen tree species with
evergreens predominant. Trees rise to over 56 m in the
mature forest (Struhsaker 1975, Wing and Buss 1970) and
exhibit a closed canopy of stratified tree crowns. .Here,
the undergrowth is sparse and characterized by shade
tolerant herbs and shrubs. Plants such as Palisota
schweinfgrthii, Pollia condensate and a variety of broad
leaved forest grasses and ferns abound. Selective felling
up to 1969 resulted in about 368 of the northern and a big
part. of the central forest block being harvested
(Struhsaker 1975). This caused giant gaps in the forest
canopy which allow direct sunlight to reach the forest
ground. Consequently, the dense growth of herbaceous and
non-woody vegetation dominated by Agggthgg species

characterizes the secondary forests.

12

The Kibale Forest Reserve supports one of the richest
arrays of fauna of any forest in East Africa (Van Orsdol.
1983, 1986). ”Prominent among these are the eleven species
of non-human primates. These include the rare and
endangered red colobus monkey (Colobus badius), black and
white monkey (Colobus guereza), the chimpanzee (gag
troglodytes), red tail monkey (Cercopithecus ascanius), blue
monkey (Cercopithecus mitig), lhoesti monkey (Cercopithecus
lhoesti), mangabey (Cercocebus albigena) and olive baboon
(£3219 anubis). The red colobus in the reserve represent
the only remaining viable population of this species in
Uganda (Struhsaker 1976).

The common terrestrial mammals in the reserve include
two species of red and blue duikers (Cephalophus harveyi and
g; monticola, respectively), bush buck (Tragelaphus
scriptus), bush pig (Potamochoeggg porous), warthog
(Phacochoerus aethiopicus) and the buffalo (Sygcerus
caffer). Kibale is also a very important sanctuary for the
African elephant (Loxodonta africana) population. After
felling operations in the Kibale forest in 1968, elephant
control by shooting was practiced. This, combined with
excessive poaching a few years after the culling, has
severely reduced the elephant populations to very low levels
(Bltringham and Malpas 1980). The waterbuck (ggbgg
defassa), sitatunga (Tragelaphus gpgkgi) and the giant
forest hog (Hylochoerus mginggtghggggi) also utilize the

reserve .

13

The carnivore population, as reported by Struhsaker
(1980) (and personal observation) include the lion (Panthera
$59 - an occasional visitor to the reserve), leopard (g;
pardus — very rare), golden Cat (Eglig aurata), serval (g;
serval), African civet (Viverra civetta), palm civet ‘
(Nandinia binotata) ratel (Mellivora capensis) and the Congo
clawless otter (Aggy; congica).

0f the small mammals, the rodents are diverse and
ubiquitous. Their role as consumers and possible predators
in the food chain is beginning to be recognized through the
studies of Isabirye-Basuta (1979) and Kasenene (1980). The
data base is still small, however, compared to the studies
on grassland rodents in the Ruwenzori (now Queen Elizabeth)
National Park (Delany, 1964b, Cheesman 1975). Except for
some non-human primates and some small rodents, no accurate
censuses or estimates have been made on the other mammal
residents of the reserve.

The avifauna of Kibale forest is also rich. Joseph
Skorupa of the University of California, Davis (unpublished)
has listed over 325 species occurring in 46 families. Some
of these include the crowned hawk eagle (Stephonoagggg
coronaggg), martial eagle (Polemaetus bellicosus), giant
blue turaco (Corythaeola cristata), gray parrot (ggittgggg
erithacus) and the black and white casqued Hornbill
(Bycanistes subcylindricus). The Kibale ground thrush
(Zoothera kibalensis) was first described in 1979 and has

not been reported from anywhere outside of Kibale forest.

14

More details and description of the forest flora and
fauna and general history of the reserve can be obtained
from Kingston, 1967; Wing and Buss, 1970; Struhsaker, 1975;
Isabirye-Basuta, 1979; and Kasenene 1980.

The Study Context, Objectives and Hypotheses

In Uganda, the greatest threats to the integrity of
tropical forest ecosystems include increased consumer and
commercial demands locally and across the borders, the
omnipresent and increasing human population and the
consequent demand for agricultural land. Specifically, the
problems of agricultural encroachment and human settlement
in the reserves, poaching, rampant firewood collection and
charcoal burning for both home and local markets and, worst
of all, uncontrolled selective logging (mechanised and
pitsawing) are prevalent. The effects of these disturbance
factors on the forest reserve are only in initial stages of
assessment and still poorly understood. The official
government policy towards the tropical forests is still
rapid exploitation for timber and the methods employed are
uncontrolled and destructive. Lack of adequate information
on how much destruction logging systems impose on the
forest prompted this study. Currently, the recommended
method for forest exploitation is polycyclic, where a given
forest tract is selectively felled once every 70 years
(Kingston 1967). Considering the fact that this system
employs heavy machinery, it definitely cannot minimize

damage in species-rich tropical forests. I therefore

15

question the appropriateness of these polycyclic systems,
which, because of too intensive felling, turn into a
monocyclic system in the actual exploitation of species-rich
tropical forests. It is inevitable that we have to exploit
our forest resources but the methods recommended for
exploitation should minimize disturbance to important
biological processes (Tie 1973, Ashton 1978).

The Kibale forest's heterogeneity in terms of
vegetation communities, topography, soils and most
importantly, level of disturbances (human induced and
natural) made it suitable for my study. Therefore, I aimed
at assessing the human impact on the forest resource base
and the forest's response to both anthropogenic and natural
forms of disturbance. Forest disturbances generate forest
gaps of various configurations and sizes. Natural tree
falls (and thus forest gaps) are considered to form an
important integral component of the forest dynamics
(Hartshorn 1978, Schenske and Brokaw 1981). Therefore, I
hypothesized that the rate and level of forest disturbance
(thus the rate and extent of forest gap formation) are
inversely related to tree species diversity and density.
rate of forest recovery and general regeneration success of
desirable canopy or timber species.

The major objectives of the study were:

1) Characterization of the mature and secondary forest

communities of the reserve with respect to plant

2)

3)

4)

16

species composition, diversity, structure and
regeneration.

Quantification and analysis of the frequency, extent
and major causes of forest disturbances or forest gaps
in the mature and secondary forest communities.
Determination of the influence of natural and human
induced forest gaps on forest vegetational parameters
such as density, diversity, frequency of occurrence,
predation levels, recruitment and mode of regeneration
(with adjacent undisturbed forest serving as the
control).

Comparison of the results of the study with related
studies conducted in other tropical forests in order to
understand the dynamics of tropical moist forest, more

generally, under the influence of disturbances.

It is hoped that the elucidation of tree death rates,

recruitment into gaps and regenerational dynamics of the

tropical moist forest under a range of disturbance regimes

should add to our understanding of their dynamics. This

should form a basis for the development of meaningful plans

for maintenance of tropical moist forest productivity and

for accelerating the recovery of exploited areas.

LITERATURE CITED

Ashton, P. S. 1978. Crown Characteristics of tropical trees.
In. Tropical trees as Living Systems pp. 591-614
(eds. P. B. Tomlinson and M. H. Zimmerman) Cambridge
University Press, New York.

Cheesman, C. L. 1975. The Population Ecology of Small
rodents in the grassland of Ruwenzori National Park.
Uganda. Ph.D. Thesis. Southampton University.

Committee on Research Priorities in Tropical Biology,
Report, 1980. Research Priorities in Tropical Biology.
National Academy of Sciences, Washington, D.C.

Daydon, J. B. 1895. Index Kewensis: An enumeration of the
genera and species of flowering plants. Vol. 1 and 2.
Oxford University Press, London.

Delany, M. J. 1964b. An Ecological Study Mammals in Queen
Elizabeth National Park, Uganda. Rev. Zool. Bot. Afr.
70:129-147.

Eggeling, W. J. and I. R. Dale, 1951. The indigenous trees
of the Uganda Protectorate. The Government Printer.
Entebbe, Uganda.

Eltringham, S. K. and R. C. Malpas. 1980. The decline in
Elephant numbers of Rwenzori and Kabalega Falls
National Parks, Uganda. Afr. F. Ecol. 18:73-86.

FAO-UNEP. 1981. Forest Resources of Tropical Africa FAO,
Rome.

Goodland, R. J. A. 1980. Environmental Ranking of Amazonia
Development Projects in Brazil. Environmental
Conservation 7:9-26.

Hamilton, A. C. 1981. A field guide to Uganda forest trees.
Privately published, Uganda.

Hamilton, C. A. 1984. Deforestation in Uganda. Oxford
University Press, Nairobi. 95pp.

Harrop, J. 1962. Soils: In: Atlas of Uganda Lands and
Surveys, Uganda, 83 pp.

17

18

Hartshorn, G. A. 1978. Tree Falls and Tropical Forest
Dynamics. In: Tropical Trees as Living Systems. pp 617-
638 (eds. P. B. Tomlinson and M. H. Zimmerman)
Cambridge Univeristy Press, New York.

Isabirye-Basuta, g. 1979. The Ecology and Biology of Small
Rodents in the Kibale Forest, Uganda. M.Sc. Thesis,
Makerere University.

Kasenene, J. M. 1980. Plant regeneration and rodent
populations in selectively felled and unfelled areas of
the Kibale Forest, Uganda. M.Sc. Thesis, Makerere
University.

Kingston, B. 1967. Working Plan for the Kibale and Itwara
Central Forest Reserves. Uganda Forest Department,
Entebbe. 142 pp.

Lang-Brown, J. R. and H. Harrop. 1962. The Ecology and Soils
of Kibale Grasslands: Uganda. E. Afr. Agr. For. J.
27:264-272.

Langdale-Brown, I., H. A. Osmaston and J. G. Wilson. 1964.
The Vegetation of Uganda and its bearing on Land Use.
Uganda Government Printer, Entebbe. 159 pp.

Lockwood Consultants. 1974. Forest Resource Development

Study, Republic of Uganda. Lockwood Consultants.
Toronto, Canada.

Osmaston, H. A. 1969. Working Plan for the Kibale and Itwara
forests. Uganda Forest Department, Government Printer,
Entebbe. 60pp.

Schenske, D. W. and N. Brokaw. 1981. Tree falls and the
distribution of understory birds in a tropical forest.
Ecology 62:938-945.

Struhsaker, T. T. 1975. The Red Colobus monkey. University
of Chicago Press, Chicago. 311pp.

Struhsaker, T. T. 1980. The Kibale Forest Project Pamphlet.
New York Zoological Society, New York.

TIE. Report. 1973. Fragile Ecosystems: Evaluation of
Research and Application in Neotropics 258pp. (eds. E.
G. Farnworth and F. B. Golley) Springer-Verlag New
York. 258pp.

Uhl, C., C. Jordan, K. Clark, H. Clark and R. Herrera 1982b.
Ecosystem recovery in Amazon Caatinga forest after
cutting, cutting and burning and bulldozer clearing
treatments. Oikos 38:313-320.

19

Van Orsdol, K. G. 1983. The status of Kibale Forest
Reserve, of Western Uganda and recommendations of its
conservation and management. A report to Ministry of
Agriculture and Forestry, The New York Zoological
Society and African Wildlife Foundation.

1986. Agricultural encroachment in Uganda's
Kibale Forest. Oryx 22:115-118.

 

Whitmore, T. C. 1975. Tropical Rain Forests of the Far East.
Clarendon Press, Oxford, 282pp.

Wing, L. D. and I. O. Buss, 1970. Elephants and Forests.
Wildlife Monograph #19. The Wildlife Society,
Washington, D.C.

PART I

THE COMPOSITION, STRUCTURE AND REGENERATION
OP SELECTIVELY CUT AND UNCUT FOREST

TRACTS OF THE RESERVE

INTRODUCTION

Tropical Forest Disturbances and the Need for Research

The rate at which tropical forest ecosystems are being
indiscriminately converted and/or transformed poses a threat
to their integrity and future existence (Richards, 1964:
Sanderson and Loth, 1965: Brenan, 1973; Fosberg, 1973; TIE
Report, 1973: Longman and Jenik 1974, Chapman 1975, IUCN.
1975: Brazier _3131. 1976, USAID, 1978; Goodland, 1980; Lugo
and Brown, 1981: and Uhl p; 3;. 1985). Estimates of rates
of tropical deforestation and other forms of forest
disturbance that lead to degradation are variable and range
from 50,000 to 300,000 sz per year (Bolin 1977, IUCN 1900,
and Woodwell 1978)., By some estimates, total elimination of
mature tropical forests from the surface of the earth is
expected to occur shortly after the year 2000 (Richards
1973, Global 2000 Report 1980). Large forest tracts in
Central America, West Africa, and South East Asia are
already approaching depletion (Myers 1980, Plumwood and
Routley 1982). In Africa, forest destruction is estimated
at approximately 2 million ha/year and the rate of forest

destruction in Uganda must be correspondingly high (FAO-UNEP

1981).

20

21

It is well established that once the structural and
functional elements of a tropical rainforest are seriously
disrupted, they become very difficult to re-establish
(Gomez-Pompa 25 31. 1974; IUCN, 1975; Whitmore, 1975; Poore,
1976; USAID, 1978; UNESCO, 1978: Ewel, 1980; Plumwood and
Routley 1982 and Uhl et al. 1982). Thus, tropical forests
qualify for designation as "fragile ecosystems" which are
highly susceptible to human forms of disturbance (TIE Report
1973, Myers, 1980).

Tropical deforestation is of special concern because of
its potential impacts on genetic diversity (Myers 1979) and
biogeochemicl cycles (Longman & Jenik, 1974, Poore 1975,‘
IUCN, 1975; Wadsworth 1978 and Jordan and Herrera 1981).

The impact of tropical forest felling and burning is
potentially of importance to world climate patterns because
of its relationship to the global carbon cycle (Bolin 1977,
Wong, 1978; 1979; Woodwell 25 21., 1978). Again, tropical
deforestation has been shown to have important local impacts
on nutrient cycles, soil organisms and stored propagules,
all of which have long term effects on site recovery (Jordan
and Herrera, 1981; Ewel 53 51., 1981; Uhl g; 3;. 1982 b.

Putz 1983, 1984, Lugo and Murphy 1986).

Most studies in the tropics have concentrated on
forests of humid climates even though they account for a
relatively small portion of the forested landscape (Brown
and Lugo 1982, Murphy and Lugo 1986 a,b). Of the total

global extent of tropical forest, Brown and Lugo (1982)

22

estimated 258 was tropical and subtropical wet and
rainforest, 338 was tropical and subtropical moist forest
and 428 was tropical and subtropical dry forest (sensu
Holdridge 1967). The tropical moist and dry forests are
very little studied in spite of their extensive global
distribution and role in supporting human populations
(Murphy and Lugo 1986 a,b). Considerable ecological
attention is now being given to formerly neglected
successional ecosystems (Ewel, 1977, Goodland 1980, Uhl g5
31. 1982 b, 1985) for all categories of tropical forests.
The major reason is that successional ecosystems have
positive net yield and therefore potential for production

of food and fiber for human needs (Ewel 1977). Secondly,
successional ecosystems occupy tremendous areas, principally
because of shifting agriculture, logging operations, and
abandonment of futile ventures, e.g. improved cattle
pastures in lowland forests (Ewel 1971, 1980; Ewel g; 21.
1977, Uhl gt 3;. 1985, Goodland 1980). Therefore, research
on how tropical forests recover from a range of disturbance
regimes should help government planners and developers avoid

costly mistakes in land use.

General Management History pf the Study Compartments and
Plots

The Kibale Forest Reserve (henceforth called the

reserve) was selected for this study (Figure 1). It

represents one of the best examples of medium altitude
tropical moist forest ecosystems in East Africa (Struhsaker.

1975). It is a transitional moist forest between dry

22A

Figure 1. Map of Uganda showing the position of Kibale
Forest Reserve. The Reserve is administratively divided
into the North Block, Central Block and South Block. The
whole reserve is a mosaic of 7 forest types.

23

 

roar PORTAL (:7? -_.-_\.
Kc:

\J- -;.\

._smnou .1;
I \—_-_
I

lEGEND \

 

r.)

\

It Roads
Rivers &

‘7 Streams

_- "Forest
_;,.Grasslands

 

  

r4

/ CE_N

/ To KAMPALA

   
  
 

om

TR:A|.

 

UGANDA

FORT PORTAL

‘, Kibale
Forest Reserve

 

 

 

 

 

 

 

24

tropical forest and wet tropical rain forest (Langdale-Brown
_3 _l. 1964, Osmaston 1959a). According to the Holdridge
(1967) Life Zone Classification, the reserve would fall
between the dry and moist tropical forest zones. But
Whittaker's (1975) classification, based on mean annual
precipitation and temperature would put the Kibale forest
formation into the tropical seasonal forest category. The.
forest in the reserve was described by Osmaston (19593) on a
1:200,000 scale vegetation map as tropical high forest which
is equivalent to tropical rain forest as described by
Richards (1952). The entire study was conducted in the
distributional limits of the Parinari forest type (Kingston
1967, Figure 2) near Kanyawara Forest Station. The presence
of mature g; excelsa with its subdominant associates of
canopy tree species such as A; altissima, Q; gorungosanum.
N; buchananii and Q; welwitschii is used as an indicator of
the climax forest between 1370 m and 1525 m altitude
(Kingston 1967).

The reserve was established to provide sustained
production of hardwood timber and production of softwoods
from plantations established in the grasslands (Kingston
1967). The forest is supposed to be exploited on a planned
felling cycle of 70 years. A system of silviculture, near
to clear felling (uncontrolled polycyclic system) is
extensively practiced and pitsawing also allowed. The

recommended logging was supposed to open up the canopy by

approXimately 50% and trees over 1.52 m (5 feet) in girth

24A

Figure 2. Map of Kibale Forest Reserve showing the
approximate position of the study area and relative location
of timber compartments included in the study. The Butanzi,
Kyakato and Nyamusika hills have been planted with exotic
species of Pines and Cypress. The shading indicates the
distributional limits of the Parinari forest type.

25

 

 

   
   

s5.
2 PARINARI
— FOREST
N KANYAW A =— TYPE
.— .1.
:
:4—
K-13
‘Fd
Rwembeita “’
Swamp Kibale Forest Reserve
Part of 0Ngogo
Kvakato
1.1353.
K-15 \
\ -- l Km 4
o ’f..\.j ’ I fi
ka~ 1.15:; .° o In
.“'5.‘-.:~l3§-;7 fit...
K-14 \ ‘\"'.‘?\ it)?"
\ h" 'A‘~_."
n. . ,. . .,
/.\ °\ .kButa'nzI.
Karambi Road

  

Nyamusika“

A“ '1'
\r.’,a

  

Km

26

(at breast height or above buttresses) were to be harvested
(Kingston, 1967, Forest Dept. Records, Fort Portal). A
total of 27 desirable or merchantable hardwood species was
recommended for harvesting. However, the recommendations
were inadequate because the unrecommended tree species and
small size classes were allowed to be cut at the sawmiller's
discretion. In some of the cut compartments, felling was
followed with chemical treatment of undesirable residual
trees (weeds) with arboricides, a process called "refining"
(Kingston 1967).

My study plots were all located in the central part of
the reserve (Central Block) where the altitude varies
between 1350 m and 1550 m (Figure 1 a 2, see general
introduction for details of climate, vegetation, geology,
soils and animals). Mechanized selective felling up to 1969
resulted in about 353 of the northern and central forest
blocks being harvested (Struhsaker 1975). In the central
block, the cut and uncut forest tracts are adjacent,
separated only by a motorable track (Figure 2). In
comparison with the uncut, the cut forest has an open upper
tree canopy, with much sunlight reaching the ground. The
understory is thick and dense forming an impenetrable tangle
of climbers, creepers and herbs (Kasenene 1980, 1984).
Repeated elephant incursions in the cut areas are
contributing to what appears to be vegetation suppressed at
an early recovery stage (Kasenene 1980). In addition, the

elephants destroy woody plants in the seedling/sapling stage

27

either by trampling, browsing, uprooting, barking or
topping. The few large trees, mostly with defects such as
hollow or crooked holes, or of undesirable species, were
left scattered in the cut areas at the time of cutting in
1969. These could serve as important seed sources for
regeneration but in the semi-open environment of felled
areas, they become susceptible to wind and storm damage
(Skorupa and Kasenene 1984).
The Stggy Plots

These were located in cut (K-13, K-14 and K-15) and
uncut (K-30) forest compartments of the reserve (Figure 2).
Each study plot had a previously established grid system of
trails used for access in primate and other research
conducted by the Kibale Forest Project. Compartment 30 (K-
30) includes about 300 ha of relatively undisturbed mature
forest. Prior to 1970, an extremely low density of 3 to 4
large stems/km2 were removed by pitsawyers (Struhsaker 1975,
Skorupa & Kasenene 1984) and this seems to have had very
little impact on the integrity of the forest. The
Compartment has been protected from further pitsawing

activities since 1970 but has been a site of extensive

research on primates and other aspects of forest biology and
ecology. Compartment K-30, therefore, is ideally suited to

serve as control. Compartment 14 (K-14) includes nearly 390
ha of secondary forest that was selectively out between May

and December 1969. Selective harvesting was light,

removing, on the average, 14 ma/ha (saleable volume only)

28

and leaving behind a lightly disturbed forest. Logs of 23
tree species were removed with only 9 species contributing
943 of the total timber harvested (Skorupa, unpublished;
Uganda Forest Dept. Records).

Compartment 15 (K-15) includes about 360 ha of forest
that was heavily selectively out between September 1968 and
April 1969. Total harvest removed, on average, 21 ma/ha
from K-15. The 100 ha area included in the study plots was
extremely heavily disturbed. A total of 18 tree species was
harvested with only 9 of them contributing more than sex of
the harvested volume. Both K-14 and K-15 residual forest
areas were untreated with arboricides after the harvest.
Compartment 13 (K-13), the northern-most part of the study
area (Figure 2), was heavily selectively harvested between
the months of April and August 1968. Total harvest averaged
17 ma/ha (i.e. actual saleable volume on record). Trees of
28 desirable or merchantable species were felled with only 9
species contributing 913 of the harvest volume. The study
plot was located in the part of K-13 that was heavily cut
and treated with arboricides. However, it should be noted
that more timber trees were destroyed through damage during
road construction, hauling of logs and crushed trees than
indicated by saleable volumes recorded by the Uganda Forest
Department at Fort Portal.

The results of timber stock enumerations prior to

forest exploitation show significant local variations in

densities of individual species, but total forest density,

29

basal area, forest canopy cover and species diversity were
relatively constant for all subtypes of Parinari forest
(Kingston 1967).

The major objective of this study was to characterize
the mature and secondary forest communities of the central
part of the reserve with respect to the effect of selective
timber harvesting on composition, structure and regeneration

of the forest.

METHODS

The forest sites selected for intensive vegetation and
regeneration studies included the selectively felled
compartments (K-13, K—14, & K—15) and the adjacent mature
and undisturbed forest (K-30). All these compartments are
located within the Parinari forest type (Kingston 1967),
(Figure 2). Compartment-30 (K-30) which shows very little
evidence of recent human disturbances was used as control in
the study of the effects of selective logging on the
composition,structure and regeneration of the forest in the
reserve. Composition, structure and regenerational
characteristics were compared between the uncut mature
forest site (K-30) and the three selectively felled
secondaryforest sites (K-13, K-14, K-15). Regeneration was
assessed through the analysis of seedling and sapling
abundance and distributions, by size class. In addition to
the major objective of the study, an attempt was made to
obtain an index of the level of forest exploitation
independently of the Forest Department's volume records.

Level of Forest Exploitation

It is often difficult to convert the forest

department's output records, in cubic meters of wood, into

density of trees harvested per forest area. Therefore,

30

31

enumeration of cut tree stumps was undertaken to estimate
the level of forest exploitation in terms of density. To
enumerate the partially or sometimes completely covered tree
stumps in cut forests, a modified form of game census method
(Struhsaker, pers. comm.) was devised and adopted. For game
censuses, the method relies on the target animal advertising
its presence by noise or flight. Then its location is
marked and.distance to the line of travel (or trail)
recorded. The width of the strip sampled is taken as twice
the observed average distance at which the animals were
flushed (Leopold, 1948). .Similarly, for purposes of this
study, all cut tree stumps that could be seen within various
distances astride the trail were counted and their
perpendicular distances from the trail measured. Censusing
of stumps involved slow walk «2-2 km/hr), visual scanning of
either side of the trail for stumps and recording. There
was no straying off the trail except when a stump was
spotted and measurements of distance to the trail were to be
taken. The number of stumps seen or observed were plotted
against their perpendicular distances from the trail. The
perpendicular distance at which the slope of frequency of

occurrence of stumps declined drastically was taken as half

the effective width of the strip sampled. Double this
distance, therefore, gave the estimated width of the strip
along the trail that was sampled for cut tree stumps. The
area of forest sampled was then calculated by multiplying

the estimated width of the strip by the length of the

32

censused trails. Consequently, the level of forest
exploitation in terms of density was calculated.
Forest Vegetation Copposition and Structure

In each forest tract, previously established trails
were followed for tree, pole and sapling enumerations and
sampling of other vegetational features. The direction of
travel along the trails was variable. At every junction,
the direction of movement was randomly determined. Sample
plots were established at every 120 m interval along the
trails. Each sample plot measured 5 m x 20 m. All trees
(3 13 cm dbh), poles (>5 cm and <13 cm dbh) and saplings (>
2 cm and < 5 cm dbh) within 2.5 m on either side of the
trail and in the sample plot were measured. For every tree,
pole or sapling, its species, diameter at breast height
(dbh) (in this study, breast height was fixed at 1.3 m from
the ground or base of the tree) and condition (e.g., dead or
alive, crown broken, wounded, number of wounds or scars)
were recorded. On-site identification of trees depended on
keys and descriptions of the common tree species of Uganda
by Eggeling and Dale (1951), and Hamilton (1981). Trees
were identified based on leaf shape, trunk shape, presence
of latex or other sap and various characteristics of outer
and inner bark including color, odor, thickness, and
texture. Unidentified trees were given their local name,
described and voucher specimens collected, dried and pressed

for subsequent identification and deposition at the Makerere

University Herbarium. Sometimes it was impossible to

33

collect specimens of giant trees. These were mapped, and
marked with aluminum tags bearing identification code names.
Mr. Katende, Curator of Makerere University Herbarium, was
hired and helped in onsite identification of marked trees
and plants in the study plots. The sampling intensity per .
forest compartment studied was based on the species area
curves for saplings, poles and trees.

Estimates of forest canopy cover were made by eye and
checked by measurements of the amount of light penetrating
into the forest using a Weston light meter. An area of
approximately 10 m x 10 m (ground measurements) of the
forest canopy, centered in the sample plot was scanned from
the ground at random points and x canopy cover estimated.
Estimates involved mental comparison of the amount of holes
in the canopy through which one could see the sky and the
area shaded by the canopy. At every canopy cover scanning
point, light readings were taken and very soon after, a
forest gap light reading was taken as well. The light meter
was calibrated in foot candles.

A subplot of 2 m x 5 m was located in the middle of

each 5 m x 20 m plot for forest understory and ground
vegetation sampling. The subplot was subdivided into 4
smaller plots of 1 m x 2.5 m. All living plants rooted

within each of the 4 smaller plots were identified, counted
and recorded. Estimates of percent ground vegetation cover
(gvc) by plants under 1.3 m height were made. This was done

by standing on the edge of the smaller plots, looking down

34

upon the ground vegetation and estimating the amount of
ground vegetation cover against open ground. Again, voucher
specimens of unidentified plants and tree seedlings were
collected, labelled and pressed for future identification
and deposition at the Makerere University Herbarium. A
compass and clinometer were used to determine both the
aspect and general slope of the plots studied.

The collected data were then used to calculate or
estimate density (#/ha) dominance (m2/ha), and importance
values (IV) and ranks of desirable and some important non-
desirable tree species that dominate the canopy. Importance
value (IV) was calculated as based on the sum of relative
frequency, relative density and relative dominance. The
widely used Shannon-Wiener and Simpson diversity indices
(Krebs 1972, Brower and Ear 1984) were calculated for trees.

poles, and sapling classes.

RESULTS

Level of Forest Exploitation

In compartment 13, (K-13) the length of the trail
censused for out tree stumps was 4.3 km. The estimated mean
strip width from trail to stumps (by frequency distribution
of stumps with respect to perpendicular distance from the
trail) was approximately 5.1 m. The sampled strip therefore
was estimated to be twice this mean strip width i.e. 2 x 5.
1 m. The area sampled in K-13 was then.estimated to be 4300
x 10.2 m2 or 4.39 ha. With a total of 43 cut tree stumps
counted within the estimated sample strip width, the
estimate of density of trees removed from K-13 was 43 9 4.39
or 9.8 trees/ha. Similarly, in K-14, 28 tree stumps were
counted in 6.94 km. of 13.5 m wide sample strip. Thus an
area of 9.37 ha was estimated resulting in a level of
harvest of 3.0 trees/ha. In K-15, the length of the census
trail was 6.53 km and the estimated width of the strip
10.1/m. Thus the forest area sampled for cut tree stumps

was 6.9 ha. With 56 tree stumps counted within the sample
strip, an index of harvest of 8.50 trees/ha was estimated.
In K-30, the length of the census strip and width were 7.18
km and 29.8 m, respectively. With only 3 cut tree stumps

observed, the estimated index of harvest was very low,

35

36

approximately 0.14 trees/ha. The majority of tree stumps
observed were in advanced stages of rotting and
decomposition. Thus, few stumps could be identified to
species level with certainty.

Table 1, shows estimates of the amount of forest
harvest in the four compartments. Estimates of density
(#lha) by conversion from volume data (Skorupa unpublished)
or by direct enumeration of cut tree stumps (this study) are
also included. Our estimates for density of forest trees
harvested varied in values but were not different in
depicting the level of forest exploitation for the forest
compartment studied (U-10, N1 - N2 - 4, P > 0.10, one tailed
test), (Sokal and Rohlf 1981). In both cases, the trend was
similar. The heavily cut forest areas showed higher indices
of forest exploitation than lighly cut or uncut forest

tracts (Table 1).

Forest Vegetation Composition and Structure
Tree speciesggiversity

 

Two common measures of species diversity are the
Simpsons index (Ds) and the Shannon-Wiener-Index (H‘) (Krebs
1972, Smith 1974 and Brower and Zar, 1984).

In this study, the Simpsons Index was defined

as D8 = 1 - pi (hi-1)
N(N-1)
In general form, the Shannon-Wiener formula is written

as: H' - -pi logz pi. However, in the calculation of

37

one maahoxmv amen on» a“ Em\m00Au MIN mo nopmo one

.Aewa~ anaconda
so po>0809 mushzmmewmo

 

sense mess es.o e.m o.m m.e
...Hn:o::. ooaooxm exmo.oI~o.os e.e H.m ~.e

ewes anaconda a nosnoxm asxeoeuu is es es

mammoo
Qaaum hm
.on\ev haemaon

035~o> loam
:oflmmo>:oo ha
.ee\e. seemeoo

om\ a
083H0> AmaOE

 

emu can eam mm:
newamlhomcm no oousom mbzmthm<mzoo
summon

UOHfll—flvmfl $903.03.“.Hdnm

.0003 #30 m0 mHGSOEm popoooou
ooadowe moooaaumo -< .mumoauhmnaoo use h~o>auomqom mom Ad:\ev
hvumcep Usualnvmo on» one Ad:\ m8. oxmvpao woman» Hmnomoafloo mo marao> .H manmh

38

.mOMOQQm coma caooummop acomouaom auexoeha ma newsman

 

 

.mm.cvum.c .uo.ovmm.o .ms.c.~m.o .um.cvmw.o moons

Aaw.cvmm.o .mm.c.ww.o Ame.cvew.c .wp.o.um.c moon» was modem

Amm.cvca.° .om.ovem.c .c.~.mm.c .om.cvmm.o mmsflqaem
.m. .seeqeoeeeaom

A~.evo.m Ac.mvn.m Am.mvm.m .a.n.m.m mocha

Ap.mvn.u Ac.mvw.n Aw.nvb.m .>.m.>.m moon» pas moaom

.o.¢.«.e ._.e.e.m .«.e.e.m .~.e.e.m nauseous
.xel I

.@.nv~.m .N.Nve.e Ac.mva.e .v.n.m.e mosh

Am.n.p.e .~.nvm.e Ac.nvm.v .m.~.o.e moves 6:6 meaom

.v.n.s.e .@.m.b.v .N.evw.e .e.m.~.m mmcddaom
.=. hswmuo>wn ao=0w3Icomco£w

.oa.c.wa.o .ap.cvmm.c .Nw.o.ea.o .wm.c.em.o momma

.mm.c.vo.o .mw.cvwm.o Amm.c.mm.c Amm.c.~a.c mean» a modem

Apm.cvma.c Aaw.o.mm.c .nm.ovmm.c Amw.c.ma.o mmcfifinom
Asa. haemh0>wn m.:omalwm

Ae.e.o.e .m.evm.o .N.m .m.m .o.m.m.> moons

Aa.n.w.n .¢.nvm.e .e.> so.» .m.wv~.o~ moons pad modem

.m.o.m.e .a.wvm.> .o.c~.c.- .~.m~.~.s~ mmawqaem
moaoonm\m~¢=pw>wpmfl no a

Aegvae Auflvmm .edvem .muvwn moons

Anuvam .Ngvme Aeuvfim Anuvnu moon» a snack

.euvsn Augvme Amdvmm Amdvuo mmcfiqaem
n mmu:a0«m mowomnm
.meucv Ameucv Aweumv .mvunv haumA0>aa no measure:

m~I& mnIM e—Iu chm

 

mfizm29m<mtoo EmmmOm

 

.l cm x I m £060 .muodn

mo amass: on» we :

uo>u0eoa use «0 mucosa ueoaou Aomlmv uses: use .mqu .quM .m~I&. v50
hqo>waooqom cw Asap lo «u A. moons omega use Asap Io may me as. moons
Adele use moqoa ..£np lo ov IONA. mmawqaem you m05q6> xopsw huwmuo>mn .N oases

39

indices (Table 2) the modified form (Smith 1974) used was

defined,as:
H - 3.322 [logic N - (ni log 10 ni)/N]
The maximum species diversity index (H max.) was also
calculated using the relationshp:
H max. - logz S or 3.322 loglo 3.
Estimation of tree species evenness or Equitability (E)
involved the relationshp:

E = H
H max.

In these indices,

Ds - Simpson's species diversity index
H - the Shannon-Wiener index of species diversity
8 u the number of species observed

3.322 a a conversion factor, loglo to logz

N = total number of individuals of all the species

ni number of individuals of a species

Table 2, exhibits six common measures of species
diversity. The simplest measure of diversity was the number
of species or species richness encountered in the four
forest compartments studied. Clearly, the uncut.forest

shows the highest tree species richness in the sapling and

pole size classes. The lightly cut forest (K-14) was ranked
second to uncut and the heavily selectively cut forests
shared the third and fourth position. However, the cut
forest tracts tended to have more species in the tree size

classes than uncut forest.

40

Uncut forest (K-30) also had the highest number of
individuals per species, for all size classes, than any of
the selectively cut forest tracts. The lightly selectively
cut forest ranked second in this respect and the heavily
selectively cut forest (K-15 and K-13) ranked in the third
and fourth positions, respectively. With respect to
Shannon-Wiener diversity index, the uncut forest had the
highest diversity of tree saplings. The lightly cut forest,
ranked second and heavily cut areas shared the third
position. This was also true for the maximum possible
diversity estimates for the four forest tracts studied.
High indices of equitability for saplings, poles and trees
were also exhibited by both the cut and uncut forest tracts
(Table 2). The desirable or canopy tree species followed
the trend of species diversity indicated above. The
unlogged forest showed the highest tree species richness,
highest numbers of individuals per species and high
equitability. As expected of lightly cut forest, K-14 was
placed second in all diversity measures and the heavily cut
(K-15) in the third and heavily cut and treated forest in
the fourth position.

Another measure of species diversity considered in this
study was the species-area curves for the four forest
compartments (Fig. 3 and 4). Species-area curves were
plotted to assure that the majority of tree species within
the study compartments had been observed, but also provided

important comparative information on the species richness of

 

40A

Figure 3. Cumulative increase in the number of species of
saplings (>2<5 cm dbh) poles and small trees (>5<13 cm dbh)
and large trees (213 cm dbh) with corresponding increase in

forest area sampled in lightly cut (K-14) and uncut (K-30)
forest tracts of the reserve.

41

cos + N5. 2. om._n_s_<m <mm< m>F<._:s_=o

pm me be me on mu ow me be m o

09v. 1|.1I< .v _..v_<i<.. .4 A 3533.. \o\ .. \
omi XIXIx {3. XIXIIxumooz s mo_on_ 0\ x1 . .
09x elele £129-90 ”moot. \o\ . x. x ..
>m¥ \O\\OIIO \ a.
\o- \
\\Q \X\\\* ¢~
\OIIOIIIOI\OIIO||Q \x\\ s
\\ \ \
p .
\O\\Q \X\\ <\\
\R \xi. . \
\°I\° \ o JIXII* \\(
0:0-on xx--. - .. .x I. .48
x... .. Lois... . I,

o.
0')

o:
1

6
ID

0

O
,—

O
N

 

saloads =Io # aAIIV'Inwno

0*
(O

41A

Figure 4. Cumulative increase in the number of species of
saplings (>2 cm <5 cm dbh) poles and small trees (>5 cm <13
cm dbh) and large trees (313 cm dbh) with corresponding
increase in forest area sampled in heavily selectively cut
and treated forest (K-13) and the heavily cut but untreated
forest (K-15) of the reserve.

42

00.. news. 2. <mm< Mahala—2:0

 

m? 0.? mm 0.” MN ON Mr O. _. m. 0
m7: 43.454 6 TV. Ilka «mm—anm N.
a 3. T IXIIX 6 TV. Illx ”moo: a «Eon. \ .
07: 90.. IO 6 TV. TI ”moo...- \ \<\

>mv— V” \ a
\Ix\\%\\ «\ .
*\x\x\®w ... \
\\X\\X\\X\\OIU V§ .

- .wawhfi. 1..

\\x||*.\ \OIIQ\

o
1-

o
co
saloads # aAIIV'Inwno

6
<1-

 

O

O
N

o
. In

.1 i‘

43

trees, poles and saplings in relation to increasing size of
forest areas sampled.

Figures 3 and 4 show species area curves for three size
classes for each forest compartment studied. Curves were
plotted for trees (2 13 cm dbh), poles (>5 cm <13 cm dbh),
and tree saplings (>2 cm <5 cm dbh) It was thought that
segregation of size classes would give a better estimate of
the level of sampling of the study plots. In the lightly
cut and uncut forest tracts (K-14 and K-30 respectively) the
curves for trees and poles approached asymptotes in just
about 0.45 to 0.50 ha of complete sampling, and within 0.25
to 0.30 ha for the sapling size classes (Fig. 3). However,
for the heavily selectively felled forest areas (K-13 and K-
15), only the sapling size classes plateaued after 0.25 to
0.30 ha of sampling. The curves for the tree and poles size
classes continued to rise even after 0.50 ha of sampling
(Fig. 4).

Except for the tree size classes, the species-area
curves for poles and sapling size classes for uncut and

lightly cut forest tracts were similar (Fig. 3). This
indicated similarity in the number of tree species per area

of forest sampled. The species-area curves for heavily cut

forest tracts (Fig. 4) were different for all tree size
classes compared. Using K-30 species-area curves as a
standard, the curves for K-13 and K-15 fall under those of
K-30 for the respective size classes (Figs. 3 and 4). The

curves also show that there were more tree species in the

44

sapling size classes per area sampled in uncut and lightly
cut than in heavily cut forest tracts. These results
suggest negative effects of logging on tree species
diversity.

