Savanna woody regeneration in response 
to different treatments of herbivory and 
fire 
 
 
A Combrink 
22751165 
 
 
 
Dissertation submitted in fulfilment of the requirements for the 
degree Magister Scientiae in Environmental Sciences at the 
Potchefstroom Campus of the North-West University 
 
 
 
Supervisor:  Dr F Siebert 
Co-supervisor: Ms JM Botha 
 
 
 
 
 
May 2017 
i 
ABSTRACT 
Woody regeneration is one of the fundamental processes responsible for structuring savanna 
systems. Herbivory and fire are two primary drivers of woody structure, and therefore also 
woody regeneration. Regeneration responses of woody species to savanna controlling factors 
can be examined and explained by using the combined competition-based and demographic 
bottleneck model. Woody encroachment has received increasing attention in savanna system 
dynamics since savannas are considered vulnerable to encroachment. Woody species can 
broadly be categorised as encroachers, which are considered as species with the ability to 
outcompete other life forms when disturbances occur, and non-encroachers, which are 
considered more desirable. The savanna demographic bottleneck model has not yet been 
applied to these two woody functional groups separately. This could bring insight into woody 
encroachment and to what extent it can be controlled by herbivory and fire without negatively 
affecting non-encroacher species.  
The main aim of this study was to test the effect of herbivory and fire (presence and release 
thereof) on woody regeneration of both encroacher and non-encroacher woody species across 
a small-scale (139 ha) heterogeneous landscape in a riparian semi-arid savanna ecosystem. 
The specific objectives were to (i) describe and compare dominant woody families and species 
and basic PFtraits of the woody layer across different treatments of herbivory and fire (presence 
and exclusion), (ii) evaluate the effects of herbivore exclusion on woody species assemblages, 
(iii) evaluate the effect of herbivory and fire on woody species abundances, and (iv) evaluate the 
effect of herbivore and fire presence and exclusion on woody community and population 
demography and stability. Woody species assemblages referred to the statistically tested woody 
species composition in the specific treatment and woody communities is the group of woody 
species within the boarders of the herbivore or fire treatment. The broad hypothesis states that 
the exclusion of both herbivory and fire from a semi-arid savanna ecosystem will enhance 
regeneration and recruitment of woody species. 
The study was conducted at the Nkuhlu long-term exclosures situated in the southern parts of 
Kruger National Park, South Africa. The exclosures are divided into different treatments of 
herbivory and fire. Herbivore treatments consist of a fully fenced exclosure (designed to exclude 
all mammalian herbivores larger than a hare); a partial exclosure (designed to exclude elephant, 
but also excludes giraffe due to their body height) and a control site (exposure to all large 
mammalian herbivores). Each herbivore treatment was devided into a fire-exposed and fire-
excluded area by means of a fire break. Woody individuals were sampled inside permanently 
marked plots located on transects initiating in the riparian vegetation zone (close to the Sabie 
River), extending across the sodic midslopes to the crest (uplands).  
ii 
Results from the floristic analyses indicated that 13 years of excluding herbivory and fire was 
long enough to initiate changes in abundance of dominant families and plant functional types, 
and excluding herbivores also changed species composition. Herbivores played an important 
role in structuring the woody layer, although fire had much less effects on woody regeneration 
than was expected in this savanna type. Woody species abundance results indicated that 
herbivore activity negatively impacted recruitment of both encroacher and non-encroacher 
species, with effects differing between the two groups. Herbivore effects were also evident in 
community and population demography. Herbivores managed to suppress regeneration of key 
encroacher species, except for Dichrostachys cinerea. Key non-encroacher species differed in 
their response to herbivore activity, with some indicating demographic resistance to herbivore 
pressure.  
Key words: recruitment; encroachers; non-encroachers, recruitment bottleneck, elephants, 
mesoherbivores. 
  
iii 
ACKNOWLEDGEMENTS 
Firsty, I would like to give all honour to God, who carried me throughout this journey.  
I would like to thank the following people for their contribution to this dissertation: 
 My supervisor, Frances Siebert, and co-supervisor, Judith Botha, for their valuable input 
and time invested in this project. 
 Helga and the rest of our field work team for assistance with data sampling. 
 Gwen Zambatis (Skukuza herbarium) for assistance in identifying specimens. 
 SANParks for general logistical support. 
 Research Unit: Environmental Sciences and Management, North West University for 
financial support. 
 My husband, Luan, for his encouragement, love and unconditional support throughout 
this journey. 
 My parents and family for their love and support. 
  
iv 
TABLE OF CONTENTS 
 
Abstract i 
Acknowledgements iii 
List of Tables vii 
List of Figures ix 
 
Chapter 1: Introduction 1 
 1.1. Background and rationale 1 
 1.2. Objectives 3 
 1.3. Hypotheses 4 
 1.4. Dissertation layout 4 
Chapter 2: Literature review 6 
 2.1. Savanna vegetation structure and dynamics 6 
  2.1.1. Plant-plant interactions 6 
  2.1.2. Plant-herbivore interactions 8 
  2.1.3. Plant-fire interactions 9 
  2.1.4. Plant-soil interactions 9 
 2.2. Regeneration of woody species 10 
  2.2.1. Models that explain regeneration of woody species 11 
  2.2.2. Savanna woody floristic changes driven by herbivory and fire 13 
  2.2.3. Savanna woody structural changes driven by herbivory, fire and the 
herbaceous layer 
13 
 2.3. Woody encroachment 15 
Chapter 3: Study area 17 
 3.1. Locality 17 
 3.2. History 18 
 3.3. Herbivore community 18 
v 
 3.4. Climate 19 
 3.5. Geology, soil and topography 19 
 3.6. Vegetation 20 
Chapter 4: Materials and Methods 22 
 4.1. Experimental design 22 
 4.2. Data sampling 24 
 4.3. Data analyses 25 
Chapter 5: Floristic, functional and species composition changes 26 
 5.1. Introduction 26 
 5.2. Methods 27 
 5.3. Results and discussion 28 
  5.3.1. Floristic changes 28 
   5.3.1.1. Responses to herbivore treatments 28 
   5.3.1.2. Responses to fire treatments 34 
  5.3.2. Basic plant functional type changes 39 
   5.3.2.1. Responses to herbivore treatments 39 
   5.3.2.2. Responses to fire treatments 44 
  5.3.3. Species composition changes 49 
   5.3.3.1. Response to herbivore exclusion 49 
 5.4. Conclusion 54 
Chapter 6: Woody abundance changes across herbivore and fire treatments 55 
 6.1. Introduction 55 
 6.2. Methods 57 
 6.3. Results 58 
  6.3.1. Complete woody community 58 
  6.3.2. Woody abundances of encroacher species 61 
  6.3.3. Woody abundances of non-encroacher species 64 
 6.4. Discussion 67 
vi 
 6.5. Conclusion 68 
Chapter 7: Demography of the woody community after 13 years of herbivore and 
fire manipulations 
69 
 7.1. Introduction 69 
 7.2. Methods 70 
 7.3. Results 73 
  7.3.1. Population structure of the complete woody community 73 
  7.3.2. Population structure of pre-selected key species 77 
 7.4. Discussion 88 
 7.5. Conclusion 92 
Chapter 8: Summary and general recommendations 94 
 8.1. Main findings 94 
 8.2. Recommendatios for future studies 96 
 
References 97 
Appendices 116 
 Appendix A 116 
 Appendix B 119 
     
 
  
vii 
LIST OF TABLES  
Table 5.1: Plant functional type ratios for dominant seedling species over time across 
herbivore treatments and vegetation zones. ................................................. 42 
Table 5.2: Plant functional type ratios for dominant established tree species over time 
across herbivore treatments and vegetation zones. ..................................... 43 
Table 5.3: Plant functional type ratios for dominant seedling species in 2015 across fire 
treatments and vegetation zones. ................................................................ 47 
Table 5.4: Plant functional type ratios for dominant established tree species in 2015 
across fire treatments and vegetation zones. ............................................... 48 
Table 6.1: Summary of significant interaction effects between treatments (vegetation 
zones, herbivore treatments, fire treatments and year (2002–2015)) for 
seedlings and established individuals separately. The complete woody 
community assessment included all woody individuals, which was then 
separated into encroacher and non-encroacher species for separate 
analyses on these functional groups. ........................................................... 58 
Table 6.2: Woody seedling abundances per year across vegetation zones and herbivore 
treatments. ................................................................................................... 59 
Table 6.3: Woody seedling abundances per year across fire and herbivore treatments. .......... 60 
Table 6.4: Woody abundances of established individuals per year across vegetation 
zones and herbivore treatments. .................................................................. 61 
Table 6.5: Woody abundances of encroacher seedlings per year across vegetation zones, 
herbivore treatments and fire treatments. ..................................................... 62 
Table 6.6: Woody abundances of established encroacher individuals per year across 
vegetation zones and herbivore treatments. ................................................. 63 
Table 6.7: Woody abundances of non-encroacher seedlings per year across vegetation 
zones and herbivore treatments. .................................................................. 64 
Table 6.8: Mean woody abundance of non-encroacher seedlings between 2002 and 2015 
in different fire treatments per vegetation zone. ............................................ 65 
viii 
Table 6.9: Woody abundances of non-encroacher seedlings per year herbivory treatments 
and fire treatments. ...................................................................................... 65 
Table 6.10: Woody abundances of established non-encroacher individuals per year 
across vegetation zones and herbivore treatments. ..................................... 66 
Table 7.1: Size-classes for woody individuals based on their height (m). ................................ 70 
Table 7.2: Pre-selected diagnostic encroacher and non-encroacher woody species per 
vegetation zone. ........................................................................................... 73 
Table 7.3: Summary of size-class distribution measures of the complete woody 
community in 2015 for each herbivore and fire treatment. Ordinary least 
square regression analyses (Slope, SE Slope, R2, p), Permutation Index 
(PI) and Simpson’s Dominance Index (SDI). ................................................ 74 
Table 7.4: Summary of size-class distribution measures for the key woody encroacher and 
non-encroacher species of the riparian vegetation zone in different 
herbivory treatments in 2015; Ordinary least square regression analyses 
(Slope, SE Slope, R2, p), Permutation Index (PI) and Simpson’s 
Dominance Index (SDI). ............................................................................... 79 
Table 7.5: Summary of size-class distribution measures for the key woody encroacher and 
non-encroacher species of the sodic vegetation zone in different 
herbivory treatments in 2015; Ordinary least square regression analyses 
(Slope, SE Slope, R2, p), Permutation Index (PI) and Simpson’s 
Dominance Index (SDI). ............................................................................... 82 
Table 7.6: Summary of size-class distribution measures for the key woody encroacher and 
non-encroacher species of the crest vegetation zone in different 
herbivory treatments in 2015; Ordinary least square regression analyses 
(Slope, SE Slope, R2, p), Permutation Index (PI) and Simpson’s 
Dominance Index (SDI). ............................................................................... 85 
Table 7.7: Summary of the herbivore effect on pre-selected woody species demography 
(i.e. regeneration) at the Nkuhlu exclosure in 2015. ..................................... 88 
  
ix 
LIST OF FIGURES 
Figure 3.1: Location of the Nkuhlu exclosures research site in the Kruger National Park, 
South Africa. ................................................................................................ 17 
Figure 3.2: Aerial image (Google Maps, 2016) of the Nkuhlu exclosures study site overlain 
by the herbivore treatments, which illustrate the position of the broader 
vegetation zones (Riparian, Sodic and Crest) across the topographic 
sequence. More detail on the herbivore treatments will be provided in 
Chapter 4. .................................................................................................... 21 
Figure 4.1: Graphic representation of the experimental design of the Nkuhlu long-term 
research exclosures (adapted from Van Coller et al., 2013). ........................ 23 
Figure 4.2: The fully fenced (a; right side of the fence is inside the exclosure) and partially 
fenced (b; right side of the fence is in the exclosure) exclosures which 
represent the herbivore treatments at the Nkuhlu exclosures ....................... 23 
Figure 5.1: Top three dominant woody families (expressed as abundance in %) and 
species per year (i.e. 2002 and 2015) for both seedling and established 
communities for each vegetation zone (riparian, sodic, crest) of the 
control treatment. ......................................................................................... 31 
Figure 5.2: Top three dominant woody families (expressed as abundance in %) and 
species per year (i.e. 2002 and 2015) for both seedling and established 
communities for each vegetation zone (riparian, sodic, crest) in the 
partially fenced treatment. ............................................................................ 32 
Figure 5.3: Top three dominant woody families (expressed as abundance in %) and 
species per year (i.e. 2002 and 2015) for both seedling and established 
communities for each vegetation zone (riparian, sodic, crest) in the fully 
fenced treatment. ......................................................................................... 33 
Figure 5.4: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the riparian zone of the control treatment............................. 35 
Figure 5.5: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the crest zone of the control treatment. ................................ 36 
x 
Figure 5.6: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the riparian zone of the partially fenced treatment. .............. 36 
Figure 5.7: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the sodic zone of the partially fenced treatment. .................. 37 
Figure 5.8: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the crest zone of the partially fenced treatment.................... 37 
Figure 5.9: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the riparian zone of the fully fenced treatment. .................... 38 
Figure 5.10: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the sodic zone of the fully fenced treatment. ........................ 38 
Figure 5.11: Top three dominant woody families (expressed as abundance in %) and 
species per fire treatment in 2015 for both seedling and established 
communities in the crest zone of the fully fenced treatment. ........................ 39 
Figure 5.12: NMDS ordination scatter plot to visually display the distribution of seedling (a) 
and established (b) woody species composition in 2002 (blue) and 2015 
(green) in the fully fenced exclosure. Different symbols represent 
different positions along a granitic toposequence (i.e. different vegetation 
zones). ......................................................................................................... 50 
Figure 5.13: NMDS ordination scatter plot to visually display the distribution of seedling (a) 
and established (b) woody species composition in 2002 (blue) and 2015 
(green) in the partially fenced exclosure. Different symbols represent 
different positions along a granitic toposequence (i.e. different vegetation 
zones). ......................................................................................................... 51 
 
 
xi 
Figure 5.14: NMDS ordination scatter plot to visually display the distribution of 2002 (a) 
and 2015 (b) woody species composition for the seedling (blue) and 
established (green) communities in the fully fenced exclosure. Different 
symbols represent different positions along a granitic toposequence (i.e. 
different vegetation zones). .......................................................................... 52 
Figure 5.15: NMDS ordination scatter plot to visually display the distribution of 2002 (a) 
and 2015 (b) woody species composition for the seedling (blue) and 
established (green) communities in the partially fenced exclosure. 
Different symbols represent different positions along a granitic 
toposequence (i.e. different vegetation zones). ............................................ 53 
Figure 7.1: Size-class distributions (plant canopy height in 1 m intervals) for mean number 
of woody individuals per plot (±SE) in successive size-classes (a) and 
quotients between successive size classes (b) for the complete woody 
community of 2015. ...................................................................................... 74 
Figure 7.2: Size-class distributions (plant height in 1 m intervals) for mean number of 
woody individuals per plot (±SE) in successive size-classes (a) and 
quotients between size-classes (b) across herbivory treatments in 2015. 
Control: all herbivores were present; Partial: elephant & giraffe were 
excluded; Full: all herbivores were excluded. Significant differences from 
ANOVA statistics are indicated with *. .......................................................... 76 
Figure 7.3: Size-class distributions (plant height in 1 m intervals) for mean number of 
woody individuals per plot (±SE) in successive size-classes (a) and 
quotients between size-classes (b) across fire treatments in 2015. 
Control: all herbivores were present; Partial: elephant & giraffe were 
excluded; Full: all herbivores were excluded. Significant differences from 
ANOVA statistics are indicated with *. .......................................................... 77 
Figure 7.4: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Gymnosporia senegalensis, Spirostachys africana, and Flueggea virosa 
in successive size-classes (a) and quotients for these individuals 
between size-classes (b) across herbivory treatments. Control: all 
herbivores present; Partial: elephant and giraffe excluded; Full: all 
herbivores excluded. Total/Overall: indicate SCD/Quotients of 
population, without distinguishing between herbivore treatments. ................ 80 
xii 
Figure 7.5: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Diospyros mespiliformis and Ziziphus mucronata in successive size-
classes (a) and quotients for these individuals between size-classes (b) 
across herbivory treatments in 2015. Control: all herbivores present; 
Partial: elephant and giraffe excluded; Full: all herbivores excluded. 
Total/Overall: indicate SCD/Quotients of population, without 
distinguishing between herbivore treatments. ............................................... 81 
Figure 7.6: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Vachellia grandicornuta and Rhigozum zambesiacum in successive 
size-classes (a) and quotients for these individuals between size-classes 
(b) across herbivory treatments in 2015. Control: all herbivores present; 
Partial: elephant and giraffe excluded; Full: all herbivores excluded. 
Total/Overall: indicate SCD/Quotients of population, without 
distinguishing between herbivore treatments. ............................................... 83 
Figure 7.7: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Pappea capensis in successive size-classes (a) and quotients for these 
individuals between size-classes (b) across herbivory treatments in 
2015. Control: all herbivores present; Partial: elephant and giraffe 
excluded; Full: all herbivores were excluded. Total/Overall: indicate 
SCD/Quotients of population, without distinguishing between herbivore 
treatments. ................................................................................................... 83 
Figure 7.8: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Dichrostachys cinerea and Combretum apiculatum in successive size-
classes (a) and quotients for these individuals between size-classes (b) 
across herbivory treatments in 2015. Control: all herbivores were 
present; Partial: elephant and giraffe excluded; Full: all herbivores were 
excluded. Total/Overall: indicate SCD/Quotients of population, without 
distinguishing between herbivore treatments. ............................................... 86 
Figure 7.9: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Senegalia nigrescens and Vachellia exuvialis in successive size-classes 
(a) and quotients for these individuals between size-classes (b) across 
herbivory treatments in 2015. Control: all herbivores present; Partial: 
elephant and giraffe excluded; Full: all herbivores were excluded. 
Total/Overall: indicate SCD/Quotients of population, without 
distinguishing between herbivore treatments. ............................................... 87 
1 
 
CHAPTER 1 
Introduction 
1.1 Background and rationale 
Savanna vegetation structure is characterized by a continuous herbaceous layer and 
discontinuous woody layer (Knoop & Walker, 1985). The complex co-existence between 
these two layers has been of interest for savanna ecologists over many years (Walter, 1971; 
Scholes and Archer, 1997; Higgins et al., 2000; Bond et al., 2003; Scholes et al., 2003; 
Sankaran et al., 2004). Savannas are heterogenous systems (Ben Shahar, 1996) that 
function at alternating stable states (Illius & O’Connor, 1999; Briske et al., 2003; Gillson, 
2004; Sankaran et al., 2005) in response to various factors controlling the transition from one 
stable state to another. These factors include climate variability, geological substrate, 
herbivory and fire (Skarpe, 1991; Van de Koppel & Prins, 1998; Van Wilgen et al., 2000). 
These factors act together to shape savanna ecosystems, depending on the specific 
characteristics of the area.  
The Kruger National Park (KNP) is one of the largest protected areas in southern Africa 
shaped by interconnectivity between climate, soil, fire and herbivory. The park hosts a great 
diversity of large mammalian herbivores and is subjected to fire events. In 2001, long-term 
research exclosures were constructed along two of the large rivers in KNP, the Letaba and 
Sabie Rivers respectively, which provided the opportunity to study temporal and spatial 
responses of savanna vegetation structure to herbivory and fire exsposure, as well as the 
effect of their exclusion from the system (O’Keefe & Alard, 2002). Research exclosures in 
the southern part of the KNP along the Sabie River, expand across a small-scale 
heterogenous savanna landscape (Siebert & Eckhardt, 2008; Siebert et al., 2010), typified 
by different soil types along a riparian topographic sequence (often referred to as the 
catena), which hosts a unique vegetation structure. These exclosures, i.e. the Nkuhlu 
exclosures site, provide unique opportunities to study vegetation dynamics in response to 
herbivory and fire across a typical granitic landscape catena. 
Woody encroachment are considered a concern to savanna systems due to its negative 
effects on the system. These negative effects include disruption in ecosystem functioning 
through altering soil carbon storage (Berthrong et al., 2012), inhibiting ground water 
recharge (Gray & Bond, 2013), affecting tourism (Gray & Bond, 2013), lowering grazing 
potential (Angassa & Baars, 2000) and affecting biodiversity (Ratajczak et al., 2012), 
therefore research on this topic is important. Woody encroachment is not only a concern for 
2 
 
private landowners but also for conservationists, which has led to extensive research on 
potential drivers of woody encroahment and suitable management practices for 
conservationists within southern African savannas. Woody encroachment have been 
identified as a consequence of the tree-grass ‘balance’ being disturbed (O’Connor et al., 
2014), with the disturbance leading to an increase in woody biomass, stem densities or 
woody cover in an ecosystem (Stevens et al., 2016). Savannas are particularly vulnerable to 
woody encroachment (Parr et al., 2014), and changes in the factors controlling savanna 
vegetation structure, i.e. climate, soils, herbivory and fire, are responsible for woody 
encroachment (O’Connor et al., 2014). Recent research suggested that increasing 
atmospheric CO2 levels should also be considered as a primary determinant and global 
driver of woody encroachment (Bond & Midgley, 2012).  
Savanna woody structure (i.e. composition, abundance and height class distribution) is 
primarily determined by the ability of woody species to regenerate, despite interactions with 
factors controlling savanna systems. Regeneration is therefore defined here as the process 
through which one (established) generation of woody plants is replaced by the following 
generation (Harper, 1979). During the regeneration process, woody seedlings grow past the 
sapling phase (juveniles) to get established (adult). This is referred to as woody recruitment 
(Harper, 1979). Savanna woody recruitment is best studied when all life stages are 
considered, in other words from seedlings to established trees, to identify specific 
demographic stage alterations which ultimately determine future woody structure.  
Regeneration responses of woody species to the most common savanna ecosystem drivers, 
i.e. herbivory and fire in combination with different soil types, can be examined by using a 
combination of assumptions and mechanisms from the competition-based and demographic 
bottleneck models (February & Higgins, 2010; O’Connor et al., 2014). These models are 
commonly used to describe the tree-grass co-existence in savanna ecosystems (O’Connor 
et al., 2014). The competition-based models refer to competition for nutrients, water and 
space between woody and herbaceous species (Walker & Noy-Meir, 1982; Walter, 1971). 
The demographic bottleneck models consider the variability of climate and disturbance, i.e. 
herbivores and fire, on growth and recruitment (Higgings et al., 2000). When combining 
these two models (Sankaran et al., 2004), one can predict that the herbaceous layer can 
directly or indirectly create demographic bottlenecks for woody plants, by (i) outcompeting 
woody seedlings, ‘trapping’ them within a high herbaceous biomass, which leads to (ii) high 
intensity fires that inhibit woody seedlings to grow past the ‘fire trap’. Another factor which 
can be combined with the demographic bottleneck model is the browse trap. This is where 
3 
 
woody plants are directly severely utilized by browsing herbivores, inhibiting them to grow 
past a certain height, keeping them within the ‘browse trap’.  
By applying these savanna models to the woody layer across the Nkuhlu exclosures, insight 
could be gained on the specific effect of presence or absence of herbivores (especially 
mega-herbivores) and fire on regeneration of woody species in a semi-arid savanna 
ecosystem. Novel insights will include the specific effects on encroachers and non-
encroachers.  
Woody structure can be altered through ‘trapped’ woody individuals inside a demographic 
bottleneck. These changes can occur as a result of alteration in woody abundances (Higgins 
et al., 2000; Staver et al., 2009; Gandiwa et al., 2012; Sankaran et al., 2013) and size-class 
distribution (Condit et al., 1998; Riginos, 2009; Kambatuku et al., 2011; Holdo et al., 2014; 
Vadigi & Ward, 2014).  
Woody structure is therefore regulated by numerous environmental factors, although it is still 
unclear how these factors affect woody species with different ecological adaptations, e.g. 
encroacher species versus more desirable non-encroaching species. When discerning 
between these two functional groups and disentangling effects of different herbivores (meso- 
and mega-herbivores) and fire on these species, insight could be gained on woody 
encroachment in savanna systems. The demographic models has not yet been used to 
examine responses of encroachers and non-encroachers to herbivore and fire 
manipulations. These results could be used to aid management practices on controlling 
encroacher species without negatively affecting the desirable non-encroaching species.  
 
1.2 Objectives 
The main objective of this study was to test the effect of herbivory and fire (presence and 
absence thereof) on woody regeneration (recruitment and rejuvenation) of both encroaching 
and non-encroaching woody species across a small-scale heterogeneous landscape along a 
riparian semi-arid savanna ecosystem. 
The specific objectives were to: 
 Describe and compare frequency of dominant woody families (and species) and 
ratio’s of basic plant functional types of the woody layer across different treatments of 
herbivory and fire (presence and release thereof) for seedlings and established 
woody individuals between 2002 and 2015 (Chapter 5). 
4 
 
 Examine and evaluate the effect of herbivore exclusion on woody species 
composition for both seedlings and established species between 2002 and 2015 
(Chapter 5). 
 Examine and evaluate the effect of herbivory and fire presence and absence on 
woody seedling and established woody individual abundances between 2002 and 
2015 (Chapter 6). 
 Examine and evaluate the effect of herbivore and fire presence and absence on 
woody community and population demography and stability (Chapter 7). 
 
1.3 Hypothesis 
The exclusion of both herbivory and fire from a semi-arid savanna ecosystem will enhance 
woody recruitment. 
 
1.4 Dissertation layout 
The layout of this dissertation conforms to the guidelines set for a standard dissertation by 
the North West University. However, the format was slightly adjusted in the respective 
results chapters to facilitate later manuscript preparations. Each results chapter therefore 
includes a short introduction, a description of methods specifically relevant to the research 
question addressed in the chapter, a description of the results and a discussion. A total of 
eight chapters are included, with a single reference list for all cited research at the end of the 
dissertation. A brief description of each chapter is provided below. 
Chapter 2: Literature review 
Literature relevant to the research topic is provided in this chapter. This includes a brief 
background on the Savanna Biome and short discussions on drivers of savanna structure 
and regeneration, as well as potential threats with special emphasis on bush encroachment.  
Chapter 3: Study area 
The study area, i.e. the Nkuhlu long-term research exclosures site, is described in terms of 
its locality, history, climate, geology, soil, topography, and vegetation.  
 
5 
 
Chapter 4: Materials and Methods 
The experimental design and sampling approach is explained in this chapter. An overview of 
statistical analyses is provided, although analyses relevant to specific objectives are 
discussed in detail in the respective chapters. 
Chapter 5: Floristic changes in response to herbivory and fire 
This chapter serves as an introduction to the floristics of the study area. Floristic responses 
to the different herbivore and fire treatments are presented based upon dominant plant 
families, species and basic plant functional types. Both seedlings and established woody 
individuals were considered in analyses. The effects of herbivore exclusion on woody 
species composition were also evaluated. This was done to compare effects of mega-
herbivore (elephant and giraffe) exclusion opposed to the exclusion of all mammalian 
herbivores. 
Chapter 6: Woody abundance changes across herbivore and fire treatments 
Changes in abundances of seedlings and established individuals of the (i) complete, (ii) 
encroacher, and (iii) non-encroacher woody communities were tested and evaluated across 
the different herbivore and fire treatments.  
Chapter 7: Demography of the woody community after 13 years of herbivore and fire 
manipulation 
In this chapter, the demography and stability of woody communities and pre-selected key 
encroacher and non-encroacher species were studied separately to evaluate the woody 
community structure after 15 years of herbivore and fire manipulations.  
Chapter 8: Conclusions 
The main findings are discussed and linked to the hypotheses tested (regarding each results 
chapter). Recommendations for future studies on regeneration of the woody layer in semi-
arid savanna ecosystems are provided. 
  
6 
 
CHAPTER 2 
Literature review 
2.1 Savanna vegetation structure and dynamics 
Savannas are widely defined as tropical or near-tropical seasonal ecosystems consisting of 
a continuous herbaceous layer, dominated by grass, and a discontinuous layer of trees 
and/or shrubs (Frost et al., 1986; Skarpe, 1992; Mucina & Rutherford, 2006; Van As et al., 
2012). Savanna ecosystems are dynamic in their vegetation structure and species 
composition, which change in response to biotic disturbances and various interactions 
between organisms and their abiotic environment (Skarpe, 1991; Van de Koppel & Prins, 
1998; Van Wilgen et al., 2000). On the African continent, many different types of savannas 
exist. These include edaphic savannas, climatic savannas and disturbance-based savannas. 
Different formations of savannas ranging from shrubby grasslands to open woodlands exists 
with the primary drivers of vegetation structure differing between them. The savanna type 
under investigation in this dissertation is a disturbance-based savanna driven primarily by 
disturbances such as herbivory and fire as well as soil-characteristics. 
 
