IHESIS

 

 

 

 

LIBRARY ,,

Michigan stare {

University "

I\ fill

I

1
l
f

This is to certify that the (

dissertation entitled
'IHE EFFECIS OF LARGE HERBIFORES ’
OF THE EOOSYSTEM OF I
QUEEN ELIZABETH NATIONAL PARK, UGANDA
presented by

GINDRO LADUMA LAKER-OJOK

 

has been accepted towards fulfillment

 

 

 

 

of the requirements for
PH.D. degree in FISHERIES & WILDLIFE
l
at?! I
Major professor .33.
2-15-85
Date E

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

rntl'fii'tlgnnifik _

MN

4
"1I

 

 

 

 

 

1V1531_J RETURNING MATERIALS:
Place in book drop to

'LJBRARJES remove this checkout from

.—;_—L your record. FINES will

 

 

 

be charged if book is
returned after the date
stamped below.

 

 

 

 

 

 

 

 

 

THE EFFECTS OF LARGE HERBIVORES
ON THE ECOSYSTEM OF
QUEEN ELIZABETH NATIONAL PARK, UGANDA

By

Gimoro Laduma Laker-Ojok

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Fisheries and Wildlife
1985

Copyright by
GIMORO LADUMA LAKER-OJOK
1985

Dedicated to my parents,

PILIPO OJOK and LUCIRA AJOK,
who taught me the meaning of conservation

and appreciation and respect for nature.

ACKNOWLEDGEMENTS

I am thankful to my major professor, Dr George A. Petrides (Fisheries and
Wildlife), and members of my guidance committee, Drs John A. King
(Zoology), Stephen Stephenson (Botany and Plant Pathology) and Mitchell
W. Taylor (Fisheries and Wildlife) for intellectual stimulation and
support.

The research for this dissertation was made while the author was a
Research Officer with the Uganda Institute of Ecology. I am most grateful
to the Uganda National Parks for financial support. The research could
not have been completed without the financial, material and moral support
of my family for which I am most thankful. I am also grateful to the Page
Foundation for financial assistance towards the preparation of the
thesis.

I am indebted to several members of the junior staff of the Uganda
Institute of Ecology who assisted with the fieldwork, especially to Peter
P. Oola, Francis Odong, Yoram Katungwesi, John Bosco Omona, Alex Ojera
and Michael Owalagala. Their contributions are gratefully acknowledged.

My wife, Rita, endured years of separation during the fieldwork. She
has been a source of encouragement as well as financial, intellectual and
emotional support during the entire project. I am deeply indebted to her.
This project could not have been possible without her trust, support and

nurturing.

ii

ABSTRACT

THE EFFECTS OF LARGE HERBIVORES
ON THE ECOSYSTEM OF
QUEEN ELIZABETH NATIONAL PARK, UGANDA

By
Gimoro Laduma Laker-Ojok

Game counts in previously studied areas of Queen Elizabeth National Park,
Uganda, during November 1981 to June 1983 showed that populations of
large and medium-sized ungulates were lower than counts made in 1956/57
and 1963-67. For big game species such as elephant (Loxodonta africafla)

and buffalo (Syncerus gaffer), the decline was attributed to recent

large-scale poaching.

The variability of animal counts from month to month was high but
showed no marked trend over the course of the study . Censuses of
individual species seemed to be influenced by such factors as body-size,
wariness, herding characteristics and mobility, the ability of observers
to see all the animals present, as well as the size of the population.

Census results showed no significant seasonal differences. This
supported the general impression that animals in the park do not
undertake large-scale seasonal movements.

Hippopotamus (Hippopotamus amphibius) have returned to the Mweya

Peninsula in large numbers following a cropping scheme between 1958 and

Gimoro Laduma Laker-Ojok

1967 during which they were virtually eliminated from the area. Their
increase has been at the expense of the other ungulate species. The
increased abundance, high biomass and diversity of various large-mammal
species which occurred during and immediately after the cropping of
hippopotamus have been lost.

In the Craters region, the present study disclosed that a woodland
which in 1968 was losing trees due to widespread elephant damage was
regenerating abundantly. Although species composition of the woodland had
remained virtually the same, major shifts had occurred in the relative
abundance 'of species. The most significant was that of Acacia qerrardii
which in 1968 constituted only 21.3% of the trees in the woodland had by
1982 increased to 82.41. Two other species, Bridelia scleroneuroides
Olea africana, which were as abundant as A, gerrardii in the earlier
survey haVe declined greatly and were no longer important components of
the vegetation.

The respective roles of fire and elephant in controlling this woody
vegetation was examined. On account of a gradient of decreasing
mature-tree density with increasing distance from disturbing influences
along the park border, it was concluded that elephant rather than fire
were the primary factor in controlling woody vegetation. In Murchison
Falls National Park, Uganda, observations on experimental plots from
which elephant, but not fire, have been excluded support such a
conclusion. The pattern of tree distribution seemed likely to have
developed as a result of differential use of the range by ungulates in
response to poaching and harrassment along the park border.

Thicket clumps constituted an important resource to wildlife

Gtmoro Laduma Laker-Ojok

providing food, shelter, shade and concealment for a wide range of fauna.
An increase in thickets was attributed to their release from frequent
fires and heavy browsing. In heavily-grazed areas such as the banks of
the Kazinga Channel and the Mweya Peninsula, the absence of fire (due to
inadequate fuel to support them) allows thickets to flourish. It was
concluded that thicket clumps originate from a nucleus of a single bush
in the grassland. The bush has to be a vigorous species in order to
persevere and survive frequent fires and persistent defoliation.

Of three grassland types, namely the short, medium-height and tall
grasslands, standing-crop plant biomass was highest in the tall and
lowest in the short. The differences were attributed both to grazing and
precipitation. The intensity of grazing was highest in the short
grassland where between 601 and 90% (with an average of 75.61) of
standing-crop biomass was consumed by herbivores compared with only 26.5%
in the medium-height grassland. Light grazing enabled the maintenance of
good vegetative cover but resulted in accumulation of litter and the risk
of hot fires. Heavy grazing severely reduced grass cover and resulted in
bare soil patches. It is, no doubt, a major factor in inducing

bush-invasion and habitat deterioration in some areas of the park.

TABLE OF CONTENTS

Chapter 1. Introduction ..................................... 1
1.1. Introduction ..................................... 1
1.2. Study objectives ................................... 3

1.3. Thesis outline

Chapter 2. Queen Elizabeth National Park and the game census areas 6

2.1. The Queen Elizabeth National Park .................. 6
2.1.1. Geology and topography ..................... 6
2.1.2. Climate .................................... 9
2.1.3. Soils .................................... 11
2.1.4. Vegetation ................................. 11
2.1.5. Fauna ..................................... 14
2.1.6. Human and wildlife history ................. 20

2.2. Census areas ...................................... 22

Chapter 3. The ungulate community .......................... 27

3.1. Introduction ...................................... 27

3.2. Methods ........................................... 28
3.2.1. Counts ..................................... 28
3.2.2. Variability in numbers ..................... 31
3.2.3. Distribution with respect to habitat ....... 31

iii

3.3.

3.4.

3.2.4. Seasonal fluctuations in animal numbers

3.2.5. Population structure

3.2.6. Long-term trends

Results

3.3.1. Counts

3.3.2. Variability in numbers

3.3.3. Distribution with respect to habitat

3.3.4. Seasonal fluctuations in animal numbers

3.3.5. Population structure

3.3.6. Population trends on Mweya Peninsula .......

Discussion

3.4.1. Counts

3.4.2. Variability in numbers

3.4.3. Distribution according to habitat ..........

3.4.4. Seasonal fluctuations in numbers

3.4.5. Population structure

3.4.6. Population trends on Mweya Peninsula .......

Chapter 4. Woodland regeneration ...........................

4.1.
4.2.

4.3.

Introduction

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

32
32
33
33
33
38
41
43

62
63
63

65
7o
74
79

81
81
82
82

85
85
86

4.4.

Chapter

5.1.
5.2.

5.3.

5.4.

4.3.1. Composition and density

4.3.2. Structure of the woodland

4.3.3. Regeneration potential ......

4.3.4. Fire, elephant and the woodlan

Discussion

4.4.1. Composition and density .....
4.4.2. Structure of the woodland

4.4.3. Recruitment potential

4.4.4. Fire, elephant and the woodland

5. Thicket abundance and expansion

Introduction

Methods

5.2.1. Clump size, density and cover
5.2.2. Composition of thicket clumps

5.2.3. Origin and development of clumps

5.2.4. Thicket expansion

5.2.5. Thicket as a wildlife resource

Results

5.3.1. Clump size, density and cover

5.3.2. Species composition

5.3.3. Origin and development

5.3.4. Expansion

5.3.5. Thickets as a wildlife resource

Discussion

V

d ............

IIIIIIIIIIIIIII

86

87
90

91
95
95
96
101
102

106
106
107
107
107
108
108
108
108
108
109
111
113
119
120
120

5.4.2. Species composition

5.4.3. Origin and development of clumps

5.4.4. Thicket expansion

5.4.5. Thickets as a wildlife resource

Chapter 6. Standing-crop biomass, species composition and grazing
intensity in three grassland types
6.1. Introduction
6.2. Methods

6.2.1. Species composition and vegetative cover
6.2.2. Standing-crop biomass and intensity of grazing
6.2.3. Vegetation within the 15-year old exclosure on
the Mweya Peninsula
6.3. Results

6.3.1. Species composition and vegetative cover
6.3.2. Standing-crop biomass and intensity of grazing
6.3.3. Vegetation within the 15-year old exclosure

6.4. Discussion

6.4.1. Species composition and vegetative cover
6.4.2. Standing-crop biomass and intensity of grazing
6.4.3. Vegetation within the 15-year old exclosure

6.5. The Mweya Peninsula: A case study of vegetation changes

induced by grazing

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Chapter 7. Conclusion and recommendations

7.1. Conclusion

128
128
129
129
129

130
131
131
135
140
145
145
148
149

150

153
153

Table 3.1.

Table 3.2.

Table 3.3.

Table 3.4.

Table 3.5.

Table 3.6.

Table 3.7.

Table 3.8.

LIST OF TABLES

Elephant abundance in Queen Elizabeth Nationaql Park,
Uganda, illustrating the recent decline in game numbers 27
The number of poaching incidents recorded by author in

QENP, Uganda, between 1980 and 1983

............. 28
Average large herbivore numbers in the study areas,
Queen Elizabeth National Park, Uganda, November 1981
to June 1983 ..................................... 33
Average large mammal densities/sq km in study areas,
Queen Elizabeth National Park, Uganda, November 1981
to June 1983 ..................................... 34
Percentage changes in densities of large herbivores
in 1981/83 from 1956/57 and 1963/67 levels in Queen
Elizabeth National Park, Uganda .................. 35
Average standing-crop biomass (kg/ha) in the study
areas, Queen Elizabeth NAtional Park, Uganda. November
1981-June 1983 ................................... 37

Summary of counts of large mammals in the study areas

in Queen Elizabeth National Park, Uganda, November 1981

to June 1983 ..................................... 39

Average standing-crop biomass (kg/ha) according to
habitat type, Queen Elizabeth National Park, Uganda,
November 1981 to June 1983

Table 3.9. Results of a 3-factor ANOVA to determine the impacts

of season, area and replication on species density in
the study areas, Queen Elizabeth National Park, Uganda,

November 1981 to June 1983 ....................... 46

Table 3.10. Results of 1-way ANOVAs on sub-season data from study
Area 1, Queen Elizabeth National Park, Uganda, November
1981 to June 1983

................................ 47
Table 3.11. Average numbers of animals in study Area 1 according
sub-seasons, Queen Elizabeth National Park, November
1981 to June 1983 ................................ 47

Table 3.12. Size distribution of buffalo bachelor groups, Queen

Elizabeth National Park, Uganda, November 1981 to June

1983 ............................................. 59

Table 3.13. Size distribution of buffalo herds in Queen Elizabeth
National Park, Uganda, November 1981 to June 1983 59
Table 3.14. Age and sex classes of eight species of herbivores in
Queen Elizabeth National Park, Uganda, November 1981
to June 1983 ..................................... 61
Table 3.15. Average densities/sq km of six herbivore species

between 1956 and 1983 on Mweya Peninsula, Queen

Elizabeth National Park, Uganda .................. 63
Table 4.1. Species composition of a regenerating woodland in

the Craters region, Queen Elizabeth National Park,

Uganda, November 1981 to February 1982 ........... 89

Table 4.2. Species densities and importance values for Acacia
woodland in the Craters region, Queen Elizabeth

ix

Table 4.3.

Table 4.4.
Table 5.1.

Table 5.2.

Table 5.3.

Table 5.4.

Table 5.5.

Table 6.1.

Table 6.2.

Table 6.3.

National Park, Uganda. November 1981-February 1982 91
Agagia regeneration potential in woodland in the

Craters region, Queen Elizabeth National Park, Uganda,
November 1981 to February 1982 ................... 92
Severity of fire damage according to tree height . 93
Comparison of thicket parameters in Mweya Peninsula

and Kikorongo, Queen Elizabeth National Park, Uganda,
May 1983

......................................... 111
Common thicket species on Mweya Peninsula, Queen
Elizabeth National Park, Uganda, May 1983 ........ 112
Mean monthly rainfall (mm) on the Mweya Peninsula,

Queen Elizabeth National Park, Uganda ............ 116
Changes in diameter (m) of selected thickets on

Mweya Peninsula, Uganda, between February 1982 and

May 1983 ......................................... 118
Qualitative assessment of large mammals use of the
thicket resource in Queen Elizabeth National Park,
Uganda, November 1981-June 1983 .................. 121

The major components of three grassland types in

Queen Elizabeth National Park, Uganda. Dry weights

are means for 14 months from May 1982 to June 1983 134
Foliar cover in the short grassland on Mweya

Peninsula, Queen Elizabeth National Park, Uganda, at

the end of the first growing season, May 1982 .... 135
Mean monthly biomass (g/sq m) harvested in grazed

and ungrazed plots in three types of grassland, Queen

X

Table 6.4.

Table 6.5.

Table 6.6.

Table 6.7.

Table 6.8.

Elizabeth National Park, Uganda, 1982-1983

Estimates (g/sq m) of vegetation removed by ungulates

from the short and medium-height grasslands in Queen
Elizabeth NAtional Park, Uganda, 1982-1983. (Figures
in parenthesis are percentages of standing—crop
biomass cropped).

Monthly rainfall (mm) at Mweya Peninsula, QENP,
Uganda, and mean dry weight (g/sq m) of live shoots
harvested in ungrazed plot
Major constituents of the grass sward in the 15-year

old exclosure on Mweya Peninsula, Queen Elizabeth
National Park, Uganda, May 1982-June 1983 ........ 145
Comparison of the numbers of trees of three woody

species in burned and unburned sectors of the 15-year

old exclosure on Mweya Peninsula, Queen Elizabeth
National Park, Uganda, May 1983.

Changes in frequency of grass species on the Mweya.

Peninsula, Queen Elizabeth National Park, Uganda . 153

xi

Fig.
Fig.
Fig.

Fig.

Fig.

Fig.

Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.

2.1.
2.2.
2.3.

2.4.

3.1.

3.2.

3.3a.
3.3b.
3.3c.
3.3d.
3.3e.
3.3f.
3.39.
3.3h.
3.3i.
3.4.

LIST OF FIGURES

Location of Queen Elizabeth National Park in Africa

Queen Elizabeth National Park, Uganda

Map of Queen Elizabeth National Park showing the
the isohyets of mean annual rainfall for 1964-1966

inclusive. Figures are in centimeters

Map of the sector between Kazinga Channel and the
equator, Queen Elizabeth National Park, Uganda,

showing game census areas

Large mammal biomass composition in the long,
medium-height and short grass habitats ...........

Comparison of seasonal densities in habitats near and

far from water

Elephant groups

Warthog groups

Hippopotamus schools

Reedbuck groups

Waterbuck groups

Kob groups

Bushbuck groups

Buffalo: bachelor groups

Buffalo herds

Game population trends on Mweya Peninsula, 1956 to

xii

10

23

42

45
49
50
51
52
53
55
56

833}

Fig.

Fig.

Fig.
Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

3.5.

4.1.

4.2.
4.3.

5.1.

5.2.

6.1.

6.2.

6.3.

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Trends in biomass composition on Mweya Peninsula,
1956 to 1983

Girth distribution of Acacia trees at 1.5 m above

 

ground

Fire damage to woodland trees

Density of mature Acacia trees between Equator Road

and Katwe Road

Mean monthly rainfall (mm), Mweya Peninsula, QENP,
Uganda

Growth of bush-clumps, Mweya Peninsula, QENP,
Uganda

Live-shoot biomass in three grassland types, QENP,

Uganda

Intensity of grazing, Queen Elizabeth Nat'l Park,
Uganda

Live-shoot production in relation to rainfall, QENP,

xiii

62

90

96

117

119

140

141

143

CHAPTER 1

INTRODUCTION

1.1. Introduction

The Queen Elizabeth National Park (QENP) in southwestern Uganda was
created by an Act of Parliament in 1952. Until the 19705 it supported the
highest biomass of wild ungulates anywhere in the world (Petrides & Swank
1965; Field & Laws 1970; Leuthold & Leuthold 1976). Intense poaching
during the 19703 led to a sharp decline in the population of many species
(Eltringham & Malpas 1976, 1980; van Orsdol 1979; Malpas 1980).

The massive decline in animal populations seriously interrupted the
dynamic equilibrium between animals and vegetation. The magnitude and
. extent of the consequences are not fully understood and may not be for
some time. Some effects are already apparent, however, and involve a
profuse regeneration of trees and increase in bush cover.

This regeneration reverses a declining trend blamed (Field 1971) on
elephant (Loxodonta africana). It is rapidly turning large areas of
previously open grassland into dense woodland (Yoaciel 8. van Orsdol
1981). Park authorities are concerned about possibly-adverse effects of
grassland reduction on grazing animals.

In many African national parks, changes in landscape have been
attributed to the effects of elephant and fire (Buechner & Dawkins 1961;
Laws 1970; Laws, Parker & Johnstone 1970,1975). Because the interactions
between elephant and fire are complex, blame for the changes cannot be
attributed to one without considering the other. Until recently both

2
elephant and fire were common in Queen Elizabeth National Park. With the
elephant virtually eliminated from large areas of the park, the situation
offered a unique opportunity to assess the effectiveness of fire as an
agent of habitat change and to determine the relative importance of both
fire and elephant in controlling woody vegetation.

One aspect of the changes in habitat involves the spread of thickets.
The increase has been particularly rapid along the shores of Lake George,
Lake Edward and the Kazinga Channel where grazing by large numbers of
hippopotamus (Hippopgtgmys amphibius) keeps the grass short and thin,
and precludes the occurrence of grass fires. In areas where fires are
frequent, the spread of shrubs appears to be modest or under control.
These observations suggest that grazing enhances the ability of a thicket
to spread while fire restricts it. Understanding the processes involved
could provide the information necessary to formulate an economic program
that would reverse the spread. The authorities have attempted clearing
bush on a limnted scale on Mweya Peninsula using hand tools, but
abandoned the project because it was slow and too costly.

Although there are large populations of hippopotamus in the waters of
the lakes and the Channel, there is a decreasing gradient of grazing
pressure on uplands as distance from these bodies of water increase (Lock
1967, 1972). Changes in grassland species composition as well as in
biomass and productivity are observed along the gradient. An attempt was
made to determine the above-ground production of dry matter and the

intensity of grazing at varying distances from water.

1.2. Study objectives
The aims of this study were to:

1. Establish the present status of the large-herbivore populations
(in terms of abundance, distribution and population structure), assess
the extent of changes in abundance over the last decade and project the
outlook for the forseable future;

2. Determine the above-ground production of dry matter and obtain
mean values for the rate of grazing along a gradient of grazing
intensity;

3. Document changes in vegetation, especially the regeneration of
Acacia species in the Craters highlands and the spread of thickets on
Mweya Peninsula; and

4. Assess the effectiveness of fire as an agent of habitat change in
the absence of elephant and determine the relative inportance of both

elephant and fire in suppressing the regeneration of trees.

1.3. Thesis outline

The thesis is organized in seven chapters each of which explores a
different topic. Each topic is presented in a methods-results-discussion
format. A brief introduction to the issues and the context in which the
study was done is made in Chapter 1 which also outlines the objectives of
the study.

Chapter 2 describes the Queen Elizabeth National Park, the reserve in
which the study was done and presents a brief overview of its climate,
geology, vegetation and fauna. Human and wildlife history relating to the

park, its relationship to other natural resources in the region and

4
current threats to its continued existence are outlined. Areas selected
for animal censuses and for the study of tree regeneration are
described.

Chapter 3 deals with the status, distribution and population
structure of large herbivores. Details of census methods are given and
results are presented and discussed. The age structure and sex ratio of
each population, as well as the abundance of large mammals in three
habitats are analyzed. The findings of the study are compared with those
of the past, and a long-term trend for one of the areas is discussed. The
outlook for each species is briefly outlined.

Chapter 4 deals with tree regeneration in the park. A brief
description of the regenerating woodland is given and the method used to
determine tree density in the woodland is also described. The composition
of the woodland, the spatial distribution of plant species and the
importance value of the major trees are given. The effects of elephant
and fire on regeneration of the woodland are evaluated and their relative
importance in controlling woodland development determined.

Chapter 5 deals with the abundance and expansion of thickets in the
park. Methods used to investigate the expansion are presented. The
results of investigation, including the origin and development of thicket
clumps are reported. Present density, size and rate of expansion of
thicket clumps in two different types of grassland, are given and the
roles of fire, grazing and trampling on vegetation are assessed. Probable
causes of thicket expansion are discussed.

Chapter 6 explores changes in species composition as well as in
height and productivity of grasslands at varying distances from the

5

shoreline, corresponding to a gradient of decreasing grazing pressure by
hippopotamus. Biomass of grass in grazed and ungrazed plots was analyzed
and compared between plots and areas to assess the intensity of grazing
in different areas. The effects of fire on species composition and
productivity in ungrazed grasslands of the Mweya Peninsula are discussed.
The ditched exclosure used to keep animals out of the plot is described.

Chapter 7 concludes the study. Its major findings are summarized and
an attempt is made to pull the different components into a coherent

overview.

CHAPTER 2

QUEEN ELIZABETH NATIONAL PARK AND THE GAME CENSUS AREAS

2.1. The Queen Elizabeth National Park

2.1.1. Geology and topography
The park is 1978 sq km in area and is situated between longitudes 29°
39'E and 30' 19'E and latitudes 0° 43' S and 0° 15'N, and is situated on
the floor of the western arm of the Great Rift Valley of East Africa
(Figs. 2.1 & 2.2). The landscape is dominated by the Rwenzori Mountains
(the fabled Mountains of the Moon) which lie immediately to the north of
the park and .include glacier-capped peaks that reach 5119 m above mean
sea level. Elevations in the park vary from 913 m at lake levels to 1378
m at Kyamatumu in the Craters highlands.

Associated with the early Pleistocene volcanic activity that gave
rise to the Rwenzoris is a series of at least 78 explosion craters which
occur in an area of about 155 sq km in the park north of Kabatoro
village. These craters are thought to have been formed some 10,000 years
ago in violent explosions of gas and steam accompanied by a vast belching
of volcanic ash and dust which settled over the area as the basis of a
rich, alkaline volcanic soil. The noxious gases and ash are believed to
have exterminated some of the Nilotic fauna such as the Nile crocodile
(Crocodylus niloticus) and Nile perch (L§E§§ niloticus) whose fossilized
remains have been found in the area (Beadle 1965). A number of the

volcanic craters contain saline lakes, and at least two support a

 

 

 

   
 

 

 

20°

Fig. 2.1.

 

 

 

 

Eq uetor

 

 

 

H

96km

Queen Elizabeth
Netlonel Perk

 

20 40°

9 1 600 320.0 km

Location of Queen Elizabeth National Park in Africa.

  
  

:2

 
   
   

 

 

 

M w e v A
nu “.0” g I
.' mm mm Loon: ’ fl 1'
l
: LAKE a"!
eowano ,) A
: .’
:0. t .I
: l
m 0 “new" (3%,- Z
t 2 y, 2
‘ : < b . w
T -' ‘
N 5 U .'
5 D )I 3
. 'wgfl'uu r\ f
' J
'3 r a
L j
a I' Q
' I.
\\ y ‘ Q
'\ . LEGEND
.\. .’ 0 maven Boundery
\. f
\. .I' o------—-lnternquonel Barder
\'v"’ 9 . Village
Q Waiver
. ~n°.d
‘fi;‘; Swamp
0 non er Poet
o Q 9 46 Km
TOMHO.

‘I

QUEEN ELIZABETH NATIONAL PARK, Uganda.

flourishing salt industry.

Lake George (389 sq km) and Lake Edward (2150 sq km) lie in the
northeast and southwest of the park, respectively. Nearly half of Lake
Edward is situated in neighboring Zaire where it reaches a depth of over
115 m. In comparison, Lake George is a large but shallow pond, only 3m
deep. The 32-km long Kazinga Channel, which connects the two lakes,
bisects the park into approximately equal halves. Water flow from Lake
George to Lake Edward along the Channel is sluggish because the two lakes
are almost at the same elevation. Several streams flowing from the

mountains and the eastern escarpment traverse the park and drain into the

lakes.

2.1.2. Climate

Considering the park's position astride the equator, temperatures are
surprisingly cool with mean maxima near 28 ’C and mean minima around 18
'C in all months. Mean hours of sunshine vary from 5.1 in October to 6.5
in July. There are two rainy seasons each year, from March to May and
September to November. Amounts of annual rainfall increase towards the
eastern escarpment from a mean of about 660 mm on the floor of the rift
valley at Mweya to 1250 mm a year in the Maramagambo Forest (Fig. 2.3).
Although the seasons are usually well-marked, appreciable quantities of
rain also may fall as showers during the dry months. The prevailing winds
are mostly northwesterly and westerly. Winds are calm during the dry
seasons but often strong and gusty during the rains. The high
temperatures, long hours of sunshine and low humidity result in high

rates of evapotranspiration, especially during the dry season. In

 

Equator

,. *1/18 .

90 ‘v 70

    

 

 

0670 80
1000
L. EDWARD
:or/ 120
1 10
10° 1 20
70
a .1. 1.9 a 1.0 mu...
80 5310 5 10 15 20km
FIGURE 2 . 3 .
90
Map of the Queen Elizabeth National Park showing the laohyets of

mean annual rainfall for 1964-1966 Inclusive. Figures are In
centimetres.

11
heavily-grazed areas with little vegetative cover, large quantities of

rainwater during thunderstorms are also rapidly lost as run-off.

2.1.3. Soils

The area is underlain by Pleistocene silts, sands and clays which are
usually covered by a thick layer of volcanic ash ejected during the
formation of the craters (Bishop 1970). The ash has weathered to produce
eutrophic soils that are generally loamy, rich in mineral nutrients and
with a high reserve of phosphorus and calcium. Because of the youthful
state of the profiles and of the rich original ash, the potential

fertility of the soils is high (Harrop 1960). The dry climate, however,
limits their agricultural potential.

2.1.4. Vegetation

The vegetation of the park has been described by Langdale-Brown (1960),
Osmaston (1965), Lock (1967, 1977) and Field (1968). It is varied and
includes open grassland, grassland with thickets, woodland, forest and
swamp.
Sims

The most extensive swamp in the park occupies an area of about 250 sq km
north and northeast of Lake George. The community is dominated by papyrus
(gypgrgs pgpygug). Growing in the shallow water among the papyrus is

woody vegetation dominated by species of Agggia, the ambatch tree

(Aechynomene elaphroxylon), aquatic species of Eiggs and the borassus

palm (Borassus aethiopium). A population of the cycad Encephalartos
hildebrandtii occurs in the swamp near the mouth of the Mpanga River.

12

The makindu palm (Phoenix reclinata) is found in inland swamps in Ankole

and on the banks of the Kazinga Channel. The Nile cabbage (Elstia,

stratiotes) flourishes in Lake George and a few large inland hippo

wallows. In the lake, it is carried by the current and blown about by the
wind. Occasionally large numbers may be seen floating down the Kazinga
Channel after a storm. Swamps are also found along the shore of Lake
Edward, especially at the mouths of major rivers.

Open grassland
For the most part, open grassland communities in the park are dominated

by the coarse, tufted, fire-encouraged species such as Themeda triandra,

 

Heteropoqon contortus “and Hyparrhenia filipgndula. Spgrobolus
pyramidalis is also widespread in most areas and S, stapfianus and

Microchloa kugthii, are common in ,the less productive grasslands. In
areas of low rainfall the grasses tend to be short. In higher rainfall
areas, such as the Crater Highlands, they are much taller. In these
’ areas, large patches of tussock grass Cymbopogon afrongggg§_and Indian
sword grass (Imperata cylindrica) commonly occur. Some authors (Osmaston
1965; Spinage 1982) have suggested that their occurrence is evidence of
former agricultural land use. The grasslands to the north of the park are
characterised by trees of Euphorbia candelabrum, which dot the
landscape. In the southern extremity of the park their place is taken by
the gnarled, large wild fig trees Eigy§,gnaphalacarpa, whose branches

form a favorite resting place for lions. Usually the grasslands are also
dotted with thickets.