Measures of species richness for the four forest
compartments were further analyzed and compared using
oneway ANOVA and Duncan's Multiple Range Test (Appendix
Table 1). With the exception of K-13 vs. K-15, and K14 vs.
K30 plot pairs, all other pairs compared had significantly
different species richness of trees (2 13 cm dbh) per sample
plot (F = 7.83, P 5 0.0001). As expected, the plots in
uncut forest had the highest mean number of tree species per
sample plot, the lightly cut forest had the second highest
and the heavily cut forests the lowest (third and fourth
position, Fig. 5). For pole tree species richness, only the
K-13 vs. K-15 plot pair showed similarities; all other pair
combinations differed (F = 22.638, P < 0.0001). Again,
plots in uncut forest had higher pole species richness than
those in cut forests. But within the cut forest category,
the plots in lightly cut forest had higher pole species
richness than plots in heavily cut forest areas. Thus the
order of importance for tree pole species richness was K-30>
K-14> K-15> K-13. Important results pertaining to forest
regeneration were exhibited by tree sapling classes
(Appendix table 1, Fig. 5). There were highly significant.
differences for sapling species richness among all plot pair

combinations of the four forest compartments (F = 35.500 P<<

44A

Figure 5. Mean species richness for sample plots in uncut
(K-30), lightly cut (K-14), heavily cut (K-15), and heavily
cut and treated (K-13) forest tracts of the reserve.

45

whzm S.._.m<n_ .200 PwmmOu.

m 3. e 3.

o

‘1’
a:

 

 

\ a \

 

 

 

 

 

 

 

 

 

 

 

fl

2:00am Easemen— V\\\k

009399-00 m0_0$0m =< I

fie :8 9 A 89F u n
can So 3 v :8 m A86“. n a
fie :8 n v :3 a A 35.3.... .2 n .

>mx

a

\

NV

 

 

 

 

 

 

imJ

 

 

IN...

 

 

stoId av“ 00 L/sseuqogu segoeds ueaw

46

0.0001).. The uncut forest had the highest mean number of
sapling species per sample plot, the lightly cut forest was
ranked second and the heavily cut forests followed in the
third and fourth positions. The lightly cut forest (K—14)
had 72.98, heavily cut forest (K-15) had 51.18 and the
heavily cut and treated forest (K-13) only 29.2x of the
sapling species richness exhibited in uncut forest (K-30).
It was also noted with concern that within the heavily
selectively cut and treated forest (K-13), the sapling
species richness was very low compared to the species
richness of poles and trees (Fig. 5). This trend was not
exhibited by any of the other forest compartments studied.
The forest management practice of treatment of residual
forest after felling with arboricides may account for the
observed depression in the sapling species richness in
compartment-13.

For the desirable trees, there were more species per
plot in uncut than any of the cut forest areas (F - 7.062,
P<0.0001). The desirable pole species showed significant
difference only in three pairs of plots (F = 6.876,
P<0.0002). The desirable pole species richness was similar
for the plot pairs of uncut and lightly cut forest and for
the two heavily cut forest tracts. However, the similarity
in the pole species richness for the plots in K-13 and K-30
was difficult to explain. The desirable sapling species

richness were significantly different for three obvious plot

pairs, K-13 and K-14, K-13 and K-30, and K-15 and K-30 (F I

47

11.07, P<0.0001). But sample plots of uncut and lightly cut
forest and those of heavily cut forests (K-13 & K-15) had
similar sapling species richness. The observed differences
in tree, pole and, more importantly, sapling species
richness suggest that selective timber harvesting could have
been the main cause or influence.
Tpee gensity and basal area

Estimation of the density of trees and juvenile size
classes for the selectively cut and uncut forest tracts of
the reserve was thought to give an insight into the degree
of forest exploitation, disturbance, and an indication of
the success of forest regeneration 20 years later. Table 3,
and Appendix table 1 show mean density estimates for tree,
poles, and sapling classes for the study compartments. The
densities within the three size classes were compared among
the compartments using ANOVA and Duncan's Multiple Range
Test (Appendix Table 1). As expected, there was significant
difference in the density of trees between uncut, lightly
cut and heavily selectively cut forest (F a 9.750,
P<0.0001). As depicted in Table 3, uncut forest (K-30)
exhibited the highest tree density with lightly cut forest

in the second position and K-15 and K-13 in the third and

fourth positions, respectively. All logged study plots,
except for K-14, differed significantly from the control
plot (K-30) in forest vegetation composition (Appendix table
1). Among the logged study plots, only the K15-K13 pair

showed similarity in tree density. A similar trend was

1'
chkm

48

dam! one no sewamw>op pumpsdum u o

 

 

 

 

Au.ee. e.oe .o.me. e.oe Ae.ee. e.oe Ame. e.oos ocean-co ~H<
is.ee. e.em .m.ee. e.ee Aa.ee. e.ee .e.ee. o.ee moons
A¢.~. o.~ Ao.~. o.~ .e.~. e.e .e.es s.e modem
he.os e.e .e.oe m.e .e.ec m.~ Ao.o. e.~ nauseous
AGAVNI‘ d0h< Admdm
.e.zc e.“ .e.s. «.3 .o.s. m.~ .m.~. e.~ moons menaeeuoo
.e.e. e.“ .~.e. N.“ .e.~s e.~ .m.~. e.n meson menaoeuoo
.o.«. e.» .o.«. e.e .e.e. e.e .e.e. m.o assesses managemoo
.«.~. «.4 .o.~. e.e .«.~. o.e .s.~. e.o moons
Ae.~. “.4 .~.e. e.e .o.e. «.o Ae.e. e.m~ modem
.o.e. ~.e Ao.e~. e.ee .e.o~v e.mu .~.e~. m.o~ nauseous
co” m anxse mmwmmmm
.eeun. issue. issue. .eeues mooeoaeeeo

«mus emum «HIM onus

 

mhzmzhm<mzoo BmMGOh

 

.E on x a m £060 mvoaa mo 9038:: one ma 2

.Asne :0 Asa.

moon» owned pad Aznp so mfiv so mAv moon» Hamam pad moaon ..436 80 mv

Io NAV mmswqamm popsdosw masoaousmmoz

.o>aom0m 0:» :w .mfiImv amonom

papacy» use #30 haw>oos one ADHINV 9:0 hafi>mon .AeHIM. #30 haanmwd
.AchM. amouom manual #50:: Ga unwanlmm newvdeomo> umomom mo madamom .m canoe

49

exhibited for the poles, where even the lightly cut and
uncut forest plots differed in their pole density (F =
39.143, P<0.0001). There was a highly significant
difference in the density of saplings among all plot pair
combinations for the four forest compartments (F - 34.596,
P<<0.0001). The order of seedling density was K30) K14>
K15> K13. The heavily cut forests (K-13 and K-15) showed
similar tree densities for thedesirable species but all
other pair combinations were significantly different (F =
8.591, P<0.0001). The existence of a difference in density
of desirable trees between K-14 and K-30 study plots could
have resulted from the effects of harvesting which was
selective with respect to size and prime tree species. The
plots in uncut and lightly cut forest had higher pole
density than those in cut forests (F - 6.832, P<0.0001) and
within the cut forests, K-i4 had higher densities. Again,
the most striking density results were those of sapling
classes where the heavily selectively cut forest plots
showed similar but very low densities and the lightly cut
and uncut forest plots very high densities (F = 12.993,
P<0.0001).

‘There was considerable variation in basal area
estimates for the four forest compartments and for the
trees, poles and sapling classes (F = 9.791, 42,033 and
42.425 respectively, all P<0.0001). The contribution by
saplings and poles to basal area estimates was very small

(table 3). By virtue of high densities of big trees, K-30

50

plots had highest basal area estimates for all size classes
considered. However, tree basal area for heavily
selectively cut forest (K-15) was slightly higher than for
lightly cut forest but the two were not significantly
different (P>0.05).

In K-30, the desirable timber trees and poles combined
and saplings contributed only 27.4% and 33.48 respectively,
of the total density. The results for K-14 are comparable
to those for K-30. The desirable timber trees and poles
contributed 32.9% of the total density while saplings
contributed only 32.5x of the total sapling density. In the
heavily selectively cut forest tracts, K-15 and K-13, the
desirable trees and poles accounted for 24.3% and 38.85,
respectively, of the total tree and pole densities. The
desirable saplings also showed relatively high percentage
contributions of 34.3% (K-15) and 47.93 (K-13) of the total
tree sapling density. The high estimate values in K-13
might have been influenced by small numbers which often tend
to inflate percentages. Although it was almost 19 years
since the forest was felled, forest composition with respect
to tree density and basal area showed great variations
between the compartments (Table 4). Great differences in
percent reduction of densities of trees, poles, and saplings
was observed for the differently managed forest compartments
(Table 4). With the exception of only one positive increase
for the density of poles of desirable species in lightly

logged forest, the general trend of density of trees, poles

.moqon mesa m0 haflmsou on» sfl
omdososw vm0>nmsIumon o>wawmon a me: cause »

 

51

 

 

 

 

e.~e «.em e.ee o.oo~ socenaoo HH<
e.He o.em ”.me e.oe moose
m.se e.oe «.em ~.e ooaoo
o.em o.oe o.m~ o.~ assesses
~msVNE. sou< Amman
m.ee m.~m >.c~ m.N moose manoswmoa
e.em o.mm ue.mus e.~ noses managemoo
m.ee ”.mm e.s~ m.e muessoem managemoo
e.~m e.em m.e m.e moose
e.ee e.me «.me e.ms modem
m.ee m.ee ~.o~ m.o~ assesses
oneooomm a oe\oz cos ioe\~s essence
Ameuos scenes issues _ issues muouoaoeeo
«mus AHIs esus I om-s
mezmzem<ozoo

 

.8 ON x i m some mvoan

mo monflas use ma 2 .dOhusoo mm uvm: me: use sacs: mm: chm

.0>s0eos as» ho masolusdmioo vmoaom 9:0 hao>waomaom sw mmswanmm
uso madam .momsu mo some demon use huwmsmursw sofluosuos psoosom .e oases

52

and saplings following logging was negative. The negative
trend in density was more pronounced in heavily out than
lightly cut forest. This was also true for the basal area
of trees, poles and saplings (Table 4). It is suggested
that the magnitude of the differences in the density or
basal area for the different size classes and compartments
were a reflection of the intensity of exploitation and
forest disturbance.

In general terms, therefore, the plots in uncut forest
(K-30) had the highest species richness and diversity and
the highest density and basal area for all size classes
considered. The lightly cut forest (K-14) ranked second
while the heavily selectively cut forest (K-15) and heavily
selectively cut and treated forest (K-13) ranked third and
fourth respectively. However, the non-desirable timber
species accounted for the majority of species richness and
diversity (Fig. 5), and a greater percentage of density and
basal area (Table 3), for all size classes in all forest
tracts studied than did the desirable or canopy tree
species. Therefore, it appears that the great differences
between the study plots in vegetation composition reflect,
to a greater extent, the intensities of logging disturbance
and in a minor way the inherent natural variations.l The
differences in tree, pole and sapling species richness for
all trees in general and desirable tree species could be

explained based on the effects of selective logging with

respect to size and species.

53

Impgrtance Values of tree species

Like most tropical forests, the forest areas studied
had high species diversity values reflecting the absence of
any tendency for individual tree species to dominate the
community. However, the best measure for the importance of
a species in a community should incorporate measures of
basal area, extent of spatial distribution and population
size (Curtis and McIntosh, 1951). Consequently the
importance values (IV.) recorded in table 5, were obtained
by summing values of relative dominance, relative frequency
and relative density. of tree species. The importance
values of desirable timber species (also dominant canopy
tree species) and a few important non-desirable species were
computed for the four tracts studied (Table 5).

In uncut and selectively cut forest tracts some non-
desirable canopy and understory tree species ranked within
the top ten in importance (Table 5). A non-desirable but
canopy tree species, Diospyros abyssinica, was ranked first
in uncut mature forest and was in second, third and seventh
rank positions in selectively cut forests of K-15, K-14 and
K-13, respectively. Similarly, Celtis durandii, and two

other understory species, Teclea nobilis and Uvariopsis
congensis ranked in the top ten species in importance for

compartments K-30, K-14 and K-15 but were very low in
importance in compartment-13. Each forest compartment
studied had a different tree species at,the top of the list

of importance values. In K-14, Markhamia plagygalyx was

 

54

.msem amazod an use seesaw: moueOAusw Au. osoe
.msaamo>hes sw uoHHom uos one: mewooam omneswmouIsOZe

 

 

 

 

 

mm m 0 6 0.0 ~.N~ ~.NN 0.6g a mwmsousoo .D
an 6g 6g cm 0.0 0.» 0.0 m.e euwusOANm .6
«a 0 N e 0.6 «.mn 0.0a 6.0a a mwawnos .9
es «a as as o.e o.o 0.“ s.o muoeeeoooeu .m
u e~ 0 ca e.ge 0.0 0.eN 6.0a wheammosom .0
0d ~N cm as «.6 0.0 6.N 0.0 eseOMHMe .m
«N 0 0H Nd ~.H o.n~ «.m m.- endeoxe .m
N a 0 0 ~.00 0.~0. n.6a 6.6a «wsoeuflzqoz .o
0 0" ca 0g m.NN ~.m 0.0“ 6.e easesesoan .z
HN 0 0a 0 0.H 0.6 N.e e.m~ flexesmuen .2
0 0 a 0 0.0g H.0~ 0.0a 0.6“ NNHeONueHm .2
ma «N «a N 0.N 0.0 e.gfi 0.e« aasoasosshzm .4
fig "a 6 m 0.0 «.cg ~.0N 0.eN ewaomwpea .0
ed 0“ as mm 0.6 0.e 0.0 ¢.~ mwmseaouse .m
m a“ an «N 0.0“ a.o~ N.~ ~.~ eHHNsmosoeI .m
6 N a n 0.0N 0.He N.0N «.mw a eOAstmNse .a
e «g «a ma 0.N 0.0 0.0 0.e wwseaawl .0
0a 0“ «N 0a 6.0 —.m 0.~ 0.6 assemousssou .0
a m 0 0 «.6N 0.0m 0.0m a.- a Mausessu .0
ON e a «a m.~ ~.~N c.¢~ 0.0a ese0wsme .0
0 6 fig 6a N.HN N.6~ 0.0 6.0 eveusnwss .m
6~ mm 0H nu 0.0 c.~ 0.6 0.0g e«me0«xov .<
c.- on 0g Ha 0.6 ~.~ «.6 ~.e~ eawmewpde .<

mfiIm 0~IH e~I& chN «HIM 0~IM efiIa omIM
execs A>Hs moeooom
00seumonflH m03~e> moseusoaaH nose

.o>homon 0s» 00 muoesa amenom AnaIM .0HIN .eaImv v30 hqe>wuooaoe
use .cMIM. #50:: now mowoomm sense» oaseswmouIsos elem use

Ahaoseo season cede. odaeuwmou mom eases use A>H0 mosde> moseusoaaH .0 eases

55

most important, glgg welwitschii and Stropbosig schefflgri
ranked in top places of importance for K-15 and K-13,
respectively. No one tree species had similar importance
value indices in each of the four forest compartments
studied. Consequently, there was no significant correlation
between importance values and ranks of tree species when
comparing K-30 and K-15, K-30 and K-13, K-14 and K-13, and
K-15 and K-13. However, there was significant positive
correlation between importance values and ranks of tree
species when comparing K-30 and K-14 (r = 0.670, P<0.01) and
K-14 and K-15 (r = 0.628, P<0.01).

The major forest canopy tree species which are also
desirable timber species, such as Parinari excelsa, Mimusops
bagshawei, govoa swynnerton_i, Newtonia buchananii,
Aningerig altissima, and Chrysophyllpp gorungosanum were
very poorly distributed and had very low densities for all
size classes, especially in selectively cut forest tracts.
Consequently, they had very low importance value indices and
ranked very low in importance. The relative frequency,
relative density and relative dominance, and consequently
importance values of desirable species or canopy tree

species, were very low in cut forest areas compared to uncut

but non-desirable tree species showed higher values for all
four forest tracts. However, some non-desirable understory
species such as Uvariopsis congensis and Teclea nobilis had

high importance values and ranks in uncut and lightly cut

56

forest tracts, low values and ranks in K-15 and were almost
absent in K-13 plots.
Other measures of forest_§tggcture and composition

An analysis of variance and Duncan's multiple range
test revealed that all logged study plots (Table 6) differed
significantly from the control plot (K-30) in forest canopy
cover (F - 52.30 , P<0.0001). All pairwise comparisons
between uncut, lightly cut and heavily cut study plots
showed that forest canopy cover was dissimilar (Table 6)
However, study plots in K-30 (control plot) had the highest
canopy cover averages, with lightly cut forest (K-14) ranked
second and heavily selectively cut forest K-15 and K-13 in
third and fourth positions, respectively. Relative to
canopy cover for the control plot (K-30), cover was 19.8%
less in K-14, and 47.9% and 61.7% less in K-15 and K-13
respectively. There was also a strongly inverse
relationship between forest canopy cover and the amount of
light reaching inside the forest for all study plots.

High amounts of sunlight reach inside the forest in
heavily cut forest areas (Table 6). An analysis of variance
and Duncan's multiple range test revealed that all logged
study plots, except K-14, differ significantly from the
control plot (K-30) in the amount of sunlight penetrating
into the forest (F =.42.895 P<0.0001). Percent canopy cover
and % gap light inside the forest were significantly
negatively correlated at P<0.05 for K-14 and K-30 only (r =

-0.359 and r = -0.306 respectively, Pearson's correlation).

57

sees es» mo scauea>ou useuseum n e

 

 

Av.o~v 0.0” A6.wv 0.0M A~.o~v 6.0a A6.6v 0.0N muodm mo omodm x
A0>0v

A0.NNV v.a6 A¢.0Nv 0.00 AN.0NV 0.0e A0.0~v ~.¢N h0>00 sofiaeuouo> ussosc X

AH.ONV N.00 Aa.m~v 0.6m A0.0. 0.0 A¢.mv 0.6 emvhom ouwmsfl Hausa new R

A0.VHV ¢.NN A0.0~v 0.00 A0.¢Hv 0.00 eA¢.Omv 0.00 s0>00 knoseo umosom R

”HI! 0HIM «film owl: mmmsmz<m<m

 

BZMZBM<QZOO Banach

 

.o>semes 0s» 00 muoesu vmesom 030s:
use 050 sw so>oo scauevome> ussosm vmesom use .ummsom as» ouwmsw
unwed new usoosem .se>oo knoseo umosom mo madsmos on» no muesssm .0 wanes

58

All logged study plots differed significantly from the
control plot in the amount of ground vegetation cover (gvc)
(F = 52.142, P<0.0001). All pairwise comparisons among
uncut, lightly cut and heavily cut study plots were
significantly different. The control plot (K-30) had the
lowest of all gvc estimates, K-14 the second lowest and the
heavily cut forest areas the highest gvc estimates (Table
6). Pearson's correlation between % gvc and amount of
sunlight reaching into the forest were highly significantly
positive for all study plots (all P<0.0001) except K-14.
However, the relationship between % gvc and the amount of
forest canopy cover were strongly negative for all study
plots (all P<0.0001).

The % lepe of study plots in cut forest tracts
differed significantly from those in uncut forest (F =
21.80, P<0.0001. ANOVA and Duncan's test). The average %
slope for K-30 plots was, 28.0% and 14.74%, 15.8% and 16.28%
for K-14, K-15 and K-13 plots respectively. However, there
was no significant difference in the slope of study plots
among the cut forest tracts, indicating similarity in
terrain. The influence of topograhpic relief on observed
differences in forest composition parameters may be small,
at least for the selectively cut forest tracts.

Intraplot vegetation parameter cggparisong

Within uncut mature forest (K-30) there was no

significant correlation between large tree and the juvenile

tree species richness or density (Table 7a). Further, there

59

 

 

 

 

z anode so oooam a
N NN mm A a0>00 .me> ussosc a
N M N season swasmAHQec N schemes!
I sesao
n se>oo haoseo umosom u
xx xx as H muessoam canouemoo
NN NN NN = moose oasesweea
es;
NN NN 0 mwsfldaem haamsen
NN N NN m moose
NN m emswasem oaseswmen
NN a moose oaseswmon
omse>sw hauseOAMNsmwm u M 0 mmsaaaem
~0.ovvm ae mmossOMH
useomesmam muses: u NN m emsaaueem 0000090
0c.cvm I
we useonwsmwm u N < moons
: A N a H z 0 m m a 0 m 4 mommeao mesa seuosesem

 

.0>s0m0s on» s« .omIN. weesom usoss
mo emcee-esem scaveueme> seesaw seozpon msoMueqmssoo m.somse0m

.ae ashes

60

was no correlation between tree species richness or density
and other measures of forest composition and structure (see
Table 7a). The relationship between seedling species
richness and the amount of light reaching inside the forest
was significantly negative (r a - 0.288, P<0.045).
Similarly, a highly significant negative correlation between
% ground vegetation cover (gvc) and tree seedling (but not
sapling) species richness was observed (r = -0.517 P<0.001,
Fig. 6). However, the amount of sunlight reaching inside
the forest was negatively related to forest canopy cover but
positive in relation to the amount of gvc inside the forest.
Therefore, this suggests that excessive destruction of the
forest canopy could be detrimental to seedling and sapling
regeneration.

For the lightly cut forest tract (K-14), the high tree
species richness and density was positively related to
forest canopy cover, as was tree seedling and sapling
species richness and density (Table 7b). Here, the amount
of light reaching inside the forest was negatively
correlated with both forest canopy cover, species diversity
and density of desirable tree species. Again, the strongly
negative relationship between % gvc and the seedling and
sapling species richness and density was realized (Table
7b, Fig. 6, 7).

For the heavily selectively cut forest tract (K-15),
the big tree species richness and density were positively

correlated with the seedling and sapling species richness.

60A

Figure 6. Correlations between seedling (<2 cm dbh) and
sapling (>2 cm <5 cm dbh) species richness and the level of
forest ground vegetation cover of the study plots in the
reserve.

Seedling Species Ftichness/100m2 Plot

Sapling Species Richness/100m2 plot.

 

61

\.\ Seedlings < 2 cm dbh
.>~..\ '\ 1 = K-13 3 = K-15
"are“ .2 2=K-14 4=K-30
“'3‘ \.

 

 

.—A
N

O 20 4O 60 80 1 00

Percent Forest Ground Vegetation Cover

L . Saplings>2 cm <5cm dbh

<\\\ \. 2

\\\ \ 1=K-13 3=K-15
‘ ' 2=K- = -
\\3\\. 14 4 K30

\\\\\.
\

    

 

O 20 4O 6O 80 1 00

Percent Forest Ground Vegetation Cover

62

 

 

 

 

z muons so ooosm x
mm m m m m. m we. ._ .28 .2; use... a
NN NN NN N veesom sw vsmwnsec a messmeOE
II II 00sec
NN NN N NN N H se>00 hsoseo aeosom a
NN NN NN NN H mmswaaem oasesaeea
NN NN NN NN NN N moose oasesamoa
es).
NN NN NN 0 mmswasem hawmsen
NN NN NN NN a moose
NN NN m mmswasem caneswmoa
NN a moose oaseswmen
emse>sw havseONmasmwm n .M NN 0 mmswasem
No.0vvm we mmess0ws
aseONMNsmwm Eases: u NN m emswauoom mewooam
0c.ovm
8e aseOMMMsUam n N < moose
z A N 0 H N c m m a 0 m < mommeqo mesa seamlesem

 

.e>semes esv sw Ae~INv season 9:0 mauswwd
00 eseaOEeAea scavepomo> season seezves escapedossoo m.sOmseem

.ne ashes

62A

Figure 7. Correlation between sapling density (#.ha 100) and
percent forest ground vegetation cover of the study plots in the
reserve.

DENSITY OF SAPLINGS (no/ha -:- 100)

63

\ Saplings > 2 cm < 5 cm dbh

40_ \ 1=K-13 3=K-15
\. 2=K-14 4=K-30

 

 

0L , . \.

O 20 4O 6O 80 1 00

PERCENT FOREST GROUND VEGETATION COVER

64

density of saplings, and forest canopy cover but negatively
related to the amount of light reaching inside the forest
and ground vegetation cover. However, the forest canopy
cover was positively correlated with the seedling and
sapling species richness and density but negatively related
to ground vegetation cover. The amount of ground vegetation
cover in the forest understory was postively correlated with
the intensity of light penetrating into the forest.

However, both the ground vegetation cover and the amount of
light penetrating into the forest were strongly negatively
correlated to seedling and sapling species richness and
density (Table 7c Figs. 6 & 7).

The heavily selectively cut and treated forest tract
(K-13) showed vegetation correlation patterns similar to K-
15. The species richness of big trees was positively
correlated, though weakly, with sapling species richness and
forest canopy cover, but negatively with the amount of gap
light reaching inside the forest (Table 7d). Again, the
juvenile tree classes were positively correlated with the
level of forest canopy cover. The amount of forest ground
vegetation cover was again strongly negatively correlated
with juvenile tree species richness and density and the
level of forest cover but positively with the amount of

sunlight penetrating into the forest.

65

 

 

 

 

xx xx x a noose so oooAm x
xx .x x m m m m m m e .38 .5 2.55 a
NN N N N umesom sx usmxaseo N mossmeoa
II I I 00:00
N NN N NN N h so>00 hsoseo pmosom R
xx xx xx xx x xx x assesses canaunaoo
N NN N NN N moose expesxmon
es:
NN NN NN NN 0 mmsxaaem huxmseo
NN N NN NN m @0099
xx x xx x muesxoem caneeemoo
N NN a moose canesxmoa
oeso>sx hueseOANAsmxm u N NN N 0 emsxdnem
Ac.cvvm as l emoss0xs
annexeseasn sagas: u xx xx x mmenxooom mosooom
00.0vm
we useOANAsmxm u N < mess?
2 A N a H N 0 m N a 0 m < eemmexo mosh seamlesem

 

.0>s0m0s esp sx .0AINV amosom 0:0 xdo>xuooaom hdx>e0s
mo msoaoaeses sexaeaomo> umehom soozaon escapedossoo m.somse0m

.06 vanefi

66

 

 

 

 

z cacao so oooxm x

N MM M MM MM MM A se>00 .mo> ussoxo R
MM. .M N Quench sx vswxxsec N mouseees
sesao

NN NN N NN NN NN H s0>00 hAoseO amouom 8

xx xx xx x H excesses exaceeuoo

NN NN NN N moose easesxeen
es\s
xx xx x o assesses xesmcoo

NN NN m moose

xx x uueexoom expansuoo

NN a moose oasesxmoa

emse>sx hxuseoAmxswxm n M N N 0 mmsxdaem
No.0vvm He mmessOAH
useOANAsmxe hAsme u NN m emsxxueem mexoesm

00.0vm

.sOAaedosxoo aseOAMAsmxm n N < moose

x A N a H N 0 m N a 0 m < mommer mesa seaoaesem

 

.0>semos use sx AnaINv ueosom uoaeosu use use x~o>xpoodom

xxx>e0s Mo uneasiexen sexaeaomo> amosom seozaon asexuexossoo m.soese0m .u6 oases

 

 

67

Regeneration 1g Mature egg Secopgary Forest Communities

Tree Size-Clgss distribution
The size class distribution of both desirable and non-

desirable timber species combined are shown in Table 8, and
Fig. 8. For all forest tracts studied, there was heavy
stocking in the smaller tree size classes (> 2 cm < 10 cm
dbh) and intermediate stocking in the middle classes (> 11
cm 5 40 cm dbh) but very low stocking densities for the
larger size classes. In all cases, and for all size classes
but one, the plots in uncut mature forest displayed the
highest stocking densities. In one intermediate size class
(3 21 cm < 30 cm dbh) the plots in lightly cut forest (K-14)
exhibited one of the highest stocking densities, 19.8% more
than the density of the same class in uncut forest. It was
suggested that light felling and slight opening of the
canopy encouraged more regeneration and growth of pre-
existing and new fast growing species which resulted in the
big increase in stocking. Again, most understory species
like g3 congensis and 33 nobilis, which exhibit no large
trees, showed extremely high stocking densities in the

smaller and intermediate size classes. They also seem to be

regenerating continuously in all compartments except the

heavily cut and treated forest (K-13). One should also note
that the effect of selective cutting and intensity of
cutting and/or treatment were detrimental to both the
seedling/sapling classes and larger tree size classes (Table

8, Fig. 8). The contribution of thedesirable or canopy

68

.hvxmsou sx omeoxosx aee>ses

amen 0>xvxmos me: chose a

 

 

e.os x.xe m.xm m.se 0.x
e.s x.e m.e s.es e.oe «Aux
s.om o.ee o.ee n.3e A.os
e.x a.» o.» e.os «.60 oAIx
n.ee o.ee .m.oxs ,e.em 0.0x
«.3 0.x o.¢ e.e e.em «Aux
cox eos cos sex sex
e.o ve.o xe.o o.x «.0o oxux
eoxo sooem

00A c0vnmA ONVHNA chuaA cavNA

snu I0 sx mommeao eaxm

 

.AxooAu. Hosasoo
e me chN sax: umo>ses eosxm noose wsxuseae lssxxel expanses

0s» 0» sewassxsasoo Hexvsoaos use o>somos one no masoluxemloo
huSAe es» s« menoose 00x0 m0 sewasnxsuexu mmeao oaxm asoosom .0 oaseh

68A

Figure 8. Stand curves for uncut forest (K-30) lightly cut
forest (K-14), heavily selectively cut forest (K-15) and the

heavily selectively cut and treated forest tract (K-13) of
the reserve.

69

+50 5.0 2. mmwmjo mN_m

00.: mIimIim

m3... ..... «:54

#3.. ll

3.0.0-510
>mN

.1
0e

9

In»

09s! 0& 06100!

.16

(at

(A.

z
.1
0e 0..

.6

°.
0

°.
1-

Q
N

O.
0

0.
<1-

 

( euI'on) mIsuap 6611 0L(So-I

70

tree species to the density and distribution of trees in the
various size classes was small (see tree density and basal
area). Clearly, selective felling disrupted the
distribution and compositon of the desirable trees in cut
areas. This was reflected in importance values and ranks of
species (Table 5) and the great fluctuations in tree stand
curves for size classes above 60 cm dbh (Fig. 8). Tree
species such as A; altissima, 33 unijugata, 93 africana, 93
gorungosanum, 93 millenii, g; bagshawei, g; buchananii, Q;
welwitschii, £3 excelsa and I; splendida which were
particularly sought during felling had the lowest
representatives in all size classes studied and for all cut
forest tracts. Even some of the largest and often most

valuable timber species such as g; excelsa and Q;

welwitschii showed significant deficiencies both in the

 

smaller and intermediate size classes in all study plots.
The greater proportion of individuals of these species was
in size classes greater than 50 cm dbh. They also show poor
regeneration under semi-open forest canopies. It was
thought that such species, deficient in seedling, sapling
and pole size classes, might require forest gaps for
successful regeneration.
Tree seedling and sapling dypamics

Throughout this chapter, it has been emphasized that
the non-desirable timber species contributed significantly
to tree species richness, distribution, density and

dominance and thus were ranked high in importance (Table 5).

71

The contribution by desirable timber species to sapling and
pole species richness and density was relatively small in
all study plots (Table 9). In this section, I focus on the
nature and level of regeneration of the 19 desirable or
canopy tree species and the four important non—desirable
species (Appendix table 2). The 19 desirable species were
mostly the canopy trees which exceed 50 cm dbh at maturity.
In heavily selectively cut and treated forest (K—13),
tree species with large numbers of saplings, poles and small
trees included 3; buchananii, g; unijugata, §3 scheffleri
and Q; abyssinica in that order of importance. Together,
they constituted 56.3% of sapling regeneration, 36.1% of
pole regeneration and 50.6% of sapling and pole regeneration
of the desirable and the four important non-desirable
species (Appendix table 2). For the lightly cut forest (K-
14), the order of abundance of tree sapling species
regeneration was £3 sgynnertonii > £3 buchananii > g;
unijugata > g; scheffleri > A; toxicaria. These species
formed 63.9% of the regeneration of desirable sapling
species but 33.9% of the regeneration of both desirable and
non-desirable sapling species in the understory. In the
heavily selectively cut forest (K—15), the desirable tree
species with large numbers of saplings and poles included 33
unijugata (in the lead) followed distantly by E; buchananii,
g; africana, E3 macrophylla and g; scheffleri in that order.
These accounted for 39.8% of desirable sapling regeneration

in the understory. Under semi—closed canopy conditions of

72

 

 

 

 

 

 

 

m0x009m\mxe§ux>xusx ho honflss see! u moxoomm\e .eN\* waxesoa x
0.6 0.6 6.xx 0.xx moxoone\e
0.00N 0.00A n.0e0 0.66m x xuxmsos xv MAIN
6.0 A.m n.0m 0.on moxoosm\a
v.00» A.Ae 6.000A 0.0A0 a xaxmsoa 0e 0AIN
6.xA 0.0A 0.00 A.~e moxoose\s
0.x00 «.0xu «.000N 0.000 x xxxmson 0e vxIN
o.NA 0.x” c.0vm m.x0 moxoosmxe
e.g60 v.60A 0.Axmmx v.00wx * xuxmsos av CMIN
easexxmousb oqnexxmon oHAesxmousD oaseswmon
Ashe :0 osv :0 ex. Anne no my so «A. sons-axe oesoaem uo0<
mmsxanem soaoflesem macam a husam

 

one new mommeHOI

.meose husum snow
onxm snu so sx sexusnxsumxu oxen use msxxsem

.0 vanek

 

73

K-30, £3 sgynnertonii had the highest amount of
regeneration, g3 bagshawei followed in second place, £3
latifolia in third and £3 buchananii in fourth. These four
formed 48.5% of the regeneration of desirable sapling
species, 30.9% of the regeneration of desirable sapling and
pole classes combined, and 32.8% of all desirable and non-
desirable sapling and pole classes.

0f the nineteen common desirable species in the study
plots, seven had large numbers of saplings and poles under
closed canopy and six under opened up canopies whereas the
rest seemed to have no special preferences (Appendix table
2). Under closed canopy, saplings of Q3 swynnertoni;, 53
bagshawei, 1; splendida, 53 toxicaria, Q; gorungosangp and
£3 latifolia were abundant. All non-desirable species,
mostly the understory types such as g; congensig and I;
nobilis, were also very well represented in the sapling
regeneration under closed canopy. However, the most
Isuccessfully regenerating desirable species under open
canopies included 33 unijugata, g; macrophylla, Q;
welwitschii and £3 buchananii. Although Q; swynnertonii, g;
buchananii, A; altissima, g; bagshawei and g; africana
species had large numbers of saplings and poles in mature
forest, a greater percentage of the mature trees was either
already dead or dying. In K-30 plots, mature E3 buchananii
trees had suffered 100% mortality and Q; swynnerton;; over
80%. The mortality of trees of the other three species was

much lower: 14% of 93 africana, 17% of A; altissima, and

74

only 3.3% of up bagshawei. Most were poorly represented in
the pole and intermediate size classes but the sapling
classes were abundantly represented.

Table 10 shows the results of intercompartment
correlation analysis of the density of the saplings and
poles of the nineteen desirable and some non-desirable
species in the study plots (Appendix table 2). The
desirable sapling species regeneration in uncut and lightly
cut study plots was similar. The lightly cut (K-14) and
heavily cut (K-15) plots also showed similar amounts of
desirable sapling regeneration. The third pair which showed
similarity in sapling regeneration was the two heavily
selectively cut forest tracts (K-15 vs. K-13). All other
paired comparisons were significantly different. Similarly,
paired comparisons between uncut forest, lightly cut forest
and heavily cut forest (K-15) showed that their desirable
pole species regeneration was similar. However, all other
paired comparisons with the heavily cut and treated forest

were significantly different.

75

 

mz oz m2
sex.uo nem.ouu see.eue

~0.0vm 00.0vm
000.0uh N0¢.Oux

He.ovo
omm.ouu
em n so Ase
mAuw mmum «mum mmnm

No.0vm 02 02
«60.0"...H N0~.cus 000.0us

Ac.0vm mz
660.0us 06N.oux

so.ovo
eee.ouu

Ax u so Ase
exum wane «mum mans

mHIN

0uIN

efllN
OMIN

mmoso
xssem

 

.200 so xxv :0 0A. moxom

Asnu so 0v so NAV mwsxxsem

 

.muoxm xusvm osu no mvom use emsxxaem ooh» oasesxmouIsos pseusosax
osom use oasesxmou mo xaxmsou on» no msomxseasoo o>xuexosx00

.oA OHQeE

DISCUSSION

The Uganda Forest Department employs a polycyclic
system in the exploitation of the natural forest in the
reserves. Natural regeneration in the cut forests was
expected without application of special silvicultural
measures if cutting was followed by intensive elephant
control (Kingston, 1967). A 70-year rotation cycle of
forest exploitation was anticipated. Only commercially
desirable species 3 1.52 m girth at breast height (gbh) were
recommended for felling (Kingston, 1967), but the sawmillers
also cut trees below 1.52 gbh. Consequently, heavy timber
exploitation ranging from 14.0 ma/ha in lightly cut to over
21 ma/ha in heavily cut forest tracts ensued. In light of
the current state of forest recovery 20 years after cutting,
e.g. basal area, 55.7%, 63.6% and 37.3% and forest canopy
cover, 80.2%, 52.1% and 38.3% of uncut forest in lightly
cut, heavily cut and heavily cut and treated forests,
respectively, the 70- year rotation may not be feasible in
heavily cut areas.

When carried out with care and on a rational basis,
selective timber havesting could be one of the most
ecologically sound options for forest management causing

minor disturbances akin to natural tree falls (Goodland.

76

77

1980; Fearnside, 1983; Uhl 53 gig, 1985). However, several
factors interact to make mechanised selective timber
harvesting much more detrimental than it would appear (Uhl
25 31;, 1985; Skorupa and Kasenene, 1984; this study). In
the first instance, the cutters are not the land owners and
have no vested interest in doing the work carefully. The
cutters are likely to fell all potentially harvestable
trees, with no regard to minimum girth recommendations, thus
resulting in undersize trees being cut. Usually, the
bulldozers that follow cause damage to many seedlings,
saplings and poles during the hauling of logs to extraction
roads. The whole method of timber harvesting is lacking in
adequate control.

To overcome the difficulty of conversion of outtake
records in cubic meters to density of trees removed per
forest plot, as a measure of intensity of harvest, a tree
stump census method was devised. Although it was almost 19
years since forest exploitation, the decomposing tree
stumps, or at least the heartwood portions, were still
standing. This indicates that the hardwood forest stumps,
under secondary forest conditions, require more than 20
years to rot away. The two methods of estimating level of
forest exploitation in terms of density, i.e. conversion
from data in cubic meters (Skorupa, unpublished), and direct
tree stump census (this study), provided similar estimates

of the level of forest exploitation. Direct tree stump

censuses can, therefore, be very useful where there are no

78

records of volume of exploitation. It is very likely that
the understory vegetation cover, which was variable for the
study compartments, had a strong influence on the number of
tree stumps seen and distance at which the tree stumps were
observed. This might have affected the estimates of the
effective width of sample strips and density of observed
tree stumps. After all, the method assumed that all cut
tree stumps were still present and easily discernible in
spite of partial decomposition.