2.1.1 Plant-plant interactions 
In savanna ecosystems, the most apparent plant-plant interaction is between trees and 
herbaceous plants (Scholes & Archer, 1997; Sankaran et al., 2004). The dominance of the 
one life form over the other is largely dependent upon geographical location, but also on 
specific environmental conditions, such as rainfall, soil type, soil nutrients, atmospheric CO2 
levels and frequency of herbivory and fire (Van As et al., 2012).  
Tree-grass co-existence in savanna ecosystems is regarded as unstable. Four models were 
summarized by House et al. (2003) to explain how these two life forms co-exist: (1) Niche 
separation, where woody and herbaceous plants occupy different regions in space (i.e. 
preferential access to deep soil water by woody plants versus shallow soil water use by 
grass); (2) Balanced composition, where intraspecific competition dominates over 
interspecific competition; (3) Competitive exclusion, where the system is eventually driven to 
a relative stable state and where one life form, for a short time, dominates and virtually 
eliminate the other until disturbances with greater effect prevent maintenance of this 
domination; (4) Multiple stable states, where the spatial and temporal heterogeneity of 
7 
 
resource availability and disturbance is incorporated into equilibrium models so that 
contrasting tree-grass ratios might exist for a given site at various times. 
New explanations arose during recent years to explain the co-existance of trees and 
grasses. These explanations recognised the effect of disturbances, i.e. herbivory, fire and 
climatic variability, on the dynamic nature of tree-grass coexistence (Menaut et al., 1990; 
Jeltch et al., 1996, 2000; Higgins et al., 2000; Gardner, 2006; Van Langevelde et al., 2011). 
These disturbances impact on the main or ‘superior’ life form at the given time, which create 
opportunities for the competitive ‘inferior’ life form to establish and persist (Warner & 
Chesson, 1985). The disturbance-based explanation of tree-grass co-existence is underlain 
by two different arguments for the dynamics (Sankaran et al., 2004), which include (i) non-
equilibrium dynamics (Higgins et al., 2000; Gardner, 2006), and (ii) disequilibrium dynamics 
(Menaut et al., 1990; Jeltsch et al., 2000). These two paradigms consider that savannas are 
continually in transition with disturbances and temporal resource availability maintaining the 
transitional state (Van Langevelde et al., 2011). The disequilibrium argument states that a 
system is prevented to reach an equilibrium state due to effects of disturbances, whereas 
the non-equilibrium argument ignores the existence of any equilibrium positions (Van 
Langevelde et al., 2011). The non-equilibrium dynamics model only considers the effect of 
disturbances on tree recruitment, stating that inter- and intra-life form competition do exist 
but is assumed to have a minor effect on tree:grass ratios compared to climatic variability 
(Van Langevelde et al., 2011). The disequilibrium dynamics assume that competition 
between trees and grasses occurs at all life stages in the particular environmental 
conditions, and internal drivers (disturbances) such as herbivory and fire prevent dominant 
vegetation structure and composition transitions (Van Langevelde et al., 2011).  
Scheiter & Higgins (2007) proposed a model that explains tree-grass co-existence by 
identifying above- and belowground competitive factors. These factors take in consideration 
the consequences of competition for multiple resources. Aboveground competition exists for 
light and belowground for water, rooting space and nutrients (Scheiter & Higgins, 2007). 
Herbivory and fire were identified as the main attributes to affect aboveground biomass. It 
was stated that tree-grass coexistence could only be achieved if belowground competition 
(roots) are held below a certain threshold and aboveground competition for light is low (Van 
Langevelde et al., 2011).  
High herbaceous biomass can suppress or inhibit tree establishment, but evidence exists 
that tree establishment can be facilitated by, for example, other trees (Holmgren et al., 1997; 
Ludwig et al., 2004) or even grasses (Brown & Archer, 1989; Davis et al., 1998; Anthelme & 
8 
 
Michalet, 2009; Van Langevelde et al., 2011). This facilitation effect is especially true for 
stressful environments (Brooker et al., 2008).  
 
2.1.2 Plant-herbivore interactions 
African savannas are known for its rich diversity of large mammalian herbivores. Savanna 
vegetation has co-evolved with these animals, which led to certain adaptations from both 
sides (Charles-Domique et al., 2016). This includes the functional relationship between the 
development of spinescent trees in the presence of herbivores, especially of medium-sized 
social mixed-feeders (Charles-Domique et al., 2016). Herbivory was responsible for the 
formation of structural defences in multiple woody lineages (Charles-Domique et al., 2016).  
More adaptations in woody plants to repel or prevent herbivory include (Van As et al., 2012): 
(i) formation of leave tannins (Scogings et al., 2011), (ii) leaf shedding during dry seasons 
(Owen, 1978), and (iii) deep root system development to prevent uprooting (Du Toit et al., 
2014). 
Interactions between trees and herbivores have led to niche diversification between the two 
groups. Several processes have been identified that could be involved in this niche 
diversification: (i) separation in height niches, where herbivores have different feeding 
strategies depending on their body size (Wilson & Kerley, 2003) and vertical separation in 
spines on plants match the herbivore body sizes present (Burns, 2014); (ii) niche 
specialisation, mammals likely to browse more have narrow muscles, longer tongues, and 
prehensile lips, allowing them to browse spiny plants better than grazers (Shipley, 2007); (iii) 
segregation of niches in time for trees, with impacts from browsers varying in season and 
between deciduous and evergreen trees (Bryant et al., 1992; Massei et al., 2000). This 
niche-diversification across evolutionary time also contributed to the diverse nature of 
African savannas (Skarpe, 1992). 
African savannas have a long history of large herbivores shaping plant communities and 
maintaining landscape and species diversity (Guldemond & van Aarde, 2008; Levick & 
Rogers, 2008; O’Kane et al., 2011; Scogings et al., 2012; Van Coller et al., 2013). For 
example, trees are attractive structural elements for a variety of small and large mammals, 
adaptation in seed dispersal and scarification, and animals are potential drivers of the spatial 
patterning of woody species. Increasing megaherbivore densities, particularly elephant may 
result in loss of woody cover, species richness and healthy regeneration (Whyte et al., 2003; 
Western & Maitumo, 2004; Guldemond & Van Aarde, 2008; O’Connor, 2010; Asner & 
9 
 
Levick, 2012; Gaugris et al., 2014; De Boer et al., 2015). Significant negative elephant 
impacts include damage to large trees by debarking them, pushing the entire tree over, or by 
tearing of leaves and branches (Scholes et al., 2003). Other smaller bodied herbivores also 
have substantial effects in changing savanna vegetation structure (Moe et al., 2009; O’Kane 
et al., 2012). This is done through direct consumption of grass and woody species, 
especially seedlings, or indirectly through trampling. 
 
2.1.3 Plant-fire interactions 
Fire-maintained savannas first appeared in the late Miocene, millions of years after 
mammals dominated savannas (Charles-Domique et al., 2016). These dyostrophic fire-
dominated savannas are characterised by seasonally humid and nutrient-poor environments 
(Lehmann et al., 2011; Charles-Domique et al., 2016). Fire has been considered as an 
ecosystem management tool for bush encroachment, especially in savanna conservation 
areas. The woody component can directly (direct consumption of woody biomass) or 
indirectly (clearance of herbaceous biomass to create establishment gaps for woody 
seedlings) be influenced by fire (Bond & Keeley, 2005; O’Connor et al., 2014). The effect of 
fire on savanna woody plants is often species-specific (Bond et al., 2001; Zida et al., 2007). 
This was observed in studies that reported that Sclerocarya birrea individuals with a height 
less than 2m were negatively affected by fire (Jacobs & Biggs, 2001), whereas individuals of 
Dichrostachys cinerea and Vachellia gerrardi were negatively affected in the height classes 
less than 3m and those with a height greater than 3m were unaffected (Jordaan, 1995).  
Trees have developed mechanisms to tolerate fire disturbances, for example, resprouting 
(López-Soria & Castell, 1992) by using stored reserves. Resprouting involves further gowing 
after fire has caused damage to the tree. This mechanism enables trees to be resilient to fire 
disturbances. The ability of trees to reprout after a fire disturbance depends largely upon tree 
size (height). If a tree is unable to reach above a certain height before a fire disturbance, it 
have a higher chance to get killed.  
A common feature whithin fire exposed savannas is the ‘fire-trap’, which creates a 
recruitment bottleneck for woody species (Higgins et al., 2000). The ‘fire-trap’ inhibits woody 
seedlings/saplings to grow above a certain height (3 m). This may lead to a decline in woody 
cover over long-term periods. 
 
10 
 
2.1.4 Plant-soil interactions 
Soil characteristics whithin a savanna can alter vegetation composition (Scholes & Walker, 
1993; Levick et al., 2010; Colgan et al., 2012). The Kruger National Park consists of a 
heterogenous landscape due to its geology. The granite/gneiss landscape in the southern 
parts of the park (at the Nkuhlu exclosure site) resides in rapid transitions between 
dystrophic and eutrophic environments, creating distinct vegetation zones. These zones are 
situated on the catena which initiate close to the Sabie River (riparian) and extend across 
the midslopes (sodic) to the crest. Rainfall commonly interacts with changes in topography to 
influence soil development. Clay particles are carried downslope, leaving the crest with 
sandy, nutrient-poor and well-drained soils, and the low-lying areas (riparian) with moist, 
nutrient-rich clayey soils (Scholes & Walker, 1993; Khomo et al., 2011). In the riparian 
bottomlands, a low percentage of rock cover is found (<5%) and the vegetation structure is 
characterised as moderately closed to closed woodland (Siebert & Eckhardt, 2008). The 
mid-slopes (sodics) are also characterised with low percentage rock cover (5%) but with 
moderately open vegetation structure (Siebert & Eckhardt, 2008). On the crest, a higher 
percentage rock cover (5-10%) is found and the vegetation structure is a moderately closed 
shrubland/savanna (Siebert & Eckhardt, 2008). These changes in soil characteristics and 
rock cover across the catena influence vegetation composition. The catena can therefore be 
identified as one of the major determinants of savanna vegetation patterns whithin a 
landscape (Milne, 1947; Morison et al., 1948; Fraser et al., 1987; Baldeck et al., 2014). 
 
2.2 Regeneration of woody species 
Regeneration can be defined as the process through which one generation of plants are 
replaced by a following generation (Harper, 1979). Through this process, various interactions 
may influence the recruitment process. Recruitment is the developing process of a woody 
plant from a seedling, past the juvenile stage, to form an adult tree (Harper, 1979).  
The seedling stage of woody plants can be considered as the most vulnerable life stage and 
therefore also the most important stage (Hoffmann, 1999; Higgens et al., 2000; Roques et 
al., 2001). If the seedling can survive in its immediate environment despite various biotic and 
abiotic influences, the plant will be able to grow and get established in that specific area. In 
theory this is a simple concept to understand, but in nature, this process is rather complex 
due to complex interactions between numerous environmental factors.  
11 
 
Among the various factors that affect regeneration of woody plants in savanna ecosystems 
are herbaceous competition (especially grass), herbivory, and fire. The herbaceous layer is 
known to have competitive effects on woody regeneration specifically on the seedling stage 
(Riginos, 2009; Vadigi & Ward, 2014). This is where competition for nutrients, water and 
space plays an important role in woody establishment (Riginos, 2009; Kambatuku et al., 
2011; Vadigi & Ward, 2014).  
Herbivory through grazing and browsing may affect woody recruitment either directly through 
consumption of woody plants, or indirectly by relieving grass competition and suppressing 
smaller herbivore effects, for instance by rodents and insects. In their study, Goheen et al. 
(2004) found evidence of large mammalian herbivores to facilitate and enhance seedling 
establishment by suppressing smaller herbivores such as rodents, insects, etc. Grazing 
herbivores may also reduce competition between woody species and the herbaceous layer 
by creating space for woody establishment and growth (Kambatuku et al., 2011). Herbivory 
may also have negative impacts on woody establishment by direct tree utilisation (Moe et al., 
2009). Elephants have received considerable attention in African savanna ecosystems due 
to their apparent destructive feeding behaviour. Limited knowledge exist of their effects on 
woody seedlings. It has been suggested that smaller bodied mammals, such as impala could 
also be responsible for changes in woody structure (Belsky, 1981; Sharam et al., 2006; Moe 
et al., 2009; O’Kane et al., 2012). Moe et al. (2009) and O’Kane et al. (2012) found a positive 
relationship between increasing densities in impala and larger impacts in terms of increasing 
mortalities of woody recruits. Lagendijk et al. (2015) suggest that elephants can possibly 
enhance woody regeneration through displacement of meso-herbivore activity. 
Fire is considered a fundamental determinant of savanna vegetation structure through its 
impact on regeneration. In the seedling stage, woody individuals are most vulnerable to 
damage from fire (Harrington et al., 1991; Casillo et al., 2012; Joubert et al., 2012; Taylor et 
al., 2012) due to their thin bark, lack of carbon storage (Casillo et al., 2012; Joubert et al., 
2012; Midgley, 2010), and exposure to high temperatures in the fire zone fuelled by the dry 
grass biomass (Miranda et al., 1993). This imposes disruption in the recruitment process of 
these individuals (Harrington et al., 1991; Casillo et al., 2012; Joubert et al., 2012; Taylor et 
al., 2012). This is one of the reasons why fire has been considered a tool for managing 
woody encroachment (Lohmann et al., 2014). By killing woody individuals in their seedling 
stage, less woody species will be able to grow into mature adult trees (Zida et al., 2007).  
 
12 
 
2.2.1 Models that explain regeneration of woody species 
Sankaran et al., (2004) suggested that the tree-grass co-existence may be explained by 
combining the competition-based and demographic bottleneck models. Therefore, the 
dominance of the ‘strongest’ life form is explained through a close interaction between 
competitive exclusion and top-down controls, such as herbivory and fire, which cause a 
demographic bottleneck for woody recruits through suppressive effects on seedling growth 
(Riginos, 2009; Vadigi & Ward; 2014). 
Walter (1971) introduced the two-layer hypothesis in the co-existence of the grass and 
woody layer in savanna ecosystems. This hypothesis states that woody and grass species 
occupy different soil layers, with roots growing much deeper (Walter, 1971). More recent 
studies contradicted this hypothesis (Riginos, 2009; February & Higgins, 2010; Tedder et al., 
2012, 2014; Wakeling et al., 2015) with a new concept that although woody roots can grow 
deeper than roots of grass, they still directly compete with each other for nutrients (Cramer 
et al., 2007, 2009, 2012; Bloor et al., 2008), water (Gordon et al., 1989; Weltzin & 
McPherson, 1997; Davis et al., 1999; Picon-Cochard et al., 2006) and space 
(McConnaughty & Bazzaz, 1991, 1992; Casper & Jackson, 1997) in the upper soil layer. 
This is however dependent on the involved species (both grasses and woodies), life history 
stages, soil properties and climatic conditions (Kambatuku et al., 2013; Ward et al., 2013) 
The grass layer commonly outcompetes woody seedlings, which inhibits their establishment 
and imposes a demographic bottleneck in their recruitment (Bond, 2008; February et al., 
2013; Wakeling et al., 2015). 
In grass-dominated ecosystems, woody individuals (less than 3 m in height) that fail to 
escape the fuel-zone are commonly believed to either be killed by hot burns, especially 
seedlings, or suppressed in terms of growth. This phenomenon is referred to as the ‘fire-trap’ 
which may lead to a recruitment bottleneck (Higgins et al., 2000; Bond et al., 2005). This 
means that fire disturbance is stage-structured by affecting saplings and not trees (Higgins 
et al., 2000; Hanan et al., 2008; Staver & Levin, 2012; Staver & Bond, 2014). 
Analogous to that of a fire-trap is the browse trap created by chronic herbivory (Fornara & 
Du Toit, 2008; Staver et al., 2009). To date, the understanding of herbivory effects on 
savanna systems has been vague (Bond & Keeley, 2005; Staver & Bond, 2014) although 
herbivory effects are in many respects similar to that of fire. The recruitment process of 
woody plants are suppressed by constant browsing at a specific height, and only when the 
pressure is reduced, the juvenile trees can grow into large trees (Prins & Van der Jeugd, 
1993; Holdo et al., 2009; Staver et al., 2009). The juvenile trees can therefore be referred to 
13 
 
as being in a browse-trap. These two disturbance traps also differ in some respects. The 
fire-trap is considered an episodic disturbance, while herbivory is more continuous. It is 
rather the release from herbivory that is considered as an episodic event (Young, 1994; 
Holdo et al., 2009). The browse-trap is also considered to be species specific, suggesting 
that woody species may be trapped in different size-classes depending on their forage 
prefences by different herbivores (Wigley et al., 2014). 
 
2.2.2 Savanna woody floristic changes driven by herbivory and fire 
Savannas first began to spread across Africa in the Miocene, with herbivory and fire 
regarded as determinants of its spread. Mammal-dominated savannas predate fire-
dominated savannas by millions of years (Charles-Dominique et al., 2016). These two 
determinants of savanna structure seem to favour contrasting savanna environmental 
settings: herbivore-dominated (eutrophic) savannas are associated with nutrient-rich 
environments and fire-dominated (dystrophic) savannas with nutrient-poor environments 
(Scholes, 1990; Lehmann et al., 2011; Hempson et al., 2015). The evolutionary history of 
African savannas suggests that savanna vegetation is to some extend adapted to the 
presence of both fire and large herbivores and that the release of these factors from 
savanna ecosystems could lead to environmental changes. Herbivores do not only shape 
vegetation structure, but may also be considered as an important driver of species 
composition changes. It is suggested that palatable species are more likely to get browsed 
upon than unpalatable species. Regeneration of unpalatable woody species is therefore 
expected to be stronger than palatable species in the the presence of herbivores, depending 
on the frequency and intensity of browsing pressure (Fornara & Du Toit, 2008). Herbivore 
exclusion could also cause changes in species composition through the removal of selective 
browsing pressure (Wiseman et al., 2004; Fornara & Du Toit, 2008; Barton & Hanley, 2013; 
Gauris et al., 2014), and therefore benefit recruitment of palatable woody species.  
 
2.2.3 Savanna woody structure changes driven by herbivory, fire and the 
herbaceous layer 
Studies on variability of woody abundances in savanna ecosystems revealed that 
demographic bottlenecks contribute to explaining patterns observed in vegetation structure. 
Staver et al. (2009) found that reduction in herbivore activity led to the escape of trapped 
seedlings (in the browse-trap), which resulted in an increase of adult tree abundances. This 
14 
 
was also observed by Gandiwa et al. (2012) who stated that it is fundamental for woody 
seedlings to escape fire- or browse-traps to increase in relative abundance. This 
phenomenon was also reported by Wakeling et al. (2011) and Hartnett et al. (2012). 
Increase in woody abundances after herbivore exclusion further demonstrated the ecological 
effects of browse-induced demographic bottlenecks (Sankaran et al., 2013).  
Woody abundance responses can also be species or functional trait specific. In their study, 
Fornara & Du Toit (2008) found that selective browsing reduced local abundances of 
palatable woody species, while unpalatable woody species abundances increased and 
eventually dominated. Species such as Senegalia nigrescens are adapted to browsing and 
their abundances did not decrease in heavily browsed sites in the study of Fornara & Du Toit 
(2008). Fire disturbances responsible for creating demographic bottlenecks in the form of a 
fire-trap are known to especially affect woody seedlings in savannas (Higgins et al., 2000; 
Higgins et al., 2007). Trapped seedlings cause a decrease in woody cover (Bond et al., 
2005; Sankaran et al., 2005) by preventing adult tree establishment (Hoffmann, 1999; 
Higgins et al., 2000) and therefore decreasing adult tree abundances.  
The release of herbivores and fire from a system causes an increase in herbaceous biomass 
(Van Coller et al., 2013). The herbaceous layer acting as a recruitment filter (Riginos, 2009; 
Vadigi & Ward, 2014) can therefore suppress seedling establishment, leading to changes in 
woody structure.  
Demographic bottlenecks can also alter size-class distribution of plant species. Size-class 
distribution, which is a measure of demography or population structure, is commonly used to 
evaluate regeneration success and population or community stability within a specific 
environment (Shackleton, 1993; Mwavu & Witkowski, 2009; Venter & Witkowski, 2010; 
Byakagaba et al., 2011; Shackleton et al., 2015). The distribution of woody size-classes 
should reveal an inverse J-shaped curve when graphically displayed. This curve indicates 
that most individuals are located in the first size-class and the successive size-classes are 
monotonically declining. Woody individuals can be divided into size-classes according to 
diameter at breast height (DBH) (Baker et al., 2005; Coomes & Allen, 2007; Wang et al., 
2009) or based on their crown height (Pellew, 1983; Holdo et al., 2014).  
Woody demography analyses highlight specific effects of environmental variables, such as 
herbivory and fire on size-class distribution. However, the seedling size-class is mostly 
affected by herbaceous competition (if herbaceous biomass is high), which inhibit woody 
growth past seedling stage (Eckhardt et al., 2000; Riginos, 2009; Kambatuku et al., 2011; 
Vadigi & Ward, 2014). Herbivory, by browsers, creating browse-traps, and fire-traps created 
15 
 
by fire exposure are identified to cause alteration in especially the sapling size-class 
distribution (1-3 m in height) of a woody population (Holdo et al., 2014; Wigley et al., 2014). 
These demographic bottlenecks ultimately cause variation to a normal inverse J-shaped 
population (Silwertown, 1992; Oliver & Larson, 1990; Condit et al., 1998; Wilson & 
Witkowski, 2003; Mwavu & Witkowski, 2009; Holdo et al., 2014). Demographic bottlenecks in 
size-class distribution analyses can be identified by a sharp drop in woody abundance 
between size-classes (Lykke, 1998), for instance where significantly more seedlings than 
saplings are recorded (Prior et al., 2009), or where size-classes are missing. This can be 
linked to the specific environmental conditions to determine what the causes for the 
bottlenecks are.  
 
2.3 Woody encroachment 
Encroachment by woody species result in an increase of woody biomass, stem densities or 
woody cover, which is a growing problem in savanna ecosystems (Roques et al., 2001; 
O’Connor et al., 2014; Parr et al., 2014; Stevens, et al., 2016) and is considered one of the 
top three apparent rangeland problems in South Africa (Hoffmann et al., 1999). Woody 
encroachment may have numerous negative effects on the environment. A combination of 
all these factors can accelerate woody encroachment. For example, an ecosystem subjected 
to high grazing levels will cause reduction in fire frequency and intensity, which will promote 
woody encroachment by releasing woody recruits from competition with the grass layer 
(O’Connor et al., 2014). The rising atmospheric CO2 concentrations will increase woody 
growth through improved water usage (Polley et al., 1997; Leaky et al., 2009) as well as 
increased rates of carbon uptake. This would extend the growing season and growth rates of 
woody species, which will increase woody cover (Bond & Midgley, 2000; Hoffmann et al., 
2000; Kgope et al., 2010; Stevens et al., 2016). Therefore, woody encroachment is a 
consequence of the tree-grass ‘balance’ in savannas being disturbed (O’Connor et al., 
2014). 
Encroachment is likely to occur in woody communities containing a high abundance of 
nitrogen-fixing woody plants, for example, woody species from the family Fabaceae 
(Stevens et al., 2016). Where these species dominate in combination with increased 
atmospheric CO2 concentrations and/or reduction in drought stress, the community is likely 
to display a rapid increase in woody biomass over time. This is because nitrogen-fixing 
species can match elevated photosynthesis rates by producing more nitrogen-fixing tissues 
16 
 
(Leakey et al., 2009; Rogers et al., 2009) which maintain high nitrogen-fixation rates (Rogers 
et al., 2009). 
Rates of woody encroachment have allegedly accelerated during recent years (Buitenwerf et 
al., 2012; O’Connor et al., 2014). Encroachment has for a long time been limited in relatively 
pristine, large natural ecosystems, such as the Kruger National Park. These areas provide 
habitat for a complex faunal diversity (meso- and mega-herbivores) and fire has been used 
as a management tool for most of the last 120 years (Govender, 2003) The removal of 
elephant from savanna ecosystems have been identified as a significant cause of woody 
encroachment in Africa (Guldemond & Van Aarde, 2008). This is because free-ranging 
elephants which controlled encroachment where they roamed, have been removed from 
many parts of Africa and are now only present in conservation areas (Owen-Smith, 1992). 
Therefore, the presence of elephant is suggested to control woody encroachment to some 
extent (Stevens et al., 2016) through limiting woody cover (Guldemond & Van Aarde, 2008).  
17 
 
CHAPTER 3 
Study area 
3.1 Locality 
The study was conducted at the Nkuhlu long-term research exclosures in the Kruger 
National Park, South Africa (Figure 3.1) which comprise 139 ha of semi-arid savanna. It is 
situated approximately 18 km south of Skukuza along the Sabie River (24°59’10”S 
31°46’24.6”E).  
 
Figure 3.1: Location of the Nkuhlu exclosures research site in the Kruger National Park, South 
Africa. 
18 
 
3.2 History 
The Nkuhlu exclosures were constructed shortly after the large flood event in 2000 to 
evaluate the natural restoration of a semi-arid savanna riparian ecosystem. In addittion to 
flood response, the research exclosures were designed to test the combined effects of 
herbivory and fire on vegetation structure, composition and diversity to provide research-
based evidence for the development or adjustment of management policies in the Kruger 
National Park (O’Keefe & Alard, 2002). The exclosures are intended to be monitored for at 
least 25 years (starting in 2002). By using this data, insight could be gained on key 
ecological processes that would otherwise be difficult to obtain (O’Keefe & Alard, 2002).  
General research objectives were provided in 2002 by O’Keefe & Alard (2002) for specific 
studies which needed to be done at the exclosures. The following research questions have 
been formulated: (i) how do herbivory and fire change the vegetation pattern? (ii) How do 
herbivory and fire affect the regeneration of vegetation following a major flooding event? (iii) 
How do herbivory and fire affect seed dispersal, seed germination, and then seedling 
survival? and (iv) What effect do animals have on the physical and biogeochemical features 
of the landscape?  
Throughout recent years (from 2002 until 2015) several vegetation studies have been 
conducted at the Nkuhlu exclosures, which include a vegetation and floristic description of 
the exclosures (Siebert & Eckhardt, 2008), responses of woody vegetation to exclusion of 
large herbivores (Scogings et al., 2012), stem growth of woody species (Scogings, 2011), 
secondary metabolites and nutrients of woody plants in relation to browsing intensity 
(Scogings et al., 2011), herbaceous species diversity patterns across herbivory and fire 
treatments (Van Coller et al., 2013; Van Coller & Siebert, 2015), deciduous sapling response 
to season and large herbivores (Scogings et al., 2013), landscape scale effects of herbivores 
on tree fall (Asner et al., 2012). No studies on woody plant species regeneration has been 
conducted at this particular study site. 
3.3 Herbivore community 
The area in and around the study site (exclosures) are rich in a diversity of herbivores. Most 
common species found in the area and surrounding areas are elephant (Loxodonta 
africana), impala (Aepyceros melampus), warthogs (Phacochoerus africanus), rhino 
(Ceratotherium simum), giraffe (Giraffa camelopardalis), kudu (Tragelaphus strepsiceros), 
bushbuck (Tragelaphus scriptus), hippo (Hippopotamus amphibious), buffalo (Syncerus 
19 
 
caffer), zebra (Equus quagga), and blue wildebeest (Connochaetes taurinus) (Reardon, 
2012). 
 
3.4 Climate 
The study area is characterised as a semi-arid savanna (Sankaran et al., 2005) with a mean 
annual precipitation average of 561 mm (as measured at Skukuza) 
(http://www.sanparks.org/parks/kruger/conservation/scientific/weather). Two distinct seasons 
characterise the area, which include a hot, wet, growing season from October to April 
(summer), and a cool to warm, dry period from June to August (winter) (Williams et al., 2009; 
Scogings et al., 2012) The Skukuza region is generally frost-free, with temperatures ranging 
from minimum means of ±6°C in June, July, and August to maximum mean temperatures of 
above 32°C in December, January and February (Gertenbach, 1983).  
 
3.5 Geology, soil and topography 
The study site is mainly underlain by Archean granite and gneiss (Gertenbach, 1983), which 
is characterised by a typical Lowveld savanna granitic toposequence with a unique 
vegetation composition at each catenal position (see 3.5). The site varies in altitude from 
210 m above sea level at the lowest catenal position (i.e. the riparian bottomlands) up to 235 
m on the upland crests (Siebert & Eckhardt, 2008). The soil is in general shallow with a few 
rock patches on the upland areas, but deeper in bottomland areas where there is an 
accumulation of clay and minerals (Gertenbach, 1983). On the mid-slopes (sodics), hyper-
accumulation of exchangeable sodium occurs as a result of reduced hydrolic conductivity 
(Khomo & Rogers, 2005). This is due to the release of sodium from granite as the parent 
geology (Khomo & Rogers, 2005). In 1972, Harmse & Van Wyk identified Mispah and 
Glenrosa as the main soil types. Other soil types located at the Nkuhlu exclosures across 
the landscape (from bottomlands to uplands) include Oakleaf in the riparian, mainly Montagu 
on the sodics and Glenrosa/Mispah and patches of Clovelly on the crest (Siebert & Eckhardt, 
2008). 
 
20 
 
3.6 Vegetation 
The study site falls within the Lowveld Bioregion and forms part of the Granite Lowveld 
vegetation type (SVI 3, Mucina & Rutherford, 2006). The vegetation of the surrounding area 
of the Nkuhlu exclosures is classified as ‘Thickets of the Sabie and Crocodile Rivers’ 
described by Gertenbach (1983) and the vegetation type as the Acacia (Senegalia) 
nigrescens - Combretum apiculatum association. The dominant woody species of the Sabie 
River granitic landscape include Combretum apiculatum, Grewia bicolor, G. flavescens, 
Dichrostachys cinerea, Euclea divinorum, Terminalia prunioides, Spirostachys africana, 
Vachellia (Acacia) grandicornuta and Senegalia (Acacia) nigrescens (Gertenbach, 1983).  
Siebert & Eckhardt (2008) described the vegetation of the Nkuhlu exclosures as a patchy 
mosaic, within a relatively small, spatially restricted, heterogeneous landscape. The 
exclosures consist of five main plant communities, which include ten sub-communities and 
few small-scale variations, all of which are linked to a specific catenal position (riparian, 
sodic and crest (Figure 3.2)) (Siebert & Eckhardt, 2008). Diagnostic woody species in the 
riparian zone include Flueggea virosa, Gymnosporia senegalensis, Peltophorum africanum, 
Euclea divinorum, E. natalensis, Combretum hereroense, C. imberbe, Phyllanthus 
reticulatus, Spirostachys africana, Kigelia africana and Bridelia cathartica (Siebert & 
Eckhardt, 2008). In the sodics Vachellia grandicornuta, Rhigozum zambesiacum, Pappea 
capensis, and Adenium multiflorum are the diagnostic, whereas Senegalia nigrescens, 
Dichrostachys cinerea, Combretum hereroense, C. apiculatum, C. zeyheri, Sclerocarya 
birrea and Vachellia exuvialis dominate the crest of this granitic landscape (Siebert & 
Eckhardt, 2008). 
21 
 
 
Figure 3.2: Aerial image (Google Maps, 2016) of the Nkuhlu exclosures study site overlain by 
the herbivore treatments, which illustrate the position of the broader vegetation zones 
(Riparian, Sodic and Crest) across the topographic sequence. More detail on the herbivore 
treatments will be provided in Chapter 4. 
  
22 
 
CHAPTER 4 
Materials and methods 
4.1 Experimental design 
The Nkuhlu long-term research exclosures are divided into different treatments of herbivory 
and fire (Figure 4.1). The herbivory treatments consist of the following: (1) fully fenced 
exclosure (designed to exclude all mammalian herbivores larger than a hare) (Figure 4.2a); 
(2) partial exclosure (designed to exclude elephant, but also exclude giraffe due to their body 
height (Figure, 4.2b); and (3) a control site, which allows for access by all large mammalian 
herbivores. To test the effect of fire on vegetation structure and diversity, each herbivore 
treatment site was subdivided into a burn and a no-burn block, separated by a fire break 
(O’Keefe & Alard, 2002). Fire as a treatment conforms to fire management of the area, which 
allows natural fires within the broader management unit to enter the fire treatments, but 
excluding them from the no-burn blocks. The last fire treatment was applied in 2012. 
Permanent transects were placed within each herbivore and fire treatment, and extend from 
the Sabie River riverbank across the granitic toposequence, i.e. from the alluvial bottomland 
(riparian zone) across the seepline (sodic zone) to the coarse sandy upland (crest zone). 
Along these transects, fixed 200 m2 plots were permanently marked with steel rods and 
positioned with the longest side (20 m) parallel to the river. These plots were used as 
replicates in the analysis of the data. 
23 
 
 
Figure 4.1: Graphic representation of the experimental design of the Nkuhlu long-term 
research exclosures (adapted from Van Coller et al., 2013). 
 