Grassland with thickets
Areas of very short grass with dense bushes dominated by the thorny

13
scramblers Capparis tomentosa and Azima. tetracantha occur near the
shores of the lakes and the Channel that are heavily grazed by
hippopotamus. This short grassland is rich in species, with 23 grasses
recorded. Chrysochloa orientalis and Spgrobolus pyramidalis are
dominant, but [raggs berteronianus, Spgrobolus stapfianus, Eragrostis

cilianensis. and Cenchrus ciliaris are common. Increase in density and
size of thickets in recent years, particularly along the shoreline, is a

major concern among park authorities and bush clearance has been
attempted on a limited basis.
Woodland

The most extensive stand of Agggia_woodland north of the Kazinga Channel
occurs in the Craters region. It is domnnated by A. gerrardii but A,
sieberiana and A. hggkil are also common. Some slopes contain a largely
unexplored xerophyllous woodland of Olea Chrysophylla, Cordia gyalis and
Euphorbia candelabrum. Southeast of Kasese, patches of Balanites
aggyptiaca form a wooded grassland which also contains trees of Agagig
sieberiana and of A, gerrardii.

Field (1971) reported that between 1954 and 1968 the Craters woodland
declined by about 89% under the combined effects of elephant and fire.
Mature trees were pushed over by the animals and regeneration was either
browsed or destroyed by fire. At present, few mature trees remain in the
southern portion of the area but a stand persists near the main road in
the extreme north. Since the recent decline in elephant numbers, profuse
regeneration of Agggig, particularly gerrardii but also sigbgglgga and
hggkil, is rapidly restoring the area to woodland. Along the eastern

escarpment north of the Maramagambo Forest, are scattered mature trees of

14

Eigu§,gnaphalocarpa and Agagla sieberiana. Ground reconnaissance surveys
indicate that A, sieberiana and A, hggkil are regenerating rapidly,
particularly in the area east and south of Katunguru village. Eiggg, on
the other hand, has not demonstrated the same vigor as its Agagia
counterparts and is not regenerating actively.

In Ishasha, in the southern tip of the park, regeneration by A,
sieberiana is brisk. Yoaciel & van Orsdol (1981) have estimated that
between 1974 and 1978, woodland there increased by 130%.

2.1.5. Forest

The Maramagambo Forest is a large forest reserve within the park. It
covers approximately 450 sq km and is contiguous with the Kalinzu Forest
Reserve outside the park to the east. Although this mahogany-rich forest
is largely unexplored floristically, it is dominated by mature stands of
the Uganda ironwood Cynometra alexandri. In the past it harbored large
numbers of elephant during the dry season. Numerous buffalo (Syncerus
gaffgg) also live in the forest. Smaller forests occur along and at the

mouths of most of the major rivers flowing through the park.

2.1.5. Fauna

The mamnalian fauna of the park is characterised by large nuubers of
several species, particularly hippopotamus, elephant and buffalo. The

area is, however, surprisingly poor in ungulate species. There are no

 

black rhinoceros (Diceros bicornis), zebra (Equus bgrchellis), giraffe
(Giraffa camelopardalis), hartebeest (Alcelaphus buselaphus), eland
(Taurotragus oryx), oribi (Ourebia oerebi) or impala (Aepyceros

15

melgmpus), although all of these species occur within 160 km to the
east. Some authors (e.g. Eltringham 1973) have suggested that the
intensive, late Pleistocene volcanic activity that gave rise to the
craters could have caused widespread mortality and extinction of some
animal species, and that subsequent recolonization was prevented by the
surrounding forest which acted as a barrier to plains game. Others (e.g.
Spinage 1982) believe that the paucity of species is, at least in part,
the result of extermnnation by man due to excessive hunting during
previous settlement of the area. It is also possible that the severe
rinderpest epidemic which swept through western Uganda earlier in this
century eliminated much of the wildlife and cattle (Lugard 1893) and
contributed to the absence of some species. At present, numerous
settlements and extensive cultivation around the park form an equally
effective barrier. Immigration of these species by natural processes is
most unlikely.

Along the Lubilia River in the northwest and the Ishasha River in the
southwest are vegetative corridors of unsettled land which are continuous
with the Kivu National Park in Zaire. Movements of animals, especially
elephant and buffalo, across the border occur frequently.

A West African influence is apparent in the composition of both the
fauna and flora, and in animals it is most marked in the invertebrates
and birds. Traces may be seen, however, among the large mammals. Buffalo
are rather smaller than elsewhere in East Africa, averaging 659 kg in
weight for the adult male and 475 kg for the female (Grimsdell 1969).
There is also a marked brownish coloration, a feature reminiscent of the

reddish color of the small West African bush cow. The adult male bushbuck

16

(Tragglaphus scriptus) are also red as in West Africa and quite unlike

the almost black males seen, for example, in Kenya. The bush pig
(Potamochoerus pgrggs) is also red like the West African hog and not

black as in the southern and eastern races. The elephants, however,
belong to the race L, africana and show no interbreeding as is often

asserted. They are, however, polymorphic in size with some being smaller
at the shoulder and with lighter tusks than the others (Laws 1966). A
West African flavor is also found in the primates with a larger range of
species than is normal in East African parks. The long-haired chimpanzee
(Egg troglodytes), a West African species, occurs in the Maramagambo

Forest in considerable nunbers.

Elephant and hippopotamus are without doubt the most important
species in the park as far as their effect on the habitat is concerned
(Laws 1968, 1970). Estimates of elephant numbers have fluctuated
considerably. The mean of 22 counts conducted between 1963 and 1972 is
2375 (SD 1 870; Malpas 1978). Numbers have fallen precipitously, however,
in recent years due to poaching.

The hippopotamus is one of the most numerous large mammals in the
park. It occurs along most of the shorelines of the lakes and Channel as
well as in inland wallows, particularly during the wet season when these
are numerous. Hippos retreat to the waters of the Channel and the lakes,
however, when the wallows dry up.

The hippo is a significant species in the management of the park.
Between 1958 and 1967 about 7000 were shot in a management program. This
program was initiated to halt the overgrazing and trampling which had

severely reduced grass cover for several kilometers inland and exposed

17

large areas of bare earth to soil erosion (Bere 1959). Vegetative
observations under conditions which simulated heavy grazing by hippo
showed that stoloniferous species were encouraged at the expense of erect
species (Lock 1967, 1972; Field 1970). The decline of the erect species
was attributed to their inability to colonize bare ground and to their
vulnerability to trampling. During the management cropping program, hippo
were completely eliminated from the Mweya Peninsula. Following their
removal the grassland showed a spectacular increase in foliar cover and
erect species, while prostrate and mat-forming species declined (Laws
1968; Thornton 1971). Other effects of the removal of hippopotamus
included increased numbers and biomass of other members of the large
mammal community and a greater species diversity (Eltringham, 1974).
After the culling ended, however, hippo quickly recolonized the Peninsula
and steadily increased in numbers. At present, the composition of the
large mammal community is once again similar to that found in 1956 with a
preponderance of hippo.

The buffalo is widespread throughout the park and seems to be equally
at home in the grassland as in the forest. Two aerial counts by
Eltringham & Woodford (1973) in 1968 and 1969 averaged 17938 buffalo.
Buffalo were heavily poached during the 19703, however, and their present
number is estimated to be only around 5000.

Topi (Damaliscus lggatgs) are restricted to the portion of the park
south of the Maramagambo Forest. In recent years, the topi population has
fallen by nearly 501 from approximately 5500 in 1975 to about 2970 in
1978 (Yoaciel & van Orsdol 1981). It is estimated that the present
population is probably not much more than 1000. Their decline is

18
attributed both to poaching and changes in habitat.

Spinage (1967) gave a figure of 3000 for the waterbuck (Egbgs
defassa) but the present population is probably closer to 2000. The
Uganda kob (AQQ£2£§.EQQ) is the most numerous antelope in the park. Its
numbers is estimated at 28000 (Malpas 1980). The kob is the only large
manmal species which has increased in nunbers in recent years.

No figures are available for warthog (Phacochoerus aethiopicus) but
they are numerous. Bushbuck (Tragelaphus scriptus) and reedbuck (8299229
[299223) are quite common but are less often seen due to their secretive
nature.

Apart from the common animals there are some rare species found
within the park. They include the giant forest hog (Hylochoerus
meinertzhageni) and bush pig (Potamochoerus pgrggs). The giant forest
hog occurs primarily in the Maramagambo Forest but has relict populations
in forested areas throughout the park. The bush pig is only found deep
within the Maramagambo Forest. The sitatunga (Tragelaphus spgkgi) is
known to occur in swamps within the park but its status is unknown.

Forest duikers, including the red (Cephalophus harveyi) and the blue (9.

monticola) are probably more abundant than is commonly realized. The

yellow-backed bush duiker (Sylvicapra grimmia) is occasionally found
within bushes in the grassland.

Several species of primates inhabit the park. The olive baboon (Eaplg
agubls) is common and frequents opengrassland and forest edge. The
vervet (Cercopithecus aethiopg) is also numerous. Less common and
restricted to the Maramagambo and riparian forests are the black and
white colobus (legbgs pglykomos), the red colobus (C, bggigs), the

19
redtail (Cercopithecus nictitggs), the Uganda blue monkey (Q, @151;
stuhlmgnni) and the chimpanzee (Egg troglodytes).

Among the large carnivores, the spotted hyena (Cygggtg gygggtg) is
probably the most numerous. Others include the lion (Eggthgga_1eg) and
leopard (E. pggggs). Though no figures are available, it is estimated
that there are about 200 lions and maybe 100 leopards in the park. The
African hunting dog (Lygggg pigtgs) was found in the park until about
1955. It was exterminated by game and park authorities because of its
method of predation (Uganda Game Department Annual Report 1948-1952;
Uganda National Parks Annual Report 1953-1955). Small cats include the
African wild cat (Eglig lygigg), and the serval (E. sgyygl). The
side-striped jackal (Q§31§_§gg§tg§) is found only in the southern part

of the park. The ratel or honey badger (Mellivora caggnsis) occurs in
the park but is rare. Other small carnivores include the civet

(Civettictis civetta), genet (Genetta tigyina) and five species of

 

mongoose: the greater grey mongoose (Herpggtes ichneumon), slender or
black-tipped mongoose (H, sanguineus), banded mongoose (Mgggg§,mgggg),
white~tailed mongoose (Ichneumia albicauda) and marsh mongoose (Atilgy
galudinosus).

The ant bear (Oryctergius afer) and the pangolin (Magi; teminckii)
are two species of large insectivores that are present but not common in
the park.

Because of varied park habitats, the avifauna is extremely rich.

Reptilian fauna includes the python (Eythgg_§gggg) and many species
of venomous and non-poisonous snakes and lizards (among the latter, the

Nile monitor Varanus niloticgg is the largest). No crocodiles occur in

20
the park, although fossilized remains recovered in Pleistocene sediments
and show that Crocodylus niloticus once existed in the area.
The fish fauna of Lake George and Lake Edward is varied and supports
a flourishing local industry. Rhodes (1965) has described the fish fauna
in detail. The smaller size of fish caught by fishermen today compared to
a decade ago and the persistent demand by fishermen for smaller-sized

nets are circumstantial evidence that the lake system is being
overfished.

2.1.6. Human and wildlife history

The region which is now the Queen Elizabeth National Park was once
densely settled. Abundant archaeological remains of stone tools of
Acheulin culture at Mweya and Katwe indicate that the region has been
inhabited for over 50000 years (Posnansky 1971). Later settlements of
late Pleistocene are represented by widespread remains of pottery and
metallic tools. The presence of hippo harpoons and lance heads suggests
that hunting was the predominant way of life. By the end of the 13th
century hunting had been replaced by a pastoral economy supplemented by
fishing and trading in salt from Lake Katwe. The first European explorers
to reach the region were impressed with the material wealth of the
people, particularly their large herds of cattle (Stanley 1890; Elliott
1896).

The park owes its existence to three catastrophic epidemics. These
were of smallpox, rinderpest and sleeping sickness. The smallpox epidemic
occurred in 1889 and affected the local people. It was followed shortly
afterwards by an epizootic of rinderpest which decimated 951 of the

21

cattle. The impact of the disease on wildlife was undoubtedly equally
substantial and Lugard (1893) observed that ”in some districts, almost
all the game including the small antelopes seem to have perished”. The
high cattle mortality precipitated a famine among the pastoralists. In
1920, the human population was stricken with sleeping sickness
(trypanosomiasis). To combat the latter, the colonial authorities
evacuated the human population from the area in 1924, leaving behind only
a few inhabitants to maintain the salt and fish industries. The
large-mammal populations benefited from the removal of hunting pressure
and increased rapidly. The area was declared a game reserve in 1934 and a
national park in 1952.

Rinderpest ceased to be a problem soon after 1920 and since then the
only epidemic reported in the park was an outbreak of blue-tongue disease
(malarial catarrhal fever) among the topi in Ishasha in 1961. In this
case population losses were minimal. Bovine tuberculosis is common
especially among the buffalo (Woodford 1972). The disease was probably
contracted from domestic stock earlier this century. It is endemic north
of the Maramagambo Forest where contact between domestic lifestock and
wild bovids was probably common in the mid-nineteenth century and remains
frequent to this day.

In the last decade, the large-mammal community has suffered a
tremendous decline at the hands of man. This was first documented in 1974
by Eltringham & Malpas (1976, 1980) with respect to the elephant
population. Another onslaught followed during and in the aftermath of the .
1979 war when roving bands of men with automatic rifles killed large
numbers of elephant, hippopotamus, buffalo and topi. The magnitude of

22
their destruction has been estimated by van Orsdol (1979) and Malpas
(1980).

Regardless of other influences, the greatest threat to the long-term
survival of the park is the continued existence of numerous settlements
within and along its borders. When the park was created, some existing
villages in the area were either excised from it or were allowed too
remain within specified boundaries. Over the past three decades, however,
other villages have been established so that there are now 22 such
communities in the park. The population of these settlements has also
grown so that there are at least 70,000 people living in the park. This
is expected to rise even more sharply when the salt factory currently
being built in Katwe by a German company becomes operational.

The presence of a large human population within the park has
intensified the demand for space and firewood. All of the original
villages have expanded beyond their legal limits and the collection of
firewood and the grazing of lifestock both accelerate the destruction of
habitat within the park (Field 1968; Musoke 1980). The villages serve as
centers for poachers and markets for their illegal trade (Edroma 1973).
The distribution of elephant carcasses mapped by Eltringham & Malpas
(1980) strongly implicated the villages as havens for poachers. Domestic

stock may also act as a reservoir for disease (Woodford 1972).

2.2. Census areas

The study was conducted in the central sector of the park bounded by the
equator and the Kazinga Channel. Ground counts of large mammals were made

at monthly intervals in five selected areas (Fig. 2.4).

3000 1K '\
\.
m \v" I V 'I'(,' Equator

 

 

[——r——r—r—|
O 5 10 km $.33 Contour (thermal)
— Perk agenda,
- Buildings

-' to study areas

Fig. 2.4. flap of the sector between Kazinga Channel and the
equator, Queen Elizabeth National Park, Uganda, showing game
census areas.

23

24

Area 1 (4.4 sq km) was the Mweya Peninsula. The peninsula lies at two
levels, with a steep fault running north and south dividing the area into
approximately two equal parts. The area is heavily grazed by
hippopotamus. The vegetation is a mosaic of short grass with bush
thickets. and is one of the areas where the spread of thickets has been
most rapid. On higher ground there is more grass and less thicket, but
the lower portion of the peninsula and the steep slopes are densely
covered with bush. Because of the absence of any significant quantity of
grasses, fire is absent. In the late 19503 the peninsula experienced
serious habitat deterioration due to overgrazing. To reverse the
situation, 128 hippopotamus were eliminated in 1958 (Bere 1959) and until
1967 recolonization was prevented by shooting. A number of transects were
established in 1958 and changes in vegetation recorded until 1965 (Laws
1968; Thornton 1971). Animal counts were made by Petrides & Swank (1965),
Field & Laws (1970), Eltringham (1974, 1980) and Yoaciel (1981). This is
the ‘seriously overgrazed' study area of Petrides & Swank (1965) and Area
4 of Field 8 Laws (1970).

Area 2 (13.0 sq km) was part of the mainland adjoining Mweya
Peninsula. It is a series of relatively open grassy ridges separated by
valleys with dense bush thickets and strings of small wallows. At the
southern end, the vegetation is a mosaic of short grass and thicket
similar to that found on the Peninsula. The grass increases in height to
the north. Beyond the short grass area is a zone of taller grassland
dominated by Smrobolus pyramidalis, with Bothriochlog spp. and
Tephrosia gggg, creeping between the tussocks of Spgrobolus. Grazing is

heavy but less intensive than further south closer to the shoreline. A

25

mediumrheight, lightly-grazed Ihgmgggrhypgrrhenia grassland with patches
of Impgrata cylindrica dominating the northern portion. Fire is frequent
in the northern part, but absent in the heavily-grazed southern portion.
This is the ‘over-grazed' study area of Petrides 8 Swank (1965) and Area
6 of Field 8 Laws (1970).

Area 3 (26.2 sq km) was an area of volcanic craters north of
Kabatoro. The vegetation is tall grassland with scattered mature trees of
Acacia sieberiana and A, ggrrardii and bush thickets. The dominant grass

 

species are Hyparrhenia filipgndula and Ihgmggg triandra. Large patches
of Qymbopogon afronggggg and I. cylindrica occur on the floors of many
craters. Profuse regeneration of trees, particularly of A, ggrrardii, is
turning large sections of the northern portion of the area into woodland.
Fire is frequent during the dry season. This is Area 2 of Field 8 Laws
(1970) and overlaps the ‘long-grass' area of Petrides 8 Swank (1965).
Area 4 (17.1 sq km) was located at the junction of the Katwe-Kasenyi
Road and the main Mbarara-Kabarole (Fort Portal) Road. It has a series of
parallel ridges and wet bushed valleys with strings of small wallows. In
the west, the vegetation is a moderately grazed open fire-climax
grassland with small patches of thicket. The grassland is dominated by

 

Heteropogon contortus. Other common grass species include Spgrobolus
stapfianus, Microchloa kunthii. Hyparrhenig,filipendulg and Spgrobolus
pyramidalis. Thickets of Capparis and clumps of regenerating Aggglg
sieberiana tree occur mostly along the valleys. Trees of Euphorbia
candelabrum are common on ridge tops. Between 1958 and 1962 the eastern
half of the area was the location of an experimental 50% reduction of
hippopotamus by eliminating some animals occupying wallows in the area

26
(Bere 1959). This is the ‘moderately overgrazed' study area of Petrides 8
Swank (1965) and Area 8 of Field 8 Laws (1970).

Area 5 (26.2 sq km) was a slightly undulating, tall to medium height
fire-climax savanna grassland dotted with thickets. The Mbarara-Kabarole
and Katwe-Kasenyi roads mark its eastern and southern boundaries,
respectively. Kikorongo Crater is included in the study area and forms
its northern boundary. The western boundary is demarcated by a neglected
fire-break along the foothills of the Craters highlands. Its southeast
corner is contiguous with Area 4. Trees of Euphorbia candelabrum are
common in thickets. Hygarrhenia filipgndula and Ihgmggg triandra.are the
dominant grasses. Their associates include Heteropoqon contortus,

Bothriochlog spp., Spgrobolus pyramidalis, S. stapfianus, Microchloa

 

kggthii and Digitaria diagonalis. The southwestern portion is dominated
by extensive patches of Impgrata. Patches of Impgrata also occur on the
western border along the foothills of the Craters highlands and inside
Kikorongo Crater. Dense woodlands infested with the tsetse fly Glossina
palpalis occur on the fringes of Lakes Nyamunuka and Murumuli southwest
of the area. A territorial ground of the Uganda kob has persisted in the
area since the study of Buechner 8 Schloeth (1965) and probably long
before then. This is Area 7 of Field 8 Laws (1970).

. The five game census areas represented three habitat types. Areas 1
and 2 comprised a habitat of short grass with dense thicket close to
water. Areas 4 and 5 were a habitat of open, medium-height grassland
somewhat distant from water. Area 3 was a habitat of tall grass distant

from water.

CHAPTER 3

THE UNGULATE COMMUNITY

3.1. Introduction

During the 19503 and 19603, QENP supported one of the highest biomass of
wild ungulates in the world (Petrides 8 Swank 1965; Field 8 Laws 1970;
Leuthold 8 Leuthold 1976). Political unrest in the country during the
19703, however, resulted in large scale poaching and a massive decline in

the populations of many species (Eltringham 8 Malpas 1976, 1980; van
Orsdol 1979; Malpas 1980) including elephant (Table 3.1).

Table 3.1. Elephant abundance in Queen Elizabeth National Park,
Ugand . illgstrgting the recent gecline in game ngmbers.

 

 

 

Year Ngmper
1963 1574
1964 1759
1966 2888
1967 3448
1968 2267
1969 3265
1970 2042
1971 2222
1972 2209
1973 2264
1974 1450
1975 1047
1976 968
1980 153
1982 300*

 

Sources: Annual Report of the Nuffield Unit of Tropical Animal
Ecology 1963-1971; Buss 8 Savidge (1966); Field (1971);
Eltringham 8 Malpas (1976, 1980); Eltringham (1977); Malpas
(1980). *An estimate from several sightings within the park.

27

28
To restore the integrity and security of the national parks the
Government of Uganda with the co-operation of the United Nations
Development Programme (UNDP), European Economic Community (EEC) and
Frankfurt Zoological Society (FZS) launched a major drive in 1980 to
stamp out poaching. The decline in poaching activity in QENP (Table 3.2)

indicates that, in the short term, the program was successful in reducing

the level of poaching.

Table 3.2. The number of poaching incidents
recorded by author in QENP, Uganda, between

 

 

_1980Vgnd 1983.
Year Observed
1980 26
1981 7
1982 1
1983 (lst 6 months) 1

 

df=3, x2=48.08, P<<.01***

A park-wide aerial survey made in March 1980 (Malpas 1980) showed
that elephant numbers had fallen dramatically. To look further into the
matter, five areas (see Fig. 2.4) where ground counts had been made in
the past were re-surveyed. All five had been studied by Field 8 Laws
(1970) and four were studied by Petrides 8 Swank (1965).

3.2. Methods
3.2.1. Counts

The census method used was similar to that of Petrides 8 Swank (1965) and

29
Field 8 Laws (1970). It involved two observers counting from the roof of
a hard-top Land Rover driven over a series of approximately-parallel

transects about 800 m apart. Each observer counted a strip 400 m wide on

one side of the vehicle. The observers instructed the driver to stop from
time to time to enable them to enumerate large herds or to make closer
inspection with binoculars. During periods of heavy smoke haze in the dry
season or in areas where visibility was poor due to dense vegetation,
transects were closer together. Each area was covered as completely as
possible.

In contrast to procedures in the other areas and because of difficult
terrain, counts in Area 3 (the Craters highlands) were made largely from
a track. In the south, the route provided good observation points
overlooking relatively open country and permitted effective area counts.
In the north, visibility was often poor due to tall grass and dense
regeneration of acacia trees. There, transect were about 400 m apart
except after fires when improved visibility allowed more normal
procedures.

Counts were made once a month at about the same time each month. All
animals seen were recorded on outline maps along with information on
their sex, age, and group size. Familiarity with the areas contributed to
considerable accuracy in locating the position of animals. Animals which
were intolerant of approach had their numbers estimated first and counted
later, if closer observation was possible. With experience, the estimates
proved to be close approximations of subsequent counts and are believed
to have introduced no large errors. The chance of recording the same

animals twice was slight since herds could easily be identified by size,

30
composition and the presence of recognizable individuals.

The assumptions of the census method were (1) that all animals
present in an area were visible to observers and (2) that they could be
counted accurately. In practice, smaller animals (e.g. warthog) and those
with retiring habits (e.g. bushbuck and reedbuck) were more likely than
others to be missed. Warthog are also sensitive to temperature
fluctuations (Sowls 8 Phelps 1965) and probably sought shelter
underground during cool periods.

Even large and readily-seen ungulates may be difficult to count
accurately. Kob, in particular, occurred in large herds which were
constantly in a state of flux and very difficult to record. Some buffalo
herds were intolerant of approach and would stampede before observers
were close enough to count accurately. The method is inappropriate for
nocturnal animals. These were mostly either carnivores or omnivores with
little impact on the vegetation. The abundance and distribution of rats,
mice and other small mammals not assessed in this investigation have been
studied in detail by Neal (1967).

In spite of its limitations, the ground census method was an
invaluable tool for estimating animal abundance, particularly those of
the large and most important members of the ungulate community.

Hippopotamus on land or in wallows were recorded during counts. These
represented only a small fraction of those using the area at night,
however, especially in Areas 1 and 2. To obtain the numbers that
regularly used these areas, counts of hippopotamus along the adjacent
shoreline were made using a boat and outboard motor. During the count,

two observers with binoculars scanned the water ahead of the boat. The

31
behavior of hippopotamus in this area is such that initially, as a school
is approached, few are seen. Then at about 50-100 m, they evidently sense

the approach of the boat and surface for a short period before they
submerge and take evasive action. During this brief time, the school is

counted repeatedly. The maximum number obtained is assumed to be the
number present and was recorded. Counts were made on a monthly basis in
Area 1 but could be conducted only occasionally in Area 2 due to limited
availability of supplies, equipment and man-power.

3.2.2. Variability in numbers

The question of whether the monthly counts were representative samples of
ungulate populations was raised during a series of ground counts similar
to those made in this study (Field 8 Laws 1970). Eltringham (1973) made
an assessment of the variability in those counts and concluded that they

were adequate. The variability of counts in this study was assessed using

one-way analysis of variance.

3.2.3. Distribution with respect to habitat
The five census areas represented three types of habitat. Area 3 was an
area of long grass distant from water. Areas 4 and 5 were relatively-open
medium-height grasslands with scattered thickets somewhat distant from
water. Areas 1 and 2 were a habitat of short grass with dense thickets
close to water.

The abundance of each ungulate species in the three habitats was
compared. A 3—way analysis of variance (ANOVA) show the relative
importance of various populations in different habitats. The density of a

32

species in one habitat relative to another was assumed to be evidence of

preference.

3.2.4. Seasonal fluctuations in animal numbers

A 1—way ANOVA was used to test the hypothesis that the wet and dry season

means were the same for each species in each study area.

3.2.5. Population structure
Due to the low numbers of animals in census areas, data on group size,
age structure and sex ratio were collected for the whole sector between
the Kazinga Channel and the equator. This was necessary to increase
sample size as well as to make the results more representative of the

sector. Trips were made regularly to various parts of the sector to
collect data.

Group sizes: A group was defined as the number of individuals found
within 100 m of each other. A single animal was treated as a group of
one. Whenever a group was found partly within and partly outside a census
area during a count, the number outside the area was included in the
group size data but excluded from the census data.

Age structure gnd sex rgtios: The sex and relative age (such as
adult, sub-adult, etc.) of each animal in a group were recorded during
counts and other sightings. Relative age was based primarily on body size
and horn development. An animal classified as adult was a fully grown
individual. Sub-adults and juveniles were about three-quarters and
one—half adult size, respectively. Calves were recent offspring.

In many instances it was not possible to determine the sex of buffalo

33
with certainty because the herds were wary and fled when approached.

Routine sexing of hippopotamus immersed in water was also not possible.

3.2.6. Long-term trends
The large-herbivore community of Mweya Peninsula (Area 1) has been
studied almost continuously between 1956 and 1983 (Petrides 8 Swank 1965;
Field 8 Laws 1970; Eltringham 1974, 1980; Yoaciel 1981; this study).
Abundance of the major species over the 27-year period was analyzed to

reveal trends in population.

3.3. Results

3.3.1. Counts
Sixty nine counts were made in Area 1 and 19 each in Areas 2, 3, 4 and 5.
The average number of animals ranged from a low of 185.0 (11.7% of total)
in Area 1 to a high of 528.9 (33.41) in Area 5 (Table 3.3). Kob,

Table 3.3. Average large herbivore numbers in the study areas, Queen
Elizabeth National Park, Uganda. November1981 to June1983.

Study areas

 

Species 1 g 3 4 5 Total 1
Elephant 13.9 11.1 0.5 0.0 0.1 25.6 1.6
Warthog 7.1 14.0 2.5 8.2 3.5 35.3 2.2
Hippopotamus 149. 4 148.5 0.0 0.1 0.0 298.0 18.8
Reedbuck 0.0 0.0 1. 3 1.7 0.8 3.8 0.2
Waterbuck 0.0 14.4 17. 8 4.3 1.5 38.0 2.4
Kob 0.1 31.7 215. 8 271.2 499.3 1018.1 64.3
Bushbuck 5.6 10.9 0. 5 1.5 1.1 19.6 1.2
Buffalo 8.9 40.7 24. 6 48.9 22.6 145.7 9.2
Total 185.0 271.3 263.0 335.9 528.9 1584.1

Percentage 11.7 17.1 16.6 21.2 33.4 100.0

 

34

hippopotamus and buffalo were the most numerous species. They constituted
64.3, 18.8 and 9.2% of total numbers, respectively. All the other species
combined accounted for less than 8% of the total numbers recorded.

Hippopotamus were abundant in Areas 1 and 2 with densities of 34.0
and 11.5/sq km, respectively (Table 3.4). Their abundance was associated
with the proximity of these areas to water. They were virtually absent
from the study areas distant from water.

Large concentrations of kob occurred in Areas 4 and 5 where densities
of 15.9 and 19.9/sq km, respectively, were recorded (Table 3.4). Their
densities in other areas were much lower. In Area 1, they were virtually

absent. No other species had a density higher than 3.2/sq km in any study
area.

Table 3.4. Average large mammal densities/sq km in study areas, Queen
Elizabeth National Park, Uganda. November 1981 to June 1983.