The estimates of level of forest exploitation in the
reserve were in accord with African tropical forest
management schemes where it is common practice to harvest
between 5 and 25 cubic meters of commercially desirable
wood/ha (Fontane g; 333, 1978). This level of exploitation
is often dictated by the need for adequate short term cash
returns typical of capital intensive forestry (Whitmore
1975; IUCN, 1975; USAID, 1978; Skorupa and Kasenene, 1984).
Usually, capital intensive mechanised timber harvesting in
the tropics destroys on the order of 50% (33 to 66%) of the
original stand (Fox, 1968; Ewel and Conde, 1976; Johns,
1983; Skorupa and Kasenene, 1984; Uhl g; 333, 1985). In
spite of all this extensive forest destruction during
felling, only 10% or less of the stand is actually harvested
(Fox, 1968; Johns 1983). So, it is difficult to assess the
full impact of forest logging based only on volume and level
of harvest of a few commercially desirable species. The

official Forest Department's extraction data or records do

79

not show the volume or density of trees, poles, saplings or
seedlings damaged or killed by tree crushes or by the
mechanical handling of logs from the forest to extraction
roads. Available evidence all indicates that even the
removal of 20 cubic meters of commercial timber or less/ha
results in considerable damage to the forest (Ewel and
Conde, 1976; Skorupa and Kasenene, 1984; Uhl g; 333, 1985).
In lightly cut areas, removal of 5 14 m3 of commercially
valuable timber/ha seems to have caused tolerable forest
destruction and disturbances, for high levels of recovery of
the forest structure and composition and high amounts of
forest tree regeneration have occurred. However, the forest
compartments K-13 and K-15, which had approximately 17 and
21 m3 of actual volume of commercial timber mechanically
removed/ha, respectively, seem to have been over-exploited,
causing extensive forest destruction and disturbance. This
level of exploitation may have been detrimental for the
medium altitude tropical moist forest, as such forests are
characterised by lower numbers of species and lower volume
of potentially harvestable timber than typical lowland
tropical forests (Richards, 1964, 1973; Whitmore, 1975).

Judging from the amount of deviation of selected

ecosystem variables, e.g. vegetation composition, structure
and level of regeneration etc., from a reference mature
forest, 20 years post cut, the heavily selectively cut
forests were substantially disturbed and degraded. Species

diversity has been one of the basic concepts of ecology used

80

to characterise communities and ecosystems (Gomez—Pompa 35
gig, 1972; Richards, 1973; DeJong, 1975; Whitmore, 1975).
Consequently, two commonly applied indices, the Simpson's
(Ds) and Shannon-Wiener (H') diversity indices were used to
characterise the cut and uncut forest tracts of the reserve.
Both indices are known to be sensitive to changes in the
numbers of species and the distribution of individuals
amongst the species (i.e. evenness or equitability, E)
(Krebs, 1972; DeJong, 1975; Barbour g3 31;, 1980 and Brower
and Zar, 1984). Species richness, as the simplest and least
ambiguous index of species diversity (Green, 1979) was also
compared between cut and uncut forest compartments.

Deducing from the maximum species diversity index (H
max), all forest compartments, cut and uncut, exhibited high
tree, pole and sapling species diversity indices for all
tree species combined (H = 4.4 to 5.1), but low diversity
indices for desirable species (H = 2.2 to 4.2). The
unlogged forest leads in the diversity of tree sapling with
lightly cut forest in second place and heavily felled areas
in last position. This seemed to agree with the intensity
of disturbance the forests had experienced. However, the
high equitablity indices for all study compartments clearly
indicates multispecies dominance of the forests studied.

The fact that the species-area curves approached or
were leveling off after only 40 or less 100 m2 sample plots,
for all size classes in mature and lightly cut forests

indicated that a representative number of species had been

81

observed within the study plots. This could be attributed
to the high density of trees for all size classes in the
mature and lightly cut forests. However, the values of tree
species richness observed in this study (37 to 57
species/0.5ha), were lower than those reported for tropical
wet forests (Richards, 1952; Whitmore, 1975; Uhl g; 333,
1980; Hartshorn, 1980), but higher than those reported for
tropical dry forests (Murphy and Lugo, 1986a, b). The
equitability component of diversity was relatively high,
except for the heavily selectively cut forests.

Significant differences in species diversity, richness
and equitability still prevail among uncut, lightly cut and
heavily selectively felled forest tracts. In the heavily
cut forests, 61.1 to 67.6% and 60 to 80% of the large tree
and pole classes, respectively, included pioneer and fast
growing secondary forest species which were uncommon in
mature and lightly cut forests. The sapling classes too (50
to 60%) were dominated by non-desirable weedy species. Pure
stands of 13553 ggineensis, a fast growing pioneer species,
were found mainly in the heavily felled forest tracts.
Timber stock enumerations prior to forest exploitation

showed that tree diversity was relatively constant for all

subtypes of Parinari forest (Kingston, 1967). Therefore, it
is very unlikely that physical variation and natural
phenomena, alone, could explain the great differences in
species richness, diversity and equitability for the cut and

uncut forest tracts within a single Parinari forest type.

82

Differential logging intensity, ensuing disturbances and
treatment, are thought to account for the large part of the
differences exhibited. The Forest Department's post felling
silvicultural practice of poisoning non-commercial residual
trees (weeds), could also account for the 21.9%, 12.9% and
2.0% further depression of floral diversity of saplings,
poles and trees, respectively, in the heavily cut and
treated forest tracts of the reserve.

The species diversity and richness were not the only
characteristics affected by logging in the reserve.
Directly and/or indirectly, heavy selective logging resulted
in decreases in the sapling, pole and tree density,
distribution and basal area. Light selective cutting
decreased the density of saplings, poles and trees by 19.1,
32.4 and 7.3%, respectively, while heavy selective cutting
by 54.3, 55.9 and 27.7% respectively. Light selective
cutting increased the density of desirable poles to 128.6%
(i.e. 28.6% more than in uncut mature forest) whereas heavy
cutting has reduced it by 34.6 to 53.9%. However, the
greatest reduction in the density of saplings, poles and
trees was observed in the heavily cut and treated forests
where percentage decreases of 75.8, 69.9 and 32.4%
respectively have been recorded. The same general trend
applies for the basal areas of saplings, poles and trees.
The reductions in densities, distribution and basal areas
seemed to correspond to the intensity of exploitation and

treatment.

83

The drastic decreases in density, distribution and
basal area have also led to low importance values of the
desirable timber or canopy species. This is due to the fact
that felling was selective with respect to size and prime
species. Selection for size and prime species during
cutting may also explain the low correlations between
importance values and ranks of tree species in uncut and
heavily cut forests or lightly cut and heavily cut forests.
However, the significant positive correlations between
importance values and ranks of tree species in uncut and
lightly cut forests were clear indications of high forest
recovery rates in lightly selectively cut than heavily cut
forests.

Accompanying drastic decreases in density and basal
areas of trees were serious reductions in forest canopy
cover, more especially in heavily cut areas. In selective
felling, the maximum recommended level of reduction of
closed canopy cover lies between 40 and 50% (Kingston, 1967,
Uhl g5 33;, 1985). But in the actual cutting of the forest
in the reserve, this was not adhered to. Twenty years
after, the forest canopy cover in lightly cut, heavily cut
and heavily cut and treated forests was still 19.8%, 47.9%
and 61.7%, respectively, less than that in uncut mature
forest. Consequently, the high intensities of sunlight
reaching into the forest encourage the growth and
development of a dense tangle of impenetrable herbs,

climbers and non-woody shrubs. The dense understory was

84

strongly negatively correlated with seedling and sapling
regeneration of both desirable and non-desirable species.
Generally, the slightly open forest canopy as in uncut and
lightly cut forests, encouraged more regeneration and
provided protection to natural regeneration of hardwood
species including the desirable timber species, at the
expense of a thick understory of herbs and non-woody shrubs.
Therefore, many of the vegetational differences between
uncut, lightly cut and heavily cut forest tracts were a
result of habitat modification by selective felling
operations and not initial differences in the habitat
conditions.

In the Budongo forest, Uganda, Philip (1965) observed
adequate natural regeneration of valuable timber species
after felling and poisoning of weedy species. But most
studies concerned with forest regeneration and recovery
after felling operations or other forms of disturbances
(e.g. Eggeling, 1947a; Richards, 1964; Whitmore 1975;
Synnott, 1975; Uhl 23 gig, 1985, 1982b and this study)
contradict Philip's observations. However, it does appear
that differences in level and type of disturbance are very
important in forest regeneration and recovery. Most studies
suggest that selective logging encourages the regeneration
of colonizing or secondary forest species with only a small
representation of primary forest species in intermediate
stages of succession (Richards, 1964; Uhl 53 £33, 1985).

Forest communities which have undergone clear cutting or

85

have had any severe disruption show decreases in numbers of
K-selected species and increases in opportunistic or r-
selected species.

In their natural habitats, characterised by low light
conditions, seedlings and saplings of primary forest species
apparently possess adaptive mechanisms which permit them to
undergo long periods of suppression and still recover after

a disturbance (Richards, 1972; Whitmore, 1975; Gomez-Pompa

If.

33;, 1976). However, these shade-tolerant species differ
with respect to optimal light intensities for growth
(Marquis g; 333, 1986; Lugo, 1970). As the results of this
study also suggest, the logged or desirable primary forest
species may have found post-logging light conditions (i.e.
3.5 to 5.0 times more in heavily cut than uncut forest)
unsuitable for their own regeneration. Tree species such as
g; congensis and I; nobilis are characteristic understory
species of the reserve and probably are shade- requiring in
habit. Their impoverished distribution, lower densities,
lower dominance and absence in heavily cut and significantly
disturbed forests in the reserve may suggest failure to
survive in altered microhabitat and forest climate. In a
red fir forest in Sierra Nevada, California, Ustin g; g};
(1984) also found that high light intensities on open sites
were inhibitory to red fir seedling regeneration. They
suggested that seedlings were absent from high light plots
because of temperature and water stresses induced by higher

irradiances early in the growing season.

86

Observations in this study and those of Langdale-Brown
g; 2;; (1964) and Kingston (1967), suggest severe lack of
regeneration of g; excelsa and 23 welwitschii in the high
forest of the reserve. These were the dominant tree species
in some parts of the reserve and often the largest trees.
The scarcity of saplings, poles and small trees in the
understory shows that their populations may be unstable and
declining as a component of the system. In African forests,
poor regeneration of dominant species, as was the case in
this study, may indicate changing composition of the
community (Langdale-Brown g; 533, 1964; Richards, 1964).

The changes may be part of the normal process of development
of a successional forest towards a stable climax state.

In the Budongo forest, Uganda, Eggeling (1947),
Brasnett (1946) and Synnott (1975) suggested that lack of
adequate seed supply after cutting was responsible for the
low species diversity and limited regeneration of trees in
the secondary forest. Over-exploitation of mature trees and
treatment of over-mature and defective desirables with
arboricides reduces the potential seed sources for forest
regeneration. Other major factors germane to heavy forest
logging that have been shown to suppress forest regeneration
include: drastic changes in biotic components (e.g. rodents
and insects, Janzen, 1971; Synnott, 1975; Isabirye-Basuta,
1979; Kasenene, 1980, 1984), changes in microlimatic
conditions of the forest (Richards, 1964; Whitmore, 1978;

Gomez-Pompa, 1976) and physical-chemical changes in the

87

soils (Armson 25 53;, 1973; Putz, 1983). However, other
observers suggest that recovery of a felled forest may
depend on either the persistence of adequate primary forest
tree species on the spot as adult refuges (Synnott, 1975;
Kasenene, 1980, 1984; this study), regeneration from
dormant soil seed banks (Whitmore, 1975; Putz, 1983a; Uhl
and Jordan, 1984; Young, 1985), survival of pre-existing
seedlings, saplings and sprouts (Uhl g; 333, 1982b, 1985 and
Murphy and Lugo, 1986a) or the possibility of colonisation
from outside (Richards, 1964, Gomez-Pompa 33 33;, 1972, Liew
and Wong, 1973; Poore, 1976 and Hartshorn, 1978). However,
seed viability for some trees may be very short and
successful regeneration may occur only under closed or
slightly altered forest conditions. In the heavily
selectively cut forest tracts of the reserve, a combination
of factors, including excessive felling of the mature trees,
the killing of undesirable and defective desirable trees
with arboricides and windthrows (Skorupa and Kasenene, 1984)
drastically reduced the number of reproductively mature
trees. Seed supply may have been locally inadequate,
contributing to low potential for natural regeneration.

Therefore, in selective felling, much effort should be

geared toward minimizing the impact of deforestation. The
crucial matter in the felling system should be to control

the damage done.

SUMMARY AND CONCLUSION

The composition, structure and regeneration of uncut
mature forest and selectively cut secondary forest tracts of
the Kibale Forest Reserve were investigated to determine the
effects of mechanized selective timber harvesting. The
mature and secondary forest communities both exhibited high
species diversity and equitability which indicated lack of
forest dominance by any one tree species. The species-areas
curves for saplings, poles and trees reached asymptotes in
less than 0.5 ha of cumulative area sampled. This suggests
high stand densities for the three size classes examined.
The species richness observed for the mature high altitude
tropical moist forest averaged 49 to 57/0.5 ha and was lower
than those reported for lowland tropical moist or
rainforests but high compared to tropical dry forests.
However, the average values of basal areas of trees, poles
and saplings combined were high (40.7 to 109.1 mz/ha)
compared to those reported for tropical wet or rainforest
(20-75 mz/ha) and tropical dry forests (17 to 40 mz/ha). A
combination of high stand densities for the three size
classes studied and inclusion of the saplings and poles in
the sample, may have caused the inflation of the values for

basal areas.

88

 

89

In high altitude tropical moist forest, mechanised
selective timber exploitation in excess of i4 ms/ha of
commercially desirable timber was observed to be destructive
to the forest structure, composition and regeneration. The
forest compartments which had 17 and 21 ma/ha of actual
volume of commercial timber mechanically removed appear to
have been overexploited, causing extensive damage and
disturbance to the forest. In contrast, the lightly cut
forest had 14 m3 of commercial timber per hectare removed,
and high amounts of forest regeneration and recovery have
occurred. Directly and/or indirectly, heavy selective
logging resulted in drastic decreases in equitability, basal
area, density and species richness of saplings, poles and
small tress and large trees. The reductions in
equitability, density and basal area also led to low
importance values of desirable timber or canopy species
since forest felling was selective with respect to size of
trees and prime species.

Corresponding to drastic decreases in density and basal
areas of trees were serious reductions in forest canopy
cover of the heavily cut areas. Consequently, high
intensities of sunlight, deleterious to primary forest
seedling and sapling species regeneration, induced a dense
tangle of herbs, climbers and non-woody shrubs. The dense
understory was itself strongly negatively correlated with
seedling and sapling regeneration of both desirable and non

valuable tree species. Thus, great structural and

90

compositonal differences between uncut, lightly cut and
heavily selectively cut forest tracts of the reserve still
existed at the time of sampling, 20 years post cut.

Physical variation and natural phenomena alone could not
explain the great differences in species richness, diversity
and equitablity for uncut, lightly cut and heavily
selectively cut, and cut and treated forest tracts of the
reserve. Differential logging intensity, ensuing problems
of forest disturbance and arboricide treatment were
evidently the major cause.

The uncut mature forest exhibited the highest levels of
species richness, density and basal areas for trees, poles
and saplings of both the desirable timber or canopy species
and non-valuable tree species. In this respect, the lightly
selectively cut forest followed closely in second place, the
heavily cut in third and the heavily cut and treated forest
in fourth position. The gradations in tree species
diversity, density, basal area and level of forest
regeneration in terms of sapling and pole species richness,
density and equitability of both desirable and non-valuable
timber species, 20 years after selective felling closely
corresponded to the intensity of felling and felling with
treatment. Treatment of the residual forest weeds with
arboricides was suspected to be the major cause of further
depression in the floristic diversity of both desirable and
non—desirable species in the heavily cut and treated forest

tracts of the reserve.

91

Conclusion

Because of the intensity of such operations, the damage
inflicted by mechanical logging operations is often much
more severe than that of pitsawyers (Kasenene 1984, Skorupa
and Kasenene 1984, Struhsaker pers. comm.). Modified forms
of a polycyclic felling system (Whitmore 1975) that are
slightly more advanced than pitsawing conceivably could be
managed in ways compatible with maintaining the dynamic
structure and functions of the forest. Timber extraction
through controlled polycyclic systems that could remove
between 2 and 4 canopy trees/ha would be akin to natural
tree fall disturbance causing small gaps in the forest
canopy (Whitmore 1975, Herbert and Beveridge 1977, Uhl g;
3;. 1985) and the damage done to the forest would be
relatively minor. Therefore an important goal of rational
forest exploitation should be to manipulate the forest
canopy to create gaps of a size that favor the growth and

development of the chosen species.

LITERATURE CITED

Armson, K.A. and R.J. Fesseden, 1973. Forest windthrows and
their influence on soil morphology. Soil Sci. Soc.
Amer. Proc. 37:781-784.

Barbour, M.G., H. Burk and W.D. Pitts, 1980. Terrestrial
plant Ecology. The Benjamin/Cummings Publishing Co.
Inc. 604 pp.

Bolin, B. 1977. Changes of land biota and their importance
for the carbon cycle. Science 196:613-615.

Brasnett, N.V. 1946. The Budongo forest, Uganda. J. Oxf.
Univ. For. Soc. 3(1)11-17.

Brazier, J.D., J.E. Hughes and C.B. Tabb, i976.
Exploitation of Natural tropical forest resources and
the need for genetic and ecological conservation. In:
Tropical trees: Variation, Breeding, Conservation. pp.
1-10. (eds. J. Hurley and B.T. Styles). Linnean
Society Symposium Series No. 2. Academic Press.

Brenan, J.P.M. 1973. A summary. In: Tropical ecosystems
in Africa and South America: A Comparative review. pp.
1-8 (eds. B.J. Meggers, E.S. Ayensu and W.D. Duckworth)
Smithsonian Institution Press, Washington, D.C.

Brower, J.E. and J.H. Zar, 1984. Field and Laboratory
methods for general ecology. W.M.C. Brown Company
Publishers, Dubuque, Iowa. 226 pp.

Brown, V.K. 1984. Secondary succession: Insect plant
relationships. Bioscience 34(11)710-716.

Brown S. and A.E. Lugo, 1982. The storage and production of
organic matter in tropical forests and their role in
the global carbon cycle. Biotropica 14:161—187.

Chapman, E.C. i975. Shifting agriculture in tropical
forest areas of S.E. Asia. In: IUCN (1975), The use
of Ecological Guidelines for development in tropical
forest areas of S.E. Asia. IUCN Series No. 32. Morges
Switzerland. 186 pp.

92

93

Curtis, J.T. and R.T. McIntonsh, 1951. An upland forest
continuum in the Prairie—forest border region of
Wisconsin. Ecology 32:476-496.

DeJong, T.M. 1975. A comparison of three diversity indices
based on the components of richness and evenness.
Oikos 26:222-227.

Eggeling, W.J. 1947a. Observations on the ecology of the
Budongo rainforest, Uganda. J. Ecol. 34:20-87.

Eggeling W.J. and I.R. Dale. 1951. The indigenous trees of
the Uganda Protectorate. The Government Printer,
Entebbe. Uganda. 491 pp.

Ewel, J.J. 1971. Biomass changes in early tropical
Succession. Turrialba 21:110-112.

1977. Differences between wet and dry
successional tropical ecosystems. Geo—eco. Trop.
1(2)163-117.

 

1980. Tropical Succession: manifold routes to
maturity. Biotropica: Supplement 12(2):2-7.

Ewel, J.J. and L. Conde, 1976. Potential ecological impact
of increased intensity of tropical forest utilization.
Final report to USDA Forest Service. Forest Products
Laboratory, Madison Wisconsin. 115 pp.

Ewel, J.J., C. Berish, B. Brown, N. Price and J. Raich,
1981. Slash and burn impacts on a Costa Rican Wet
forest site. Ecology 62:816-829.

FAO-UNEP, 1981. Forest Resources of Tropical Africa, FAO,
Rome.

Fearnside, P.M., 1983. Development alternatives in the
Brazilian Amazon: An Ecological Evaluation.
Interciencia 8:65-78.

Fosberg, F.R., 1973. Temperate forest influence on tropical
forest land use: A plea for sanity. In: Tropical
ecosystem in Africa and Southern America: A
comparative review. pp. 245—350 (eds. B.J. Meggers,
E.S. Ayensu and W.D. Duckworth). Smithsonian
Institution press, Washington, D.C.

Fox, J.E.D. 1968. Logging damage and the influence of
climber cutting in the lowland Dipterocarp forest of
Sabah. The Malayan Forester 31:326—347.

94

Global 2000 Report, 1980. The technical report Vol. 11.
Department of Environmental quality and Department of
State. Washington, D.C.

Gomez-Pompa, A., C. Vazquez-Yanes and S. Guevara, 1972. The
tropical rainforest: A non renewable resource. Science
117:762—765.

Gomez-Pompa, A., A.L. Asada, F. Golley, G. Hartshorn, D.
Janzen, M. Kellman, L. Nevling, J. Penalosa, P.
Richards, C. Vazques and P. Zinke, 1974. Recovery of
tropical ecosystems: In: Fragile ecosystems:
Evalution of Research and applications in the
Neotropics. pp. 113-138 (eds. E.G. Farnworth and P.B.
Golley). Springer Verlag, New York.

Goodland, R.J.A. 1980. Environmental ranking of Amazonian
development projects in Brazil. Environmental
Conservation 7:9—25.

Green, R.H. 1979. Sampling design and statistical methods
for environmental biologists. 257 pp. John Wiley and
Sons, New York.

Hamilton, A.C. 1981. A field guide to Uganda forest trees.
Privately Published, Uganda.

Hartshorn, G.S. 1978. Tree falls and tropical forest
dynamics. In: Tropical trees as living systems. pp.
617-638 (eds. P.B. Tomlinson and M.H. Zimmerman).
Cambridge University Press, Cambridge.

Hartshorn, G.S. 1980. Neotropical forest dynamics.
Biotropica 12 (Supplement) 23—30.

Herbert, J. and A.E. Beveridge, 1977. A selective logging
trial in dense Podocarp forest in Central North Island,
New Zealand J of Forestry, 22:81—100.

Holdridge, L.R. 1967. Life Zone Ecology. 206 pp. Tropical
Science Center, San Jose, Cost Rica.

Isabirye-Basuta, G.M. 1979. The ecology and biology of
small rodents in Kibale Forest, Uganda. M.Sc. Thesis,
Makerene University.

IUCN-1975. The use of Ecological guidelines for development
in tropical forest areas of South East Asia. IUCN
Publication: New Series, No. 32,1110 Morges,
Switzerland.

Janzen, D.H. 1971. Seed predation by animals. Ann. Rev.
Ecol. Syst. 2:465-492.

95

Johns, A.D. 1983. Effects of commerical logging in a wet
Malaysian Primate Community. In: Proceedings of the
IXth Congress of the International Primatological
Society, Atlanta Georgia, USA. (eds. F.A. King and
D.M. Taub) Van Nostrad—Reinhold, New York, New York.

Jordan, C.F. and R. Herrera, 1981. Tropical rainforests:
Are nutrients really critical? American Naturalist
117(2)167-179.

Kasenene, J.M. 1980. Plant regeneration and rodent
populations in selectively felled and unfelled areas of
the Kibale Forest, Uganda. M.Sc. Thesis, Makerere
University.

1984. The influence of selective logging on
rodent populations and the regeneration of selected
tree species in the Kibale forest, Uganda. Trop. Ecol.
25:179-195.

Kingston, B. 1967. Working plan for the Kibale and Itwara
Forest reserves. Uganda Forestry Department, Entebbe.
142pp.

Krebs, C.J. 1972. Ecology. Harper and Row, New York. 694
PP-

Langdale-Brown, I., H.A. Osmaston, J.G. Wilson, 1964. The
Vegetation of Uganda and its bearing on land use.
Uganda Government Printer, Entebbe. 159 pp.

Leopold, A. 1948, Game management. Charles Schribner's
Sons, New York. 481 pp.

Liew, T.C. and F.O. Wong, 1973. Density, recruitment,
mortality and growth of Dipterocarp seedlings in virgin
and logged over forest in Sabah. Malay. Forester
36(1)3—15.

Longman, K.A. and J. Jenik, 1974. Tropical forest and its
environment. Longman Group LTD. London. 196 pp.

Lugo, A. 1970. Photosynthetic studies on four species of
rainforest seedlings. In: Tropical rainforests. (eds.
H.T. Odum and R.F. Pigeon). U.S. Atomic Energy
Commission, Div. Tech. Info. Oak Ridge Tennessee.

Lugo, A.E. and S. Brown, 1981. Ecological monitoring in the
Luquillo Forest Reserve. Ambio Vio No. 2-3:102-107.

Lugo, A.E. and P.G. Murphy, 1986. Nutrient dynamics of a
Puerto Rican subtropical dry forest. Journal of
Tropical Ecology 2:55-72.

 

 

96

Marquis, R.J. and H.J. Young, 1986. The influence of
understory vegetation cover on germination and seedling
establishment in a trOpical lowland wet forest.
Biotropica i8(4)273-278.

Murphy, P.G. and A.E. Lugo, 1986a. Structure and biomass of
a subtropical dry forest in Puerto Rico. Biotropica
18(2)89-96.

 

and A.E. Lugo, 1986b. Ecology of tropical dry
forest. Ann. Rev. Ecol. Syst. 17:67-88. .

Myers, N. 1979. The Sinking Ark. Pergamon Press, Elmsford,
New York, USA. 307 pp.

1980. Conversion of tropical moist forests.
National Academy of Science, Washington, D.C., USA. 205

PP- ‘

 

Odum, E.P. 1985. Trends expected in stressed ecosystems.
Bioscience 35(7)419-422.

Osmaston, H.A. 1959a. Working plan for the Kibale and
Itwara forest, Uganda. UGANDA Forestry Department,
Entebbe. 60 pp.

Philip, M.S. 1965. Working plan for the Budongo central
forest reserve. Government Printer, Entebbe, Uganda.

Plumwood V. and R. Routley, 1982. World rainforest
destruction: The Social Factors. Ecologist 12:4-22.

Poore, D. 1975. Retention of the natural forest as a
resource. In: IUCN: The use of ecological guidelines
for development in tropical forest areas of S.E. Asia.
IUCN Series No. 32. Morges, Switzerland.

Putz, F.E. 1983. Treefall pits and mounds, buried seeds
and the importance of soil disturbance to pioneer trees
on Barro Colorado Island, Panama. Ecology 64:1069-
1074.

1984. The natural history of lianas on Barro
Colorado Island, Panama. Ecology 65(6)1713-1724.

 

Richards, P.W. 1952. The tropical rainforest. Cambridge
University Press, Cambridge. 450 pp.

1964. The tropical rainforest; An Ecological
Study. Cambridge University Press, Cambridge. 450 pp.

 

97

1973. Africa, the "Odd man out" In. Tropical
Forest ecosystems in Africa and South America: A
comparative review. pp. 21—26 (eds. B.T. Meggers, E.S.
Ayensu and W.D. Duckworth) Smithsonian Institution
Press, Washington, D.C.

Sanderson, I.T. and D. Loth, 1965. Great Jungles. Julian
Messner, New York. 480 pp.

Skorupa J.P. and J.M. Kasenene, 1984. Tropical forest
management: Can rates of natural tree falls help guide
us? Oryx 18(2)96—101.

Smith, R.L. 1974. Ecology and field biology, 2nd Edition
Herper and Row Publishers, New York. 849 pp.

Sokal, R.R. and F.J. Rohlf, 1981. Biometry, W.H. Freeman
and Co., San Francisco. 859 pp.

Struhsaker, T.T. 1975. The Red Colobus Monkey. University
of Chicago Press, Chicago. 311 pp.

Synnott, T.J. 1975. Factors affecting the regeneration and
growth of seedlings of Entandrophragma utile (Dawe and
Sprague) Sprague, in Budongo forest and Nyabyeya
nursery, Uganda. Ph.D. Thesis, Makerere University.

TIE Report, 1973. In: Fragile ecosystems: Evaluation of
research and application in Neotropics (eds. E.G.
Farnworth and F.B. Golley). Springer Verlag, New York.
258 pp.

Uhl, C. and C.F. Jordan, 1984. Succession and nutrient
dynamics following forest cutting and burning in
Amazonia. Ecology 65(5)1476-1490.

Uhl, C. and R. Buschbacher, 1985. A disturbing synergism
between cattle ranch burning practices and selective
tree harvesting in the Eastern Amazon. Biotropica
17(4):265-268.

Uhl, C., C. Jordan, K. Clark, H. Clark, and R. Herrera,
1982b. Ecosystem recovery in Amazon Caatinga forest
after cutting, cutting and burning and bulldozer
clearing treatments. Oikos 38:313-320.

UNESCO. 1978. Tropical forest ecosystems: A state of
knowledge report. Natural Resources Research No. XIV,
United Nations, Paris.

USAID. 1978. Proceedings of the United States conference
on tropical deforestation Washington, D.C. USA.

98

Ustin, S.L., R.A. Woodward, M.G. Barbour and J.L. Hatfield,
1984. Relationship between sunfleck dynamics and
Redfir seedling distribution. Ecology 65(5)1420—1428.

Wadsworth, F. 1978. In: USAID preceedings of the United
States strategy conference on tropical deforestation
Washington, D.C. USA.

Whitmore, T.C. 1975. Tropical rainforests of the far east.
Clerendon Press, Oxford. 282 pp.

Whittaker, R.H. 1975. Communities and Ecosystems.
MacMillan Publishing Co. Inc. New York. 385 pp.

Wong, 0.8. 1978. Atmospheric imput of carbon dioxide from
burning wood. Science 200:197-200.

1979. Carbon input to the atmosphere from
forest fires. Science 204:210.

Woodwell, G.M. 1983. The blue planet of wholes and parts
and man. In: Disturbance and Ecosystems. pp. 2-10
(eds. R.A. Mooney and M. Godron). Springer Verlag,
Berlin.

Woodwell, G.M., R.H. Whittaker, W.A. Reiners, G.E. Likens,
C.C. Delwiche and D.B. Botkin, 1978. The biota and
Carbon budget. Science 199:141—146.

Young, K.R. 1985. Deeply buried seeds in a tropical wet
forest in Costa Rica. Biotropica 17(4):336-338.

 

99

 

 

x x x x u
.sozoA so Ao>oA 00.0 I a I m
0:0 00 eeouoxees I e I x x oueeram
xAuseoAmAsmAm I I A oAnesAmon
essosm Mo msAem u a
oxIN u e x x e
0AIN u m I x I x m
eAIN u N I x I x N evom
NAIN u A I I A oAnesxeon
I...» use xxx A.
I x I a I m
I I a I N moose
I I I A oasesxmoo
I...» use use A.
I. I... I...” a
I x I u I x N mmsAAnem
I I I A
In». a.” 4x» A.
I» I. I» x
I a I a I a N mvom
I I I A
Inxx x a u A.
I x e In N
I I x I x N moose
I I I A
v m N A m N A m N A
es\ s es\e mmosAOAs msouosesem
eoxe Aemmm xaAoso0 mvoon

 

.mxouoseses soAaesoflsso soAaevomo> amosom oaom so umou

omsem oAnAaAsz m.se0ssa use ooseAxe> 0o exexxese mo xseassm .A oAneB xxusonm<

100

moxoonm soAsAv oAnesAmouIsoz

nu

 

 

 

 

 

 

o.o e.o A.Ax o.me e.xx e.mox e.mx A.eox . mAmeomeoo .:
e.AA x.ex x.x x.xx o.o o.oo. A.~ A.xe eomoeoAmm .e
e.AA x.xx e.on e.mm m.xo x.xmx e.Ae m.exA I . mAAAnoo .e
0.8 o.eA o.o o.o o.o e.eA A.x A.xm euoeAAsnowu .m
x.ox m.oe v.4 A.Am 0.x» x.ee e.om e.ee Amereoeom .m
e.« s.« o.o «.4 e.e e.m o.o o.o coroeeee .o
o.o o.o x.~ «.mx 4.4 ,e.eA o.o x.eA oonoxo .o
x.x x.o ~.x A.Ax o.o e.ee A.x x.m AAsmmeAzAmm .o
x.e o.eoA x.~ N.~e o.oA 4.4oA o.o x.eo AAeecarosn .z
6.0 e.¢ o.o x.xx m.e m.ee ~.xA «.mAA Arrogance .s
e.eA e.e e.eA «.mA A.oe A.ox m.ee e.mx xonoxmwAm .2
5.0 x.o x.« 6.6 e.m e.xeA e.ox e.eex AssessmemNMm .4
o.eA x.o x.xA A.Ax a.me e.A~ o.ee e.mx 3AAoeA0wA .x
m.x x.o e.e o.m A.xA A.xA A.x x.m ImAsevoxmm .x
e.¢ e.eA A.AA «.4e s.o A.mA A.~ ~.m «AAxrooeooa .x
x.e e.oe v.4x e.ee e.Ax o.oe m.oe A.ooA . eoecxemxna .o
«.6 o.o A.AA o.m x.x «.3 x.m ~.xA AseoAAAa .o
o.o e.mA o.o ~.x~ u.e «.eA A.x m.oe eaeomoxeseou .o
x.om 0.xA e.ex s.ex A.o~ ¢.eA m.ex m.oe . AAoceuzo .o
x.x 6.0 x.xx o.m« A.ox o.oe A.~ o.ee acaoAswm .o
m.xx e.me A.AA e.mxA e.o A.mAA A.x e.mx ammusers .m
o.e e.ex 0.0 ”.mA A.xA o.xe m.m e.emA «AoeoAon .<
o.AA x.o u.~ x.x x.x e.Ax A.x m.oe eaArmAAAa .<
on em on am on em on em

moxoonm

«AIx oAIx eAIx oeIx more

mezxzex<ozoo

 

.mquo xosem 0:» you .00. mvoo

use .em. mmsAAsem moxoose oasesxmou No buxmsou mo moaeaxxmm .N oAneB xAusosn<

PART II

MODE OF TREE DEATH AND GAP CREATION IN SELECTIVELY
CUT AND UNCUT FOREST TRACTS OF THE RESERVE

INTRODUCTION

Tropical forests have often been described as mosaics
of patches of different sizes and ages of regrowth
originating in gaps created by falling trees (Richards,
1952; Whitmore, 1975, 1978; Brokaw, 1982a, b; Pickett, 1983;
Putz 53 gig, 1983). Forests are often subject to many
scales of disturbance ranging from the fall of an individual
tree or branch to the flattening of large areas by
earthquakes, hurricanes or wind and storm throws (Leigh,
1982; Foster and Brokaw, 1982). Tropical forest patchiness
has also been largely attributed to several other factors
including microtopography, landslides, various forms of tree
death, animal activity, wind and storm throws, forest
disturbances from selective logging and clearing for
agriculture, settlements and roads amongst others (Eggeling,
1947; Longman and Jenik, 1974; Hartshorn, 1978; Denslow,
1980; Brokaw, 1982a; Putz and Milton, 1982; Foster and

Brokaw, 1982; Putz 53 al., 1983; Pickett, 1983; Sousa, 1984;

Skorupa and Kasenene, 1984; Brokaw, 1985a, b; and Lieberman
_5 al., 1985).
Several workers have suggested that the natural rate of

tree falls constitutes a very important integral component

of the climax forest dynamics (Richards, 1952; Gomez-Pompa

101

I”

102

_3 gig, 1972, 1974; Hartshorn, 1978, 1980; Schenske and
Brokaw, 1981; and Pickett, 1983). For example, many canopy
tree species have been observed to depend on gaps for
establishment and growth to maturity (Richards, 1952;
Whitmore, 1975; Hartshorn, 1978, 1980; Brokaw, 1982b;
Pickett, 1983). Seeds of some forest tree species do
germinate under closed forest canopy but their seedlings
must subsequently be in a gap to ensure further growth and
establishment (Denslow, 1980; Brokaw, 1982b). Other groups
of tree species survive as suppressed saplings in closed
forest until a gap is formed above them. Then they respond
by either reducing growth and eventualy dying (Burgess,
1970; Meijer, 1970; Liew and Wong, 1973) or accelerating
growth and surviving (Richards, 1952; Whitmore, 1975; and
Gomez-Pompa 53 33;, 1972).

Pioneer tree seedlings emerge in gaps of all sizes
(Brokaw, 1982a) but their optimum growth and survival are
restricted to large forest gaps (Richards, 1952; Whitmore,
1975; Hartshorn, 1978; Denslow, 1980; Brokaw, 1982a; and
Pickett, 1983). Therefore, tree falls and consequent forest
gap formations are a very important source of environmental
heterogeneity (Whitmore, 1975, 1978; Hartshorn, 1978) which
has ramifications for ecological diversification and
evolution of rainforests (Connell, 1978; Denslow, 1980; and
Sousa, 1984). Denslow (1980) and Pickett (1983) have

written comprehensive literature synopses of tropical tree

species ecological limitations, adaptations and response to

103

gaps of specific size ranges and modes of formation.

In many forest communities, the most common natural
disturbance is the death of canopy trees. These deaths
occur in a variety of ways which may include tree uproots,
snaps, or death while standing (Brokaw, 1982a; Putz, 1983;
Putz and Milton, 1982; Putz 33 gig, 1983). Each of these
types of disturbance has a different effect on the species
composition in regenerating patches of the forest (Brokaw,
1982a; Putz and Milton, 1982; Putz 33 gig, 1983). Richards
and Williamson (1975) regard gaps, both natural and
artificially generated, as dynamic patches for forest
regeneration and recovery. However, selective felling has
been observed to foster more frequent and severe windfalls
(Herbert and Beveridge, 1977).

The existence of large gaps or exposed areas, even in
mature forest, has been suggested to be the cause of fast
turnover rates for some tropical rainforests (Brokaw, 1982a;
Foster and Brokaw, 1982). Denslow (1980) suggested that
human induced changes in the size frequency distribution of
gaps, by either increasing or decreasing the rate of gap
formation, will result in the gradual loss of species. The
preceding observations and predictions suggest high tree
fall rates for forests where there has been extensive forest
disturbance through felling. This could be detrimental to
forest regeneration and recovery since the few, usually
defective desirable tree species left behind after felling

are exposed and may not withstand wind pressure. Their

104

falls would also contribute to the damage of young
regeneration and to the further expansion of gaps (Longman
and Jenik, 1974; Brokaw, 1982). Accordingly, if
conservative use of a tropical forest ecosystem is to be
achieved, the utilization scheme should be formulated with
respect to likely impacts upon some important dynamic
processes of the forest. Consequently, the study of tree
mortality rates, with particular emphasis on the dynamics of
tree falls in mature and secondary forest communities, was
undertaken in the Kibale Forest Reserve. Thus, the main
objective for conducting the study was to establish whether
selective timber harvesting (light and heavy) had any
significant influence on the rates of tree falls, thus gap
formation. Secondly, the study was intended to check on
whether there were any correlations between rainy seasons
and tree falls and possible causes of falls. Finally, an
objective was to compare mortality rates observed in this
study to mortality rates reported for Neotropic and
Malaysian forests to determine whether there were any
similarities which might imply applications of similar or

different forest management plans.

METHODS

Intensive systematic censusing of freshly fallen trees
and large branches began in November, 1984, and ran through
June, 1986. Tree and big branch fall censuses were
conducted in the lightly selectively cut (K-14), heavily
selectively felled (K-15), and uncut mature forest (K-30)
compartments of the reserve. All three study compartments
are located within Parinari forest type (Kingston, 1967).
The mature, undisturbed forest (K-30), located adjacent to
the lightly cut forest (K-14), was used as a control.
Treefall census routes were permanently marked out in the
three forest compartments indicated above. The census
routes followed previously established grid lines or trails
(Fig. 1 and 2) which are maintained for primate studies. A
strip enumeration of all fallen trees encountered on either
side of census route was conducted every second and third
day of the month following the previous counts. Only new

falls, subsequent to the previous, were counted. All tree

and big branch falls which caused an observable hole or
light gap in the forest canopy and all other trees brought
down by the major fall were included in the counts.

Actual censusing of treefalls involved slow walk

(3 2Km./hr.), visual scanning of either side of the trail

105

105A

Figure 1. Sectors of study compartments 14(K-14) and 30(K-30)
showing tree fall census routes along which tree falls were
enumerated for 21 consecutive months. Only a few numbered or

lettered grid lines or trails have been included in order to
enhance clarity of the census route (thick line).