 
Figure 4.2: The fully fenced (a; right side of the fence is inside the exclosure) and partially 
fenced (b; right side of the fence is in the exclosure) exclosures which represent the herbivore 
treatments at the Nkuhlu exclosures 
 
 
24 
 
4.2 Data sampling 
Data sampling was done according to standardized protocol designed by O’Keefe & Alard 
(2002). First data were sampled by SANParks in 2002 and in 2015 a group from the North 
West University, Potchefstroom sampled the data.  
Two brightly coloured 30 m ropes were extended around the corners of the fixed 200 m2 
plots to clearly mark the sampling plot. All woody individuals of size >10 cm in height and 
rooted inside a plot were identified up to species level, counted and various measurements 
were recorded per individual (see below for details). Where it was not clear what constitutes 
an individual inside a clump of vegetation with multiple stems or whether stems are a result 
of vegetative reproduction, stems further than 0.5 m apart were treated as different 
individuals (O’Keefe & Alard, 2002).  
For each woody individual canopy height was measured using a measuring pole marked at 
0.5 m intervals. Basal diameter of each individual was measured using a calliper (for stems 
<10 cm in diameter) or a measuring tape (for stems >10 cm in diameter). Basal diameter 
was measured at 5 cm aboveground. The basal diameter of multi-stemmed species were 
measured for all stems.  
Woody individuals were divided into two life stages: seedlings or established trees and 
shrubs. Seedlings and saplings were not differentiated, but lumped into a single seedling 
category. The term ‘seedling’ in this particular study refers to all woody individuals with a 
height between 10 cm up to 1 m, with a basal diameter of less than 1 cm. Established 
individuals represents shrubs or trees with a canopy height of over 1 m and basal diameter 
larger than 1 cm. If basal diameter was above 1 cm and height less than 1 m, the individual 
was regared as an established tree or shrub. 
A woody population in this study represent all individuals from the same species inside a 
specific treatment. A community in this study represent all woody and shrub individuals from 
the different species inside the specific treatment. 
All woody species recorded in the study site were classified as either encroachers or non-
encroachers (Wiseman et al., 2004; Gaugris et al., 2014). Encroachers are considered the 
‘increasers’ of woody plants because they are the first to increase in the presence of 
disturbances and the non-encroachers are classified as the ‘decreasers’, usually unable to 
tolerate disturbance (Moe et al., 2009).  
 
25 
 
4.3 Data analyses 
Data analyses were performed to test the effect of (i) catenal position, (ii) herbivore 
treatment, and (iii) fire treatment, on the structure and composition of the whole woody 
community as well as for encroacher and non-encroacher woody species. A summary of the 
analyses is presented below. Please note that a detailed account of statistical analyses is 
provided within each of the corresponding chapters. 
Chapter 5: Floristic analyses based upon multi-dimensional ordinations were done to assess 
species composition changes. A description of the dominant families, species and basic 
plant functional types (eg. growth form, deciduousness, palatability, and encroaching ability) 
of the woody plants is presented. Comparisons between woody species composition among 
seedlings and established trees and shrubs were made through the application of Non-
Metric Multidimensional Scaling (NMDS) in Primer 6 (Clarke, 1993). Significant differences in 
species assemblages were determined through the application of a One-way ANOSIM (in 
Primer).  
Chapter 6: Linear modelling (Hancock & Meuller, 2010) was used to compare the responses 
of woody abundances (seedlings and established individuals respectively) across different 
treatments of herbivory and fire, as well as across vegetation zones between 2002 and 
2015. Interaction effects were tested for herbivore treatments, fire treatments and the 
vegetation zones between 2002 and 2015. Effect sizes (Ellis & Steyn, 2003) were used to 
test for significant temporal changes in woody abundances within herbivore and fire 
treatments.  
Chapter 7: The demography of woody species was analysed by using size-class 
distributions (SCD’s). Woody individuals were grouped into 6 size-classes based upon their 
canopy height (i.e. ≤1 m; >1–2 m; >2–3 m; >3–4 m; >4-5 m; >5 m). SCD’s were graphically 
displayed for visual comparisons. Ordinary Least Square Linear regressions (Hutcheson, 
2011) were performed on the size-class distributions to determine the slope significance. 
The Permutation Index (Wiegand et al., 2000) was used to evaluate the deviation of the 
community or population from a monotonic decline. Simpson’s Index of Dominance 
(Wiegand et al., 2000) measured the evenness of size-class occupation. Population stability 
was evaluated by using quotient analyses (Shackleton, 1993).  
  
26 
 
CHAPTER 5 
Floristic changes in response to herbivory and fire 
5.1 Introduction 
The Nkuhlu exclosures site is characterised by a rich woody diversity in co-existence with a 
herbaceous layer dominated primarily by grass, but rich in forb species (Siebert & Scogings, 
2015). Studying the floristic composition of an ecosystem is often neglected, although it often 
reveals important knowledge on changes that can not always be depicted from structural 
data analyses. Note that this chapter serves only as an introductory chapter on woody 
regeneration and therefore the results of the floristics and plant functional types are not 
discussed but only described.  
Several studies have indicated that herbivore and fire effects on trees are species specific 
(Levick & Rogers, 2008; Wigley et al., 2014). Whether herbivore and fire effects can be 
linked to changes in dominant woody families are relatively unknown. Plant functional traits 
have become a major research topic in savanna vegetation studies (De Deyn et al., 2008; 
Hoffmann et al., 2012; Clarke et al., 2013). Therefore, basic plant functional traits were also 
considered to test for trait dominance changes over time across herbivore and fire 
treatments.  
African savanna flora evolved in association with herbivores and fire for millions of years 
(Charles-Domique et al., 2016). Therefore, it is predicted that herbivory and fire exclusion 
from a savanna ecosystem will lead to changes in dominant woody families and traits. This 
chapter will test whether 13 years of herbivore and fire exclusion is long enough to reveal 
any changes. 
Herbivore activity or exclusion may cause changes in woody species composition (Levick & 
Rogers, 2008; Barton & Hanley, 2013). Herbivores can negatively impact recruitment of 
woody species, resulting in unsuccessful regeneration. This can especially be true for more 
palatable woody species, which are more likely to get browsed upon, while unpalatable 
species can grow into established size-classes (Fornara & Du Toit, 2008). It is predicted in 
this descriptive chapter that seedling and established communities will differ in species 
composition when comparing the two communities where herbivores are present and the 
absence of herbivores will cause similarity in seedling and established community 
composition. 
27 
 
5.2 Methods 
A detailed description of the experimental design and woody data sampling appraoch is 
provided in Chapter 4 (Materials and Methods). Each recorded woody individual (seedling or 
established tree) was identified up to species level and then assigned to its plant family 
according to Germishuysen & Meyer (2003). Each herbivore and fire treatment was 
assessed in terms of dominant families and species in the seedling and established 
communities separately. This was done to test whether 13 years of herbivore exclusion 
(either exclusion of only mega-herbivores or exclusion of all herbivores) or fire treatments 
affected dominant families and species for seedlings and established community 
respectively.  
The dominant species were further assessed in terms of their basic plant functional types to 
test whether herbivore and/or fire treatments caused changes in trait dominance between 
2002 and 2015. The traits that were evaluated included (i) encroacher or non-encroacher, (ii) 
growth form, i.e. tree, shrub or liana, (iii) single- or multi-stemmed, (iv) deciduous or 
evergreen, and (v) palatable (commonly browsed upon) or unpalatable (not preferred by 
browsers). These plant functional types were assigned to the species based on expert 
knowledge and literature (Germishuysen & Meyer, 2003). 
Statistical analyses were not done for the dominant families and plant functional types, since 
a detailed account was considered beyond the scope of this study. Differences in the 
percentage contribution based on individual abundances by each dominant family and ratio’s 
between dominant traits were used to test whether treatments of herbivory and fire affected 
floristic dominance over time. Results of herbivory effects on dominant families and traits 
were dispalyed for both 2002 and 2015, while results of fire effects were only provided for 
2015. This is because significant patterns between fire and no-fire treatments were only 
revealed in the 2015 community. The 2002 results are provided in Appendix B.  
Woody species composition was analysed to test effect of sampling year (between 2002 and 
2015) as well as for effects within the same year between seedlings and established 
individuals in herbivore exclusion treatments. The 2002 dataset (abundance per plot per 
treatment) was used as a control treatment where species composition was a result of 
herbivore presence, while the 2015 dataset (abundance per plot per treatment) was used as 
the variable explaining the effect of herbivore exclusion on species composition. These 
analyses were performed for the full and partial exclosures only because it addressed the 
specific objective to determine the effect of herbivore exclusion (either all herbivores or only 
mega-herbivores) on woody species composition. Analyses on species composition were 
28 
 
performed through the application of Non-Metric Multidimensional Scaling (NMDS) in Primer 
6 (Clarke, 1993). The data was fourth-root transformed to reduce weighting of the more 
abundant species, while not losing information on their relative abundances (Clarke, 1993; 
Scogings et al., 2012). Bray-Curtis similarity resemblance measure was used, with 50 
restarts. These procedures were followed for both the seedling and established woody data 
sets - firstly analyses were done between years to test temporal effects on seedling and 
established species composition in the herbivore exclusion treatments, followed by a 
combined analysis of seedlings and established trees of the 2002 and 2015 data sets 
(respectively). This was done to test for differences in species composition between the two 
life stages within each year. Each vegetation zone hosts its own unique vegetation 
composition (Siebert & Eckhardt, 2008) and therefore species assemblage changes were 
also tested across the different vegetation zones (i.e. riparian, sodics and crest) for both 
seedlings and established individuals between 2002 and 2015. 
One-way ANOSIM (in Primer 6) (Clarke, 1993) was applied to the species assemblage 
datasets to test for significant differences in species composition between woody seedlings 
and established community in 2002 and 2015. Bray-Curtis coefficient of similarity was used 
together with 9999 permutations of the data. The R-statistic was used to calculate significant 
differences in species assemblages in the ordinal space. R=1 indicate species assemblages 
between two groups were totally different and R closer to 0 indicate groups to be 
indistinguishable. For this experiment, R-values >0.3 suggest observed differences in woody 
species assemblages. 
 
5.3 Results and discussion 
5.3.1 Floristic changes 
5.3.1.1 Responses to herbivore treatments 
5.3.1.1.1 Herbivore presence; Control treatment: 
Seedling dominant families did not change over time across any of the catenal positions, but 
changes did occur in the established community (Figure 5.1). These changes may be 
ascribed to continued herbivore activity in the control site. Euphorbiaceae replaced the 
frequently occurring Tiliaceae (Grewia) in 2015 in the riparian vegetation zone (Figure 5.1). 
In the sodic zone, herbivore presence favoured occurrance of Rubiaceae over the more 
palatable Sapindaceae (Pappea capensis) (Figure 5.1). The abundant species in the riparian 
29 
 
zone (Bridelia cathartica, Flueggea virosa and Spirostachys africana), which are all in the 
family Euphorbiaceae are less palatable than species from other families, such as Fabaceae 
species, but are known to be utilized by elephant and black rhino (Scogings et al., 2011). 
Although F. virosa (Euphorbiaceae) is one of the top ten most browsed species in the study 
area, especially by elephants (Scogings et al., 2012), results revealed that F. virosa remains 
under the most abundant riparian species in the presence of all herbivores. This suggests 
that F. virosa can tolerate elephant herbivory. Utilization of Tiliaceae (Grewia spp.) by 
herbivores is usually the same as that of F. virosa (Scogings et al., 2012), but the Grewia 
spp. in this study were more suppressed by elephants. Elephants are likely to topple Grewia 
spp. or break their branches (Boundja & Midgley, 2010). Results suggest little floristic 
changes in crest vegetation zone over time.  
5.3.1.1.2 Elephant and giraffe exclusion; Partial exclosure: 
Changes in the floristics of the riparian zone were minor, opposed to more pronounced 
changes in the sodics zone. Dominance in abundance by seedlings from the Tiliaceae family 
(Grewia spp.) replaced Rhigozum spp. seedlings (Bignoniaceae), and the palatable Pyrostria 
hystrix (Rubiaceae) seedlings in 2002 were replaced by less palatable Maerua parvifolia 
(Capparaceae) in 2015 (Figure 5.2). These shifts in floristics of seedlings are suggested to 
be associated with an increased mesobrowser effect when elephants are excluded from the 
landscape (Fritz et al., 2002; Valeix et al., 2008; Hilbers et al., 2015; Lagendijk et al., 2015). 
In the established woody community of the sodic zone, there was a strong shift towards 
Fabaceae species dominance in the abundance of individuals in the absence of elephants 
(Figure 5.2), which suggests elephant control of Fabaceae species. The unpalatable Euclea 
divinorum (Ebenaceae) was no longer one of the most abundant species after 13 years of 
elephant exclusion (Figure 5.2). On the upland crest the Grewia spp. (Tiliaceae) became 
dominant in abundance in the established woody community (Figure 5.2), which suggest that 
elephants have suppressive effects on Grewia spp.  
5.3.1.1.3 Total herbivore exclusion; Full exclosure: 
The exclusion of all herbivores for 13 years had the largest effect on floristic abundance 
dominance of all herbivore treatments in both the seedling and established woody 
community. In the riparian zone, the plant families that dominated, in abundance, the 
seedling community in 2002 were totally different than the dominant abundancesin 2015 
(Figure 5.3). Fabacae seedlings were replaced by Ebenaceae (Euclea spp.) (Figure 5.3). 
The appearance of Jasminum fluminese as a dominant abundant species in the seedling 
community in 2015 (Figure 5.3) illustrates the positive effect of herbivore exclusion on lianas. 
30 
 
Lianas are therefore suppressed by herbivores because when herbivores were excluded 
Lianas became more abundant. Herbivores, especially browsing herbivores, utilize plant 
material directly from or near trees and/or shrus, therefore lianas are suppressed by 
herbivore activity either through direct utilization or trampling. Euphorbiaceae seedlings were 
replaced by Celastraceae (Gymnosporia spp.), which suggest stronger competitive ability of 
the latter to increased herbaceous biomass. However, in the established woody community, 
Euphorbiaceae appeared as a dominant abundant family to the expense of Grewia 
flavescens (Figure 5.3). In the sodic zone the more palatable Sapindaceae (Pappea 
capensis) seedlings appeared and the Capparaceae (Maerua parvifolia) disappeared as 
being dominant (Figure 5.3). In the established community, Rhigozum abundances 
increased in the absence of all herbivores, while Ebenaceae spp., which are usually 
associated with the sodic vegetation zone, became less abundant (Figure 5.3). In the crest 
zone, the dominant families changed only in the seedling community. Combretaceae 
seedlings were not listed as one of the top three abundant families and were replaced by 
Tiliaceae (Grewia flavescens) (Figure 5.3), which are probably more sensitive to herbivores 
and more tolerant to high herbaceous biomass competition.  
In summary, less dynamic floristic changes occurred in the crest vegetation zone than in the 
riparian or sodic zones. The crest vegetation zone is considered as a dystrophic savanna 
and therefore less intense herbivore activity is expected. The eutrophic bottomlands (riparian 
and sodic vegetation zones) indicated the most floristic changes.  
31 
 
 
Figure 5.1: Top three dominant woody families (expressed as abundance in %) and species per 
year (i.e. 2002 and 2015) for both seedling and established communities for each vegetation 
zone (riparian, sodic, crest) of the control treatment.  
Burseraceae; 20 
Combretaceae; 
28 
32 
 
 
Figure 5.2: Top three dominant woody families (expressed as abundance in %) and species per 
year (i.e. 2002 and 2015) for both seedling and established communities for each vegetation 
zone (riparian, sodic, crest) in the partially fenced treatment. 
33 
 
 
Figure 5.3: Top three dominant woody families (expressed as abundance in %) and species per 
year (i.e. 2002 and 2015) for both seedling and established communities for each vegetation 
zone (riparian, sodic, crest) in the fully fenced treatment. 
34 
 
5.3.1.2 Responses to fire treatments 
5.3.1.2.1 Herbivore presence; Control treatment:  
In the control site, only the riparian and the crest zone could be tested for fire effects since 
the sodic vegetation zone did not extend into the fire exposed area. Euphorbiaceae 
seedlings was most abundant in the riparian vegetation zone despite fire or no-fire 
treatments (Figure 5.4). Fire did not cause any remarkable changes in the woody seedling 
floristics on the crest zone (Figure 5.5). The established community responded to fire 
treatments when 2002 and 2015 communities were compared (Appendix B). Burseraceae 
(Commiphora spp.) were amongst the top three dominant families in 2002 but not in 2015, 
which indicated that Commiphora spp. are sensitive to fire exposure. Grewia spp. were 
dominant in fire exsposed areas, but this could be due to low intensity (cold) fire, because 
adult Grewia spp. are usually more sensitive to fire.  
5.3.1.2.2 Elephant and giraffe exclusion; Partial exclosure: 
In the riparian vegetation zone, Fabaceae and Ebenaceae seedlings were most abundant 
despite fire exposure or absence of fire (Figure 5.6). In the sodic zone Tiliaceae and 
Capparaceae seedlings were more fire resistand than the palatable Boraginaceae (Ehretia 
amoena) and Celastraceae (Gymnosporia buxifolia) (Figure 5.7). No changes were 
observed in the riparian and sodic vegetation zones in the established community (Figure 
5.6 and 5.7). In the crest zone, Tiliaceae (Grewia spp.) were more tolerant to fire than 
Celastraceae (especially the palatable G. buxifolia and Hippocratea longipetiolata) in both 
the seedling and established communities (Figure 5.8). No changes were observed whithin 
fire treatments between 2002 and 2015 (Appendix B). 
5.3.1.2.3 Total herbivore exclusion; Full exclosure: 
In the fully fenced treatment, the high herbaceous biomass (Van Coller et al., 2013) was 
expected to have an impact on floristic composition and dominance of woody species due to 
the competitive exclusion effect. In the riparian zone, Ebenaceae was abundant in both the 
presence and absence of fire (Figure 5.9). Euphorbiaceae and Fabaceae seedlings 
indicated tolerence to fire while the absence of fire favoured Oleaceae and Celastraceae 
(Figure 5.9). In the sodic vegetation zone, Tiliaceae and Capparaceae seedlings were more 
fire resistant than the palatable Rhigozum zambesiacum and especially Pappea capensis 
seedlings, which are seemingly not fire resistant in their seedling stage (Figure 5.10). In the 
established community, Grewia spp. were more sensitive to fire than Spirostachys africana 
(Euphorbiaceae) and Euclea divinorum (Ebenaceae) (Figure 5.9). In areas where fire was 
35 
 
excluded, Grewia spp. dominated (Figure 5.9). Rhigozum zambesiacum was also dominant 
in the absence of fire (Figure 5.10). On the crest zone, Burseraceae family (Commiphora 
africana) responded negatively to fire exposure (Figure 5.11). C. africana is a low-growing 
species which usually remains whithin the fire-trap. Ebenaceae (Euclea divinorum) 
dominanted in the presence of fire, but Grewia spp. dominated where fire was excluded 
(Figure 5.11). This suggests that Grewia spp. are sensitive to fire in their seedling stage. 
Grewia spp. in the established community were also sensitive to fire. The palatable 
Celastraceae family indicated tolerence to fire, but not during the seedling stage. 
Fabaceae was the most successful family in this system because it was abundant in most of 
the treatments. This family also indicated most tolerence to fire. Fabaceae species are 
mostly palatable and therefore herbivores are expected to affect their abundance. It was 
however only in the riparian vegetation zone in the fully fenced exclosure (Figure 5.9) where 
Fabaceae was not listed as one of the top three abundant families in 2015.  
 
 
Figure 5.4: Top three dominant woody families (expressed as abundance in %) and species per 
fire treatment in 2015 for both seedling and established communities in the riparian zone of 
the control treatment. 
 
36 
 
 
Figure 5.5: Top three dominant woody families (expressed as abundance in %) and species per 
fire treatment in 2015 for both seedling and established communities in the crest zone of the 
control treatment. 
 
 
Figure 5.6: Top three dominant woody families (expressed as abundance in %) and species per 
fire treatment in 2015 for both seedling and established communities in the riparian zone of 
the partially fenced treatment. 
37 
 
 
Figure 5.7: Top three dominant woody families (expressed as abundance in %) and species per 
fire treatment in 2015 for both seedling and established communities in the sodic zone of the 
partially fenced treatment. 
 
 
Figure 5.8: Top three dominant woody families (expressed as abundance in %) and species 
per fire treatment in 2015 for both seedling and established communities in the crest zone of 
the partially fenced treatment. 
38 
 
 
Figure 5.9: Top three dominant woody families (expressed as abundance in %) and species 
per fire treatment in 2015 for both seedling and established communities in the riparian zone 
of the fully fenced treatment. 
 
 
Figure 5.10: Top three dominant woody families (expressed as abundance in %) and species 
per fire treatment in 2015 for both seedling and established communities in the sodic zone of 
the fully fenced treatment. 
39 
 
 
Figure 5.11: Top three dominant woody families (expressed as abundance in %) and species 
per fire treatment in 2015 for both seedling and established communities in the crest zone of 
the fully fenced treatment. 
 
5.3.2 Basic plant functional type changes 
5.3.2.1 Responses to herbivore treatments 
The dominating functional types for the seedling community at the study site were palatable, 
deciduous, single-stemmed encroacher tree and shrub species. The established community 
was dominated by palatable encroacher multi-or single-stemmed shrubs. 
5.3.2.1.1 Riparian zone; Seedling functional type changes 
Little changes in dominant plant functional types were observed in the control site (Table 
5.1). In the partially fenced exclosure, the ratio between encroachers and non-encroachers 
increased and the dominant growth form changed from single stemmed trees in 2002 to 
multi-stemmed shrubs in 2015 (Table 5.1). Dominance by deciduous and palatable species 
slightly decreased from 2002 to 2015. In the fully fenced exclosure, the ratio between 
encroachers and non-encroachers decreased, which suggests dominance of non-
encroacher over encroacher species after 13 years of herbivore exclusion (Table 5.1). In 
2015, shrubs and lianas became more dominant (Table 5.1). This was also observed in the 
appearance of Oleaceae family in 2015 (see section 5.3.1.1; Figure 5.3) Multi-stemmed 
40 
 
species dominated in 2002, but the dominance changed to single-stemmed species in 2015 
(Table 5.1). A decrease in dominance by palatable species occurred between 2002 and 
2015 (Table 5.1) where Ebenaceae family (Euclea spp.) became more dominant in the 
riparian zone of the fully fenced exclosure (see section 5.3.1.1; Figure 5.3).  
5.3.2.1.2 Riparian zone; Established community functional type changes 
The only remarkable change in plant functional types in the control site was the increased 
abundance of unpalatable over palatable species (Table 5.2). This could be explained by the 
changes in dominance from Ebenaceae and Fabaceae in 2002 to Euphorbiaceae and 
Ebenaceae in 2015 for the established woody community (see section 5.3.1.1; Figure 5.1). 
In the partially fenced exclosure, the ratio between encroachers and non-encroachers 
decreased between 2002 and 2015 (Table 5.2). Elephants commonly suppress tree 
dominance, which could not be confirmed in the trait results. Dominance by single-stemmed 
trees was replaced by multi-stemmed shrubs after 13 years of elephant exclusion (Table 
5.2), although the ratio of palatable species to unpalatable species slightly increased in the 
absence of mega-herbivores.  
In the absence of all herbivores, the ratio between encroachers and non-encroachers as well 
as between palatable and unpalatable woodies increased. A small decrease in shrub 
dominance was observed between 2002 and 2015, while dominance by deciduous species 
increased.  
5.3.2.1.3 Sodic zone; Seedling functional type changes 
The sodic zone in the control site did not have major changes in traits over time (Table 5.1). 
The partial exclosure was dominated by encroachers over non-encroachers over time – the 
ratio increased from 1:0.1 to 1:0.02 (Table 5.1). Single-stemmed trees remained dominant 
over multi-stemmed shrubs between 2002 and 2015 (Table 5.1). Dominance by palatable 
species increased over time. In the absence of all herbivores, encroaching species remained 
dominant over time (Table 5.1). A higher tree to shrub ratio was observed in 2015 and multi-
stemmed species dominated over single-stemmed species (Table 5.1). Deciduous and 
palatable species were suppressed by the absence of herbivores (Table 5.1). 
5.3.2.1.4 Sodic zone; Established community functional type changes 
In the established community of the partially fenced exclosure, encroachers dominated over 
non-encroacher species in the sodic zone (Table 5.2). An increase was observed in the ratio 
between trees and shrubs (Table 5.2). A decrease in dominance by multi-stemmed species 
41 
 
occurred over time, which led to co-dominance by single- and multi-stemmed species in 
2015 (Table 5.2). Evergreen woodies dominated in 2002, but were replaced by the 
dominance of deciduous woodies. Decreased palatability ratio was observed in the 2015 
community as to the 2002 community. In the absence of all herbivores, the ratio between 
encroachers and non-encroachers increased more than three-fold (Table 5.2). The dominant 
growth form changed from single-stemmed trees to multi-stemmed shrubs (Table 5.2). The 
absence of herbivores promoted deciduous species to dominate over evergreen species 
(Table 5.2). The ratio between palatable and unpalatable woodies increased which again 
indicated protection for palatable woodies where all herbivores were excluded (Table 5.2). 
5.3.2.1.5 Crest zone; Seedling functional type changes 
The ratio of non-encroacher to encroacher species increased between 2002 and 2015 in the 
fully fenced area (Table 5.1). Shrubs remained dominant over trees and lianas in all 
herbivore treatments (Table 5.1). Multi-stemmed individuals increased in abundance over 
time in the presence of elephant and decreased in their absence where the ratio of single-
stemmed to multi-stemmed individuals remained higher (Table 5.1). There was a decrease 
in deciduousness of species in the control site (Table 5.1). In all treatments palatable woody 
species dominated over unpalatable species (Table 5.1).  
5.3.2.1.6 Crest zone; Established community functional type changes 
In the established community of the partially fenced exclosure, non-encroachers dominated 
in abundance in 2002 and in 2015 abundance of encroachers (Table 5.2). In the fully fenced 
exclosure, the ratio of encroachers to non-encroachers was much higher than in 2002 (Table 
5.2). An increase in shrub dominace occurred over time (Table 5.2). There was an increased 
dominance of deciduous woody abundances over abundances of evergreen species in the 
absence of mega-herbivores (Table 5.2). Abundance of palatable species to unpalatable 
species increased from 2002 to 2015 in the treatments where herbivores were excluded, 
irrespective whether all herbivores, or only elephants and giraffes have been excluded 
(Table 5.2).  
 
42 
 
Table 5.1: Plant functional type ratios for dominant seedling species over time across herbivore treatments and vegetation zones. 
Vegetation 
zone 
Herbivore 
treatment 
Year Plant Functional Types Dominant traits* 
Encroachers: 
Non-encroachers 
Tree:Shrub:Liana Single-stemmed: 
Multi-stemmed 
Evergreen:Deciduous Unpalatable:Palatable 
Riparian Control 2002 1 : 0.5 1  :   2      : 0 1 : 1.8 1 :     2 1 :     5 En/S/Mu/D/P 
2015 1 : 0.6 1  :   2      : 0 1 : 1.4 1 :     3 1 :     3 En/S/Mu/D/P 
Partial 2002 1 : 0.08 1  :   0.7   : 0 1 : 0.7 1 :     5 1 :     9  En/T/Si/D/P 
2015 1 : 0.06 1  :   3      : 0 1 : 3 1 :     4 1 :     4 En/S/Mu/D/P 
Full 2002 1 : 0.3 1  :   4      : 0 1 : 3 1 :     3 1 :   76 En/S/Mu/D/P 
2015 1 : 1.8 1  : 14      : 10 1 : 0.6 1 :     0.04 1 :     1.3 N/S/Si/E/P 
Sodic Control 2002 1 : 0.2 1  :   0.2   : 0 1 : 0.5 1 :     3 1 :     8 En/T/Si/D/P 
2015 1 : 0.2 1  :   0.3   : 0 1 : 0.8 1 :     2 1 :     6 En/T/Si/D/P 
Partial 2002 1 : 0.1 1  :   0.6   : 0 1 : 0.6 1 :   11 0 : 148 En/T/Si/D/P 
2015 1 : 0.02 1  :   0.8   : 0 1 : 0.7 1 :     7 0 : 322 En/T/Si/D/P 
Full 2002 1 : 0.01 1  :   1      : 0 1 : 1.4 1 :   27 0 : 340 En/TS/Mu/D/P 
2015 1 : 0.2 1  :   0.02 : 0 1 : 3 1 :     5 0 : 183 En/T/Mu/D/P 
Crest Control 2002 1 : 0.6 1  :   0.5   : 0 1 : 0.2 1 : 239 0 : 240 En/T/Si/D/P 
2015 1 : 0.9 1  :   0.7   : 0 1 : 0.3 1 :   41 0 : 251 En/T/Si/D/P 
Partial 2002 1 : 0.7 1  :   1.8   : 0 1 : 0.3 1 :   15 1 :     7 En/S/Si/D/P 
2015 1 : 0.6 1  :   2      : 1 1 : 0.2 1 :   10 1 :     9 En/S/Si/D/P 
Full 2002 1 : 0.07 1  :   1.2   : 0 1 : 0.8 1 : 107 0 : 325 En/S/Si/D/P 
2015 1 : 0.2 1  :   2      : 0 1 : 0.6 1 : 167 0 : 336 En/S/Si/D/P 
*En=encroachers; N=non-encroachers; T=trees; S=shrubs; Si=single-stemmed; Mu=multi-stemmed; D=deciduous; E=evergreen; P=palatable; U=unpalatable;    
Bold=dominant trait 
.
43 
 
Table 5.2: Plant functional type ratios for dominant established tree species over time across herbivore treatments and vegetation zones. 
Vegetation 
zone 
Herbivory 
treatment 
Year Plant Functional Types Dominant traits* 
Encroachers: 
Non-encroachers 
Tree:Shrub:Liana Single-stemmed: 
Multi-stemmed 
Evergreen:Deciduous Unpalatable:Palatable 
Riparian Control 2002 1 : 0.2 1 : 5    : 0 1 : 4 1 :     1.6 1 :    1.7 En/S/Mu/D/P 
2015 1 : 0.3 1 : 4    : 0 1 : 2 1 :     4 1 :    4 En/S/Mu/D/P 
Partial 2002 1 : 0.1 1 : 1    : 0 1 : 0.8 1 :     3 1 :    3 En/TS/Si/D/P 
2015 1 : 0.2 1 : 1    : 0 1 : 1 1 :     4 1 :    5 En/TS/SiMu/D/P 
Full 2002 1 : 0.8 1 : 4    : 0 1 : 1.2 1 :     1 1 :    1.2 En/S/Mu/ED/P 
2015 1 : 0.5 1 : 3    : 0 1 : 1.2 1 :     2 1 :     3 En/S/Mu/D/P 
Sodic Control 2002 1 : 0.2 1 : 0.6 : 0 1 : 1.3 1 :     0.8  1 :     1.5 En/T/Mu/E/P 
2015 1 : 0.1 1 : 0.9 : 0 1 : 0.8 1 :     2 1 :     4 En/T/Si/D/P 
Partial 2002 1 : 0.1 1 : 1.7 : 1 1 : 1.4 1 :     0.8 1 :     0.6 En/S/Mu/E/U 
2015 1 : 0.1 1 : 1.1 : 1 1 : 1 1 :     1.2 1 :     0.9 En/S/SiMu/D/U 
Full 2002 1 : 0.01 1 : 0.3 : 0 1 : 0.3 1 :     5 1 :     5 En/T/Si/D/P 
2015 1 : 0.003 1 : 1.2 : 0 1 : 1.2 0 : 566 0 : 566 En/S/Mu/D/P 
Crest Control 2002 1 : 0.8 1 : 0.4 : 0 1 : 0.3 1 :   71 0 : 432 En/T/Si/D/P 
2015 1 : 0.4 1 : 0.5 : 0 1 : 0.5 1 : 138 0 : 418  En/T/Si/D/P 
Partial 2002 1 : 1.1 1 : 0.8 : 0.1 1 : 0.6 1 :     8 1 :     9 N/T/Si/D/P 
2015 1 : 0.6 1 : 1    : 0 1 : 0.9 1 :   52  0 : 424 En/TS/Si/D/P 
Full 2002 1 : 0.3 1 : 1.2 : 0 1 : 1.2 0 : 245 0 : 245 En/S/Mu/D/P 
2015 1 : 0.1 1 : 1.8 : 1 1 : 1.2 1 : 341 0 : 683 En/S/Mu/D/P 
*En=encroachers; N=non-encroachers; T=trees; S=shrubs; Si=single-stemmed; Mu=multi-stemmed; D=deciduous; E=evergreen; P=palatable; U=unpalatable;    
Bold=dominant trait. 
 