Study area 1 2 3 4 5

 

Area (sq km) 4.4 12.9 26.2 17.1 26.2
No. of counts 19

Elephant
Warthog
HWm
Reedbuck
Waterbuck
Kob
Bushbuck
Buffalo

Total

01

(O
H
to
H
to

19

0.)
99

C’F’F’é’f‘?”
H
OH?P

H NQmHOU'IH‘D

H
009999?

O OOOOOOOO'H"
H PPP?

0 99999999
to OONVOOHO

H

9 P99???

\| whammy-some
N QQHHOOHO

0° P°f‘

.3:
N
$3 53

N
H

A When the average densities obtained in this study are compared with

35
those of 1956/57 and 1963/67, it is apparent that the abundance of large
mammals have changed considerably (Tables 3.5). It is also evident that

for most species and in most study areas, the changes have been

reductions. Elephant and buffalo show the largest decline. For elephant,
the magnitude of the reduction has amounted to percentage changes of
-47.9 and -74.3 from average densities in 1956/57 and 1963/67,
respectively. Corresponding figures for buffalo are -60.9 and -86.2. The
decline of large mammals generally, and of big game specifically, has
been due to intensive poaching in the park between 1973 and 1980 (Edroma
1973; Eltringham 8 Malpas 1976, 1980; van Orsdol 1979; Malpas 1980).

Table 3.5. Percentage changes in densities of large herbivores in 1981/83
from 1956/57 and 1963/67 levels in Queen Elizabeth National Park,
Uganda. ‘

 

 

Compared with 1956/57* Compared with 1963/67**
, S t u d y a r e a 3
Species 1 2 3 4 1 2 3 4 5
ELE +81.6 -74.4 -98.7 -100.0 +2.9 -75.5 -99.0 -100.0 -100.0
WHG -78.8 -59.2 +138.5 -43.6 -78.8 -62.6 +377.1 -56.8 -74.8
HIP - - - - +3631.3 -43.9 0.0 -99.9 -100.0
RBK 0.0 0.0 +24.0 -17.2 0.0 0.0 +65.4 -33.7 -82.0

WBK -100.0 -66.8 +195.5 -59.4 -100.0 -68.3 +83.6 -79.4 -96.9
K08 0.0 +434.2 +6763.9 -14.1 +13.6 +667.9 +715.5 +58.8 -56.6
BBK -57.7 -56.5 +INFIN +119.3 -53.7 +14.2 +90.8 -53.8 +39.9
BUF -45.5 -21.5 -92.4 -84.3 -91.8 -82.1 -80.1 -83.7 -93.3

 

*Petrides 8 Swank (1965); **Field 8 Laws (1970).

Despite the general decline in abundance, there has been an increase
in game in Area 3 (the Craters highlands) (Table 3.5). The largest
increases were for kob, with percentage changes of +6763.9 and +715.5

36
from 1956/57 and 1963/67 levels, respectively. Warthog, too, show a large
percentage increase in this study area. The numbers of warthog were
initially extremely low, however, so that even a modest increase resulted
in a large percentage change.

Kob increased in most study areas and decreased significantly only in
Area 5 (Table 3.5). The decrease in this study area represented a major
change in kob distribution. Large concentrations of kob associated with
communal territorial grounds have been a feature of this area for a long
time (Buechner 1961; Buechner 8 Schloeth 1965; Modha 1972).

The decline in kob abundance in Area 5 is believed to have been due
to poaching. The study area is bounded by heavily-travelled roads that
provide poachers with easy access. From January 1980 to June 1983, the
author recorded more incidents of poaching in this study area than in the
others combined. It is believed that harrassment by poachers forced some
animals to move to less accessible areas. The large increase in the
number of kob in Area 3 supports this view.

There were also examples of local exceptions to the general decline.
The most dramatic involved the increase in hippopotamus in Area 1 where a
percentage increase of +3631.3 occurred relative to their 1963/67 density
(Table 3.5). The large increase was due, in part, to the fact that counts
in 1963/67 were made during a period of hippopotamus cropping which
virtually eliminated the species from the area (Field 8 Laws 1970). It
also emphasizes the remarkable recovery that the hippopotamus population
has made since the cropping program ended. Hippopotamus density during

this study was higher than the 1956/57 and 1963/67 levels and occurred at
the expense of all other species.

37
Area 1, the Mweya Peninsula (see Section 2.2), is the only study area
in which elephant showed a modest increase (Table 3.5). This study area
is, however, only 4.4 sq km in size. Given the propensity of elephant to

range over wide areas, such a small study area could not be expected to
‘have nor did it have a resident elephant herd. Elephant were, however,
found there more frequently than in any other study area. The reason for
the restricted movement was probably security. Area 1 was the site of the
park headquarters. For this reason, the area and its surroundings
(including Area 2) were avoided by poachers despite the breakdown of law
and order in the rest of the park. As the most secure site in the entire

park, it is suggested that it was used more frequently and intensively
than it would otherwise have been.

Table 3.6. Average standing-crop biomass (kg/ha) in the study areas,
Queen Elizabeth National Park, Uganda. November 1981-Juneg1983.

Unit
Species weight Study areas

(kg) 1 2 4 5
Elephant 1700 53.53 14.52 0.31 0.00 0.03
Warthog 50 0.80 0.54 0.05 0.24 0.07
Hippo 1000 339.59 114.23 0.00 0.02 0.00
Reedbuck 50 0.00 0.00 0.02 0.05 0.02
Waterbuck 160 0.00 1.77 1.09 0.40 0.09
Kob 65 0.01 1.58 5.35 10.31 12.39
Bubshbuck 50 0.64 0.42 0.01 0.04 0.02
Buffalo 395 7.98 12.36 3.71 11.29 3.41
Total 402 55 145.42 10 54 22.36 16.03

 

The species of ungulates involved in the study vary greatly in size
and weight, and consequently in their energy demands on the environment.

38

Due to these differences, comparisons of biomass rather than numbers may
be a more valid measure of the relative importance of a species in the
environment (Table 3.6). The unit weights used in calculation were taken
from Field 8 Laws (1970) and were derived from studies of animal
populations in the park. The age structure of the populations was taken
into consideration in estimating average individual weights. These
weights are substantially lower than those used by some authors (e.g.
Petrides 8 Swank 1965).

Since these weights were determined, many populations have suffered
high mortality as a result of the intensive poaching of the 19703. There
is a possibility, therefore, that the age structures and unit weights of
the populations studied were quite different. Until new average weights
are determined, however, these remain the best estimates available.

From the data (Table 3.6), it is evident that elephant, hippopotamus
and buffalo dominated the standing-crop biomass of ungulates in the study
areas. Although hippopotamus accounted for only 18.8% of animal numbers
(Table 3.3), they constituted 76.01 of the biomass. Similarly elephant
constituted only 1.6! of the numbers but 11.5% of the biomass. Kob, on
the other hand, contributed 64.31 of the numbers but only 5.0% of the

biomass. The ungulate biomass in the study areas evidently was dominated
by the few large species.

3.3.2. Variability in numbers
Overall variability in numbers of most species was high (Table 3.7).
Hippopotamus numbers along the shoreline showed little variability in

contrast to large fluctuations in numbers seen on land. This is not

39

Table 3.7. Summary of counts of large mammals in the study areas
in Queen Elizabeth National Park, Uganda, November 1981 to June

 

 

1983.
Study areas
1 3 5

No. of counts 69 19 19 19 19
Elephant

Mean 13.86 11.11 0.47 - 0.05

SD 32.37 34.60 2.06 0.23

Range 0-143 0-150 0-9 - 0-1
Warthog

Mean 7.07 13.95 2.47 8.21 3.47

SD 2.33 7.60 3.99 6.16 6.23

Range 1-12 3-28 0-14 0-20 0-28
Hippopotamus (on land) .

Mean 17.07 0.37 0.05 -

SD 16.55 0.60 - 0.23 -

Range 0-75 0-2 0-1 -
Reedbuck

Mean - 1.26 1.68 0.84

SD - 2.98 2.69 1.77

Range - 0-12 0-12 0-6
Waterbuck

Mean - 14.42 17.84 4.26 1.47

SD 9.20 19.67 8.93 3.44

Range - 0-39 0-76 0-33 0-14
Kob

Mean 0.06 31.68 215.79 271.16 499.32

SD 0.34 59.88 185.73 239.91 361.76

Range 0-2 0-260 7-629 27-1122 37-1417
Bushbuck

Mean 5.62 10.89 0.47 1.47 1.11

SD 3.59 6.45 0.96 1.54 1.76

Range 0-16 2-27 0-3 0-4 0-6
Buffalo

Mean 8.88 40.68 24.58 48.89 22.63

SD 11.91 36.94 39.33 39.01 70.64

Range 0-42 0-125 0-124 0-98 0-305
Hippopotamus (along shoreline)

Mean 149.42 148.50 - - -

SD 23.44 17.03 - - -

Range 111-198 122-169 - - -

 

surprising as hippopotamus habitually return to the water by day. The

numbers of warthog, too, were relatively stable. The warthog is largely a

40
sedentary animal and individuals in an area were probably resident. A
major cause of variability in warthog was the tendency of individuals to
go underground during cool weather (Sowls 8 Phelps 1965; Clough 8 Hassam
1970). They also lay under thickets during the heat of the day and could
have been overlooked.

Numbers of kob were less variable than those of waterbuck. This was
probably because kob tend to aggregate in large herds which remain in one
area for considerable periods. Waterbuck, on the other hand, were less
common and could be more easily overlooked. They also wander over a
larger home range (Spinage 1967, 1982). The numbers of bushbuck were more
variable than would be expected from their restricted home range (Waser
1975; Allsopps 1978; Okiria 1980). Variability of this species is, at
least in part, due to the low numbers recorded. Areas with particularly
low numbers (3, 4 and 5) showed the greatest variability. Bushbuck are
retiring in habit and are easily overlooked. Their numbers were estimated
to be up to 2.6 times higher than those normally recorded (Waser 1975).
Reedbuck occurred in only a few areas and in such low numbers that few
conclusions can be drawn from the results.

Counts of buffalo were quite variable. The large variability was due
to periodic movement of herds into and out of study areas. Irregular
arrivals or departures of herds of various sizes affected the variability
of counts.

Variability in counts of elephant was probably due to their ability
to range over large areas. This result was not expected in view of their
apparent confinement to Areas 1 and 2. None of the census areas was,

however, large enough to support a resident population of elephant.

41

The preceding discussion suggests that variability was influenced by
several factors. In the case of small animals or those with retiring
habits, variability was partly due to observers failing to see all the
individuals present. Species in this category included bushbuck, reedbuck
and warthog. For other species, variability was due to movement which
took animals out of an area in which they were counted or, conversely,
into an area from which they were previously absent. Examples of such
species are elephant and buffalo. There were species, such as kob, which
were conspicuous and relatively sedentary but were difficult to count

accurately. Kob frequently occurred in large herds which were constantly

in a state of flux and difficult to enumerate.

3.3.3. Distribution according to habitat
The various species of large herbivores were not uniformly distributed in
the study areas. Each species was relatively abundant in some areas and
uncommon in others (Tables 3.4 8 3.6). Of the three habitat types
studied, the site closest to water (Areas 1 and 2) supported the highest
and the tall grassland (Area 3) the lowest biomass (Table 3.8).

Elephant and hippopotamus occurred almost exclusively in the habitat
of short grass with dense thicket close to water (Table 3.8). There,
hippopotamus and elephant accounted for 82.9 and 12.41, respectively, of
total biomass. Neither contributed significantly to biomass 'in the
habitats away from water.

The composition of standing-crop biomass in the habitats distant from
water show a remarkable simnlarity (Table 3.8; Fig. 3.1). In both

habitats, kob constituted over one-half and buffalo more than one-third

LARGE MAMMAL BIOMASS COMPOSITION

NMWMWM

  

WALD (33.13)

 

m (3.7:) “a (son)

 

 

lNTHEIlEDINGRASSl-MTAT “WM/MUM”
mums (2.7x) mun (3.7:)
wows)
com (um
sumo (awn
was (mm
m (nu)
Fig. 3.1.

42

43

Table 3.8. Average standing crop biomass (kg/ha) according to habitat
typeLQueen Elizabeth Ngtiongl Pgrk. Uganda. November 1981 to Jgne 1983.

 

Habitat type

Mediumrheight Short grass and
Tall grass, grass, somewhat thicket bushes,

 

 

distant from distant from close to water
Species water (1) water (2) (3)

Biomass 1 Biomass 1 Biomass 1
Elephant 0.31 2.9 0.02 0.1 34.00 12.4
Warthog 0.05 0.5 0.16 0.8 0.65 0.2
Hippopotamus 0.00 0.0 0.02 0.1 227.52 82.9
Kob 5.35 50.8 11.34 59.0 0.80 0.3
Buffalo 3.70 35.1 7.36 38.3 10.25 3.7
Others 1.12 10.6 0.32 1.7 1.39 0.5
Total 10.53 19.22 274.61

 

the biomass. In contrast, kob contributed only 0.31 of the biomass in the
habitat close to water and buffalo accounted for only 3.71.

The results of a 3-way ANOVA to test for the impacts of replication,
area and seasonality on species density and total biomass (excluding
hippopotamus) show that most of the variability in species densities and

total biomass were explained by differences between areas (Table 3.9)

supporting the general impression of habitat preferences among ungulate

populations.

3.3.4. Seasonal fluctuations in numbers
While the 3-factor ANOVA (Table 3.9) revealed no significant differences
in species density between the dry and wet seasons (except for reedbuck,

df = 1, F = 6.28, P =1.37), one-way ANOVAs showed significantly different

44

values for bushbuck in Area 1 (df=1, F=2.84, P=9.29), warthog in Area 4
(df=1, F=5.49, P=2.99) and reedbuck in Areas 4 (df=1, F=3.16, P=9.01) and
5 (df=1, F=4.40, P=4.87). Error terms within seasons were generally
large, however, indicating that the simple designation of dry versus wet
was probably too crude a measure. Testing for each month of the dry and
wet seasons as a sub-season, mean densities for sub-seasons were
calculated from weekly counts made in Area 1. The results (Table 3.10)
revealed significant differences for elephant (df=5, F=7.10, P=0.01),
hippopotamus on land (df=5, F=2.52, P=3.77) and warthog (df=5, F=1.97,
P=9.39).

The results of the one-way ANOVA confirmed that considerable
variability occurred within seasons. Differences in results of the
seasonal and sub-seasonal ANOVAs indicate that animal movement associated
with moisture levels were hidden by the large variability within both dry
or wet seasons. The results provide empirical evidence that large
herbivores in the park were sensitive to and responded to changing
conditions within seasons.

Despite the results of the 3-factor ANOVA (Table 3.9) which showed no
significant differences in animal densities between seasons, field
observations indicated that some species undertook local migrations of 10
to 15 km in response to environmental conditions. Kob and buffalo, for
example, showed a tendency to move closer to permanent sources of water
during the dry season. As a consequence, their dry season densities in
habitats further from water were lower than wet season densities (Fig.

3.2), although the differences were not statistically significant.

Fig 3.2 Comparison of season densities
in habitats near and for tcr.

 

 

 

 

 

 

W1

A
A
7
A

/////
////////

§////

W

\\\\\\\l

A

/

i

 

 

 

 

 

 

 

 

 

 

 

W

\

 

/////////////41

 

 

 

5W]

 

 

 

gV///

a

 

C3
Q9 a:
3% 5.
e...

O

E
[Z

46

Table 3.9. Results of a 3-factor ANOVA to determine the impacts of
season, area and replication on species density in the study areas,

Queen Elizabeth Nation§1_Park, annda, November 1981 to June 1983.

 

Elephant Warthog Hippopotamus
df F p! F p! F p!
Replication 9 0.91 1.81 7.88% 0.96
Area 4 4.54 0.27*** 74.04 0.0*** 36.45 0.0***
Season 1 0.36 1.58 21.01 0.35
Area x Season 4 0.13 1.86 12.48 0.35
Residual error 77
Reedbuck Waterbuck Kob
df F p! F p1 F pt

Replication 1.19 31.45 1.00 44.69 0.30

9
Area 4 4.64 0.24*** 16.49 0.0*** 13.48 0.0***
Season 1 6.28 1.37** 2.60 10.73 0.21

Area x Season 4 1.67 16.46 0.70 0.55
Residual error 77

Bushbuck Buffalo Total biomass
df F p! F p1 F p1
Replication 9 1.29 25.67 0.80 0.77
Area 4 36.04 0.0*** 3.77 0.77*** 3.60 0.97***
Season 1 0.60 0.21 0.11
Area x Season 4 0.32 0.50 0.23

Residual error 77

 

*Significant at 901 level; resignificant at 95% level; ***significant
at 991 level.

When weekly-count data in Area 1 were organized according to monthly
sub-seasons (Table 3.11), the average numbers for warthog, bushbuck and
hippopotamus along the shore were fairly constant, indicating negligible
seasonal movement. These populations were, therefore, largely resident. A
large influx of elephant occurred at the end of the dry season and was
sustained into the beginning of the wet. Averages for hippopotamus on
land were higher during the second and third months of the dry season,
indicating an increased tendency to stay on land when forage was reduced
in quantity. Buffalo increased at the end of the wet and the beginning of

the dry season, the reverse of the pattern shown by elephant.

47

Table 3.10. Results of 1-way ANOVAs on sub-season data from study Area 1,
Queen Elizabeth Nationglyflgrk. Uggggg. November 1981 to June 1983.

Elephant Warthog Hippopotamus
df F p1 F p% F p!
Total 68
Between 5 7.10 0.01***1.97 9.39% 2.52 3.77**
Within 63
Kob Bushbuck Buffalo
df F p! F p1 F p!
Total 68
Between 5 1.80 12.5 1.67 15.3 0.20
Within 63
Hippo along shore
df F p1
Total 18
Between 5 1.60 22.86
Within 13

 

*Significant at 90: level; arsignificant at 93: level.

Table 3.11. Average numbers of animals in study Area 1 according to

sag-seasons, Queen Elizabeth National Park, November 1981 to June
1 3.

Sub-
seasonsa Elephant Warthog Hipposa Hippo*** Bushbuck Buffalo

6.47 6.07 9.80 171.50 5.00 10.07
2.92 7.00 27.67 138.33 5.00 7.17
64.50 8.75 26.13 143.00 4.63 6.38
21.10 6.30 15.90 131.00 4.40 9.80
1.63 7.88 12.00 146.33 6.13 8.00
5.25 7.31 14.69 159.00 7.69 10.19

0301.5me

 

*1, 2 and 3 are respectively the lst, 2nd and 3rd months of the dry,
and 4, 5 and 6 are the lst, 2nd and 3rd months of the wet season.

Hippo** and Hippo*** are respectively counts of hippopotamus on land
and along the shoreline.

3.3.5. Population structure

Group sizes:

Elephant: Fifty nine elephant groups were recorded. Their sizes
ranged from 1 to 201 (Fig. 3.3a), with a mean of 39.6. Solitary animals
comprised over 131 of the groups. More than half (53%) the groups were
between 2 and 50. About 121 of the groups were larger than 100. The large
groups probably represented multiples of family units which had banded
together for safety and leadership (Laws pp p1. 1975).

Warthog: Some 449 groups ranging in size from 1 to 13 were recorded
(Fig. 3.3b). Fewer than 15% of the groups involved more than 4
individuals. Average group size was 2.5. Groups of one to 4 were common,
with nearly 371 of the groups composed of solitary animals.

Hippopotamus: Of 385 groups of hippopotamus recorded along the shore,
201 were solitary animals and nearly 45% included 2-10 individuals (Fig.

3.3c). Those between 11 and 20 made up 151 and groups >20 comprised only
. 201 of total.

Reedbuck: Data on reedbuck were limited but indicated that
individuals are predominantly solitary. The mean size for 67 groups was
1.4, with group sizes one and two comprising 66 and 321, respectively, of
the total observed. Groups larger than 2 comprised less than 21 of the
total recorded (Fig. 3.3d). Reedbuck numbers and group size were
influenced considerably by the occurrence of fire when then, temporary
concentrations developed in relatively open areas. In the open, too,
these cryptic antelopes were easier to see. The larger groups recorded in

counts, including one of 12, were seen after fires.

Waterbuck: Some 178 groups, ranging in size from one to 38, were

Fig. 3.30. Elephant groups
n=59

 

 

 

 

 

 

 

\\\\\\\

 

 

 

 

 

 

 

 

 

11-20 21-30 31—
Size classes

 

Fig. 3.3b. Warthog groups
n=449

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T
1111111

S ssssssssss

 

Fig.3.3c. Hippopotamus schools
ll - 385

 

 

 

 

 

 

Jm
\\\\\

 

 

 

 

 

 

 

 

 

 

 

 

S ssssssssss

Fig.3.3d. Reedslguck rou s
70
60‘ .
w- /

0 T l l
1

2
Group siz

26

18-
x

24+ /
2:: A

7
g
l: g 7? '7
'2: W M A
:: WM « i
: 443446633/5,

Fig. 3.3e. Waterbuck groups
n=l78

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T l T I l T T l I
2 3 4 5 6 7 8 9 10 >10
Gmup sizes

 

54
recorded (Fig. 3.3e). Solitary animals were the most common, accounting
for nearly 261 of the groups recorded. Groups of 2 to 4 individuals were
also frequent. Groups larger than 4 were increasingly less common. The
average group size was 6.0. The preponderance of solitary animals was due
to relatively large numbers of solitary adult males in the sample.

Kob: For most species, deciding what constituted a group presented
little problem because groups tended to be quite distinct. In dealing
with kob, however, it was not always easy, especially when confronted
with large numbers of animals spread out over a large area. The social
organization‘ of these antelopes involves communual breeding territories
(Buechner 1961). There were large numbers of animals associated with each
territorial ground and large numbers of females and young circulated
among territorial grounds located several km apart. These large
congregations accounted for some very large groups recorded. Large groups
also occurred in areas of tender green grass flushes.

Group sizes ranged from 1 to 1152 (Fig. 3.3f). In spite of the
tremendous variability, average group size was surprisingly small (81.8).
This was due to the presence of a large number of relatively small groups
as well as to a large number of solitary males holding single
territories.

Bushbuck: A total of 493 bushbuck groups was recorded (Fig. 3.39).
Groups ranged in size from 1 to 7, with a mean 1.2. Eighty three percent
of the sightings were of solitary animals. The most frequent association
was that of a female with calf, although mature animals of the same or
opposite sex were also recorded. Groups larger than two were relatively

rare. The largest group comprised 4 adult males and three adult females.

Fig. 3.3i. Kob groups
n=347

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Per-cc nt freq ue racy

90

Fig. 3.3g. Bushbuck groups

 

80-

701

504

5OJ

404

20-

10-

 

 

 

 

 

/lz771

 

“W

l l l l l l

2 3 4 5 5 7
Group size (n = 493)

56

 

28

251
24..

20-
)x
g 18-
8
U 16-
2
«- 144
5 12-
2
g 101

OKs-Fa:

Fig. 3.3h. Buffalo: bachelor groups
n=36

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4   44 7

/ / Z Z

 

 

 

 

 

 

 

 

1 \\\\\\\\\\\\\X\.\\\\\\\\\

l l l l T l l l l T l l l
l 2 3 4 5 6 7 8 9 10 ll 12 l3 14
Group sizes

 

Flg. 3.3i. Buffalo ::::::
55

 

 

 

 

 

 

 

 

 

 

 

71/), W

 

59

Table 3.12. Size distribution of buffalo bachelor
groups, Queen Elizabeth National Park, Uganda,
November 1981 to June 1983.

Herd size Freggency Percentage

13.9
13.9
16.7
27.8

5.6

2.8

H
ODQVOU'I-waI-I

H
H

HHI—n
chi-0N
HHNNH I I H I NOO’U‘IU‘I

wwwww
$030030)

 

Table 3.13. The size distribution of buffalo herds
in Queen Elizabeth National Park, Uganda, November
1981 to June 1983.

 

Herd size Frequency Percentpge
1 - 50 26 47.3
51 - 100 24 43.6
101 - 150 4 7.3
>150 1 1.8

 

Buffalo: Buffalo occur in two kinds of social units. One contains
members of both sexes and are classified as ‘herds,’ while the other is
comprised of males only and are called ‘bachelor groups' (Eltringham 8

60
Woodford 1971). Of 91 buffalo groups recorded, 36 (39.6%) were bachelor
groups (Table 3.12; Fig. 3.3h) and 55 (60.4%) were herds (Table 3.13;
Fig. 3.31). Among bachelors, group sizes ranged from one to 14. Four was
the most common group size and was recorded 27.8% of the time. The data
show a slight increase in frequency with size up to 4. Few bachelor
groups were larger than 4. The average size of bachelor groups was 4.8.

Among herd groups, sizes ranged from 2 to 305, with a mean of 55.3.
Groups between 2-50 and 51-100 were about equally common. Only 5 of the
55 herds recorded were larger than 100.

Large herbivores typically exhibited three broad categories of
groupings: (1) species such as elephant, buffalo, waterbuck, kob and
hippopotamus occurred primarily in large groups; 2) others (e.g. warthog)
commonly occurred in smaller groups or family units and 3) a few, such as
bushbuck and reedbuck were predominantly solitary. These observations
probably indicate that ungulate species large enough to see one another
above the top of the grass tended to form herds, while those which are

small in stature live solitarily or in small family groups.

Age structureypnd sex ratios:

In all six species for which sex ratios were determined, the proportion
of females was higher than that of males (Table 3.14). Bushbuck showed
the greatest disparity with a male-to-female ratio of 1 to 3.2. The ratio
was closest to parity in reedbuck. The sex ratio in this species was
1:1.6. ’

The percentage of calves in the population ranged from a low of 0.2
for bushbuck to a high of 19.3 for warthog. Warthog was the only species

61
among those studied which produced more than one offspring in a litter.
The high proportion of early juveniles in its population relative to
those of the other species was, at least in part, a reflection of this
capacity.

Elephant, hippopotamus and buffalo had surprisingly high proportions
of calves compared with the smaller species, except for warthog. This was
unexpected given the greater longevity and lower reproductive rates of
the larger mammals.

Sixty six percent of bushbuck recorded were adult females. The
preponderance of adult females appears to be a feature of small to
medium-sized antelopes. Reedbuck, kob and waterbuck also exhibited the

phenomenon, with adult females constituting 52.1, 62.7 and 50.71 of the
population, respectively.

Table 3.14. Age and sex classes of eight species of herbivoresr in Queen
Elizabeth Nptionpl Pprk, ngpgp. November 1981 to June 1983.

Percentage of gge/sex clpsses in populgtion
Number

in Adult Sub-adult Juvenile ratio
Species sglllple mple female mple femgle lp_ale fema_le Cjalves m-f

Elephant 1856 12.7 33.4 11.3 21.5 6.6 5.9 8.6

1:2.3
Warthog 1106 30.4 37.4 3.4 5.2 1.5 2.8 19.3 1:1.8
Hippo 4458 '63.9- '17.3- '8.9- 9.9 -
Reedbuck 94 36.1 52.1 2 8 4.9 0.0 2.1 2.1 1:1.6
Waterbuck 1068 17.9 50.7 5.4 10.3 5 6 4.2 5.8 1:2.5
Kob 28397 14.8 62.7 7.8 4.7 4 3 1.1 4.6 1:2.7
Bushbuck 505 18.1 66.0 4 6 6.8 1 3 3.1 0.2 1:3.2
Buffalo 3218 -73.4- '11 3' -4.5- 10.8 -

 

*Some age/sex criteria used were adopted from Grimsdell (1969) for
buffalo, Modha 8 Eltringham (1976) for kob, Child, Sowls 8 Mitchell

(1965) and Bradley (1972) for warthog and Spinage (1967, 1968) for
waterbuck.

Average density (/sq km)

Fig. 3.4. (Some population trends

on Mweya Peninsula. 195610 1983.

 

20-

15-

 

 

 

T l l l T T l T l

'56/5763/67 '68 ’59 ’70 ’71 ’72 '73 74/7881/83

Yeor(s) of census
+ Hi 4 Wh A Wb X Bb v Bu

62

 

63
3.3.6. Population trends on Mweya Peninsula
Populations of large herbivores on Mweya Peninsula have been monitored
almost continuously from 1956 to 1983 (Petrides 8 Swank 1965; Field 8
Laws 1970; Eltringham 1974, 1980; Yoaciel 1981 and this study). Elephant,
buffalo and waterbuck increased between 1956/7 and 1963/7 (Table 3.15) at
a time when hippopotamus were virtually eliminated as part of a
management program (Bere 1959). The gain was sustained and densities of
the various species remained more or less stable until 1973. Counts since
1974 showed a dramatic decline in all species except hippopotamus. The
hippopotamus population recovered quickly after the cropping program
ended in 1967 and by 1972 had exceeded the 1956/57 level (Table 3.15;
Fig. 3.4). Losses by buffalo and waterbuck accounted for most of the

Table 3.15. Average densities/sq km of six herbivore species between 1956
and 1983 on Mweya Peninsula, Queen Elizabeth National Park, annda.

1956/7* '63/7 '68 '69 '70 '71 '72 '73 '74l6 '81/3

Elephant 1.92 3.07 5.16 2.22 3.20 2.13 1.87 1.69 1.9 3.1
Hippo 19.01 0.91 4.80 6.40 9.78 15.56 19.73 22.49 23.7 34.0
Warthog 8.36 7.62 5.96 5.07 4.53 5.42 4.71 6.49 3.4 1.6
Waterbuck 4.04 10.46 8.80 7.56 7.91 7.47 6.76 7.56 1.1 0.0
Bushbuck 3.31 2.75 2.13 1.51 1.87 2.58 2.13 2.58 1.7 1.3

0.8 2.0

Buffalo 4.08 24.75 35.