106

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

'pu snsuaa "9

 

 

 

 

 

 

 

 

 

 

 

106A

Figure 2. Sectors of study compartment 15(K-15) showing the
treefall census route (thick line) along which tree fall
enumerations were conducted for 21 consecutive months.

107

K-15

 

 

 

 

8E \

 

 

 

7E

 

 

 

6E . /

5E
4E N

.. 1

2E .

1E \
Start .L

88

W8 15 25 35 45 55 65 75 85 95 EB

 

GAE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6
1‘.Ie\\‘5“$‘ma (1 290
I m fl

 

108

for treefalls, and recording. There was no straying off the
census trail until a tree or branch fall was spotted.
Measurements of the perpendicular distance from the trail to
the tree or branchfall were made using a tape measure. The
diameter of the fallen trunk(s) at 1.3m from the root plate
(for uproots), and the diameter of the snapped tree(s) at
1.3m from the ground, were measured. The tree was
identified to species following Eggeling and Dale (1951) and
Hamilton (1981). The fallen tree condition with regard to
whether the tree was live, dead, had heart rot or hollowed
center in case of snaps, was noted.

The modes of tree death were categorised as tree snaps,
uproots and dead standing. Large limbfalls which caused
gaps were classified as branchfalls. Tree snaps included
treefalls which had breaks anywhere above the ground level
but below the branches. Tree uproots involved soil
disturbance and presence of root plate. Trees which broke
in the canopy above the lower branches were recorded as
branchfalls. Where a combination of both tree snaps and
uproot occured, the fall was designated as uproot. Both
live and dead trees involved in the falls were counted but
recorded separately. The location of the fallen tree was
mapped and the fall marked with tapes and aluminum tags
bearing dates of first observation and major tree species
involved in the fall. A similar method, following the same
census trails as for tree falls, was employed in the

censusing of dead standing trees, at three-month intervals.

109

The measurements of the perpendicular distance from the
trail to tree falls, branch falls or dead standing trees and
the number offalls or dead standing trees observed were
used to plot graphs of frequency distribution such as those
in Figure 3. The distance at which the slope of occurrence
of falls or dead standing trees began to decline permanently
to 2% or less of the total counts was taken as half the
effective width of the strip sampled. The area of forest
strip sampled was then calculated by multiplying the
estimated width of the strip by the length of the census
trail. The density of falls and dead standing trees in
terms of #lha was then calculated.

The measurements of diameter of trees involved in
falls and dead standing were used in estimating the size-
class range of standing stock that was susceptible. The
susceptible species population densities and density of
susceptible trees per study plot were obtained from the tree
enumeration data of Part One of this study. The percent
rates of falls (snap, uproot and branchfalls) and dead

standing trees per study area were then calculated thus:

DX x 100
Monthly rate = ----------
DS
120x x 100
Annual rate = --------------
20 x DS

Where

DX = Monthly density of tree fall or dead standing trees.

109A

Figure 3. Frequency distribution of tree falls with respect
to their perpendicular distance of location in the census
strip from the census trail. The 2% or lower cut off points

for K-14 and K-15 treefalls were estimated at 10m and at 12m
for K-30.

110

=erot. on :9... 23:00 Eo: .s: 002305 .e_:o_ucooeon_

 

on owe. :1 a. a. 1,. 0.. e1 e1 «10.. e. a... e1 e... ,..o
Eugen“: Isl.II......x... - - - - I . .
A , . ...... a...................:I..-./,
I/xI: lel- Ix
—. 41:14:31.4 ’5
,
4 ..././4.:. 4
xi.
m—qu ..... .4
32x3...
03.1

.06

ON

.00

 

sneiaeu Io Aouenbal :1

111

Ex 2 Mean monthly density of tree falls or dead standing
trees.

DS = Density of susceptible trees per study area.

12 and 20 = Months in a year and months of sampling,
respectively.

The time it would take all susceptible trees in each
compartment to have fallen (average life expectancy) was
estimated by dividing the density (No./ha) of susceptible
trees by total live tree falls/ha/year. However, it was
assumed that constant rates of tree falls per compartment
per year would be maintained.

The data for the first month of tree death sampling
(October, 1984) was only used as base line for subsequent
samples. It was never included in the statistical
computations due to uncertainties regarding the time over

which observed tree falls had occurred.

RESULTS

 

The lengths of the census routes were 7.18km, 6.94Km,
and 6.64Km for uncut (K-30), lightly cut (K-14), and heavily
selectively cut (K-15) forests, respectively. Estimates of
census strip widths were 20m for both lightly and heavily
cut forests (K-14 and K-15 respectively) and 24m for the
uncut mature forest. Consequently, forest areas of 13.88ha,
13.28ha and 17.23ha for K-14, K-15 and K-30, respectively,
were sampled for tree falls.

Intra-Compartment Comparisons

The monthly densities (#/ha) of tree and branch falls
were compared (Appendix Tables 1, 2 and 3) within each of
the three compartments. The Mann-Whitney U-test for large
samples (n=20 months) was used to compare the mode of tree
and branch falls within the study compartments. The U-value

was computed from the relationships:

n1 (1114-1)
U = n1n2 + -------- - R1 or
2
n2(n2+1)
U = nlnz + ---------- - R2
2

112

113

where R1 and R2 were the sums of the ranks assigned to
groups whose sizes were n1 and n2 respectively (see Siegel,
1956: pp. 119-120).

Within the lightly cut forest tract (K-14), there was
no significant difference between the densities of live and
dead branch falls, tree snaps, tree uproots or a combination
of tree snaps and uproots involved in the falls (all P>0.05,
Mann-Whitney U-test-one tailed test) among months. However,
there were more monthly tree snaps than uproots (U=106,
n1=n2=20, P<0.01 one tailed test). Both tree snaps and
uproots were independent of whether the tree was live or
dead (K2=0.333, P>0.455). With the exception of total tree
falls (combined snaps and uproots, r=0.420, P<0.05) all
other forms of falls i.e. branchfalls, tree snaps and tree
uproots were not significantly correlated with monthly
rainfall (Fig. 4).

The heavily selectively cut forest (K-15) displayed a
different pattern of tree falls (Appendix Table 2). The
densities of live and dead branchfalls and live and dead
tree uproots were similar. There were more live tree snaps
than dead tree snaps (U=119. n1=n2=20 P<0.025 one tailed
test) and the total tree falls (snaps + uproots) had more
live than dead trees involved in the falls (U=120, n1=n2=20
P<0.025). As for K-14, there were also more tree snaps than
uproots (U=134 n1=n2=20, P<0.05). However, in the heavily
selectively cut forest, live trees exhibited a higher

tendency to snap than uproot (K2=3.34, x20.o5 [1]=3.84)

113A

Figure 4. Monthly density of total tree falls (snaps +
uproots) for the lightly cut (K—14), heavily selectively cut
(K-15) and uncut mature forest (K-30) compared with monthly
rainfall (e————e).

0.900562 5 35:00 .0009... .03...
5. 0 5

 

 

 

114

 

 

 

 

 

MONTHS

 

 

 

NDJFMAMJJASONDJEMAMJ

 

 

 

 

 

 

5 0 5 . . 5 0. 5
k
5 O n
4 1 3 ‘0.
1 I I 000
K K K V:
.I... m .
.m m .....
a e
f d
N." u
h a v
t f o o
m x :
M n
H a VD.........
W...
m...”
n p b b .0 >1 n n 00000
m m m m o m m m o
3 3 2 2 1

 

1 3
29052:: E .323. 25:05.

115

compared to lightly cut forest (K-14) though the tendency
was not strongly significant. For the heavily cut forest,
the correlations between rainfall and tree falls was
extremely low and not significant (Fig. 4).

In the control plot (x-ao) the density of dead branch
falls was higher than live branch falls (U=93.5, n1=n2=20,
P<0.01) and the density of dead tree uproots was also higher
than that of live tree uproots (U=120, n1=n2=20, P<O.25).
However, there were no differences between the density of
live and dead tree snaps or total tree falls (all P>0.05,
Mann-Whitney U-test, Appendix Table 3). In contrast with K—
14 and K-15 tree falls, the density of tree snaps and
uproots was similar. As with lightly cut forest, the tree
snaps and uproots were independent of tree condition, dead
or live (x2=2.82, P>0.05). However, there were significant
positive correlations between big branch falls and rainfall
(r=0.428, P<0.05) and also for tree uproots and rainfall
(r=0.427, P<0.05). All other treefall correlations with
rainfall were not significant.

Of the total trees dead or dying in the lightly
selectively cut forest (i.e. treefalls + dead standing stock
a 142), 34.58 died because the trunks snapped, 14.13 of them
were uprooted and 52.13 died standing. or the tree falls
and dead standing stock, 79.5% was by already dead trees at
the time of sampling and 21.1% by live treefalls. Of the
134 trees involved in tree falls and dead standing trees in

the heavily selectively cut forest (K-15), 53.03 died from

116

tree snaps, 26.1% from uproots and 18.7% died standing.
Live tree falls constituted 54.5% of tree falls and dead
standing stock and dead trees the remaining 45.5%. For
uncut mature forest, the proportion of dead standing trees
was comparable to lightly cut forest. 0f the 237 tree falls
and dead standing trees recorded, 22.8% were tree snaps,
27.8% tree uproots and 48.5% were dead standing trees.
Again, the proportion of live tree falls (21.1%) and dead
trees (78.9%) involved in tree falls and dead standing stock
were comparable to lightly cut forest. Heavy selective
felling seems to have encouraged high rates of tree snaps,
most of which were live, but lowered tree deaths through
uproots and dead standing. This could be the real case
since most large trees, some of which were overmature and
thus prone to die standing or uproot, were removed by the
felling process.
Intercompartment Comparisons

Diameter measurements of trees that were involved in
falls and dead standing trees were used to fix the size
class range and density of standing stock that was prone to
tree fall death (Table 1). All trees above 15cm dbh were
susceptible to falls or death while standing. However,
trees below the 15cm dbh cut off point were also involved in
tree falls, but trees greater than 15cm dbh were most
frequently subject to tree falls and dead standing trees in

the majority of study plots.

117

Table 1. Parameters of tree death rates for the lightly cut

forest (K-14). heavily selectively cut (K-15) and

uncut mature forest (K—30) of the reserve.

 

 

 

 

 

Parameters Study Plots

K—14 K-15 K-30
Forest area (ha) sampled/month 13.88 13.28 17.23
Density (#/ha) of susceptible trees 470.3 351.1 501.7
Total trees fallen (L+D) 68 109 122
Tree snaps/ha/yr (L+D) 2.08 3.25 1.91
Percent snaps /ha/yr (L+D) 0.44 0.93 0.38
Tree uproots/ha/yr (L+D) 0.86 1.67 2.33
Percent uproots/ha/yr (L+D) 0.18 0.48 0.46
Total falls/ha/yr (L+D) 2.94 4.92 4.24
Percent of falls/ha/yr (L+D) 0.63 1.41 0.85
Total live tree falls 30 73 50
Total live tree falls/ha/yr 1.30 3.30 1.74
Percent of deaths/ha/yr 0.27 0.94 0.35
Average life expectancy‘ (yrs) 361.8 106.4 288.4
Branchfalls/ha/yr 1.90 2.53 0.42
Percent branch falls 0.40 0.72 0.42
Dead standing trees (DST)
Forest area sampled 24.98 30.54 28.72
Mean DST/month 73.81 25.01 115.02
DST/ha/month 2.95 0.82 4.0
Percent DST/ha/month 0.63 0.23 0.80

 

* Computed by dividing the susceptible tree density by total

live treefalls/ha/year. L=Live,

D=Dead.

118

Tree and branchfall density data (Appendix Tables 1, 2.
and 3) were transformed into rates (as percent) of tree
fall/ha/susceptible tree and then compared between
compartments. The non-parametric comparison by STP
(Simultaneous Test Procedure of Dwass and Gabriel, an
unplanned test for equal sample sizes), a substitute for
single classification ANOVA (Sokal and Rohlf, 1981; p. 438)
was used in comparisons. The calculated critical value of
“30.05[3v2°] was 286.63 and only values of Us equal or
greater than the critical value were considered significant.

In paired comparisons, the heavily selectively cut
forest (K—15) had higher monthly rates of tree snaps than
the lightly cut (Us=285) and uncut forest (Us=293). There
were no significant differences between the rates of tree
snaps in lightly cut and uncut forest tracts (also see Fig.
5, Table 1). For tree uproots, again, the heavily
selectively cut forest had higher rates than the lightly cut
forest (Us=283, weakly significant). The rates of tree
uproots in uncut mature forest were also higher than in
lightly cut forest (Us=305). However, the heavily
selectively cut and uncut forest had similar rates of tree
uproots (Us-201 Ns, Fig. 5, Table 1). The rates of total
tree falls (i.e. uproots and snaps combined) for the heavily
selectively cut forest were also significantly higher than
for lightly cut (Us=295.5) or uncut forest (Us=276, weakly
significant). But the rates of total tree falls in lightly

cut and uncut forest tracts were not significantly different

118A

Figure 5. Rate of tree and branch falls in terms of the
density of tree uproots, stem snaps and big branch falls
(i.e.No/ha/year) expressed as the percentage of the density
(#/ha) of susceptible trees in lightly cut (K-14), heavily

selectively cut (K—15) and uncut (K-30) forest tracts of the
reserve.

119

    

Big Branc
Falls

Total
Treefalls

Stem
Snaps

KEY
Uproots

Tree

<0. 0! «a <r. 0
1- 1- o 0
new we 9L < swans alqndeosns
10 0/0 se nemouela 10 691110 9123

120

(Us 8 244.5, NS, see also Fig. 5, Table 1). However the
rates for total branch falls (i.e. live+dead) for all paired
plot comparisons showed no significant differences among
compartments (all Us<272, NS, see also Fig. 5 and Table 1).
Comparisons of live and dead trees or big branches
involved in falls were also performed (Fig. 6 Table 1). 0f
the total tree falls, live tree falls formed 67% in heavily
cut, 44% in lightly cut and 41% in uncut mature forest. The
heavily selectively cut forest also exhibited higher
monthly rates of live tree falls than the lightly cut forest
(Us=295) or uncut mature forest (Us=274 weakly significant).
The montly rates of live tree falls for uncut and lightly
cut forest tracts were not significantly different (Us=219
NS). Consequently, estimate of the time it would take all
susceptible trees in K-14, K-15 and K-30 to have fallen
(average life expectancy) followed a pattern opposite to
that of the rates of live trees falls. The average life
expectancy for trees in the lightly selectively cut forest
was the highest (361.8 years), followed by uncut mature
forest (288.4 years) and heavily selectively cut forest
(106.4 years) (x2=134.o1, P<0.005). These results should be
taken as crude estimates due to the short sampling period
and lumping of several age classes of trees including
saplings, poles, small trees and large trees. However, with
the average life expectancy of trees in uncut mature forest
(288.4 years) as a standard, the average expectancy of

further life of trees in heavily cut forest seems to have

120A

Figure 6. Rates of live and dead tree and branch falls
(No/ha/yr) and monthly estimates of the density of dead
standing trees (No/ha/month) expressed as percentage of the
density of susceptible trees in lightly cut (K—14), heavily

selectively cut (K—15) and uncut (K-30) forest tracts of the
reserve.

 

121

woos...

zon—

mE—ocflm 50:95 «:5. m=mn_ m=mu
omen

coma cocoa 03.. 09... omen. ooh... o>3

 

 

>mv_

O

v.0

wd

N._.

qu Luo 9L < $991.]. alqndaosns lo % se
sued LICUBJQ pue sell 91m 10 peaa 10 33:93

122

been substantially reduced, while that for trees in lightly
cut forest increased. Natural variation alone could not
explain these great differences in average life expectancy
of trees from differently managed forest compartments.
Differential selective logging was suspected to have
contributed significantly to these differences and could
account for the greatest part of the differences.

For dead tree fall rates, there were no significant
differences among the study compartments (all Us.5263.5 NS)
(Figure 6). Although live and dead branch fall rates were
not significantly different for all paired comparisons, the
tendency for higher rates of live branch falls in the
heavily cut forest was apparent (Fig. 6). Clearly, the
lightly cut and uncut forest tracts had similar higher
densities of dead standing trees (Us=47 n1=n2=11, P>0.05)
than the heavily selectively cut forest (Us=26, n1=n2=11,
P<0.025, one tailed test, also see Fig. 6). There was no
correlatiOn between rates of tree falls for all study
compartment pair combinations (all r50.112, n=20, P>0.05).
This suggests that differences in forest management (in this
case, different levels of felling intensity) alters the
forest's dynamics of tree falls.

Table 2 shows the density of desirable and some
important non—desirable timber or canopy species involved in
tree mortality. Excessive removal of large desirable tree

species from K-15 apparently did not lead to reductions in

treefalls but the escalation of it. All the seventeen major

Table 2.

123

Density (#/ha) of desirable and some important

non-desirable* timber or canopy tree species

involved in tree falls, and dead standing trees in
cut and uncut forest tracts of the reserve.

 

Tree Species

Tree Falls

 

Dead standing

 

 

 

 

 

 

K14 K15 K30 K14 K15 K30

A. altissima 0.0 0.15 1.33 0.04 0.0 0.56
A. toxicaria 0.0 0.08 0.0 0.0 0.0 0.0
B. unijugata 0.0 0.15 0.06 0.0 0.03 0.0
C. africana 0.57 0.38 0.64 0.08 0.03 0.10
C. durandii‘ 0.36 0.12 0.23 0.04 0.0 0.0
C. gorungosanum 0.07 0.08 0.46 0.0 0.03 0.07
C. millenii 0.07 0.08 0.06 0.08 0.07 0.0
D. abyssinica‘ 0.50 1.66 0.41 0.08 0.03 0.0
F. macrophylla 0.07 0.08 0.0 0.0 0.0 0.07
F. angolensis 0.0 0.23 0.11 0.0 0.0 0.07
F. latifolia 0.50 0.08 0.52 0.04 0.0 0.03
L. swynnertonii 0.0 0.08 0.35 0.0 0.0 0.59
M. platycalyx 0.58 0.08 0.29 0.36 0.10 0.14
M. bagshawei 0.07 0.08 0.75 0.04 0.0 0.07
N. buchananii 0.22 0.08 0.06 0.08 0.07 0.03
N. macrocalyx* 0.36 1.66 0.12 0.04 0.16 0.03
0. welwitschii 0.29 0.90 0.34 0.08 0.23 0.0
P. excelsa 0.14 0.23 0.23 0.0 0.0 0.0
P. africana 0.0 0.08 0.06 0.0 0.0 0.0
S. scheffleri 0.29 0.15 0.29 0.08 0.0 0.30
T. guineensis* 0.50 0.68 0.0 1.64 0.0 0.0

i 0.22 0.34 0.30 0.13 0.04 0.10

Si 0.21 0.49 0.32 0.35 0.06 0.17

 

124

canopy or desirable tree species were among tree falls in
heavily cut forest; fifteen tree species were involved in
uncut and only eleven species in lightly cut forest. The
density of individuals of desirable species involved in
treefalls was significantly higher for heavily cut, relative
to uncut forest (T=54, n=20, P<0.025, one tailed Wilcoxon's
signed rank test) or lightly cut forest (T=32.5, n-21,
P<0.005). However, the density of individuals of desirable
species involved in treefalls for uncut and lightly cut
forest was not significantly different. This may imply that
for the heavily selectively cut forest (K-15), the few large
desirable trees which were left behind because of defects,
e.g. crooked boles, heart rot or hollowed center, could not
withstand post felling conditions.

In lightly cut forest, 90% of the dead standing trees
in the small size classes (36<20cm dbh) were of the pioneer
species, 1; guineensis. M; platycalyx formed the highest
percentage of the dead standing desirable tree species
(12.16%) while species of Cordia, Newtonia, Q; welwitschii.
and g; scheffleri were in the second place, each
contributing about 2.7% of dead standing stock. Species of
A; altissima, M; bagshawei and g; latifolia together
contributed 4.05% of the dead standing stock.

In the heavily cut forest, the total number of dead
standing trees was low for obvious reasons; the most
susceptible mature and overmature trees were removed during

felling and by consequent tree falls. However, one of the

125

most prized timber species, 915; welwitschii (or Elgon
olive), had the highest percentage (28%) of dead standing
stock while 5; platycalyx was second with 12% and N;
buchananii and g; millenii together formed 16%. Trees of N;
macrocalyx, a pioneer species, formed 18.28% of the dead
standing stock while the major portion of the remaining 26%
was formed by non-desirable tree species. Here, the
percentage estimates for dead standing species are inflated
because of the small number of trees involved in dead
standing stock.

In the mature forest (K-30), the dead standing species
were dominated by Logo; (14.9%), Aningeria (14.0%) and
Strombosia (7.9%). These species also had the largest
number of large individuals in the samples. In combination,
N; buchananii, M; bagshawei, g; lalifolia, g; africana, M;
platycalyx, L macrophylla and g; gorungosanum constituted
13.16% of the dead standing stock.

Diameter glass Distribution 9; £353 £311;

Within plot comparison of size classes of tree snaps
and uproots showed no significant differences for all the
study compartments (all 00.05[2,11] 2 42.5, Mann-Whitney U-

test, Siegel, 1956). Intercompartment comparisons of size

class distribution of trees involved in tree fall deaths
also showed no differences in all possible plot pair
combinations (all US$74, 00.05[3,11]-96.19, STP, Sokal and

Rohlf, 1981, p. 438). However, the pattern of distribution

126

of tree falls was variable between compartments (Fig. 7,
Appendix Table 4).

Figure 7shows the frequency of the various categories
of tree snaps, uproots and total tree falls (snaps +
uproots) in each diameter class. Tree fall peaks differed
for the cut and uncut forest tracts but the peaks for out
forests were generally similar (Fig. 7). However, Figure 8
presents the same information of total tree falls in a more
instructive form; the density of tree falls (No/ha) as a
percentage of the density of living stems in the
corresponding diameter classes. The heavily selectively cut
forest had one minor tree fall peak in the lower size
classes, between 20 and 40 cm dbh and a major peak in the
intermediate size classes, between 60 and 70 cm dbh. The
lightly cut forest also had two tree fall peaks, 3 minor one
between 40 and 50 cm dbh and a major peak between 70 and 80
cm dbh. However, some high tree fall levels were recorded
for size classes above 100 cm dbh. For uncut forest, tree
fall peaks included a minor one between 30 and 40 cm dbh and
a major peak between 70 and 80 cm dbh. The major tree fall
peaks for uncut and lightly cut forests were similar in dbh
range but not in magnitude. The major tree fall peaks for
both cut and uncut forest tracts occurred between 60 and 80
cm dbh with the heavily selectively cut forest having the
highest, lightly cut forest second highest and the uncut
mature forest in third position. The high tree falls in the

lower size classes (20 to 40 cm dbh) could be largely a

126A

Figure 7. Size class (10 cm dbh interval) distribution of
tree falls (No/ha) attributed to stem snaps and tree uproots
in lightly selectively cut (K-14), heavily selectively cut
(K-15) and uncut (K-30) forest tracts of the reserve.

Density (No./ ha) of Treefalls

Density (No./ ha) of Treefalls

Density (N0./ ha) of Treefalls

 

127

,0. K14

 

 

 

 

_* K15 KEY

‘\ li——-X——-X Stem snaps
\\ ANNA ...... A Tree uproots
.._ -.-—-. Uproot & snaps

  
   

 

 

,' ‘\ K30
/ \
ll..A """" A3.. \\
/ A ., ~«A..‘ , \ I
‘ 1 “NA"... .~.....~....,?m H’. .1

 

2-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 91-100 >100

Size Classes in cm dbh

127A

Figure 8. Frequency distribution of diameter at breast
height (10 cm dbh interval) of total tree falls (snaps +
uproots) expressed as percentages of the density (No/ha) of
living stems in the same diameter class in lightly cut (K—

14), heavily selectively cut (K-15) and uncut (K-30) forest
tracts of the reserve.

 

128

OOFA OO Twm 0a.. Pm

.o..... “.0. u,"

 

   

o9. III.
mFx 4.4.40.4 oooooo q
es. $3.111
. >w¥
2.8.8; .83

coo Eu E $320 3.5

cm. wk Oh. rm Om. Fm 0m. we 0?. pm on. PN ON- w P 0 TN
1 o<0 1 4 I d ‘ 11‘ ‘

 

\ X.

\xl..x.. .M.
\

 

SWBLS ONIAI'I :IO SSVLNEOHHd

129

function of large tree falls which may also break many
smaller trees. Whenever there was a major tree fall,
several saplings, poles and small trees were brought down as
well. With the exception of lightly cut forest, very few
trees 390 cm dbh were involved in tree falls. Selective
tree harvesting could explain such observation for heavily
cut forest (K-15), but the uncut mature forest also
exhibited lower mortality in the large size classes. This
may suggest that intact forests have the ability to guard
against large tree falls.

For dead standing trees, the differences between size
class distribution in lightly cut and heavily cut forest, or
in lightly cut and uncut forest, were not significant. But
the size class distribution of dead standing trees in
heavily selectively cut and uncut mature forest were weakly
significantly different (Us=95 while Uso.o5[3,11]= 96.19,
STP). As with tree falls, the uncut forest showed a peak in
dead standing trees between 70 and 80 cm dbh, but in this
case the larger size classes were also well represented in
‘tree deaths with a peak between 90 and 100 cm dbh. Higher
percentages of tree deaths were also recorded for

intermediate size classes between 30 and 60 cm dbh (Fig. 9,

Appendix table 4). The heavily selectively cut forest

showed higher levels of tree deaths between 60 and 80 cm dbh
with a peak in the large size classes between 90 and 100 cm
dbh. As in uncut mature forest, the lightly selectively cut

forest had high levels of tree deaths in the intermediate

129A

Figure 9. Frequency distribution of the diameter at breast
height (10 cm dbh interval) of dead standing trees expressed
as the percentage of the density (No/ha) of living trees in
the same diameter class in lightly cut (K-14), heavily

selectively cut (K-15) and uncut (K-30) forest tracts of the
reserve.

130

can Eu E 3320 05m

OOFA 00 w. ..m cm. pm Owéh Oh. ..m cm. Fm 0m. :4 0?. pm oméN ON._.P CTN

 

J

z

,4

I
’

1x4

.1 ..4‘V‘
..4 ........... ..

\kfx... 1!

¥\\ ..4 // \

IX\

\

x
4. 4V . .4

\
\

oak Ills
m T: <.....<......4
4 3. TIT;
>5.
woos... 95:65 noon

 

SWSLS IE)Nll\l'l :IO 39V1N3083d

131

size classes between 30 and 60 cm dbh with a peak between 70
and 80 cm dbh. The first major peaks of dead standing trees
for both out and uncut forest tracts occurred between 60 and
80 cm dbh with uncut forest having the highest peak, lightly
cut forest in second place and heavily selectively cut
forest in the third position. However, the second major
peaks of tree deaths occurring between 90 and 100 cm dbh
were dominated by the heavily selectively cut forest, uncut
forest in the second and lightly cut forest in the third
position. In general, all tree mortality curves were skewed

to the left.

DISCUSSION

Comparison 9; Tree Falls in Cut and Uncut Mature Forests

The study of tree fall dynamics employed a strip census
method in order to include representative samples of the
topography and vegetation of the Central Kibale Forest
Reserve. Brokaw (1982a) used a similar method for the same
reasons, to cover as much of the representative forest
vegetation and topography as possible. Unlike other workers
who predetermined the size class ranges of trees and tree
falls to be examined (Brokaw, 1982a Putz and Milton, 1982) I
considered all tree size classes that were involved in the
falls. As in Putz gt a1; (1983), fallen trees in small size
classes (510cm dbh) as well as large canopy type trees or
"gap formers" were all considered in sampling. The small
size classes were included because of their importance in
forest regeneration and potential recovery after various
forms of natural and human disturbances.

There is extreme variability in the predominant mode of
tree deaths for different forests in the same or different
regions (Brokaw, 1982a; Putz and Milton, 1982; Putz g3 £14,
1983). The results of this study also suggest that even
within the same forest type, especially where there have

been differential management practices, variations in rates

132

133

and modes of tree mortality were extremely high (Appendix
tables 1 to 3). However, most studies, including the
present one show three prevalent types of tree death,
including tree snaps, tree uproots and death while standing
(Runkle, 1982; Putz and Milton, 1982; Putz 55 gig, 1983;
Harcombe and Marks, 1983). In this study, the predominant
modes of tree death included tree snaps for the heavily
selectively cut forest, and tree snaps, tree uproots and
death while standing for the lightly cut and uncut mature
forests.

More dead trees than live were involved in tree falls
in uncut and lightly cut but not in heavily selectively cut
forest. In the heavily selectively cut forest, mostly live
trees (67%) were involved in tree falls. In the lightly
selectively cut forest, 34.5% and 14.1% of tree falls
occurred through tree snaps and uproots, respectively. The
heavily selectively cut forest had 53% and 26.1% of the tree
falls occurring through tree snaps and uproots,
respectively, while the uncut mature forest had 22.8% and
27.8% of tree falls as snaps and uproots, respectively.

Similarly, Runkle (1982) found that of gap making

trees, 67% and 19% were tree snaps and uproots,

respectively. The damage to trapical hardwood plantations
in Western Samoa, mainly through tree snaps and uproots
accounted for only 15% (Wood, 1979). In a Fagus-Magnolia
forest in Texas, a five year sample showed 21% and 0% of the

trees that fell were snaps and uproots, respectively

134

(Harcoabe and Marks, 1983). On Barro Colorado Island,
Panama, of the fallen trees in a specified area and period,
70% were snapped and 25% uprooted (Putz 53 al;, 1983) while
in an old forest on the same island, 52% had snapped trunks
and 17% were uprooted (Putz and Milton, 1982). At La Selva
in Costa Rica, of the dead tree individuals over a 13 year
period, 26% died standing, 31% had fallen and 7% were found
buried under tree falls (Lieberman gt gig, 1985). In other
forests, however, the major mode of tree deaths was tree
uproots instead of snaps. In a tropical forest in Nigeria,
uprooting accounted for 30% of tree deaths (Jones, 1956)
while tree snaps were 27.2% (extrapolated from frequency
curves). For Tilia-Carpinus and ginus- uercus forests in
Poland, tree uproots (48%) predominated over tree snaps (7%)
(Falinski, 1978). The results of this study were in concert
with the observations of Lieberman 55 a1; (1985) that
mortality rates were independent of tree sizes 3 10 cm dbh.
However, small to middle-story trees (3 20 cm < 80 cm dbh)
were more susceptible to tree fall death than large trees 3
90 cm dbh. With the exception of heavily selectively cut
forests, this study could also add that tree falls were
independent of tree condition, dead or alive but slightly
biased towards dead trees.

0f the tree deaths involving tree snaps, tree uproots
and dead standing trees in lightly cut and uncut forest
tracts, 79.6% and 78.9%, respectively, were already dead

trees at the time of sampling. But for the heavily

135

selectively cut forest, dead tree falls and trees which died
standing formed 45.5% of total tree deaths. With the
exception of heavily selectively cut forest (54.5%), the
percentages of live tree falls for the lightly cut (20.4%)
and uncut mature forest (21.1%) were relatively low. It was
obvious that heavy selective logging removed most of the
large desirable trees from the heavily cut and lightly cut
tracts and that there was no felling in the mature forest.
Therefore, high levels of dead tree falls and dead standing
trees would only be expected for uncut mature forest.

Several studies in other tropical and temperate forests
also show variable and high amounts of trees which died
while standing (Sousa, 1984). For example, in mesic forest
in eastern U.S.A., Runkle (1982) found that 10% of the gap-
making trees died standing. In a Fagus-Magnolia forest in
Texas, U.S.A., 77% of the trees that died did so before they
fell over (Harcombe and Marks, 1983). In a Tilig-Carpinus
and Pinus-Quercus forest in Poland, of the trees that died,
45% were dead standing (Falinski, 1978). Also, in lowland
Dipterocarp forest in Malaysia, trees mostly died standing
and very few died through snaps or uproots (Putz and

Appanah, Ms). In an old forest on Barro Colorado Island,

14% of tree mortality was by trees which died while standing
(Putz and Milton, 1982) and all others through snaps and
uproots. In Costa Rica, 26% of the trees that died over a
13 year period did so while standing (Lieberman 3; a14,

1985). So the predominant modes of tree deaths, including

136

tree snaps, tree uproots and death while standing can be
variable locally and regionally, but ubiquity implies their
importance in the dynamics of forests.

The preliminary results of ongoinglong term research
(now in 13th year of observation) on tree mortality,
encompassing a larger area (42 ha) of uncut study
compartment (K-30) in the reserve show higher levels of die-
offs of some primary forest species (Struhsaker, Geither and
Kasenene--unpublished). Large, mature individuals of five
canopy tree species including N; buchananii, L;
swynnertonii, A; altissima, g; africana and M; bagshawei are
particularly involved in the die-offs. These are some of
the upper canopy species that attain heights at maturity of
from 30 m to 60 m (Eggeling, 1951) and are common to
abundant in some parts of the reserve. The mature
populations of N; buchananii and L; swynnertonii have
already had 100% and 80% mortality, respectively. The
mortality levels in A; altissima (17%), g; africana (14%)
and M; bagshawei (3.3%) are lower than in the previous two
species. The majority of dead Newtonia and nggg trees had
already fallen by the time of the study and thus made little
contribution to the estimates of tree falls and trees which
die while standing. However, the questions regarding the
cause of the massive die-offs of these important timber or
primary forest species remain unresolved and call for

another study.

137

In the reserve, the proportions of tree falls and dead
standing trees for all three compartments showed increases
and high levels from 20 to 40 cm dbh class; at which point
there was a break and no further increase until the 60 cm
dbh class. This indicates high death rate among trees less
than 40 cm dbh. This may be because of the fast growing and
short lived species of small size in cut forests or the
death of many small trees caused by large tree falls. The
short lived pioneer species such as T; guineensis and N;
macrocalyx in cut forest tracts seem to have matured and
were in senescence only 20 years after selective felling.
However, their death (most died while standing) was more
pronounced in lightly cut, thus less disturbed, forest, than
heavily felled areas where there was extensive forest
disturbance and large gaps.

In uncut and lightly cut forest, the high death rates
among trees of 60 cm dbh upwards, the mature middle-story
and emergent individuals could be attributed, mainly, to old
age and senescence. But in heavily cut areas, the high
death rates through live tree snaps may be taken as
indicating high incidences of wind damage. However, the

death rates among trees 2 60 cm dbh were irregular, perhaps

because of the small number of trees involved. The rapid
rises of death rates with increasing size of trees, in both
cut and uncut forests, suggest unstable, all-aged

populations.

138

The predominant modes of tree deaths clearly vary from
forest to forest, suggesting that many factors, in simple or
complex interactions, were in control. Variations in modes
of tree deaths are regarded as important for rain forests
because they create heterogeneous sites vital for forest
regeneration, ecological diversity and evolution (Armson and
Fesseden, 1973; Connell, 1978; Denslow, 1980 and Sousa,
1984). In the Kibale Forest, it appears that selective
felling altered the mode and rate of tree deaths in favor of
tree snaps, where more live than dead trees were involved.
For example, the annual rate of live tree falls/ha was 1.30
for the lightly cut, 3.30 for the heavily selectively cut
and 1.74 for uncut mature forest.

Selective forest felling previously has been observed
to foster more frequent and severe wind throws (Herbert and
Beveridge, 1977). The existence of large gaps or exposed
areas even in mature forest has been suggested to result in
high turnover rates for some tropical rainforests (Brokaw,
1982a; Foster and Brokaw, 1982). Denslow (1980) suggested
that human-induced changes in the size frequency
distribution of gaps by either increasing or decreasing the
rate of gap formation will result in gradual loss of
species. Human-induced changes in forest structure through
felling, for example, might account for the high rates of
live tree falls in the heavily selectively cut areas of the
Kibale forest. The few large trees which were left behind

after felling (usually the defective desirable species) were

139

exposed to the vagaries of nature and suffered heavy
mortality. This could be detrimental to forest regeneration
and recovery because excessive mortality of the few mature
desirable trees could reduce the potential seed sources in
addition to crushing young trees or creating more extensive
gaps (Longman and Jenik, 1974; Foster and Brokaw, 1982;
Kasenene, 1984).

In spite of the short period of sampling tree falls in
the Kibale Forest Reserve, mortality rate estimates
(no/ha/yr) for disturbed forest sites were comparable to
those of Malaysian forests but not forests on Barro Colorado
Island (Table 3). At La Selva, Hartshorn (1978) estimated
rates of tree fall at 1/ha/year. Brokaw (1982a) also
arrived at an estimate of one gap/ha/year which could be
taken as one canopy tree/ha/year for a forest on Barrov
Colorado Island. These rates are comparable to Kibale
forest estimates of 1.30 and 1.74 trees/ha/year for lightly
cut and uncut mature forests, respectively. Previously,
mortality rates of 1.30 and 1.40 trees/ha/year had been
established for the same forest tracts (Skorupa and
Kasenene, 1984) and are consistent with the current rates.
However, the heavily selectively cut forest exhibited higher
mortality rates (6.2 trees/ha/year) in the previous study
(Skorupa and Kasenene, 1984) and the current rate is also
high (3.30 trees/ha/year) compared to lightly cut and uncut
forests. This indicates almost constant mortality rates for

the less disturbed (uncut and lightly cut) forests but

 

 

 

 

51530 each oo.~ can Ap.~vcm.« Now n.b~ he anouou 09:00! uses:
5‘3». sash no." new Am.—vo~.u cpv m.n~ av eschew #50 Masai“;
56:00 swab cm.~ ecu An.nvm.m fie” «.m— uv econou 8:0 huu>so=
spasm:
.amvuom causal
«cad cu Na .o.mvmm ave w.~ um eschew hudlwum
MW coed“: 6:0 wasm us ca as AN.~VN~ mm v.6 do ovum oonhsaaun
11 mead Aaulmluudhz sunbeds:
uncle; uwxam
«can N_ on .m.uvaw van o.~ mm eschew husluhm
coeds: can spam as «a —v Aw.uv~n sou v.c ~o Guam poshzamqa
oca~ sa~lmnua¢53 dfinhedst
sushcot “Olcsm
«was m cm Av.nvb~ can « om amouom ado
Naa~ a pm .m.m~va «mm m cm umohom w::o>
coeds: 0:6 seam panama ouchOAOU ouuam
.mp5 .mp5 Ausch\v .sg gnu
flashes Olaaomwa .d:\m£udov some» nodules owceu
huauogvs< scavc>homno ousuo>< no .52 no .02 sou< maedo ouwm oumm

 

.00whu< diam edvcdla vac dumhsndz .v:¢~mn openedoo canon loam mvHOOOH huuddouoz OOQB .m OHDGH

141

variable rates over time in the more disturbed (heavily cut)
forest. The different mortality estimates for the heavily
cut forest in 1984 (6.2 trees/ha) and 1987 (this study, 3.30
trees/ha) could suggest that there have been drastic
reductions in the amount of susceptible trees over a short
period of time. Similarly the pattern of tree deaths on
Barro Colorado Island shows higher mortality rates for young
or disturbed forests than for undisturbed mature forest
while the converse was true for Malaysian forests (Putz and
Milton, 1982; Table 3). There were great variations in
forest areas sampled, lower size class limits for trees
studied, and the length of observation period for the
different studies of tree mortality (e.g. Hartshorn, 1978;
Brokaw, 1982a; Putz and Milton, 1982; Putz gt g;;, 1983;
Harcombe and Marks, 1983: Skorupa and Kasenene, 1984;
Lieberman gt g;;, 1985; this study). These variations might
have influenced the results differently for the different
forests thus possibly leading to questionable comparisons
and incorrect conclusions.