44 
 
5.3.2.2 Responses to fire treatments 
Consistant abundant functional types for both seedlings and established communities in fire 
exposed areas were deciduous, palatable encroachers (Table 5.3; Table 5.4). Fire 
suppressed the abundance of shrubs (Table 5.3; Table 5.4). In the seedling community, 
multi-stemmed species were favoured by the presence of fire, while in the established 
community fire suppressed this trait (Table 5.3; Table 5.4). 
5.3.2.2.1 Riparian zone; Seedling functional type changes 
In the control site, main ratio changes occurred between fire treatments in the 
tree:shrub:liana ratio, with an increase in especially shrubs and lianas in the absence of fire 
(Table 5.3). Deciduous species dominated in the presence of fire, while the absence of fire 
led to dominance by evergreen species (Table 5.3). In the absence of elephant, a stronger 
encroacher to non-encroacher ratio as well as an increase in multi-stemmed shrubs was 
observed in the fire-excluded areas (Table 5.3). Palatability seem to be favoured by fire, but 
not in the absence of elephant where browsing by mesoherbivores negatively affect 
palatable seedlings (Table 5.3). In the fully fenced exclosure, fire favoured encroacher 
dominance over non-encroachers, whereas the opposite was observed in the absence of fire 
(Table 5.3). The no-fire area had a higher shrub:tree ratio compared to the area exposed to 
fire as well as more abundant evergreen individuals than deciduous individuals (Table 5.3). 
5.3.2.2.2 Riparian zone; Established community functional type changes 
In the control site, the ratio of encroachers to non-encroachers was higher in the fire-
excluded area than where fire was present (Table 5.4). Single-stemmed shrubs dominated in 
abundance in the area where fire was present, while multi-stemmed trees dominated the 
area where fire was excluded (Table 5.4). The abundance ratio of deciduous to evergreen 
individuals increased in the presence of fire (Table 5.4). More palatable species were 
located in the fire exposed area (Table 5.4). In the partially fenced exclosure, a higher 
encroacher to non-encroacher ratio was observed in the presence of fire compared to 
absence of fire (Table 5.4). Single-stemmed trees dominated in abundance in the fire-
exposed ares, while abundance of multi-stemmed shrubs dominated in the absence of fire 
(Table 5.4). Abundance of deciduous and palatable individuals dominated over evergreen 
and unpalatable individuals in the absence of fire (Table 5.4). In the fully fenced exclosure, 
the ratio between encroacher and non-encroacher abundances was less in the presence of 
fire than in the absence of fire (Table 5.4). In the absence of fire multi-stemmed shrubs 
45 
 
dominated in abundance (Table 5.4). Abundances of deciduous and palatable individuals 
was higher in the presence of fire compared to the no-fire treatment (Table 5.4).  
5.3.2.2.3 Sodic zone; Seedling functional type changes 
In the partially fenced exclosure, a higher ratio of encroacher to non-encroacher aundances 
was observed in the fire exposed areas compared to the areas excluded from fire (Table 
5.3). The most abundant individuals in the area where fire was applied, were palatable 
(Table 5.3). The abundance ratio of deciduous to evergreen individuals was higher in areas 
protected from fire than the areas exposed to fire (Table 5.3). In the fully fenced exclosure, 
the encroacher to non-encroacher as well as the deciduous to evergreen ratio increased in 
the presence of fire (Table 5.3). Abundance of multi-stemmed shrubs dominated in both the 
fire and no-fire treatments, but dominance was stronger in the fire exclusion areas of the fully 
fenced exclosure (Table 5.3). 
5.3.2.2.4 Sodic zone; Established community functional type changes 
In the partially fenced exclosure, higher abundance of encroachers was observed in the 
presence of fire, with the main growth form being trees (Table 5.4). The abundance of the 
single-stemmed functional type dominated in the presence of fire, while abundance of multi-
stemmed individuals dominated in the absence of fire (Table 5.4). Deciduous and palatable 
species were most abundant in fire exposed areas, while evergreen and unpalatable species 
were most abundant in the no-fire areas (Table 5.4). In the fully fenced exclosure, high 
encroacher to non-encroacher abundance ratio was observed in the presence and absence 
of fire (Table 5.4). Palatable, deciduous multi-stemmed, encroacher, shrub species 
dominated in abundance where herbivores and fire were excluded from this nutrient hot-spot 
(Table 5.4). 
5.3.2.2.5 Crest zone; Seedling functional type changes 
In the control site, non-encroacher abundance dominated over abundance of encroachers in 
fire-exposed areas. In the absence of fire, encroacher abundances dominated. Single-
stemmed trees were most abundant in the presence of fire (Table 5.3). The areas protected 
from fire were dominated by higher abundances of deciduous and palatable individuals 
(Table 5.3). In the partially fenced exclosure, a slightly higher abundance ratio in encroacher 
to non-encroacher was observed in the absence of fire (Table 5.3). In the absence of fire, 
single-stemmed shrub abundances was higher than in areas where fire was excluded (Table 
5.3). The presence of fire favoured deciduous and palatable species in the presence of 
mesoherbivores (control site and partial exclosure) (Table 5.3). In the fully fenced exclosure, 
46 
 
the absence of fire promoted encroacher to non-encroacher abundance ratio (Table 5.3). 
Dominance by shrubs were much stronger in the no-fire areas, while fire-exposed areas 
were dominated by trees (Table 5.3). More abundant deciduous individuals than evergreen 
individuals as well as more abundant palatable individuals than unpalatable individuals was 
observed in the absence of fire (Table 5.3).  
5.3.2.2.6 Crest zone; Established community functional type changes 
In the control site, the absence of fire favoured dominance of encroachers above non-
encroachers (Table 5.4). Dominance in abundance of single-stemmed trees was higher in 
the presence of fire than in the areas where fire was excluded (Table 5.4). The ratio of 
deciduous to evergreen species increased in the absence of fire. Palatable species 
dominated both fire and no-fire areas (Table 5.4). In the partially fenced exclosure, 
encroachers dominated both fire and no-fire treatments (Table 5.4). The tree to shrub as well 
as single-stemmed to multi-stemmed ratio was more or less similar in both fire treatments 
(Table 5.4). Higher abundances of deciduous (than evergreen) and palatable (than 
unpalatable) individuals was observed in the presence of fire (Table 5.4). In the fully fenced 
exclosure, dominance of encroachers above non-encroachers increased in the presence of 
fire (Table 5.4). Shrub dominance was also favoured by the presence of fire (Table 5.4). The 
multi-stemmed functional type dominated in abundance in both the fire and no-fire 
treatments, while deciduous and palatable individuals had higher abundance dominance in 
the absence of fire (Table 5.4). 
 
47 
 
Table 5.3: Plant functional type ratios for dominant seedling species in 2015 across fire treatments and vegetation zones. 
Vegetation 
zone 
Fire treatment Plant Functional Types Dominant 
traits* Encroachers: 
Non-encroachers 
Tree:Shrub:Liana Single-stemmed: 
Multi-stemmed 
Evergreen: 
Deciduous 
Unpalatable:Palatable 
Riparian Control; Fire 1 : 1 1 :   1.1  :   0 1 : 0.8 1 :    6 1 :     7 EnN/S/Si/D/P 
Control; No Fire 1 : 1.1 1 : 14     : 11 1 : 0.8 1 :    0.7 1 :     8 N/S/Si/E/P 
Partial; Fire 1 : 0.2 1 :   2     :   0 1 : 1.6 1 :    7 1 :     8  En/S/Mu/D/P 
Partial; No Fire 1 : 0.09 1 :   5     :   0.5 1 : 3 1 :    3 1 :     3 En/S/Mu/D/P 
Full; Fire 1 : 0.2 1 :   1.3  :   0 1 : 0.8 1 :    2 1 :     2 En/S/Si/D/P 
Full; No Fire 1 : 2 1 : 12     : 10 1 : 0.4 1 :    0.1 1 :     1.4 N/S/Si/E/P 
Sodic Partial; Fire 1 : 0.03 1 :   0.8  :   0.02 1 : 0.8 1 :    4 0 : 191 En/T/Si/D/P 
Partial; No Fire 1 : 0.2 1 :   0.9  :   0.03 1 : 0.5 1 :    8 1 :     7 En/T/Si/D/P 
Full; Fire 1 : 0.1 1 :   5     :   0.7 1 : 3 1 :   29 0 :   91 En/S/Mu/D/P 
Full; No Fire 1 : 0.3 1 :   1.1  :   0 1 : 3 1 :     4 0 : 117 En/S/Mu/D/P 
Crest Control; Fire 1 : 1.1 1 :   0.5  :   0 1 : 0.2 1 :   23 0 : 141 N/T/Si/D/P 
Control; No Fire 1 : 0.6 1 :   1     :   0 1 : 0.7 0 : 110 0 : 110 En/TS/Si/D/P 
Partial; Fire 1 : 0.8 1 :   1.2  :   0 1 : 0.5 0 : 213 0 : 213 En/S/Si/D/P 
Partial; No Fire 1 : 0.4 1 :   3     :   0.3 1 : 0.1 1 :   13 1 :     6 En/S/Si/D/P 
Full; Fire 1 : 0.4 1 :   0.7  :   0 1 : 0.7 1 :     5 1 :     5 En/T/Si/D/P 
Full; No Fire 1 : 0.1 1 : 11     :   0 1 : 0.7 1 : 220 0 : 221 En/S/Si/D/P 
*En=encroachers; N=non-encroachers; T=trees; S=shrubs; Si=single-stemmed; Mu=multi-stemmed; D=deciduous; E=evergreen; P=palatable; U=unpalatable;    
Bold=dominant trait. 
 
48 
 
Table 5.4: Plant functional type ratios for dominant established tree species in 2015 across fire treatments and vegetation zones. 
Vegetation 
zone 
Fire treatment Plant Functional Types Dominant traits* 
Encroachers: 
Non-encroachers 
Tree:Shrub:Liana Single-stemmed: 
Multi-stemmed 
Evergreen: 
Deciduous  
Unpalatable: 
Palatable 
Riparian Control; Fire 1 : 0.5 1 : 2      : 0 1 : 0.9 1 :   67 0 : 135 En/S/Si/D/P 
Control; No Fire 1 : 0.2 1 : 0.1   : 0 1 : 5 1 :     1.1 1 :     3 En/T/Mu/D/P 
Partial; Fire 1 : 0.1 1 : 0.7   : 0 1 : 0.7 1 :     3 1 :     4 En/T/Si/D/P 
Partial; No Fire 1 : 0.3 1 : 1.6   : 0 1 : 1.7 1 :     5 1 :     6 En/S/Mu/D/P 
Full; Fire 1 : 0.3 1 : 1.6   : 0 1 : 1.2 1 :     4 1 :     5 En/S/Mu/D/P 
Full; No Fire 1 : 0.4 1 : 6      : 0 1 : 2 1 :     1.6 1 :     1.7 En/S/Mu/D/P 
Sodic Partial; Fire 1 : 0.1 1 : 0.8   : 0.1 1 : 0.6 1 :     1.6 1 :     1.1 En/T/Si/D/P 
Partial; No Fire 1 : 0.1 1 : 1.9   : 0.2 1 : 1.6 1 :     1 1 :     0.7 En/S/Mu/ED/U 
Full; Fire 1 : 0.001 1 : 1      : 0 1 : 1 1 :   10 1 :   10 En/S/SiMu/D/P 
Full; No Fire 1 : 0.0003 1 : 1.8   : 0 1 : 1.8 0 : 379 0 : 379  En/S/Mu/D/P 
Crest Control; Fire 1 : 0.5 1 : 0.3   : 0 1 : 0.3 1 : 103 0 : 208 En/T/Si/D/P 
Control; No Fire 1 : 0.3 1 : 0.7   : 0 1 : 0.7 1 : 209 0 : 210 En/T/Si/D/P 
Partial; Fire 1 : 1 1 : 0.9   : 0 1 : 0.8 1 :   28 0 : 145 EnN/T/Si/D/P 
Partial; No Fire 1 : 0.6 1 : 1      : 0.2 1 : 0.8 1 :   12 1 :     7 En/T/Si/D/P 
Full; Fire 1 : 0.04 1 : 1.2   : 0.05 1 : 1.1 1 :   22 1 :   16 En/S/Mu/D/P 
Full; No Fire 1 : 0.2 1 : 3      : 0 1 : 3 1 : 258 0 : 259 En/S/Mu/D/P 
*En=encroachers; N=non-encroachers; T=trees; S=shrubs; Si=single-stemmed; Mu=multi-stemmed; D=deciduous; E=evergreen; P=palatable; U=unpalatable;    
Bold=dominant trait. 
 
49 
5.3.3 Species composition changes 
5.3.3.1 Response to herbivore exclusion 
In both the fully and partially fenced exclosure, no significant changes in species assemblages 
of either life stage communities (i.e. seedling- and established woody community) were revealed 
when comparing the 2002 and 2015 communities of each life stage (Anosim, R<0.3) (Figure 
5.12; Figure 5.13). In 2002, when all herbivores had access to both the fully and partially fenced 
exclosure sites, seedling and established species composition were dissimilar in the riparian 
vegetation zone (Full: R=0.57, Figure 5.14a; Partial: R=0.72, Figure 5.15a) and the sodic zone 
(Full: R=0.39, Figure 5.14a; Partial: R=0.31, Figure 5.15a). Where all herbivores were excluded 
species composition was similar between seedlings and established community in all three 
vegetation zones (R<0.3 for all three vegetation zones) (Figure 5.14b), whereas exclusion of 
only elephants led to dissimilar species composition between seedlings and established 
community in the riparian vegetation zone (R=0.32) (Figure 5.15b). Meso-herbivores therefore 
affected species composition in especially the riparian vegetation zone. 
Scogings et al. (2012) evaluated woody species composition change over a short term period at 
the Nkuhlu exclosures in the absence of large mammalian herbivores. The study revealed no 
significant change in woody species composition over 5 years (Scogings et al., 2012). After 13 
years of large mammalian herbivore exclusion, woody species composition still remained 
unchanged, when comparing seedlings and established life stages separately between 2002 
and 2015. Long-term absence of herbivores, however, reduces the dissimilarity in species 
assemblages between the seedling and established woody communities. Dissimilar species 
composition between the seedling- and established woody community could be ascribed to 
browsing effects (Bond et al., 2001; Guldemond & Van Aarde, 2009; Staver et al., 2009; 
Sankaran et al., 2013). Direct utilisation by browsing herbivores inhibits woody recruitment, 
either by removing palatable woody seedlings from the ecosystem, or by suppressing their 
growth (Sankaran et al., 2013). Palatable species are most likely to get trapped inside a 
botlleneck while unpalatable species can get established (Fornara & Du Toit, 2008). Results 
from this study revealed that the release from herbivore pressure led to woody establishment 
and hence to similar seedling and established woody communities. Results suggest that mega-
herbivores especially impacted species composition in the uplands while meso-herbivores drive 
changes in the bottomlands.  
50 
 
Figure 5.12: NMDS ordination scatter plot to visually display the distribution of seedling (a) and 
established (b) woody species composition in 2002 (blue) and 2015 (green) in the fully fenced 
exclosure. Different symbols represent different positions along a granitic toposequence (i.e. 
different vegetation zones). 
51 
 
Figure 5.13: NMDS ordination scatter plot to visually display the distribution of seedling (a) and 
established (b) woody species composition in 2002 (blue) and 2015 (green) in the partially fenced 
exclosure. Different symbols represent different positions along a granitic toposequence (i.e. 
different vegetation zones). 
52 
 
Figure 5.14: NMDS ordination scatter plot to visually display the distribution of 2002 (a) and 2015 
(b) woody species composition for the seedling (blue) and established (green) communities in the 
fully fenced exclosure. Different symbols represent different positions along a granitic 
toposequence (i.e. different vegetation zones). 
53 
 
Figure 5.15: NMDS ordination scatter plot to visually display the distribution of 2002 (a) and 2015 
(b) woody species composition for the seedling (blue) and established (green) communities in the 
partially fenced exclosure. Different symbols represent different positions along a granitic 
toposequence (i.e. different vegetation zones). 
 
  
54 
5.4 Conclusion 
It was expected that the floristic composition of the woody layer at the study site would change 
in response to exclusion of herbivory and fire. Changes in most abundant woody families in 
response to herbivore exclusion were especially observed in the eutrophic bottomlands, which 
are usually favoured by herbivores. PFT’s also indicated changes in response to herbivore 
exclusion, which further contribute to support the hypothesis. Woody floristic responses to fire 
were especially recognised in areas where herbivores were excluded. This means that 
herbivory, in this study site, override effects of fire. Most changes occurred in the upland areas 
which are characterised as a dystrophic savanna. This corresponds to studies which suggested 
that especially dystrophic savannas mostly indicate fire responses in its floristic composition 
(Scholes, 1990; Lehmann et al., 2011; Hempson et al., 2015). Results from this chapter 
therefore suggest that 13 years of excluding fundamental drivers of savanna systems (herbivory 
and fire), is long enough to initiate changes in dominant families and basic PFT’s. The nature of 
these changes can however not yet be classified as permanent as it could be in a transitional 
state. 
Herbivore activity suppresses seedling growth and inhibiting woody healthy recruitment. It is 
therefore expected that only those species adapted to herbivore utilisation can get established. 
Therefore, dissimilar seedling and established species composition was found in the presence 
of herbivores (except for the sodic vegetation zone), while similar species compositions were 
found in the absence of herbivores, with the exception of the riparian in the partial exclosure. 
The release of all large mammalian herbivore activity from this system could therefore result in a 
more homogeneous woody community by creating similar species composition groups for 
seedlings and established trees. Herbivory can therefore be identified as a driver of 
heterogeneity in this semi-arid savanna ecosystem. 
  
55 
CHAPTER 6 
Woody abundance changes across herbivore and fire treatments 
6.1 Introduction 
Seedlings are considered the fundamental stage in regeneration. Seedling survival and/or 
mortality define future woody community composition and structure (Barton & Hanley, 
2013).The seedling stage is the most vulnerable due to their small size, limited structural 
support as well as inter- and intraspecific competition (Miranda et al., 1993). Both bottom-up 
and top-down controls may affect successful seedling establishment. 
High herbaceous biomass, leading to competition between the herbaceous and woody layer, 
can be considered a constraint for seedling recruitment and establishment (Kambatuku et al., 
2011). Extensive below-ground grass root biomass may compete with woody seedlings for 
resources, inhibiting successful establishment of woody individuals (Eckhardt et al., 2000; 
Riginos, 2009; Kambatuku et al., 2011; Vadigi & Ward, 2014).  
While the herbaceous layer performs bottom-up competition with seedling establishment, 
certain top-down controls (i.e. herbivory and fire) may also contribute to changes in woody 
community structure, including woody regeneration success. Both herbivory and fire may inhibit 
sapling growth into established adult tree size-classes through direct consumption, imposing a 
recruitment bottleneck in the form of ‘browse or fire traps’ (Bond & Keeley, 2005). Recent 
studies indicated that herbivory, or rather herbivore exclusion, will cause change in woody 
species composition (see results from previous chapter; Levick & Rogers, 2008; Barton & 
Hanley, 2013). Such changes may be attributed to the suppression effect of browsers on woody 
seedlings and saplings, keeping them in the browse trap (Bond et al., 2001; Guldemond & Van 
Aarde, 2009; Staver et al., 2009; Sankaran et al., 2013). Palatable woody species are more 
likely to get trapped, while unpalatable species have the ability to grow into higher size-classes 
(Fornara & Du Toit, 2008). When woody species are being released from this trap through the 
exclusion of herbivore activity (Staver et al., 2009), the once suppressed seedlings and 
saplings, often referred to as gullivers (Bond & van Wilgen, 1996) may grow into established 
size-classes (Asner et al., 2009). This will contribute to increases in the abundance of 
established woody individuals. Impacts of fire on woody species are expected to be similar to 
that of herbivory since fire also consumes woody biomass, killing trees/shrubs or limits their 
growth (Bond & Keeley, 2005; Staver et al., 2009). Woody seedlings and saplings are limited in 
growth through fire keeping them within the fire-trap (0-3 m in height). Only fire-prone or fire-
adapted species will survive and grow beyond the fire-trap to reach heights unaffected by fire 
(Bond & Keeley, 2005). This may contribute to changes in woody structure over the longer term.  
56 
Therefore, both bottom-up and top-down controls may create two major demographic 
bottlenecks in savanna tree populations (Higgings et al., 2000; Sankaran et al., 2004): (i) 
recruitment bottleneck, where seedlings are trapped within the grass sward, and (ii) sapling 
release bottleneck, where herbivory and fire inhibit growth past sapling stage to form large adult 
trees (also known as the ‘browse or fire traps’). These bottom-up and top-down controls may in 
the long-term contribute to changes in woody abundances (Fornara & Du Toit, 2008; Barton & 
Hanley, 2013).  
Woody encroachment is a growing concern for rangeland scientists and conservationists of 
savanna ecosystems (Archer et al., 1995; Scholes and Archer, 1997; Roques et al., 2001; 
Wiegand et al., 2006; Wigley et al., 2010). Global and regional drivers have been identified to 
explain this phenomenon (Stevens et al., 2016). At a global scale, (Asner et al., 2003; Cabral et 
al., 2003; Laiolo et al., 2004; Wiegand et al., 2006), increasing atmospheric CO2 levels have 
been identified as one of the major contributors to increasing woody encroachment (Stevens et 
al., 2016). At a regional scale, increasing atmospheric CO2 levels combined with disturbances 
such as herbivory and fire, and to some extent the release thereof from a system, are 
considered attributes to encroachment (Bond & Midgley, 2000; Stevens et al., 2016). Herbivory 
and fire can also, if managed well, control or even suppress woody encroachment in savanna 
systems. Responses of the woody layer to herbivory and fire are often species-specific (Levick 
& Rogers, 2008; Migley et al., 2014), which ultimately lead to a species being categorized as 
either an encroaching or a non-encroaching woody species.  
During their seedling stage, both encroachers and non-encroachers are affected by herbivory 
and fire, either directly through consumption or indirectly through grazing herbivores or fire 
events that create gaps for seedling establishment. How herbivory and fire control 
encroachment of woody species without negatively affecting non-encroacher species in a 
natural, protected semi-arid savanna ecosystem still remains relatively unexplored.  
The Nkuhlu exclosures provided the opportunity to test how the presence of, or release from 
herbivores and fire affected woody community changes by comparing abundances of woody 
encroacher and non-encroacher species between 2002 and 2015 for both seedlings and 
established woody individuals. 
It is hypothesised that exposure to herbivory and fire will not significantly reduce encroacher 
seedling abundances due to their adaptive strategies, such as resprouting, in savanna 
ecosystems (Moe et al., 2009). It is rather expected that encroacher seedling abundances will 
increase in sites that are open to herbivore disturbance. Herbivore activity prevents herbaceous 
biomass accumulation (Van Coller et al., 2013), therefore inhibit warm fires and hence reduce 
the competitive exclusion effect on encroacher seedlings. Non-encroacher seedling 
57 
abundances are expected to be lower in the presence of herbivores and fire and more 
specifically in the nutrient-rich bottomlands (i.e. riparian and sodic vegetation zones), which are 
favoured by most browsers (Scholes, 1990). Elephant effects are expected to be strongest on 
the abundance of established trees rather than on seedlings. 
 
6.2 Methods 
A detailed account of data sampling methods is provided in the Methods chapter (Chapter 4). 
Only the methodology that is designed to meet the specific objectives stated in this chapter will 
be discussed here. 
Woody seedlings are referred to those individuals with a height of less than 1 m and a basal 
diameter of less than 1 cm, while established individuals have a height of ≥1 m and a basal 
diameter of ≥1 cm. 
Hierarchical Linear Modelling (Hancock & Meuller, 2010) was used to determine whether 
herbivore/fire treatments affected mean woody abundance per plot across different vegetation 
zones between 2002 and 2015. The interaction between fixed effects, which included herbivore 
treatments, fire treatments, vegetation zones, and years, was tested for (i) all woody seedlings 
and established individuals and (ii) encroacher and non-encroacher seedlings and established 
individuals respectively. Significant interaction effects were identified at p<0.05. To test for 
significance within effects, particularly temporal changes in abundance, the data were further 
analysed through the calculation of effect sizes. Effect size is normally used to determine 
practical significance, i.e. difference between two means divided by the estimate of standard 
deviation, by making the difference independent of units and sample size, and relate it to the 
spread of the data (Ellis & Steyn, 2003). Small effect sizes (eg. d<0.5) indicated non-significant 
changes, while medium effect sizes (d=0.5–0.6) and large effect sizes (d>0.6-1) indicated 
significant changes. Highly significant changes in abundances were reported by an effect size of 
>1.  
 
6.3 Results 
Across both life stages (seedlings and established woodies) as well as woody communities 
(complete, encroachers and non-encroachers), various combination of interaction effects were 
revealed (Table 6.1). The only two treatments that consistently interacted significantly were 
herbivory and fire - in some cases with herbivory and in other cases with vegetation zone (Table 
58 
6.1). Fire treatment revealed no significant interaction effects with any other treatments in the 
established woody community. Seedling abundances, however, were stronger affected by fire. 
Both encroacher and non-encroacher seedlings were affected by fire, which suggests their 
sensitivity to fire (Table 6.1). The only four-way interaction effect was revealed for the 
encroacher seedling community (Table 6.1). Linear modelling results suggest that 13 years of 
herbivory and fire manipulations were adequate to observe changes due to the significant year 
effects in both life stages and all communities (Table 6.1). Herbivore treatment can be 
considered as the main driver of system dynamics at the study site since it interacted 
significantly with all other effects (Table 6.1). Vegetation zone also played an important role in 
the system, but is considered secondary to the interactions between herbivory and year.  
 
Table 6.1: Summary of significant interaction effects between treatments (vegetation zones, 
herbivore treatments, fire treatments and year (2002–2015)) for seedlings and established 
individuals separately. The complete woody community assessment included all woody 
individuals, which was then separated into encroacher and non-encroacher species for separate 
analyses on these functional groups. 
Life stage Woody community Fixed effects F p 
Seedlings Complete woody community Vegetation zone*Herbivory treatment*Year 7.72 < 0.01 
Herbivory treatment*Fire treatment*Year 4.8 0.011 
Encroachers Vegetation zone*Herbivory treatment*Fire treatment*Year 5.63 0.001 
Non-encroachers Vegetation zone*Herbivory treatment*Year 6.14 0.000 
Vegetation zone*Fire treatment*Year 4.61 0.012 
Herbivory treatment*Fire treatment*Year 8.43 0.000 
Established Complete woody community Vegetation zone*Herbivory treatment*Year 4.44 0.003 
Encroachers Vegetation zone*Herbivory treatment*Year 2.92 0.026 
Non-encroachers Vegetation zone*Herbivory treatment*Year 4.15 0.004 
Vegetation zone: Riparian, Sodic, Crest; Herbivore treatment: Control site, Partial exclosure, Full exclosure; Fire treatment: exposure to fire, fire 
exclusion; Year: between 2002 and 2015. F: variance in group means, p: significant when p<0.05. 
 
6.3.1 Complete woody community 
Seedlings 
There was a significant interaction effect between year, vegetation zone and herbivore 
treatment for woody seedling abundances at the study site (Table 6.1). These results suggest 
that seedling abundance changes between 2002 and 2015 varied across the different 
vegetation zones and herbivore treatments, regardless whether fire was applied or not. 
Contrasting patterns were revealed between herbivore treatments in the riparian zone since 
seedling abundance between 2002 and 2015 increased significantly in both the fully fenced 
exclosure and the control site (Table 6.2). Seedlings in the sodic zone were favoured by the 
presence of meso-herbivores (control and partial exclosure sites), regardless of the presence or 
59 
absence of elephants (Table 6.2). The exclusion of elephants on the crest zone led to a highly 
significant increase in seedling abundance over time (Table 6.2). Therefore, the largest 
increases in seedling abundance were in the absence of elephant, but only on the midslopes 
and upland areas (sodic and crest zones).  
 