‘D

1 27.29 18.04 21.78 23.82 20.27 1

*Authorities: Petrides 8 Swank (1965) for 1956-57; Field 8 Laws (1970)

for 1963-67; Eltringham (1980) for 1968, 1969, 1970, 1971, 1972 and 1973;
Yoaciel (1981) for 1974-76; this study for 1981-83.

 

recent decline of large mammals other than hippopotamus from the

Peninsula. The dramatic increase of buffalo and waterbuck during the

period of hippopotamus cropping and their recent decline following

 

0

|0mGSS COI’TlpOSI lOfl

0

Fig. 3.5. Trends in b

on Mweya Peninsula, 195610 1983.

 

l

 

- ‘ ‘ ‘ ‘ ‘ l I

1000000000000» .0 /
v000000000000. -_

v 0 0 0 0 0 0

“00000000000. -_

0 0 0 0 0 00. 5

100000000000 I n.

T

 

“““““4‘ -
00000000000000000000006 _ g .
. 0 0 0 0 0 0 0 0 0

0 0 0 0 0 A .
V 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 G .
1.000000000000006 .
.0000000000000000000000l .

l

'73 ’74/6 ’81/3

 

I

’72

 

- 0 000000000000000000000900

oN0N00010000t4000004040.101010“ _ a
0
0 0 0 0 0

0.0

v-‘-“‘-“““.

.000000000000000000000000000. _ - /
.000000000000000 00000000600. .

000000000000000N00000’00000. _ \

.0000000000”’0000000000 0000b \ n

0000000000 0000000000000D'0t _ \L -

vN0H00000000000N0H0H0N0H0H00A.u 1”
a000000000000000000000000. 00o \

.0000000000000000000000000000._ k

 

.‘0.“‘{“‘0.‘.‘

V00000000000000000000000000000. I /
000000000000000

.0000000000000000000000000'00.. 1

.000000000000004 I

.000000000000000000000000.0600s '

.0.0000000000000000000000000000I

 

' I
'1
'i
o
4
o

6 8
'15::

f
0
0
~:~
0

 

4 ‘ 0 0 .’ ‘ ‘ - 0 0 -

V 0 0 0 00000000000000000000000000000

u0000000000000000000b000 000000000000000000000000
0 000 0 0 000 0 00A

0 0 0 0 0 0 . 0 0 0
0 0 0 0 0 0 0 0 0 0
000000000000000 000 0 0
p00000000000000000000000000000000000000000000000.

l

 

‘ 0 0 0 0 0 0
404040040N0040000N0 0006H0010006M00000000000008 I //.
\\%}//////A
.000000.00‘4‘0‘000‘.000‘0‘00
0000 0 0000000 00000 000 0 0 00000 0 0 0 0 00000000000.

000000
00000000000000.000000000
I. 0 0 0 0 0 0 0 0 0 0 . .000. 0.00.00000”0"0.

00 0 0 000 00000 - /
0.00.00.000.00.000.00000.00
nounununu&nunnnnnunununuuuuuuuuun«MunnuuuuunNunnnnuuiu. ” /

000000000000 0 \t” k,

000 O 0 0 0
‘00 0 0 000 0 0 0 0 0 0 00
I0000000000000H0000000000000000000000000000000000000...“ —

V00000N0000060000000000000000000 0 0 0 0 0 0006.
v0.0 .
mnnnwnmm /
. con.“ \l
.ODOioLkl ‘ q _
o 0

 

8888.???»88.”88888889.... \\1
q d _ u

l

’56/7 '63/7 '68

 

_
mwwmwwwwzio

fiOfiUOQfl £000 \Afl C0.¥]D.L¥C00 OQUuCOULOQ

’70 '71

'69

_ Yearsig
’Wh k\\‘ KEBb mBu

mm

ZZIEI

64

65
hippopotamus recovery indicates that hippopotamus compete directly with
these species.

It is apparent from the changes in biomass composition between 1956
and 1983 (Fig. 3.5) that there has been a rapid and steady rise in
hippopotamus biomass since the end of the cropping program. During the
period of cropping, hippopotamus constituted only 5.0% of the biomass.
Five years after the program ended they had risen, once again, to a
position of prominence accounting for 58.51 of the biomass. Hippopotamus
gained dominance at the expense of the other species, accounting for
84.3% of the biomass in 1981/83. The situation observed during the study

was, therefore, similar to that of 1956/57 with a virtual monoculture of
hi ppopotamus.

3. 4. Discussion

3.4.1. Counts
Counts in this study demonstrated that for most species of large

herbivores and in most study areas, numbers were considerably lower than
in 1956/57 and 1963/67. (The few notable exceptions were kob, which
increased in most study areas, and hippopotamus in Area 1.

Poaching was the major cause of the decline in game abundance. Bands
of armed men roamed the park between 1973 and 1980 and inflicted a heavy
toll on animal populations, particularly elephant and buffalo (Edroma
1973; Eltringham 8 Malpas 1976, 1980; van Orsdol 1979; Malpas 1960;
Yoaciel 8 van Orsdol 1981). Eltringham 8 Malpas (1976, 1980) showed that
the distribution of elephant carcasses and other animal remains were

correlated with the location of villages, providing evidence that game

66
mortality was due to human activities. Edroma (1973) found that the
villages in and around the park provided a haven for poachers and a
lucrative market for their illegally acquired commodities.

3.4.2. Variability in numbers

Variability of the counts was high. For most species there were no marked
trends in the monthly counts so that any one observation was as likely to
be representative as any other. Variability was influenced by several
factors. In the case of small animals or those with retiring habits,
variability was partly due to observers failing to see all animals
present. Species in this category included bushbuck, reedbuck and
warthog. For other species, variability was due to movement which took
animals out of a study area in which they were counted or conversely,
into one from which they were previously absent. Examples of such species
are elephant and buffalo. There were species, such as kob, which were
conspicuous and relatively sedentary but were difficult to count
accurately. Kob frequently occurred in large herds which were constantly
in movement and difficult to enumerate. The size of the population also
contributed to the variability of counts for some species. Reedbuck, for
example, occurred in such low numbers that differences between counts by
only a few individuals resulted in large percentage change, whereas
similar numerical differences in a large population resulted in only
negligible variability.

Single monthly counts were Judged to be adequate in sampling the
population present in an area. Eltringham (1973) reached a similar

conclusion in his assessment of variability in the counts made by Field 8

67
Laws (1970). As a technique for detecting long-term trends in animal
populations, a monthly ground count is probably adequate provided
precautions are taken in drawing the boundary of study areas to minimize
movement of animals across borders. It is, however, likely that there is
a mutual cancellation of errors over a long period as was found by

Matthews (1960) in an analysis of extended counts of wildfowl on a

reservoir.

3.4.3. Distribution according to habitat
Elephant, hippopotamus, warthog and bushbuck were most numerous in the
areas of short grass with dense thicket close to water. Hippopotamus were
dependent on water to keep cool by day and to support their weight as
they rested. Elephant, buffalo and warthog, too, used water to cool
themselves. They drank regularly but were less dependent on water than
hi ppopotamus.

Elephant are normally extremely mobile and roam over a wide area.
Analysis of elephant distribution in this study showed that they were
confined almost exclusively to Areas 1 and 2, which together were only
about 18 sq km in size. Given the elephant's propensity for movement,
these results appeared rather paradoxical. Observations confirmed that
the same elephant spent some time elsewhere in the park. Yet they were
recorded almost exclusively in Areas 1 and 2 and evidently spent most of
their time there.

The restricted distribution of elephant probably reflected the
species' instinctive search for security. Poaching had been a serious

problem in the park. Despite a breakdown of law and order elsewhere in

68
the reserve, security had been maintained in Area 1 and its vicinity,
which included Area 2, because it was the site of the park's
headquarters. Elephant had evidently retreated to the most secure area
they could find.

Warthog were more abundant in the short grass and thicket areas than
in other habitats. warthog preferred this habitat probably because in the
short grass they could see their surroundings and avoid surprise attack
from predators. Furthermore, the dense thickets provided shelter from
tropical storms, the heat of the sun, as well as cover and concealment
from predators. In Areas 1 and 2, the conditions favored by warthog
existed primarily because of intense hippopotamus grazing pressure.
Harthog occurred in other habitats throughout the park, but in much lower
densities. High densities of warthog seem to be an early indicator of
overgrazing (Petrides, pers. comm.) Warthog owed their abundance in the
short grass and thicket areas to hippopotamus.

Kob and buffalo were most numerous in medium-height and tall
grasslands. The low densities of kob in areas of short grass and thicket
suggest that they particularly avoided this habitat. Factors which were
likely to be important included availability of adequate quality forage,
access to water and avoidance of predation. Proximity to water seemed
relatively important since kob were observed to drink frequently. But
this would be a factor in favor of using the habitat, not avoiding it.
Grass species favored by kob include Heteropoqon contortus, Themed;
triandra and Hyparrhenia filipgndula (Field 1972). Because of heavy
hippopotamus grazing in Areas 1 and 2, these species were absent or

present only in trace amounts (see Chapter 6). Lack of favored food

69

plants was probably an important factor causing kob to avoid the habitat.
Finally, how important could predation have been a factor in denying kob
use of the short grass and thicket habitat? Potential predators included
lion, leopard, hyena and man. With the exception of the nocturnal hyena,
the other predators hunt by sight. Kob, too, rely primarily on sight to
detect predators. In areas of dense bush visibility would be seriously
compromised and the ability to detect a predator in time to avoid being
caught severely impaired. Dense bushes also provide cover and concealment
from which a predator can stalk and launch attacks. To kob, the relative
usefulness of cover to a potential predator is probably the significant
difference between an open and a bushed terrain. Areas of dense bushes
would place kob in a situation of double Jeopardy with their defences
curtailed while the offensive abilities of the predator would be enhanced
by the abundant cover. Whether the absence of suitable foods or the
danger of predation in a high-risk habitat is the more important factor
affecting the spatial distribution of the species is unknown.

Buffalo were fairly evenly distributed throughout the study areas.
The key to their wide distribution appeared to be a feeding strategy that
allowed them to be relatively selective and to maximize their benefits in
a wide range of environment. In the wet season when seasonal inland pools
and wallows were available, buffalo seemed to prefer to graze the new
growth in grasslands distant from permanent water. When the inland
wallows dried up during the dry season and drying upland grass made poor
grazing, they moved closer to permanent sources of water and increased
their intake of browse which at this time of the year contained a higher
percentage of protein than did hay (Field 1972). Even normally coarse

70
grasses such as Cymbopogon afronardus and Imperata cylindrica, which

were avoided when growing rank, were grazed when tender new shoots
sprouted following a burn.

Bushbuck were most numerous in areas of short grass and dense
thicket. Bushbuck affinity for areas of dense bush indicated that shrubs
were an important resource to the species. Possible benefits included
food, shade, concealment and proximity to water. Bushbuck were often
observed to browse the thickets. According to Waser (1975) and Okiria
(1980) the thickets provide a rich variety of grasses, herbs and woody
plants, many confined to this habitat. It is likely, therefore, that
bushbuck depended on the bushes for a diversity of food plants not
otherwise available.

Since bushbuck were not observed to drink, proximity to water seemed
relatively unimportant. Shade, on the other hand, could have been a
factor given the intensity of equatorial solar radiation. But shade could
be obtained in the other habitats also and was unlikely to be an
important factor in bushbuck distribution.

Concealment from predators appeared to be an important potential
benefit to bushbuck in the densely bushed habitat. While predators could
also be concealed in bush, the ability of bushbuck to hide themselves and
the effectiveness of the strategy were demonstrated repeatedly during the
20 months of game counts. When motionless, bushbuck blended in so well
with the surroundings of bushes that they were difficult to detect. On
one occasion a leopard was seen to pass a pair of bushbuck at a distance
of less than 10 m without apparently detecting them. Yet the pair was
only partially concealed by bushes. The bushbuck had detected the

71
predator and frozen in their tracks but remained alert. As the predator
passed by it gave no indication of having detected the bushbuck which
were closer to it than to the census crew of whose presence the cat was
certainly aware. Concealment, too, was therefore probably an important
factor in bushbuck preference for a habitat with dense bushes.

While association with thickets may be of advantage in avoiding
diurnal predators relying on sight, this advantage is lost at night since
association with thickets at night greatly limits the area in which a
nocturnal potential predator needs to hunt. Waser (1975b) found that
bushbuck on Mweya Peninsula showed a pattern of movement away from areas
of dense thicket to areas of open grassland at night. Such subtle shifts
demonstrate that predation is a major factor in distribution at a macro
as well as a micro level.

The distribution of some herbivores was also influenced by seasonal
changes involving fire and rainfall. Fire acted indirectly through its
effects on the vegetation. By removing mature unpalatable vegetation of
low food value and replacing it with nutritious tender growth, fire
affected not only the quantity and quality of forage available, but also
the distribution of animals attracted to the new growth. From January to
March 1982, a large concentration of kob hop-scotched a succession of
burned areas in the Craters highlands as each sprouted with new growth
following a series of fires earlier in the dry season. Less spectacular
concentrations were observed elsewhere in the park. Some isolated burned
areas were grazed so intensively that they had not recovered two growing

seasons after the fire. Similarly intensively-grazed areas in Kenya took

two years to recover (Pratt 1967).

72

Fire removes vegetation cover. While a mosaic of burned and unburned
areas results in habitat diversity and is potentially beneficial to a
spectrum of animals, extensive fires leave large areas without cover. To
small mammals and the young of many large ungulates that depend on cover
to avoid predation, extensive fires may pose a threat to survival.
Wide-ranging fires occurred frequently in such areas as the Craters
highlands. Such fires appeared to affect temporarily the social
organization and distribution of some species, particularly reedbuck.

Areas’ which were ordinarily devoid of permanent fresh water were
inhabited during the wet season by large numbers of animals when
temporary pools were created after rains. When the pools dried up during
the dry season, animals dependent on water left for areas closer to
permanent water. Buffalo, waterbuck, kob and to a lesser extent,
hippopotamus were involved in such seasonal movements.

It is concluded that the distribution of various ungulate species in

different habitats was not random.

3.4.4. Seasonal fluctuations in numbers
In general, large mammals were little affected by changes in seasons.
Elephant, hippopotamus along the shoreline, waterbuck, kob and buffalo
all showed no significant seasonal differences in density. Statistically
significant differences were demonstrated in one or two study areas for
warthog, bushbuck and reedbuck but there was no pattern of seasonal
differences for any species.

In the case of bushbuck, observations during game counts in Area 1

indicated that larger numbers tended to be recorded during rainfalls. As

73
most of rain in the park occurred in convectional rainstorms in the wet
season, many more counts were carried out during or immediately after
rainfall in the wet season than in the dry. The seasonal differences
indicated by the data were probably due more to the behavioral response
of bushbuck to rainfall than to seasonal differences in abundance.

Reedbuck are adept at hiding, especially in tall brown grass. Their
rufuous coat provides a cryptic blend which makes them difficult to spot.
In the park, the grass was tall and brown primarily in the dry season.
During the rains, especially early in the season when the grass was still
short and green, it was harder for reedbuck to hide against a contrasting
landscape. The seasonal differences demonstrated for the species in Areas
4 and 5 probably reflected differences in observer ability to see all the
animals present rather than seasonal differences in abundance. Seasonal
differences in visibility were maintained by frequent fires which also
resulted in tender, nutritious new growth to which reedbuck were
attracted.

Warthog are short and easily hidden by tall and mediumeheight grass.
Their dark skin also made them difficult to see against a burned
landscape. During the wet season when new growth turned the landscape
green, they too, were easier to see. They were also attracted to the new
growth and consequently were in the open and easier to enumerate. The
seasonal differences indicated for the species in Area 4 were again
probably due more to inability of observers to see all the animals,
present during the dry season counts and to improved conditions for

counting during the wet season rather than to seasonal differences in

abundance.

74

Analyses of the sub-seasonal data for Area 1 showed that the numbers
of hippopotamus on land during the day increased as the dry season
progressed. Grass usually ceased to grow, dried up and turned brown by
the end of the first month of the dry season. Heavy grazing, principally
by hippopotamus, quickly removed much of the standing forage creating a
food shortage. The grass was reduced to a uniformly cropped stubble close
to the soil surface. Further and further into the dry season, an
increasing number of hippopotamus probably failed to meet their daily
dietary requirements during the usual nightly feeding. Because of the
shortage of forage an animal had to search over a wider area to collect
sufficient bulk to fill its stomach. By the second and third months of
the dry season apparently an increasing proportion of the hippopotamus
population reach a threshold at which the usual nightly feeding was no
longer adequate. The increased number of hippopotamus on land during the
day is considered to be due to the extension of feeding time in order to
obtain enough stomach fill.

Although the average number of hippopotamus along the shore did not
change significantly between sub-seasons, it was interesting to note that
the averages varied inversely with those of hippopotamus on land, as
would be expected.

Prior to the recent escalation of poaching, many elephant would move
into forested areas during the dry season or even leave the park
altogether so that dry season counts were generally about half of those
recorded in the wet (NUTAE 1963-1971; Eltringham 1973). The influx of
elephant in the third month of the dry season was unexpected and
indicated that poaching and harrassment had probably also denied elephant

75
access to the forests.

I When grazing is poor, browse becomes the elephant's primary source of
sustenance (Field 1971). Area 1 had dense Capparis tomentosa and Azima
tetracantha bushes, both of which were relished by elephant.
The availability of these foods and of water probably provided an
acceptable alternative to forest resources during this period. Elephant
were not resident in study Area 1. This is indicated by their low numbers
early in the dry season and later in the wet. During game counts,
elephant were found almost exclusively in this area and in the adjacent
Area 2. If they were not resident in Areas 1 and 2 and yet were rarely
found in the other study areas as the counts showed, then it was logical
to assume that they frequently used areas other than those in which
counts were made.

Sightings of elephant in the sector between the Kazinga Channel and
the equator_ placed most groups within a triangular area with apices at
Mweya Peninsula, Katunguru and Kabatoro. The greatest concentration of
sightings was along the Kazinga Channel. Areas 1 and 2 comprised the
southern apex of the triangle with borders on the Channel. These areas
with permanent water had higher elephant densities relative to the other
study areas which were distant from the Channel.

In Area 1, the average numbers of warthog remained more or less the
same between sub-seasons (Table 3.11). A one-way ANOVA of the data
indicated, however, that there were marginally significant differences
between sub-season means (Table 3.10). Though statistically significant,
these differences were numerically unimportant. Statistical significance

was possible with such small differences between means only because

76
variability between counts were so small.

Despite large variability in numbers between counts, this study has
demostrated that the abundance of large mammals in QENP were little
affected by seasons. Apparent seasonal differences in the abundance of
smaller species such as warthog, reedbuck and bushbuck were shown to be
the result of improved visibility during the wet season due to the
removal of dense vegetation by fire late in the dry season and inability
of observers to see animals in dense vegetation during much of the dry

season. Because .fire is primarily a seasonal phenomenon, its effects
tended to be seasonal.

3.4.5. Population structure
Group sizes:

Elephant: The increase in average group size observed in this study
indicates that banding-together has occurred in recent years. Laws (1974)
has postulated that group size is a measure of the ecological health of
an elephant population. He suggested that banding is a symptom of stress.
The stress could be the result of nutritional deficiency due to habitat.
degradation (Laws at al. 1970; Malpas 1978) or of the breakdown of
social structure through excessive disturbance (Eltringham 1977). QENP
has been found to be a good habitat for elephant on a year-round basis
(Malpas 1977, 1978). The stress leading to aggregation was, therefore,
likely to be of social rather than physiological origin.

Given the recent history of the park, it is concluded that the
increase in elephant group size reflected the high level of poaching in
the park in recent years. Eltringham (1977) observed similar banding in

77
the neighboring Chambura Game Reserve, a hunting area that was visited
frequently by professional safaris and animal capture units as well as by

gangs of poachers. The average group size in that reserve was 28.8.
Bushbuck: The observation that bushbuck were predominantly solitary

confirms the conclusion reached in other studies (Waser 1975, Allsopp
1978). Although commonly found alone, bushbuck associate with other
conspecifics. The most-frequent association was that of a female and her
calf. Adult animals of the same or opposite sex also frequently occur
together. Waser (1975) found that encounters between individuals were
more frequent than would be expected by chance. Allsopp (1978) observed
that an area was utilized by a group, members of which were mutually
tolerant of each other. Females even engaged in mutual grooming. Similar
observation suggested that when adult males met, they tended to stay
together. Serious aggression was shown only in the presence of females
and probably only if the females were in estrus (Allsopp 1978). These
observations indicate that the bushbuck is a social species whose members
choose to live separately.

The significance of solitary life in bushbuck society may be related
to predation. As a browser in dense bush habitat the bushbuck would
derive little benefit from associating in large herds, and dispersal
probably offers a better insurance against predation.

Buffalo: The average size of buffalo herds observed in this study was
55.3. This was considerably lower than the 101.5 and 105.8 estimated by
Eltringham 8 Woodford (1973) in park-wide aerial surveys in 1968 and
1969, respectively. The mean herd size for the sector was 86.3 in 1968
and 88.0 in 1969. Because of recent poaching in the park, the decline in

78

herd size was unexpected. As with elephant, disturbance by poachers was
thought to result in herd amalgamation for protection. That would result
in fewer but larger herds. mile fewer herds were found than in the past,
they were smaller not larger. Smaller herd size could have resulted from
fragmentation of larger herds, but fragmentation would also have resulted
in a larger number of herds. It was concluded that the decline in the
number and size of herds was probably the result of the high mortality
due to poaching in recent years.

Galton (1871) first proposed the theory that gregarious behavior in
animals is a response to predation. This idea has since received
widespread attention with regard to avian flocking (Wynne-Edwards 1962;
Lock 1968; Lazarus 1972; Diamond 8 Lazarus 1974: Bertram 1980), fish
schooling (Brock 8 Riffenburgh 1960: Cushing 8 Harden-Jones 1968) and
ungulate herding (Estes 1974; Jarman 1974: Leuthold 1977). Hamilton
(1971) argued that by aggregating, animals lower their probability of
being captured during an attack by a predator. His ‘selfish herd' model
was based on the concept of a ‘domain of danger', defined as the area
around each animal from which a concealed predator might spring. The
domain of danger is the area enclosed by a set of points equidistant
between the animal and its surrounding neighbors. As animals form larger
groups, this domain, and the probability of a specific individual being
captured, decreases. Thus, scattered animals would benefit by forming
groups.

The Hamilton model was based on a hypothetical situation in which a
predator attacks from within a group. Pulliam (1973) argued that the
Hamfilton model cannot explain herding behavior because predators usually

79
attack from the periphery of a prey group. Pulliam pointed out that the
benefits accruing to individuals within a herd are unequal because

animals on the periphery are more vulnerable than those in the center,

and suggested that such animals would reduce their risk by abandoning the
group. As each successive layer of animals become peripheral, they in
turn would disperse until the group disintegrated. This argument assumes,
however, that the risk of predation to peripheral animals in a herd is
greater than that for dispersed ones. This assumption would be valid only
if the same individuals are on the periphery of the group during
successive predator attacks (Rubenstein 1978).

Vine (1971) and, Treisman (1975a, b) evaluated the probability of a
predator detecting prey, and demonstrated mathematically that the
probability is lower when prey are aggregated than when scattered.
Treisman (1975a, b) further showed that if crypsis is the only means of
avoiding predation (that is, once prey is detected its chances of
escaping are low) solitary living is favored. If animals are able to
detect and escape from a predator, however, grouping is advantageous to
the [individual regardless of prey density or cover characteristics. The
reason is that the increased collective vigilance by a group of prey
against predators outweighs any tendency of predators to detect prey
groups more easily than solitary animals. If the habitual predator
becomes satiated before consuming the entire group or must pause for
digestion before further attacks, grouping would also reduce an
individual's risk (Robertson 1973; Treisman 1975a; Taylor 1976).

Gregarious behavior may reduce the probability of an individual being
caught in a number of ways. These have been reviewed by Bertram (1978)

80

and may be summarized as (1) increased predator detection due to
collective vigilance against predators, even though individual investment
in vigilant behavior may be lower; (2) communal defence due, in some
species, to the greater ability of a group to defend itself by attacking
the predator, as do buffalo (Schaller 1972: Sinclair 1977) and musk ox
(Qyibgs, noschatus) (Mech 1970); (3) lowered probability of being caught

since, as in the mobbing of predators by birds, the probability of anyone
animal being captured is inversely proportional to group size; and (4)
predator confusion, since a predator usually selects a single individual
to chase, the confusing effects of large numbers of fleeing animals may
distract the predator, thus allowing all prey to escape.

Age structure and sex ratios:

Adult sex ratios for kob, waterbuck and bushbuck all show a paucity
of males in their respective populations. The reported foetal sex ratio
for populations in the park and elsewhere have been shown to be 1:1
(Modha 1972 for kob; Spinage 1968, 1973 for waterbuck; Allsopp 1978 for
bushbuck). The large disparity in foetal and adult sex ratios indicates
that survival and mortality rates are different in the sexes, with higher
survival for females and greater mortality for males.

It has been suggested that the differences in survival rates is
largely due to greater predation on males (Buechner 1961: Spinage 1968;
Modha 1972). The preponderance of bucks as compared to does, which are
killed by predators, may be partly because does are more alert since

solitary animals present mostly less-aware targets, but it may also be

partly attributable to greater boldness on the part of the buck.

81

3.4.6. Population trends on Mweya Peninsula
The spectacular increase in the numbers of buffalo and waterbuck on the
Mweya Peninsula during the period of hippopotamus cropping and their
decline following the recovery of the hippopotamus population indicates
that hippopotamus competed with them. The most likely resource for which
they would compete would be food. Field (1972) showed that buffalo and
waterbuck along the shore grazed the same grasses eaten by hippopotamus.
The dynamics of buffalo and waterbuck populations on Mweya Peninsula
vis-a-vis that of hippopotamus over the past 27 years confirm that did
indeed compete. It seems that when grazing reaches a certain intensity,
buffalo and waterbuck are unable to compete with hippopotamus and are
forced to seek grazing elsewhere.

During the period of hippopotamus cropping, the relative proportions
of a diverse spectrum of large mammal species were more equitable than in
previous years when hippopotamus dominated the large manlnal comnunity
(Eltringham 1974). The diverse community was considered more desirable
than the virtual monoculture of hippopotamus because it made better use
of available resources without deleterious effects on the habitat.

The relentless expansion of the hippopotamus population following
cessation of the cropping program has been the most dramatic feature of
the dynamics of the large mammal community on Mweya Peninsula. The
expansion has been persistent and at the expense of the forage and the
other large mama] species.

Elephant increased slightly between 1956 and 1968 but declined
significantly in 1969. Since then, the trend is one of gradual decline

82
until 1974/76 when they increased slightly. Their density in this study
was the highest since 1970.
The abundance of warthog and bushbuck have changed little over the
years although numbers recorded in recent counts have been low,

indicating that both populations have declined recently.

CHAPTER 4

WOODLAND REGENERATION

4.1. Introduction

For nearly 25 years, there has been considerable debate over the relative
importance of fire and elephant in controlling woody vegetation. Some
authorities (Glover 1963; Lamprey gt a1. 1967; Laws 1969:
Vesey-Fitzgerald 1973; Harrington 8 Ross 1974; Laws gt a1. 1975;
Thomson 1975; Caughley 1976) consider elephant to be the principal
agents, while others (Harthoorn 1966: Lawton 8 Gough 1970) are equally
convinced that fire is the primary controlling factor.

Fire is a common phenomenon in QENP and until recently elephant were
also abundant. Where both occur, however, it is difficult to separate the
effects of one from the other. By opening up thickets and allowing
grassland to penetrate, elephant encourage the spread of fire. Similarly,
trees damaged by elephant are rendered more susceptible to destruction by
fire.

The recent elimflnation of elephant from large areas of the park
created a natural experiment. By comparing the current situation with
that of a period when elephant were abundant, the relative importance of
fire and elephant in controlling the survival and regeneration of woody
vegetation could be determined by inference.

The woodland surveyed in the present study was bounded by Equator
Road, longitude 29°56’E, Katwe Road and River Nyamugasani. It was similar
to and incorporated much of the woodland studied by Field (1971). The

83

84
dominant grasses were tall, tussocky, fire-resistant species including
Cymbogegon, Hyparrhenia, [nemege and Imperata.

The woodland had been virtually free from elephant use for nearly
eight years prior to this study. Fire was seasonal but frequent due to
abundant grass in the region. It occurred on average at least once every
two to three years (Eltringham 1976; this study) and temporarily opened
up the area to grazing. Kob, buffalo and waterbuck fed particularly on
the tender, young and nutritious grass shoots which sprouted following a

burn. Intense fires were reported to occur ocassionally, but none was

recorded during this study.

4.2. Methods

4.2.1. Composition and density
The point-centered quarter (PCQ) method developed by Cottam and Curtis
(1956) was used to determine species composition of the woodland, as well
as density and relative frequency of the principal species. The method
was chosen for three reasons: (1) it requires only simple, readily
available equipment to apply effectively (an important consideration in a
situation where equipment was in short supply); (2) it is easy to perform
and can be applied rapidly and accurately, and (3) it provides reliable
data which are free from subjective biases (Beasom 8 Haucke 1975).

The woodland was sampled along four randomly-selected transects
running approximately east and west. These were chosen from a set of 17
by drawing two-digit numbers from a table of random digits. Similarly,
the first sampling point along a given transect was determined by a
three-digit number representing distance in meters from the beginning of

85
the transect. Subsequent sampling points were 30 m apart.

The area around each sampling point was divided into quarters by
drawing two imaginary lines through the point, one along the transect and
the second at right angles to the first. Within each quarter the nearest
tree to the sampling point was recorded by species. Distance from the
sampling point to the center of the tree base and girth of the tree at
1.5 m above the ground were also recorded.