In the forest tracts studied, there were few trees
above 61cm gbh and very few of them were involved in tree

falls. Therefore, it was irrelevant to calculate the life

expectancy of trees using only the large size classes as has
been the practice in other studies (Putz and Milton, 1982).
I propose that the best estimates of life expectancies of
trees should be the ones that reflect the life expectancy of

the whole forest as well. So the inclusion of all tree size

142

classes that were physically involved in treefall deaths was
emphasized. After all, when a large tree falls, it brings
down with it many small to intermediate size class trees
because of their usually high density. The life
expectancies for trees in the heavily selectively cut,
lightly cut and uncut forest tracts were variable and high
(Table 3). The comparatively low expectancy of further life
(106 yrs.) for trees in the heavily selectively cut forest
could be accounted for by the low densities of trees in the
small to intermediate size classes and the high rates of
live tree falls. However, the high expectancy of further
life in lightly cut forest (361 years) could also be
attributed to the preponderance of small to intermediate
size trees and lower mortality in the large tree classes.
Although the small to intermediate size class trees were
abundant, the high mortality in the large size classes gave
the uncut mature forest a relatively higher turn over rate
(288 years) than the lightly cut forest. Again, variations
in forest areas sampled, tree size class limits and the
length of observation period might have influenced the
results from the different forests studied including Kibale,
thus limiting comparisons and conclusions. However, the
general trend was longer expectancy of further life for
trees in lightly disturbed and primary forests and a shorter
one for trees in heavily disturbed forests. Natural

variation alone may not explain the great difference in tree

mortality and life expectancy of trees in lightly and

143

heavily disturbed forest sites of the reserve. Human
disturbances through differential felling intensities were
suggested to have been the major initiator of observed
variations. The influence of heavy selective felling on the
dynamics of tree fall deaths was still strong 20 years
later.

A number of workers have observed clear cut seasonal
peaks in tree falls for other tropical and temperate forests
(Hartshorn, 1978; Sarukhan, 1978; Falinski, 1978 in Brokaw
1982a; Brokaw, 1982a). These observers attributed
seasonality in tree mortality to seasonally wet soil and
consequent loosening of roots or seasonal high winds
associated with heavy rain. We would expect high
correlations between tree falls and rainfall. However,
total tree falls (snaps + uproots), tree uproots and branch—
falls were positively correlated with rainfall only in the
lightly cut and uncut mature forest. There were no
significant correlations between rainfall and tree falls for
the heavily cut forest where tree falls seemed to occur at
high levels for all months in the year and asynchronous with
rainfall. In both cut and uncut forest compartments, tree

snaps had no significant correlations with rainfall. This

strongly suggests that factors other than rainfall and
associated wind were also influencing tree falls, especially
in the heavily selectively cut forest. If seasonally wet
soil and consequent loosening of roots are important for

tree uproots (Hartshorn, 1978; Whitmore, 1978; Brokaw, 1982;

144

Pickett, 1983; Putz gt g;;, 1983) then gusty winds are also
important for tree snaps. The high incidences of live tree
snaps in the heavily cut forest, higher rates of tree snaps
than uproots in both cut and uncut forest tracts and the low
correlations between tree falls and rainfall suggest gusty
winds to be an important control of tree falls in the
reserve. In the Kibale forest, it was common for windy
weather to precede periods of rainfall, and within the wet
season not all windy weather was necessarily followed by
heavy rain (personal observation). This could partly
explain the low correlations between rainfall and all modes
of tree falls. Brokaw (1982a) suggested that rain trapped
by leaves and associated vines, epiphytes and bark of trees
creates substantial "rain load" which could also account for
more tree snaps than uproots. A combination of factors,
including rainfall, wind and possibly rain load, may have
contributed to the variations in tree falls for the
differently managed forest tracts of the reserve. Thus as
predicted, more tree falls (including snaps, snaps and
uproots together) occurred in the heavily selectively cut
forest where scattered large trees were not buffered from
wind pressure. In lightly cut and uncut mature forest, the
presence of many large trees may have provided protection to
large and small trees against gusty winds, explaining the
fewer tree falls relative to heavily cut forest.

Several ultimate factors, including shallow rootedness,

insect and fungus damaged stems and roots, steepness of

145

terrain and landslides (Brokaw, 1982a; Putz and Milton,
1982; Putz gt g;;, 1983; Sousa, 1984) and animal damage
(Eggeling, 1947) have also been observed to strongly
influence treefalls. Shallow rootedness was common among
big tree uproots. Incidences of heart rot or hollowed boles
were not widespread among tree snaps. However, elephant
damage to trees through uproots (sapling classes) and tree
snaps (pole classes) was widespread, especially in the
heavily selectively cut forest. It has been reported that
trees on the edges of large gaps are subject to more wind
induced mechanical stresses than trees in the forest
interior. Mechanical damage suffered during earlier
treefalls and sun scald of exposed tree trunks all lead to
greater likelihood of treefalls in heavily cut forests
(Lawton and Putz, manuscript).

Another dimension to factors influencing tree snaps,
uproots or mortality rates is wood density, wood strength
and tree size (Putz gt g;;, 1983). Larger trees are more
prone to uproot than snap, and trees of high wood strength
and density would also be expected to uproot more frequently
than snap (Putz gt gl;, 1983). The pattern and level of
tree snaps and uproots in the selectively cut forests seemed
to agree with the prediction. There were more tree snaps
than uproots, possibly because of the abundance of low
density, low wood strength species characteristics of
secondary forests. However, the mature forest with high

densities of large trees and primary forest species of high

 

146

density and strength had similar levels of tree uproots and
snaps. This did not contradict the predictions of Putz gt
g;; (1983) completely if the three study areas were
considered together. There were more tree deaths through
uproots in the mature forest than selectively cut secondary
forests. One other variable introduced in my tree fall
samples was the small tree size classes. It could be that
only the large tree classes (360cm gbh) used in earlier
studies were more responsive to the effects of heavy rain
and wind and result in high correlations between treefalls
and rain fall. Evidently, human disturbance was very
important in the alteration of the seasonal rhythms of
treefall dynamics of the studied forests of the reserve.
The degree of alteration or variation in treefall dynamics

depended on the level of forest disturbance.

SUMMARY AND CONCLUSIONS

Twenty months of intensive systematic censusing of
freshly fallen trees and big branch falls in lightly cut (K—
14), heavily selectively cut (K-15) and uncut mature forest
(K-30) were undertaken for the purpose of esbablishing
whether heavy selective timber harvesting had any influence
on the dynamics of tree falls and trees which died while
standing. The forest areas sampled were estimated at
13.88ha for K-14, 13.28ha for K-15 and 17.23ha for K-30.
Censuses were conducted every second or third day of the
month following the previous counts and only new tree falls
subsequent to previous falls were enumerated.

Within lightly cut forest (K-14), the densities of dead
and live tree snaps and dead and live tree uproots and thus
dead and live tree falls were similar. However, the density
of tree snaps was higher than that of the uproots. In the
heavily selectively cut forest, tree falls were dominated by

live tree snaps. For the mature forest, the density of dead

tree uproots was higher than for live but the density of
dead and live tree snaps were similar. In contrast with K-
14 and K-15, the density of tree uproots and snaps was

similar.

147

148

Of the total tree mortality observed in K—14 (142
trees), 34.5% were tree snaps, 14.1% were uproots and 52.1%
died standing. Live tree falls formed 21.1% of the deaths
while 79.6% was by trees already dead at the time of
sampling (i.e. dead tree falls + dead standing trees).

Total tree mortality for K-15 included 134 trees where 53%
were tree snaps, 26.1% were uproots and 18.7% died standing.
However, live tree falls constituted 54.5% of tree fall
deaths while already dead trees formed 45.5%. The
partitioning of 237 tree deaths for K-30 was 22.8% tree
snaps, 27.8% uproots and 48.5% dead standing trees. The
proportion of live tree falls (21.1%) and dead trees (78.9%)
was comparable only to K-14.

The die-offs of several canopy tree species in the.
mature forest of the reserve seemed a cause for alarm and
calls for immediate research and action. For example, the
species populations of Newtonia and Egggg have already had
100% and over 80% mortality, respectively, and the causes of
the death have not been established. The high percentages
of dead tree falls or dead standing trees in K-14 or K15
could be accounted for largely by the die-offs of the short
lived pioneer species of Ttggg and Neoboutonia. They seem
to be dying off only 20 years after selective felling.
However, their deaths were more pronounced in lightly, as
compared to heavily, disturbed forest. The populations of

pioneer species in the mature forest was insignificant,

149

which implies low levels of disturbances or a preponderance
of small gaps.

Among the study plots, the rates of tree snaps and
uproots were higher in K-15 than K-14 or K-30, but K-14 and
K-30 had similar rates. Again, the rates of total tree
falls (uproots + snaps) for K-15 were much higher than for
K-14 or K-30, whereas K-14 and K-30 had similar rates.
Again, K—15 exhibited higher rates of live tree falls than
K-14 or K-30 but live tree fall rates were similar for K-14
and K-30. It appears that selective felling altered the
mode and rate of tree deaths in favor of tree snaps where
more live than dead trees were involved. Consequently, the
average life expectancy for trees in K-14 was highest (361.8
years) followed by K-30 forest trees (288.4 yrs) and K-15
trees (106.4 years).

The mortality rate estimates for the mature or primary
forests were very variable. The annual rates of live tree
falls/ha were 1.30 for K—14, 3.30 for K-15 and 1.74 for K-
30. The pattern of tree size class distribution of tree
fall deaths were also variable between compartments. In
general, the tree mortality curves for both out and uncut
mature forests were skewed to the left. It was suggested
that the damage caused by a few large trees to many small
ones during tree falls contributed to high mortality rates
in the small tree classes.

The mortality of desirable timber species in K-15 was

significantly higher than that in K-14 or K-30. Selective

\-‘\'_

150

removal of many large desirable trees from K-15 did not lead
to reductions in later tree falls but seemed to escalate it.
This implies that the few large desirable trees which were
left behind after felling, mainly because of defects, were
exposed to various natural forces and suffered heavy
mortality. This could be detrimental to the forest
regeneration and recovery since excessive mortality of such
trees would drastically reduce the seed source and
populations necessary for ample forest regeneration and
recovery. In addition, excessive falls of remnant trees
would crush the young regeneration and create more extensive
gaps.

Natural variation alone could not explain the great
variations and differences in the mode of tree deaths, tree
mortality rates and life expectancy of trees from the
forests which were differentially managed 20 years earlier.
Human disturbances through differing felling intensities
were suggested to have been the major initiator of observed
variations upon which the proximate and ultimate natural
forces acted to produce current differences and variations

in tree mortality 20 years later.

CONCLUSION

Can rates of natural tree falls help guide us in the
management of natural forests? Except for disturbed forest
sites, the natural rate of tree falls for primary and mature
tropical forests so far reported range from one tree/ha/year
to 5.8 trees/ha/year. For the reserve, the number of dead
standing trees/ha was 4.0 while the rate of tree falls for
the lightly selectively cut forest was 1.30 trees/ha/year.
The lightly selectively cut forest looked less disturbed and
maintained low mortality levels over the years. With
respect to variations among the differently managed forests
in the reserve, an average harvest intensity of 3 to 4
canopy trees/ha is suggested and could be compatible with
minimizing damage to the forest structure and composition in
Ugandan forests in general and the Kibale forest in
particular. This would simulate the natural rates of tree
falls which constitute an important integral component of

the mature forests' dynamics.

151

LITERATURE CITED

Armson, K. A. and R. J. Fesseden 1973. Forest wind throws
and their influence on soil morphology. Soil Sc. Soc.
Amer. Proc. 37: 781-784.

Brokaw N. V. L. 1982a, Tree falls: frequency, timing and
consequences. In: The ecology of a tropical forest:
Seasonal rhythms and long term changes. pp. 101-108
(Eds. E. G. Leigh Jr. A. S. Rand and D. M. Windsor).
Smithsonian Institution Press, Washington DC.

1982b The definition of treefall gap and its
effect on measures of forest dynamics. Biotropica
14:158—160.

 

1985a. Treefalls, regrowth and community
structure in tropical forests. In: The Ecology of
Natural disturbance and Patch dynamics. pp. 53-69 (Eds.
S.T.A.Pickett and P. S. White) Academic Press, New
York, New York.

 

1985b Gap phase regeneration in a tropical
forest. Ecology 66(3) 682—687.

 

Burgess P. F. 1970. An approach towards a silvicultural
system for the hill forests of the Malay Peninsula
Malay Forest, 33:126-134.

Connell, J. H. 1978. Diversity in tropical rainforests and
coral reefs. Science 199:1302-1310.

Denslow J. S. 1980. Gap partitioning among tropical
rainforest trees. Biotropica 12 (Suppl.) 47-55.

Eggeling W. J. 1947. Observations on the ecology of the
Budongo rainforest, Uganda. J. Ecol. 34:20—87.

Eggeling W. J. and I. R. Dale 1951. The indigenous trees of
the Uganda protectorate. 2nd Edition. The Govt.
Printer Entebbe, Uganda.491 pp.

Falinski, J. B. 1978. Uprooted trees, their distribution
and influence in the primeval forest biotope.
Vegetatio 38:175-183.

152

153

Foster, R. B. and N. Brokaw 1982. Structure and history of
the vegetation of Barro Colorado Island. In: The
Ecology of a tropical forest. pp. 67-81. (Eds. E. G.
Leigh Jr. A. S. R. and D. M. Windsor) Smithsonian
Institution Press, Washington DC.

Gomez-Pompa, A.; C. Vazques—Yanes and S. Guevara 1972. The
tropical rainforest: A non-renewable resource. Science
117:762-765.

Hamilton, A. C. 1981. A field guide to Uganda forest trees.
(Privately Published).

Harcombe, P. A. and P. L. Marks 1983. Five years of tree
death in a Fagus-Magnolia forest, Southeast Texas,
U.S.A. Oecologia 57:49-54.

Hartshorn, G. S. 1978. Treefalls and tropical forest
dynamics. In: Tropical trees as living systems. pp.
617-683. (Eds. P. B. Tomlinson and M. H. Zimmermann).
Cambridge University Press, Cambridge.

1980 Neotropical forest dynamics. Biotropica
12(suppl.):23-30.

 

Herbert, J. and A. E. Beveridge 1977. A selective logging
trial in dense podocarp forest in Central North Island,
New Zealand J. of Forestry 22:81-100.

Jones E. W. 1956. Ecological Studies on the rainforest of
southern Nigeria, IV: The reproduction and history of
the Okuma Forest Reserve, Part II. The reproduction
and history of the forest. J. Ecol. 44:83—117.

Kasenene, J. M. 1984. The influence of logging on rodent
populations and the regeneration of selected tree
species in the Kibale Forest, Uganda. Tropical Ecology
25:179-195.

Kingston, B. 1967. Working plan for the Kibale and Itwara
Central Forest Reserves. Uganda Forest Department,
Entebbe. 142 pp.

Lawton, R. 0. and F. E. Putz (Manuscript) Elfin forest
dynamics in Monte Verde, Costa Rica.

Leigh, E. G. 1982. Why are there so many kinds of tropical
trees? In: The Ecology of a tropical forest: Natural
rhythms and long term changes. (Eds. E. G. Leigh, A.
S. Rand and D. M. Windsor) Smithsonian Institution
Press. Washington DC.

154

Lieberman D., M. Lieberman, R. Peralta and G. S. Hartshorn
1985. Mortality patterns and stand turnover rates in a
wet tropical forest in Costa Rica. J. Ecol. 73:915-924.

Liew, T. C. and F. 0. Wong 1973. Density, recruitment
mortality and growth of dipterocarp seedling in virgin

and logged over forest in Sabah. Malay Forester 36:3-
15.

Longman, K. A. and J. Jenik 1974. Tropical forest and its
environment. Longman Group Ltd. London 196pp.

McIntosh, R. P. 1961. Windfalls in forest ecology. Ecology
48:834.

Meijer, W. 1970. Regeneration of tropical lowland forest in
Sabah, Malaysia, forty years after logging. Malay
Forester 33:204-229.

Pickett, S. T. A. 1983. Differential adaptations of
tropical trees species to canopy gaps and its role in
community dynamics. Tropical Ecology 24(1):69-84.

Putz, F. E. 1983. Treefall pits and mounds, buried seeds
and the importance of disturbed soil to pioneer trees
on Barro Colorado Island, Panama. Ecology 64:1069-
1074.

and S. Appanah (manuscript) Buried seeds, newly
dispersed seeds and the dynamics of a lowland forest in
Malaysia.

 

, P. D. Coley, K. Lu, A. Montalvo and A. Aiello,
1983. Uprooting and snapping of trees: Structural
determinants and ecological consequences. Can. J. For.
Res. 13:1011-1020.

 

Putz, F. E. and K. Milton 1982. Tree mortality rates on
Barro Colorado Island, Panama. In: The ecology of a
tropical forest: Seasonal rhythms and longer term
changes pp. 95—100. (Eds. E. G. Leigh Jr. A. S. Rand
and D. M. Windsor) Smithsonian Institution Press,
Washington D. C.

Richards, P. W. 1952. The tropical rainforest. Cambridge
University Press, Cambridge. 450pp.

Richards, P. W. and G. B. Williamson 1975. Treefalls and
patterns of understory species in a wet lowland
tropical forest. Ecology 56:1226—1229.

Runkle, J. R. 1982. Patterns of disturbance in some old
growth mesic forests of Eastern North America. Ecology
63:1533-1546.

 

155

Schenske, D. W. and N. Brokaw 1981. Treefalls and the
distribution of understory birds in a tropical forest.
Ecology 62:938-945.

Sarukhan, J. 1978. Studies on the demography of tropical
trees. In: Tropical trees as living systems, pp. 163—
184, (Eds. P. B. Tomlinson and M. H. Zimmermann).
Cambridge University Press, Cambridge.

Siegel S. 1956. Non parametric statistics for the
behavioral Sciences. Mcgrow Hill, New York.

Skorupa, J. P. and J. M. Kasenene 1984. Tropical forest
management: Can rates of natural treefalls help guide
us? Oryx, 18:96-101.

Sokal, R. R. and F. J. Rohlf 1981. Biometry. W. H. Freeman
and Co. San Francisco. 859 pp.

Sousa, W. P. 1984. .The role of disturbance in natural
communities. Ann. Rev. Ecol. Syst. 15:353-391.

Whitmore, T. C. 1975. Tropical rainforests of the Far East.
Clarendon Press, Oxford 282pp.

1978. Gaps in the forest canopy. In Tropical
trees as Living Systems. pp 639-655. (Eds. P. B.
Tomlinson and M. H. Zimmermann) Cambridge University
Press, Cambridge.

 

Wood, T. W. W. 1970. Wind damage in the forest of Western
Samoa, Malaysia For. 33:92-99. ‘

Wyatt-Smith, J. 1966. Ecological Studies on Malayan forest.
Composition of and dynamic studies in lowland evergreen
rainforest in two 5 acre plots in Bukit Lagong and
Sungei Menyala forest reserve. 1947-1959. Research
pamphlet No. 52, Kepong, Malaysia. Forest Research
Institute, Forest Department.

156

 

 

 

O«. vN. ON. OO. O «N. NO. ON. NO. ON. NO. Om. on. «N. v«. ON. NO. Nv. NO. v«. NO. DO. ON. «690%
v«. n«. «N. OO. O NO. O NO. NO. v«. NO. NN. NO. NO. v«. v«. O .«N. NO. v«. NO. O0. v«. teen _ Ocean:
««. ««. NO. ON. O v«. NO. NN. O NN. O ON. ON. v«. O v«. NO. «N. O O O NO. vO. 0>«a + Baum

O«. OO. v«. ON. O O O O«. O O O ON. O NO. NO. «N. O O O O O NN. v«. «Ouch
OO. vO. NO. v«. O O O O O O O O«. O NO. NO. NO. O O O O O O«. O 6609 Ocean:
OO. vO. NO. O«. O O O O«. O O O v«. O O O v«. O O O O O NO. v«. o>«a 0099

v«. N«. v«. OO. O «N. NO. v«. NO. On. NO. «N. on. v«. NO. NO. NO. Nv. NO. v«. NO. Ov. vO. «Bach
O«. O«. v«. NN. O NO. O NO. NO. v«. NO. NO. NO. O NO. NO. O «N. NO. v«. NO. «v. v«. Udon Adam
OO. NO. O v«. O v«. NO. NO. O NN. O v«. ON. v«. O O NO. «N. O O O O OO. o>«A cosh
v«. O«. v«. v«. O NO. NO. O NO. v«. «N. NO. v«. «N. ON. 0O. ON. «N. «N. NO. NO. v«. vv. «0009 O««cu
O«. NO. NO. NO. O O NO. O NO. NO. v«. O O NO. v«. ON. ON. «N. v«. NO. NO. O NN. 660: nocshn
OO. NO. NO. NO. O NO. O O O NO. NO. NO. v«. v«. v«. ON. O O NO. O O v«. NN. o>«a O«O
NO X can but un< hi! son :06 00a >02 300 show I5< «an :58 ha: an< as! non :In con >02 80° AOGGHA co«vdlu0h
I no can» AGO
OOO« OOO« vOO« no ocean no ova!
..v«uuv anchou #50 h«o>«aoo«00 h«u:‘«« :« O««du socshn tad och» N0 Ad£\.oz. baflmcov One :o«a:n«uua«v N«£uco: .« O«adb

x«v:onm<

157

 

 

 

an. «v. O«. O OO. «O. vO. NN. vN. OO.« «O. «O. ON. O«. O«. vO. ON. vN. NN. me. O«. OO. OO. «such
OO. O«. O«. O O nN. ON. OO. OO. OO. OO. ON. OO. OO. OO. OO. 0N. O«. O«. OO. O«. OO. OO. icon voouns
«O. ON. O O OO. OO. «O. O«. O«. ON.« ON. ON. OO. OO. OO. Ov. no. OO. «O. ON. ON. O O o>«a + Beam

N«. v«. OO. O O O ON. ON. OO. OO. O«. ON. «a. O O O«. O«. OO. «O. OO. O OO. O «sacs
NO. OO. OO. O O O O«. OO. O O OO. O«. OO. O O O OO. OO. OO. OO. O OO. O econ uooun:
OO. OO. O O O O OO. O«. OO. On. OO. OO. ON. O O O«. OO. O NN. OO. O O O o>«a some

ON. ON. OO. O OO. «O. «n. O O«. OO.« O«. OO. ov. O«. O«. OO. OO. O«. Ov. O«. OO. O OO. «ouch
OO. NO. OO. O O ON. OO. O OO. OO. O OO. O OO. OO. OO. O«. OO. OO. O O«. O OO. teen neon
«N. «N. O O OO. OO. NN. O OO. OO. O«. On. Ov. OO. OO. O». mv. OO. OO. O«. ON. O O o>«a soak
O«. «N. O OO. O«. ON. OO. O O«. On. O«. «N. OO. O«. OO. O«. «n. O«. NO. ON. O«. OO. O«. «each a««su
N«. ««. O O OO. ON. O O O O OO. OO. On. OO. OO. OO. NN. OO. O«. O«. O OO. O«. 0009 access
OO. O«. O O OO. O OO. O O«. OO. OO. O«. ON. OO. NN. OO. OO. OO. O«. OO. O«. O O o>«q O«O
um N 65% he! 99¢ no: son can con >02 coo deem O5< «on can he: un< as: non can 00: >02 000 socaua co«aslhom
I no Dona moo
OOO« OOO« vOO« no oasaO no 06o:
..O«:Iv eachou e30 N«o>«aoo«oe h««>so: :« e««su acorns use sous no .d£\.ozv bu«moov One :o«a:£«uom«v N«sacot .N O«OdN

x«vcoon<

 

158

 

 

O«. OO. ON. NO. OO. Ov. «O. OO. OO. vN. N«. Ov. O N«. O». Nv. Nv. NN. OO. OO. vN. NO. OO.« «duos
O«. «N. ON. OO. ON. Ov. vn. O N«. N«. OO. vN. O N«. N«. O«. «v. OO. N«. O«. O«. ON. OO. OGOO aooun:
««. O«. OO. N«. OO. OO. N«. OO. «v. N«. OO. vN. O O vN. vN. OO. OO. O«. O«. OO. ON. ON. o>«a + Beam

O«. ON. N«. ON. N«. ON. N«. O vN. N«. N«. OO. O OO. O«. O«. OO. vO. «N. O«. OO. v». ON. «0808
OO. O«. N«. N«. N«. ON. N«. O N«. N«. OO. O«. O OO. OO. OO. ON. On. OO. N«. OO. N«. Ov. anon OOOHAS
NO. NO. O N«. O O O O N«. O OO. N«. O O N«. N«. OO. vN. N«. OO. O N«. ON. o>«A ovhb

OO. O«. «such
NO. OO. OO. ON. N«. N«. N«. O O O O OO. O OO. OO. N«. N«. OO. OO. OO. N«. N«. N«. van: nose
NO. OO. OO. O OO. OO. N«. OO. ON. N«. O N«. O O N«. N«. O N«. OO. N«. OO. OO. Ov. o>«a Gone
OO. N«. O«. O«. vN. OO. N«. O N«. ON. O«. ON. OO. OO. vN. O«. ON. ON. O«. OO. O«. ON. ON. «6009 m««su
OO. N«. O«. N«. N«. OO. N«. O OO. ON. O«. ON. N«. O N«. O«. ON. OO. OO. O O«. N«. ON. Odoa season
OO. OO. O OO. N«. O O O OO. O O O OO. OO. N«. O O N«. N«. OO. O N«. O o>«a O«O
XO N can has hn< but non can 00° >02 coo anew I=< «an :55 but ua< as: no. can con >oz aoo nocmun :o«adlhom
I go ocuO ado
OOO« OOO« vOO« No oasum No 060:
..Onlfl. unchou ensue! 830:: aw m««sN cocsun val coma No Asg\.oz. Nu«mcov vac ao«»=o«hum«v N«Aacot .n O«Osb

x«v:0nn<

159

 

 

 

vm.O O0.0 N«.O vN.O «v.o N0.0 N0.0 NO.« OO.« O0.0 O«.O mean + access
N«.O O0.0 N«.O N«.O N«.O O«.O O«.O O«.« O0.0 Ov.O O0.0 noose: cone
N«.O O0.0 O0.0 N«.O ON.O N«.O N0.0 N0.0 ON.O NN.O N«.O Adam cone Om1O
N0.0 N«.O N0.0 ON.O «m.o O«.O O0.0 v0.0 Ov.O Ov.O m0.0 mc«vsmam Odom
O«.O O0.0 O«.O O0.0 Ov.O NN.O «0.0 mo.« «N.« OO.« OO.« mean + poouan
O0.0 O0.0 O0.0 O0.0 O«.O O0.0 NN.O mv.o O0.0 O0.0 Ov.o noose: owns
O0.0 O0.0 O0.0 O0.0 O«.O O«.O O«.O O0.0 m«.« OO.« mv.« Assn oops O«um
O0.0 O0.0 O0.0 O0.0 O0.0 O«.O N0.0 O«.O ON.O O«.O N0.0 unacceam coon
NN.O O0.0 O0.0 «N.O N0.0 N0.0 NN.O mv.O O0.0 Om.« NN.O mean + uooumO
O0.0 O0.0 O0.0 N0.0 O0.0 N0.0 ON.O v«.O NN.O av.O NN.O uoonm: oops
NN.O O0.0 O0.0 v«.O N0.0 O0.0 Ov.O ON.O 40.0 O«.« O0.0 nmcm some v«um
v0.0 v0.0 O0.0 O0.0 O0.0 O0.0 Ov.O ON.O v0.0 NN.O NN.O meaucsum Odom
OO«A OO«I«O OO:«O OO:«N ON:«O OOIOO Omn«v Ovn«m Omu«N ON:«« O«IN .snv so as momms«o ou«m
.«Omlmv umouom ou=¢ma
#30:: can «O«Ix .v«lmv £30 h«0>mvoo«om :« moon» unavamvm 6606 6:6 muoouas Oman .mnacm

Beam me Ad£\.ozv Newmnmv Oo mauve cw mocha mo Gowpzn«hamwv mmd«o ouwm .v manna

KNOGOAQ<

PART III

POST LOGGING PLANT DYNAMICS IN FOREST GAPS

INTRODUCTION

Forest Gaps and Forest Regeneration

Gap creation and filling has been recognized as a major
influence on community structure and species interaction.
Many moist or rain forest species require or are stimulated
by the gaps (Richards, 1952; Gomez-Pompa gt g;;, 1972;
Whitmore, 1975; Denslow, 1980; Brokaw, 1982a; Pickett, 1983;
Putz, 1983; Uhl gt g;;, 1985). The ecological effects of
gaps can be related to their structure, mode of formation
and size (Richards, 1952; Whitmore, 1975; Denslow, 1980 and
Pickett, 1983). Major causes of gaps and larger
disturbances in tropical forests include physical and
biological factors such as landslides, gusty winds, storms
and lightning, animal activity, death of trees, selective
logging (mechanized and pitsawing), agricultural
encroachment and human settlement, firewood collecting and
charcoal burning (Longman and Jenik, 1974; Whitmore, 1975;
IUCN, 1975; UNESCO, 1978; Hartshorn, 1978, 1980; USAID.
1978; Brokaw, 1982a; Putz and Milton, 1982; Putz, 1983a, Uhl
_t gt;, 1984, 1985; Dittus, 1985; Lieberman gt gt;, 1985 and
Pimentel gt g;;, 1986).

In Ugandan tropical moist forests, the major

disturbance factors include agricultural encroachment,

160

161

uncontrolled timber exploitation for timber and firewood.
energy, construction and trade and rampant poaching of game.
These problems are bound to increase in light of the ever
increasing human population growth rate, currently estimated
to be between 3.2% and 3.5% per annum (Hamilton, 1984). The
omnipresent social and economic problems, poor land use
practices and lack of sound management of natural resources
such as forests (Hamilton, 1984; Van Orsdol, 1983) are bound
to aggravate the situation.

In Uganda as elsewhere in the tropical forest regions,
there appears to be no clear solution to the increases in
tropical forest disturbances that accompany the demands for
more tropical forest land and forest products, both of which
are directed by the human population problems (Richards,
1964; Brenan, 1973; Fosberg, 1973; Chapman, 1975; Orington,
1975; Brazier gt g;;, 1976 and USAID, 1978). Amidst these
challenges, there is great lack of knowledge regarding the
minimum disturbance levels that can be allowed during forest
exploitation without destroying the integrity of the
structure and function of the tropical forest for sustained
use (TIE Report, 1973; Downes, 1974 and IUCN, 1975).

Richards and Williamson (1975) and Webb gt gt; (1972)
regard forest disturbances and gaps, both naturally and
human generated, as dynamic patches of forest regeneration
and recovery. However, the disruption of tropical
rainforest composition, structure and function through a

range of disturbance regimes has often been observed to lead

162

to impoverished secondary forests (Gomez-Pompa gt gt;, 1972;
Hartshorn, 1980; Uhl gt g;;, 1984 and Odum, 1985).
Consequently, tropical rainforests were appropriately
described as "fragile ecosystems" sensitive to disturbance
(TIE Report, 1973). Once disturbed, the only hope of their
regeneration and some recovery depends on either coppicing,
regeneration from soil seed banks and established seedlings,
or recolonization from outside the disturbance site
(Richards, 1952; Whitmore, 1975; Gomez-Pompa gt gl;, 1974;
Ewel gt gt;, 1981; Uhl gt gt;, 1982b; Putz, 1983; Uhl gt
gt;, 1985, Young, 1985; Marquis gt gt;, 1986). However,
Richards (1964) and Whitmore (1975) observed that seedlings
and saplings of more extreme shade tolerant species do not
usually survive in extensive gaps and that seedling
establishment and survival was an inverse function of gap
size.

In Uganda, commercial timber harvests involve a
mechanized polycyclic system (Whitmore, 1975) where the
forest canopy is supposed to be opened up by 50% (Kingston,
1967; Uganda Forest Dept. Records). In the Kibale Forest,
the northern 1/3 of the central block where this research
was conducted was mechanically selectively felled between
1968 and 1969 creating giant gaps in the forest canopy.
However, the southern 2/3 is still intact mature forest with
small to medium gaps created by natural tree falls. There
is a paucity of information on the impact of mechanized

polycyclic felling systems on the structure, composition,

163

function and the regeneration of the resultant secondary
forest. However, in other forests, a number of workers have
observed significant reductions in species diversity of
seedlings and saplings following selective forest felling
and equate the reductions to ecological instability (Odum.
1963, 1969; Westhoof, 1971; Herbert gt g;;, 1977; Poore,
1976).

My hypothesis was that the level of forest disturbance
and gap size were inversely related to tree sapling and pole
abundance, diversity, and the general regeneration success
of desirable timber or canopy tree species. Therefore the
study of the level of forest regeneration in natural tree
fall gaps and comparison with the regeneration in human
induced gaps should help us understand the effects of
mechanized selective logging on species rich tropical moist
forest and the requirements for regeneration of commercially
desirable or canopy tree species. Our understanding of the
dynamics of tropical moist forest regeneration under a range
of disturbance regimes, including natural tree fall
disturbances, should also help us in the establishment of
minimum disturbance levels the forest could sustain and yet
maintain its natural function. Development of appropriate
management plans for sustained productivity of existing
forest and success in improving the regeneration of already
heavily exploited and disturbed areas may depend heavily on
the knowledge of the dynamics of forest regeneration after

various intensities and modes of disturbances.

164

This study was designed to determine the ecological
effects of forest gaps, natural and human induced, on the
floristic composition, abundance and regeneration of
saplings and poles of desirable timber or primary forest
species and important non-desirable tree species in a
tropical moist forest. Secondly, the relative contribution
of coppicing and sprouts, soil seed banks, and seed rain

from outside the disturbance sites was to be evaluated.

METHODS

Field work began in August 1984 with a survey of the
four forest compartments (K-13, K-14, K-15 and K—30), each
of which had had a different history of management (see part
I). All four compartments were located in the central part
of the Kibale Forest Reserve near Kanyawara Forest Station.
In the lightly cut (K-14) and uncut (K-30) mature forests,
the existing system of trails running south-east and north-
west were walked searching for gaps within 10 to 20m of the
trail. For the heavily felled forest (K-15), the 100m
interval system of trails running east and west were walked
and all discernible gaps within 10 to 20m of the trail
marked and mapped. The heavily selectively cut and treated
forest (K-13) had no trail grids. A grid system of 100m x
200m was established in late August, 1985 covering an area
of 30 ha and a linear distance of 5.8 Km.

The entire linear distance of the grid was walked and
all recognizable gaps marked and mapped. For the
selectively cut forests (K-13, K-14 and K-15), the gaps with
cut tree stumps and/or waste tree trunks were marked. In
uncut mature forest (K-30), gaps with tree or big branch
falls were considered for marking, mapping and later

selection. This study concentrated on late gap phase-early

165

166

building phase (see Whitmore, 1975; Hartshorn, 1980 and
Brokaw, 1985) stage of forest gap succession. This stage
was characterized by young trees, mostly shade tolerant
(since the majority of sapling species in the gaps also
appeared in the adjacent forest understory), and a few shade
intolerant tree species.

Gap ages for K-13, K-14 and K-15 were deduced from the
time the forest trees were cut. However, for K-30, the
natural forest gaps which had characteristic gap vegetation
species of size-ranges similar to those in selectively cut
forest gaps were selected for sampling. A total of 52, 51
and 56 gaps were identified and marked with identification
numbers in K-13, K-14 and K-15, respectively, whereas 86
gaps were marked in uncut mature forest (K-30). A table of
random numbers was used to select 40 gaps per forest
compartment which were measured, mapped and permanently
marked with metal tags bearing identification numbers and
date of first sampling (Figures 1 a 2). These were
designated for intensive two—year and longer term study.
All new gaps encountered after the selection of the 40
study gaps in each forest compartment were also marked with
identification numbers and dates of first observation and
mapped for future studies.

The size (in area, m2) of a selected gap was determined
by a combination of two methods: a) visual estimation of
the edges of the gap based on such cues as presence and

extent of light demanding vegetation, e.g. Brillantaisia

166A

Figure 1. Approximate location of the 40 randomly selected
forest gaps per forest compartment used for gap vegetation

studies in lightly selectively cut (K-14) and uncut (K-30)
forest tracts of the reserve.

167

2225 53::

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(0

1m snsuaa "

 

 

 

 

 

 

 

 

 

 

 

167A

Figure 2. Approximate location of 40 randomly selected
forest gaps per compartment used for gap vegetation studies
in the heavily selectively cut forest (K-15) and heavily
selectively cut and treated forest (K-13) in the reserve.

 

7E

56E
14E

44AE

1653

 

 

 

 

 

 

 

 

 

 

 

 

 

 

29 27
30 (p 28 g
26
\.
22 \
2
25
21 ’4 20
l
40
17 1
12 13 16
1s """
N
51" 14

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

19
liAE

 

300

.200

70

 

 

 

53$; g 6;???“
Eucnmws‘” "
PLANTATION

 

- 3
- - " e 5" a
‘ ' \fix‘ ' v
% g l _ A r o
1‘1: 1 " ’nJ‘ v ’

 

 

169

nitens, Mimulopsis solmsii and gtgpg guineensis, which
often filled the gaps to the forest/gap interface; b) light
meter readings at random points in the forest surrounding
the gap. The mean of 10 light readings was taken as
representative of the surrounding forest. The interface
between gap edge and forest where the gap light declined
dramatically to almost surrounding forest levels was marked
as the gap edge. Light readings were taken between 11:30 am
and 2 pm on days when the sky was relatively clear. Trails
were then cut in the gap or the tape measure was pulled
through the gap vegetation from a fixed position to the gap
edges and measurements taken. Access trail cutting depended
on the shape of the gap. If the gap was circular, at least
eight radii were cut from the fixed position in the center
to the forest edge. If it was long and wide, a base line
trail was out following the longest axis of the gap and
other trails crossed it at right angles ending at the forest
edge. The size of the gap (area in m2) was determined from
scale drawing of gap configuration on graph paper.
ggp ttgg Sapling gpg Pgtg Enumeration

After estimation of the gap size, an attempt was made

to sample 40% of the gap area for saplings and poles using

the trails cut in the gaps. A five meter wide strip along
selected trails (2.5 m either side of trail) was sampled.
All tree saplings and pole class trees (range 31.5cm to 10cm
dbh) encountered in the sample strip were measured (dbh, cm)

and identified to species following Eggeling 1951 and

170

Hamilton 1981. Voucher specimens of unfamiliar tree
saplings and poles were collected, labeled and pressed for
future identification at, and deposition with, the Makerere
University Herbarium. The effect of logging on forest
sapling and pole regeneration was studied by comparing
species richness and abundance in cut forest gaps with the
same parameters in uncut mature forest gaps. The effect of
logging intensity on tree sapling and pole regeneration was
also studied by comparing species richness and density of
saplings and poles in forest gaps of similar sizes. The
effect of post felling treatment with arboricides was
studied by comparing species richness and density of
saplings and poles in gaps of similar sizes in the two
heavily felled forest tracts (K-13 and K-15), one of which
(K-13), was treated with arboricides. The effect of gap
conditions on forest sapling and pole regeneration was
studied by comparing sapling and pole species richness and
density in gaps and adjacent surrounding forest. For each
of the 20 randomly selected gaps (out of the 40 gaps/
compartment) four 10m x 10m quadrats were established
directly opposite the two major axes that crossed the gap
perpendicularly in the middle. The quadrats were located 7m
to 10m into the forest. Data collection was as described
earlier for gap sapling and pole enumerations.
Qgp Ground Vegetation Cover vc.

Within each gap, two 4m2 quadrats were established and

marked with metal tags for long term sampling. Each quadrat

171

was subdivided into four 1m2 subplots. All living plants
rooted within each of the four 1m2 subplots were identified
and their number and frequencies recorded. Visual estimate
of the percentage GVC by plants under 1.3m height was made.
Voucher specimens of unfamiliar plants were collected,
labeled and pressed for subsequent identification at the
Makerere University Herbarium. The mean of at least three
gvc recordings, 2 dry season (January and June) and one wet
season (October) was obtained.