Table 6.2: Woody seedling abundances per year across vegetation zones and herbivore 
treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size d 
(2002 vs 2015) 
Riparian Control 2002 33.58 6.98 1.20 *** 
2015 57.25 6.98 
Partial 2002 33.75 5.70 0.27 
2015 28.33 5.70 
Full 2002 31.60 5.69 0.98 ** 
2015 50.86 5.69 
Sodic Control 2002 19.41 8.04 0.81 ** 
2015 35.30 8.04 
Partial 2002 21.83 5.70 1.27 *** 
2015 46.83 5.70 
Full 2002 35.68 5.68 0.43 
2015 27.25 5.68 
Crest Control 2002 21.56 5.69 0.29 
2015 27.36 5.69 
Partial 2002 35.83 5.70 1.95 *** 
2015 74.25 5.70 
Full 2002 35.86 5.68 0.47 
2015 45.20 5.68 
d: d<0.5=non-significant; d=0.5–0.6=medium effect*; d>0.6–1 =large effect**;                                       
d>1=significant large effect***. 
 
Fire interacted significantly with herbivory and year, irrespective of vegetation zones (Table 6.1). 
The fire exposed plots in the control site and partially fenced exclosure increased significantly in 
seedling abundance between 2002 and 2015 (Table 6.3). In the plots where fire was excluded 
from 2002 onwards, all herbivore treatments (control site, partially and fully fenced exclosures) 
revealed significant increases in seedling abundance 13 years later (Table 6.3). The highest 
increase was observed in the absence of elephant (Table 6.3). 
 
 
 
60 
Table 6.3: Woody seedling abundances per year across fire and herbivore treatments. 
Fire 
treatment 
Herbivore 
treatment 
Year Mean 
abundance per 
plot 
Std.Error Effect size d 
(2002 vs 2015) 
Fire Control 2002 25.42 6.98 1.08 *** 
  2015 47.67 6.98  
 Partial 2002 32.28 4.65 0.57 * 
  2015 43.56 4.65  
 Full 2002 36.32 4.64 0.06 
  2015 35.05 4.64  
No Fire Control 2002 26.29 4.64 0.55 * 
  2015 37.07 4.64  
 Partial 2002 28.67 4.65 1.39 *** 
  2015 56.06 4.65  
 Full 2002 32.44 4.64 0.75 ** 
  2015 47.15 4.64  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d>1=significant large effect***. 
 
Established individuals 
For the established woody community a significant three-way interaction effect was revealed 
between vegetation zone, herbivore treatment and year (Table 6.1). A strong elephant exclusion 
effect was observed in the riparian zone where significant increases in established individuals 
between 2002 and 2015 were observed, despite the presence (partial exclosure) or absence 
(full exclosure) of other large herbivores (Table 6.4).  
No herbivory effects were observed in the sodic zone, whereas a strong elephant effect was 
noticed on the crest (Table 6.4). This was detected in the decrease of established tree 
abundances in the control site, while an increase in abundance was observed in the absence of 
elephant, regardless of the presence of other herbivores (Table 6.4).  
 
 
 
 
 
 
61 
Table 6.4: Woody abundances of established individuals per year across vegetation zones and 
herbivore treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size d 
(2002 vs 2015) 
Riparian Control 2002 6,29 2,44 0,17  
  2015 7,49 2,50  
 Partial 2002 1,42 2,05 0,73 ** 
  2015 6,58 2,05  
 Full 2002 5,70 2,05 2,30 *** 
  2015 22,02 2,04  
Sodic Control 2002 4.27 2.87 0.45 
  2015 7.49 2.87  
 Partial 2002 6.17 2.05 0.12 
  2015 5.33 2.05  
 Full 2002 8.33 2.05 0.01 
  2015 8.25 2.05  
Crest Control 2002 12.42 2.05 0.51 * 
  2015 8.83 2.05  
 Partial 2002 8.50 2.05 0.92 ** 
  2015 15.00 2.05  
 Full 2002 9.67 2.05 0.78 ** 
  2015 15.17 2.05  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d>1=significant large effect***. 
 
6.3.2 Woody abundances of encroacher species 
Seedlings 
There was a significant four-way interaction effect of vegetation zone, herbivore treatment, fire 
treatment and year on encroacher seedling abundances (Table 6.1). Significant increases in the 
abundance of encroacher seedlings were present in four of the six treatments in the riparian 
zone, two with fire and two without fire (Table 6.5). Exclusion of elephants had the largest effect 
on encroacher seedling increases, since significant increases were revealed irrespective of fire 
treatment (Table 6.5). Therefore, for encroacher species seedling abundances, the exclusion of 
elephants predominate fire effects. Accumulation of herbaceous biomass in the fully fenced 
exclosure in combination with fire seem to have favoured encroacher seedlings (Table 6.5), 
which suggests their resistance to fire, even to assumingly high intensity fires produced by high 
fuel loads. 
In the sodics vegetation zone, fire did not play an important role in the control of encroacher 
seedlings, except where all herbivores were excluded. Encroacher seedlings in the sodic zone 
were favoured by the presence of meso-herbivores, regardless of the presence or absence of 
elephant (Table 6.5).  
62 
The prolonged activity of all herbivores on the crest zone (i.e. control site) seem to have 
controlled encroacher seedling abundances, irrespective of fire treatment (Table 6.5). Where 
elephants were excluded (partial exclosure), significant increases in encroacher seedling 
abundances were observed, irrespective of fire treatment (Table 6.5). These observations 
therefore suggest elephant control over increase in encroacher seedling abundances, but 
mostly on the upland crest. 
 
Table 6.5: Woody abundances of encroacher seedlings per year across vegetation zones, 
herbivore treatments and fire treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Fire 
treatment 
Year Mean 
abundance 
per plot 
Std.Error Effect size d 
(2002 vs 
2015) 
Riparian Control Fire 2002 19.00 7.19 0.43  
   2015 24.33 7.19  
  No Fire 2002 20.50 5.08 0.58 * 
   2015 27.67 5.08  
 Partial Fire 2002 40.17 5.08 1.85 ** 
   2015 17.17 5.08  
  No Fire 2002 11.50 5.08 0.95 ** 
   2015 23.33 5.08  
 Full Fire 2002 17.67 5.08 0.64 ** 
   2015 25.67 5.08  
  No Fire 2002 23.26 5.08 0.48  
   2015 17.34 5.08  
Sodic Control No Fire 2002 13.55 5.08 1.13 *** 
   2015 27.64 5.08  
 Partial Fire 2002 16.67 5.08 2.13 *** 
   2015 43.17 5.08  
  No Fire 2002 12.83 5.08 1.63 *** 
   2015 33.17 5.08  
 Full Fire 2002 24.53 5.08 0.49  
   2015 18.48 5.08  
  No Fire 2002 37.24 5.08 1.33 *** 
   2015 20.72 5.08  
Crest Control Fire 2002 8.00 5.08 0.12 
   2015 9.50 5.08  
  No Fire 2002 8.37 5.08 0.17 
   2015 10.49 5.08  
 Partial Fire 2002 8.33 5.08 1.22 *** 
   2015 23.50 5.08  
  No Fire 2002 14.67 5.08 0.74 ** 
   2015 23.83 5.08  
 Full Fire 2002 43.45 5.08 1.39 *** 
   2015 26.16 5.08  
  No Fire 2002 10.38 5.08 0.75 ** 
   2015 19.71 5.08  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
63 
Established individuals 
A significant three-way interaction effect was revealed between established encroacher species 
abundances and vegetation zone, herbivore treatment and year (Table 6.1). The strongest 
elephant effect on encroacher abundances was observed in the riparian and crest zones. In the 
presence of elephant (control site), increases in established encroacher individuals occured, 
although these increases were negligible compared to treatments where elephants have been 
excluded for 13 years. Significant increases in established encroacher species abundances 
were noticed in the absence of elephant, regardless the presence of other herbivores (partial 
exclosures) or without them (full exclosure) (Table 6.6). In the nutrient-rich sodics zone, 
encroacher species abundances did not increase significantly over time in any of the treatments 
(Table 6.6). 
 
Table 6.6: Woody abundances of established encroacher individuals per year across vegetation 
zones and herbivore treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size d 
(2002 vs 2015) 
Riparian Control 2002 3.750 1.827 0.35 
  2015 5.583 1.827  
 Partial 2002 1.083 1.492 0.79 ** 
  2015 5.167 1.492  
 Full 2002 2.417 1.492 1.97 *** 
  2015 12.583 1.492  
Sodic Control 2002 4.167 2.110 0.35 
  2015 6 2.110  
 Partial 2002 4.500 1.492 0.06 
  2015 4.167 1.492  
 Full 2002 7.500 1.492 0.06 
  2015 7.167 1.492  
Crest Control 2002 4.500 1.492 0.39 
  2015 6.500 1.492  
 Partial 2002 2.500 1.492 0.98 ** 
  2015 7.583 1.492  
 Full 2002 7.667 1.492 0.92 ** 
  2015 12.417 1.492  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
 
 
 
64 
6.3.3 Woody abundances of non-encroacher species 
Seedlings 
Abundances of non-encroacher seedling interacted significantly with vegetation zone, herbivore 
treatment and year (Table 6.1), which suggests a limited fire effect. Increases in non-encroacher 
seedlings in the riparian zone cannot be attributed to herbivore effects since seedling 
abundances increased significantly in both the presence (control site) and absence (full 
exclosure) of all herbivores (Table 6.7). On the uplands (crest zone), non-encroacher seedling 
abundances increased significantly where elephants have been excluded for 13 years (partial 
exclosure), and also where all herbivores have been excluded (Table 6.7). Similar to encroacher 
seedlings, non-encroacher seedlings of the dystrophic upland crests also seem to be controlled 
by elephant. 
 
Table 6.7: Woody abundances of non-encroacher seedlings per year across vegetation zones and 
herbivore treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size d 
(2002 vs 2015) 
Riparian Control 2002 13.83 4.78 1.29 *** 
  2015 31.25 4.78  
 Partial 2002 7.92 3.90 0.01 
  2015 8.00 3.90  
 Full 2002 10.99 3.89 1.36 *** 
  2015 29.43 3.89  
Sodic Control 2002 5.68 5.47 0.14 
  2015 7.61 5.47  
 Partial 2002 7.08 3.90 0.12 
  2015 8.67 3.90  
 Full 2002 5.06 3.86 0.20 
  2015 7.78 3.86  
Crest Control 2002 13.31 3.89 0.32 
  2015 17.70 3.89  
 Partial 2002 24.33 3.90 1.89 *** 
  2015 49.83 3.90  
 Full 2002 8.51 3.86 1.01 *** 
  2015 22.14 3.86  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
 
A second significant interaction effect was observed between vegetation zone, fire treatment 
and year (Table 6.1). Non-encroacher seedlings had better recruitment in the no-fire treatments, 
specifically in the riparian zone, while species of the crest zone did not indicate sensitivity to fire 
in their seedling stage. (Table 6.8).  
65 
Table 6.8: Mean woody abundance of non-encroacher seedlings between 2002 and 2015 in 
different fire treatments per vegetation zone. 
Vegetation 
zone 
Fire 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size d 
(2002 vs 2015) 
Riparian Fire 2002 12.444 3.681 0.42 
  2015 18.167 3.681  
 No Fire 2002 9.383 3.178 1.35 *** 
  2015 27.620 3.178  
Sodic Fire 2002 4.917 3.875 0.25 
  2015 8.301 3.875  
 No Fire 2002 6.706 3.168 0.09 
  2015 7.967 3.168  
Crest Fire 2002 10.518 3.172 0.91 ** 
  2015 22.809 3.172  
 No Fire 2002 20.247 3.168 1.24 *** 
  2015 36.974 3.168  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
 
In the significant three-way interaction between herbivore treatment, fire treatment and year 
(Table 6.1), non-encroacher seedling abundances did not increase in all the fire exclusion 
treatments. Significant increases in non-encroacher abundances was observed in the full and 
partial exclosures, where herbaceous biomass have accumulated more significantly over time. 
The fire exposed treatments in the full and partial herbivore exclosures would have produced 
warmer fires, which can suppress seedling recruitment (Table 6.9). Significant increases in 
seedling abundances in the fire-exposed area of the control site can possibly be ascribed to 
very low fire intensity (low fuel load) due to prolonged grazing in the control site. 
 
Table 6.9: Woody abundances of non-encroacher seedlings per year herbivory treatments and fire 
treatments. 
Herbivore 
treatment 
Fire 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size over 
years (Es) 
Control Fire 2002 11.92 4.78 1.32 *** 
  2015 29.75 4.78  
 No Fire 2002 12.04 3.17 0.24 
  2015 15.34 3.17  
Partial Fire 2002 10.61 3.19 0.33 
  2015 15.06 3.19  
 No Fire 2002 15.61 3.19 1.01 *** 
  2015 29.28 3.19  
Full Fire 2002 7.69 3.16 0.29 
  2015 11.62 3.16  
 No Fire 2002 8.68 3.16 1.42 *** 
  2015 27.95 3.16  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
66 
Established individuals 
A significant three-way interaction effect was observed for vegetation zone, herbivore treatment 
and year in the established non-encroacher community (Table 6.1). The abundances of 
established non-encroacher individuals did not increase where elephants have been excluded 
for 13 years, but only in the riparian and upland crest zones (Table 6.10). The only significant 
increase was revealed in the riparian bottomlands in the absence of all herbivores. Declining 
abundances of non-encroacher species were observed in the area exposed to all herbivores 
(control site) with significant decreases on the upland crests (Table 6.10). Since individuals 
increased in areas where only elephants have been excluded (partially fenced site), elephants 
are suggested as the main driver of declining abundances of established non-encroacher 
species. The impacts of giraffes are not being neglected, although their feeding behaviour give 
evidence to minor destructive effects, but rather of selective defoliation of the leaf canopy 
(Furstenburg & Van Hoven, 1994). 
 
Table 6.10: Woody abundances of established non-encroacher individuals per year across 
vegetation zones and herbivore treatments. 
Vegetation 
zone 
Herbivore 
treatment 
Year Mean abundance 
per plot 
Std.Error Effect size over 
years (Es) 
Riparian Control 2002 2.60 1.22 0.18 
  2015 1.99 1.22  
 Partial 2002 0.33 1.00 0.31 
  2015 1.42 1.00  
 Full 2002 2.90 1.00 1.87 *** 
  2015 9.36 1.00  
Sodic Control 2002 0.29 1.40 0.39 
  2015 1.63 1.40  
 Partial 2002 1.63 1.00 0.15 
  2015 1.12 1.00  
 Full 2002 0.83 1.00 0.07 
  2015 1.08 1.00  
Crest Control 2002 7.92 1.00 1.61 *** 
  2015 2.33 1.00  
 Partial 2002 6.00 1.00 0.39 
  2015 7.33 1.00  
 Full 2002 2.00 1.00 0.22 
  2015 2.75 1.00  
d: d<0.5=non-significant; d between 0.5 and 0.6=medium effect*; d>0.6–1=large effect**;                          
d >1=significant large effect***. 
 
 
67 
6.4 Discussion 
Established trees are commonly debarked, pushed over, and broken by elephants leading to a 
decline in tree densities in semi-arid savannas (Cumming et al., 1997; Shannon et al., 2008; 
Woolley et al., 2011). In both the riparian and crest vegetation zones, exclusion of elephants led 
to increases in established tree abundances. Stronger elephant effects on the upland crests 
correspond with findings by Lagendijk et al. (2015), who also reported elephant impacts to be 
more pronounced on upland areas. Stronger elephant effects on the crests compared to the 
riparian bottomlands may also be explained by findings by Smit et al. (2007) who reported that 
the riparian zone is able to maintain higher elephant densities due to higher forage availability. 
Results did not reveal significant elephant effects in the sodics vegetation zone. It is therefore 
suggested that elephants do not actively browse in the sodic zone, but rather pass through this 
area on their way to water. Although the sodics zone is considered a nutrient hotspot that 
produces vegetation that is more palatable than surrounding vegetation types (Grant & Scholes, 
2006), studies confirmed higher palatability only for the herbaceous layer (Siebert & Scogings, 
2015) with limited evidence of increased palatability in the woody layer. Results obtained 
through this study, which suggest that herbivore impact on the woody layer is lowest in the sodic 
zone, call for further investigation into the nutritious value of woody plants in this nutrient 
hotspot. 
The suggested high impact of elephants on the crest was not restricted to larger woody 
individuals. Seedling abundances increased where only elephants were excluded. Similarly, 
non-encroacher seedlings, which are all palatable species favoured by herbivores, increased in 
the absence of elephant, regardless of the presence or absence of other herbivores. This 
corresponds to the findings by Lagendijk et al. (2015) which stated that meso-herbivore effects 
were less severe in upland areas where elephant impact was high. Elephants are known to 
displace meso-herbivore activity (Fritz et al., 2002; Valeix et al., 2008; Hilbers et al., 2015; 
Lagendijk et al., 2015). Results from this study therefore support this hypothesis. 
Guldemond & Van Aarde (2008) stated that presence of elephants, especially in African 
savanna conservation areas, could cause a decline in woody cover. It was predicted by Stevens 
et al. (2016) that the combined effect of mega-herbivores and fire could slow down woody 
encroachment. This was partially supported in the results, but mainly on the crest zone where 
encroacher species abundances (seedlings and established individuals) significantly increased 
in the absence of elephant. Fire, however did not play a significant role in the suppression of 
encroacher species abundances. Encroaching woody seedlings in the lowland areas indicated 
to be best controlled by the combined presence of meso-herbivores and fire. The exclusion of 
elephants led to higher meso-herbivore activity (Fritz et al., 2002; Valeix et al., 2008; Hilbers et 
al., 2015; Lagendijk et al., 2015) in the riparian zone which, in combination with fire exposure 
68 
significantly decreased encroacher species abundances. The high abundance of encroacher 
species seedlings in the sodic vegetation zone asks for ecological monitoring of this valuable 
nutrient hotspot. 
Other than expected, the non-encroaching seedling abundances were not significantly reduced 
by meso-herbivore activity and fire, which suggests resistance towards bush encroachment 
controls (eg. fire and herbivory). However, non-encroacher species abundance increased 
significantly in the absence of fire, which suggests some sensitivity to fire, especially in the 
seedlings stage. Results further suggest that non-encroacher seedling species of the crest zone 
are less sensitive to fire. 
To conclude, elephants remain one of the strongest drivers of woody species abundances (for 
both encroachers and non-encroachers) (Asner et al., 2016) in both life stages. Encroacher 
species will be best controlled during seedling stage (Hoffmann et al., 2009), but particularly by 
the presence of elephant on the crest, and the presence of meso-herbivores in lowland areas. 
Non-encroachers can, to some extent, resist impacts from bush encroachment controls 
(herbivory and fire). 
 
6.5 Conclusion 
Considering woody abundances, the entire woody community, which includes established trees, 
seedlings, encroacher and non-encroacher species was negatively impacted by elephants, 
especially on the upland (crest) areas. In the presence of elephants, meso-herbivore effects 
were less severe because elephants are known to displace meso-herbivores across the 
landscape. Elephant browsing is suggested to act as a bush encroachment control measure in 
the upland areas, regardless of the presence or absence of fire, while the combination of 
intense meso-herbivore activity (where elephants are excluded) together with fire exposure 
could decrease encroaching woody seedling abundances without negatively suppressing non-
encroaching species in lowland areas. It is further suggested that elephants only pass through 
the sodic vegetation zone towards water, without significantly affecting the woody vegetation. 
  
69 
CHAPTER 7 
Demography of the woody community after 13 years of herbivore 
and fire manipulations 
7.1 Introduction 
Size-class distributions (population demography) are commonly used to assess woody 
population structures and the stability thereof (Shackleton, 1993; Mwavu & Witkowski, 2009; 
Venter & Witkowski, 2010; Byakagaba et al., 2011; Shackleton et al., 2015). In areas exposed 
to environmental disturbances, woody plant population demography may provide information on 
the most affected size classes for a particular species (Staver & Bond, 2014). 
Herbivory and fire are important factors in the maintenance of the tree-grass balance in semi-
arid savanna ecosystems. These top-down controls are known to suppress bush encroachment 
(Roques et al., 2001; Joubert et al., 2012; Lohmann et al., 2014), maintain heterogeneity in 
vegetation structure (Holdo et al., 2009; Levick et al., 2009; O’Kane et al., 2012) and recycle soil 
nutrients (Fornara & Du Toit, 2008). In savannas where tree-grass co-existence is a 
characteristic feature (Frost et al., 1986; Skarpe, 1992), herbivory and fire play an important role 
in controlling the tree-grass balanace directly or indirectly, which eventually determines 
vegetation structure and composition (Skarpe, 1991).  
Demography of woody species in a savanna may be affected by several factors, such as (i) the 
grass layer (Sankaran et al., 2008), which is considered as a fundamental filter for woody 
seedling recruitment; (ii) herbivory (Goheen et al., 2010) through grazing, browsing and 
trampling, and (iii) fire (Prior et al., 2010). All these factors may alter size-class distributions, 
causing changes from the commonly accepted normal inverse J-curve shape (Silwertown, 1992; 
Oliver & Larson, 1990; Condit et al., 1998; Wilson & Witkowski, 2003; Mwavu & Witkowski, 
2009; Holdo et al., 2014).  
In this chapter, woody community and population demography patterns will be assessed to 
address the effect of herbivory and fire, including their exclusion, on size-class distributions. 
This assessment will be undertaken at the community level, as well at species level where pre-
selected dominant encroacher and non-encroacher populations will be assessed. The aim of 
this study was to test the effects of herbivore and fire treatments at the Nkuhlu exclosures site 
on woody regeneration success, therefore evaluating whether or not on-going regeneration 
occured and whether the woody community or population can be considered stable or not. 
70 
The presence of herbivores and fire may affect size-class distribution of woody species by 
creating demographic bottlenecks through browse and fire traps (Higgins et al., 2000; Bond et 
al., 2001; Guldemond & Van Aarde, 2009; Staver et al., 2009; Sankaran et al., 2013; Staver & 
Bond, 2014; Levick et al., 2015). This may cause community and population instability. It is 
therefore expected that the exclusion of large mammalian herbivores and fire from the system 
will release trapped woody individuals (mostly seedlings or saplings), which will lead to changes 
in size-class distributions and ultimately reveal an inverse J-shaped demography and stable 
communities and populations. Encroacher woody species that have been shown to be 
suppressed by herbivores and fire in this savanna ecosystem (Chapter 6) are expected to revert 
to stable regeneration after the exclusion of these bush encroachment controls. Non-encroacher 
regeneration is expected to be favoured by the exclusion of all herbivores. 
 
7.2 Methods 
A detailed account of the data sampling approach is presented in chapter 4 (Material and 
methods). Analyses relevant to woody community regeneration patterns are discussed below. 
To meet the objectives of this particular chapter, only the 2015 data sets were analysed to 
assess woody community demography after 13 years of herbivore and fire manipulations. All 
woody individuals were grouped into six size-classes according to their maximum canopy height 
(Table 7.1). Basal stem diameter of size-class 1 was <1 cm and ≥1 cm for size-classes 2-6.  
 
Table 7.1: Size-classes for woody individuals based on their height (m). 
Size-class Height division 
1 <1m 
2 1-2m 
3 >2-3m 
4 >3-4m 
5 >4-5m 
6 >5m 
 
Size-class distributions (SCD) were constructed for: (i) the complete woody community, which 
included all woody species, (ii) pre-selected encroacher species and (iii) pre-selected non-
encroacher species. SCD were graphically displayed for visual comparisons. Ordinary least-
squares regressions (OLS) (Hutcheson, 2011) were performed on the SCD data set of 2015 to 
test SCD of woody species. Mean individuals per plot for each size class was used for the 
complete woody community, whereas the total number of individuals was plotted for the pre-
71 
selected encroacher and non-encroacher species (also referred to as the ‘key species’). These 
were set as the dependent variables (Ni), whereas the midpoint of each size-class was used as 
the independent variable (mi). Variables were transformed using ln(Ni+1) and ln(mi) prior to 
analyses (Lagendijk et al., 2011). SCD slopes were used as indicators of community and 
population structure (Obiri et al., 2002; Mwavu & Witkowski, 2009; Venter & Witkowski, 2010; 
Lagendijk et al., 2011). Interpretation of SCD slopes was based on Obiri et al. (2002) who 
reported negative slopes as indicatative of good recruitment with more individuals recorded in 
the lower size-classes than in the larger size-classes and positive slopes related to poor 
recruitment with more individuals in the larger size-classes than in the lower size-classes. 
Steepness of the slopes was used to describe recruitment trends. A steep negative slope is 
commonly used to indicate healthier recruitment than shallow slopes (Lykke, 1998; Obiri, 2002; 
Mwavu & Witkowski, 2009; Venter & Witkowski, 2010). The R2 coefficient determines the 
variability of the dependent variable. R2-values closer to 1 indicate that the variability of the 
dependent variable can be explained by the independent variable (Hutcheson, 2011). If the 
information provided by the explanatory variables (SCD slope) was significant, it was indicated 
by p=<0.05.  
Population stability was evaluated by determining quotients between successive size-classes 
and by displaying the results graphically. The interpretation of the quotients follows Shackleton 
(1993), Botha et al. (2004), and Venter & Witkowski (2010), who stated that constant quotients 
between successive size-classes indicate a stable population, while fluctuations in quotients 
display unstable populations.  
The Permutation Index (Equation 1) evaluates the deviation of a population from a monotonic 
decline, which can be expected to be present in a stable population. It is the sum of the 
absolute distances between the expected and real location (rank) of all size classes (Wiegand 
et al., 2000). The permutation index is higher in a discontinuous size-class distribution than in a 
continuous, monotonically declining population, which indicates stability. 
𝑃 =  ∑|𝐽𝑖 − 𝑖|;         𝐽𝑖 = 1, . . . , 𝐾
𝐾
𝑖=1
 
           Equation 1 
where Ji is the rank of size class i (i = 1 for the smallest trees), with the highest rank (Ji = 1) 
given for the most frequent size class, and K is the total number of size classes. 
Simpson’s Index of Dominance (Equation 2) measures the evenness of occupation of the size 
class, ignoring the order in which size classes are arranged (Wiegand et al., 2000). Values 
72 
above 0.1 reveal steeper size frequency than would be expected from an exponentially 
declining population, suggesting that the size classes are arranged in descending order, and 
values below 0.1 shows that size classes are more evenly distributed.  
𝐶 =
1
𝑁(𝑁 − 1)
∑ 𝑁𝑖(𝑁𝑖 − 1)
𝐾
𝑖=1
 
           Equation 2 
where N is the total number of trees, Ni the number of trees in class i, and K is the total number 
of size classes. 
Analyses of variance (ANOVA) were used in Statistica 13 to determine whether woody 
individuals in the respective size classes differed among the herbivory and fire treatments for 
the complete woody community at the Nkuhlu exclosures. The data set was firstly tested for 
normality by using Kolmogorov-Smirnov & Lilliefors tests. For data sets that met normality, a 
One-way ANOVA was applied, whereas Log10(x+1) transformation was applied to data that did 
not meet normality. In cases where normality could still not be revealed after transformation, a 
non-parametric Kruskal Wallis ANOVA was applied. To test for significant woody abundance 
differences (p<0.05) between treatments for each size-class, unequal N Tukey’s t-tests were 
applied. 
The species considered for demographic analyses were selected according to their dominance 
and ecological importance in this savanna ecosystem. Species dominance in each vegetation 
zone across the riparian topographic sequence were taken into consideration during species 
selections (Figure 3.2) (Siebert & Eckhardt, 2008). They include Diospyros mespiliformis, 
Ziziphus mucronata, Pappea capensis, Senegalia nigrescens, Vachellia exuvialis, Gymnosporia 
senegalensis, Spirostachys africana, Flueggea virosa, Vachellia grandicornuta, Rhigozum 
zambesiacum, Dichrostachys cinerea, and Combretum apiculatum (Table 5.2). The abundance 
of a species was also considered in the selection since some ecologically important species are 
sparsely represented in the study area. The species were also classified as either an 
encroacher or non-encroacher species based upon their specific ecology (i.e. ability to respond 
to disturbance or to occupy open spaces) and their palatability (Skarpe, 1990; Hudak & 
Wessman, 1998; Eggemeyer & Schwinning, 2009; Van Auken, 2009; Eldridge et al., 2011; 
Twidwell et al., 2013) (Table 7.2).  
Regeneration of the communities and populations was considered good when the following was 
observed: SCD indicated a negative slope together with the R2-value closer to 1; PI of 0 
(monotonic decline in SCD); and SDI >0.1–0.4, which suggests slightly uneven size-classes 
73 
frequency distribution without the complete dominance of a few size classes. These parameters 
graphically display a typical inverse J-shaped SCD curve, which is considered good 
regeneration, with highest mean or most abundant individuals in size-class 1 and successive 
size-classes declining in abundance. 
 
Table 7.2: Pre-selected diagnostic encroacher and non-encroacher woody species per vegetation 
zone. 
Vegetation zone Encroachers Non-encroachers 
Riparian Gymnosporia senegalensis Diospyros mespiliformis 
 Spirostachys africana Ziziphus mucronata 
 Flueggea virosa  
Sodic Vachellia grandicornuta Pappea capensis 
 Rhigozum zambesiacum  
Crest Dichrostachys cinerea Senegalia nigrescens 
 Combretum apiculatum Vachellia exuvialis 
 
7.3 Results 
7.3.1. Population structure of the complete woody community 
The demography of the 2015 Nkuhlu woody community revealed good regeneration through a 
typical inverse J-shaped pattern in its size-class distribution (Figure 7.1a). The highest mean 
number of individuals per plot was found in the seedling height class (<1 m) and declined in 
abundance within successive height classes (Figure 7.1a). This is also confirmed by the steep 
negative slope of -1.204 (Table 7.3). The Simpson’s dominance index for the overall woody 
community indicated slightly uneven distribution of size-class occupation (SDI=>0.1) whereas 
the permutation index of 0 indicated that the occupation of size-classes was declining 
monotonically (Table 7.3).Quotient analyses revealed stability in the lower size classes, 
whereas the higher size-classes (between 4 & 5 and 5 & 6) were unstable (Figure 7.1b). 
74 
 
Figure 7.1: Size-class distributions (plant canopy height in 1 m intervals) for mean number of 
woody individuals per plot (±SE) in successive size-classes (a) and quotients between successive 
size classes (b) for the complete woody community of 2015. 
 