Cottam, Curtis 8 Hale (1953) have demonstrated that for the PCQ
method, mean area per tree is derived from the average of distances
between trees and sampling points. The average distance is found by
summing individually measured distances and dividing by the number of
trees (or four times the number of points sampled). The average distance
squared is equal to mean area per tree, and density is the reciprocal of
the mean area. In terms of trees per hectare, density equals 10,000 (the
number of square meters in one hectare) divided by the mean area per

tree. These calculations are demonstrated below using data from Table
4.2:

d1+d2+d3+"-+dn
Average distance

 

47x Number of sampling points

13,732.51
= = 7.80 m;
4 x 440

Mean area = (average distance)2 = (7.80)2 = 60.88 m2:
Total density

1/60.88, which in terms of trees ha‘1
10,000/60.88 = 164.26.

Total density is used to calculate the absolute density for each species

86
by multiplying by the relative density of the species concerned.

Cottam 98. e1. (1953) and Cottam 8 Curtis (1956) determined that
at least 30 points in a sample were necessary for an accurate or
reasonable estimate. According to this criterion, the three species of
Aeeeie, were adequately sampled (Table 4.2). A prohibitively number of
sampling points would be required, however, to adequately sample the
rarer species. Conversely, it is evident that density estimates for
Aeeeie gerrardii and A, sieberiana could be obtained with fewer samples.

Importance value index (IVI), a measure of a species' importance
within its community (Curtis 8 McIntosh 1951), was calculated for the
major woodland species. IVI is obtained by adding together relative
frequency (a measure of spatial dispersion), relative density (a measure
of abundance) and relative dominance (a measure of cover). Each sampling
point of the PCQ method is treated as a quadrat. Relative frequency is
calculated as the percentage of all sampling points at which a species
occurred. Relative density is the percentage which a species comprises of
all trees in the sample. Similarly, relative dominance of a species is
the percentage of basal area contributed by the species to the combined
total basal area for all species. The three relative values for each

species are added to give the importance value (Table 4.2).

4.2.2. Structure of the woodland
Preliminary observations indicated that the dominant species exhibited
two distinct tiers. Girth was assumed to be strongly correlated with

height. In a stand of nearly 0.5 ha judged to be representative of the
woodland, the girth of Acacia gerrardii trees at a height of 1.5 m above

87
ground was measured and plotted to show its distribution.

4.2.3. Regeneration potential
During the application of the PCQ method (Section 4.2.1), every tenth
sampling point was used as the center of a circular plot of radius 5 m.
Tree seedlings (<0.5 m growing from seed as opposed to rootstock) and
saplings (<1.5 m other than seedlings) in each plot were tallied by

species. The samples were used to estimate sapling and seedling abundance
in the woodland.

4.2.4. Fire, elephant and the woodland
In December 1981, a 0.3 ha plot was established in the woodland after a
major fire. Within three days of the fire, damage to trees was recorded
by species and height classes (51.0, 1.1-3.0, 3.1-5.0 or >5.0'm). Damage
to each tree was recorded as one of four categories of increasing
severity corresponding to the destruction of <11, 1 to 25$, 26 to 501,
and >50! of the tree, respectively. The plot was monitored for 18 months
to observe the longer term effects of fire. The effects of fire also were
observed briefly at several other sites within the woodland.

Sometimes, too, it was possible to obtain indirect evidence of the
relative importance of fire and elephant by studying existing vegetation.
From an analysis of serial photographic transects, Laws §£.§l- (1975)
showed that the density of trees in Murchison Falls National Park
increased towards the boundaries of the park. This was attributed to
game, particularly elephant, avoiding park borders because of attacks and

harrassments by people. The result with regard to woody vegetation, is

88
less intensive browsing with greater regeneration along the park borders.
Conversely, the more-intensive browsing in the interior results in
greater seedling suppression. Thus, where elephant are the principal
factor in controlling regeneration, there should be a gradient of
decreasing tree density from' the park boundaries corresponding to an
increasing intensity of use in the interior.

The incidence of fire should not be correlated with tree density in
this way. Because grass occurs throughout the woodland, fires are usually
extensive and their effects likely to be general. There would be no
gradient of tree density as postulated with regard to elephant.

To test the hypothesis that elephant and not fires were the principal
factor in controlling woody vegetation, mature Aeeeie gerrardii and A,

sieberiana trees in 1-kmrwide consecutive transects between Equator Road

and Katwe Road were tallied and converted to average densities (/sq km).

4. 3. Results
4.3.1. Composition and density

Qualitatively, species composition of the woodland has remained largely
unchanged since 1968 (Table 4.1). Major shifts occurred, however, in the
relative abundance of species. The most significant is that of Aeeeie

gerrardii which in 1968 constituted 21.3% of the trees in the woodland,

but by 1982 had increased to 82.4%. In comparison, increases by Aeeele

sieberiana and Eb hee511_were small numerically but large in terms of

percentages. Two other species which were as abundant as A. gerrardii in

the earlier survey have declined greatly. Bridelia scleroneuroides and

89

Table 4.1. Species composition of a regenerating woodland in the
Craters region, Queen Elizabeth National Park, Uganda. Nov. 1981

 

 

to Feb. 198g.
1968* 1982

Tree species No. % No. %
Acacia gerrardii 109 (21.3) 1451 (82.4)
A. hockii 0 (0.0) 45 (2.6)
A. sieberiana 26 (5.1) 251 (14.3)
Albizia coriaria 5 (1.0) 5 (0.3)
Bridelia scleroneuroides 107 (20.9) 2 (0.1)
Cordia ovalis 8 (1.6) 0 (0.0)
Euphorbia candelabrum 19 (3.7) 1 (0.1)
Ficus gnaphalocarpa 25 (4.9) 3 (0.2)
Maytenus senegalensis 47 (9.2) 0 (0.0)
Olea africana 103 (20.2) 1 (0.1)
Rhus natalensis 18 (3.5) 0 (0.0)
Tarenna graveolens 43 (8.4) 1 (0.1)
Turraea sp. 1 (0.2) 0 (0.0) '
Total 511 (100.0) 1760 (100.2)

 

*From Field (1971).

Olea africana constituted 20.9 and 20.2%, respectively, of the trees in
the woodland in 1968 but only 0.1% each in 1982 (Table 4.1).
The average tree density of all species taken together in 1981-82 was

164.3/ha (Table 4.2). The most abundant species were Acacia gerrardii,
A» sieberiana and A, hockii; with densities (lha) of 135.4, 23.4 and

4.2, respectively. The density of all others species combined was only
1.2 (Table 4.2).

4.3.2. Structure of the woodland

The distribution of girth in the sample of Acacia gerrardii trees was
bimodal (Fig. 4.1), and demonstrated that two distinct generations were

trees oi 1.5 m above ground.

Fig. 4.1. Girth distribution of Acacia

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m ..
mm... m
m .. m
nMMr
E. m
r
T
m .. n
a w
E .. :0.
7?///
T////// i w
V/M/// r
w/x/é///i w
r////////// ..
T//////////// i m
7////////// .
7//r/é//////////W/ i w
7//////////////i///M///// r
r...
_ . 4 a d a fi 5 u

 

1
mnwmxunmwwuu wwwwww

\AOC curl—Unfit

Girth classes in cm.

90

91

Table 4.2. Species densities and importance values for Acacia
woodland in the Craters region, Queen Elizabeth National Park,
Uganda. Novemper 1981- Februery 198;.

 

 

No. of No. I of I of
sampling of dis- basal
Species points trees tances* area'
Acacia
gerrardii 414 1451 9815.5 6.24
A. sieberiana 129 251 3430.8 4.18
A. hockii 32 45 394.5 0.11
Others 7 13 91.8 0.14
Total 582 1760 13732.5 10.67
*in m
'in m2

Teple 4.2 (cont'd).

Rel. Rel. Rel. Impor- Absolute
fre- den- domi- tance density
Species quency sity nance value (trees/ha)

 

Acacia

gerrardii 71.13 82.44 58.48 212.05 135.41
A. sieberiana 22.16 14.26 39.18 75.60 23.42

A. hockii 5.50 2.56 1.03 9.09 4.20
Others 1.20 0.74 1.31 3.25 1.22
Total 99.99 100.00 100.00 299.99 164.26

 

represented in the population. Visual observations also identified a
linnted cohort of tall, sparsely-scattered, mature trees and an abundant

population of young plants. There were virtually no trees of intermediate

92
height.

The two generations represented discrete periods of tree growth
separated by a period of suppression which ended only recently. The
paucity of intermediaries, representing recruitment between the two
generations, testified to the severity of the suppression. This view
substantiates that of Field (1971) who determined that between 1954 and
1968, the same woodland had lost trees at an annual rate of 14.6%, one of

the highest reported for any natural woodland.

4.3.3. Regeneration potential

. A total of 44 plots with a cumulative area of 3455.75 sq m were sampled.
Tallies of 349 rootstocks and 27 seedlings were obtained from 44 plots
with a cumulative area of 3455.75 sq m. These are equivalent to densities

of 1009.9 and 78.1/ha, respectively (Table 4.3), for a regeneration
potential of 1088 young trees/ha.

Table 4.3. Acacia regeneration potential in woodland in the Craters

region. Queen Elizabeth National Park, Uganda, November 1981 to
Febggery 1982.*

Seedling Rootstock
Total number tallied 27 349
Density (ha'l) 78.1 1009. 9

 

*n = 44, cumulative area sampled = 3455.75 m2.

0f the 349 rootstocks, 302 were of Acacia gerrardii, 46 of A,

sieberiana and one each of Ficus gnaphalocarpa and Albizia coriaria. No
attempt was made to tally seedlings by species.

93

These results show conclusively that the potential for regeneration
was extremely high, due overwhelmingly to abundance of rootstocks. The
contribution of seedlings was insignificant. Saplings of A, gerrardii
from rootstocks were the most abundant and numerically surpassed those of
A, sieberiana by a ratio of 6.6:1.

4.3.4. Fire, elephant and woodlands
Fire moves over the landscape as a narrow strip of flames. Observations
showed that flames never persisted in one spot long enough to ignite live
trees or other woody vegetation. Leaves and tender young shoots, however,

were commonly killed. Fire was, therefore, effective in pruning woody

vegetation but not killing it.

Itee 4.4. Severity of fire damage according to tree height.
Extent of Number (percentages) of trees by height class

 

damage <1.0 1.0-3.0 3.1:5.0 >5.0 (m)
Light 0 (0.0) 1 (1.1) 31 (39.7) 23 (53.5)
Moderate 1 (1.5) 17 (18.5) 23 (29.5) 14 (32.6)
Heavy 27 (39.7) 26 (28.3) 16 (20.5) 5 (11.6)
Severe 40 (58.8) 48 (52.2) 8 (10.2) 1 (2.3)
n=281

 

Observations indicated that the woody vegetation most severely
affected by fire were plants small in size and low in stature. In the
‘ woodland, this was principally immature trees. Due to their low height,
inmature trees were conIiionly completely engulfed in flames. These same
characteristics (small size, low stature and lack of an adequate

insulating bark) which rendered immature trees more vulnerable to fire

Ti
.. 40‘

2
i“ 20-

Fig.4.2. Fire damage to waadland trees

60

 

 

 

O 30.1

 

 

 

 

 

 

 

 

 

 

 

i\ I
50- 7
7
/
7 7
7 7
7 i
7 \
7 0
7 \7
7 \ i7
lO- / § \/ i
7 0 070 i.
0 mék rr\\ f \/\

 

e sass in m). .
[Z light fl moderate i heavy severe

95
than mature trees, also made some woodland species more vulnerable than
others. Among the most susceptible was A£§£11.02£Eii-

An assessment of the damage inflicted on individual trees showed that
fire-damage was not uniform. A strong inverse correlation was found
between tree height and severity of damage (Table 4.4, Fig. 4.2). More
than 50% of trees in the lower height classes suffered serious losses.
Saplings up to 2 m high were killed back to ground level. In tall trees,

damage was consistently greatest among the lowest branches. Tall trees

rarely had more than 10.2% of their tissues destroyed.

Table 4.5. Average density of mature Acacia trees in
consecutive parallel transects between Equator Road
and Katwe Road, Queen Elizabeth National Park, Uganda.

 

November 1981.
Distance from Average density (lsq km)
Equator Road (km) of mature Acacia trees

 

NNb

HHHHHH
ngbwwuoomuomwaH
ooooooospppwewrrw

owoonOHmowmet-OODN

 

Trees per sq km

 

Fig. 4.3 Density of mature Acacia trees
between Equator Road and Katwe Road

 

 

 

 

Distance in km from Equator Road

96

97

The average density of mature Acacia within a kilometer of Equator

 

Road was 45 trees/sq km but dropped to 5 at 4 km from Equator Road and

were essentially totally destroyed beyond the 7-km point (Table 4.5; Fig.
4.3).

4.4. Discussion

4.4.1. Composition and density
Species composition of the Craters woodland has remained virtually
unchanged since the survey by Field (1971) 14 years earlier. Constituent
species evidently are well-adapted to frequent fires and persistent
browsing, the two most significant ecological factors in the habitat.

Changes in the relative abundance of species probably reflect
differences in the rates of regeneration. Species that had increased in
the short-term were probably those which possess rootstocks. In the
presence of heavy browsing, plants of these species are unable to grow
out of the stage at which they are most vulnerable to fire. Though shoots
produced from rootstocks may be cut down by animal browsing and destroyed
by fire, they are apparently able to produce sufficient food to develop
large root systems. When there is a lapse in browsing pressure, the
shoots are able to grow rapidly, using the well-established root system.

Such adaptation would account for the short-term spectacular increase of

Acacia gerrardii.

While the prolific regeneration of A, gerrardii ensures that the
species will continue to be an important member of this woodland

community, its future status will depend on the intensity of competition
with other woodland species.

98
Species that declined in abundance were regenerating slowly or not at
all. Their persistence, however, indicates that the species will probably
survive in reduced numbers as the prolific species continue to increase.

Their decline in importance is illustrated by what has happened with

Bridelia scleroneuoroides and glee africana.

If the present densities of trees were maintained, the stand would
close and develop into forest. This is unlikely, however, because of
frequent fires, inadequate precipitation, animal browsing and competition
for resources. Some individuals will be eliminated and tree density will

decline from the present high level to a more moderate level. A new

equilibrium composition will be established.

4.4.2. Structure of the woodland
The bimodal girth distribution exhibited by Acacia gerrardii is evidence

 

that, for a period of time, the woodland was being destroyed at least as
fast as it was regenerating. Field (1971) showed conclusively that
elephant were responsible for the damage. Using a long-term annual rate
of tree destruction for this woodland which he calculated to be 14.6%,
the period of regeneration suppression was calculated as approximately 40
years. This would mean that the process began before the area was
declared a national park in 1952, possibly while it was depopulated as a
sleeping-sickness zone.

In recent years, a number of hypotheses have been advanced to explain
why elephant destroy trees, and in the process, wreck their own habitat.
Caughley ,(1976) has suggested that elephant and trees may oscillate in

numbers, rather like the lynx and snowshoe hare in the arctic. His

99
evidence was based on the age distribution of trees in the Luangwa
Valley, Zambia, where a large number of baobab trees were around 140
years old, suggesting a maximum survival of saplings produced that long
ago. On the assumption that the seedlings survived because there were few
elephant around to destroy them, a cycle of 280 years (2 x 140) was
postulated, since elephant are presently at a maximum.

On the other hand, it could be argued that as the baobab is a
grassland tree, woodland and forest were at a minimum 140 years ago. This
suggests that elephant were then at a maximum, or rather that they had
been so some time before, since their numbers trail those of trees by
about a quarter of a period. Hence the cycle would occupy 187 years.
Caughley suggests a round figure of 200 years as a compromise.

Wildlife cycles of 4 or 10 years' duration are well-known but they
have in common a simple ecosystem with a low input of energy. Such
oscillations, or stable-limit cycles, are unlikely in the tropics, or
with species that are as long-lived and slow-breeding as elephant and
trees. The characteristics of a species that indulges in booms and
crashes are rapid reproduction and a short generation time, so that when
conditions permit, recovery of numbers is rapid. Elephant have many
admirable qualities but profligate breeding is not one of them. It seems
unlikely that they could ever become locked into a stable-limit cycle.
Indeed, there is no example of an ungulate that cycles in this way.
Another objection to this theory is that elephant and trees do not show a
simple predator-prey relationship. Trees are certainly destroyed by
elephant but elephant do not die in the absence of trees. They can and do

turn to other foods. Trees are important to elephant. Elephant health

100
declines when trees are gone, but they survive nevertheless.

Another hypothesis is that of Phillipson (1975), who has observed
that, at least in Tsavo National Park, Kenya, elephant numbers have often
surged and died down again. He has suggested that numbers follow the
pattern of vegetative production, which in turn fluctuates with rainfall.
Droughts are certainly periodic in Kenya and every 50 years or so there
is a particularly severe one. Phillipson sees this as a mechanism for
regulating elephant numbers and has suggested that the cycle should be
allowed to run its course, and that human interference in the form of
culling to reduce temporary over-abundance is unnecessary.

Phillipson's (1975) estimate of 43-50 years as the interval between
severe droughts is far too short a time for an elephant population to
complete a cycle. Phillipson supposes that the severe mortality caused by
the drought would be followed by a period of 35 years of respite, during
which the population would build up to a critical level again, leaving 8
to 15 years of danger from elephant damage. For a species which at best
can produce one calf every 4 years, 35 years is a very short time in
which to recover. Neither Caughley (1976) nor Phillipson (1975) produces
any evidence that elephant undergo a cycle.

The evidence for a climatic cycle upon which Phillipson's hypothesis
is based, is slight and tenuous. There are also theoretical objections to
the concept of an inherent cyclical rhythm in elephant numbers. MacArthur
8 Wilson (1967) propose two types of evolutionary strategy in animals
which they term r-selection and K-selection. An r-selected population
fluctuates widely in size and has the capacity to multiply very rapidly,
while the essence of K-selection is that the population is stable (Pianka

101
1970). The elephant is an extreme K-strategist. It is highly unlikely
that cyclical changes in reproductive rates, as assumed by both
Phillipson and Caughley, would evolve in such an animal.

Petrides (1974) and Eltringham (1982) have suggested that despite
lack of solid evidence that elephant migrate over long distances in
modern times, they must have done so before the advent of modern man.
According to this hypothesis elephant wandered freely over a wide domain
which was probably hundreds of kilometers across. Within this range there
would be areas offering better forage and other resources than elsewhere.
Eltringham (1982) refers to these as ‘hot spots', to which elephant would
be attracted. Within these hot spots elephant behaved much as they do
today, systematically wrecking the habitat. Trees were debarked or pushed
over, bushes broken up and grasses uprooted. But the destruction would go
only so far. Before the habitat was completely destroyed, the elephant
would move on to look for another such hot spot.

The reason for assuming this follows from studies on optimal foraging
in birds and other animals. When a rich food source is found, the animal
does not remain with it until all the food is consumed but moves away to
find another similar food source well Ibefore the original food is
depleted. This is due mainly to the law of diminishing returns. The less
food there is the more difficult it is to find it and the longer it takes
to complete a meal. The time would be better spent in seeking out another
food source where the original eating rate can be restored. Eltrigham has
suggested that this was the case with elephant. But even if they did
wreck the area completely, they would then leave and not return for many

years, by which time the vegetation would have recovered and could

102
withstand another session of abuse.

Seen in this light, elephant behavior is not unadaptive. It is only
when elephant are prevented from moving to new pastures that irreversible
damage to vegetation occurs. Eltringham suggested that not only is
elephant behavior not unadaptive under the assumed circumstances, it is
positively adaptive. For in wrecking trees the elephant convert primary
forest to secondary forest, which is a superior habitat for them.

The significance of this hypothesis is that, if it is correct, it
would be futile for park authorities to wait for nature to take its
course and for elephant in a national park to come into balance with
their resources. Under present-day conditions where national parks are
''islands" of wild habitats, the more likely outcome would be for the
elephant to continue to wreck the area until the whole ecosystem came
crashing down, leading to the reduction or extinction not only of
elephant but of many other large species as well.

Caution should be exercised in interpreting a phenomenon such as the
bimodal structure demonstrated for the Craters woodland. Clearly it
provides evidence of wholesale destruction of a cohort of tree sprouts
and/or seedlings sometime in the past. It does not, however, provide an
answer as to why the failure occurred. Because of recent experiences it
is tempting to conjecture that the elephant were responsible. Even if
they were, it would not be proof that population cycles exist in
elephant. All that can be said is that prior to the current regeneration,
tree replacement was prevented for a period estimated to have lasted
nearly 40 years. The repression was caused by elephant (Field 1971), a
view that is further supported by aerial and ground counts since 1963

103
(Eltringham 1973: Table 3.1), which show that elephant numbers in the
entire park were high. This does not, however, imply that there is

necessarily an elephant population cycle and that habitat restoration is

likely under modern conditions.

4.4.4. Recruitment potential
The study disclosed that the Craters woodland is regenerating abundantly
where earlier a heavy loss of trees due to elephant had been reported
(Field 1971). At densities of 1088 saplings/ha, regeneration potential
was exceedingly high and almost entirely from rootstock. A great number
of seeds were produced but seedlings contributed little towards stand
regeneration. The lack of seedlings could have been for a variety of
reasons including seed infestation, poor conditions for germination,
herbivory and fire but these factors were not investigated.

The current profuse regeneration indicates that the trend towards
grassland is not irreversible and that woodland, at least Aeeeie
gerrardii woodland, can be restored by reducing elephant densities (in
this case, by poaching). This has important management implications, not
only for QENP, but also for many African national parks where too many
elephant have led to drastic changes in landscape and habitat. It shows
explicitly that to halt further destruction and to restore woodlands and
forests, it is necessary to reduce the number of elephant.

The objective of a national park is to maintain a natural landscape.
Over-population, no matter what the cause, usually results in overgrazing
and degradation of the habitat. Is overgrazing natural? It is imperative,

therefore, that it should be recognized as an unnatural situation when it

104
occurs and corrective measures taken to deal with it.

The restoration of woodlands and forests should be an integral part
of elephant conservation. Trees are essential to elephant (Laws ee e1.
1975; Malpas 1978) as well as to other wildlife species and perform many
vital ecological functions. Their deep roots are able to tap water
sources far below the ground so that they are able to provide succulent
browse long after the grass has withered and dried. Nutrients are quickly
leached from the surface of many African soils and the deep-rooted
vegetation is important to retrieve and recycle them. There is more to
conserving trees, therefore, than a pretty landscape and their
destruction by elephant should be cause for concern as much for the

welfare of elephant themselves as for the rest of the flora and fauna.

4.4.5. Fire, elephant and the woodland
There is substantial evidence that fire can damage or destroy trees
(Glover 1963; Lamprey ee,e1, 1967; Lawton 8 Gough 1970; Vesey-Fitzgerald
1973; Harrington 8 Ross 1974: Thomson 1975). This study confirmed that
this is the case and showed that immature trees were the most severely
affected. Observations indicated that this was due, at least in part, to
their small size, low stature and lack of thick protective bark.e
Over-all, fire had the effect of pruning but not killing all woody
vegetation.

In spite of fires or because of it, strong regeneration of vegetation
has continued. Frequent fires prevented accumulation of large quantities
of combustible material which result in intense and devastating fires.

Such fires would burn most saplings back to ground level and destroy most

105
of the lower branches of taller trees. Fires of this kind were not
recorded during this study. Their absence was attributable to frequent
regular grass fires which are generally cooler and less destructive. The
conclusion from these data and observations is that although fire could
ocassionally have caused serious damage, particularly to immature trees,
it could not have been the sole cause of the decline of the woodland or
its failure to regenerate in recent times.

More direct evidence that elephant were more important in controlling
woody vegetation than was fire is provided by the experimental enclosures
set up by Spence 8 Angus (1971) in Murchison Falls National Park. They
excluded fire and elephant in a variety of combinations. They found that,
in bushlands, fire did not prevent regeneration of woody vegetation
whereas browsing did. In woodlands, on the other hand, either fire or
browsing prevented tree regeneration.

Harrington 8 Ross (1974) carried out a similar series of experiments
in Kidepo Valley National Park, Uganda, and found that either factor
could prevent tree regeneration. They considered browsing to be the more
important factor. Kidepo Valley differs from both Murchison Falls and
Queen Elizabeth National Parks. It is much drier and hot fires are more
frequent. The growing season is also much shorter. Rootstocks, therefore,
may have less food reserves to sustain repeated attempts at regeneration.
Precipitation is higher and the growing season longer in the other Uganda
national parks. These factors apparently provide additional resilience to
withstand fire.

One can conclude from these published data that in most cases tree

destruction is due to elephant. Although fire may exacerbate the effects

106
of browsing, it is rarely the sole cause of woodland decline. Evidence to
the contrary, such as produced by Lawton 8 Gough (1970) in Zambia, has
come from restricted areas and their conclusions cannot be extrapolated
to cover all eventualities. In their case, the study area was not even
typical of the Luangwa Valley (Astle 1971).

In fairness to elephant, it should be stated that these animals were
not the only browsers in the experimental plots. Some of the damage to
trees and woody vegetation could have been caused by other species.
Elephant may also be blamed for damage for which they may be innocent.
Trees die from a variety of other causes, including old age. In the
Akagera National Park, Rwanda, Spinage 8 Guinness (1971) described tree
destruction caused by rainstorm and lightning which might have been
attributed to elephant were the species not absent from the park. Western
and Van Praet (1973) also revealed the danger of jumping to conclusions
by showing that trees supposedly killed by elephant in the Amboseli
National Park, Kenya, actually died from an increase in soil salinity.

The present study has demonstrated that active and profuse
regeneration has occurred and is continuing in the Craters woodland
despite persistent, and occasionally intense, fires. It showed that the
current situation was the reverse of that documented by Field (1971) in
1968/69 when elephant were destroying trees and preventing regeneration.
The timing of the regeneration, which began after a massive decline in
elephant numbers caused by poaching, indicated that elephant were much
more important than fire in suppressing regeneration of woody vegetation
there. The study has also demonstrated that there were sufficient
rootstocks present to enable the re-establishment of the woodland. If

107
fire could also be controlled, re-establishment probably would be even
more rapid and more diverse.

The problem with elephant in even the largest of present-day national
parks is that they cannot, through self-regulation, adjust their
population size rapidly enough to remain within the carrying capacity of
the habitat. The rationale of cropping is to help them do so. Whether the
resulting equilibrium with the vegetation is natural or not, it is an
essential management measure that must be maintained if elephant are to
be kept within the confines of a national park and if a totally unnatural
landscape is to be avoided. By allowing elephant numbers to increase
without control within the park, the elephant is condemned to years of
stressful life with an inadequate diet, a disruption of its social life
and often serious damage to other large mammals. in the community
(Petrides 8 Swank 1974). Cropping has been shown to be efficient and to

cause little suffering to elephant populations when carried out by
professional biologists (Laws €2.11- 1975).

CHAPTER 5

THICKET ABUNDANCE AND EXPANSION

5.1. Introduction

Thickets are a prominent component of the vegetation in Queen Elizabeth
National Park. Thickets were once much more extensive, particularly along
the low-lying lake shores. There were extensively cut between 1912 and
1914, however, to remove the habitat of the tsetse fly (Glossina
palgalis) following an outbreak of sleeping sickness (Spinage 1968).
Today, isolated clumps occur in the grassland, each bush consisting of a
virtually impenetrable tangle of thorny scramblers, particularly Caggaris
tomentosa and Agige tetracantha.

The concensus among park authorities (6. Okot, personal
communication) is that, since about 1973, thickets throughout the park
have increased. The increase has coincided with a decline in large
herbivores and has fueled speculation that it was caused by a reduction
in browsing. This view was enlarged by the Acting Chief Warden in the
April 1975 monthly report to the Director of the National Parks. He
attributed the increase in thickets to three factors: (1) overgrazing and
lack of grass fires, (2) reduction in the number of elephant and (3)
higher rainfall, resulting in a more vigorous thicket growth.

0 The probable causes of thicket proliferation and expansion as well as
thicket ecology in general were investigated in the course of this study.
The Mweya Peninsula and Kikorongo census area were chosen for intensive

study. The two areas differed considerably in the incidence of fire and

‘108

109

intensities of browsing and grazing. The peninsula was heavily grazed,
the grass was short and thin, and there had been no fire in the area
since 1963. Thickets, too, were heavily browsed by elephant and buffalo.
Kikorongo, on the other hand, was an area of abundant mediumvheight to
tall grasses where elephant were virtually absent and fires were common.

Comparison of thicket characteristics between the Upper Peninsula (at
a higher elevation and generally drier) and Lower Peninsula (generally
wetter and closer to the water table) (see Section 2.2 under Area 1) also

were made to evaluate the influence of moisture.

5.2. Methods
5.2.1. Clump size, density and cover

A modified version of the point-centered quarter (PCQ) method of Cottam 8
Curtis (1956) was used to determine clump density. The method was applied
in a manner similar to that described for trees in Section 4.2.1. The
distance from the sampling point to the margin of the bush was measured
at ground level along a line from the sampling point to the perceived
center of the bush. The radius of the bush (r) was obtained indirectly by
measurement of the clump circumference (c) at ground level, (r=c/2u).
Bush-clump height was measured with a calibrated pole. Assumptions were
made that the bushes were circular and that the centers of the clumps
were properly located. It is believed, however, that the calculated

values for r were approximations of the true values.

5.2.2. Composition of thicket clumps

Plant species present in each bush clump were recorded. The presence or

110

absence of termite mounds was also tallied.