Thirty marked gaps in forest compartments K-14, K-15
and K-3O were visited at the end of each month to examine
animal disturbance and intensity of browsing on tree
seedlings, saplings and poles. All ninety gaps were sampled
on the same day. Sampling involved recording of the
presence of incursions and the nature of vegetation
disturbance, e.g. trampling, browsing, snapping, uprooting
or barking, by tree species.

Seed predation trials on four seed species which could
be obtained in large numbers were conducted. Twenty to 30
marked gaps in each forest compartment were used for seed

placement. Two seed placement sites, one in the gap center

and other at the gap edge, were located in each gap. The
number of seeds per placement or pile varied depending on
seed size and abundance of available seeds. Only one tree
seed species was used per predation trial. The experiment

was run until 80% or more of the seeds had been eaten or had

172

become infested with fungus. Recordings for all three
forest tracts were done on the same day. Recording involved
the number of days elapsed since placement, number of seeds
eaten and number remaining per pile. Then the rates of seed

disappearance or predation were calculated using the

 

relationship:
E
r:
Pts
Where r = rate of seed predation (seeds/pile/day)
E = total seeds eaten
P = total number of seed placements
t = time (days) since day of placement and
s = number of seeds per pile.

Sources pt Egant Recrpttpent tnto Forest 9gpg

Soils were collected from forest gaps (gap soil) and
from under closed or semi-closed forest adjacent to the gap
(forest soil) in uncut and selectively cut forest
compartments. Samples of the top 10cm of surface soil with
bits of litter were collected from 1m2 plots. The soil was
transferred to screened nursery beds fixed on a raised stand
( 1.3m above ground) to avoid contamination and seed or
seedling predation. Two soil samples from K-14 (one from
forest and other from gap) were each thinly spread on a flat
iron sheet and thoroughly heated over a fire to sterilise
them. These were also placed in the screened nursery bed.
The littered surface soil and sterilized soils were spread
thinly on top of soil removed from below 10cm depth of each
sample. Two replicate trials per gap and adjacent forest

soil were run. Each germination trial included 14 soil

173

samples, seven from the forest gaps and a corresponding
seven from the forest surrounding the gaps. Fiberglass
netting of 1mm x 1mm mesh was used to screen the nursery.
The nursery was then lightly thatched with young elephant
grass to reduce exposure of the soil to excessive
insolation. The plots were watered every evening except on
rainy days.

Two 4m2 plots (plant regeneration plots = PRP) were
also established in 20 of the selected forest gaps in each
forest compartment to examine the contribution of buried
seeds and seed rain to forest regeneration. The four

treatments of the 1m2 quadrants in the 4m2 PRP were:

a) 1m2 subplot with vegetation removed

b) 1m2 subplot with vegetation removed and screened

c) 1m2 subplot with vegetation and surface soil removed

d) 1m2

subplot with vegetation left intact.

Monthly checks of the nursery and PRP and counts and
classification of seedlings germinated from buried seeds or
seeds dispersed into the gaps were made. Unfamiliar plants
were tagged with identity numbers and left to mature for
easier identification. For each forest or forest gap plot
sampled, two soil seed germination trials were conducted.
One was begun in December of 1984 and run through June of
1985. The second one was set up in late June, 1985, and

also ran for seven months. Each trial ran for nearly seven

months, under the assumption that the majority of seeds

174

should have germinated by then. However, 20 months of
observation on the PRP in forest gaps were accomplished.
Mpgg gt Itgg Sapling gpg ggtg Eggengggtion ;_ Qgpg

A total of 25 forest gaps was sampled in each forest
compartment. Five meter wide strips following the two major
or longest axes that crossed the gap perpendicularly in the
middle were sampled for tree saplings and poles which were
either growing normally (N), coppicing (C) or stem sprouting
(SS). The saplings and poles which had no signs of breaks
or bark injury were classified as normal (N). Where the
main stem was doubly bent, dry or with signs of
deterioration while bent, and a dominant side branch taking
off from the stem, the sapling or pole was designated stem
sprout (53.). In coppices, the main stem was snapped or
broken with sprouts taking off from the stock. The number
of breaks and coppices per desirable species sapling or pole
stem was also counted. No attempt was made to determine the

contribution of root sprouts to gap regeneration because

destructive sampling was prohibited in the reserve.

RESULTS

Natural and Artificial Forest Gaps

The non-parametric multiple comparison by STP
(Simultaneous Test Procedure of Dwass, Sokal and Rohlf,
1981, p. 437) was used to analyze gap sizes of uncut and
selectively cut forest tracts (see Appendix table 1, 2, 3
and 4). The calculated critical value for 00.05[4v°V] was
1066.7. Except for the forest gaps of pair comparison K-13
Vs K-15 (U=880), the forest gap sizes for all other plot
pair comparisons were significantly different (all U21135).
As expected, the heavily selectively felled forest (K-13 and

K-15) exhibited the highest gap size means (938m2 and

1307m2, respectively) and ranges (2271112 to 3313m2 and 73m2
to 7100m2, respectively). The lightly cut forest tract (K-

2

14) showed a lower mean (467m2) and range (75m to 1800m2)

while the naturally formed gaps in uncut mature forest (K-

2

30) had the lowest mean (256m2) and range (100m to 663m2).

Figure 3 clearly shows that only a small percentage of

gaps, 55% and 40%, in the heavily cut forests of K-13 and K-
15 respectively, were within the 200m2 to 800m2 gap size
class range. However, the greater percentage of forest
gaps, 90% and 100%, in lightly cut (K-14) and uncut (K-30)

forest tracts, respectively, were within the same gap size

175

175A

Figure 3. Size class distribution of 40 randomly selected
gaps in uncut mature forest (K-30), lightly selectively cut
forest (K—14), heavily selectively cut forest (K-15), and
heavily selectively cut and treated forest (K—13) tracts of
the reserve.

 

176

«wE OON “mosses. «0320 35 now

OOON A OOON OOVN CONN OOON COOP COO FOO»: CON _. 000—. GOO 000 8:0

 

 

EEHUIEEEEHI

 

 

 

ONO—D
3%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EB

 

 

 

 

 

 

 

. 212%2... 1.
2. .
m .2
. 2m.
o
\
m...
=. .
one. -. .O_.
3.:V\
& .8
222
ms

 

SdVE) :lO HHSINflN

177

class range. The gaps in K-13, K-14, and K-15 were a result
of multiple tree removals during selective felling and
consequent wind and storm throws. The gaps in uncut forest,
K-30 were mainly due to tree falls caused by wind and/or
heavy storms. Consequently, the majority of forest gaps in
the heavily felled forests were larger than those in the
lightly cut and uncut forest tracts. Therefore, selective
forest felling resulted in the formation of gaps larger than
those characteristic of natural forest disturbances such as
windfalls.
Forest ggp Stggg gpg Sapling gpg ggtg Regeneration

In the heavily selectively cut and treated forest (K-
13) the correlation between gap size and sapling and pole
species richness or density was not significant. It was
likely that the low correlation was influenced by the low
species richness and density due to treatment effects,
rather than factors controlled by gap size. In K-15, both
the species richness and density of saplings and poles of
all species combined and desirable tree species were

significantly negatively correlated with gap size (all r5-
0.459, n - 40, P<0.01). Even the density of desirable tree

saplings and poles per species was negatively correlated

with gap size (rs-0.488, P<0.01, Fig 4). For K-14, the
species richness and density of saplings and poles of all
species combined and desirable tree species were also
significantly negatively correlated with gap size (all

r50.514, P<0.005). Again, the density of desirable saplings

177A

Figure 4. The density (No/ha) of saplings and poles of all
tree species combined and desirable tree species (21.5 cm to
10 cm dbh) per species in the forest gaps of lightly
selectively cut (K-14), uncut mature forest (K-30) and the
heavily selectively cut (K-15) and heavily selectively cut
and treated (K-13) forest tracts of the reserve.

Density of Saplings and Poles Per Species

Density of Desirable Sapling/Poles Per Species

250

200-

150»

100

50

178

 

 

 

200

150

100,

50_

 

 

250

200 »

150

100.

.50.

 

 

 

 

200

150

100',

50_

 

* ‘1
O
r o ..... 0- K-15
-A---A- K-13
\ O... . o
’I’ ‘8.~‘8---A~‘ '__A\ 0....O.,... no A” ‘\
x’. . ”ac-"Tin, AI’AO‘.,$(.°’ft-o-°: 0: 622
-O—-O- K-3O
L "X"“'"k' K'14
A
\\ I, \x
\ / 'X
. 2X. . g. 2 .
’ Q
- ' .0 on K-15
'. -A-—-A- K'13
£~”.8:::3;-4“3”A Av” \
’I . \\.O'°- o,..:8:. 0.." \\6
A! . . 2"~o'"'.\A__.A/f EWGC’; 2

 

 

200 400 600 .800 ‘1 0001 2001 40016001 800 200022002400 2600 >2600
Gap Size-Classes (nearest 200 m2)

179

and poles per species was strongly negatively correlated
with gap size (r=-0.536, P<0.005, Fig 4). In spite of the
small gap sizes and low range, the tree saplings and poles
in the gaps of K-30 were also responsive to variations in
gap size. The species richness of tree saplings and poles
of all tree species combined and desirable tree species were
highly negatively correlated with gap size (r=-0.601 and r=—
0.619, respectively, all P<0.005). The density of saplings
and poles of all species combined, desirable tree species,
and the density of desirable tree saplings and poles per
species, were also significantly negatively correlated with
gap size (all r5-0.435 and P50.01).

Tree sapling and pole species richness, and density of
saplings and poles per species, tended to fluctuate most,
tending downwards in larger (>650m2) than smaller gaps less
than 650m2 in size (Appendix tables 1-4, Fig. 4). The
majority of natural gaps (97.5%) in uncut mature forest (K-
30) fell below the 650m2 gap size range. The gap sizes with
the highest tree sapling and pole species richness and
density, and number of individual saplings and poles per
species, for K-13, K-14, K—15, and K-30, also fall within
the 200m2 and 650m2 size range.

There was a highly significant difference between mean
tree sapling and pole species richness in the gaps of cut
and uncut forest tracts (F=40.43, P<0.005). The Duncan's

multiple range test also showed that the mean species

richness of saplings and poles of all species combined were

180

lower and similar for K—13 and K-15 gaps whereas they were
higher and similar for K-14 and K—30 forest gaps. For
desirable tree sapling and pole species, the gaps in uncut
and lightly cut forest had higher levels of species richness
than gaps in heavily cut or heavily cut and treated forests
(F=37.82, P<0.005). Again, the means of desirable tree
sapling and pole species in the gaps of K—13 and K-14 were
low and similar while those in K-14 and K-30 gaps were high
and similar. The gaps in the heavily felled and treated
forest (K-13) had the least while those in uncut forest
tract had the highest means of desirable tree sapling and
pole species richness.

Similarly, the mean densities of saplings and poles of
all tree species and desirable species in the gaps of uncut,
lightly cut and heavily cut forests were significantly
different (F=43.33 and F=47.70, respectively, all P<0.001).
The Duncan's test also showed that the mean densities of
tree saplings and poles in the gaps of heavily cut forests
were similar and lower than the mean densities in the gaps
of lightly cut and uncut mature forest. Therefore, ranking
of the densities of saplings and poles in the forest gaps
was K-30>K-14>K-15>K-13. Except for the K-13 vs K-15 plot
comparison, all other gap plot pair comparisons of the
density of sapling and poles per species were significantly
different (all U31130, critical Uo.os(4,40)=1066.7. STP.).
The gaps in uncut mature forest had the highest densities

per species followed closely by the gaps in lightly cut

181

forest and those in heavily selectively cut forest (Appendix
tables 1-4, Fig. 4).

The percent ground vegetation coverage of the study
gaps in the four forest compartments was also compared using
STP (see Sokal and Rohlf, 1981). As before, the calculated
critical value of 00.05”! as) was 1066.7. All gap plot pair
comparisons of percent ground vegetation cover (gvc) showed
highly significant differences (all U31133.5). The
hierarchy in gap gvc ranking was K-13>K—15>K-14>K-30. In
general, gaps in selectively cut forest tracts supported
higher densities of herbaceous and non woody plant species
forming denser gvc than gaps in uncut forest. Light-
demanding species were frequently more dominant in cut
forest gaps than uncut forest. In some cases, a few gap

species such as Fleuryg urophyllg, Brillantaisia nitens,

 

Aspilia africana, Monechmg and Mimulopsis sp. dominated the

 

dense gvc almost to the exclusion of other species in gaps
of cut forests. The gvc was poorly correlated with forest
gap sizes in K-13, K-15, and K-30 but not in K—14 (r=0.382,
P<0.05). In K—13 and K-15 most forest gaps were large with
relatively high and uniform gvc. In K-30 gaps, the gvc was
mostly low and variable among gaps. However, the
correlations between forest gap gvc and the species richness
and density of saplings and poles of all tree species,
desirable tree species, and the density of desirable tree
saplings and poles per species, in K-13 and K-15, were all

significantly negative (all r5-0.339 and r5-0.383

182

respectively, and P<0.05). This was also true for K-14
forest gap gvc but the correlations in this case were more
strongly negative (all r5—0.560, P<0.01). For the naturally
formed gaps in uncut forest, the gap gvc had no significant
relationship with the sapling and pole species richness and
density. However, the tendency towards an inverse
relationship between gvc and tree sapling and pole species
richness and density was present.

The effect of logging intensity and chemical treatment
on forest regeneration was analyzed by comparing the species
richness and density of desirable tree saplings and poles in
forest gaps of similar sizes. Sixteen gaps of similar sizes
(all "(16,16)2117'5 while UO.05(16,16)=83 Mann-Whitney U-
test, Siegel, 1956) were selected from the 40 study gaps for
each of the three forest tracts (K-14, K-15 and K-30).

These were used to study the effects of logging intensity on
forest gap tree sapling and pole regeneration for lightly
cut, heavily selectively cut and uncut forest tracts.

The sapling and pole species richness for all tree
species combined in the gaps of the heavily cut forest and
the lightly cut forest (K-15 vs K-14), and lightly cut and
uncut forest (K-14 vs K-30), were not significantly
different (U=91 and U=104, respectively and P>0.05).
However, the gaps in heavily selectively cut forest (K-15)
had lower sapling species richness than the gaps in uncut
forest (U=85, P=0.05). These results indicate that both

light and heavy selective logging tended to reduce tree

183

sapling and pole species richness of the gaps created by
felling. But the level of reduction depended heavily on the
intensity of felling. Thus, light selective felling similar
to that in K-14 was similar to the natural tree falls and.
less destructive to sapling and pole regeneration of all
tree species in the gaps created.

The influence of logging was more clearly demonstrated
by the changes in species richness and density of saplings
and poles of desirable or canopy tree species. The forest
gaps in lightly cut and uncut forest tracts had similar
species richness of desirable tree saplings and poles
(U=90.5, P>0.05). However, the forest gaps in heavily
selectively felled forest had lower species richness of
desirable tree saplings and poles than gaps in lightly cut
or uncut forest (U=34 and U=22.5, respectively, all
P<0.001). These results suggest that logging affected the
species richness of saplings and poles of desirable timber
or forest canopy species more than the species richness of
saplings and poles of all tree species in general.
Consequently, the gaps of heavily selectively cut forest (K-
15) had lower density of desirable tree saplings and poles
per species than the gaps in lightly cut (U=54, P<0.01) or
uncut mature forest (U=20, P<0.001). In contrast, the gaps
in lightly cut and uncut forest tracts had similar densities
of desirable tree saplings and poles per species (U=94,

P>0.05).

184

To test the effect of post felling chemical treatments,
the sapling and pole regeneration in eighteen pairs of gaps
of similar sizes (U=155, P>0.05) from the 40 marked gaps in
heavily cut and heavily cut and treated forests were
compared. The species richness of tree sapling and pole
regeneration in the gaps of heavily cut forest (K-15) was
higher than that in the gaps of heavily cut and treated
forest (U=60, P<0.01). The sapling and pole species
regeneration in the gaps of lightly cut forest was 90.1% of
the species richness in K-30 gaps, that in the heavily
selectively cut forest 78.4%, and that in the heavily cut
and treated forest gaps only 40.8%. Post felling treatment
with arboricides could account for most of the 37.6%
depression of tree sapling and pole species in treated
forest gaps, relative to heavily cut but untreated plots.
Correspondingly, the density of sapling and pole
regeneration, and density of saplings and poles per species,
was higher in the gaps of heavily cut forest than in heavily
cut and treated forest (all 0551, P<0.01).) However, the
species richness and density of desirable tree saplings and
poles per species were not significantly different, although
21.2% and 25.7% lower, respectively, in K-13 than K-15
forest gaps.

Finally, the effect of gap conditions on tree
regeneration was examined by comparing the species richness
and density of saplings and poles in the gaps and plots

located in the forest tracts surrounding the gaps (Table 1).

185

.n-Uv. 03H NO :0d&6%>0” ”Ra‘s—60m O

 

 

«NN«O O.vO« A0.000 O.nN« AN0.00 O0.0 «O0.0. O«.O O
AN.ONO v.OON .0.0N. v.OON .N0.00 N0.0 .O0.00 «N.O N OnaO
.O.NOO O.NO .Oa.OOv v«.OO .O«0.00 O«0.0 .v0.0. O0.0 O
Av.NOv O.«O« .Ov.va v0.00« .O«0.00 «v0.0 .N0.00 v«.O m O«1O
An.OO«O 0.0n« .N.ON. O.n«« .N0.0. v0.0 .N0.00 O«.O O
AN.OOO O.«NN Av.NO. v.ONN .«0.0v O0.0 “«O0.00 O«.O m v«um
ao«oonm\ao«oa a O6««n6e 60«00&6\0«on d leemoc£0«u .nnm v«on NI\660::o«u .096 muo«a stum
O«n6u«aoa no ha«a=0O O6««n6a No hv«mcoa d u:««n66 O«neuweoa O«on a I:««A6e ««<

 

.o>hoeou 0:» no 6&06ua umouou
.OOIN. 0:06: 6:6 «O«Iu 6:6 v«INv 0:0 N«o>«uoo«oa o:& No £060 hon «a. 600«& anonou 86006506 666 AO. one. vouoo«oa

N«louceh ON one hou «6£\O. ha«m:ov 666 Nl\6ao:£0«u noncone .sAO IOO«vl0O.«MO O«on 0:6 I:««n6e 0I6ho>< .« O«aeb

 

186

For each forest compartment, the tree saplings and poles in
20 pairs of gaps and corresponding forest plots were
compared. The Wilcoxon matched pairs signed rank test (T)
was used.

In lightly cut forest (K-14) the forest plots adjacent
to the gaps harbored higher species richness of all tree
species, and desirable tree species, than did the
corresponding forest gaps (T=16.5, P<0.005 and T=22, P<0.01
respectively). The species richness of all tree species
combined, and desirable tree species in forest gaps was
31.6% and 33.3% lower, respectively, than that in adjacent
forest plots. Similarly the forest plots in K-14 had higher
densities of sapling and pole species for all tree species
and desirable tree species than the forest gaps (T=7 and
T=9, respectively, all P<0.005). The density of all tree
species combined and desirable tree species in forest gaps
was 49.6% and 37.6% lower, respectively, than that in
adjacent forest plots.

Evidently, the gaps in heavily selectively cut forest
(K-15) experienced serious reductions in tree species
richness and density. The species richness of saplings and

poles of all three species and desirable tree species in

forest gaps and forest plots were highly significantly
different (T=4 and T=23 respectively all P<0.05). The
species richness of saplings and poles of all tree species
and desirable tree species in the forest gaps was 57.2% and

53.7% lower, respectively, than that in adjacent forest

187

plot. Consequently, the density of tree saplings and poles
per species for all species and desirable tree species was
higher in the forest plots than corresponding gaps (T=1 and
T=O, respectively, all P<0.005). Furthermore, the density
of saplings and poles per species, for all tree species and
desirable tree species, in forest gaps was 63.4% and 64.8%
lower, respectively, than that in adjacent forest plots.

For the uncut mature forest, the forest gaps also
exhibited lower sapling and pole species richness for all
tree species and desirable tree species than the
corresponding forest plots (T=1.5, P<0.005 and T=24, P<0.02,
respectively). However, the percent reduction of tree
saplings and poles in the forest gaps was relatively lower
than in selectively cut forests which had a predominance of
large gaps. The species richness of saplings and poles of
all tree species and desirable tree species in the forest
gaps was 23.8% and 28.6% lower, respectively, than that in
surrounding forest plots. The density of saplings and poles
per species for all tree species and desirable tree species
was also lower in the forest gaps than corresponding forest
plots (T=10 and T=48.5 respectively, all P<0.05). The
density of saplings and poles per species for all tree
species and the desirable tree species, in forest gaps was
40% and 24.2% lower, respectively, than that in adjacent
forest plots.

Forest disturbances through both light and heavy

felling and even natural tree falls had a tendency to reduce

   

188

gap sapling and pole species richness and density of
saplings and poles. However, the level and significance of
reduction was related to the intensity of felling and the
nature of disturbance as well. It seems that felling alone
could not explain the paucity of saplings and poles in the
gaps of selectively cut forest tracts. Other consequences
of forest felling may have significantly contributed to
impoverished regeneration of saplings and poles in forest
gaps.

Except for the non-desirable pioneer species or forest
gap specialists such as I£2E§ guineensis and species of
Neoboutonia, most tree species regenerating in the gaps were
also present in adjacent forest tracts. Even the 18 common
desirable timber or canopy tree species in the central
forest reserve were common both in the forest gaps and plots
in adjacent forest surrounding the gaps (Table 2). The
density of the common, desirable timber or canopy tree
species was similar for forest gaps and forest plots in K-14
but significantly different in K-15 where forest plots had
higher densities than forest gaps (T=32.5, P<0.01).

However, there was strong evidence of higher tree sapling
and pole densities of the common desirable timber or canopy
tree species in the naturally formed forest gaps than in
forest plots (T=9, P<0.005) in uncut mature forest. These
results imply that most of the sapling and pole regeneration
in the forest gaps was from pre-existing seedlings and

saplings which survived the impact of gap creation. Many

189

 

 

 

 

«.ee No.4m e.- e.e~ n.—4 4.ee e—.e~
e.ee e.ee o.~_ 4.x" s.n4 «.u4 e~.e~ m
o.me e4 e.4 o3 ~.e4 «.me e.o_ deco 4 .e
e.ma~ as“ ~.~ e.s m.__ o.e a.4 auouO42moqu .m
e.oe e.—« e.4_ e.—~ e.oo~ ».oe 4.e~ «moduuosmu .m
o.u o.u n.e o.e~ e.~ o.e 4.m audooxo .m
o.4 N.e «.5 e.en e.- o.e «u «anouemzsmz .o
NON e..~— e.ee a.ee e.g4m n.ne em aescsososa .z
4.ee n.e4 e.e s.e~ n.ee n.e4 o.o ”censuses .2
~.e4 e.ge o.~_ N.ee e.om e.ee o.o mu«eoue¢qm .z
eo— e.en~ e.m~ e.e o.ao e.ee o.o “aqmeeoeswze .a
4.eo. e.gs m.e e.~n m.ee n.em e.~ as«oegea4 .m
4.e o.e e.e o.m e.e N.e_ ~.e usncoqousu .e
~.- o.e «.u a.» e.e a.» 4a.: a««usmosocl .e
4.» o.o ~.~ e.e e.e e.e o.e assoasfie .o
c.44 e.e~ a.» o.» n.e~ e.e~ ~.4 Insane-sauce .o
e.en an o.- e.- e.m~ o4 4e.o ascoaeee .o
«.ue. N._e e.e4 es ofid e.~o~ o.me~ «assesses .m
4.4m— e.oe ~.m e.~ n.4e we e.e« assmowxoe .<
e.ee e.e~ 4.6— o— «a as m.e4 alsuuse4¢ .<
o m o a o a o nowooem ones
can: 3.1: 4—1: aqua

 

.maoeh» amouOO AOOIOV vaocs O66 «O«ua .v«lx .N«1Ov 0:0 h«0>«000«0m
:« 696- 0.3 0» #:00636 C: maodu ..«m0hou «466 «O. 696' umouou .:« 8255.09.03:
0.50600 ue0hom «6.4.2.00 0.3 :« 60«00nm 00.5 mac—.60 .no 00362.. 0«£6h«m0«. sac-loo

use as .sps noe_wnoe._. oases. sodas sea muesqeuu eo .as\.oz. manages use .s ofisae

190

more tree seedlings and saplings of the common desirable
timber or canopy species were able to survive and continue
growth into the sapling and pole classes in the commonly
small to medium natural gaps than the characteristically
larger gaps created by logging.

In all forest compartments the saplings and poles of
Blighia pnijugata, Newtonia buchangptg, Lovoa swynnertonii,
Funtpnia latifolia and Antiaris toxicarig seem to be
regenerating more successfully than other species in both
forest gaps and forest understory. However, they exhibited
some tendency towards higher densities in forest gaps than
forest understory. §ypphonia globulifera was poorly
regenerating in all cut forest tracts, gaps, or forest plots
while abundant regeneration in gaps and understory of uncut
mature forest was observed. Some dominant canopy tree

species such as Olea welwitschii, Parinari excelsa, Fagara

 

macrophylla, and Fagaropsis angolensis showed no
regeneration site preferences and had very low sapling and
pole regeneration in both forest gaps and forest understory.
Herbivory and Seed Predation tp Forest Gaps

Preliminary data on forest gap utilization by elephants
suggest that their impact on the regeneration of some

important timber or canopy tree species in large gaps
created by felling can be significant (Table 3). The
frequency of elephant incursions into the logged forest
areas and the number of forest gaps used as browsing sites

were higher in selectively out than uncut mature forests

 

191

.u6wuooun6 666 w6«an66m .u0>o:m=n .6o«v6««on06 .u6nxu6a "06:«06«

0.6666 no monhfi u

 

O O O
O O O
.6.« m 4o

O O O
O.m 0.0n «.O

v.v 0.0v N.O«

O O O
N O N
« O O

O OOO«
N« OOO«

O vOO«

 

OOIN O«II v«lx

60>h06no 0'6l66 Odom d
O6««n6a 006» no 60959

OOIM O«Il v«lm

60»«6«> 666‘
601666 N 660:

OOIM O«ll v«lm

m6o«06606«
$66690H0 no .02

6o«u6>6066o no
m£a6o0 666 660%

 

.o>uouou one :6 .ee-u. enouou uses: use .e~-n. eso eqo>neooqon 5~O>4oe .14316. use edesuss :6 moans ass-sous

m6 0&663A0«0 0:» ha 606: one. ueouon 60:66! no 0O6u600001 666 m6o«66606« a66600«0 no hocoaoohn oak .O 0«n6N

192

(all T52, P<0.005, Wilcoxon's test). Consequently, elephant
damage to gap vegetation, including herbs, shrubs and tree
species was high for the selectively cut forest tracts. The
seedlings, saplings and poles of a few canopy tree species,
such as A; toxicaria, t; swynnertonii, N; buchananii, g;
gummifera and B; phoberos seemed more prone to elephant
damage than other species. The seedlings, saplings and
poles of these species were killed through all five modes of
feeding damage, i.e. barking, total defoliation, pushing
over, snapping and uprooting. The saplings of Eggpg.
Antiaris and Bosguiea were particularly subject to
uprooting. However, more observations and data are still
needed before firm conclusions can be made.

Seed predation trials were intended to show whether
there were different levels of seed predation at increased
intensities of forest exploitation. Within the study
compartments, the rates of seed predation, with respect to
gap center or edge, were not different. With the exception
of Aphania seeds, the rates of seed predation were slightly
higher in the gaps of selectively cut forest than in uncut
forest gaps (Table 4). The half life of seed piles (i.e.
the time it took 50% of the seeds to disappear) were
generally shorter, and very short in the case of thpggpg
seeds, in the gaps of selectively cut forest relative to
uncut forest gaps. There were weakly significant
differences between the rates of seed predation of Trichilia

in the gaps of cut and uncut forest tracts but no

193

 

 

 

N«.O O.N O OOIN
ON.O O.N O O«Iu mwanolmmo m«0fl0«06>:
vN.O O.N O v«ul
ON.O O.« a Omnu
Ov.O O.« O O«uu 6uw6m0«mm d«««mfl«hh
Ov.O O.« O v«IN
vo.O n« v OO1N
ON.O v.N v O«II «03666lflm a oe:l«r
O«.O O.N v v«ul
Ov.O «.« O onuu
O«.O O.N O O«lu 0«a60«6 0606 6n666m<
N0.0 N.O O v«lu

...s\.~.s\.soo.. ..4... ..

6o«066oho 6000 00««& 6006 06«e »o«m eonoone 6000 006s
no .6. 0&66 660: no 0nn«1n«6= 0««n600O h6saO

 

.0>60006 0:0 no 00066» 000uon .O«1u. $60 h«0>«a00«0m h«n>606

666 «onuuv a3066 ..v«1M. v60 N«»:O«« no 606' ueouon 6n 6o«u66066 6006 no 0006m .v O«n6h

194

differences in the predation rates of Aphania and Uvarigpsis
seeds. However, the gaps in cut forest tracts exhibited
higher rates of Mimusops seed predation than gaps in uncut
forest [U=2, P<0.001 (K—14 vs K-30), and U=0, P<0.001 (K-15
vs K-30), Mann-Whitney U-test]. There was no difference
between Mimusops seed predation rates in the gaps of lightly
cut and heavily cut forest (P>0.159). Deducing from the way
seeds were eaten, the chips of seed or seed coat, and
sometimes fecal pellets left on seed pile sites, rodents
appeared to be the main seed predators. Again, the data-
base was small; more tree seed species need to be tested
before firm conclusions may be drawn.
Sources gt Plant Recruitpgnt Into Forest ggpg

Table 5 shows the summary of the species richness and
density of plants germinated in 1m2 plots of forest soil and
forest gap soil from compartments K-14, K-15 and K-30.
Single classification ANOVA and Duncan's multiple range test
(Sokal and Rohlf, 1981) were used to compare the species
richness of plant seedlings germinated in 1m2 plots of the
top 10cm of forest and gap soils. For the forest soils,
there were no significant differences in species richness of

tree seedlings sprouted from the soils of lightly cut (K-14)
uncut (K-30) or heavily cut forest (K-15). Except for the

forest soils from K—14 and K—30 (F=9.05, P<0.005), the
forest soils from K-14 and K-15, K-15 and K-30, had similar
numbers of shrub species. The soils from uncut forest had

the least, whereas those from the lightly cut, intermediate

 

 

195

 

«no. 66. aeohom n O
««06 vm0uom u n
ha«66on
0.0«0.0.00N O.NON.N.OON N.N«O.N.O«O 0.0 .O.«N O.v«.O.NO 0.0«.v.NN d 60«00&O .oz M
O.v .O.v O.N .0.0 O.v .0.0 O.« .N.N v.« .O.» 0.0«.«.« 6&0£I««0 666 006«>
0.0NN.0.0N« 0.00N.O.nvN N.ONv.v.OON 0.0 .O.«« 0.0 .0.0N v.O .0.0« 60«0096 660006260:
0.0N .N.N« v.Ov .O.N« 0.0N .O.N v.« .0.0 0.0 .v.N O.N .O.« noncomm azhsm
«.O« .n.NO O.N .N.OO O.« .N.Ov O.« .v.n O.« .0.0 O.« .O.N 6I6««6006 eons
O n O n O b O b O n O m
«ONu6vOO1O AONu6OO«1M .ONu6vv«1l .ONu6OOOIO AONu6.O«IO .ONn6.v«1O luon 0n«« v66«m
6066«n no ha«060n M 60«009O a66«m .02 M

 

.5006666 60600600 6n 63ouu 666 OOlO 666 O«1N .v«IO loan .O. ««oma6w 666 .m.

««om um0hon no l0O« no» 0:» no avo«n I« 6« 60066«l60O .NI\.oz. huwo606 666 0606:0«0 60n00n6 a66«n 660! .O 0«£6h

N

196

and heavily selectively cut forest had the highest species
richness of herbaceous seedlings (F=20.60, P<0.005). The
vine and climber species richness was higher for the soils
from uncut and heavily cut forest than for lightly cut
(F=4.57, P<0.025). When all plant seedlings were considered
together, the soils from uncut and lightly cut forest had
lower plant species richness than soils from heavily cut
forest (F=9.96, P<0.005). However, all forest soil seed
bank germination trials were dominated by non-woody seedling
species with soils from the heavily cut forest in the lead
(Fig. 5).

The sterilized soils from K-14 forest had the lowest
plant species richness of all soils examined. Only two
plants of elephant grass (Pennisetum sp.) and ten of black
jack (Bidens sp.) sprouted from the sterilized soil. These
two species never reappeared in the plots of other soils
from the forest or forest gaps enclosed in the screened
nursery. This was possibly a case of contamination. 0f the
germinated seedlings from the forest soils of K-14, K-15 and
K-30, tree seedling species richness was only 10.3%, 9.3%
and 16.0% of the species richness, respectively, while
herbaceous plants constituted 80.4%, 71,2% and 53.1%,
respectively (Fig. 5).

Similarly, the forest gap soils from uncut, lightly cut
and heavily selectively cut forests exhibited no significant
differences in the species richness of germinated tree, vine

and climber seedlings. For the shrub and herbaceous plants,

196A

Figure 5. Percent contribution of the four plant life forms
to plant species richness and density of plants (No/m )
germinated in 1m plots of the top 10 cm of forest soil and
gap soil collected from forest gaps and forest plots
adjacent to the gaps, respectively, in lightly cut (K-14),
heavily selectively cut (K-15) and uncut mature forest (K-
30) and grown in screened nursery. T=tree seedlings, S=Shrub
species seedlings, H=herbaceous species seedlings and

=vines and climber seedlings.

197

K-15

H 80.4%

 

a. Mean seedling species richness (Forest Soil)

K-14 K-15

H 61.0%

   

b. Mean seedling species richness(Forest-gap Soil)

K-14 K-15
V1.2°/o-\ V3.2%\

S 2.5% \ S 4.2%

H 82.7% H81.3%

?

c. Mean seedling density (#lm2)(Forest Soil)

K-14 K-15

 

 

 

H 92.4%

 

d. Mean seedling density (#lmz) (Forest-gap Soil)

 

K-30

 

 

198

the soils from uncut forest gaps had the lowest species
richness while those from heavily cut forest had the
highest, with soils from lightly cut forest intermediate
(F=6.84, F=7.19, respectively, all P30.01). When all plant
seedlings were considered together, the soils from uncut and
lightly cut forest gaps had lower plant species richness
than those from heavily cut (F=5.1, P<0.025). The majority
of germinated plant seedlings in the forest gap soils were
also of herbaceous species (Fig. 5). Of the germinated
seedling species from the gap soils of K-14, K-15 and K-30,
only 11.9%, 8.9% and 17.7% were tree species, respectively.
In contrast, the herbaceous species constituted 58.9%, 61.0%
and 55.6% of the germinated seedling species in K-14, K—15
and K—3O forest gap soils, respectively. The vines,
climbers and shrub species formed the remainder (Fig. 5).
The Mann—Whitney U-test (Sokal and Rohlf, 1981) was
also used in paired comparisons of species richness of
plants germinated in forest soils and forest gap soils. The
forest soils from under the forest tracts, irrespective of
whether the forest was heavily cut, lightly cut or uncut,
exhibited higher tree species richness (all 0348, P<0.001),

higher herbaceous species richness (all U=49, P<0.001) and

higher vine and climber seedling species richness (all U339,
P<0.05) than forest gap soils. For shrubs, gap soils in the
heavily cut forest tended to have higher species richness
than forest soils from lightly cut and uncut forest.

Comparisons of all germinated plant species combined also

199

showed that forest soils had higher plant species richness
than forest gap soils (all U349, P<0.001).

Comparison of the density of seedling sprouts from the
forest soils and gap soils using ANOVA and Duncan's test
showed a different pattern from that of plant species
richness. The densities of tree seedlings, shrub seedlings
and non-woody plants germinated from the forest soils of K—
14, K—15 and K—SO were not different. The overall mean
densities of plants germinated from the forest soils of the
three forest compartments were also not different. Only the
densities of vine and climber seedlings from the soils of K-
15 were higher than those from K-14 or K—SO forest soils
(F=4.24, P<0.05). With the exception of tree seedling
densities, where K-SO gap soils had the highest (F=4.33,
P<0.05), the densities of seedlings of shrubs, herbaceous
plants, vines and climbers germinated from the gap soils of
the three forest compartments were not different. Again,
the overall mean density of plants germinated from the gap
soils from the three forest compartments was similar.

However, paired comparisons using the U-test showed
that the densities of tree seedlings germinated from forest
soils were generally higher than from forest gap soils (all
U338, P50.05). The density of shrubs, vines and climbers
was not different between forest and forest gap soils.
However, the forest gap soils from heavily felled forest

tracts had higher densities of herbaceous seedlings than

forest soils from uncut, lightly cut or heavily cut forest

200

tracts (all U340, P50.05) whereas all other plot pair
comparisons had similar densities. Because of the extremely
high non-woody species densities, the forest gap soils from
the heavily cut forest also had higher densities of
seedlings of all four life forms considered together than
forest soils from under uncut, lightly cut or heavily cut
forest tracts (all U344, P30.01). All other possible plot
pair comparisons had similar densities.

Within the study compartments, the mean species
richness and density of seedlings of vines, shrubs and trees
germinated from the forest soils of K-30 were low and
significantly different from the species richness and
density of herbaceous seedlings (F=23.8, and 24,07,
respectively, all P50.005, ANOVA and Duncan's' test, Fig.
5). For lightly cut forest (K—14) the trend was similar to
K-30 forest soil. The species richness and density of
vines, shrubs and tree seedlings were similar and lower than
the richness and predominance of herbaceous seedlings
(F=272.9 and 7.9 respectively all P50.005, ANOVA). The mean
species richness and density of vines, shrubs and trees
germinated from forest soils of K-15 were also very low
compared to species richness and density of herbaceous
seedlings (F=77.82 and 29.85, respectively, all P<0.005,
Fig. 5).

Within-compartment comparisons of seedling sprouts from
gap soils of K-14, K-15, and K-30 showed that the species

richness and densities of tree, shrub, vine and climber

201

seedlings were similar but quite low and different from the
extremely high species richness and density of herbaceous
seedlings (all F321.3, P<0.005, ANOVA and Duncan's test,
Fig. 5). This implies that seed banks in both forest and
gap soils were abundantly rich in seeds of herbaceous
species but extremely impoverished in seeds of tree species.
In addition, the forest gap soils were more impoverished in
both herbaceous plant species richness and tree species
richness and abundance than forest soils.

The majority of tree seedling species germinated from
the forest soil and forest gap soils belonged to the so
called 'non—desirable' timber species (Table 6). Most were
the fast growth, gap or secondary forest colonizers with a
few characteristic of middle and understory species. 0f the
commercially desirable timber species, only Celtis africana
appeared in the forest and forest gap soils and Cordia
millenii in forest soils from uncut forest, but all in
extremely low densities. Trema guineensis was the only
pioneer gap species well represented in the soil seed pools
of forest and forest gaps. However, the forest soils had
higher seedling densities of Trema than forest gap soils
(Table 6). In general, the forest soils from uncut mature
forest had higher tree seedling species richness and
abundance than forest soils from selectively cut forest
tracts or gap soils (Table 6). However, the rarity of

seedlings of commercially desirable timber or canopy species

Table 6. Mean density (No./a2) and frequency of occurence (100X=l) of tree seedlings germinated

froa 1-2 plots of the top 100- of forest soil and forest gap soil tended in screened nursery.

No. of la2 plots for K-14, [-16 and K-30 were equal, n=l4.