Table 7.3: Summary of size-class distribution measures of the complete woody community in 
2015 for each herbivore and fire treatment. Ordinary least square regression analyses (Slope, SE 
Slope, R2, p), Permutation Index (PI) and Simpson’s Dominance Index (SDI). 
Treatment Size class distribution 
Slope SE Slope R2 p PI SDI 
Complete community -1.204 0.210 0.892 0.005* 0 0.32 
Control -1.286 0.230 0.886 0.005* 0 0.33 
Partial -1.315 0.141 0.956 0.001* 0 0.41 
Full -1.032 0.261 0.796 0.017* 0 0.27 
Control, Fire -1.232 0.252 0.857 0.008* 0 0.32 
Control, No Fire -1.285 0.230 0.886 0.005* 0 0.34 
Partial, Fire -1.347 0.101 0.978 0.000* 0 0.45 
Partial, No Fire -1.296 0.170 0.936 0.002* 0 0.38 
Full, Fire -0.950 0.264 0.763 0.023* 0 0.26 
Full, No Fire -1.113 0.273 0.806 0.015* 0 0.28 
Size-class distribution Slope: negative slope=healthy regeneration, SE Slope: standard error of slope, R2: closer to 
1=size-class explain variability of mean woody abundance, Significant SCD slope when p<0.05 indicated by *. PI: 
Permutation Index of 0=monotonic decline in SCD. SDI: Simpson’s Dominance Index >0.1=uneven size-class 
occupation. Treatments: Control=all herbivores were present, Partial=elephant and giraffe excluded, Full=all 
herbivores excluded. 
75 
The effect of herbivory (control site) and release from herbivory (in the partially and fully fenced 
exclosure) on the demography of the woody community revealed an inverse J-shaped 
distribution of the size-classes for all treatments (Figure 7.2a). The Simpson’s dominance index 
value indicated that the size classes of the woody community was unevenly distributed, 
although values are less than or close to 0.4, which is still within the ‘stable’ zone (Table 7.3). 
The permutation index values of 0 for all treatments indicated a monotonic decline in number of 
individuals present in each successive size-class (Table 7.3). Therefore, all three treatments 
hosted good regeneration, although the exclusion of elephant revealed the best regeneration. 
This was indicated by the significant negative slope and highest R2 value for the partially fenced 
exclosure compared to the values of the other treatments (Table 7.3). 
Quotient analyses of the complete woody community across herbivory treatments revealed 
uneven distributions for all communities, although highest stability was revealed for the woody 
community in absence of all herbivores (full exclosure) (Figure 7.2b). The SDI for this 
community was between 0.1 and 0.3, which further support community stability. 
Significant variance in mean woody abundances per plot was revealed for size classes 2, 3, 5 
and 6 across herbivore treatments. Post hoc tests revealed significantly more (p<0.05) 
individuals in these size classes in the full exclosure compared to the control site and partially 
fenced exclosure (Figure 7.2a). This explains the higher stability of woody regeneration in the 
absence of all herbivore activity (Figure 7.2b).  
 
76 
 
Figure 7.2: Size-class distributions (plant height in 1 m intervals) for mean number of woody 
individuals per plot (±SE) in successive size-classes (a) and quotients between size-classes (b) 
across herbivory treatments in 2015. Control: all herbivores were present; Partial: elephant & 
giraffe were excluded; Full: all herbivores were excluded. Significant differences from ANOVA 
statistics are indicated with *. 
 
Good regeneration was revealed for both the fire and no-fire treatments (Table 7.3a). Quotient 
analyses, however, revealed unstable woody communities in all three herbivore treatments, 
irrespective of fire treatment (Figure 7.3b). No significant effect of fire on woody abundances 
could be detected for any of the size-classes within the different herbivore treatments, with the 
exception of woody individuals between 4 m and 5 m (size class 5). Post hoc tests revealed 
significantly higher abundance (p=0.045) of woody individuals in the partially fenced exclosure 
where fire was absent (Figure 7.3).  
77 
 
Figure 7.3: Size-class distributions (plant height in 1 m intervals) for mean number of woody 
individuals per plot (±SE) in successive size-classes (a) and quotients between size-classes (b) 
across fire treatments in 2015. Control: all herbivores were present; Partial: elephant & giraffe 
were excluded; Full: all herbivores were excluded. Significant differences from ANOVA statistics 
are indicated with *. 
 
7.3.2. Population structure of preselected key species 
To investigate the population demography of key species, differentiation between fire 
treatments were not considered, due to limited evidence of a significant fire effect (see 7.3.1) 
and low numbers of individuals per species for each fire treatment respectively. Patterns 
observed through analyses therefore only represent herbivore treatment and catenal position 
effects. 
Riparian vegetation zone: 
When all measures of size-class distribution were considered (ordinary least square (OLS) 
regression, PI and SDI), regeneration of encroacher species Gymnosporia senegalensis and 
Spirostachys africana was favoured by the exclusion of all herbivores (Table 7.4). Size-class 
78 
distribution for individuals of these two species revealed R2 values above 0.9 in the fully fenced 
exclosure, which indicated that size-class explained more than 90% of the variability in number 
of individuals (Table 7.4). An inverse J-shaped SCD was displayed for G. senegalensis (Figure 
7.4a) with more individuals in the lower size classes, which was confirmed by a PI value of 0 
(Table 7.4). This was also evident from the quotient analyses. SCD of S. africana population in 
the fully fenced exclosure was altered from the J-shaped curve, which was supported by a PI 
value of 2 (monotonic decline was interrupted) (Table 7.4). Recruitment bottlenecks occurred in 
size-classes 2, 3, and 4. The S. africana population was less stable than G. senegalensis (lower 
R2 values, but higher p and PI values, Figure 7.4), although highest population stability was in 
the fully fenced exclosure (negative slope, R2 = 0.9, p<0.005, SDI = 0.21).  
Results from the ordinary least square (OLS) regression results indicated that regeneration of 
the encroacher Flueggea virosa was healthy in the partially fenced exclosure where only mega-
herbivores (in this experiment, elephant and giraffe) were excluded (Table 7.4). The negative 
slope of SCD together with a PI of 0 and SDI of 0.31 indicated an inverse J-shaped SCD slope 
(Table 7.4). Alteration in an inverse J-shaped SCD curve occurred in the presence (PI=2) and 
absence (PI=2) of all herbivores due to recruitment bottlenecks (Figure 7.4). Population stability 
decreased for individuals between 4 m and 5 m in the partially fenced exclosure.  
The non-encroacher species of the riparian vegetation zone were considerably less abundant 
than the encroacher species (Figure 7.4 and 7.5). SCD of Diospyros mespiliformis did not 
conform to the typical J-shaped curve across any of the three herbivore treatments (Figure 7.5). 
The PI values of all herbivore treatments were above 0 which indicated that SCD’s did not 
decline monotonically (Table 7.6). Despite a recruitment bottleneck imposed by low numbers of 
individuals in size-class 4 (PI=2) (Figure 7.5), the fully fenced exclosure hosted the most stable 
population for D. mespiliformis. Variability explained by size-classes for the full exclosure was 
94.2% (R2-value) with a significant SCD slope (p=0.001) (Table 7.4). Quotient analysis further 
revealed that the most stable population occurred in the fully fenced exclosure (Figure 7.5). 
Ziziphus mucronata had bad regeneration in all herbivore treatments with high PI values, low R2 
and p-values > 0.05 (Table 7.4), skewed size-class frequencies and fluctuating quotients 
(Figure 7.5). Instability was revealed for the population inside the full exclosure where no 
individuals in four of the six size-classes were recorded (positive SCD slope; PI=10, SDI=0.5 
(Table 7.4)).  
 
 
79 
Table 7.4: Summary of size-class distribution measures for the key woody encroacher and non-
encroacher species of the riparian vegetation zone in different herbivory treatments in 2015; 
Ordinary least square regression analyses (Slope, SE Slope, R2, p), Permutation Index (PI) and 
Simpson’s Dominance Index (SDI). 
Vegetation 
zone 
Ecological 
status 
Species Treatment Size class distribution 
Slope SE Slope R2 p PI SDI 
Riparian Encroachers Gymnosporia 
senegalensis 
Overall -2.080 0.499 0.813 0.014* 0 0.42 
Control -1.627 0.321 0.865 0.007* 0 0.41 
Partial -1.339 0.342 0.793 0.017* 0 0.32 
Full -1.829 0.297 0.905 0.004* 0 0.49 
Spirostachys 
africana 
Overall -0.597 0.069 0.949 0.001* 2 0.22 
Control -0.638 0.166 0.787 0.018* 2 0.21 
Partial -0.769 0.253 0.698 0.038* 2 0.26 
Full -0.518 0.077 0.919 0.003* 2 0.21 
Flueggea 
virosa 
Overall -1.942 0.670 0.678 0.044* 2 0.33 
Control -1.756 0.609 0.675 0.045* 2 0.34 
Partial -1.479 0.462 0.719 0.033* 0 0.31 
Full -1.511 0.442 0.745 0.027* 2 0.34 
Non-
encroachers 
Diospyros 
mespiliformis 
Overall -0.571 0.148 0.789 0.018* 4 0.21 
Control -0.710 0.240 0.687 0.041* 4 0.23 
Partial 0.115 0.156 0.119 0.503 12 0.11 
Full -0.580 0.072 0.942 0.001* 2 0.22 
Ziziphus 
mucronata 
Overall -0.445 0.415 0.224 0.343 4 0.28 
Control -0.307 0.235 0.299 0.262 4 0.13 
Partial -0.458 0.352 0.297 0.264 4 0.39 
Full 0.141 0.321 0.046 0.684 10 0.5 
Size-class distribution Slope: negative slope=healthy regeneration, SE Slope: standard error of slope, R2: closer to 1=size-class 
explain variability in number of individuals, Significant SCD slope when p<0.05 indicated by *. PI: Permutation Index of 
0=monotonic decline in SCD. SDI: Simpson’s Dominance Index >0.1=uneven size-class occupation. Treatments: Control=all 
herbivores present, Partial=elephant and giraffe excluded, Full=all herbivores excluded; Overall=across all three herbivore 
treatments.  
  
80 
Encroachers 
 
Figure 7.4: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Gymnosporia senegalensis, Spirostachys africana, and Flueggea virosa in successive size-
classes (a) and quotients for these individuals between size-classes (b) across herbivory 
treatments. Control: all herbivores present; Partial: elephant and giraffe excluded; Full: all 
herbivores excluded. Total/Overall: indicate SCD/Quotients of population, without distinguishing 
between herbivore treatments. 
 
 
 
 
 
81 
Non-encroachers 
 
Figure 7.5: Size-class distributions (plant height in 1 m intervals) for the abundance of Diospyros 
mespiliformis and Ziziphus mucronata in successive size-classes (a) and quotients for these 
individuals between size-classes (b) across herbivory treatments in 2015. Control: all herbivores 
present; Partial: elephant and giraffe excluded; Full: all herbivores excluded. Total/Overall: 
indicate SCD/Quotients of population, without distinguishing between herbivore treatments. 
 
Sodic vegetation zone: 
Vachellia grandicornuta, an encroaching species dominating the sodic vegetation zone, showed 
the best regeneration in the absence of mega-herbivores (Table 7.5). A monotonic decline in 
SCD were present for size-classes 1 to 5 followed by a slight increase in abundance in size-
class 6 (therefore PI=2) (Figure 7.6). It was also evident from the quotient analyses that the 
population in the partially fenced exclosure was more stable than populations in the control site 
and fully fenced exclosure (Figure 7.6). Recruitment bottlenecks were present in the fully fenced 
exclosure, indicated by the high PI value (PI=8, Table 7.5; Figure 7.6). 
Regression and SCD analyses of the encroacher species Rhigozum zambesiacum revealed 
revealed bad regeneration in all herbivore treatments (PI >2, SDI >0.4, fluctuating quotients and 
low R2 values). The total number of individuals in size-class 2 (saplings) were more than double 
than that in size-class 1 (seedlings) (Figure 7.6), which furthermore support that the R. 
zambesiacum population is unstable at this study site (Figure 7.6).  
82 
Only one non-encroacher species, Pappea capensis, had enough individuals to compare across 
herbivore treatments. As expected, the exclusion of all herbivores led to good regeneration 
(Slope=-1.554; R2=0.91; p=0.003) (Table 7.5) and a stable population structure (inverse J-
shaped curve, PI =0), although uneven SCD’s were revealed (SDI>0.4, Table 7.5, Figure 7.7). 
Little fluctuation occurred in quotients for this species in the full exclosure, suggesting 
population stability in the absence of herbivores (Figure 7.7). Herbivores inevitably led to 
unstable population structure of P. capensis, with or without elephant (Table 7.5, Figure 7.7). 
Seedling recruitment was high in all herbivore treatments, although seedling establishment, 
measured as abundances in size class 2 (i.e. from seedlings to saplings) was lower in die 
presence of herbivores, including elephant.  
 
Table 7.5: Summary of size-class distribution measures for the key woody encroacher and non-
encroacher species of the sodic vegetation zone in different herbivory treatments in 2015; 
Ordinary least square regression analyses (Slope, SE Slope, R2, p), Permutation Index (PI) and 
Simpson’s Dominance Index (SDI). 
Vegetation 
zone 
Ecological 
status 
Species Treatment Size class distribution 
Slope SE Slope R2 p PI SDI 
Sodic Encroachers Vachellia 
grandicornuta 
Overall -0.960 0.216 0.832 0.011* 4 0.28 
Control -1.008 0.131 0.936 0.002* 2 0.42 
Partial -1.202 0.104 0.971 0.000* 2 0.4 
Full -0.615 0.391 0.382 0.191 8 0.26 
Rhigozum 
zambesiacum 
Overall -1.618 0.504 0.720 0.033* 2 0.42 
Control -0.652 0.304 0.534 0.099 6 0.71 
Partial -1.211 0.313 0.789 0.018* 4 0.52 
Full -1.495 0.546 0.652 0.052 2 0.48 
Non-
encroachers 
Pappea 
capensis 
Overall -1.463 0.158 0.955 0.001* 0 0.48 
Control -1.339 0.249 0.878 0.006* 2 0.7 
Partial -1.008 0.183 0.884 0.005* 2 0.37 
Full -1.554 0.244 0.910 0.003* 0 0.45 
Size-class distribution Slope: negative slope=healthy regeneration, SE Slope: standard error of slope, R2: closer to 1=size-class 
explain variability in number of individuals, Significant SCD slope when p<0.05 indicated by *. PI: Permutation Index of 
0=monotonic decline in SCD. SDI: Simpson’s Dominance Index >0.1=uneven size-class occupation. Treatments: Control=all 
herbivores present, Partial=elephant and giraffe excluded, Full=all herbivores excluded; Overall=Population across all three 
herbivore treatments. 
 
 
 
 
 
83 
Encroachers 
 
Figure 7.6: Size-class distributions (plant height in 1 m intervals) for the abundance of Vachellia 
grandicornuta and Rhigozum zambesiacum in successive size-classes (a) and quotients for these 
individuals between size-classes (b) across herbivory treatments in 2015. Control: all herbivores 
present; Partial: elephant and giraffe excluded; Full: all herbivores excluded. Total/Overall: 
indicate SCD/Quotients of population, without distinguishing between herbivore treatments. 
 
Non-encroacher 
 
Figure 7.7: Size-class distributions (plant height in 1 m intervals) for the abundance of Pappea 
capensis in successive size-classes (a) and quotients for these individuals between size-classes 
(b) across herbivory treatments in 2015. Control: all herbivores present; Partial: elephant and 
giraffe excluded; Full: all herbivores were excluded. Total/Overall: indicate SCD/Quotients of 
population, without distinguishing between herbivore treatments. 
 
84 
Crest vegetation zone: 
Regeneration of Dichrostachys cinerea was most stable in the control site (Table 7.6; Figure 
7.8). This was indicated by a significant negative SCD slope (p=0.039; Slope=-1.59) together 
with a high R2 value, a low PI value (2) and SDI values between 0.1 and 0.2 (Table 7.6). 
Although D. cinerea was most abundant in the fully fenced exclosure (Figure 7.8), its negative 
SCD slope was insignificant (p=0.119) and had a PI of 4 (Table 7.6), which indicated 
recruitment bottlenecks. 
Regeneration of Combretum apiculatum did not indicate good regeneration in any herbivore 
treatments as revealed by OLS regression analyses (Table 7.6). This was also confirmed by the 
high PI for populations in all three treatments (Table 7.6). The partially fenced exclosure hosted 
the most stable population with a steep negative SCD slope (-0.744, although insignificant with 
p=0.065) and highest R2 value (0.613) (Table 7.6; Figure 7.8).  
Demography of the non-encroacher Senegalia nigrescens revealed abundant seedlings (size-
class 1) followed by over 70% decline in sapling individuals (size-class 2) across all three 
herbivore treatments (Figure 7.9). Good regeneration was not revealed for any of the three 
herbivore treatments (PI>0 and unstable quotients) (Table 7.6; Figure 7.9). The control site 
hosted the most stable population (slope=-1.518; R2=0.876; p=0.006; PI=2; SDI=0.25) (Table 
7.6). 
Regeneration of Vachellia exuvialis was good in the control site (PI=0, uneven SCD’s 
(SDI=0.48)), conforming to an inverse J-shaped curve (slope=-1.784; R2=0.852; p=0.009) 
(Table 7.6; Figure 7.9). Fluctuation of quotients was however present in all three herbivore 
treatments (Figure 7.9).  
 
 
 
 
 
 
85 
Table 7.6: Summary of size-class distribution measures for the key woody encroacher and non-
encroacher species of the crest vegetation zone in different herbivory treatments in 2015; 
Ordinary least square regression analyses (Slope, SE Slope, R2, p), Permutation Index (PI) and 
Simpson’s Dominance Index (SDI). 
Vegetation 
zone 
Ecological 
status 
Species Treatment Size class distribution 
Slope SE Slope R2 p PI SDI 
Crest Encroachers Dichrostachys 
cinerea 
Overall -1.799 0.779 0.571 0.082 0 0.27 
Control -1.590 0.526 0.696 0.039* 2 0.31 
Partial -1.794 0.666 0.645 0.054 4 0.35 
Full -1.621 0.818 0.495 0.119 4 0.29 
Combretum 
apiculatum 
Overall -0.573 0.300 0.477 0.129 4 0.22 
Control -0.471 0.351 0.310 0.251 6 0.22 
Partial -0.744 0.295 0.613 0.065 2 0.25 
Full -0.507 0.239 0.530 0.101 6 0.2 
Non-
encroachers 
Senegalia 
nigrescens 
Overall -1.327 0.246 0.879 0.006* 4 0.65 
Control -1.518 0.285 0.876 0.006* 2 0.82 
Partial -1.161 0.268 0.824 0.012* 4 0.63 
Full -1.111 0.203 0.882 0.005* 4 0.55 
Vachellia 
exuvialis 
Overall -1.392 0.363 0.786 0.019* 2 0.37 
Control -1.784 0.372 0.852 0.009* 0 0.48 
Partial -1.115 0.347 0.721 0.032* 4 0.32 
Full -0.816 0.310 0.635 0.058 6 0.24 
Size-class distribution Slope: negative slope=healthy regeneration, SE Slope: standard error of slope, R2: closer to 1=size-class 
explain variability in number of individuals, Significant SCD slope when p<0.05 indicated by *. PI: Permutation Index of 
0=monotonic decline in SCD. SDI: Simpson’s Dominance Index >0.1=uneven size-class occupation. Treatments: Control=all 
herbivores present, Partial=elephant and giraffe excluded, Full=all herbivores excluded; Overall=Population across all three 
herbivore treatments. 
 
  
86 
Encroachers 
 
Figure 7.8: Size-class distributions (plant height in 1 m intervals) for the abundance of 
Dichrostachys cinerea and Combretum apiculatum in successive size-classes (a) and quotients 
for these individuals between size-classes (b) across herbivory treatments in 2015. Control: all 
herbivores were present; Partial: elephant and giraffe excluded; Full: all herbivores were 
excluded. Total/Overall: indicate SCD/Quotients of population, without distinguishing between 
herbivore treatments. 
  
87 
Non-encroachers 
 
Figure 7.9: Size-class distributions (plant height in 1 m intervals) for the abundance of Senegalia 
nigrescens and Vachellia exuvialis in successive size-classes (a) and quotients for these 
individuals between size-classes (b) across herbivory treatments in 2015. Control: all herbivores 
present; Partial: elephant and giraffe excluded; Full: all herbivores were excluded. Total/Overall: 
indicate SCD/Quotients of population, without distinguishing between herbivore treatments. 
 
Regeneration of encroacher species was better in the absence of herbivores (either only 
elephant or all herbivores) (Table 7.7). Response of non-encroacher regeneration was species 
specific. Regeneration of broad-leaved palatable species was better where elephants or all 
herbivores were excluded, while microphyllous palatable species were adapted to herbivores 
(through spinescent traits) and had better regeneration in the presence of herbivores (Table 
7.7). 
 
 
 
 
 
88 
Table 7.7: Summary of the herbivore effect on pre-selected woody species demography (i.e. 
regeneration) at the Nkuhlu exclosure in 2015. 
Vegetation zone Species Ecological status Treatment with healthy 
regeneration 
Riparian Gymnosporia senegalensis Encroacher Full exclosure 
 Spirostachys africana Encroacher None; aFull exclosure 
 Flueggea virosa Encroacher Partial exclosure 
 Diospyros mespiliformis Non-encroacher None; aFull exclosure 
 Ziziphus mucronata Non-encroacher None; aPartial exclosure 
Sodic Vachellia grandicornuta Encroacher Partial exclosure 
 Rhigozum zambesiacum Encroacher None; aPartial exclosure 
 Pappea capensis Non-encroacher Full exclosure 
Crest Dichrostachys cinerea Encroacher None; aControl site 
 Combretum apiculatum Encroacher None; aPartial exclosure 
 Senegalia nigrescens Non-encroacher None; aControl site 
 Vachellia exuvialis Non-encroacher Control site 
Herbivore treatments: Control site=all herbivores present, Partial exclosure=elephant and giraffe excluded, Full 
exclosure=all herbivores excluded; a treatment with most stable population. 
 
7.4 Discussion 
This study revealed strong evidence of positive community responses to herbivore exclusion. All 
herbivore treatments hosted significant SCD slopes, but the absence of all herbivore activity for 
13 years resulted in a more stable woody community, especially in higher size-classes. These 
increases in woody SCD stability was ascibed to a higher woody abundance in higher size-
classes. Therefore, where trees were released from browsing suppression, i.e. browse-trap, 
more woody individuals could grow into higher size-classes (Staver & Bond, 2014). It is still 
unclear whether the patterns observed are a possible mega-herbivore (specifically elephant) 
effect, which have previously been considered a main attribute to woody structure change 
(Eckhardt et al., 2000; Whyte et al., 2003; Wiseman et al., 2004; Asner et al., 2009), or a meso-
herbivore (i.e. impala, kudu, nyala etc.) effect. Recent studies found evidence that meso-
herbivores have an equal or even a stronger effect than mega-herbivores on woody population 
structure in savanna ecosystems (Moe et al., 2009; O’Kane et al., 2012). This study revealed 
that meso-herbivores were responsible for instability in higher size-classes. In the presence or 
absence of elephants, populations were unstable due to instability in numbers in the size 
classes >4 m. Although meso-herbivores do not directly affect higher size-classes (>2 m) 
through browsing, they created a browse-trap in lower size-classes which over a long-term 
period inhibited woody growth (Prins & Van der Jeugd, 1993; Augustine & McNaughton, 2004; 
Moe et al., 2009). This browse-trap is evident in the quotient results of the partially fenced 
exclosure. Exclusion of elephants assumingly led to increased meso-herbivore activity (Fritz et 
89 
al., 2002; Valeix et al., 2008; Hilbers et al., 2015; Lagendijk et al., 2015), which prompted 
instability between size class 1 and 2. Seedlings (size-class 1) were much more abundant than 
saplings (size-class 2). This severe drop in abundance indicated a bottleneck in recruitment of 
seedlings (Lykke, 1998; Prior et al., 2009). These size-classes were directly subjected to 
utilisation by browsers, which inhibited woody growth past seedling stage. Only those species 
able to tolerate browsing (either they were unpalatable or fast growing (Fornara & Du Toit, 
2008)), have grown into higher size-classes. This could have resulted in the instability noted in 
higher size-classes in both the control site (presence of elephant) and partially fenced exclosure 
(absence of elephant).  
The fire treatments applied at the Nkuhlu exclosure revealed no significant effect on the woody 
community demography. The fire-mediated recruitment bottleneck hypothesis states that fire 
disturbances restrict the number of juveniles growing past the seedling size class to become fire 
tolerant trees (Bond, 2008; Lehman et al., 2009). This was not observed in the results from the 
fire treatments in this specific experimental setting (Nkuhlu exclosures), which revealed an 
inverse J-shaped curve, with the highest abundance of individuals in the seedling height class, 
in both the fire and no fire treatments. Herbivory is therefore suggested as the main driver of 
woody community structure in this particular savanna ecosystem. This corresponds with 
Sankaran et al. (2013) who reported strong herbivore control of woody vegetation, even in the 
absence of fire. This is especially true for arid and semi-arid savannas (such as this study site) 
and not mesic savannas, where fire has a substantial effect on woody recruitment (Sankaran et 
al., 2013). 
Species-specific demographic analyses revealed unique responses to herbivory and the 
exclusion thereof, irrespective of being an encroacher or non-encroacher species. The different 
vegetation zones along the catenal sequence are known to host unique assemblages of plant 
species and vegetation structure (Siebert & Eckhardt, 2008). The riparian zone on the banks of 
the Sabie River, which is also considered nutrient-rich, is particularly favoured by herbivores 
due to access to water and nutritious vegetation (Scholes, 1990). The presence of herbivores in 
this zone resulted in the suppression of key encroacher and non-encroacher regeneration. For 
the encroacher populations Spirostachys africana and Flueggea virosa were suppressed by 
presence of all mammalian herbivores (mega- and meso-herbivores), while Gymnosporia 
senegalensis was suppressed by especially meso-herbivores (absence of mega-herbivores). 
Therefore, encroachment by these species could be controlled by the presence of herbivores. 
However, results from this study indicated that herbivory is likely to cause negative regeneration 
responses of the desirable non-encroaching key species Diospyros mespiliformis. All four these 
key species are commonly browsed by herbivores (Sauer et al., 1977; Bowland, 1990; Pooley, 
90 
1997; Watson & Owen-Smith, 2002; Makhabu et al., 2006; Scogings et al., 2012; Penderis & 
Kirkman, 2014; Wigley et al., 2014). 
The non-encroacher Ziziphus mucronata is a palatable woody species (Owen-Smith & Cooper, 
1987) highly favoured by both meso- (Sauer et al., 1977; Bowland, 1990; Pooley, 1997) and 
mega-herbivores (Owen-Smith & Cooper, 1987; Wiseman et al., 2004). The population of this 
species was in general unstable, due to low numbers of individuals recorded across the 
treatments. In the presence of all herbivores and the absence of only mega-herbivores, 
regeneration of Z. mucronata was better compared to where all herbivores were excluded. This 
species is generally fast-growing and could therefore be expected to recruit into higher size-
classes despite utilisation by herbivores (Rooke et al., 2004). Poor recruitment was revealed for 
the population inside the fully fenced exclosure. Sufficient number of individuals between 5 m 
and 6 m present in 2015 could be ascribed to rapid recruitment (Rooke et al., 2004) after fences 
of the exclosure have been raised. The lack of individuals in the lower size-classes could be the 
result of competition with increased herbaceous biomass (Sankaran et al., 2008).  
The sodic vegetation zone is known its nutrient rich vegetation associated with high herbivore 
activity (Grant & Scholes, 2006). These areas are commonly characterized by grazing lawns 
and open sites, and overutilization by herbivores can be expected in these nutrient ‘hotspots’ 
(Grant & Scholes, 2006; Khomo & Rogers, 2005). The non-encroacher Pappea capensis, which 
is a palatable woody species, mainly utilized by herbivores such as elephant, giraffe, kudu, 
nyala and bushbuck (Owen-Smith, 2010; Schmitt et al., 2016), indicated good regeneration in 
the absence of all herbivores. Regeneration of the two encroacher species Rhigozum 
zambesiacum and Vachellia grandicornuta were favoured by the presence of meso-herbivores 
and absence of mega-herbivores (partially fenced exclosure). Mega-herbivores, specifically 
elephant was therefore identified as the browsing guild to control regeneration of these two 
encroacher species, which correlated with the study done by Scogings et al. (2012) that 
recognised elephant effects on these two species through tree felling, debarking, branch 
breaking and complete uprooting. 
On the dystrophic, drier uplands (the so-called crest zone), the well-known encroacher 
Dichrostachys cinerea revealed the most stable population in the presence of all herbivores, 
which corresponds with the study of Roques et al. (2001) where a positive relationship occurred 
between D. cinerea encroachment and grazing pressure. Where all herbivores were released, 
an increase in abundance (and therefore cover) was noticed for this shrub across the size-
classes (Levick et al., 2009) as well as a bottleneck in the lower size-classes. This could be 
explained by increased interspecific competition with the herbaceous layer (Sankaran et al., 
2008) as well as for intraspecific competition, or self-thinning. Encroachment by this species will 
therefore be difficult to control because of a more stable population in the presence of 
91 
herbivores but an increase in abundance (leading to densification) where herbivores have been 
excluded. Considering these two challenges, it is suggested that the presence of herbivores, 
and especially elephant would be the best option to control D. cinerea encroachment because it 
will suppress densification of the species across the size-classes.  
Combretum apiculatum, also considered an encroacher species at this study site, did not exhibit 
significantly goodregeneration in any herbivore treatment. The most stable population was 
located in the partially fenced exclosure, where mega-herbivores were excluded. In their study, 
Scogings et al. (2012) noted that C. apiculatum was one of their studied species that was most 
affected by elephants. This explains why elephant exclusion in this study favoured regeneration 
of C. apiculatum. It could furthermore be concluded that regeneration of C. apiculatum is 
stimulated by the presence of especially meso-herbivores because in the absence of all 
herbivores, recruitment of this species was suppressed. This can be due to competition with the 
herbaceous layer or with the intense increase in D. cinerea individuals, as hypothesized by 
Scogings et al. (2012). 
Senegalia nigrescens is a well-known keystone forage species in the Kruger National Park. This 
non-encroacher species is well adapted to browsing by developing both tolerance (high 
regrowth rates, extensive branching) and resistance (close thorn spacing) traits (Fornara & Du 
Toit, 2007). The population demography for S. nigrescens revealed evidence for this adaptation 
through indicating highest stability in the presence of all herbivores. High numbers of seedlings 
(size-class 1) were present followed by a sharp decline in number of individuals in size-class 2. 
According to Lykke (1998), this decline can be indicative of a recruitment bottleneck. This 
pattern in SCD was present in all three herbivore treatments, and therefore the bottleneck in the 
control site and partially fenced treatment can be attributed to a meso-herbivore induced 
browse-trap, while in the fully fenced treatment it was the herbaceous layer that imposed a 
recruitment filter (Sankaran et al., 2008) in which most individuals were trapped.  
Vachellia exuvialis, also a non-encroacher species that is commonly browsed upon by 
herbivores, such as black rhino and elephants (Scogings et al., 2012), revealed most stable 
regeneration in the presence of all herbivores, whereas the fully fenced treatment hosted very 
few individuals overall in each size-class. This suggests that V. exuvialis can be adapted to 
herbivore activity (especially meso-herbivores) and mega-herbivores do not explicitly affect their 
regeneration. 
 