5.2.3. Origin and development of clumps
Termite mounds, Eughorbia candelabrum trees and several species of
shrubs were evaluated as possible foci of clump origins. The appraisals
included determining the frequency of each component in clump samples,
their resilience to browsing, trampling, fire and moisture deficiency, as
well as their ability to regenerate quickly when damaged. Some
observations were made in experimental enclosures where fire was

controlled and browsing, grazing and trampling by large herbivores
eliminated.

5.2.4. Thicket expansion

Of 20 thicket clumps on Mweya Peninsula, ten were on the Upper and ten on
the Lower Peninsula. The height and circumference of each bush was
measured every three months between February 1982 and May 1983. Changes
in diameter were averaged to give seasonal and annual rates of expansion.

The apparent effects of fire, grazing and browsing were noted.

5.2.5. Thickets as a wildlife resource

Animal use of thickets was noted.

5.3. Results
5.3.1. Clump size, density and cover

The average number of thicket clumps on Mweya Peninsula and in
Kikorongo was determined to be 53.6 and 4.4/ha, respectively (Table 5.1).

111

Thus, thickets were over 12 times more abundant on the Peninsula than in

Kikorongo.

Table 5.1. Comparison of thicket parameters in Mweya Peninsula and
Kikorongo. Queen Elizabeth Natiogel Perk. Uganda, Mex 1983.

 

Mweya Kikorongo
Sample size (n) 428 296
Average height (m) 2.63 3.84
Average basal area (sq m) 51.47 102.75
Mean diameter (m) 8.10 11.44
Mean area/thicket (sq m) 186.59 2291.24
Density (lha) 53.59 4.36
Cover (%) 27.58 4.45

 

Thickets in Kikorongo were individually larger than those on the
Peninsula. The mean diameter and height for Kikorongo were 11.4 and 3.8
m, respectively, while corresponding figures for the Peninsula were 8.1
and 2.6 m. At 102.8 sq m, the average basal area for thickets in
Kikorongo was almost exactly twice as large as the 51.1 sq m for those on
the Peninsula. Thicket cover over the study areas was calculated to be

27.6% on the Peninsula and 4.5% in Kikorongo (Table 5.1).

5.3.2. Species composition
A thicket clump commonly contained 20 to 40 plant species (Table 5.2).
Bushes on Mweya Peninsula were dominated by the thorny scrambler Cagearis
tomentosa and the even more-thorny Agige tetracantha. The two species

were commonly co-dominant in the same clump. Even in clumps where one was

clearly more abundant, the other was almost always present. Generally,

112
Caggaris dominated on higher ground while Azima dominated on lower

ground closer to the water table and on saline seepages.

Table 5.2. Common thicket species on Mweya Peninsula, Queen Elizabeth
Natione] Park. Uganda, May 1983.

Trees
Euphorbia candelabrum Trem. ex Kotschy Euphorbiaceae
Shrubs
Azima tetracantha Lam. Salvadoraceae
Capparis tomentosa Lam. Capparidaceae
Erythrococca bongensis Pax. Euphorbiaceae
Hoslundia opposita Vahl Labiateae
Indigofera arrecta A. Rich. Leguminosae

Jasmi ni um emi ni i Gilg.

subfam. Papilionoideae

Oleaceae

Maerua triphylla A. Rich. Capparidaceae

Pavetta albertina S. Moore Rubiaceae

Tarenna graveolens (S. Moore) Brem. Rubiaceae
Climbers

Ceropegia sp. Asclepidiaceae

Cissus adenocaulis Steud. ex A. Rich. Vitaceae

Cissus quadrangularis L. Vitaceae

Cyphostemma adenocaule (Steud. ex Vitaceae

A. Rich.) Descoings
Glycine javanica Leguminosae

subfam. Papilionoideae

Momordica sp. Curcubitaceae
Sarcostemma viminale R. Br. Asclepidiaceae
Herbs
Achyranthes aspera L. Amaranthaceae
Ageratum conyzoides L. Compositae
Anthericum subpetiolatum Bak. Liliaceae
Barleria sp. Acanthaceae
Commelina africana L. Commelinaceae
Commelina benghalensis L. Commelinaceae
Cyanotis foecunda DC. ex Hassk. Commelinaceae
Dyschoriste radicans (Hochst.) Nees Acanthaceae
Justicia flava (Forsk.) Vahl Acanthaceae
Ocimum americanum L. Labiateae
Ocimum suave Willd. Labiateae
Sansevieria dawei Stapf. Agavaceae
Sida alba L. Malvaceae
Sida ovata Forsk. Malvaceae

Solanum incanum L.

Solanaceae

113
Table 5.; (contd.).

Grasses
Brachiaria eminii (Mez.) Robyns Paniceae
Cenchrus ciliaris L. Paniceae
Digitaria velutina (Forsk.) Beauv. Paniceae
Panicum deustum Thunb. Paniceae
Panicum maximum Jacq. Paniceae
Setaria kagerensis Mez. Paniceae
Setaria sphacelata Stapf. 8 C.E. Hubb. Paniceae
Sporobolus pyramidalis P. Beauv. Sporoboleee

 

 

The floristic composition of thickets north of the Kazinga Channel
exhibited a remarkable degree of constancy with Euehorbia candelabrum,

Q. tomentosa, A, tetracantha, Tarenna ggeveolens. Hoslundie opposite and

 

Eegeege albertina present over wide areas. Differences in dominants,
however, were evident. For instance in Mweya Peninsula, Securigega yiegee
was only an occasional member of the thicket community but some 16 km to
the north, in Kikorongo, it was frequently the dominant species.

5.3.3. Origin and development

Termite mounds were present in 46.4 and 51.9% of the clumps sampled in
Mweya Peninsula and Kikorongo, respectively. If one accepts the theory of
clump origin around termite mounds (Osmaston 1961; Jackson 8 Gartlan
1965), the origin of half of the clumps sampled would still be
unaccounted for. Furthermore, if termite mounds were the foci of clump
development, a sizable proportion of small clumps in the early stages of
development could be expected to be on mounds. In the survey, though,
only 4.3% (2 out of 46) of such clumps were associated with mounds.

Since nearly half the number of clumps sampled had mounds in them, it

114
is reasonable to assume that mounds commonly developed in a clump once it
was established. In a sub-sample of 200 of the largest clumps, 178 or 89%
were found to have mounds. This demonstration that mounds were most
common in the largest clumps indicates that thicket clumps are frequently
the foci of termite activity and mound formation rather than the
reverse.

The frequency of Euggorbia candelabrum within clumps was 89.5 and
96.3% in Mweya and Kikorongo, respectively. Despite the high incidence of
the species inside thicket clumps, it was virtually absent from the
heavily grazed and trampled grassland between clumps. Furthermore, inside
an experimental exclosure devoid of large mammal activity there were 1600
seedlings/ha. Thus the species evidently can, in fact, grow in grasslands
but, seedlings did not occur there because they were unable to survive
trampling and/or browsing by large herbivores. Examination of a large
sample of saplings showed that a young Eughorbia was fleshy, weak and
would snap easily should a large mammal even brush against it. Seedlings
in the open could not, it seemed, provide nuclei for clump formation.

Observations inside the exclosure showed that several other thicket
species, including Securigega yiegee, Meegee triggylla and Hoslundia
opposita, were capable of establishing themselves in the grassland.
Small individuals of these species were also frequent in the
heavily-grazed short grass area north of the Kazinga Channel. During the
dry seasons, these and most other thicket species commonly wilted and
lost their leaves completely depending on the severity of moisture
deficiency. '

Of the common thicket species, only Caggaris figmentosa and Azima

 

115
tetracantha were capable of vigorous and sustained growth. Their
seedlings were firmly rooted, making them resistant to uprooting as a
result of browsing or trampling. During the re-construction of vehicle
tracks in the park in 1981, numerous thicket clumps were removed. Taking
advantage of the opportunity, it was observed that both 9. tomentosa and
EL tetracantha developed extensive root systems and sprouted readily
after damage. Small bushes quickly developed from root sprouts along
tracks and in ditches around enclosures. The most remarkable trait in the
two species was, however, their ability to remain green and lush and to

continue to grow actively even during a severe dry season when all others

species essentially ceased to grow.

5.3.4. Expansion

Rainfall data (Table 5.3; Fig. 5.1) show that the period 1965-1978 was
wetter than that of 1954-1963. The meteorological data, therefore, tend
to support the claim that the increase in thicket abundance and expansion
could be due, at least in part, to increased precipitation.

Observations in Kikorongo indicated that frequent fire was a major
factor limiting both the abundance and expansion of thickets. During
grass fires, young shrub seedlings often were killed and small clumps
burned back to ground level.

Some established thicket plants such as Caggaris, were extremely
resilient to fire. Even when severely damaged, they often put forth new
shoots within two to three weeks after a fire. Rarely did fire totally
demolish large bush clumps. Usually only the shoots along the margin were

burned, arresting growth and horizontal expansion. Fire played an

116
important role in limiting thicket abundance and expansion. The absence

of grass fires would eliminate a potent controlling factor.

Table 5.3. Mean monthly rainfall (mm) on the Mweya Peninsula.
Queen Elizabeth Netiggel Perk. Ugegge.

 

1954-1963 1965-1978
January 26.2 26.7
February 24.6 49.0
March 64.5 74.4
April 75.7 123.2
May 62.2 87.1
June 29.5 42.9
July 25.4 60.6
August 71.1 73.3
September 84.6 76.4
October 82.8 78.8
November 71.1 109.2
December 51.8 39.6
Totals 669.5 841.2

 

On the Mweya Peninsula, there have been no grass fires since 1963 due
to overgrazing by hippopotamus. The area contained one of the highest
densities of thickets in the park, with 27.6% of the peninsula estimated
to be under thickets. The absence of fire seemed certain to have been a
contributing factor to the proliferation of thickets in the area.

On the Mweya Peninsula, the average diameter for 20 thicket clumps
increased from 6.2 m in February 1982 to 7.7 m in May 1983, an expansion
rate of 9.7 cm per month (Table 5.4). Seasonal monthly growth rates

differed significantly. The wet season rate was 12.0 cm/mo. in contrast

to the dry season rate of 6.2 cm/mo.

Rainfall in mm

Fig. 5.1. Mean monthly rainfall (mm)

30 Mweya Peninsula. QENP. Uganda.
1

 

120 -

1101

100-

701

60-

50..

 

 

 

 

iriiiiiiirii
JFMAMJJASOND

M o n t h '
0 1954-1963 + 1965-1978

117

118

Table 5.4. Changes in diameter (m) of selected thickets on Mweya
Peninsule. Uganda. between Februery 198g_end Mey 1983.

 

Thi cket Feb May Aug Nov Feb May
I 198g 198g 198g 1982 1983 1983
1 6.27 6.25 6.13 6.55 7.08 7.17
2 3.16 3.39 3.29 3.32 6.54 6.90
3 3.15 4.12 3.54 3.69 6.75 6.89
4 5.80 6.37 5.82 6.15 4.06 4.29
5 5.05 5.93 5.56 5.73 8.08 8.14
6 5.28 5.78 6.27 6.75 6.64 6.97
7 6.89 7.32 7.56 7.86 8.83 9.21
8 8.06 8.69 8.67 9.08 8.96 8.34
9 5.73 6.74 7.60 7.99 6.89 5.79

10 5.04 5.69 5.87 8.33 6.91 7.21
11 7.50 7.47 7.51 7.66 7.60 7.70
12 7.51 7.62 7.63 8.08 8.26 8.43
13 7.66 8.01 8.04 8.10 8.47 8.66
14 5.15 5.49 5.88 6.08 6.19 6.64
15 5.61 5.90 6.31 6.37 6.44 7.64
16 6.27 6.89 6.75 7.71 7.86 7.92
17 10.91 11.60 11.82 12.66 12.99 12.79
18 4.68 6.11 5.06 5.53 6.03 6.83
19 6.84 7.24 7.33 7.41 7.73 7.75
20 7.35 7.85 7.77 8.08 8.22 8.40
X .6.20 6.72 6.72 7.16 7.53 7.65
SD 1.77 1.72 1.85 1.98 1.72 1.66
Growth rate: wet and dry season = 12.0 and 6.2 cm/mo., respectively;

an annuel rate of 9.7 cm/mo.

 

Despite the steady increase in average diameter (Fig. 5.2), some
individual clumps experienced temporary reductions in diameter between
measurements (Table 5.4). These losses followed substantial browsing,
primarily by elephant, but frequently also by buffalo. Evidently,
browsing by ungulate could also exercise considerable control on thicket

expansion.

The average diameter of thickets on the Lower Peninsula was much

Fig. 5.2. Growth of bush-clumps

Mweya Peninsula. QENP, Uganda.

 

820 s
800 -
780 -i
760 -
740 d
720 -
700 ~
680 s
660 -

Mean diameter (cm).

620 -
600 s
580 -+
560 -

 

 

540 l l r l l

-—-i

. 1982/83 .
D Total + Upper Peninsula 9 Lower Peninsula

119

 

120

larger than of those on the Upper (Fig. 5.2). This was thought to be due
to a higher rate of expansion on account of a presumed higher level of
soil moisture on the Lower Peninsula. The data, however, contradict this
hypothesis. While the wet season growth rates for the Upper and Lower
Peninsula thickets were remarkably similar (12.4 and 12.5 cm per month,
respectively), those for dry season were quite different. The rate for
the Upper Peninsula was 8.9 cm/mo., while that for the Lower Peninsula
was 3.4 cm/mo.

In QENP, the single factor most limiting to plant growth is
availability of water. Thickets on the Lower Peninsula are closer to the
shoreline and watertable and are better situated to exploit soil moisture
on a year-round basis. They were thus expected to have a higher rate of
expansion than those on the Upper Peninsula.

The paradox appears to be the result of seasonally intense browsing,
particularly by elephant but frequently also by buffalo. During the
second dry season of 1982, elephant activity centered on the Lower
Peninsula where browse was more abundant and shade trees provided relief
from the sun. Water for drinking and bathing was also more readily
available than on the Upper Peninsula. On the Upper Peninsula, there were
no trees to provide shade and elephant activity was limited to visits
during the night, occasionally lingering into the morning. The Upper
Peninsula also contained the Headquarters of the park and camps for the
staff. As a consequence, there was always some human activity on the
Upper Peninsula even at night. Out of a total of 99 hours (h) that
elephant groups were on the Peninsula, 62 (62.6%) were spent on the Lower
Peninsula. If this is the general pattern during the dry season, the low

121

rate of thicket expansion recorded on the Lower Peninsula during the dry

season may have been, at least in part, due to heavy browsing by

ungulates.

5.3.5. Thickets as a wildlife resource

A wide range of animals commonly used thickets for food, shelter, shade

and concealment (Table 5.5). Thickets were more important to some species

than to others. For example, the bushbuck and vervet monkey depended

heavily on them for several aspects of their well-being including food,

Table 5.5. Qualitative assessment of large mammal use
of the thicket resource in Queen Elizabeth National
Park, Uganda. November 1981-June 1983.

Cetegory of use
Species

 

Food Shelter Shede Conceelment

Elephant Common - Common -
Hippo Occasional - - -
Warthog Rare Common? Common Common
Reedbuck - - - -
Waterbuck Common - Common Occasional?
Kob Occasional - Frequent -
Bushbuck Common Common Common Common
Buffalo Frequent - Common -
Baboon Common Common? Common Common
Vervet

monkey Common Common Common Common
Lion - - Common Common
Monitor

lizard - Frequent Common Common

 

shelter and concealment. The reedbuck and kob, on the other hand, hardly

used them at all. Warthog and lion used them extensively for shade and

122
concealment, but not for food. In this brief survey, the diversity of the
fauna that benefited from thickets indicated that they were an important
wildlife resource. In East Africa, this role has generally been

overlooked.

5.4. Discussion

5.4.1. Clump size, density and cover
Thicket clumps in Kikorongo were larger than those on Mweya Peninsula.
This was attributed to two causes. First, precipitation was higher in
Kikorongo than on the Peninsula and Kikorongo was free of elephant

activity. Also fires prevented the establishment of thicket seedlings in .

Kikorongo.

5.4.2. Species composition
Although species composition in thickets north of the Kazinga Channel
showed remarkable constancy, this is apparently not characteristic of the
entire park. Lock (1977) reported that on the Ishasha River Flats in the
extreme south, many of the associates are replaced by species that are
absent in the north such as Caggaris fascicularis.

The distribution of dominants perhaps emphasizes the enormous

variability in precipitation, drainage and topography as well as the
influences of fire and herbivory within the park.

5.4.3. Origin and development of clumps
Several theories have been advanced to explain the origin of thicket
clumps. Lebrun (1947) suggested that they develop from the shrub layer

123

below isolated trees. Osmaston (1961) suggested that clumps develop
around termite mounds and that the shade of a §g2flge§1e,eeggeleeggg
tree, once established, is beneficial to the process on Lolui Island in
Lake Victoria, Jackson and Gartlan (1965) found that clumps developed
from seeds in monkey droppings around rocks and termite mounds frequently
used by the primates. The present study does not support these theories
of clump origin around termite mounds or Euggorbia trees.

For a plant to flourish in QENP, it has to be adapted to a large
moisture deficit, frequent fires and persistent browsing. The last factor
could result in uprooting or breakage. In general, bushes and thickets
are better adapted to withstand browsing than are trees. When a tree is
broken or uprooted, the physical damage is serious. Thickets are more
difficult to destroy. Parts may be severed but regeneration is rapid. No
lasting damage is done. When a tree is killed another will not
necessarily grow in its place. It seems logical, therefore, that species
acting as nuclei for clump formation are shrubs rather than trees.

An alternative theory is that a single bush in the grassland acts as
a nucleus from which a clump develops (Lock 1977). Given the persistent
browsing and frequent fires, the bush has to be able to grow continually
and vigorously in order to survive. Of the common thicket species, only
Caegeris and Agige, had the characteristics necessary to be the nuclei
around which thicket clumps could develop.

Termite mounds occur frequently in thicket clumps probably because
termites are attracted to the sites by dead stems and accumulation of
litter. Similarly, clumps do not develop around Euggorbia trees, but

rather the trees are able to establish themselves and to develop within

124
thickets because of the protection provided by thicket clumps. Without
the clumps the trees would not survive the trampling and browsing of
large herbivores. Rather than being the origin of thicket clumps, termite
mounds and Eugeorbia trees occur in thickets because the clumps provide
a supportive environment.

Once established, a thicket evidently gained species and expanded
outward from its origin. Usually the first to establish themselves under
the young thicket were some of the common grasses and herbs. They
included Brachiaria emigii, Cenchrus ellieris, Sporobolus eyramidalis,
Achyranthes eeeege, Phyllanthus spp., Qeimgg americanum, Q. egege, §1ge
glee, S, QMQEQ. and Seleggg iggleeg and the sedge Kyllinga. These
associates, too, contribute to the changing environment. Gradually woody
plants appeared, common among them were Erythrococee bongensis.
Euggorbia candelabrum, Meeege,triggylla, fleerEe albertina and Ieeegge
geaveolens. In time a tree or two, usually a Euggorbia, may emerge
through the bush. Observations indicated that emergent trees were at
greater risk of destruction by elephant than non-emergents, yet some
survived to maturity.

Caggaris or Agige often remained the major clump constituents.
Some very large thickets (30 to 40m in diameter) recorded in this study
were thought to have formed by adjacent thickets coalescing. Eventually
animals (especially elephant, buffalo and hippopotamus) could break into
the thicket and may split it up or even destroy it.

Usually fires scorch only the periphery of the clump and recovery is
rapid. Even when all aerial parts were destroyed, bases sprouted

vigorously and, in time, probably restored the clump. Under certain

125
circumstances, fire was observed to destroy thickets. Occasionally, the

mass of dead wood from a fallen Euphorbia would produce a very hot flame
which killed much or all of the thicket.

The continued active growth of many thicket constituents during the
dry season may be, at least in part, due to the soil moisture available
to them. In West Africa, Okali e3 e1. (1973) have shown that Caggaris
and other constituents of similar thicket clumps, are able to conserve
water because they have deep roots and enjoy restricted water-loss
through the stomata.

Certain grass species, including Cenchrus ciliaris, Eragrostis
tenuifolia and Sgorobolus gyramidalis, were common in both grasslands
and thickets. Individuals in the grassland were stunted and feeble, while
those within thickets grew taller, stouter, more robust and remained
green longer into the dry season, after those in the grassland had wilted
and dried. Persistent defoliation of plants in the grassland by
hippopotamus was thought to be the reason for the disparity in stature.
It would not, however, explain why plants inside thickets remained green
longer. This is evidence that clumps are able to retain and conserve
moisture much better than is the open grassland. If this is true, the
greater amount of moisture available to plants growing inside thickets

may also contribute to the noted disparity in stature.

5.4.4. Thicket expansion
The impact of rainfall, fire and browsing pressure were discussed with
respect to differences in thicket size and abundance in Mweya Peninsula

and Kikorongo. These factors also influence the overall trend towards

126
thicket proliferation and expansion in the park as a whole.

Although higher rainfall in recent years may have resulted in more
vigorous thicket growth, it is not considered the likely cause of the
trend towards increased thicket density. Rainfall data indicates that the
increase in rainfall levels began in 1961. By all accounts, however, the
increase in thicket size and density for the park as a whole began about
1973 (E.L. Edroma, personal communication). It is unlikely that it took
thickets more than ten years to respond to increased rainfall.

Observations in Kikorongo showed that fire is a major limiting factor
in certain areas of the park. In heavily grazed areas, the absence of
fire due to lack of fuel, has been a factor in thicket expansion. There
is no indication, however, that in recent years the incidence of fires
has declined in sectors of the park other than the Mweya Peninsula. For
this reason the absence of fire, is also eliminated as the major cause of
thicket proliferation.

The only other plausible explanation for the rapid trend towards
thicket abundance and expansion is a reduction in the overall level of
browsing pressure. This is supported by the fact that the decline in
populations of large herbivores in the park was followed shortly by
reports of thicket expansion. It is concluded, therefore, that thicket

proliferation was a result of the massive decline in large herbivores.

5.4.5. Thickets as a wildlife resource
Thickets clumps were found to be important for the food, shelter, shade
and concealment which they provide to a diverse spectrum of animals. The

thicket flora is comprised of a rich variety of grasses, herbs and woody

127

plants not otherwise available.

Few ungulates are as dependent on thickets as the bushbuck and bush
duiker. These animals live in and feed on thickets and use them
effectively to hide or escape from predators (Waser 1975; Allsopp 1978;
Okiria 1980). Their entire ecology is dependent on the thicket and its
resources.

A number of other large herbivores used thickets primarily as a
source of food. They included elephant, buffalo and waterbuck. Field
(1971) has shown that the young and tender shoots of Caggaris are rich

in protein. During the dry season when the coarse, dry hay provided poor
grazing, such browse constituted a valuable source of protein. At such
times even species such as kob which are primarily grazers, were observed
occasionally to browse thickets. Thickets provide grazing and browse to
livestock which are kept by inhabitants of the many villages in the park,
despite an ordinance forbidding the raising of domestic animals. Goats

have been particularly destructive, overbrowsing thickets around a number
of the villages (Musoke 1980).

Two common species of grassland primates, the olive baboon and the
vervet monkey, obtained a considerable portion of their food from thicket
plants, feeding on fruits, berries, seeds and foliage. Baboons on the
southern banks of the Kazinga Channel even feed on shoots of Euphorbia
candelabrum (Lock 1972). Few animals fed on Euggorbia because they
contain a noxious latex, but somehow these baboons have developed a
mechanism to detoxify the latex.

Numerous birds and insects also derived sustenance from thickets. All

the major woody species in the thickets have edible fruits. These were

128
often eaten by birds which were probably instrumental in dispersing their
seeds. From June to August Caggaris was frequently completely defoliated
by caterpillars of Belenois eurote Fab. (Kenyi 1980).

Lion, waterbuck, warthog, the monitor lizard and guineafowl (Ngmige
meleageis), regularly sought comfort away from the hot equatorial sun in
the shade of thickets.

In conclusion, this study has rejected the hypotheses of the possible
origins of thicket clumps on termite mounds and around Euphorbia trees
in favor of clump development around the nucleus of a single plant in the
grassland. That plant had to be a vigorous species to persist in spite of
browsing, trampling and burning. The trend in the park towards increased
thicket size and abundance was evidenced to be the result of decreased
browsing pressure. The increase in rainfall levels in recent years may
have resulted in a more vigorous thicket growth but was not directly the
cause of the proliferation. Fire was judged to be a major factor in
checking thicket abundance and expansion in areas where they were
frequent. It is postulated that fire is particularly effective in
preventing seedling establishment, resulting in lower thicket density.
Differences in thicket size between areas were attributed to varying
levels of precipitation as well as to the activities of large herbivores
which may break up or destroy thickets.

Although thickets and the grassland have been treated as separate
vegetational units, they clearly form parts of a single ecosystem. They
are dynamically related since new thickets apparently become established
on grassland, while old ones are eventually destroyed and replaced by

grassland. Large herbivores not only grazed and browsed the thickets but

129
often took shelter in them, with resulting concentration of fecal
material. As a consequence of these relations, the physical and chemical
properties of the soil are probably changed during the development of a
thicket. These changes, at least temporarily, influence the grassland
which subsequently replaces the thicket.

CHAPTER 6

STANDING-CROP BIOMASS, SPECIES COMPOSITION AND GRAZING INTENSITY IN THREE
GRASSLAND TYPES

6.1. Introduction

In Queen Elizabeth National Park, grassland types are associated with
rainfall (Field 1970) and intensity of grazing (Brooks 1957; Heady 1966:
Lock 1972). In the present study, three physiognomic types namely the
short, medium-height and tall grasslands were studied to determine
species composition, vegetative cover, standing-crop biomass and
intensity of grazing. The vegetation inside a 15-year old exclosure on
the Mweya Peninsula was compared with that of an adjacent grazed plot.
Changes in the vegetation over the past 25 years induced by grazing were
reviewed briefly. Present conditions were compared with those reported in
earlier studies.

In the late 19503, excessive grazing by hippopotamus (Bere 1959;
Petrides 8 Swank 1965) resulted in widespread deterioration of habitat.
As discussed in Chapter 3, the present hippopotamus population on Mweya
Peninsula is substantially higher than that which earlier resulted in
habitat deterioration.

Some observers (Yoaciel 1981) have postulated that grazing might not
have been responsible for the habitat degradation in the late 19503. They
point to rainfall data which shows that the period 1954 to 1963 was drier
than the period since 1963 as evidence that habitat deterioration was
probably a climatic episode. This study disagrees, however, and shows

‘130

131
that while widespread deterioration similar to that reported for the

19505 has not recurred, hippopotamus are presently overgrazing the range

and if allowed to continue could result in range degradation.

6.2. Methods

6.2.1. Species composition and vegetative cover
Species composition was determined by recording plants present within
1-sq-m quadrats taken at 10 m intervals along randomly selected
transects.

In the short grassland on Mweya Peninsula, data on vegetative cover
and species composition were collected simultaneously. The 1-sq-m quadrat
used to determine species composition of the grassland had steel wires
stretched between opposite sides at 20-cm intervals. The wires formed a
grid with 16 intersections each of which was used as a sampling point to
determine vegetative cover. The plant species directly below each
sampling point was recorded. Vegetation was not present at every sampling
point, of course, in which case litter or the substratum was recorded.

Cover for a given species was calculated as the percentage of sampling

points at which the species occurred.

6.2.2. Standing-crop biomass and intensity of grazing
The amount of forage available to herbivores in each grassland type was
determined monthly by taking samples from a 30-m—square plot established
at a site judged to be representative of the grassland. The plot was open
to grazing. At about the same time every month, eight 1-sq-m samples were

selected by random coordinates and harvested at ground level from within

132
the plot. Live shoots were separated from dead material and further
sorted by species. The entire crop was dried and weighed.

In the short and medium-height grasslands, an ungrazed plot protected
by barbed-wire fencing 1.5 m high was established adjacent to the grazed
plot. Limited resources prohibited construction of a similar exclosure in
the tall grassland. Differences between the quantities harvested in the
grazed and ungrazed plots at each site were assumed to measure the
amounts of forage consumed by large herbivores.

Rainfall data collected at the plot site in the short grassland
enabled an evaluation of the relationship between monthly precipitation

and live-shoot biomass.

6.2.3. Vegetation within the 15-year old exclosure on the Mweya

Peninsula
The experimental exclosure on the Mweya Peninsula was constructed in 1968
and consists of a circular trench around a barbed-wire fencing 80 m in
diameter (Strugnell 8 Pigott 1978). A survey of the vegetation within the
exclosure was made to determine the nature of the vegetation the area
would carry in the absence of grazing. Species composition, standing-crop
biomass and vegetative cover within the exclosure were compared with that
of an adjacent grazed plot outside.

The exclosure was also used to appraise the long-term effects of fire
on the grassland. Since 1968, one half of the exclosure has been burned
annually (except in 1979 and 1980), while the other half has been kept
unburned. By excluding animals, the exclosure provided a controlled

situation in which fire was the only experimental factor. The species

133

composition of the two halves were compared to determine the long-term

effects of fire on the flora.

6.3. Results

6.3.1. Species composition and vegetative cover
The short grassland was found to be the richest in grass species with a
total of 23. In comparison, the number recorded in the medium-height and
tall grasslands were 14 and 12, respectively.