 

Forest Gap Soil
K-15 K-30 K-14 [-15 [—30

Forest Soil

K-14

 

Tree species

202

OTOOONOOOOOO
OOOOOOOOOOOO

ONOOOGOOOOOO
éOzOOOOOOOOO

ION—:OMOOOOOOO
OOOOOOOOOOOO

..OOOOOOOOOO

NDOOOOOOOOOO
OOv-IOOOOOOOOO

flNOfiOOOs-OOOOOO
OOOOOOOOOOOO

..QOOOOQOOO

”Q'OC'OOONOOOO
F'lOOOOOOOOOOO

OOOOOF'OO—ol'zlfl'v':

OO—‘OOOOOOOOO

OO.-Q OO....

N‘NOOOOOOOOQ‘
NNgGOéOOv-zOOO

OOOOQO'TONOO':
dOv-UOOOOOOOOO

NOOOOOOOOOON

OOIDOHOOOOOOO

ONONOOOOOOOO

v-IOOOOOOOOOOO

NOOOOOOOO
OOOOOOOOO

'
N

23.8.
16.8.

uineensis
ta

Teclea nobilis
roe ab s ini a

C ho andra betace
Cordia ailleni
la ceolata

Unidentified

Chaetacaea aria ata
ios

Celtis durandii
Celtis african
Neoboutonia g2;
C ause a anis

Treaa
Haes

 

Total Tree Species

 

15.1

33.8 57.7 .

42.2

Mean Density

 

203

in the forest and forest gap soil seed banks was of concern
with regard to forest regeneration.
gap §Qil §ggg gangs and Incoming Seed Dispersal

This experiment was designed to demonstrate the
potential contribution of buried seeds and seeds newly
dispersed into the gaps, and the influence of existing
vegetation, on forest gap regeneration. Appendix Table 5
shows averages for plant life forms germinated in the 20 PRP
in forest gaps of the selectively cut (K-14 and K-15) and
uncut forest tracts. The general trend was similar to that
of forest gap soil seed germination trials where tree
seedlings showed the lowest species richness and density
while the herbaceous seedlings showed the highest. However,
the plant species richness and density observed in PRP were
much lower than those realized from forest soil seed pools
or forest gap soils seed pools germinated under screened
nursery conditions (all P<0.01). This may suggest seed or
seedling predation or failure of some seeds to germinate
under forest gap conditions which were different from the
nursery conditions. In K-14, K-15 and K—30 gaps, the mean
seedling species richness of trees, shrubs, vines and
climbers that germinated from the PRP was lower than that of
herbaceous sprouts (F=8.36, 13.53 and 12.47, respectively,
all P50.01 and Duncan's test). This was another clear
indication of the abundance of viable seeds of herbaceous
compared to tree species in the forest gap soils. The

results also imply that successful tree regeneration in

204

forest gaps could only come from seed dispersal into the
gaps or further growth and development of seedlings already
established at the time of disturbance.

There were significant differences between treatment
means as well. For all the study plots (K-14, K-15 and K-
30), the PRP with vegetation removed, vegetation removed and
screened, or vegetation and surface soils removed, had
similar and higher plant species richness than PRP with all
vegetation left intact (F=4.10, 4,42 and 7.06 respectively,
all P50.05 and Duncan's test). The density of herbaceous
sprouts in the PRP of K-14, K—15 and K-30 was also very high
compared to that of tree, shrub, vine or climber sprouts
(all F320.93, P<0.005). There were significant differences
between treatment means with Q1 and Q2 having the highest
densities, Q3 intermediate and Q4 the lowest. Surprisingly,
the quadrants with vegetation and surface soils removed also
had higher plant species richness and density of newly
germinated seedlings than quadrants with vegetation left
intact (all P<0.025, Mann-Whitney U-test, Fig. 6). This
strongly suggests that gap vegetation had a strong
suppressive influence on the germination and establishment

of newly dispersed and buried seeds in forest gaps.

In total, seven tree species germinated in K-14 PRP.
These included 3; guineensis, g; durandii, Q; africana, !;
Lanceolata, Q; abyssinica, T; nobilis and fl; bagshawei. The
PRP in K-30 gaps had five of the species in K—14 (except M;

lanceolata and Q4 africana) plus 9; gorungosanum and Q;

204A

Figure 6. Comparison of tree segdling species richness and
density of tree seedlings (No/m ) newly germinated in 1m
manipulated gap plots in forest gaps of K—14, K—15 and K-30
after 6 months and 20 months.

 

 

205

   

03 Q4

 

 

 

 

 

 

 

 

 

 

 

8

x 1..
O
5
., ,v
.2 0
fig In ‘- l0 (9
£2 - ‘- \0

5.98 X ' X k
¥V= ~ N
&- §_ \,0
m m 5
co
0
'N
E o
x 1-
O

 

- ('0 v-
v- 1- O .
ssauqoga 'dds Buupaas 6311 x ZW/ Sfiunpaas 6611 X

206

mukole. The PRP in K—15 had only four tree seedling species
of Q; durandii, Q; africana, T; guineensis and g; ggga.
However, only two tree seedling species, 94 durandii and Q;
abyssinica, were more frequent, but in low numbers, in most
PRP of the three forest tracts. All PRP quadrants had lost
the greater percentage of newly germinated plants, more
especially tree seedlings by the twentieth month of
experimentation (Fig. 6). For both the six month and 20
month period of observations, there were no newly germinated
and established tree seedlings in the quadrants with all
vegetation left intact in K-15 gaps. However, by the
twentieth month, all PRP quadrants in the forest gaps of K-
15 had lost all new tree seedlings established in them
during the experimentation (Fig 6). This demonstrated an
extremely high mortality of tree seedlings in forest gaps of
heavily selectively felled forest relative to gaps in
lightly cut or uncut forest tracts.

Mode 9: Tree Sapling and Pole Regeneration in Gaps

 

The modes of sapling and pole regeneration (normal
growth, coppicing and stem sprouts) were compared within and
between gaps in forest tracts which had had different

management histories by using ANOVA and Duncan's multiple

range test. Within the heavily cut and heavily cut and
treated forests (K-15 and K-13) the differences between the
mean density of saplings and poles with normal growth,
coppices and stem spouts were highly significant (F=22.55

and F=79.08, respectively, all P<0.001). In both plots the

 

 

207

means of coppices were higher than the means of normal
growth or stem sprouts (2.0(3),4.2(SS) 9.5(C)) K-15 and
1.9(u),g.9(331 19.9(C)K-13, see Table 7, Fig 7).
However, the means of normal growth and stem sprouts were
similar. For the lightly cut forest (K-14), the differences
among the mean density of normal growth, coppices and stem
sprouts were significant (F=10.06, P<0.005) and larger mean
densities of coppices among the sapling and pole
regeneration, relative to normal growths or stem sprouts,
were observed (11.5(SS)15(N) 26(0)). In uncut forest
gaps as well, the differences among the means of normal,
coppices and stems sprouts were significant (F=9.31 P<0.005)
and the means of normal and coppices were similar while stem
sprouts were low but the three were not separated by
Duncan's test (3.96(SS) 8.2(C)8.24(N)).

Between-plot comparisons showed highly significant
differences between the means of normal sapling and pole
regeneration in the gaps of K-13, K-14, K-15 and K—3O
(F=32.36 P<0.001). The ranking of the mean density of
normal growth sapling and poles was K-30 > K-14 > K—15 > K-
13 (Fig. 7).

Although the means of coppices in the gaps of K-14 and
K—30 looked high (the effect of high sapling and pole
densities), the proportions of gap tree sapling and pole
regeneration as coppices were lower than in gaps of K-13 and
K-15 (Table 7, Fig. 7). However, the gaps in cut forest

tracts exhibited higher percentages of tree sapling and pole

208

 

 

 

someone hon nuances u ne\u:\o:

Am.v~v v.9“ o.o Ame. «.un Am.~ev a.aN o.~p onux
Aw.oN. «.0 m.~ Aw.oov n.@ .p.-. O.N p.nn o—ux
.o.—~v m.- ..q AN.OuV p.wN AN.QN. c.0— «.03 v—nu
.o.n—v Q.“ u.~ Ac.om. a.a~ Ap.bv m.— p.v~ n—ah

ns\s:\o: louexo: Ae\d:\o: nuxs£\oc nuxd£\0:

sa50hns love :UIOHI sofioonn hon yeah
loan excess :so: soouonoo Heluoz yea-sou coo: hoSum

 

.0>hon0h 020 :« .n—IN. eschew

VoasOHv use aao kdo>uaoo~00 hum>son one .o—va aso hqo>uvoo~os hqa>soa use Av~lnv azo hays-MA ..onlm. asocs no one:

eschew 0:» :u «an. museums loan as no ADV secunaoo so .sz alba- «sluo: gag: ucaushocouOH .oaoxoshn cu. cemasaanoa

sodoons 0:00hon i=6 segoons hon As£\.o:v hauscov Asap loo—MIOe.—Av Demon 6:8 Inwdnds cone unocso :60: .5 Quack

 

 

208A

Figure 7. Mean percent canopy tree species population of
saplings and poles regenerating with normal stems (N), as
coppices (C) or as stem sprouts (SS) in the forest gaps of
uncut (K—SO), lightly cut (K—14), heavily selectively cut
(K—15), and heavily selectively cut and treated forest (K—
13) in the reserve.

 

209

«003.30 a 9.35
m 5. m 5. v 5.09.

O

N 1-
wSISIO DUB SXBGJQ “BOW

(‘0

 

museum Edam
m C. m C. w 559.

 

$02300 526.0 .9502
m C.

v 55$! m 5. m 5. v C. env—

 

o

O O
8 v N

0
co

'dOd alod ’8 Buudes 6611 KdOUBQ IBIOJ. °/o

210

coppices than the gaps in uncut mature forest. Similarly,
with the exception of K—13, the gaps in cut forest tracts
also had higher percentages of stem sprouts than gaps in
uncut mature forest. The low mean of stem sprouts in K—13
gaps was due to lower sapling and pole densities and higher
percentages of tree regeneration as coppices.

In summary, the mode of canopy tree sapling and pole
regeneration in forest gaps was variable. The gaps in
lightly cut and uncut forest tracts had high percentages of
normal sapling and pole growth while the regeneration in
gaps of heavily cut forests was predominantly coppices.
Even the mean number of breaks or snaps per stem followed by
successful coppicing was higher for the saplings and poles
in selectively cut plots than for uncut mature forest gaps
(Fig. 7). However, forest gap regeneration by stem sprouts
was small compared to coppices or normal growth saplings and
poles but relatively significant in gaps of selectively cut
forests.

The seedlings and saplings of most canopy tree species
had great potential for coppicing after breaking or
snapping. Appendix Table 6 was included to demonstrate the

phenomenon of coppicing which was widespread and important

in the regeneration of saplings and poles of canopy tree
species. However, some canopy tree species seedlings and
saplings seemed more prone to breakage or snaps than others
but were successful in coppicing. In the gaps of heavily

cut forests (K-13 and K—15) canopy tree species of Newtonia,

 

211

Egggg, Aningeria, Blighia, Antiaris, and Funtumia, were high
on the list as coppices, with 70% to 90% of their saplings
and poles showing signs of breaks or snaps and coppicing.
Almost the same species as in K—13 and K-15 were affected in
the gaps of lightly cut and uncut forests. However, only a
few of their saplings and poles (mean = 51%), relative to K-
13 and K-15, had signs of breaks and coppicing. These
results seem to indicate frequent gap seedling and sapling
disturbances and provide indirect evidence of higher animal
usage of larger forest gaps in selectively cut forest
relative to the small to medium gaps in uncut forest tracts.
Selective browsing by animals may explain why only a few,
perhaps preferred, saplings and poles of primary forest

species were prone to breaking and coppicing.

DISCUSSION

Natural and Artificial Forest §§2§

The forest gaps studied were heterogeneous in terms of
size, configuration and mode of creation, not just "vertical
holes" in the forest canopy extending to within 2 meters of
the forest floor (as defined by Brokaw, 1982a, 1982b, 1985).
In this study, forest gaps were distinguished both by the
extent of distribution of light demanding gap vegetation,
and gradients of gap light from gap center to surrounding
forest some distance inside the forest. Variations in the
definitions of forest gaps make comparisons of gap sizes,
shapes and frequency somewhat difficult (Brokaw, 1982b).
However, some idea of variations in gaps formed in different
tropical regions can be achieved.

Most studies have dealt with natural tree fall gaps
(Whitmore, 1978; Hartshorn, 1978; Brokaw, 1982a, 1982b,
1985) and important regional differences do occur. In
Malaysian forests single tree falls of large canopy
individuals may cause gaps of 400 m2 (Whitmore, 1975, 1978)
while mean gap size at La Selva, Costa Rica was only 89:88

1112 (Hartshorn, 1978). Brokaw (1985) studied gap phase

regeneration in tree fall gaps ranging from 20 to 705 m2 in

a tropical moist forest on Barro Colorado Island, Panama.

213

214

Larger forest gaps have also been known to result from
entanglements by lianas or rare catastrophic events
(Whitmore, 1978; Pickett, 1983; Hart, 1986). The most
extensive gaps in tropical rain forests occur in the cyclone
belts between 100 and 200 North and South of the Equator
(Whitmore, 1978) where cyclones can destroy many square
kilometers in a single storm.

The mean size of gaps in Malaysian forests is 400m2
with a maximum of 600m2 formed by trees falling
simultaneously (Poore, 1968). Longman and Jenik (1974)
reported that liana entangled tree falls could even result
in gap areas of 0.5ha. In Cameroon forests, Bullock (1980)
reports the mean natural gap size as 400 m2 where the main
tree fall damaged 5.2 to 6.2 adjacent trees of 15 to 25 m
tall. In this study the mean gap size for lightly cut
forest was 467m2 (range 75 to 1800m2) and the naturally
formed gaps in mature forest averaged 256m2 (range 100 to
663m2). These were similar to those reported from other
tropical forest studies. However, all these were small
compared to the mean sizes of gaps created by heavy
selective logging. The mean gap size in heavily cut forest
was 1307m2 (range 73 to 7100m2) while that for heavily cut
and treated forest was 938m2 (range 227 to 3313m2). In the
reserve, natural tree falls and gap creation were random
events well separated in time and space. In contrast,

selective logging was systematic, highly selective with

respect to tree species and size, and occurred within a

215

short span of time over large areas of forest. Thus, giant
forest gaps, unparalleled in the natural tree fall gap size
ranges, were created.

In the central forest reserve, the mid slopes of a
hilly terrain were richer in desirable timber species
(Isabirye—Basuta, 1979; Kasenene, 1980) and were felled the
heaviest, leading to a patchwork of regrowth and relatively
large remnant forest tracts as was the case in lightly cut
forest. In the heavily cut forests, a combination of high
desirable tree species richness and predominance of glgg
welwitschii (a highly prized timber species) in large size
classes led to excessive exploitation, almost clear felling.

We have to recall the fact that in the reserve,
selective logging was uncontrolled (Skorupa and Kasenene,
1984). The sawmillers cut trees below the recommended girth
classes thus leading to extensive destruction of the canopy
and forest stand. Clearly, uncontrolled selective logging
results in the formation of larger gaps than are
ecologically beneficial for forest dynamics (Denslow, 1980).
Surprisingly enough, the resultant secondary forest was
expected to have naturally recovered and be ready for
another cycle of harvesting in only 70 years. However, 20
years later, the forest canopy in heavily selectively felled
forests was still wide open, allowing plenty of light into
the gaps while adequate regeneration of desirable tree

species appears to be in jeopardy.

216

Sapling gag Pole Eggeneration Relative 39 Forest gap giggg

Mechanized selective logging resulted in the formation
of large forest gaps quite different in size, configuration
and mode of formation from the small to medium natural tree
fall gaps. The differences between tree sapling and pole
regeneration in logged forest gaps and natural gaps could
provide useful information regarding the requirements for
regeneration of commercially desirable or primary forest
tree species (Pickett, 1983; Whitmore, 1975, 1978; Hart,
1986). Thus, forest gap regeneration of saplings and poles
in late gap phase-early building phase of the forest growth
cycle (Whitmore, 1975; Hartshorn, 1980; Brokaw, 1985) was
studied. Whitmore (1978) emphasized that because of the
overlap, the three gap phases are only abstractions and not
separate entities. However, the gap phase consists mainly
of juvenile trees, seedlings and saplings while the building
phase is mainly a pole forest which grows into the mature
phase of the cycle.

After forest gaps are formed there are four possible
modes of revegetation. In the first instance, the seedlings
and saplings present in the shaded forest understory, if
relatively undamaged by the disturbance, contribute
substantially to forest regeneration (Ewel, 1977; Pickett,
1983; Uhl, 1983; Augspurger, 1984; this study). Secondly,
the sprouts from cut or crushed stems of seedlings and
saplings produce new growth for gap regeneration (Uhl, 1983;

this study). Thirdly, plants recolonise the forest gaps by

217

the germination of seeds in the soil, where the majority of
seeds may be of pioneer or weed species. Fourthly, the
seeds from surrounding forest or distant trees may be
dispersed into the gaps for tree regeneration (Richards,
1952; Whitmore, 1975, 1978; Putz, 1983; Augspurger, 1984).

Most studies of tropical forest gaps and forest
regeneration, and the results of this study, show that
different tree species succeed in gaps of different sizes
(Knight, 1974; Hartshorn, 1978, 1980; Whitmore, 1975, 1978;
Denslow, 1980; Pickett, 1983). There is also a growing
consensus that successful regeneration of many forest trees
may actually depend on the combined influences of gap size,
gap configuration and mode of gap formation (Whitmore, 1975,
1978; Hartshorn, 1978; Denslow, 1980; Pickett, 1983; this
study). The strongly inverse relationship between gap—size
or intensity of forest exploitation and the species richness
and density of saplings and poles, for all tree species in
general and desirable tree species in particular, seems to
support this View.

As this study indicates, small to medium forest gaps
caused by natural tree falls have been observed to favor the
growth and success of seedlings and saplings of advance
regeneration or suppressed juveniles of primary forest
species in the understory (Richards, 1964; Knight, 1974;
Hartshorn, 1978; Whitmore, 1975, 1978; Gomez-Pompa g3 El;
1972). However, in very large gaps, advance regeneration

may often respond with reduced growth and eventual death

 

218

(Richards, 1964; Burgess, 1970; Meijer, 1970; Liew and Wong,
1973; Whitmore, 1975, 1978; Sousa, 1984; this study), while
pioneer plants invade and grow vigorously (Eggeling, 1947;
Meijer, 1970; Whitmore, 1975, 1978, 1980; Synnott, 1975;
Ewel, 1980; Putz, 1983; Brokaw, 1985). Brokaw (1982b,
1985), while studying gap phase regeneration on Barro
Colorado Island, Panama, observed dramatic shifts in density
and heavy mortality of primary forest seedling and sapling
species in forest gaps >150m2. Whitmore (1975, 1978),
observed successful growth of surviving individuals of
primary forest tree species in artificially generated gaps
of 0.1ha while they were suppressed by lush growth of
secondary forest pioneers in 0.2 to 0.3ha gaps.

Other studies too have indicated significant reductions
in species diversity of seedlings and saplings following
selective forest felling operations and point to post
felling stress and/or ecological instability as the major
cause (Westhoof, 1971; Poore, 1976; Herbert and Beveridge,
1977; Odum, 1985). It is not fully understood as to which
gap conditions stimulate the seedlings and saplings of
canopy tree species to commence and continue growth upwards
(Whitmore, 1975, 1978; Pickett, 1983). These conditions are
believed to vary from species to species (Lugo, 1970; Sousa,
1984; Marquis g3 al;, 1986) or even for single species at
different stages of development.

For the majority of forest gaps, including those in

this study, the principal environmental change that could

 

 

219

easily influence seedlings and sapling regeneration was
alteration of the radiation balance (Whitmore, 1975, 197a;
Pickett, 1983; Lee, 1978, 1987). This has also been shown
to initiate changes in a number of related factors embracing
the physical, chemical and biological environments important
for gap seedling and sapling regeneration (Armson g3 al;,
1973; Whitmore, 1978; Bullock, 1980 and Putz, 1983).

Amounts of daylight reaching into the closed tropical forest
have been estimated at 2% of ambient light (Pickett, 1983)
and large forest gaps receive between 50 and 100% of ambient
daylight. In addition, the duration of insolation, which is
variable for the small and large gaps, may also influence
forest gap tree regeneration. Usually, small forest gaps
caused by natural tree falls are directly insolated for
approximately 2hrs per day compared to 6 or 8hrs for large
clearings (Pickett, 1983; Lee, 1978, 1987).

Generally, the seedlings and saplings of primary forest
species prefer natural habitats of low light conditions and
have adaptive traits for prolonged suppression and recovery
after a gap is created above them (Richards 1952, 1964;
Gomez—Pompa gt gig, 1972; Whitmore, 1975, 1978; Pickett,
1983). Gap creation increases the availability of
resources, including light which stimulates increased
performance of either advance regeneration or pioneer and
weed species depending on the size of the gap (Whitmore,
1978; Pickett, 1983). Therefore, the fact that shade

tolerant species require optimal light intensities for

 

220

growth and survival and that the optimas differ among
species (Lugo, 1970; Augspurger, 1984; Sousa, 1984; Marquis
_3 31;, 1986) may partly explain the reduced seedling,
sapling and pole regeneration in large gaps relative to
small or medium gaps. The majority of seedlings and
saplings of the logged desirable or primary forest species
do not usually find large forest gaps characteristic of
mechanised selective logging suitable for their own
regeneration (Whitmore, 1975, 1978; Sousa, 1984; Uhl and
Buschbacher, 1985; this study). The extremely low seedling,
sapling and pole regeneration, in terms of species richness,
density and number of individuals per species, for all tree
species in large gaps but abundant regeneration of the same
species in forest plots adjacent to the gaps, strongly
supports the preceding observations.

Differential logging intensities and associated forest
disturbances and arboricide treatments might also account
for the low regeneration potential, especially in the gaps
of heavily selectively cut forest. During mechanized
selective felling, sawmillers in Uganda usually cut most
potentially harvestable trees, felling many more than are
actually harvested (Skorupa and Kasenene, 1984). Excessive
felling of large trees alone directly affects the would-be
advance regeneration through crushes and burial of seedlings
and saplings in debris. Subsequently, bulldozers usually
follow, smothering and killing many seedlings and saplings

during the hauling of logs to extraction roads (Kasenene,

 

221

1980; Struhsaker, pers. comm.). Several factors interact
and make mechanized selective logging more destructive than
it was designed to be. It is clear that the heavily felled
forest tracts suffered the greatest disturbance and damage,
still reflected in extremely low levels of forest gap
regeneration 20 years later.

Furthermore, the poor correlations between gap size and
tree seedling, sapling and pole regeneration in gaps of
heavily felled and treated forest could suggest the
compounded effect of gap size, intensity of disturbance and
arboricide treatment. The magnitude of the differences in
forest tree regeneration in gaps of similar sizes in two
heavily cut forests, where one of them had post felling
chemical treatment to kill "weed trees", were too big to be
explained by natural variation only. The post felling
chemical treatments with arboricide could account for most
of the 37.6% depression of tree sapling and pole species in
treated forest gaps, relative to heavily cut but untreated
forest gaps. In concert with the predictions, the small to
medium gaps in unlogged mature forest had the highest
regeneration of tree seedlings, saplings and poles, while
the gaps in lightly cut forest were ranked second. The
level of tree seedling, sapling and pole regeneration in the
large forest gaps of heavily selectively cut forest were
ranked third and those in heavily felled and treated forest
a distant fourth. These findings were not unique to this

study. During an ecosystem recovery study in Amazonia, Uhl

 

222

53 gig, (1985) also observed that increased intensity of
forest disturbance changed the early vegetation succession
from primary forest tree species (cut treatment) to
successional woody species (cut and burned treatment) and to
forbs and grasses (cut and bulldozed treatment).

The low potential regeneration of desirable or canopy
tree species in large gaps has occasionally been blamed on
lack of adequate seed populations dispersed to the gaps
(Brasnett, 1946; Eggeling, 1947; Synnott, 1975 and this
study). Selective logging usually removes most seed bearing
trees, and leftovers of overmature and defective individuals
of desirable species which would otherwise provide seeds are
commonly frilled, poisoned and killed to help forest
regeneration (Forest Dept. Records, F/P). This so called
"refining" process might have contributed to serious
reductions in the local seed supply, thus resulting in low
potential for desirable tree species regeneration,
especially in gaps of heavily selectively felled forest
sites.

Although the recommended level of reduction of forest
canopy during selective felling lies between 40 and 50%
(Kingston, 1967; Uhl g3 gig, 1985), more extensive
reductions often occur because of the problems already
cited. Consequently, high light intensities reach into the
forest and gaps and encourage the growth and development of
a dense tangle of herbs, climbers and non-woody shrubs. The

strongly inverse relationship between the dense gap ground

 

223

story growth and the tree seedling and sapling regeneration
suggests strong competition and suppression of tree
regeneration by weeds. Earlier studies by Brasnett, (1946),
and Synnott (1975) in Budongo forest, Uganda, also showed
that the thick ground—vegetation cover that normally follows
forest felling was extremely inhibitory to successful tree
regeneration. They suggested strong competition for light
between the juvenile trees and the thick vegetation cover,
root competition for space and nutrients, and actual
physical damage of climbers and lianas (also Putz, 1984), to
be contributing to reduced regeneration of trees in major
forest gaps. The whole method of mechanized selective
timber harvesting, if uncontrolled and carelessly carried
out as has always been the case in Ugandan forest, can be
extremely damaging to the species-rich tropical forest. It
has the potential to impoverish and degrade the species
composition and spatial arrangements of future secondary
forest communities.

The Ideal Gap Size

 

In the reserve 100% of the forest gaps formed by
natural tree falls and 83% of those formed by light forest
cutting were between 75 m2 and 700 m2 in area. The mean
gap size in lightly cut forest (467 m2), and the gap size
range of naturally formed gaps in mature forest (loo-663
m2), were similar to commonly formed natural gaps in

tropical forests of other regions (Whitmore, 1975, 1978;

Poore, 1968; Hartshorn, 1978; Brokaw, 1985). In contrast, a

 

224

lower percentage of forest gaps formed by heavy selective
felling (40% in K-15 and 47% in K—13) fell within the same
gap size range as natural tree fall gaps; the larger
percentage (60% in K-15 and 53% in K-13) were over 700 m2.
The gap sizes with the highest tree sapling and pole

regeneration in terms of species richness, density and
number of individuals per species for uncut, lightly cut and
heavily cut forest tracts were all below 650 m2 in size.
Tree seedling, sapling and pole regeneration for all species
and desirable or primary forest species was low and followed

a downward trend in forest gaps exceeding 650 m2.

Therefore, for the reserve, forest gaps between 500 m2 and
600 m2 seem to form the upper limits of forest gap sizes
where after a disturbance, successful regeneration of trees
including the desirable timber or primary forest species
could be expected naturally. Therefore, if a maximum of
four distantly located canopy trees are to be selectively
harvested per hectare of forest, then the total gap area
created by felling should never exceed 2400 m2/ha. This
also implies that four adjacent mature trees should never be
felled at the same time in the same felling cycle, in order
to avoid the detrimental effects of large gaps on forest
regeneration.

In earlier studies, timber extraction through strictly
controlled polycyclic systems that create 3 to 6 canopy gaps

per hectare have been found almost akin to natural tree fall

disturbance and result in the formation of small gaps in the

225

forest canopy (Whitmore, 1975; Herbert and Beveridge, 1977;
Uhl 23 gig, 1985) and the damage done to the forest is
relatively minor. In the lightly cut forest, removal of 3
to 5 canopy trees per hectare seems to have had little
effect on natural forest regeneration. The lightly cut and
uncut mature forest tracts of the reserve have approximately
3 to 4 dead-standing canopy trees/ha. The removal of four
canopy trees/ha may also resemble natural tree death, and
result in the formation of small gaps in the forest canopy
and minor damage to the forest. Previously established
seedlings and saplings of primary forest species have often
been observed to grow to maturity in small to medium sized
natural forest gaps where weedy species are naturally
excluded (Richards, 1964; Knight, 1974; Hartshorn, 1978;
Whitmore, 1978; Pickett, 1983 and Brokaw, 1985).

Herbivory and Seed Predation in Forest Gaps

 

Compared to temperate forests, tropical evergreen or
semi-deciduous forests do not provide strong, regular
seasonal pulses of resources (Muller, 1978) and forest gaps
might provide important although partially unpredictable
pulses (Pickett, 1983). Therefore, forest gaps might
attract animals which interact either positivelyor
negatively with gap plant regeneration (Wing and Buss, 1970;
Janzen, 1970, 1971; Evans, 1974; Kasenene, 1980; Pickett,
1983; and this study). The luxuriant ground vegetation that
normally follows heavy selective logging provides palatable

fodder, cover, and ideal nesting sites for animals and

 

226

insects (Wing and Buss, 1970; Isabirye-Basuta, 1979;
Kasenene, 1980, 1984; Merz, 1983; Short, 1983). Evans
(1974) has shown that insect populations usually increase in
forests which have been degraded and left to regenerate
naturally. Insects are known to damage substantial amounts
of seeds both in pre and post dispersal condition (Janzen,
1971). Although insect damage to forest tree seeds and
seedlings is not easily discernible, Janzen (1970, 1971),
Synnott (1975), Bradford g; a}; (1977) and Flowerdew and
Gardner (1978) have demonstrated that insects can destroy
significant amounts of seeds in a single seed crop. Insects
are ubiquitous, highly diverse in food habits and several
magnitudes greater in population density than any of the
common mammalian seed predators (Orians 33 gig, 1974;
Struhsaker, and Matti—Nummelin, pers. comm.).

Like insects, forest rodents are also ubiquitous and
can sustain high population densities in modified forest
environments (Delany, 1971; Fleming, 1975; Jeffrey, 1977;
Isabirye-Basuta, 1979; Kasenene, 1980, 1984). Modification
of primary forest by felling has often been associated with
drastic increases in small rodent diversity, population
densities and biomass (Delany, 1971; Jeffrey, 1977;
Isabirye-Basuta, 1979; Kasenene, 1980). Associated rodent
increases vary with the type and degree of modification of
the primary forest (Delany, 1971; Jeffrey, 1977) and
extensive changes in forest ecology following felling are

presumed to be responsible for the changes in rodent

 

227

population dynamics. As in Delany (1971) and Synnott
(1975), earlier studies by the author (Kasenene, 1980, 1984)
and Isabirye—Basuta (1979) on rodent ecology and biology in
the Kibale forest also show that heavily selectively felled
and regenerating forests support higher rodent species
diversity and density than lightly cut and uncut forest
tracts. Earlier studies in the reserve (Isabirye-Basuta,
1979; Kasenene, 1980, 1984) and those of Delany (1975),
Fleming (1975), and Synnott (1975) clearly demonstrated that
most rodent species were ecologically important forest seed
as well as young seedling predators. Seedlings were
selectively browsed and young seedlings with cotyledons
still attached were more susceptible to rodent predation and
more easily killed than other seedlings (Kasenene, 1980,
1984).

The results of this study and several previous studies
(Janzen, 1970, 1971; Delany, 1971: Synnott, 1975, 1977;
Field, 1975; Bradford gt git, 1977; Isabirye—Basuta, 1979;
Kasenene, 1984) emphasize the economic and ecological
importance of rodent seed predation with regard to forest
regeneration. Synnott (1975) observed very high mortality
rates among the seeds and seedlings of Entandrophragma gtglg
due to predation by rodents. Bradford gt El; (1977); also
found that rodents could very easily damage 75% of Scheelea
palm seeds in a single seed crop. Flowerdew gt El; (1978)
attributed the major portion (75%) of the loss of Ash seeds

prior to germination to small rodents. The preliminary

 

228

results of the forest gap seed pile predation trials
strongly suggest that rodent seed predation can be more
significant in the large gaps of the heavily selectively
felled forests than in the small to medium gaps in uncut or
lightly cut forests. Where heavy timber exploitation has
occurred, the seed supplies and seed populations of
desirable tree species could be locally inadequate (Synnott,
1975). Therefore the characteristic increases in rodent
populations and density (due to invasion and multiplication
of colonizing species) following felling might drastically
decrease the seed and seedling populations, thus limiting
adequate natural regeneration. However, several factors
including seed pile size, seed chemical defense, presence of
other preferred foods and season of seed placement might
have also influenced the results and rates of seed predation
in forest gaps (Janzen, 1978; Wilson and Janzen, 1972, Ng
1978, Kasenene, 1980).

Very few studies have focused on the influence of big
game on the composition, structure and regeneration of
tropical forests. In the Ituri forest of Zaire (Hart,
1986), the large forest gaps were frequently visited by
Okapi (Okapia johnsoni) and none of the marked pioneer
seedlings were present after two years of observation. In a
large scale survey of the Kibale forest, Uganda, Osmaston
(1959a) and Wing and Buss (1970) found no significant effect
of big game such as elephants (Loxodonta africana

Blumenbach) on the forest structure and composition.

229

However, the studies of Short (1983) and Merz (1986) in West
African forests show that big game, including elephants,
prefer secondary forests and/or forest gaps (Hart, 1986) to
undisturbed mature forests. All these observations agree
with the preliminary findings of this study and my earlier
observations (Kasenene, 1980, 1984). It is probable that
the luxuriant ground vegetation that develops following
forest exploitation provides attractive and palatable fodder
for the elephants for they frequent and browse in large
forest gaps more often than in small tree fall gaps in
mature forest (Kasenene, 1980, 1984, this study).

Eggeling (1947) found that elephants could be a
limiting factor in the natural development of vegetation
wherever they occur in big numbers. While browsing, they
destroy woody plants in the seedling, sapling and even pole
stage by a combination of methods including trampling,
defoliation, topping, uprooting and barking (Eggeling, 1947;
Kasenene, 1964; this study). Therefore, elephants can be
important ecological pests of regeneration areas. Their
frequent incursions into secondary forests and gaps
encourage the perpetuation of the herbaceous tangle
"Elephant climax" (Wing and Buss, 1970) where juvenile trees
are smothered by climbers and sometimes killed, thus
impeding forest regeneration. Therefore, the economic and
ecological importance of browsers such as elephants on the
forests should be assessed with respect to specific

interactions with forest regeneration and consideration of

 

230

all discernible forms of animal damage to trees. In the
case of elephants, we should also note that intensity of
utilization of forest varies with the degree of modification
or disturbance of the primary forest.

The large forest gaps in heavily selectively felled
forests were preferred and more frequently used by elephants
than small to medium size gaps in lightly cut or uncut
forest tracts. This implies that the impact of elephants on
forest gap regeneration could be controlled if the methods
of forest exploitation were controlled to create small to
medium gaps. After all, in addition to fertilizing the
forests with dung-manure, elephants form an indispensable
forest component as large seed dispersers. However, my data
on herbivory and seed predation in forest gaps are still
limited and fragmentary and pose many problems, but I feel
they demonstrate the potential impacts animals can have on
large forest gap regeneration and emphasize the need for
long term research.

Sources gt Plant Recruitment into Forest Gaps

In the study of soil seed banks, germination is usually

selected for evaluation of soil seed crop instead of direct

counts and identification of seeds in the soil which are

extremely tenuous and difficult (Guevara & Gomez-Pompa,
1972). The raised and screened germination stand or house
used in this study was meant to protect the soil samples
from seed contamination and seed and seedling predation.

Earlier studies and the results of this study all indicate

231

the presence of significant dormant seed pools beneath
tropical mature forests (Gomez—Pompa gt git, 1972; Guevara
and Gomez-Pompa, 1972; Harcombe, 1977; Ewel gt git, 1981;
Pickett, 1983; and Young, 1985, among others). The number
of seeds stored in the soil often referred to as seedcrop,
seedpool, seedbanks or floristic potential (Guevara and
Gomez-Pompa, 1972; Pickett, 1983), are an important factor
in the initiation of secondary succession after forest
clearance (Guevara and Gomez-Pompa, 1972; Augspurger, 1984;
Young, 1985). However, the results of most studies,
including this one, clearly show that by far the most
important floristic components of the soil seed banks are
large-gap (pioneer, weed or secondary) species (Guevara and
Gomez-Pompa, 1972; Ewel gt git, 1981; Pickett, 1983).

It is difficult to compare the populations of viable
seeds in tropical forest soils from the same or different
regions, for there are no standard experimental procedures
(Ewel, 1981; Young, 1985). Most often, the germination
experiments by different researchers have involved different
numbers and sizes of samples, different durations of
germination trials, and most importantly, different soil
sample depths (Table 8). However, the results clearly
suggest that soil seed banks of different tropical forests
were highly variable in plant species richness (Table 8).
Again, the soil seed banks of forests which had had some
forms of human disturbance had even higher species richness

than those under mature forest. But where there had been

232

 

 

bps»: case use m9 o~lc voqaou h~fi>¢o=
hvaan made «no u“ out: teqqou hues-«A
muse: u«£b mu» m o—no anouou Guava:
.0900 aoouom
buses mask can an @910 vauqou huu>ao=
5650. Odds can nu onlo vouqou huasuua
an«0l ~d0«9oua
sun». case .6» «a cane ovsaauua gnu: ocean:
gmo— coon en cosmuam unashaa +
«sou coco 9o oocuhsm leasedsl +
«0&9 #00909
adt mu dozm coca pm coouusm hhlvnooom seam cameo
pals:
macavwoov
cow" can nv «no «lacunae 0:056
one— ocudzm nsoapwoov
d Add: mam ow vno «lo-nunmoz mango
«us an «.mno angel
.coau haul can u— n.clo ~¢o~9oub dauouuz
mama ua59 «vs av auto ~00¢9ous Oldcd9
Nua—
¢9lo9uu0loc amonopu n— «9.0 coouwho>o
a sucrose «couvvo on Nuns an“: 009x02
NI .99. Ale.
00:0u0902 \Ivoom no .02 A8909 o9hu auouom nowadooa

 

.uauouow cash

mucu9oua Olom hon monsoon :0«0¢:«luou too: qqon uo conuuc9loo o Quack

233

maximum forest disturbance, both natural and artificial,
e.g. in forest gaps, the soil seed banks were extremely
impoverished in species richness compared to those under
mature or secondary forest. However, the low species
richness in the gaps is compensated for by high population
densities.

As with species richness, the mean number of seeds
germinated from soils sampled from beneath the forests were
also variable for the different tropical areas. Soil
samples from tropical second growth forest and agricultural
areas had generally higher numbers of viable seeds than
soils from mature forest sites. However, all the values of
number of viable seeds/m2 indicated in the table were
considerably lower than the value reported from 3—year—old
vegetation on abandoned farm land in the Everglades of south
Florida, 30,125 seeds/m2 (Ewel and Conde, 1979). Their
average values for the mature communities were also high,
ranging from 500 to 6000 seeds/m2. These observations
therefore suggest that the dormant seeds in the soil seed
banks under forest conditions germinate in response to
improved resources in the gaps. However, most of the
germinated plant species are survived by only a few which
successfully establish themselves in the gaps. And the gap
soil seed banks continue to be under heavier seed rain,
mainly from the few gap species that almost exclusively

establish themselves in the new sites or gaps.

234

In the Kibale the preceding observations were strongly
supported by the preponderance of herbaceous plants, vines
and climbers, and non—woody shrubs that germinated from the
surface soils of manipulated plots located within the forest
gaps. The contribution by incoming dispersal to large gap
tree seedling and sapling recruitment was very small and
extremely lacking in seeds of primary forest species. It
seemed that only a few secondary forest trees on the
periphery of the gaps were able to disperse their seeds into
the gap plots although only a few seeds actually germinated
and contributed to the seedling pools in the manipulated
plots. However, for the whole 20-month period of
observation, no new tree seedlings were observed to
germinate and establish in the gap plots with all vegetation
left intact. Again, the tree seedlings that were able to
germinate in the forest gap plots with vegetation removed,
vegetation and top soil removed, and previously screened
plots with screens removed six months later, did not survive
20 months of observation. Clearly these observations
suggest high seed and seedling mortality, which could have
been caused by predation or over exposure to high insolation

in large forest gaps. Furthermore, the dense gap vegetation

of herbs, shrubs, vines and climbers competitively excludes
or suppresses the germination and establishment of all tree
species including pioneer, secondary and primary forest

species.