92 
7.5 Conclusion 
Healthy, on-going regeneration is evident for the complete woody community at the Nkuhlu 
exclosures site, regardless of the different herbivore and fire treatments. Results revealed that 
13 years of all mammalian herbivore exclusion had significant effects on the abundances of 
woody individuals in the successive size-classes, hosting a woody community demographically 
more stable than communities in the control site and partially fenced exclosure. The instabilities 
revealed for woody community demography of the partial exclosure suggest a meso-herbivore 
effect, which imposed a recruitment bottleneck in the form of a browse trap. Fire did not have a 
profound effect in this specific experimental setup, which revealed that herbivory is the main 
contributor to woody population structure in this study site. 
Herbivory is considered as an encroachment control in savanna ecosystems by suppressing 
woody growth as well as densification. Therefore, it was expected that the absence of herbivore 
activity, whether only elephant or all herbivores would contribute to stable regeneration of 
encroacher species. This was true for most of the encroacher species. Results suggest that 
regeneration of G. senegalensis and S. africana was suppressed by meso- and mega-
herbivores, and therefore the absence of all herbivores contributed to more stable populations. 
Populations of F. virosa, V. grandicornuta, R. zambesiacum, and C. apiculatum experienced 
regeneration suppression by mega-herbivores (elephants). This was observed in the more 
stable populations where only mega-herbivores were excluded. These species could therefore 
be a problem on game farms located in similar savanna ecosystems where no elephants are 
present to control their regeneration. 
The only encroacher species that did not conform to the hypothesis, which states that herbivore 
exclusion will cause stable regeneration of encroacher species, was D. cinerea. Its population 
structure was less stable in the absence of herbivores, while the presence of all herbivores led 
to more stability in its regeneration. Although the population structures of this species indicated 
that release from all herbivores negatively affected its SCD, this same scenario also led to an 
increase in abundance of these individuals. This, in turn, could negatively affect regeneration of 
other woody species. It is therefore suggested that encroachment by D. cinerea cannot be 
controlled by herbivores only. 
Regeneration of three desirable non-encroaching species (Diospyros mespiliformis, Pappea 
capensis and Ziziphus mucronata) was sensitive to herbivory, with or without elephants, while 
two species, Senegalia nigrescens and Vachellia exuvialis, were adapted to herbivory. The two 
species adapted to herbivory will most likely, in the long term, be more abundant than the other 
key non-encroaching species in this semi-arid savanna area because of their ability to 
regenerate despite presence of herbivores.  
93 
The selected species are only a few of the sensitive key non-encroaching species located in the 
southern parts of Kruger National Park. Other important key species include Sclerocarya birrea, 
Lannea schweinfurthii and Comretum imberbe. Unfortunately, due to limited number of 
individuals across the herbivore treatments in this particular study site, their regeneration status 
could not be assessed, but should be considered in population demography studies. 
  
94 
CHAPTER 8 
Summary and general recommendations 
8.1 Main findings 
Floristic composition assessment revealed both herbivore and fire effects. Findings 
corresponded and contributed to the understanding that herbivores and fire are key 
determinants of savanna structure. After 13 years of herbivore and fire manipulations in this 
savanna system, changes occurred in the woody floristic composition, which supported the 
hypothesis that herbivore and fire exclusion will cause alteration in floristic composition. It was 
evident that the catena interplayed effects of herbivory and fire on the floristics. The eutrophic 
bottomlands were mostly affected by herbivores in terms of changes in dominant plant families 
and traits, while the dystrophic uplands were more susceptible to fire-induced changes.  
Herbivore exclusion affected woody species composition when seedling and established 
communities were compared within the same year. The hypothesis tested stated that exclusion 
of herbivore activity will cause seedling and established communities to be more similar in 
species composition. This hypothesis was supported by the results from this study. It was 
derived from the results that the presence of herbivores contributed to dissimilar seedling and 
established woody community assemblages. The release of herbivores is therefore suggested 
to create a more similar woody community. 
The demographic bottleneck model of savanna systems takes in consideration herbivory, fire 
and the herbaceous layer to create recruitment bottlenecks in woody regeneration. In this study, 
results managed to disentangle the effects of meso- and mega-herbivores in combination with 
fire on encroaching and non-encroaching species. Recruitment bottlenecks were used to 
interpret many of the patterns observed in the results.  
Comparison of seedling and established individual abundances of both encroacher and non-
encroacher species revealed that herbivores play an important role in structuring the woody 
layer. Herbivory negatively affected recruitment of both encroacher and non-encroacher 
species. Elephants primarily impacted woody abundance, especially in the riparian- and on the 
crest vegetation zones. It was expected that elephants would affect established individuals 
more than seedling abundances. Results, however, indicated that elephants affected both these 
development stages. Meso-herbivore effects were documented in the absence of elephant, 
especially in the riparian vegetation zone. These results confirmed the hypothesis that 
elephants indirectly facilitate healthy regeneration through displacement of meso-herbivores. 
These conclusions were made after results revealed bad woody regeneration trends in the 
95 
absence of elephants. The hypothesis that exposure to herbivory and fire will not negatively 
affect encroacher species abundances, but rather abundances of non-encroachers, was 
partially supported by the results obtained in this study. Abundance of encroaching seedling 
species increased in the presence of all herbivores and decreased in the absence of only 
elephants. Non-encroaching seedling abundances were not significantly reduced by meso-
herbivore activity and fire.  
Demography analyses in terms of size-class distributions revealed that herbivore exclusion 
positively affected regeneration of the woody community at the study site. Exclusion of only 
elephant led to good regeneration patterns, while exclusion of all herbivores resulted in the most 
stable woody community. Fire did not have a profound effect on size-class distributions in this 
experimental setup, and therefore herbivory was identified as the main contributor to woody 
population structure in this savanna type.  
The hypothesis that herbivore exclusion will enhance regeneration and population stability of 
encroacher species was supported by the results, which revealed that herbivory by both meso- 
and mega-herbivores suppressed the regeneration of key encroacher species. Meso-herbivores 
were responsible for the suppression of Gymnosporia senegalensis and Spirostachys africana, 
while mega-herbivores suppressed Flueggea virosa, Vachellia grandicornuta, Rhigozum 
zambesiacum and Combretum apiculatum. The hypothesis was, however, not supported for 
Dichrostachys cinerea populations. The presence of all herbivores enhanced population stability 
of Dichrostachys cinerea, while the absence of all herbivores led to higher abundances by this 
species. Encroachment by Dichrostachys cinerea can therefore not be controlled by herbivory 
alone. Future studies in similar semi-arid savanna ecosystems, where fire is present, should 
focus on Dichrostachys cinerea demography to determine whether the combination of herbivory 
and fire is suitable to control regeneration. 
The demography of three palatable non-encroaching species, Diospyros mespiliformis, Pappea 
capensis and Ziziphus mucronata, revealed sensitivity to herbivore exposure, while Senegalia 
nigrescens and Vachellia exuvialis were adapted to herbivore activity. It can therefore be 
expected that the latter two species would be more successful in a savanna ecosystem 
exposed to herbivory than D. mespiliformis, P. capensis and Z. mucronata. 
The results of this study support the disturbance-based explanation of savanna dynamics. In 
particular it support the disequilibrium dynamic argument which assume that competition 
between trees and grasses occur at all life-stages and disturbances prevent dominant life-form 
transitions (Van Langevelde et al., 2011). Competition between trees and grasses was noticed 
in the fully fenced exclosure, where high biomass could explain suppressed recruitment of 
96 
woody seedlings. In the presence of herbivores, where herbaceous biomass is low, 
disturbances such as herbivory and fire prevented woody recruitment.  
Facilitation of woody recruitment is a topic of interest for savanna ecology. Results from this 
study revealed evidence that the catena played an important role in structuring and facilitating 
regeneration of the woody layer. The effect of facilitation by nursing sites is however still 
unexplored in this study site. 
 
8.2 Recommendations for future studies 
Future studies on regeneration of woody species in semi-arid savanna ecosystems exposed to 
herbivory and fire could consider the following: 
 Test the effects of herbivores on woody species in the sodic vegetation zone, as this 
zone is suggested to be a limited elephant-browsing zone; they seem to use this area as 
a corridor between the upland crest and riparian zones.  
 Investigation of the nutritious value of woody plants in the sodic vegetation zone. 
 It is expected that elephant effects on woody abundance (without fire) can possibly be 
analogous to that of the combined effects of meso-herbivores and fire. This can 
especially be true for effects on encroaching woody species. 
 Regeneration of D. cinerea should be studied to determine whether the combination of 
herbivore presence (both meso-and mega-herbivores), as well as fire exposure could 
control encroachment by this species. 
 Demography studies in southern parts of Kruger National Park should specifically 
address regeneration status of desirable non-encroaching species such as Sclerocarya 
birrea, Lannea schweinfurthii, Combretum imberbe and Commiphora species, since 
elephant selectively feed on these species. 
 Test the hypothesis that abiotic amelioration is stronger than facilitation (protection 
against herbivory) where water and nutrients are limited, i.e. on the upland crests, but 
protection from herbivory in the form of nursing sites is expected to assist in good 
recruitment on nutrient-rich sites, such as the sodic zone. 
  