The main constituents of the short grassland were Chrysochloa
orientalis and Sporobolus gyramidalis. Several other species including
Tragus berterioniegge, Sporobolus stapfianus. Egegrostis ciliepensis,

E. tenuifolia. Cenchrus cilieris. Chloris gayana and Dactyloctenium
aegygtium were common (Tables 6.1 and 6.2). Cynodon dactylon was

abundant locally along the lake shares. The tufted, pyrophilous grasses
[negege triandra and Heteropoqon contortus which were common throughout

the greater part of the park were conspicuously absent. Conmon herbs
included Alternenthera pungens, Euphorbia spp., Tribulus terrestris and

Teghrosia spp.

The short grassland was dominated by stoloniferous and short,
mat-forming perennials and prostrate herbs and shrubs. The more common
stoloniferous species included 9. orientalis, Q. ciliaris, Q. gexege and
Q. dactylon. Sward height was 5 to 40 cm, with a mean of 7.2.
Mat-forming grasses were represented by Microchloa kunthii, e.
stagfianus and e. festivus.

In contrast, the other two grasslands were dominated by taller,

tufted grasses such as Hyparrhenia spp., I. triandra, A. contortus and

134

Table 6.1. The major components of three grassland types in
Queen Elizabeth National Park, Uganda. Dry weights are means
for 14 months from May 198g to June 1983.

Plent teee q/sqfigj %

Short grassland
Chrysochloa orientalis
Sporobolus pyramidalis
Sporobolus stapfianus
Harpachne schimperi
Eragrostis spp.
Microchloa kunthii
Aristida adoensis
Tragus berteronianus
Other grasses
Forbs

HH
mmauwmucoooo
M NO.)
-~stssws
wOGmVI-‘btot-‘H

Mediumeheight grassland
Heteropogon contortus 5
Themeda triandra 4
Sporobolus pyramidalis 2
Bothriochloa spp. 2
Hyparrhenia filipendula 2
Other grasses
Forbs

HHHNN
HHN‘OV-DO’

Tall grassland

Hyparrhenia filipendula 219.5
Themeda triandra 151.5
Cymbopogon afronardus 46.8
Imperata cylindrica 22.5
Other grasses 8.7
Forbs 10.1

Hub
NCOCDNOG

 

Cymbogggon afronardus. Despite similarities in flora and structure, the
medium-height grassland was of lower stature than the tall. Sward height
in the medium-height grassland was 45 to 105 cm with a mean of 89,
compared with a range of 80 to 180 cm and a mean of 153 in the tall.

The three grasslands were zoned according to distance from permanent

135
water in which hippopotamus, the dominant herbivores, lived. The

intensity of grazing by hippopotamus decreased with distance from the
water. It seemed, therefore, that the life-forms of the grasses

reflected, at least in part, the intensity of grazing. The stoloniferous

and mat-forming grasses being the most tolerant and the tall, tufted
species being the least tolerant of heavy grazing.

Several species of annuals were common in the short grassland but not

in the other two. They included Tragus berterionienus. Aristida adoensis
and Chloris pycnothrix.

Table 6.2. Foliar cover in the short grassland on Mweya
Peninsula, Queen Elizabeth National Park, Uganda, at the end
of the first growing season, May198g.

SpeCies* Percent cover

Chrysochloa orientalis 11.1
Sporobolus pyramidalis
Alternanthera pungens
Euphorbia spp.

Sporobolus homblei

Chloris gayana

Herbs (mostly legumes)
Tribulus terrestris

Sporobolus stapfianus/festivus
Eragrostis spp.

Tragus berterionianus

Cynodon dactylon

Litter

Bare ground

5°
N

00*‘*‘*‘T‘9°9°5°9°5’9°
aoubaatvamao

co
9°
.5

*Only species accounting for at least 1% cover have been
tabulated.
The sward in the short grassland was not only low but also thin. Q.
orientalis, §. gyramidalis and the prostrate herb Alternanthera gungens
provided the most foliar canopy, contributing 11.1, 9.2 and 8.9% of total

136
cover (Table 6.2).

Sward height and vegetative cover reached their maxima at the end of
the growing season. At this time the sward in the short grassland varied
in height from 5 and 40 cm with a mean of 7.2. Even at this time,
however, as much as 38.4% of the ground surface was still exposed (Table
6.2). This proportion increased as the dry season advanced. Due to
intense grazing, the sward was reduced to a more or less uniform height
<5 cm early in the dry season. Therefore, during much of the dry season
the proportion of exposed surface was higher than the 38.4% indicated by
this study.

While no quantitative data were obtained for the taller grasslands,
observations indicated that their vegetation was sufficiently dense that
foliar cover was nearly 100%. The only notable exceptions were localities
of kob communal territorial grounds which were kept short by heavy
grazing and trampling.

Due to the abundance of grass in the tall and medium-height
grasslands, they were always at great risk of fire during the dry season
and burned frequently. Much of the short grassland, on the other hand,
was never at risk of fire for lack of fuel to support one.

Although mat-forming species were not ordinarily significant
components of the medium-height and tall grasslands, it was observed that
following fires Sgerobolus stagfianus and Microchloa Aggegii, two
mat-forming species, briefly dominated large parts of these grasslands.
They accomplished this by growing quickly and flowering gregariously
early in the growing season before the taller, tufted species could

compete and overshadow them. In full bloom, the inflorescence of §.

137
staefianus gave the sprouting grassland a lavender hue. Gregarious

flowering similar to the post-burning phenomenon was not observed in the

short grassland.

6.3.2. Standing-crop biomass and intensity of grazing
Standing-crop biomass was highest in the tall and lowest in the short
grasslands (Table 6.3; Fig. 6.1). During the period of study, the mean
monthly standing-crop biomass was 39.8 g/sq m in the short, 172.3 g/sq m
in the mediumrheight and 459.2 g/sq m in the tall grassland.

In the short grassland, liveshoot biomass in the ungrazed plot was
consistently higher than in the grazed, indicating that the intensity of
grazing was continuously high. This was true also for the mediumtheight
grassland except for three months in which biomass in the grazed plot
marginally exceeded that in the ungrazed (Table 6.3). The larger quantity
of live shoots in the ungrazed plots reflected the intensity of grazing
pressure on the open range. ‘ 0

When grazing intensities in the short and mediumrheight grasslands
were compared, not only was the quantity grazed in any given month in
most cases higher in the short grassland but the percentage of
standing-crop biomass consumed was consistently much larger, showing that
the short grassland was grazed more intensively. The proportion of
standing-crop biomass grazed in the short grassland varied from a low of
about 60% in the growing season to a high of over 90% in the dry (Table
6.4; Fig. 6.2), with an average of 75.6%. Variations within season were

small. The high proportion of biomass cropped by ungulates explains why

138

Table 6.3. Mean monthly biomass (g/sq m) harvested in grazed and ungrazed

plots in three types of grassland, Queen Elizabeth National Park, Uganda,
1982-1983.

 

Exclosure (E) Grazed (6)
Live Live
shoots Litter Total shoots Litter Total
A. Short grassland
May 172.0 34.3 206.3 69.8 16.3 86.1
June 322.1 91.5 413.5 68.6 57.9 126.6
July 132.8 227.7 360.5 12.3 43.7 56.0
August 123.7 212.2 335.8 38.5 76.7 115.1
September 94.9 85.9 180.8 27.2 17.6 44.9
October 276.5 180.1 456.7 66.9 53.9 120.8
November 188.3 114.5 302.8 78.1 72.7 150.8
December 325.7 199.5 525.2 43.9 52.9 96.7
January 170.8 309.3 480.1 20.9 23.4 44.3
February 51.0 241.2 292.2 3.5 24.1 27.7
March 15.4 200.9 216.3 2.6 33.8 36.4
April 78.2 184.8 263.0 27.0 21.9 48.8
May 187.1 82.8 269.9 53.7 12.0 65.7
June 136.8 36.8 173.6 43.8 13.5 57.3
Total 2275.3 2201.4 4476.7 556.7 520.3 1077.0
Average 162.5 157.2 319.8 39.8 37.2 76.9
a; Medium-height grassland
V
June 561.9 312.4 874.3 183.5 126.8 310.3
July 387.1 306.1 693.1 360.9 326.4 687.3
August 443.6 537.6 981.1 219.3 432.0 651.4
September 228.1 691.3 919.4 231.8 593.9 825.8
October 197.1 314.2 511.3 73.8 314.2 387.9
November 233.9 342.6 576.5 220.1 296.6 516.7
December 277.9 232.9 510.9 164.4 304.9 469.3
January 198.1 699.9 898.0 208.1 352.8 560.8
February 136.2 460.1 596.3 142.7 450.5 593.1
March 123.7 484.2 607.8 69.3 216.6 286.0
April 145.3 426.3 571.6 105.3 120.1 225.3
May 136.8 218.5 355.3 117.8 137.5 255.3
June 223.5 257.8 481.4 142.5 114.2 256.7
Total 3293.0 5283.9 8576.9 2239.5 3786.4 6025.9
Average 253.3 406.5 659.8 172.3 291.3 463.5

139
Ieple 6.3. (Contd.).

Exclosure (g1 Greged (6)

Live Live

shoots Litter Total shoots Litter Total
C. Tall grassland

May
June 869.1 548.9 1418.0
July 650.4 454.2 1104.6
August 513.3 922.8 1436.0
September 472.6 408.3 880.8
October 653.8 446.1 1100.0
November 335.2 796.7 1131.9
December 298.6 466.1 764.7
January 309.3 405.5 714.8
February 274.9 571.4 846.3
March 341.6 371.4 713.0
April 347.9 690.3 1038.2
May 499.2 344.1 843.3
June 403.1 249.2 652.3
Total 5969.0 6674.9 12643.9
Average 459.2 513.5 972.6

 

the sward was thin and why nearly 40% of the surface was exposed. Such a
high level of off—take could not be sustained indefinitely. The question
arises whether there is a need for management intervention to prevent the

herbivores, predominantly hippopotamus, from altering the habitat from a

natural condition.

In the medium-height grassland, the percentages of standing-crop
biomass grazed showed large and erratic fluctuations between zero and
about 65% (Table 6.4; Fig. 6.2), with an average of 26.5%. In this and
the tall grassland, forage production was clearly in excess of what the
herbivores were able to consume. They could, therefore, support a higher

biomass of ungulates than at present.

Fig. 6.1. Live-shoot biomass

in three grassland types. QENP. Uganda.
900

 

800 ~

700s

600-

500+

Mean dry weight (ea/sq m)

ZOO-i

100‘

I

 

 

 

 

l T l i i ' " i l i

A S O N O J F M A M J
. 1982/83

0 Short grass + Medium-height grass 0 Tall grass

T40

Fig. 6.2. Intensity of grazing

Queen Efizabeth Natl Park, Uganda.

 

 

 

 

100

a q q _ i. A . 4 1
W w m .noo w w w N mw

VONULO 0005.319 DOLUIUC:U.anuD ”.0 uCOULOQ

FMAMJ

JASONOJ

J

M

Medium-lit grassland

1983/83)

{1 Short grassland

141

142

Table 6.4. Estimates (g/sq m) of vegetation removed by ungulates
from the short and medium—height grasslands in Queen Elizabeth
National Park, Uganda, 1982-1983. (Figures in parenthesis are
percentages of standinq'crop biomess cropped).

Grasslend type

 

Short Medium-height

May 102.2 (59.4) - -

June 253.5 (78.7) 378.3 (67.3)
July 120.5 (90.7) 26.2 (6.8)
August 85.2 (68.9) 224.2 (50.5)
September 67.7 (71.3) -3.7 (-1.6)
October 209.7 (75.8) 123.3 (62.6)
November 110.1 (58.5) 13.8 (5.9)
December 281.8 (86.5) 113.5 (40.8)
January 149.9 (87.8) -9.9 (-5.0)
February 47.5 (93.1) -6.5 (-4.8)
March 12.8 (83.1) 54.3 (43.9)
April 51.3 (65.6) 40.0 (27.5)
May 133.4 (71.3) 19.0 (13.9)
June 93.0 (68.0) 81.0 (36.2)
Average 122.8 (75.6) 81.0 (26.5)

 

Live-shoot standing-crop biomass fluctuated seasonally, indicating
that productivity was a function of rainfall. This relationship was
demonstrated (Table 6.5; Fig. 6.3) for the site in the short grassland
for which rainfall data was also collected. There was a close

relationship between amount of precipitation and vegetative

productivity.

6.3.3. Vegetation within the 15-year old exclosure
At the time the exclosure was constructed, it contained a grassland with
a mosaic of short (5 cm), closely-cropped and mediumeheight (40 cm),
lightly-grazed patches (Strugnell 8 Pigott 1978). Sgorobolus gyramidalis
and inggie gexege were the dominant grasses and provided 56 and 30% of

Percent of total

Fig. 6.3. Live-shoot production

in relation to rainfall. QENP. Uganda.

 

 

 

 

5e
4- '
._ 7
2~ l
1‘ h
0 i i i T r i l i i i i T T
M J J A S 0 N D J F M A M J
. 1982/8;
u Shoot production + Rainfall. 1 mo. lag

143

144
foliar cover, respectively. Associated with them were significant
quantities of Bothriochloa inscglpte, Cenchrus ciliegis, Heteropogon

contortus, Aristida adoensis and Sporobolus festivge,

Table 6.5. Monthly rainfall (mm) at Mweya Peninsula, QENP, Uganda, and
mean dry weight (g/sqym) of live shoots hervested in ungreged plot.

Rainfall Live shoots

mm % g %
May 139.6 15.6 172.0 7.6
June 60.0 6.7 322.1 14.2
July 22.1 2.5 132.8 5.8
August 34.5 3.9 123.7 5.4
September 74.2 8.3 94.9 4.2
October 54.6 6.1 276.5 12.2
November 93.6 10.5 188.3 8.3
December 90.2 10.1 325.7 14.3
January 4.7 0.5 170.8 7.5
February 23.5 2.6 51.0 2.2
March 26.8 3.0 15.4 0.7
April 82.9 9.3 78.2 3.4
May 126.1 14.1 187.1 8.2
June 62.4 7.0 136.8 6.0

 

After 15 years, Q. eyramidalis and Q, geyege_have been replaced by
Q. ciliaris as the dominant grass. Digitaria scalarum was co-dominant
with Q. ciliaris in the burned sector (Table 6.6). Between them the two
species accounted for nearly 65% of live-shoot biomass in the sector
which also contained moderate quantities of Hyparrhenie filipendule, Q.
some and Malta "am-

In the unburned sector, Q. ciliaris accounted for nearly 63% of the
biomass. Q, scalarum, E, geglggg_and Brachiaria decumbens were present

but in relatively low quantities. Commelina africege and Q. benghalensis

 

145
occurred in moderate quantities in both sectors. The average live-shoot
biomass in the burned sector was 370.3 g/sq m compared with 414.0 g/sq m
in the unburned. Litter in the unburned sector was considerably higher
than that in the burned (Table 6.6). This is not surprising since fire

consumes litter and thus prevents it from accumulating.

Table 6.6. Major constituents of the grass sward in the 15-year old

exclosure on Mweya Peninsula, Queen Elizabeth National Park, Uganda,
Hey 198g-June 1983.

 

 

 

Treatment
Burned Unburned

.g/sq m % q/sq m ’%
Cenchrus ciliaris 147.6 39.9 260.0 62.8
Digitaria scalarum 91.0 24.6 26.2 6.3
Commelina spp. 43.4 11.7 37.9 9.1
Hyparrhenia filipendula 34.6 9.3 0.4 0.1
Chloris gayana 20.2 5.5 4.3 1.0
Panicum maximum 7.4 2.0 23.0 5.5
Aristida adoensis 4.2 1.1 - -
Brachiaria decumbens 2.9 0.8 7.1 1.7
Other grasses 7.8 2.1 2.1 0.5
Forbs 11.2 3.0 53.3 12.9
Live-shoot total 370.3 100.0 414.0 100.0
_Litter g§;.3 - 435.3 -

 

 

These results indicate that in the absence of grazing, the

species-rich short grassland is replaced by one that is poorer in species
but higher in stature and vegetative cover and dominated by Q. ciliaris.

In the absence of both grazing and fire, the grassland would contain
numerous species of woody shrubs. Most of the shubs are susceptible to
fire, however, and would be eliminated if fires were frequent. This is

evident from their absence in the burned sector of the exclosure.

146

Frequent fires encourage the growth of a number of pyrophilous
species including A, filipendula which occurred only in the burned
sector. Such species would replace those which would be eliminated by
fire. [Aegege triandra, a common species in the park, which has been
reported to be encouraged by frequent fires (Lock 8 Milburn 1971) did not
occur in any appreciable quantity in either the burned or unburned sector
of the exclosure.

In the heavily grazed grassland outside the exclosure, plants of Q.
ciliaris were common as short, stunted and isolated individuals. Q.
ciliaris was heavily grazed (Brooks 1957; Heady 1966; Field 1972) and
was adversely affected by intense grazing. This view was supported by the
observation that plants growing within thicket bushes and protected from
frequent grazing grew to be taller and more robust, much like those
protected within the exclosure.

Forbs and shrubs were more abundant in the unburned sector than the
burned. In the burned sector, only 5 species, mostly herbs such as
Achyranthes eegege and Boerhaavia gifgee, were recorded in the grass
layer. In comparison, 18 were found in the unburned sector. Most of the
latter were woody shrubs, such as §1ge gyeee, Indigofera eeeeEe,
Agei‘egim conyzoides. Barleria sp., (3% sp. and MM _in_di_c_um. This
miscellaneous group made up only 3% of the biomass in the former as .
opposed to nearly 13% in the latter (Table 6.6).

Eughorbia candelabrum, Aeeele sp. and Caggaris tomentosa were the
major woody species in the exclosure. Chi-squared tests show significant

differences between the burned and unburned sectors with respect to the

numbers of trees and bushes (Table 6.7).

147

Table 6.7. Comparison of the numbers of trees of three woody species in
burned and unburned sectors of the 15-year old exclosure on Mweya
Peninsula, Queen Elizabeth Nationel Perk. Uqanfl. My 1983.

 

 

Treetment
Species earned Unburned df x2
Euphorbia candelabrum O 14 1 14.0**
Acacia sp. 13 76 1 44.6**
Capparis tomentosa 1 9 1 6.4*

 

itSignificant at 95% level; eesignificant at 99% level.

In addition to differences in numbers, there were also significant
differences in 'size and height. The average height of Aeeeie sp. trees
in the burned sector was 82.1 cm (n=13, SD 145.1) as compared with 190.7
cm (n=76, SD 176.4) in the unburned. Similarly, Q. tomentosa bushes in
the unburned sector were larger and consisted of many thicket species
including Q. candelabrum, Aeime tetracantha, Hoslundia oggosita, EeyetEe
albertina and Erythrococcus bongensis. The bush in the burned sector, on
the other hand, was small and consisted entirely of Caggaris.

These results indicate that fire has adverse impact on woody

vegetation, and that frequent fires result in lower species diversity.

6.4. Discussion

6.4.1. Species composition and vegetative cover

In the short grassland, the effect of heavy grazing and trampling has

148

been to eliminate tall, tufted species and to allow a high number of
predominantly stoloniferous species and annuals to be established. Some
authors (e.g. Edroma 1981) have implied that the mechanism of change
involve the selective grazing of tufted species in preference to the
non-tufted. The possession of a wide mouth, however, would seem to make
the hippopotamus a non-selective grazer. The elimination of tufted
species would not be due, therefore, to their being preferentially and
selectively grazed. The author postulates that tufted species are
eliminated because their life-form puts them at greater risk of being
grazed by hippopotamus than the species with prostrate or stoloniferous
life-forms that replace them.

Hippopotamus graze by grasping the vegetation between its horny lips
and cropping each mouthful with a jerk of the head. This pulling action
often uproots whole plants, some of which may be found along the trail of
a grazing hippopotamus. Because the hippopotamus is short-legged, it is
able to graze extremely close to the ground surface. Plants that grow
upright are at greater risk of being grazed or uprooted than those that
are prostrate or adhere closely to the ground surface. The mat—forming,
prostrate or stoloniferous life-form makes a plant less readily.
accessible to hippopotamus and as a result is less frequently grazed.
Over a period of time, the less-readily accessible plants come to
dominate. The creeping and stoloniferous life-form of the many species in
the short grassland is the result of this process.

The abundance of Sggrobolus gyramidalis, a tufted tussock grass, in
the short grassland demonstrated, however, that not all upright-growing

species were vulnerable to hippopotamus grazing. To determine what makes

149
this species resistant to grazing, the author attempted to duplicate

hippopotamus grazing by plucking leaves of Q. gyramidalis by holding the
leaves between his thumb and index finger and pulling with a jerk. Not

only were the leaves difficult to sever because of the fibers, but they
were slippery and easily passed between his gloved fingers. Although the
thick lips of the hippopotamus would seem to permit a firm grasp of grass
leaves, the leaves of Q. pyramidalis may slip also from between the lips
of a grazing hippopotamus, thus making the species less vulnerable.

It was noted earlier that the hippopotamus mode of grazing often
uproots plants. The advantage bestowed by tough, slippery leaves would be
lost if the plants were easily uprooted. While the root system of Q.
pyremidalis was not thoroughly investigated, attempts by the author to

 

uproot a large sample of these plants showed that they were firmly and
securely anchored in the soil. The combination of tough, slippery leaves
and firmly-rooted plants evidently has made Q. pyramidelis an exception

and allowed it to flourish where other upright-growing grasses have been
eliminated. Since other tufted grasses are eliminated, the implication is
that they have shallow root-systems. This remains to be confirmed.

One of the reasons the short grassland was richer in grass species
than the medium—height and tall grasslands was the presence of several
annuals in the award. Many of the annuals are probably normal components
of a seral stage and would be replaced in the sequence of plant
succession. Under conditions of intensive grazing and trampling, however,
the process of sucession is arrested and the seasonal annuals persist in
the sward indefinitely, serving as indicators of heavy grazing..

The large proportion of exposed surface also provided strong evidence

150
that the short grassland was overgrazed. This is supported by the fact
that many of the constituent species including Q. gyramidalis, A,
adoensis and I, terrestris are classed plants of disturbed or overgrazed
areas (Bews 1917; Rattray 1954; Chipindall 1955; Heady 1966).
Structurally and floristically, the mediumrheight and tall grasslands

were much more similar to each other than either was to the

heavily-grazed short grassland.

6.4.2. Standing-crop biomass and intensity of grazing
The close relationship of live-shoot biomass to rainfall in the semi-arid
environment of the park emphasizes the importance of moisture to plant
growth. It also indicates that moisture may be a factor limiting growth
during periods of severe water deficiency such as during parts of the dry
season. _

Isohytes over the park (see Fig. 2.3) show that the center of the
rift valley floor is the driest area in the park. This, in general, was
the area of short grassland. The tall grassland, on the other hand,
occurred to the west towards the foothills of the Rwenzori range where
precipitation is higher. The medium-height grassland was intermediate in
location. The gradient in precipitation indicates that even in the
absence of grazing, there would probably be differences in standing-crop
biomass on account of rainfall.

Live-shoot biomass on a range is not, however, solely a function of
moisture levels. As demonstrated by Lock (1967, 1972) and confirmed by
this study, grazing also plays an important role. Where grazing is light
as in the medium-height and tall grasslands, it has little effect on net

151

production and the natural composition and appearance of the range. Where
grazing is intensive as in the short grassland where between 60 and 90%
of standing-crop biomass is consumed, grazing may result in changes in
botanical composition, production and structure of the range. Excessive
defoliation has been shown to exhaust food reserves in plants causing
them to lose vigor, deteriorate and eventually die prematurely (Milthorpe
8 Davidson 1965; Davidson 8 Milthorpe 1966a, 1966b). Plants that are
grazed frequently, whether selectively or otherwise, are eliminated and
those that are inaccessible, avoided or grazed less frequently come to
dominate.

The changes that result from such a high level of consumption are
likely to lead to unnatural conditions such as overgrazing and soil
erosion. Since a primary objective of a national park is to maintain
natural conditions, there may be a need for management intervention to
prevent herbivores from altering the habitat from a normal state.

It is generally accepted that grazing intensity in excess of 50% of
standing-crop biomass leads to overgrazing and range deterioration. By
this criterion, the short grassland was being overgrazed. The
determination that nearly 40% of the grassland was bare ground is further

evidence.

6.4.3. Vegetation within the 15-year old exclosure
The striking differences between the vegetation within the 15-year

exclosure and the adjacent grazed plot underscores the role that grazing
plays in maintaining the short grassland. In the absence of grazing, a

taller grassland dominated by Cenchrus ciliaris replaces the short

152
grassland.

In the absence of both grazing and fire, numerous shrubs flourish in
the grassland and trees and thicket bushes get established. Fire hinders
the establishment of trees and shrubs but encourages a number of
pyrophilous perennial grasses such as Aygerrhenie filipendule. Fire also
removes vegetative cover and allows some of the short species, presumably
intolerant of shade, to maintain vigor and to persist in small
quantities. Grazing and fire in moderate frequency and intensity,

therefore, tend to maintain species diversity in the grassland.

6.5. The Mweya Peninsula: A case study of vegetation changes induced by
grazing
The recent history of game populations in Queen Elizabeth National Park
has been one of booms and bursts. Before the turn of the century, an
epidemic of rinderpest decimated a large proportion of the game (Lugard
1893). Then a scourge of sleeping-sickness affected the human inhabitants
of the area who either left or were evacuated by the colonial government
(Hale-Carpenter 1921). Game benefitted from the removal of hunting
pressure and increased rapidly. In recognition of their abundance, the
area was declared a game reserve in 1934 and a national park in 1952.
Soon after the area became a national park, it became clear that the
hippopotamus had exceeded the carrying capacity of the range and were in
danger of destroying the habitat (Petrides 8 Swank 1965). A decision was
made to cull the species (Bere 1959) and approximately 7000 animals were
cropped between 1958 and 1967 ‘(Field 8 Laws 1970). The vegetation
responded with a spectacular recovery (Laws 1968; Thornton 1971).

153
On Mweya Peninsula, the cropping scheme resulted not only in improved
vegetation cover but also a more diverse and balanced large-mammal
community (Eltringham 1974, 1980). After the end of the cropping scheme,
hippopotamus increased rapidly at the expense of other ungulates on the
Peninsula so that during the study there was once again a virtual
monoculture of hippopotamus which threatened habitat degradation. The

number of hippopotamus in the park as a whole, however, is estimated to
be lower than the pre—cropping level of 14000.

Table 6.8. Changes in frequency* of grass species on the Mweya Peninsula,
Queen ElizabethNetionel Park, Uganda.

Sjecies 3757 5/58 5759 575a 57§g 9/66 1971-4 1974-6 578e_

 

CO 3.5 37.7 21.1 22.9 6.6 (5.0 3 23 67.8
CD 2.3 9.2 2.1 3.2 11.7 8.9 - 17 6.9
SP 12.9 8.0 25.2 18.9 27.6 31.0 100 68 62.8
C6 0.8 6.4 6.8 7.2 19.4 12.2 92 23 21.2
DA - 5.2 0.1 2.2 0.5 <5.0 3 3 20.2

 

Species: CG, Chloris gayana; CO, Chrysochloa orientalis; CD, Cynodon
dactylon; DA, Dactyloctenium aegyptium; SP, Sporobolus pyramidalis.
*Frequency values between 1958-62 are expressed as % of basal cover. All
others are % of occurrence in random quadrats. Authorities: Petrides 8
Swank (1965) for 1957; Thornton (1971) for 1958-1962; Spinage (1968) for

1966; Edroma (1977) for 1971-1974; Yoaciel (1981) for 1974-1976; This
study for 1982.

Since 1957, a series of studies has recorded the frequency of five
common grasses (Table 6.8). Of these, one or a combination of three
(Chrysochloe, orientalis, Sporobolus pyramidalis and Qfllgeie,geyege) has
always been dominant. At the beginning of the series, Q. orientalis was
the dominant grass. Its record shows that between 1958 and 1974, it

declined in abundance. Much of its decline was coincidental with a period

154
of hippopotamus cropping. But has recovered remarkably since 1974 and in
the present study was found to be the most abundant species in the
sward.

Sggrobolus pyramidalis, on the other hand, increased through out the
years of hippopotamus cropping replacing Q. orientalis as the most
abundant species in 1962 (Table 6.8). It continued to increase in
abundance until 1974. Since then, it has declined slightly as Q.
orientalis increased dramatically to regain dominance as the most
abundant species. Its decline has been simultaneous with an increase in
hippopotamus density.

The rise of Q. orientalis to dominance is a recent event. A survey
made in 1976 (Yoaciel 1981), reported that Q. gyramidalis was still the
dominant grass on the Peninsula. The species has been dominant in the
area since 1961 when it replaced Q. orientalis following the elimination
of hippopotamus (Thornton 1959, 1962, 1971; McKay 1961; Lock 1964; Field
8 Laws 1970).

Q. orientalis is a carpet grass and is encouraged by hippopotamus
grazing. The abundance of stoloniferous species and annuals in the sward
is indicative of overgrazing. The fluctuations in Q. orientalis and Q.
gyramidalis indicate that the former dominates the grassland when
hippopotamus grazing becomes excessive, and the latter dominates when
hippopotamus grazing is moderate or relatively low. Conversely, the
absence of the tufted perennials I, triandra, A. contortus and A,
filipendula is evidence that grazing has been heavy and persistent; and

that fire has been absent for lack of fuel for sometime.

 

CHAPTER 7

CONCLUSION AND RECOMMENDATIONS

7.1. Conclusion

Based on 18 months of field work in Queen Elizabeth National Park,
Uganda:

1. Game counts in selected areas of the park showed that the
populations of large and medium—sized herbivores in 1981/83 were the
lowest recorded since comparable counts were first made in 1956. The
magnitude of the decline was particularly large for elephant, buffalo and
waterbuck. This reduction is attributed primarily to poaching.