235

The most striking observation about all soil seed banks
for different tropical rainforests, including those from the
reserve, was the lack of primary forest or canopy tree
species (Guevara and Gomez—Pompa, 1972; Ewel gt git, 1981;
Pickett, 1983; Augspurger, 1984). It is believed that lack
of seed dormancy among most primary forest species causes
them to germinate within a few weeks of shedding (Whitmore,
1975, 1978; Ng, 1978), and this accounts for their scarcity
in forest soil seed banks. However, under certain
circumstances, this quick germination is advantageous in
ensuring survival power of such species through avoidance of
seed predation and fungal attack (Gomez-Pompa gt git, 1972).
Again, a combination of other factors, including temporal
variation in fruiting behavior (e.g. Whitmore, 1975), low
seed population, and large seed size (relatively inefficient
for dispersal) could have made some primary forest species
unavailable for large forest gaps, especially where forest
exploitation selected for particular prime species.

The failure of primary forest tree species to invade
and regenerate in large forest disturbances in the Ituri
forest of Zaire (Hart, 1986) also suggested several
controlling factors. Hart (1986) found that even after 40
years of abandonment, the seedlings of shade tolerant
dominants from mature forest had not invaded the secondary
growth and abandoned garden sites, although the remnant
mother trees were no more than several meters away. It is

inevitable that primary forest species build up seedling and

 

236

sapling populations within the shaded understory until gaps
are formed, allowing further growth (Pickett, 1983;
Augspurger, 1984; this study). Therefore, the importance of
seed dormancy and seed banks in forest regeneration after a
large disturbance is primarily restricted topioneers or
large-gap species while the regeneration of small gap or
primary forest species is mainly by suppressed seedling and
sapling pools in the shaded forest understory.

In the reserve, the majority of seedlings and saplings
of canopy tree species that were regenerating in the forest
gaps were also found in the understory of adjacent forest
tracts. This further supports earlier observations that
non-pioneering primary forest species were limited to
suppressed seedlings as the primary mode of regeneration
after major disturbances. However, these seedlings and
saplings suffered great stem damage, probably during gap
formation, and were subject to even more damaging episodes
after gap formation.

The majority of canopy tree seedlings, saplings and
poles in large forest gaps were classed as coppices because
they had had several well spaced stem breaks or snaps and

successful coppicing. The higher frequency of well spaced

stem breaks and coppicing for seedlings and saplings in
forest gaps than in adjacent forest understory suggested
frequent episodes of forest gap seedling and sapling

disturbance. The seedlings and saplings in large forest

gaps were more susceptible to damage, most of which could

237

be attributed to frequent tree and branch falls into the
gaps or animal browsing.

However, the most important observation was that the
majority of seedlings and saplings of desirable timber or
primary forest species, in the tropical moist forest
studied, demonstrated great ability to recover by stem
coppices or sprouts after a disturbance and injury.
Therefore, the low populations of seedlings and saplings of
primary forest tree species in large forest gaps suggest
their inability to survive in extremely open gap conditions.
poor competition with the aggressive weeds that take over
the gaps, or destruction through severe damage, such as
complete defoliation and barking or uprooting. Thus the
results of soil seed bank studies and the observed mode of
tree seedling and sapling regeneration all indicate that a
primary forest cannot be restored if all trees from adjacent
areas are destroyed. If it is absolutely necessary to
harvest a forest, protection of suppressed seedlings and
saplings of primary forest species should be given priority
for the successful regeneration of the desirable tree

species in the secondary forest.

SUMMARY

In Uganda, the three major threats to the integrity of
tropical forest resources include agricultural encroachment,
uncontrolled timber exploitation, and rampant poaching of
game. However, this study concentrated on the influence of
mechanized selective felling on the development of forest
gaps and the consequent ecological effects of gaps, on the
regeneration of tree species.

A total of 40 randomly selected gaps were examined in
each of the four forest compartments comprising the uncut
mature forest (K-30), lightly cut forest (K-14), heavily
felled forest (K-15), and heavily felled and treated forest
(K-13). In selectively felled forest, gaps with cut tree
stumps and/or waste tree trunks were studied, while in
mature forest, gaps which had tree or big branch falls and
characteristic gap vegetation of size ranges (height and
dbh.) similar to that in selectively cut forest gaps were

considered. Since it had been almost 20 years since

felling, the majority of forest gaps selected were in the

late gap phase-early building phase stage of gap succession,

characterized mainly by tree seedlings, saplings and poles.
2

The mean gap size for lightly cut forest was 467 m

(range 75 to 1800 m2) and the natural tree fall gaps in

238

 

239

mature forest averaged 256 m2 (range 100 to 663 m2). The
mean gap size in heavily cut forest was 1307 m2 (range 73 to
7100 m2) while that for heavily felled and treated forest
was 938 m2 (range 227 to 3313 m2). An attempt was made to
sample 40% of each gap area. All tree saplings and poles
encountered within a 5m wide belt transect following the cut
access trails were enumerated, measured (dbh), and
identified to species level and mode of origin (coppices,
stem sprouts or normally growing). The influence of
herbivory on forest gap tree regeneration was assessed from
monthly recordings of animal incursions and the nature of
associated damage (e.g. trampling, browsing of tree
seedlings, saplings and poles). Seed predation trials were
also run for some primary forest tree species to test the
fate of seeds that come into the forest gaps through
dispersal. Finally, other potential sources of recruitment
into forest gaps, including soil seed banks, were examined
by soil seed germination experiments in protected nursery
beds.

Multiple tree removals during selective felling and
consequent wind and storm throws in felled forests resulted
in the formation of gaps larger than the small to medium
gaps characteristics of natural tree falls. All forest
gaps created by natural tree falls and 83% of those produced
by light forest cutting ranged between 75 and 700 m2 in
size. These were similar in size to most commonly formed

natural gaps observed in other forests in different tropical

240

regions. However, only a small percentage of forest gaps
formed by heavy selective felling (40% and 47% in K-15 and
K—13, respectively) were within the natural tree fall gap
size range while 53 to 60% included large gaps greater than
800 m2.

Success of tree seedling and sapling regeneration was
inversely related to gap size. Tree seedling, sapling and
pole regeneration was high in gap sizes below 650 m2 but
extremely low, impoverished or suppressed in forest gaps
greater than 650 m2. Consequently, the large forest gaps
that resulted from heavy selective felling and associated
disturbances had the poorest levels of regeneration of tree
species.

However, gap size alone did not seem to account for the
extremely low regeneration of tree species in the extensive
gaps from heavy selective logging. Heavy logging definitely
changed the structure and composition of the mature forest
through the removal of most primary forest and large seed
bearing trees. Logging intensity was related to degree of
physical damage and disturbance to suppressed seedling and
saplings in the forest understory. The post harvest
chemical treatments and resultant deaths of treated non-
commercial and overmature, or defective primary forest trees
also seem to have contributed significantly to reduced gap
regeneration of all tree species and primary forest species

in particular. It has not been ascertained whether the

arboricide mixture used in spraying (3% Finopal D/T, i.e.

 

 

241

0.5 liter of Finopal D/T per 18 liters of diesel) was
species specific or easily biologically degradable. Other
factors germane to heavy selective felling, such as dense
gap ground vegetation cover, rodent seed predation, and
destructive browsing by big game, may have played a larger
role in the reduction of tree regeneration in large forest
gaps than in small to medium natural tree fall gaps.

The soil seed banks from underneath the forest canopy
and in forest gaps in cut and uncut forests were dominated
by large-gap plants, including pioneer, herbaceous, shrub,
and vine species. 0f the seedlings germinated from forest
soils of K-14, K-15 and K—30, only 10.3%, 9.3% and 16% were
of tree species while herbaceous plants formed 80.4%, 71.2%
and 53.1%, respectively. The density of the sprouted
seedlings followed the same trend. For the gap soil seed
banks in K—14, K-15, and K-30, tree species formed 11.9%,
8.9% and 17.7% of the germinated seedling species,
respectively. The density of the sprouted tree seedlings in
K-14, K-15 and K-30 soils formed only 0.3%, 0.3% and 4.9%,
respectively, and the herbaceous plants constituted the
largest remaining percentages. Except for gt africana and
Qt millenii, which germinated from the forest soil seed
banks and M; bagshawei and gt toxicaria in the manipulated
gap plots in K—30 only, the majority of tree seedling
species which sprouted from the forest and gap soil seed
banks belonged to a few non-desirable secondary forest

species. The seeds of the only clearly defined gap

 

242

specialist and pioneer tree species in the reserve, Ttgmg
guineensis, were well represented in both forest and gap
soil seed banks. The only predictable mode of successful
regeneration of desirable timber or primary forest species
after felling disturbance was the release of suppressed

seedlings and sapling pools in the forest understory.

 

CONCLUSIONS

When carried out with great care, creating on the
average 2 to 4 canopy gaps/ha, selective tree harvesting
could be a minor disturbance, akin to natural tree falls
which are an important component of the tropical forest
dynamics. The disturbances associated with selective
felling are supposed to be moderate enough to open a gap but
not so much that the suppressed understory tree seedlings
and saplings are killed. However, respect for the
recommendations for logging and careful logging practices
are the exception in tropical rain forests. Thus,
mechanised felling in species rich, and uneven growth,
tropical moist forest is incompatible with minimizing
damage. Large forest gaps of a size that hinder the
regeneration of most desired or primary forest species are
created. Available evidence suggests that mechanized
selective timber harvesting has detrimental effects on
tropical forest ecosystems by causing major disturbances and
destruction that lead to serious reductions in diversity of
tree species, including the desirable timber or canopy tree
species, and even the herbaceous and non-woody shrubs. The
following list includes a few consequences of uncontrolled

selective logging:

243

 

244

- drastic changes in the structure and composition of
the mature forest;

- extensive destruction and reduction of the forest
canopy;

- intensive and extensive forest disturbances;

- increased tree mortality through windthrows and
widening of gaps;

- formations of unnaturally large forest gaps;

- increased daylight penetration, intensity and
duration in the forest gaps;

- development of dense and persistent gap vegetation
composed of herbs, shrubs, vines and climbers;

- intense competition between gap vegetation and tree
seedlings and saplings;

- increased seed and seedling predation, mostly by
rodents;

- increased frequency of gap utilization by browsers
thus causing sapling destruction;

- reduced tree and pole species diversity;

- reduced tree seedling and sapling species diversity;

- impoverished forest regeneration in general; and

- very low regeneration of desirable timber or primary
forest species.

These responses seem to be gradually jeopardizing the

structure, condition, stability and appearance of the
residual secondary forest community. Twenty years after the

felling operations, there appears to be no positive sign of

 

245

recolonization of the large forest gaps by primary forest
species. The gaps are open, densely covered with herbs,
vines and climbers and non-woody and woody shrubs and are
thus still subject to frequent disturbances by browsers. In
the most heavily exploited areas of the reserve it appears
very unlikely that another harvest of valuable timber
species could take place, as previously anticipated
(Kingston, 1967), 70 years from the last forest harvest.
This makes uncertain the very first management objective of
establishing the forest reserve, i.e. to realize maximum
economic, sustained production of natural hardwood timber.
The time required for degraded forests to recover varies
depending on the intensity of disturbance, and most studies
(e.g. Uhl gt El; 1982, 1985) suggest very long periods
stretching from hundreds to thousands of years. Therefore,
the prognosis is that in the Kibale tropical moist forest
reserve, extremely long periods of time, longer than the
proposed 70 years of the harvest cycle, will be required for
anything resembling a forest worthy of exploitation to
reappear, after large, high intensity disturbances.
Therefore, whenever it is necessary to harvest the
forest reserve or any other forest, more.effort should be
put into minimizing the impact of exploitation to reduce the
need for intensive silviculture to ensure future timber
harvests. The methods employed in the exploitation of
species-rich tropical moist forest should emphasize low

disturbance levels where small to medium forest gaps are

246

created. This will partially ensure protection of the
suppressed seedling and sapling pools of the primary forest
species and enhance the successful natural regeneration of
the desirable timber species in the secondary forest. Since
mechanized timber exploitation is incompatible with
minimizing destruction and disturbances in species-rich
tropical forests, it should be discouraged. Pitsawing and
other forms of polycyclic felling systems that are more
labor intensive than machine-based systems, and which would
create small to medium gaps akin to natural tree falls,
could if properly controlled be very compatible with long
term sustained yield timber exploitation. Meanwhile, a
modified form of the polycyclic felling system, slightly
more advanced than pitsawing, should be studied and
developed. The basis for forest exploitation should be to
create forest gaps of a size that favor the growth and
development of desirable timber or primary forest species
without creating the need for intensive post felling
silvicultural treatments. Such treatments would be costly,
and their implementation uncertain, due to the usual lack of

capital and adequately trained and interested man power.

P}?

 

LITERATURE CITED

Armson, K. A. and R. J. Fesseden. 1973. Forest windthrows
and their influence on soil morphology. Soil Sci. Soc.
Amer. Proc. 37:781-784.

Augspurger, C. K. 1984. Light requirements of Neotropical
tree seedlings: A comparative study of growth and
survival. J. of Ecol. 72:777-795.

Bradford, D. F. and C. C. Smith. 1977. Seed predation and
seed number in Scheelea palm fruits. Ecology 58:667-
673.

Brasnett, N. V. 1946. The Budongo forest, Uganda. J. Oxf.
Univ. For. Soc. 3(1):11-17.

Brazier, J. D.; J. F. Hughes and C. B. Tabb. 1976.
Exploitation of natural tropical forest resources and
the need for genetic and ecological conservation. In:
Tropical trees: Variation, Breeding, Conservation. pp.
1-10 (eds. J. Burley and B. T. Styles). Linnean
Society. Symposium Series No. 2. Academic Press.

Brenan, J. P. M. 1973. A summary. In: Tropical Ecosystems in
Africa and South America: A comparative review. pp. 1-8
(eds. B. J. Meggers, E. S. Ayensu and W. D. Duckworth).
Smithsonian Institution Press, Washington, D.C.

Brokaw, N. V. L. 1982a. Treefalls: frequency, timing and
consequences. In: The Ecology of a tropical forest:
Seasonal rhythms and long—term changes.pp. 101-108
(eds. E. G. Leigh, Jr. A. S. Rand and D. M. Windsor).
Smithsonian Institution Press, Washington D.C.

1982b. The definition of tree fall gap and its
effect on measures of forest dynamics. Biotropica
14:158—160.

 

. 1985. Gap phase regeneration in a tropical
forest. Ecology 66(3):682-687.

 

Bullock, S. H. 1980. Impact of logging on littoral Cameroon.
Commonwealth Forestry Review 59:208-209.

Burgess, P. F. 1970. An approach towards a silvicultural

system for the hill forests of the Malay Peninsula.
Malay Forester 33:126-134.

247

 

248

Chapman, E. C. 1975. Shifting agriculture in tropical forest
areas of S. E. Asia. In: IUNC (1975) The use of
Ecological guidelines for development in tropical
forest areas of S. E. Asia. IUNC. Series No. 32.
Morges, Switzerland.

Delany, M. J. 1971. The biology of small rodents in Mayanja
forest, Uganda. J. Zool. Lond. 165 85—129.

1975. The rodents of Uganda. Trustees of the
British Museum (Natural History) London.

Denslow, J. S. 1980. Gap partitioning among tropical
rainforest trees. Biotropica 12 (Suppl.):47—55.

Dittus, W. P. J. 1985. Influence of cyclones on the dry
evergreen forest of Sri-Lanka. Biotropica 17(1)1-14.

Downes, R. G. 1974. Land use planning in economic
development. In: IUCN (1975), the use of ecological
guidelines for development in tropical forest areas of
south East Asia. IUNC. Series No. 32. Morges,
Switzerland.

Eggeling, W. J. 1947. Observations on the ecology of the
Budongo rainforest, Uganda. J. Ecol. 34:20-87.

, and I. R. Dale. 1951. The indigenous trees of the
Uganda Protectorate. 2nd Edition. The Government
Printer, Entebbe, Uganda. 491 pp.

Evans, H. C. 1974. Natural control of arthropods with
special reference to ants (Formicidae) by fungi in the
tropical high forests of Ghana. J. Appl. Ecol. 2(1):37—
49.

Ewel, J. J. 1977. Differences between wet and dry
successional tropical ecosystems. Geo-Eco. Trop.
1(2)163—177.

1980. Tropical succession: manifold routes to
maturity. Biotropica 12(suppl.)2-7.

Ewel, J. J.; C. Berish, B. Brown, N. Rice and J. Raich.
1981. Slash and burn impacts on a Costa Rican wet
forest site. Ecology 62:816-829.

Ewel, J. J. and L. Conde. 1976. Potential ecological impacts
of increased intensity of tropical forest utilization.
Final Report to USDA. Forest Service. Forest Products
Laboratory, Madison Wisconsin.

 

249

Field, A. C. 1975. Seasonal changes in reproduction, diet
and body composition of two equatorial rodents. E. Afr.
Wildl. J. 13:221-235.

Fleming, T. H. 1975. The role of small mammals in tropical
ecosystems. In: Small mammals, their productivity and
population dynamics. (eds. F. B. Golley, K. Petrusewicz
and L. Ryszkowski) Cambridge University Press.

Flowerdew, J. R. and G. Gardner. 1978. Small rodent
populations and food supply in a Derbyshire Ash
Woodland J. Animal Ecol. 47:725-740.

Fosberg, F. R. 1973. Temperate zone influence on tropical
forest land use: A plea for sanity. In: Tropical
ecosystems in Africa and Southern America. A
comparative review. pp. 345—350 (eds. B. J. Meggers, E.
S. Ayensu and W. D. Duckworth). Smithsonian Institution
Press, Washington D.C.

Gomez-Pompa, A. C. Vazquez-Yanes and S. Guevara. 1972. The
tropical rainforest. A non renewable resource. Science
117:762—765.

Gomez-Pompa, A. A. L. Asada, F. Golley, G. Hartshorn, D.
Janzen, M. Kellman, L. Nevling, J. Penalosa, P.
Richards, C. Vasquez and P. Zinke. 1974. Recovery of
tropical ecosystems. In: Fragile ecosystems: Evaluation
of research and applications in the Neotropics. pp.
113~138 (eds. E. G. Farnworth and F. B. Golley).
Springer—Verlag, New York.

Guevara, S. S. and A. Gomez—Pompa. 1972. Seeds from surface
soils in a tropical region of Varacruz, Mexico. J.
Arnold Arbor. 53:312—335.

Hall, J. B. and M. D. Swaine. 1980. Seed stocks in Ghanaian
forest soils. Biotropica 12(4):256-263.

Hamilton, A. C. 1981. A field guide to Uganda forest trees.
Privately Published, Makerere University, Uganda.

1984. Deforestation in Uganda. Oxford University
Press. Nairobi.

Harcombe, P. A. 1977. The influence of fertilization on some
aspects of succession in a humid tropical forest.
Ecology, 58:1375-1383.

Hart, T. B. 1986. The ecology of a single-species—dominant

forest and of a mixed forest in Zaire, Africa. Ph. D.
dissertation, Michigan State University.

L‘s.

 

250

Hartshorn, G. S. 1978. Tree falls and tropical forest
dynamics. In: Tropical trees as living systems. pp.
617-638 (eds. P. B. Tomlinson and M. H. Zimmermann).
Cambridge University Press, Cambridge.

. 1980. Neotropical forest dynamics. Biotropica
12(Suppl.)23-30.

Herbert, J. and A. E. Beveridge. 1977. A selective logging
trial in dense podocarp forest in central North Island.
New Zealand J. of Forestry. 22:81-100.

Isabirye—Basuta, G. M. 1979. The ecology and biology of
small rodents in the Kibale forest, Uganda. M.Sc.
Thesis, Makerere University.

IUCN. 1975. The use of ecological guide lines for
development in tropical forest areas of S. E. Asia.
IUCN Publication: New Series No. 32. 1110 Morges
Switzerland. 186 pp.

Janzen, D. H. 1970. Herbivores and the number of tree
species in tropical forests. American Naturalist
104:501—528.

1971. Seed predation by animals. Ann. Rev. Ecol.
Syst. 2:465-492.

1978. Seeding patterns of tropical trees. In:
Tropical trees as living systems. (eds. P. B. Tomlinson
and M. H. Zimmermann) Cambridge University Press, New
York.

Kasenene, J. M. 1980. Plant regeneration and rodent
populations in selectively felled and unfelled areas of
the Kibale forest, Uganda. M.Sc. Thesis, Makerere
University.

1984. The influence of selective logging on
rodent populations and the regeneration of selected
tree species in the Kibale forest, Uganda. Trop. Ecol.
25:179-195.

Keay, R. W. J. 1960. Seeds in forest soils. Nigerian
Forestry Information Bulletin (New Series) 4 1—4.

Kingston, B. 1967. Working plan for the Kibale and Itwara
forest reserves. Uganda Forest Department, Entebbe. 142
PP-

 

—+—

251

Knight, D. H. 1974. An analysis of late secondary succession
in species rich tropical forest. In: Tropical
ecological systems: trends in terrestrial and aquatic
research. pp. 53-59 (eds. F. B. Golley and E. Medina).
Springer-Verlag, New York.

Lee, R. 1978. Forest Microclimatology. Columbia University
Press, New York.

Lee, D. W. 1987. The spectral distribution of radiation in
two Neotropical rainforests. Biotropica 19(2):161-166.

Lieberman, D., M. Lieberman, R. Peralta and G. Hartshown.
1985 Mortality patterns and stand turn over rates in a
wet tropical forest in Costa Rica. J. Ecol. 73:915-924.

Liew, T. C. and F. O. Wong. 1973. Density, recruitment
mortality and growth of dipterocarp seedlings in virgin
and loggedover forest in Sabah. Malay Forest 36:3-15.

Longman, K. A. and J. Jenik. 1974. Tropical forest and its
environment. Longman Group Ltd. London. 196pp.

Lugo, A. 1970. Photosynthetic studies on four species of
rainforest seedlings. In: Tropical Rainforests. (eds.
H. T. Odum and R. F. Pigeon). U.S. Atomic Energy
Commission, Div. Tech. Inf. Oak Ridge, Tennessee.

Marquis, R. J. and H. J. Young. 1986. The influence of
understory vegetation cover on germination and seedling
establishment in a tropical lowland wet forest.
Biotropica 18(4):273-278.

Meijer, W. 1970. Regeneration of tropical lowland forest in
Sabah, Malaysia, forty years after logging. Malay
Forester 33:204-229.

Merz, G. 1986. Counting elephants (toxodonta africana
Cyclotis) in tropical rainforest with particular
reference to Tai National Park, Ivory Coast. Afr. J.
Ecol. 24:61-68.

Muller, R. N. 1978. The phenology, growth and ecosystem
dynamics of Erythronium americanum in the Northern
hardwood forests. Ecological Monogr. 48:1-20.

Ng, F. S. P. 1978. Strategies of establishment in Malayan
forest trees. In: Tropical trees as living systems. pp.
129-162 (eds. P. B. Tomlinson and M. H. Zimmermann).
Cambridge University Press, New York.

Odum, E. P. 1963. Ecology-Modern Biological Series Holt:
Rinehart and Winston, New York 152pp.

 

 

 

 

 

252

1969. The Strategy of ecosystem development
science 164:262-270.

. 1985. Trends expected in stressed ecosystems
Bioscience 35(7):419—422.

Orians, A.; J. L. Apple, R. Billings, L. Faunier, L.
Gilberty, B. McNab, J. Sarukhan, N. Smith, G. Stiles,
and T. Yuill. 1974. Tropical Population ecology. In:
Fragile ecosystems: Evaluation of research and
application in the Neotropics. pp. 5-65 (eds. E. G.
Farnworth and F. B. Golley). Springer Verlag, New York.

Osmaston, H. A. 1959a. Working plan for the Kibale and
Itwara forests, Uganda. Uganda Forestry Department,
Entebbe. 60 pp.

Ovington, J. D. 1975. Forest management in relation to
ecological principles. In:IUCN(1975) The use of
ecological guidelines for development in tropical
forest areas of S. E. Asia. IUCN series No. 32, Morges,
Switzerland.

Pickett, S. T. A. 1983. Differential adaptation of tropical
tree species to canopy gaps and its role in community
dynamics. Tropical Ecology 24(1)69—84.

Pimentel, D., W. Dazhong, S. Eigenbrode, H. Lang, D.
Emerson, and M. Karasik. 1986. Deforestation:
Interdependency of fuelwood and agriculture. Oikos
46:404-412.

Poore, M. E. D. 1968. Studies in Malaysian rainforest. I.
The Forest on Triassic sediments in Jengka forest
reserve. J. Ecol. 56:193-196

Poore, D. 1976. The values of tropical moist forest
ecosystems and the environmental consequences of their
removal. Unsylva 28:112—113.

Putz, F. E. 1983. Tree fall pits and mounds, buried seeds
and the importance of soil disturbance to pioneer trees
on Barro Colorado Island, Panama. Ecology 64:1069—1074.

. 1984. How trees avoid and shed lianas.
Biotropica 16(1)19—23.

Putz, F. E. and C. Milton. 1982. Tree mortality rates on
Barro Colorado Island, Panama. In:The ecology of a
tropical forest: Seasonal rhythms and long term
changes. pp. 95—100 (eds. E. G. Leigh, Jr. A. S. Rand
and D. M. Windsor). Smithsonian Institution Press,
Washington D. C.

253
Richards, P. W. 1952. The tropical rainforest. Cambridge
University Press, Cambridge. 450 pp.

1964. The tropical rainforest: An ecological
study. Cambridge University Press, Cambridge. 450 pp.

 

, and G. B. Williamson. 1975. Tree falls and
patterns of understory species in a wet lowland
tropical forest. Ecology 56:1226-1229.

 

Short, J. C. 1983. Density and seasonal movements of forest
elephants (Loxodonta africana Cyclotis) in Bia National
Park, Ghana. Afr. J. Ecol. 21:175-184.

Siegel, S. 1956. Non-parametric statistics for the
behavioral Sciences. Mcgrow Hill, New York.

Skorupa, J. P. and J. M. Kasenene. 1984. Tropical forest
management: Can rates of natural tree falls help guide
us? Oryx 18:96-101.

Sokal, R. R. and F. J. Rohlf. 1981. Biometry. W. H. Freeman
and Co. San Francisco. 859 pp.

Sousa, W. P. 1984. The role of disturbance in natural
communities. Ann. Rev. Ecol. Syst. 15:353-391.

Synnott, T. J. 1975. Factors affecting the regeneration and
growth of seedlings of Entandrophragma utile (Dawe and
Sprague) Sprague, in Budongo forest and Nyabyeya
Nursery, Uganda. Ph.D. thesis, Makerere University.

TIE Report. 1973. In: Fragile ecosystems: Evaluation of
research and application in Neotropics. (eds. E. G.
Farnworth and F. B. Golley). Springer Verlag, New York.
258 pp.

Uhl, C. and P. G. Murphy. 1981. Composition, structure and
regeneration of a tierra firme forest in Amazon Basin
of Venezuela. Tropical Ecology 22:219-237.

, and C. F. Jordan. 1984. Succession and nutrient
dynamics following cutting and burning in Amazonia.
EcolOgy 65(5)1476—1490.

 

, and R. Buschbacher. 1985. A disturbing synergism
between cattle ranch burning practices and selective
tree harvesting in Eastern Amazon. Biotropica
17(4):265-268.

 

UNESCO. 1978. Tropical forest ecosystems: A state of
knowledge report. Natural resources research No. XIV.
United Nations, Paris.

254

USAID. 1978. Proceedings of the United States conference on
tropical deforestation. United States Department of
State, washington D.C.

Van Orsdol, K. G. 1983. The status of Kibale Forest Reserve
of Western Uganda and recommendations for its
conservation and management. A report to the Ministry
of Agriculture and Forestry, Uganda, the New York
Zoological Society and African Wild life Foundation.
Washington, D. C. (Unpublished).

Webb, L. J., J. G. Tracey and W. T. Williams. 1972.
Regeneration and patterns in the subtropical rainforest
Journal of Ecology 60:675-696.

Westhoof, R. 1971. The dynamic structure of plant and animal
communities. 11th Symposium of the Brit. Ecol. Sco.

15pp.

Whitmore, T. C. 1975. Tropical rainforests of the Far East.
Clarendon Press, Oxford. 282 pp.

1978. Gaps in the forest canopy. In: Tropical
trees as living systems pp. 639-655 (eds. P. B.
Tomlinson and M. H. Zimmermann) Cambridge University
Press, New York.

 

Wilson, D. E. and D. H. Janzen. 1972. Predation on Scheelea
palm seeds by bruchid beetles: Seed density and
distance from the parent palm. Ecology 53:954-959.

Wing, L. D. and I. O. Buss. 1970. Elephants and forests.
Wildlife Monographs. No. 19. The Wildlife Society
Washington D.C.

Young, K. R. 1985. Deeply buried seeds in a tropical wet
forest in Costa Rica. Biotropica 17(4)336—338.

 

255

 

 

em on co o oo o a» on em am «a no G ho>oo .uo> 1:509. I

v.n~ m.va a.¢v c m.aN c c.9w v.pn n.0w n.mo m.wm o.ao o noaoo9uxhaannov
9\um Cunahalon

w.—~ a.~m —.cn o v.9N o a.wn °.pn m.uv a.nv n.9w c.~> o Isuoo9-
\hvuncov .9xum

N.ep N.~mv c.90w o w.¢p— c u.con n.vnN h.von 0.Na~ N.vnN N.nan o A¢£\ov haw-sou
9\.I onadhquou N

p.pa p.poe >.nn¢ c e.gun o o.oam >.¢ov Q.won own «.mcv ”.umv o lax.
haun:ov 9x6» M

n a e a w o @ m.c a.» Q.” o.v n.v o neosaouh
99- oaaouuuov M
v 09 N9 a mm o ‘9 - p.> 5.5 9.x p o Ino:£o«n .9: some m
I I a c N o n e o a v— N o u9so no .02

cows cows | coca n ccvq coma coca con coo ocv can I

oowNA coca can" I coon I IOON— loco" loom tooc Iccv icon Ion noumdaouowmml9nc

 

.Anulxv awoken voadouv can aso hdo>wuoodon kdu>¢on no '9‘. 0:9 :9 Asap loo—v lOm.—Mv Amy no~o9

v:d Ace. ulcuq9ua no newasnmuanav on» use m9dn vouOOHoo hqlovcah 0v no :Oaasnwhunav umaaovonfim .— cunt?

vano99<

256

 

 

on as o o no as no M.So 0v uo>oo .wo> teach. I

p.on v.90 o o o.np c.0o «.0» a.on~ no" Ismoo9axhuuunov
9\¢m eqnlufilon

~.um w.uu o o v.mo m.nm v.90 nu— a.uou usaoonn
\huanaov 9xnm

N.«mn ~uo o o o.>~o v.0nu ¢.noo p.ano~ a.u>- .IA\.. houncop
.9\nn cone vaneuueov N

o.>w¢~ v.avo c o —.pmp c.0na. o.oa~— when pane sax.
haansov 9xcm m

o a e o o.s ¢.o ~.> m.p «.O unocaofiu
.99. oouv canahunov N
N» ha 9 o o.~— w.vu 9.09 o.n~ n.0u uno:&o«h .9- ooua m
9 u o o N o 09 ad a .900 no .02

coca ooo— 00:9 oou~ coo— oow coo oov can «I

ooo— noov— Icouu noco— loco loco noev Icon too non-uncuouumn9ao

 

..vulm. anouou aao h~o>ma00qon Angus-«u Mo I9¢I one :u Asst Ioodv Ibo.~N.

.9. uo~o9 v66 Ado. u-qu9o- no codaapmuaeav use 6:1 09¢! vouOOHoa halov:¢u a. no nodaapmhanuv nodaOuouam .N candy

Rauco99<

257

 

 

om u.pn me oh on c.9p o.>e «.up no up s.mp p.0n .uo>oo .uo> pascal n

“.m9 «.m9 N.Nn m.on ”.9” N..» we p.¢m v.9» cw —.v> son leuoo9uxbuaeaov
9\¢m cabana-09

e.o~ a.s— o.>N n.NN u.an N.Nn Nv 0.90 0.00 ~.mw ¢.vp n.~p~ eo«009u.
\huqenov .9\¢m

o.—o n.9p m.¢- “.mh N.N~N m.@o— u.uuN man n.0Nv won o.¢~n N.¢co .0£\Nv haw-cop
9\¢I o~acuueov N

—.NNN ~.NvN m.Non can u.eon v.v~s N.ONN ¢.on9~ n.aon— «.mNn~ N.QNN~ o.N~oN usxe
meanest 9xdn m

u.v m.v u.» N a u.v o «.9 o o.v n.v n.v neocao«u
99¢ unsung-op N
o..« n.99 «a m— o— «.uN N.~N m.m_ mN o.o~ v.c— o.b— encasemu .9. cone M
v a a a s v a v a v s a cane so .02

oocN ooNN OOON can“ coo. OON~ coon coo coo so. ocN I

oooNA oovN oooN coon tcoen nocvg looog loco nooo Ioov looN Ion nonconononmmu9ao

 

Ilcuq9ae ooua

..unllv anchOu Goduvu Ado>aaoodon hqu>uoa No I9GI 0:» flu Asap IOO—v IOD.~NV .9. Io~o9 6:0 ..nm.

as :0«o:&«uoeav Ode vac e96. v0a00qoe halogens ov No segusaauanuv audacioune 9am summon .0 cases

x«u:o99<

258

 

 

I I o.Om o.mN N.Om w.w¢ ho>oo acauduom0> vcaoum x

I I m.ms ~.~m~ m.nm~ m.onm mofloodm
\huflmsop 9\mm manahfimon
¢.¢w m.mm9 m.>¢~ w.mHN mmwom9m\huwmswp .o9\om

I I o.>>w «.mwma «.mmmfi m.e¢w~ haemaov
9\dm coma oaadhwmov x

I I N.HooN e.mooe ¢.wmwm O.Nmam Am:\*v
Swanson o~o9\ucfl~9am N

I I 6.99 o.m b.m c.w unannofiu
.990 mean manahwmop x
I I o.NN c.0m o.mN m mmocsofih mowoo9m menu N
o c L L «a 49 mass mo ponasz

ooNH coo“ com com ooe oON

Ioood loom Ioco Iooe IooN Ion NE .mommdao oufim 9mm

 

.AomImv umvhom mazes! usoc: mo m9aw damn

some poauom haaahzada aw Azfiv Eoofiv 60m.HMV medo9 pad munda9am mo Gawusnwuumflp
use odd 09¢! poo00dom haaoucdh cs 90 :oflusnflhamwfi cacao ouwm .v wands

wwfico99<

n4

 

 

 

 

 

 

.aoaaca awed :oMuauouo> and sum: ma:¢: s50 NI— u v0

.vo>0Iou sauce ooauuan u=H9 G :q as n ma

.u:«:oonon usu9 do :u on u N0

.vo>OIon :o«a¢oouo> sad: av:¢uvafla NI" u ~0

>.n 6.~— 6.— 6.—n v.6vn 6.66 N.6~— «N6 6.hnN v.NN~ p.66N 6.66N h£wlfl06 m
6.6 «.6 N.6 v.6u 6.N ~.~ 6.v 6 5.6 N.6 6.6 6d OHOAIwHO 5:1 00:“)
6.n um 6.6 n.6u N.¢6~ ~.66 66 65M MNN N.N6n 66N n.66n Ivfildfi DSOOOGDHOI
6.6 «.6 —.6 6.6 6.~N 6.6m h.N 6.v N.6 6.6m 6.6 —.6N 00d00§l Bflhflm
6.~ 6.6 6.6 6.6 6.6M c.6— h.6 «.un h.6 6.6a b.~— 6.vn IlfluHGOOI OOHP
Nflm.°z Hafiflflon

~.~ 6.6 6.6 6.1 6.6 6.6 6.b 6.¢ 6.6 5.6 6.6 6.6 NI\IOM0¢QD M
N.6 ~.6 ~.6 v.6 >.6 6.~ 6.~ 6.~ v.6 6.6 6.6 6.6 Ihvnlwdo 6:6 IOEM>
5.6 6.6 «.6 ¢.N 6.6 6.6 6.? v.N ~.v N.‘ N.n ~.n nfiflddfl l50006390=
6.6 n.6 «.6 6.6 6.6 6.6 b.6 h.6 6.— N.~ ~.~ —.— .UmOOQI nfiflhzm
N.6 6.6 «.6 b.6 N.6 6.6 6.~ 6.6 ~.— v.— 0.6 m.~ Iflflddvoou 009%
ouo:&o«h cowommm

6nll 6mlg Q—lfl 66IM 6~ll c—Ifi 66!! 6—Ifl tall 66!: 6HIN v—Ii naflvlvhflflloo
«O NO NO 90 00::hvazo

 

.ONIa v:u m—u: .e—IM you u9ou ON": .ONI: v:d m9IM .v9I: no noo~9
9du uba:~=9«::l :u voa::wlhou hqzo: nu:ua9 no .Nl\.oz. huun:ov 6:: nmo::o~u neuoo9u :80: .u canch

Xavflo99<

260

 

 

 

0.09 .0.N9 00.0 5.0N .0.N0 0.0N .0.90 0.00 x0
0.09 .0.09 00.0 00 .N.N0 0.90 .0.0N 00.95 M
5.09 .0.09 0.9 0.00 .00 0N .5.9N 0.50 0090:M990 .9
0N ...0 0.0 0.0N .0.00 N.00 .0.009 0.009 000M99=009u .0
09 .5.N 0.9 0N .0.0 05 .0.09 9.5N 9009990000 .0
0.0 0.9 009.0.0 . 0.0 0.0 009oox0 .9
0.0 0.0 0.0 0.0 0.0 99aomu9z9om .O
09 .0.N0 0.0 0.90 .0.00 5.N0 .9.50 00N 9900:00000 .z
00 .09 0.9 0N .0.09 00 .0.9N 0.00 903050.00 .:
0N .5.N 0.0 00 .0.0 00 .0.0 0.09 mH90ouo099 .2
09 .0.09 N.9 0.00 .0.00 5.00 .9.50 0.009 99:0000::N30 .9
0.0N .9.00 9.9 0.50 .0.05 0.0N .0.00 0.509 099099009 .9
0.0 0.0 009.0.0N 0.0 0.0N 00090900x0 .9
0.0 0.0 00 .0.09 00 .0.09 0.9N 090:09ou:0 .9
0.0 0.9 009.0.09 0.0 0.09 009N£mouo0l .9.
00 .5.N 0.0 0.0 00 .5.N 0.0 99:09990 .0
0.00 .N.5N 0.0 0.5N .0.9N 0.50 .0.0N 0.05 000000-0300. .0
0.0N .0.09 5.0 0.0N .0.09 0.50 .0.0N 5.90 0:009090 .0
0.N9 .0.09 0.0 0.00 .0.0N 9.00 .N.00 0.50 0a0um99m: .0
5.0 90.0 0.9 0.00 .N.00 0.90 .9.00 5.009 09MN09xoa .<
0.0 0.9 00 .0.09 00 .0.09 N.5N mfl9009090 .0
..:\oc. ..o..\o=. ..g\o:. .ua\o=. ..=\o:.
0000
0950090 010009 £03000 haw-:00 0090090 0009
0000 :00: 00099900 900002 90009

 

.900I06 000009 00:00: 000:: 90 0909 000009 0N :9

.009 0050090 0000 00 00 906 00099900 00 .92. 00000 90:00: £993 9:90000:0800
9:90 :009v 000.9M 00:00. 00909 0:0 00099900 0090090 0009 990:00 90

.9509: 0:0 :06 :090095909 0090090 9:00:09 0:0 .0£\.029 9090:00 0:5 .0 09005

x9v:099<