2. Average population densities and standing-crop biomasses have
fallen in all except one of four study areas. The only increase occurred
on Mweya Peninsula is due entirely to the expansion of hippopotamus. All
other species declined.

3. The hippopotamus population on Mweya Peninsula was the highest
ever recorded, with a density of 34.0/sq km and a standing-crop biomass
of 33,955 kg/sq km. As this example illustrates, there are local
exceptions to the general decline in animal numbers. Excessive decreases
and increases in wildlife abundance have different but equally-serious
management implications.

4. When the first counts were made in the late 19503, Mweya Peninsula
was over-populated with hippopotamus. To arrest habitat deterioration and
restore a more balanced community, between 1958 and 1967 a cropping

program was implemented. Since the end of the program, hippopotamus have

155

156
returned to the peninsula in large numbers. The large-mammal community
has degenerated once again to a virtual monoculture of hippopotamus. The
abundance, high biomass and great species diversity of large mammals
which occurred after the cropping of hippopotamus have been lost.

5. The variability of animal counts from month to month was high but
showed no marked trends over the course of the study. Censuses of
individual species seemed to be influenced by such factors as body-size,
wariness, herding characteristics, mobility, the size of the population,
as well as the ability of observers to see all the animals present.

6. Census results showed no significant seasonal differences. This
supported the general impression that park animals do not undertake large
scale seasonal movements.

7. The distribution of ungulates indicated strong species preferences
for certain habitats. Hippopotamus was confined to the short grass and
thicket-bush mosaic close to water. Others, such as kob, occurred mostly
in open, medium-height grasslands but were present also in the tall grass
and in the short grass with bush habitats. Only buffalo, and to a lesser
extent warthog, were fairly evenly distributed in all three habitats.

8. A woodland vegetation type in the Craters region in which a heavy
loss of trees due to widespread elephant damage was reported when it was
first surveyed in 1968 was re-surveyed. The present study disclosed it to
be regenerating abundantly. The current profusion of regeneration
indicates that the trend towards grassland is not irreversible and that
woodland, at least Aeeeie gerrardii woodland, can be restored by
reducing elephant densities (in this case by poaching). Regeneration was

found to be almost entirely from root stocks. Seedlings contributed

 

157
little, if any, towards stand regeneration.

9. Average tree density of all species taken together was 164.3/ha.
The most abundant were Aeeeie gerrardii, A, sieberiana and A, Agegli;
with densities (lha) of 135.4, 23.4 and 4.2, respectively.

10. Qualitatively, species composition of the woodland has remained
virtually unchanged since 1968. Major shifts have occurred, however, in
the relative abundance of species. The most significant is that of
Aeeeie gerrardii which in 1968 constituted 21.3% of the trees in the
woodland, but by 1982 had increased to 82.4%. Two other species which
were as abundant as A. gerrardii in the earlier survey have declined
greatly and are longer important components of the vegetation. Bridelia
scleroneuroides and glee africana constituted 20.9 and 20.2%,
respectively, of the trees in the woodland in 1968 but only 0.1% each in
1982.

11. Aeeeie_gerrardii trees in the regenerating woodland exhibited a
bimodal girth distribution, confirming the existence of two distinct
sub-populations. One consisted of abundant young regeneration and the
other of a limited cohort of Sparsely-scattered mature trees. In terms of
height, the former constituted a lower and the latter an upper tier, with
virtually no intermediaries. A period of severe suppression between the
two generations is estimated to have lasted two to four decades.

12. Evidence as to the respective roles of fire and elephant in
controlling this woodland vegetation and thicket expansion was examined.
It was concluded that elephant, not fires, were the primary factor in
controlling woody vegetation. This conclusion was based on the observed

gradient of decreasing mature-tree density with increasing distance from

158

the park border. This pattern of tree distribution seemed likely to have
developed as a result of differential use of the range by ungulates in
response to poaching and harrassment along the border. In Murchison Falls
National Park, Uganda, analysis of serial photographic transects showed
that similar increases in tree density occurred at the boundaries of
elephant ranges. Observations on experimental plots from which elephant,
but not fire, have been excluded also support the view that elephant are
more important than fire in controlling woody vegetation.

13. The concensus among park authorities is that thickets have
increased throughout the park over the last decade. The density of
thicket clumps was determined to be 26.4/ha on the heavily-grazed Mweya
Peninsula and 2.8 in moderately-grazed and frequently burnt mediumrheight
grassland at Kikorongo. Individual clumps in Kikorongo were generally
larger than those on the peninsula.

14. In general, the increase in thickets is probably the result of
their release from fires and heavy browsing. In heavily-grazed areas,
such as the banks of the Kazinga Channel and the Mweya Peninsula, the
absence of fires (due to inadequate fuel to support them) allows thickets
to flourish.

15. A thicket clump commonly contained 20 to 40 species of plants.
The composition of clumps varied as a result of variations in topography,
soil and precipitation within the park. Common clump constituents

included Caggaris tomentosa Securinega virosa and Azima tetracantha.

 

16. It was concluded that thicket clumps originate from the nucleus
of a single bush in the grassland. The bush has to be a vigorous species

in order to persevere and survive frequent fires and persistent

159
defoliation by large herbivores.

17. The continued active growth of many thicket species during the
dry season when the surrounding grassland is parched and brown is
probably because they have deep roots and can restrict water-loss through
the stomata.

18. Thicket clumps constitute an important resource to wildlife. They
provide food, shelter, shade and concealment for a wide range of
animals.

19. Vegetative standing-crop biomass was positively correlated with
increasing distance from the hippopotamus-populated waters. Light and
moderate grazing enabled the maintenance of good vegetative cover whereas
heavy grazing severely reduced grass abundance and resulted in bare soil
patches. It is concluded that grazing is responsible for creating and
maintaining the mosaic of grasslands in the park. Light grazing results
in the accumulation of large quantities of plant litter and the risk of a
hot fire. Moderate grazing results in considerably less litter and much
reduced risk of fire. Heavy grazing exposes the soil to agents of erosion
and is, no doubt, the major factor inducing bush-invasion and habitat
deterioration in some areas of the park. f

20. In the mixed short grassland on Mweya Peninsula, Chrysochloa
orientalis has replaced Sporobolus pyramidalis as the dominant grass
species. Chrysochloa is a prostrate species which tends to escape heavy
grazing. It was dominant during the late 19503 when the peninsula
experienced severe habitat deterioration induced by hippo overgrazing and
unusually low rainfall. Subsequent cropping of the hippopotamus resulted

in improved cover, a decline in Chrysochloa and a rise to dominance of

160
Segrobolus. The current return of Chrysochloa to dominance is evidence

that grazing had once again become excessive.

7.2. Recommendations

1. In the short term, an effort should be made to expand, equip and
maintain a disciplined anti-poaching game-ranger force to prevent the
further decline of wildlife populations.

2. Reform is necessary to make poaching a more serious crime than it
is at present. Stiffer sentencing of offenders could be a deterrence. At
present, a poacher needs to succeed only once to make enough money to pay
for four to five convictions. Consequently, conviction has not been a
deterrence.

3. In the long term, the key to the park's survival as a viable
conservation area will be the support of the local citizenry. It is
recommended that funds be made available for national conservation
education. It is further recommended that schools and communities
surrounding national parks be principal targets of the campaign.

4. Immediate action to reduce hippopotamus population density is
necessary to halt and reverse the process of habitat degradation in areas
such as Mweya Peninsula. Previous experience indicates that a cropping
project would result in improved habitat conditions and increased variety
of attractive ungulates. Such a program would pay for itself and enable
research on the biology of target species. It would also, however, raise
questions as to whether poaching can be wrong if culling is correct. A
parallel program of conservation education would be critically important

to establish a recognition of the objectives and values of national

161
parks, of national economic benefits through tourism and of modern

concepts of resource management.

LITERATURE CITED

Allsopp, R. 1978. Social biology of bushbuck (Tragelaphus scriptus
Pallas) in Nairobi National Park, Kenya. E. Afr. Wildl. J.
16:153-165.

Astle, W. L. 1971. Management in the Luangwa Valley. Oryx 11:135-139.

Beadle, L. C. 1965. Uganda National Parks in prehistoric times. In
Uganda National Parks Handbook, 4th. edition, pp. 89-92. Trustees of
Uganda National Parks, Kampala.

Beasom, S. L. 8 Haucke, H. H. 1975. A comparison of four distance
sampling techniques in South Texas live oak mottes. J. Range Manage.
28:142-144.

Bere, R. M. 1959. Queen Elizabeth National Park, Uganda: the hippopotamus
problem and experiment. Oryx 5:116-124.

Bertram, B. C. 1978. Living in groups: Predators and prey. In J. R. Krebs
8 N. 8. Davies (eds.), QeAeviorel ecology: An evolutionery approeeA,
pp. 64-96. Blackwell, Oxford.

Bews, J. W. 1913. An ecological survey of the midlands of Natal with
special reference to the Pietermaritzburg District. ann. Nata. Mus.
2:485-545.

Bishop, W. W. 1970. Pleistocene stratigraphy in Uganda. Mem. geol. Surv.
Uganda, 10.

Bishop, W. W. 8 Posnansky, M. 1960. Pleistocene environments and early

man in Uganda. Uganda J. 24:44-61.

162

 

163
Bradley, R. M. 1968. Some aspects of the ecology of the warthog
(Phacochoerus aethiopicus (Eellee)) in Nairobi National Park. MSc
thesis, University of Nairobi.
Brooks, A. 1957. Notes on some ecological studies in the Queen Elizabeth
National Park with particular reference to grasses. Game Department,

Entebbe. Unpubl. m.s., pp. 24.

Buechner, H. K. 1961. Territorial behavior in Uganda kob. Science
133:698-699.

Buechner, H. K. 8 Dawkins, H. C. 1961. Vegetation change induced by
elephant and fire in Murchison Falls National Park, Uganda. Ecology
42:752-766.

Buechner, H. K. 8 Schloeth, R. 1965. Ceremonial mating behavior in Uganda
kob (Adenota kob thomasi Neumann). Z. Tierpsychol. 22:209-225.

Caughley, G. 1976. The elephant problem - an alternative hypothesis. E.
Afr. Wildl. J. 14:265-283. .

Chippindall, L. K. A. (Ed.) 1955. A guide to the identificetjon of
geesses in South Africa: The gresses and pastures of South Africe.
Central News Agency, Johannesburg.

Clough, G. 1969. Some preliminary observations on reproduction in the
warthog, Phacochoerus aethiopicge Pallas. J. Reprod. Fert., Suppl.
6:323-337.

Clough, G. 8 Hassam, A. G. 1970. A quantitative study of the warthog in
the Queen Elizabeth National Park, Uganda. E. Afr. Wildl. J.
8:19-24.

Cottam, G. 8 Curtis, J. T. 1956. The use of distance measures in

phytosociological sampling. Ecol. 37:451-460.

164

Cottam, 6., Curtis, J. T. 8 Hale, B. W. 1953. Some sampling
characteristics of a population of randomly dispersed individuals.
Ecology 34:741-757.

Curtis, J. T.8 McIntosh, R. P. 1951. An upland forest continuum in the
praire-forest border region of Wisconsin. Ecol. 32:476-496.

Davidson, J. L. 8 Milthorpe, F. L. 1966a. Leave growth in Dactylis
glomerata following defoliation. Ann. Bot. (Lond.) 30:173-184.

Davidson, J. L. 8 Milthorpe, F. L. 1966b. The effect of defoliation on
the carbon balance in Qectylis glomerata. Ann. Bot. (Lond.)
30:185-198.

Dix, R. L. 1961. An application of the point-centered quarter method to
the sampling of grassland vegetation. J. Range Manage. 14:62-69.

Edroma, E. L. 1973. Poaching and human pressures in Rwenzori National
Park. Uganda J. 37:9-18.

Edroma, E. L. 1975. Influences of burning and greging on the
productivity and dynemics of grasslands in Rwenzori Netionel Park,
Agegge. ScD thesis, University of Giessen, West Germany.

Edroma, E. L. 1977. Population structure and management of grassland
types in Rwenzori National Park, Uganda. Geo. Eco. Trop. 4:277-294.

Edroma, E. L. 1981. Some effects of grazing on the productivity of
grassland in Rwenzori National Park, Uganda. Afr. J. Ecol.
19:313-326.

Elliot, G. F. 1896. A naturalist in mid-Africa. A. D. Innes, London.

Eltringham, S. K. 1973. The large mammal community of the Queen Elizabeth
National Park, (Uganda). Ann. Univ. Abidjan, serie E 6:189-219.

165
Eltringham, S. K. 1974. Changes in the large mammal community of Mweya
Peninsula, Rwenzori National Park, Uganda, following removal of
hippopotamus. J. appl. Ecol. 11:855-865.
Eltringham, S. K. 1976. The frequency and extent of uncontrolled fires in
the Rwenzori National Park, Uganda. E. Afr. Wildl. J. 14:215-222.
Eltringham, S. K. 1977. The numbers and distribution of elephants
Loxodonta africana in the Rwenzori National Park and Chambura Game
Reserve, Uganda. E. Afr. Wildl. J. 15:19-39.

Eltringham, S. K. 1980. A quantitative assessment of range usage by large
African mammals with particular reference to the effects of elephants
on trees. E. Afr. Wildl. J. 18:53—71.

Eltringham, S. K. 1982. Eleghants. Blandford Press, Dorset.

Eltringham, S. K. 8 Din, N. A. 1977. Estimates of population size of some
ungulate species -in the Rwenzori National Park, Uganda. E. Afr.
Wildl. J. 15:305-316.

Eltringham, S. K. 8 Malpas, R. C. 1976. Elephant slaughter in Uganda.
Oryx 13:334-335.

Eltringham, S. (K. 8 Malpas, R. C. 1980. The decline in elephant numbers
in Rwenzori and Kabalega Falls National Parks, Uganda. Afr. J. Ecol.
18:73-86.

Eltringham, S. K. 8 Woodford, M. H. 1973. The numbers and distribution of
buffalo in the Rwenzori National Park, Uganda. E. Afr. Wildl. J.
11:151-164.

Estes, R. D. 1974. Social organization of the African Bovidae. I. U. C.
N. Publ. N. S. 24:166-205.

 

166

Field, R. C. 1968. The food habits of some unguletes in Ugande. PhD
thesis, Cambridge University.

Field, R. C. 1970. A study of the feeding habits of the hippopotamus
(Hippopotamus amphibius Linn.) in the Queen Elizabeth National Park,
Uganda, with some management implications. Zool. Afr. 5:71-86.

Field, R. C. 1971. Elephant ecology in the Queen Elizabeth National Park,
Uganda. E. Afr. Wildl.. J. 9:99-123.

Field, R. C. 1972. The food habits of wild ungulates in Uganda by
analyses of stomach contents. E. Afr. Wildl. J. 10:17-42.

Field, C. R. and Laws, R. M. 1970. The distribution of the larger
herbivores in the Queen Elizabeth National Park, Uganda. J. appl.
Ecol. 7:273-294.

Galton, F. 1871. Gregariousness in cattle and man. Macmillan's Mag. Lond.
23:353-357.

Glover, J. 1963. The elephant problem at Tsavo. E. Afr. Wildl. J.
1:30-53.

Grimsdell, J. J. R. 1969. Ecology of the buffelgL Syncerus caffer, ifl
Western Uganda. PhD thesis, Cambridge University.

Hale-Carpenter, G. D. 1921. Unpubl. report of the S.M.O.S.S. 10.1.21.
Uganda Medical Department Archives, Entebbe.

Hamilton, W. D. 1970. Selfish and spiteful behavior in an evolutionary
model. Nature 228:1218-1220.

Hamilton, W. D. 1971. Geometry for the selfish herd. J. theor. Biol.
12:295-311.

Harthoorn, A. M. 1966. The Tsavo elephants. Oryx 8:233-236.

167

Harrington, G. N. 8 Ross, 1. C. 1974. The savanna ecology of Kidepo
Valley National Park. I. The effects of burning and browsing on the
vegetation. E. Afr. Wildl. J. 12:93-105.

Harrop, J. 1960. The soils of the Western Province of Uganda. Mem. Res.
Div. Dep. Agric. Uganda. Series 1:6.

Heady, H. F. 1966. Influence of grazing on the composition of [Aegege
triandra grassland, East Africa. J. Ecol. 54:705-727.

Jackson, 6. 8 Gartlan, J. S. 1965. The flora and fauna of Lolui Island,
Lake Victoria. J. Ecol. 53:573-597.

Jarman, P. J. 1974. The social organization of antelope in relation to
their ecology. Behaviour 48:215-267.
Junguis, H. 1971. The biology end behevior of the reegguck (Redunca
arundinum Agggeert. 1785) in the Kruger Netionel PeLA. Hamburg:Verlag
Paul Parey.
Kenyi, J. M. 1978. Aerial counts of buffalo herds in Rwenzori National
Park, Uganda. E. Afr. Wildl. J. 16:221-222.
Kenyi, J. M. 1980. Seasonal defoliation of thickets by larvae of
Belenois eurote, Afr. J. Ecol. 18:123-126.

Lamprey, H. F. 1967. Notes on the dispersal and germination of some tree
seeds through the agency of mammals and birds. E. Afr. Wildl. J.
5:179-180.

Langdale-Brown, I. 1960. The vegetation of the Western Province of
Uganda. Mem. Res. Div. Dep. Agric. Uganda, Series 2:4.

Laws, R. M. 1966. Age criteria for the African elephant, Loxodonta
africana. E. Afr. Wildl. J. 4:1-37.

168

Laws, R. M. 1968. Interactions between elephant and hippopotamus
populations and their environments. E. Afr. agric. For. J.
33:140-147.

Laws, R. M. 1969. Aspects of reproduction in the African elephant,
Loxodonte africana. J. Reprod. Fert. Suppl. 6:193-217.

Laws, R. M. 1970. Elephants as agents of habitat and landscape change in
East Africa. Oikos 21:1-15.

Laws, R. M. 1974. Behaviour, dynamics and management of elephant
populations. I. U. C. N. Publ. N. S., 24:513-529.

Laws, R. M., Parker, I. s. c. a Johnstone, R. c. a. 1970. Elephants and
habitats in north Bunyoro, Uganda. E. Afr. Wildl. J. 8:163-180.

Laws, R. M., Parker, I. S. C. 8 Johnstone, R. C. B. 1975. Qlephents and

 

their habitats. Clarendon Press, Oxford.

Lawton, R. M. 8 Gough, M. 1970. Elephants or fire - which is to blame?
Oryx 10:244-248.

Lebrun, J. 1947. La vegetation de laplaine alluviale au Sud du lac
Edouard. In Exploration du Parc National Albert, vol. 1. Institut
des Parcs Nationaux du Congo Belge, Brussels.

Lemon, P. C. 1968. Fire and wildlife grazing on an African plataue. Proc.
Tall Timbers Fire Ecol. Conf. 8:71-88.

Leuthold, W. 1974. Observations on home range and social organization of
lesser kudu, Tragelaphus imberbis (Blyth, 1869). I.U.C.N. Publ. N.S.
24:206-234.

Leuthold, W. 1977. African ungulates: A comparative review of the

ethology and behavioral ecology. Springer Verlag, Berlin.

169
Leuthold, W. 8 Leuthold, B. M. 1976. Density and biomass of ungulates in
Tsavo East National Park, Kenya. E. Afr. Wildl. J. 14:49-58.
Lock, J. M. 1967. Vegetation in relation to grazing and soils in Queen

Elizabeth National Park, Uganda. PhD thesis, University of
Cambridge.

Lock, J. M. 1970. The grasses of Queen Elizabeth National Park. Uganda J.
34:49-63.

Lock, J. M. 1972. The effects of hippopotamus grazing on grasslands. J.
Ecol. 60:445-467.

Lock, J. M. 1977. The vegetation of Rwenzori National Park, Uganda. Bot.
Jahrb. Syst. 98:372-448.

Lock, J. M. 8 Milburn, T. R. 1971. The seed biology of [Aegege triandra
Forsk., in relation to fire. Symp. Brit. Ecol. Soc. 11:337-349.

Lugard, F. O. 1893. The rise of our Qest Africanggpire. 2 vols.
Blackwood, Edinburgh.

MacArthur, R. H. 8 Wilson, E. O. 1967. The theory of islegg
biogeograghy. Princeton University Press, Princeton.

Malpas, R. C. 1977. Diet and the condition and growth of elephants in
Uganda. J. appl. Ecol. 14:489-504.

Malpas, R. C. 1978. The ecology of elephants in Rwenzori and Kabalega
Falls National Parks, Uganda. PhD thesis, University of Cambridge.

Malpas, R. C. 1980. Wildlife in Ugande_;980: A survey. A report to the
Minister of Tourism 8 Wildlife, Uganda, New York Zoological Society 8
WWF/IUCN.

Matthews, G. V. T. 1960. An examination of basic data from wildfowl
counts. Proc. int. Orn. Congr. Helsinki (1958) 12:483-491.

170

McKay, A. D. 1961. Unpubl. report, Nuffield Unit of Tropical Animal
Ecology, Uganda.

Mech, L. D. 1970. The wolf. Nat. Hist. Press, New York.

Milthorpe, F. L. 8 Davidson, J. L. 1965. Physiological aspects of
regrowth in grasses. pp. 241-255. Lg F. L. Milthorpe and J. D. Ivins
(eds.), The growth of cereals end grasses. Butterworths, London.

Modha, K. L. 1972. The ecology of the Agenda kob in the Queen Elizabeth
Nationel Park, Ugande. MSc thesis, Makerere University, Kampala.

Modha, K. L. 8 Eltringham, S. K. 1976. Population ecology of the Uganda
kob (Aeeegge Age Neumann) in relation to the territorial system in
the Rwenzori National Park, Uganda. J. appl. Ecol. 13:453-473.

Musoke, M. B. 1980. Overbrowsing of Capparis tomentosa bushes by goats
in Rwenzori National Park, Uganda. Afr. J. Ecol. 18:7-10.

Neal, 8. R. 1967. The ecology of small rodents in the grassland
community of the Queenyglleabeth National ParkLUgande. PhD thesis,
Southampton University.

Okali, D. U. U., Hall, J.B. 8 Lawson, G. W. 1973. Root distribution under

a thicket clump on the Accra plains, Ghana: Its relevance to clump
localization and water relations. J. Ecol. 61:439-455.

Okiria, R. 1980. Habitat exploitation by the bushbuck in Rwenzori
National Park. Afr. J. Ecol. 18:11-17.
Osmaston, H. A. 1965. The vegetation. In UgandefNationeJ Perks Handbook,

4th. edition, pp. 99-123. Trustees of Uganda National 'Parks,
Kampala.

171
Osmaston, H. A. 1971. The vegetation of Murchison Falls and Queen
Elizabeth National Parks. In Uganda Nationel Parks Handbook, 5th.

edition, pp. 95-112. Trustees of Uganda National Parks, Kampala.
Petrides, G. A. 1974. The overgrazing cycle as a characteristic of

tropical savannas and grasslands in Africa. Proc. First Intl. Congr.
Ecol., Wageningen, Centre Agric. Publ. Doc., 86-91.

Petrides, G. A. and Swank, W. G. 1965. Population densities and the
range-carrying capacity for large mammals in Queen Elizabeth National
Park, Uganda. Zool. Afr. 1:209-225.

Phillipson, J. 1975. Rainfall, primary productivity and ‘carrying
capacity' of Tsavo National Park (East), Kenya. E. Afr. Wildl. J.
13 171—201. '

Pianka, E. R. 1970. On r- and K-selection. Am. Nat. 104:592-597.

Posnansky, M. 1971. In UgeggeNational Parks Handbook. Trustees of the
Uganda National Parks, Kampala.

Pulliam, H. R. 1973. On the advantages of flocking. J. theor. Biol.
38:419-422.

Rattray, J. M. 1954. Some plant indicators in Southern Rhodesia.
Rhodesian Agric. J. 51:176-186.

Rhodes, D. H. 1965. Fish and fishing. In Uganda National Parks Handbook.
4th. edition, pp. 81-88. Trustees of Uganda National Parks, Kampala.

Robertson, R. J. 1973. Optimal niche space of the red-winged blackbird:

Spatial and temporal patterns of nesting activity and success.
Ecology 54:1085-1093.

172

Rubenstein, D. I. 1978. On predation, competition and the advantages of
group living. In P. P. Bateson 8 P. Klopfer (eds ), Perspectives in
ethology, pp. 205-231. Plenum Press, New York.

Sadleir, R. M. F. 1969. The ecology of reproduction in wild and domestic
meggele. Methuen, London.

Schaller, G. B. 1972. The Serengeti lion. Univ. Chicago Press, Chicago.

Sidney, J. 1965. The past and present distribution of some African
ungulates. Trans. zool. Soc. Lond. 30:1-397.

Sinclair, A. R. E. 1977. The African buffalo. Univ. Chicago Press.
Chicago.

Sinclair, A. R. E. 1977. Lunar cycle and timing of mating season in
Serengeti wildebeest. Nature, Lond., 267:832-833.

Sowls, L. K. 8 Phelps, R. J. 1965. Body temperature of juvenile warthogs
and bushpigs. J. Mammal. 47:134-137.

Spence, D. E. N. 8 Angus, A. 1971. African grassland management - burning
and grazing in Murchison Falls National Park, Uganda. Symp. Brit.
Ecol. Soc. 11:297-317.

Spinage, C. A. 1967. The eutecology of the Uganda waterbuck (Kob defassa
ugandae Neumann). PhD thesis, London University.

Spinage, C. A. 1982. A territorial antelope: The Uganda waterbuck.
Academic Press, London.

Spinage, C. A. 8 Guinness, F. E. 1971. Tree survival in the absence of
elephants in the Akagera National Park, Rwanda. J. appl. Ecol.
8:723-728.

Stanley, H. M. 1890. In darkest Africa. Vol. II. Sampson Law, London.

173

Strugnell, R. G. and Pigott, C. D. 1978. Biomass, shoot production and
grazing of two grasslands in Rwenzori National Park, Uganda. J. Ecol.
66:73-96.

Taylor, R. J. 1976. Value of clumping to prey and the evolutionary
response of ambush predators. Am. Nat. 110:13-29.

Thomson, P. J. 1975. The role of elephants, fire and other agents in the
decline of a Brachystegia boehmii woodland. J. S. Afr. Wildl. Manage.
Assoc. 5:11-18.

Thornton, D. D. 1959. Unpubl. report, Nuffield Unit of Tropical Animal
Ecology, Uganda.

Thornton, D. D. 1962. Unpubl. report, Nuffield Unit of Tropical Animal
Ecology, Uganda.

Thornton, D. D. 1971. The effect of complete removal of hippopotamus on
grassland in the Queen Elizabeth National Park, Uganda. E. Afr.
Wildl. J. 9:47-55.

Treisman, M. 1975a. Predation and the evolution of gregariousness. 1.
Models for concealment and evasion. Anim. Behav. 23:779-800.

Treisman, M. 1975b. Predation and the evolution of gregariousness. II. An
economic model for predator-prey interaction. Anim. Behav.
23:801-825.

Uganda Game Department Annual Reports. 1948-1952. Report of the Game 8
Fisheries Department (Game Section). Government Printers, Entebbe.
Uganda National Parks Annual Reports. 1953-1955, 1960. Report and

accounts of the Trustees of the Uganda National Parks. Uganda Argus

Ltd., Kampala.
van Orsdol, K. G. 1979. Uganda National Parks. Swara 2:14-16.

174

Vesey-Fitzgerald, D. 1973. Browse production and utilization in Tarangire
National Park. E. Afr. Wildl. J. 11:291-305.

Vine, I. 1971. Risk of visual detection and pursuit by a predator and the
selective advantage of flocking behavior. J. theor. Biol.
30:405—422.

Woodford, M. H. 1971. Tuberculosis in the African buffalo (Syncerus
caffer) in the Queengligabeth National Park, Uganda. D.V.M. thesis,
University of Zurich.

Wyatt, J. R. and Eltringham, S. K. 1974. The daily activity of the
elephant in the Rwenzori National Park, Uganda. E. Afr. Wildl. J.
12:273-289.

Waser, P. 1975a. Diurnal and nocturnal strategies of the bushbuck
Tragelephus scriptus (Pallas). E. Afr. Wildl. J. 13:49-63.

Waser, P. M. 1975b. Spatial associations and social interactions in a
‘solitary' ungulate: the bushbuck Tragelaphus scriptus (Pallas). Z.
Tierpsychol. 37:24-36.

West, 0. 1965. Fire in vegetation and its use in pasture management with
special reference to tropical and sub-tropical Africa. Mimeo. Publ.
No. 1/1965. Commonwealth Bureau of Pastures 8 Field Crops, Hurley,
Berks.

Western, D. 8 van Praet, P. C. 1973. Cyclical changes in the habitat and
climate of an East African ecosystem. Nature, Lond. 241:104-106.
Worthington, S. 8 Worthington, E. B. 1933. Inland waters of Africa.
Macmillan, London. A
Wynne-Edwards, V. C. 1962. Animal dispersion in relation to social

behavior. 653 pp. Hafner Publ. Co., New York.

175
Yoaciel, S. M. 1981. Changes in the populations of large herbivores and

in the vegetation community in Mweya Peninsula, Rwenzori National

Park, Uganda. Afr. J. Ecol. 19:303-312.
Yoaciel, S. M. and van Orsdol, K. G. 1981. The influencce of

environmental changes on an isolated topi (Damaliscus lunatus jimele
Matschie) population in the Ishasha Sector of Rwenzori National Park,

Uganda. Afr. J. Ecol. 19:167-174.

 

5518

93 030

miiii”Lilli!liliililililliiil