97 
References 
 
Angassa, A. & Baars, R.M. 2000. Ecological condition of encroached and non-encroached 
rangelands in Borana, Ethiopia. African Journal of Ecology 38:321–328. 
Anthelme, F. & Michalet, R. 2009. Grass-to-tree facilitation in an arid grazing environment (Air 
Mountains, Sahara). Basic and Applied Ecology 10:437-446.  
Archer, S., Schimel, D.S. & Holland, E.A. 1995. Mechanisms of shrubland expansion: land use, 
climate or CO2? Climate Change 29:91–99. 
Asner, G.P., Archer, S., Hughes, R.F., Ansley, R.J. & Wessman, C.A. 2003. Net changes in 
regional woody vegetation cover and carbon storage in Texas drylands, 1937-99. Global 
Change Biology 9, 316–355. 
Asner, G.P., Levick, S.R., Kennedy-Bowdoin, T., Knapp, D.E., Emerson, R., Jacobson, J., 
Colgan, M.S. & Martin, R.E. 2009. Large scale impacts of herbivores on the structural diversity 
of African savannas. PNAS 106(12):4947-4952. 
Asner, G.P., Vaughn, N., Smit, I.P.J. & Levick, S. 2016. Ecosystem-scale effects of mega-fauna 
in African savannas. Ecography 39:240-252. 
Augustine, D.J. & McNaughton, S.J. 1998. Ungulate effects on the functional species 
composition of plant communities: Herbivore selectivity and plant tolerance. The Journal of 
Wildlife Management 62(4):1165-1183. 
Augustine, D.J. & McNaughton, S.J. 2004. Regulation of shrub dynamics by native browsing 
ungulates on East African rangeland. Journal of Applied Ecology 41:45-58. 
Baker, P.J., Bunyavejchewin, S., Oliver, C.D. & Ashton, P.S. 2005. Disturbance history and 
historical stand dynamics of a seasonal tropical forest in western Thailand. Ecological 
Monographs 75(3):317–343. 
Baldeck, C.A., Colgan, M.S., Feret, J.B., Levick, S.R., Martin, R.E. & Asner, G.P. 2014. 
Landscape-scale variation in plant community composition of an African savanna from airborne 
species mapping. Ecological Applications 24(1):84-93. 
Barton, K.E. & Hanley, M.E. 2013. Seedling-herbivore interactions: insights into plant defence 
and regeneration patterns. Annals of Botany 112:643-650. 
98 
Bayakagaba, P., Eilu, G., Bosco, J., Okullo, L., Tumwebaze, S.B. & Mwavu, E.N. 2011. 
Population structure and regeneration status of Vitellaria paradoxa (C.F. Gaertn.) under 
different land management regimes in Uganda. Agricultural Journal 6(1):14-22. 
Belsky, A.J. 1984. The role of small browsing mammals in preventing woodland regeneration in 
the Serengeti National Park, Tanzania. African Journal of Ecology 22:271–279. 
Berthrong, S.T., PIneiro, G., Jobb_agy, E.G. & Jackson, R.B. 2012. Soil C and N changes with 
afforestation of grasslands across gradients of precipitation and plantation age. Ecological 
Applications 22:76–86. 
Bloor, J.M.G., Leadley, P.W. & Barthes, L. 2008. Responses of Fraxinus excelsior seedlings to 
grass-induced above- and below-ground competition. Plant Ecology 194:293–304. 
Bond, W.J. & van Wilgen, B.W. 1996. Fire and Plants. Chapman & Hall, London. 
Bond, W.J. & Midgley, G.F. 2000. A proposed CO2-controlled mechanism of woody plant 
invasion in grasslands and savannas. Global Change Biology 6:865–869. 
Bond, W.J., Smythe, K. & Balfour, D.A. 2001. Acacia species turnover in space and time in an 
African savanna. Journal of Biogeography 28:117-128. 
Bond, W.J. & Keeley, J.E. 2005. Fire as a global ‘herbivore’: the ecology and evolution of 
flammable ecosystems. Trends in Ecology and Evolution 20(7):387-394. 
Bond, W. J., Woodward, F. I. & Midgley, G. F. 2005. The global distribution of ecosystems in a 
world without fire. New Phytologist 165:525–538. 
Bond, W.J. & Midgley, G.F. 2012. Carbon dioxide and the uneasy interaction of trees and 
savannah grasses. Philosophical Transactions of the Royal Society B: Biological Sciences 
367:601–612. 
Bond, W.J. 2008. What limits trees in C4 grasslands and savannas? Annual Review of Ecology, 
Evolution, and Systematics 39:641–659. 
Botha, J., Witkowski, E.F.T. & Shackleton, C.M. 2004. Impact of communal harvesting of 
Warburgia salutaris (‘pepper-bark tree’) in Mpumalanga, South Africa. Biodiversity conservation 
13:1675-1698.  
Boundja, R.P. & Midgley, J.J. 2010. Patterns of elephant impact on woody plants in the 
Hluhluwe-Imfolozi park, Kwazulu-Natal, South Africa. African Journal of Ecology 48(1):206-214. 
99 
Bowland AE. 1990. The response of red duikers Cephalophus natalensis to drive counts. 
Koedoe 33: 47–53 
Brits, J., Van Rooyen, M.W. & Van Rooyen, N. 2002. Ecological impact of large herbivores on 
the woody vegetation at selected watering points in the eastern basaltic soils in the Kruger 
National Park. African Journal of Ecology 40(1):53-60. 
Brooker, R.W., Maestre, F.T. & Callaway et al. 2008. Facilitation in plant communities. The past, 
the present, and the future. Journal of Ecology 96:18-34. 
Brown, J.R. & Archer, S.R. 1987. Woody plant seed dispersal and gap formation in a North 
American subtropical savanna fauna woodland: the role of domestic herbivores. Vegetation 
73:73-80. 
Buitenwerf, R., Bond, W.J., Stevens, N. & Trollope, W.S.W. 2012. Increased tree densities in 
South African savannas: >50 years of data suggests CO2 as a driver. Global Change Biology 
18:675–684. 
Burns, K.C. 2014. Are there general patterns in plant defence against megaherbivores? 
Biological Journal of the Linnean Society Impact Factor 111(1):38–48. 
Bryant, J.P., Reichardt, P.B. & Clausen, T.P. 1992. Chemically mediated interactions between 
woody plants and browsing mammals. Journal of Range Management 45(1):18–24. 
Cabral, A., De Miguel, J., Rescia, A., Schmitz, M. & Pineda, F. 2003. Shrub encroachment in 
Argentinean savannas. Journal of Vegetation Science 14:145–152. 
Casillo, J., Kunst, C. & Semmartin, M. 2012. Effects of fire and water availability on the 
emergence and recruitment of grasses, forbs and woody species in a semi-arid Chaco savanna. 
Australian Ecology 37:452-459. 
Casper, B.B. & Jackson, R.B., 1997. Plant competition underground. Annual Review of Ecology 
and Systematics 28:545–570. 
Charles-Dominique, T., Davie, T.J., Hempson, G.P., Bezeng, B.S., Daru, B.H., Kabongo, R.M., 
Maurin, O., Muasya, A.M., Van der Bank, M. & Bond, W.J. 2016. Spiny plants, mammal 
browsers, and the origin of African savannas. PNAS 113(38):E5572-E5579. 
Clarke, K.R. 1993. Non-parametric multivariate analyses of changes in community structure. 
Australian Journal of Ecology 18:117-143. 
100 
Clarke, P.J., Lawes, M.J., Midgley, J.J., Lamont, B.B., Ojeda, F., Burrows, G.E., Enright, N.J. & 
Knox, K.J.E. 2013. Resprouting as a key functional trait: how buds, protection and resources 
drive persistence after fire. New Phytologist 197(1):19-35. 
Colgan, M.S., Asner, G.P., Levick, S.R., Martin, R.E. & Chadwick, O.A. 2012. Topo-edaphic 
controls over woody plant biomass in South African savannas. Biogeosciences 9:1809–1821. 
Condit, R., Sukumar, R., Hubbell, S.P. & Foster, R.B. 1998. Predicting population trends from 
size distributions: a direct test in a tropical tree community. The American Naturalist 152:495-
509. 
Coomes, D.A. & Allen, R.B. 2007. Mortality and tree-size distributions in natural mixed-age 
forests. Journal of Ecology 95(1):27–40. 
Cramer, M.D., Chimphango, S.B.M., Van Cauter, A.,Waldram, M.S. & Bond,W.J. 2007. Grass 
competition induces N 2 fixation in some species of African Acacia. Journal of Ecology 
95:1123–1133. 
Cramer, M.D., Wakeling, J.L. & Bond, W.J. 2012. Belowground competitive suppression of 
seedling growth by grass in an African savanna. Plant Ecology 213:1655–1666. 
Cramer,M.D., Van Cauter, A. & Bond, W.J. 2009. Growth of N2-fixing African savanna Acacia 
species is constrained by below-ground competition with grass. Journal of Ecology 98:156–167. 
Cumming, D.H.M. 1982. The influence of large herbivores on savanna structure in Africa. In: 
Huntley, B.J. & Walker, B.H. (eds), Ecology of tropical savannas, pp. 217-245. Springer, Berlin. 
Cumming, D.H.M., Fenton, M.B., Rautenbach, I.L., Taylor, R.D., Cumming, G.S., Cumming, 
M.S., Dunlop, J.M., Ford, G.S., Hovorka, M.D., Johnston, D.S., Kalcounis, M.C., Mahlanga, Z. & 
Portfors, C.V. 1997. Elephants, woodlands and biodiversity in miombo woodland in southern 
Africa. South African Journal of Science 93:231–236. 
Davis, M.A., Wrage, K.J. & Reich, P.B. 1998. Competition between tree seedlings and 
herbaceous vegetation: Support for a theory of resource supply and demand. Journal of 
Ecology 86:652-661. 
Davis, M.A., Wrage, K.J., Reich, P.B., Tjoelker, M.G., Schaeffer, T. & Muermann, C. 1999. 
Survival, growth, and photosynthesis of tree seedlings competing with herbaceous vegetation 
along a water-light-nitrogen gradient. Plant Ecology 145:341–350. 
De Deyn, G.B., Cornelissen, J.H.C. & Bardgett, R.D. 2008. Plant functional traits and soil 
carbon sequestration in contrasting biomes. Ecology Letters 11(5):516-531. 
101 
Du Toit, J.T. & Cumming, D.H.M. 1999. Functional significance of ungulate diversity in African 
savannas and the ecological implications of the spread of pastoralism. Biodiversity and 
Conservation 8(12):1643-1661. 
Du Toit, J.T., Moe, S.R. & Skarpe, C. 2014. Elephant-mediated ecosystem processes in 
Kalahari-Sand woodlands. In, Skarpe, C., Du Toit, J.T. & Moe, S.R. (eds). Elephants and 
Savanna Woodland Ecosystems: A Study from Chobe National Park, Botswana. Wiley 
Blackwell Publishers, New Jersey, United States. 
Eckhardt, H.C., Van Wilgen, B.W. & Biggs, H.C. 2000. Trends in woody vegetation cover in the 
Kruger National Park, South Africa, between 1940 and 1998. African Journal of Ecology 38:108-
115. 
Edwards, D. 1984. Fire regimes in the biomes of South Africa. In: Booysen, P. de V. & Taiton, 
N.M. (eds), Ecological effects of fire in South African ecosystems, pp. 19-37. Springer, Berlin. 
Eggemeyer, K.D. & Schwinning, S. 2009. Biogeography of woody encroachment: why is 
mesquite excluded from shallow soils? Ecohydrology 2:81-87. 
Eldridge, D., Bowker, M., Maestre, T., Reynolds, J. & Whitford, W. 2011. Impacts of shrub 
encroachment on ecosystem structure and functioning: towards a global synthesis. Ecological 
Letters 14:709-722. 
Ellis, S.M. & Steyn, H.S. 2003. Practical significance (effect size) versus or in combination with 
statistical significance (p-values). Management Dynamics 12(4):51-53. 
February, E.C. & Higgins, S.I. 2010. The distribution of tree and grass roots in savannas in 
relation to soil nitrogen and water. South African Journal of Botany 76:517-523.  
February, E.C., Cook, G.D. & Richards, A.E. 2013. Root dynamics influence tree-grass 
coexistence in an Australian savanna. Austral Ecology 38:66-75. 
Fornara, D.A. & Du Toit, J.D., 2007, ‘Browsing lawns? Responses of Acacia nigrescens to 
ungulate browsing in an African savanna Ecology 88, 200–209.  
Fornara, D.A. & Du Toit, J.T. 2008. Community-level interactions between ungulate browsers 
and woody plants in an African savanna dominated by palatable-spinescent Acacia trees. 
Journal of Arid Environments 72:534-545. 
Fraser, S.W., Van Rooyen, T.H. & Verster, E. 1987. Soilplant relationships in the central Kruger 
National Park. Koedoe 30:19–34. 
102 
Fritz, H., Duncan, P., Gordon, I. & Illius, A. 2002 Megaherbivores influence trophic guilds 
structure in African ungulate communities. Oecologia 131:620–625. 
Frost, P., Medina, E., Menaut, J.C., Solbrig, O., Swift, M. & Walker, B. 1986. Responses of 
savannas to stress and disturbance. Biology International 10:1-82. 
Furstenburg, D. & Van Hoven, W. 1994. Condensed tannin as anti-defoliate agent against 
browsing by giraffe (Giraffa camelopardalis) in the Kruger National Park. Comparative 
Biochemistry and Physiology Part A: Physiolosy 107(2):425-431. 
Gandiwa, E. 2011. Effects of repeated burning on woody vegetation structure and composition 
in a semi-arid southern African savanna. International Journal of Environmental Sciences 
2(2):470-483. 
Gandiwa, E., Gandiwa, P. & Mxoza, T. 2012. Structure and composition of Spirostachys 
africana woodland stands in Gonarezhou National Park, southern Zimbabwe. International 
Journal of Environmental Sciences 2(4):2076-2089. 
Gaugris, J.Y., Vaicek, C.A. & Van Rooyen, M.W. 2014. Woody vegetation utilisation in Tembe 
Elephant Park, KwaZulu-Natal, South Africa. International Journal of Biology 6(3):24-35. 
Germishuizen, G. & Meyer, N.L. (eds) 2003. Plants of Southern Africa; an annotated checklist. 
Strilitzia 14. National Botanical Institute, Pretoria. 
Gertenbach, W.P.D. 1983. Landscapes of the Kruger National Park. Koedoe 26:9-121. 
Goheen, J.R., Keesing, F., Allan, B.F., Ogada, D. & Ostfield, R.S. 2004. Net effects of large 
mammals on Acacia seedling survival in an African savanna. Ecology 85(6):1555-1561. 
Goheen, J.R., Palmer, T.M., Keesing, F., Riginos, C. & Young, T.P. 2010. Large herbivores 
facilitate savanna tree establishment via diverse and indirect pathways. Journal of Animal 
Ecology 79:372-382. 
Google Maps. 2016. Map of Nkuhlu exclosures. [online]. Google. Available from:  
https://www.google.co.za/maps/@-24.9884939,31.7737826,3276m/data=!3m1!1e3?hl=en 
Gordon, D.R., Welker, J.M., Menke, J.W. & Rice, K.J. 1989. Competition for soil water between 
annual plants and blue oak (Quercus douglasii) seedlings. Oecologia 79:533–541. 
Govender, N. 2003. Fire management in Kruger National Park. Arid Lands Newsletter 54. 
103 
Grant, C.C. & Scholes, M.C. 2006. The importance of nutrient hot-spots in the conservation and 
management of large wild mammalian herbivores in semi-arid savannas. Biological 
Conservation 130(3):426-437. 
Gray, E.F. & Bond, W.J. 2013. Will woody plant encroachment impact the visitor experience and 
economy of conservation areas? Koedoe 55. 
Guldemond, R. & Van Aarde, R. 2008. A meta-analysis of the impact of African elephants on 
savanna vegetation. The Journal of Wildlife Management 72:892–899. 
Hanan, N.P., Sea, W.B., Dangelmayr, G. & Govender, N. 2008. Do fires in savannas onsume 
woody biomass? A comment on approaches to modelling savanna dynamics. American 
Naturalist 171:851-856. 
Hancock, G.R. & Mueller, R.O. 2010. The reviewer's guide to quantitative methods in the social 
sciences, Routledge, New York.432p. 
Harmse, H.J.V.M. & Van Wyk, P. 1972. Verkenningsgrondkaart van die Suidelike Distrik and die 
Nasionale Krugerwildtuin. Unpublished map. University of Potchefstroom for CHE, 
Potchefstroom. 
Harper, J. 1979. Population Biology of Plants. School of Plant Biology University College of 
North Wales, Bangor: Academic Press. 
Harrington, G.N. 1991. Effects of soil moisture on shrub seedling survival in a semi-arid 
grassland. Ecology 72:1138-1149. 
Hartnett, D.C., Ott, J.P., Sebes, K. & Ditlhogo M.K. 2012. Coping with herbivory at the juvenile 
stage: responses to defoliation and stem browsing in the African savanna tree 
Colophospermum mopane. Journal of Tropical Ecology 28(2):161-169. 
Hempson, G.P., Archibald, S. & Bond, W.J. 2015. A continent-wide assessment of the form and 
intensity of large mammal herbivory in Africa. Science 350(6264):1056–1061. 
Higgins S. I., Bond W. J. & Trollope W. S. 2000. Fire resprouting and variability: a recipe for 
grass-tree coexistence in savanna. Journal of Ecology 88:213–29. 
Higgins, S.I., Bond, W.J., February, E.C., Bronn, A., Euston-Brown, D.I.W., Enslin, B., 
Govender, N., Redeman, L., O’Regan, S., Potgieter, A.L.F., Scheiter, S., Sowry, R., Trollope, L. 
& Trollope, W.S.W. 2007. Effects of four decades of fire manipulation on woody vegetation 
structure in savanna. Ecology 88(5):1119-1125. 
104 
Hilbers, J.P., Van Langevelde, F., Prins, H.H.T., Grant, C.C., Peel, M.J.S., Coughenour, M.B., 
De Knegt, H.J., Slotow, R., Smit, I.P.J., Kiker, G.A. & De Boer, W.F. 2015. Modeling elephant-
mediated cascading effects of water point closure. Ecological Application 25(2):402-415. 
Hoffmann, W.A. 1999. Fire and population dynamics of woody plants in a Neotropical savanna: 
matrix model projections. Ecology 80:1354-1369. 
Hoffman, T., Todd, S., Ntshona, Z. & Turner, S. 1999. Land degradation in South Africa. Cape 
Town: National Botanical Institute. 
Hoffmann, W. A., Adasme, R., Haridasan, M., De Carvalho, M.T., Geiger, E.L., Pereira, M.A.B., 
Gotsch, S.G. & Franco, A.C. 2009. Tree topkill, not mortality, governs the dynamics of savanna 
forest boundaries under frequent fire in central Brazil. Ecology 90:1326–1337. 
Hoffmann, W.A., Geiger, E.L., Gotsch, S.G., Rossatto, D.R., Silva, L.C.R., Lau, O.L., Haridasan, 
M. & Franco, A.C. 2012. Ecological thresholds at the savanna-forest boundary: how plant traits, 
resources and fire govern the distribution of tropical biomes. Ecology Letters 15(7):759-768. 
Holdo, R.M., Holt, R.D.& Fryxell, J.M. 2009. Grazers, browsers, and fire influence the extent and 
spatial pattern of tree cover in the Serengeti. Ecological Applications 19(1):95-109. 
Holdo, R.M., Sinclair, A.R.E., Dobson, A.P., Metzger, K.L., Bolker, B.M., Ritchie, M.E. & Holt, 
R.D. 2009. A disease-mediated trophic cascade in the Serengeti and its implications for 
ecosystem C. PLoS Biology 7:e1000210. 
Holdo, R.M., Anderson, T.M. & Morrison, T. 2014. Precipitation, fire and demographic 
bottleneck dynamics in Serengeti tree populations. Landscape Ecology 29:1613-1623. 
Holmgren, M., Scheiter, M. & Huston, M.A. 1997. The interplay of facilitation and competition in 
plant communitites. Ecology 78:1966-1975. 
House, J.I., Archer, S., Breshears, D.D. & Scholes, R.J. 2003. Conundrums in mixed woody-
herbaceous plant systems. Journal of Biogeography 30:1763-1777. 
Hudak, A.T. & Wessman, C.A. 1998. Textural Analysis of Historical Aerial Photography to 
Characterise Woody Plant Encroachment in South African Savanna. Remote Sensing the 
Environment 66:317-330. 
Hutcheson, G.D. 2011. Ordinary Least-Squares Regression. In L. Moutinho and G.D. 
Hutcheson, The SAGE Dictionary of Quantitative Management Research. Pages 224-228. 
105 
Jacobs, O.S. & Biggs, R. 2001. The effect of different fire treatments on the population structure 
and density of the Marula, Sclerocarya birrea (A.Rich.) subsp. caffra (Sond.) Kokwaro (Kokwaro 
& Gillett, 1980) in the Kruger National Park. African Journal of Range and Forage Science 
18:13-23. 
Jordaan, J.J. 1995. The short-term effect of fire on the woody component of Sourish Mixed 
Bushveld. African Journal of Range and Forage Science 12:128-130. 
Joubert, D.F., Smit, G.N. & Hoffman, M.T. 2012. The role of fire in preventing transitions from a 
grass dominated state to a bushthickened state in arid savannas. Journal of Arid Environments 
87:1-7. 
Kambatuku, J.R., Cramer, M.D. & Ward, D. 2011. Savanna tree-grass competition is modified 
by substrate type and herbivory. Journal of Vegetation Science 22:225-237. 
Kgope, B.S., Bond, W.J. & Midgley, G.F. 2010. Growth responses of African savanna trees 
implicate atmospheric [CO2] as a driver of past and current changes in savanna tree cover. 
Austral Ecology 35:451–463. 
Khomo, L.M. & Rogers, K.H. 2005. Proposed mechanism for the origin of sodic patches in 
Kruger National Park, South Africa. African Journal of Ecology 43:29-34. 
Khomo, L., Hartshorn, A.S., Rogers, K.H. & Chadwick, O.A. 2011. Impact of rainfall and 
topography on the distribution of clays and major cations in granitic catenas of southern Africa. 
Catena 87:119–128. 
Kotze, D.C. & Zacharias, P.J.K. 1993. Utilization of woody browse and habitat by the black rhino 
(Diceros bicornis) in western Itala Game Reserve. African Journal of Range and Forage 
Science 10:36-40. 
Lagendijk, D.D.G., Mackey, R.L., Page, B.R. & Slotow, R. 2011. The Effects of Herbivory by a 
Mega- and Mesoherbivore on Tree Recruitment in Sand Forest, South Africa. PLoS ONE 6(3): 
e17983. doi:10.1371/journal.pone.0017983. 
Lagendijk, D.D.G., Thaker, M., De Boer, W.F., Page, B.R., Prins, H.H.T. & Slotow, R. 2015. 
Change in Mesoherbivore Browsing is Mediated by Elephant and Hillslope Position. PLoS ONE 
10(6): e0128340.doi:10.1371/journal.pone.0128340 
Laiolo, P., Dondero, F., Ciliento, E. & Rolando, A. 2004. Consequences of pastoral 
abandonment for the structure and diversity of the alpine avifauna. Journal of Applied Ecology 
41:294–304. 
106 
Leakey, A.D., Ainsworth, E.A., Bernacchi, C.J., Rogers, A., Long, S.P. & Ort, D.R. 2009. 
Elevated CO2 effects on plant carbon, nitrogen, and water relations: six important lessons from 
FACE. Journal of Experimental Botany 60:2859–2876. 
Lehmann, C.E.R., Prior, L.D. & Bowman, D.M.J.S. 2009. Fire controls structure of four dominant 
tree species in a tropical savanna. Oecologia 161:505-515. 
Lehmann, C.E., Archibald, S.A., Hoffmann, W.A. & Bond, W.J. 2011. Deciphering the 
distribution of the savanna biome. New Phytologist 191(1):197–209. 
Levick, S. & Rogers, K. 2008. Patch and species specific responses of savanna woody 
vegetation to browser exclusion. Biological Conservation 141:489-498. 
Levick, S.R., Asner, G.P., Kennedy-Bowdoin, T. & Knapp, D.E. 2009. The relative influence of 
fire and herbivory on savanna three-dimensional vegetation structure. Biological Conservation 
142:1693-1700. 
Levick, S.R., Asner, G.P., Chadwick, O.A., Khomo, L.M., Rogers, K.H., Hartshorn, A.S., 
Kennedy-Bowdoin, T. & Knapp, D.E. 2010. Regional insight into savanna hydrogeomorphology 
from termite mounds. Nature Communications 1:1–7. 
Levick, S.R., Baldeck, C.A. & Asner, G.P. 2015. Demographic legacies of fire history in an 
African savanna. Functional Ecology 29:131-139. 
Lohmann, D., Tietjen, B., Blaum, N., Joubert, D.F. & Jeltsch, F. 2014. Prescribed fire as a tool 
for managing shrub encroachment in semi-arid savanna rangelands. Journal of Arid 
Environments 107:49-56. 
Loópez-Soria, L. & Castell, C. 1992. Comparative genet survival after fire in woody 
Mediterranean species. Oecologia 14: 267–278. 
Ludwig, F., De Kroon, H., Berendse, F. & Prins, H.H.T. 2004. The influence of savanna trees on 
nutrient, water and light availability and the understory vegetation. Plant Ecology 170:93-105. 
Lykke, A.M. 1998. Assessment of species composition change in savanna vegetation by means 
of woody plants size class distributions and local information. Biodiversity Conservation 7:1261-
1275. 
Makhabu, S.W., Skarpe, C., Hytteborn, H. & Mpofu, Z.D. 2006. The plant vigour hypothesis 
revisited – how is browsing by ungulates and elephant related to woody species growth? Plant 
Ecology 184:163-172. 
107 
Massei, G., Hartley, S.E. & Bacon, P.J. 2000. Chemical and morphological variation of 
Mediterranean woody evergreen species: Do plants respond to ungulate browsing? Journal of 
Vegetation Science 11(1):1–8. 
McConnaughay, K.D.M. & Bazzaz, F.A. 1991. Is physical space a soil resource? Ecology 
72:94–103. 
McConnaughay, K.D.M. & Bazzaz, F.A. 1992. The occupation and fragmentation of space: 
consequences of neighbouring shoots. Functional Ecology 6:711–718. 
Menaut, J.C., Gignoux, J., Prado, C. & Clobert, J. 1990. Tree community dynamics in a humid 
savanna of the Cote-d’Ivoire: Modeling the effects of fire and competition with grass and 
neighbours. Journal of Biogeography 17:471-481. 
Midgley, J.J., Lawes, M.J. & Chamaille-Jammes, S. 2010. Savanna woody plant dynamics: the 
role of fire and herbivory, separately and synergistically. Australian Journal of Botany 58:1-11. 
Milne, G. 1947. A soil reconnaissance journey through parts of Tanganyika Territory December 
1935 to February 1936. Journal of Ecology 35:192–265. 
Miranda, A.C., Miranda, H.S., Dias, I.F.O. & Dias, B.F.S. 1993. Soil and air temperatures during 
prescribed cerrado fires in Central Brazil. Journal of Tropical Ecology 9:313–320. 
Moe, S.R., Rutina, L.P., Hytteborn, H. & Du Toit, J.T. 2009. What controls woodland 
regeneration after elephants have killed the big trees? Journal of Applied Ecology 46:223-230. 
Morison, C.G.T., Hoyle, A.C. & Hope-Simpson, J.F. 1948. Tropical soil-vegetation catenas and 
mosaics: a study in the south-western part of the Anglo-Egyptian Sudan. Journal of Ecology 
36:1–84. 
Mucina, L. & Rutherford, M.C. (eds). 2006. The vegetation of South Africa, Lesotho and 
Swaziland. Strelitzia 19. South African National Biodiversity Institute, Pretoria. 
Mwavu, E.N. & Witkowski, T.F. 2009. Population structure and regeneration of multiple-use tree 
species in a semi-deciduous African tropical rainforest: Implications for primate conservation. 
Forest Ecology Management 258:840-849. 
O’Connor, T.G., Puttick, J.R. & Hoffman, M.T. 2014. Bush encroachment in southern Africa: 
changes and causes. African Journal of Range and Forage Science 31(2):67-88. 
108 
O’Kane, C.A.J., Duffy, K.J., Page, B.R. & Macclonald, D.W. 2012. Heavy impact on seedlings 
by the impala suggests a central role in woodland dynamics. Journal of Tropical Ecology 
28:291-297. 
O’Keefe, T. & Alard, G. 2002. Effects of herbivores and fire on riparian and upland savanna 
ecosystems. Field Operations Manual for Herbivore & Fire Exclosures on the Sabie and Letaba 
Rivers in the Kruger National Park. (Unpublished). 
Obiri, J., Lawes, M. & Mukolwe, M. 2002. The dynamics and sustainable use of high value tree 
species of the coastal pondoland forests of the Eastern Cape Province, South Africa. Forest 
Ecology Management 166:131-148. 
Oliver, C.D. & Larson, B.C. 1990. Forest Stand Dynamics. McGraw-Hill Publishing Company, 
New York. 
Owen, D.F. 1978. The effect of a consumer Phytomyza ilicis on seasonal leaf-fall in the holly 
Ilex aquifolium. Oikos 31:268–271. 
Owen-Smith, N. & Cooper, S.M. 1987. Palatability of woody plants to browsing ruminants in a 
South African savanna. Ecology 68(2):319-331. 
Owen-Smith, R.N. 1992. Megaherbivores: The Influence of Very Large Body Size on Ecology. 
Cambridge University Press, Cambridge. 
Owen-Smith, N. 2010. The ecological potential of the kudu for commercial production in 
savanna regions. Journal of the Grassland Society of Southern Africa 2(3):7-10. 
Parr, C.L., Lehmann, C.E., Bond, W.J., Hoffmann, W.A. & Andersen, A.N. 2014. Tropical grassy 
biomes: misunderstood, neglected, and under threat. Trends in Ecology and Evolution 29:205–
213. 
Pellew, R.A.P. 1983. The impacts of elephant, giraffe and fire upon the Acacia-Tortilis 
woodlands of the Serengeti. African Journal of Ecology 21(1):41–74. 
Penderis, C.A. & Kirkman, K.P. 2014. Savanna browse production. 2: Prediction. African 
Journal of Range and Forage Science 31(1):25-36. 
Picon-Cochard, C., Coll, L. & Balandier, P. 2006. The role of below-ground competition during 
early stages of secondary succession: the case of 3-year-old Scots pine (Pinus sylvestris L.) 
seedlings in an abandoned grassland. Oecologia 148:373–383. 
109 
Polley, H.W., Mayeux, H.S., Johnson, H.B. & Tischler, C.R. 1997. Viewpoint: atmospheric CO2, 
soil water, and shrub/grass ratios on rangelands. Journal of Range Management 1:278–284. 
Pooley, E. 1997. The complete guide to trees of Natal, Zululand and Transkei. Natal Flora 
Publications Trust, Durban. 
Prins, H.H.T. & van der Jeugd, H.P. 1993. Herbivore population crashes and woodland structure 
in East Africa. Journal of Ecology 81:305–314. 
Prior, L.D., Murphy, B.P. & Russell-Smith, J. 2009. Environmental and demographic correlates 
of tree recruitment and mortality in north Australian savannas. Forest Ecology and Management 
257:66–74. 
Prior, L.D., Williams, R.J. & Bowman, D.M.J.S. 2010. Experimental evidence that fire causes 
tree recruitment bottleneck in an Australian tropical savanna. Journal of Tropical Ecology 
26:595-603. 
Ratajczak, Z., Nippert, J.B. & Collins, S.L. 2012. Woody encroachment decreases diversity 
across North American grasslands and savannas. Ecology 93:697–703. 
Reardon, M. 2012. Shaping Kruger: The dynamics of managing wildlife in Africa’s premier game 
park. Struik Nature:Cape Town, pp. 5-208. 
Riginos, C. 2009. Grass competition suppresses tree growth across multiple demographic 
stages. Ecology 90(2):335-340. 
Rogers, A., Ainsworth, E.A. & Leakey, A.D. 2009. Will elevated carbon dioxide concentration 
amplify the benefits of nitrogen fixation in legumes? Plant Physiology 151:1009–1016. 
Rooke, T., Bergström, R., Skarpe, C. & Danell, K. 2004. Morphological responses of woody 
species to simulated twig-browsing in Botswana. Journal of Tropical Ecology 20:281-289. 
Roques, K.G., O’Connor, T.G. & Watkinson, A.R. 2001. Dynamics of shrub encroachment in an 
African savanna: relative influence of fire, herbivory, rainfall and density dependence. Journal of 
Applied Ecology 38:268-280. 
Sankaran, M., Ratnam, J. & Hanan, N.P. 2004. Tree–grass coexistence in savannas revisited – 
insights from an examination of assumptions and mechanisms invoked in existing models. 
Ecology Letters 7:480–490. 
Sankaran, M., Ratnam, J. & Hanan, N. 2008. Woody cover in African Savannas: The Role of 
Resources, fire and Herbivory. Global Ecology and Biogeography 17(2):236-245. 
110 
Sankaran, M., Augustine, D.J. & Ratnam, J. 2013. Native ungulates of diverse body sizes 
collectively regulate long-term woody plant demography and structure of a semi-arid savanna. 
Journal of Ecology 101:1389-1399. 
Sauer, J.J.C., Theron, G.K. & Skinner, J.D. 1977. Food preferences of giraffe Giraffa 
camelopardalis in the arid bushveld of the Western Transvaal. South African Journal of Wildlife 
Research 7:53–59. 
Scheiter, S. & Higgins, S.I. 2007. Partitioning of root and shoot competition and the stability of 
savannas. American Naturalist 170:587-601. 
Schmitt, M.H., Ward, D. & Shrader, A.M. 2016. Incorporating secondary metabolites, tannin-
binding proteins, and diet breadth into carrying capacity models for African elephants. 
Ecological Modelling 332:8-18. 
Scholes, R.J. 1990. The influence of soil fertility on the ecology of southern African dry 
savannas. Journal of Biogeography 17:415-419. 
Scholes, R.J. & Walker, B.H. 1993. An African savanna: synthesis of the Nylsvley study. 
Cambridge University Press, Cambridge, UK. 
Scholes, R. J. & Archer, S. R. 1997. Tree-grass interactions in savannas. Annual Review of 
Ecological Systems 28:517-544. 
Scholes, R.J., Bond, W.J. & Eckhardt, H.C. 2003. Vegetation dynamics in the Kruger 
ecosystem. In: Du Toit, J.T., Rogers, K.H. & Biggs, H.C. (eds), The Kruger experience, pp. 242-
262. Island Press, Washington DC. 
Schulze, R.E. & McGee, O.S. 1978. Climatic indices and classifications in relation to the 
biogeography of southern Africa. In: Werger, M.J.A. & Van Bruggen, A.C. (eds), Biogeographhy 
and ecology of southern Africa, pp. 19-52. Dr W. Junk, The Hague. 
Scogings, P.F. 2011. Stem growth of woody species at the Nkuhlu exclosures, Kruger National 
Park: 2006-2010. Koedoe 53(1): doi: 10.4102/koedoe.v53i1.1035 
Scogings, P.F., Hjåltén, J. & Skarpe, C. 2011. Secondary metabolites and nutrients of woody 
plants in relation to browsing intensity in African savannas. Oecologia 167(4):1063-1073. 
Scogings, P.F., Johansson, T., Hjåltén, J. & Kruger, J. 2012. Responses of woody vegetation to 
exclusion of large herbivores in semi-arid savannas. Austral Ecology 37:56-66. 
111 
Shackleton, C.M. 1993. Demography and dynamics of the dominant tree species in communal 
and protected areas of the eastern Transvaal Lowveld. South African Journal of Botany 59:569-
574. 
Shackleton, R.T., Le Maitre, D.C. & Richardson, D.M. 2015. Prosopis invasions in South Africa: 
Population structures and impacts on native tree population stability. Journal of Arid 
Environments 114:70-78. 
Shannon, G., Druce, D.J., Page, B.R., Eckhardt, H.C. Grant, R. & Slotow, R. 2008. The 
utilization of large savanna trees by elephant in southern Kruger National Park. Journal of 
Tropical Ecology 24(3):281-289.  
Sharam, G., Sinclair, A.R.E. & Turkington, R. 2006. Establishment of broadleaved thickets in 
Serengeti, Tanzania: the influence of fire, browsers, grass competition, and elephants. 
Biotropica 38:599–605. 
Shipley, L.A. 2007. The influence of bite size on foraging at larger spatial and temporal scales 
by mammalian herbivores. Oikos 116(12):1964–1974. 
Siebert, F. & Eckhardt, H.C. 2008. The vegetation and floristics of the Nkuhlu exclosures, 
Kruger National Park. Koedoe 50(1):126-144. 
Siebert, F., Eckhardt, H.C. & Siebert, S.J. 2010. The vegetation and floristics of the Letaba 
exclosures, Kruger National Park. Koedoe 52(1):1-12. 
Silwerton, J. 1982. Introduction to Plant Population Ecology. Longman, New York. 
Skarpe, C. 1990. Structure of the woody vegetation in disturbed and undisturbed arid savanna, 
Botswana. Vegetation 87:11-18. 
Skarpe, C. 1991. Impact of Grazing in Savanna Ecosystems. Forestry and the Environment 
20(8):351-356.  
Skarpe, C. 1992. Dynamics of savanna ecosystems. Journal of Vegetation Science 3:293-300. 
Smit, I.P.J., Grant, C.C. & Devereux, B.J. 2007. Do artificial waterholes influence the way 
herbivores use the landscape? Herbivore distribution patterns around rivers and artificial surface 
water sources in a large African savanna park. Biological Conservation 136:85–99. 
Sop, T.K., Oldeland, J., Schmiedel, U., Ouedraogo, I. & Thiombiano, A. 2010. Population 
structure of three woody species in four ethnic domains of the Sub-Sahel of Burkina Faso. Land 
Degradation & Development DOI: 10.1002/ldr.1026. 
112 
Staver, A.C., Bond, W.J., Stock, W.D., Van Rensburg, S.J. & Waldram, M.S. 2009. Browsing 
and fire interact to suppress tree density in an African savanna. Ecological Applications 
19(7):1909-1919. 
Staver, A.C. & Levin, S.A. 2012. Integrating theoretical climate and fire effects on savanna and 
forest systems. American Naturalist 180:211-224. 
Staver, A.C. & Bond, W.J. 2014. Is there a ‘browse trap’? Dynamics of herbivore impacts on 
trees and grasses in an African savanna. Journal of Ecology 102:595-602. 
Stevens, N., Lehmann, C.E.R., Murphy, B.P. & Durigan, G. 2016. Savanna woody 
encroachment is widespread across three continents. Global Change Biology doi: 
10.1111/gcb.13409. 
Taylor, C.A., Twidwell, D., Garza, N.E., Rosser, C., Hoffman, J.K. & Brooks, T.D. 2012. Long-
term effects of fire, livestock herbivory removal, and weather variability in Texas semi-arid 
savanna. Rangeland Ecology and Management 65:21-30. 
Tedder, M., Morris, C., Fynn, R. & Kirkman, K. 2012. Do soil nutrients mediate competition 
between grasses and Acacia saplings? Grassland Science 58:238–245. 
Tedder,M., Kirkman, K., Morris, C. & Fynn, R. 2014. Tree–grass competition along a catenal 
gradient in a mesic grassland, South Africa. Grassland Science 60:1–8. 
Tomlinson, K.W., Van Langevelde, F., Ward, D., Prins, H.T., De Bie, S., Vosman, B., Sampaio, 
E.V.S.B. & Sterck, F.J. 2016. Defence against vertebrate herbivores trades off into architectural 
and low nutrient strategies amongst savanna Fabaceae species. Oikos 125:126-136. 
Trollope, W.S.W., Trollope, L.A., Potgieter, A.L.F. & Zambatis, N. 1996. SAFARI-92 
characterization of biomass and fire behaviour in the small experimental burns in the Kruger 
National Park. Journal of Geophysical Resistance 101(D19):23531-23539. 
Twidwell, D., Fuhlendorf, S.D., Taylor, C.A. & Rogers, W.E. 2013. Refining thresholds in 
coupled fire – vegetation models to improve management of encroaching woody plants in 
grasslands. Journal of Applied Ecology 50:603-613. 
Vadigi, S. & Ward, D. 2013. Shade, nutrients, and grass competition are important for tree 
sapling establishment in a humid savanna. Ecosphere 4(11):1-27. 
Vadigi, S. & Ward, D. 2014. Herbivory effects on saplings are influenced by nutrients and grass 
competition in a humid South African savanna. Perspectives in Plant Ecology, Evolution and 
Systematics 16:11-20. 
113 
Valeix, M., Fritz, H., Chamaillé‐Jammes, S., Bourgarel, M. & Murindagomo, F. 2008. 
Fluctuations in abundance of large herbivore populations: insights into the influence of dry 
season rainfall and elephant numbers from long‐term data. Animal Conservation 11:391–400. 
Van As, J., Du Preez, J., Brown, L. & Smit, N. 2012. The story of life and the environment: an 
African perspective. Struik Nature:Cape Town. Pp. 10-455. 
Van Auken, O.W. 2009. Causes and consequences of woody plant encroachment into western 
North America grasslands. Journal of Environmental Management 90:2931-2942. 
Van Coller, H., Siebert, F. & Siebert, S.J. 2013. Herbaceous species diversity patterns across 
various treatments of herbivory and fire along the sodic zone of the Nkuhlu exclosures, Kruger 
National Park. Koedoe 55(1):1-6. 
Van Coller, H. & Siebert, F. 2015. Heraceous biomass-species diversity relationships in nutrient 
hotspots of a semi-arid African riparian ecosystem. African Journal of Range and Forage 
Science 32(3):213-223. 
Van de Koppel, J. & Prins, H.H.T. 1998. The importance of herbivore interactions for the 
dynamics of African savanna woodlands: an hypothesis. Journal of Tropical Ecology 14:565-
576. 
Van de Vijver, C.A.D.M., Foley, C.A. & Olff, H. 1999. Changes in the woody component of an 
East African savanna during 25 years. Journal of Tropical Ecology 15:545-564. 
Van Rooyen, N., Theron, G.K., Bezuidenhout, D., Bredenkamp, G.J. & Van Rooyen, M.W. 
1993. Patterns of change in the herbaceous layer of a mesic savanna, South Africa. South 
African Journal of Botany 59:74-80. 
Van Wilgen, B.W., Biggs, H.C., Mare, N. & O’Regan, S.P. 2000. A fire history of the savanna 
ecosystems in the Kruger National Park, South Africa, between 1941 and 1996. South African 
Journal of Science 96(4):167-178. 
Venter, F.J. 1990. A classification of land for management planning in the Kruger National Park. 
Unpublished Ph.D. thesis, University of South Africa, Pretoria, South Africa. 
Venter, F.J., Scholes, R.J. & Eckhardt, H.C. 2003. The abiotic template and its associated 
vegetation pattern. In: Du Toit, J.T., Rogers, K.H. & Biggs, H.C. (eds), The Kruger experience, 
pp. 242-262. Island Press, Washington DC. 
114 
Venter, S.M. & Witkowski, E.T.F. 2010. Boabab (Adansonia digitata L.) density, size-class 
distribution and population trends between four land-use types in Northern Venda, South Africa. 
Forest Ecology Management 259:294-300. 
Wakeling, J.L., Staver, A.C. & Bond, W.J. 2011. Simply the best: the transition of savanna 
saplings to trees. Oikos 120:1448–1451. 
Wakeling, J.L., Bond, W.J., Ghaui, M. & February, E.C. 2015. Grass competition and the 
savanna-grassland tree-line: A question of root gaps? South African Journal of Botany 101:91-
97. 
Walker, B.H. & Noy-Meir, I. 1982. Aspects of the stability and resilience of savanna ecosystems. 
In: Huntley, B.J., Walker, B.H. (Eds.), Ecology of Tropical Savannas. Springer-Verlag, Berlin, 
pp. 556–590. 
Walker, B.H. 1985. Structure and function of savannas: an overview. In: Tothill, J.C. & Mott, J.J. 
(eds), Ecology and management of the world’s savannas, pp. 83-91. Australian Academy of 
Science, Canberra. 
Walter, H. 1971. Ecology of Tropical and Subtropical Vegetation. Oliver and Boyd, Edinburgh, 
UK. 
Wang, X., Hao, Z., Zhang, J., Lian, J., Li, B., Ye, J. & Yao, X. 2009. Tree size distributions in an 
old-growth temperate forest. Oikos 118(1):25–36. 
Watson, L.H. & Owen-Smith, N. 2002. Phenological influences on the utilization of woody plants 
by eland in semi-arid shrubland. African Journal of Ecology 40: 65–75. 
Weltzin, J.F. & McPherson, G.R., 1997. Spatial and temporal soil moisture resource partitioning 
by trees and grasses in a temperate savanna, Arizona, USA. Oecologia 112:156–164. 
Whyte, I.J., Van Aarde, R.J. & Pimm, S.L. 2003. Kruger’s elephant population: its size and 
consequences for ecosystem heterogeneity. In: du Toit, J.T., Rogers, K.H., Biggs, H.C. (Eds.), 
The Kruger Experience. Ecology and Management of Savanna Heterogeneity. Island Press, 
Washington, DC, USA, pp. 332–348. 
Wiegand, K., Ward, D., Thulke, H. & Jeltsch, F. 2000. From snapshot information to long-term 
population dynamics of Acacia by a simulation model. Plant Ecology 150:97-114. 
Wiegand, K., Saltz, D. & Ward, D. 2006. A patch-dynamic approach to savanna dynamics and 
woody plant encroachment – Insights from an arid savanna. Perspectives in Plant Ecology, 
Evolution and Systematics 7:229-242. 
115 
Wigley, B.J., Bond, W.J. & Hoffman, M.T. 2010. Thicket expansion in a South African savanna 
under divergent land use: local vs. global drivers? Global Change Biology 16:964-976. 
Wigley, B.J., Fritz, H., Coetsee, C. & Bond, W.J. 2014. Herbivores shape woody plant 
communities in the Kruger National Park: Lessons from three long-term exclosures. Koedoe 
56(1):http://dx.doi.org/10.4102/ koedoe.v56i1.1165. 
Wilson, S.L. & Kerley, G.I. 2003. Bite diameter selection by thicket browsers: The effect of body 
size and plant morphology on forage intake and quality. Forage Ecology and Management 
181(1):51–65. 
Williams C. A., Hanan N., Scholes R. J. & Kutsch W. 2009. Complexity in water and carbon 
dioxide fluxes following rain pulses in an African savanna. Oecologia 161:469–80. 
Wilson, B.G. & Witkowski, E.T.F. 2003. Seed banks, bark thickness and change in age and size 
structure (1978-1999) of the African savannah tree. Burkea Afr. Plant Ecology 167:151-162. 
Wiseman, R., Page, B. R., & O’Connor, T. G. 2004. Woody vegetation change in response to 
browsing in Ithala Game Reserve, South Africa. South African Journal of Wildlife Research 
34:25-37. 
Woolley, L.A., Page, B. & Slotow, R. 2011. Foraging strategy within African elephant family 
units: why body size matters. Biotropica 43:489–495. 
Young, T. 1994. Natural die-offs of large mammals: implications for conservation. Conservation 
Biology 8:410-418. 
Zida, D., Sawadogo, L., Tigabu, M., Tiiveau, D. & Odén, P.C. 2007. Dynamics of sapling 
population in savanna woodlands of Burkina Faso subjected to grazing, early fire and selective 
tree cutting for a decade. Forest Ecology and Management 243:102-115. 
 
116 
Appendix A: Nkuhlu exclosure woody species list 
 
Table A-1: Complete species list of all recorded woody species in 2002 and 2015 at the Nkuhlu 
exclosures, Kruger National Park. 
Family Species 
Achariaceae Kiggelaria africana L. 
 Flacourtia indica (Burm.f.) Merr. 
Anacardiaceae Lannea schweinfurthii (Engl.) Engl. 
 Sclerocarya birrea (A.Rich.) Hochst. 
 Searsia gueinzii (Sond.) F.A.Barkley 
 Searsia tenuinervis (Engl.) Moffett 
Aristolochiaceae Aristolochia elegans Mast. 
Apocynaceae Adenium multiflorum Klotzsch 
Asparagaceae Asparagus africanus Lam. 
 Asparagus laricinus Burch. 
Bignoniaceae Rhigozum zambesiacum Baker 
Boraginaceae Cordia ovalis R.Br. ex A.DC. 
 Ehretia amoena Klotzsch 
 Ehretia obtusifolia Hochst. ex A.DC. 
 Ehretia rigida (Thunb.) Druce 
Burseraceae Commiphora africana (A.Rich.) Engl. 
 Commiphora edulis (Klotzsch) Engl. 
 Commiphora neglecta I.Verd. 
Cannabaceae Trema orientalis (L.) Blume 
Capparaceae Cadaba natalensis Sond. 
 Cadaba termitaria N.E.Br. 
 Capparis fascicularis DC. 
 Maerua juncea Pax 
 Maerua parvifolia Pax 
 Maerua racemulosa (A.DC.) Gilg & Gilg-Ben. 
Celastraceae Elaeodendron transvaalense (Burtt Davy) R.H.Archer 
 Gymnosporia buxifolia (L.) Szyszyl. 
 Gymnosporia senegalensis (Lam.) Loes. 
 Gymnosporia tenuispina (Sond.) Szyszyl. 
 Hippocratea longipetiolata Oliv. 
117 
Combretaceae Combretum apiculatum Sond. 
 Combretum collinum Fresen. 
 Combretum hereroense Wawra 
 Combretum imberbe Wawra 
 Combretum microphyllum Klotzsch 
 Combretum molle R.Br. ex G.Don 
 Combretum mosambicense (Klotzsch) Engl. 
 Combretum zeyheri Sond. 
 Terminalia prunoides M.A.Lawson 
Ebenaceae Diospyros mespiliformis Hochst. ex A.DC. 
 Euclea divinorum Hiern 
 Euclea natalensis A.DC.  
Euphorbiaceae Bridelia cathartica G.Bertol. 
 Flueggea virosa (Roxb. ex Willd.) Voigt 
 Spirostachys africana Sond. 
Fabaceae Albizia anthelmintica (A.Rich.) Brongn. 
 Albizia forbesii Benth. 
 Albizia harveyi E.Fourn. 
 Cassia abbreviate Oliv. 
 Dalbergia melanoxylon Guill. & Perr. 
 Dichrostachys cinerea (L.) Wight & Arn. 
 Ormocarpum trichocarpum (Taub.) Engl. 
 Peltiphorum africanum Sond. 
 Philenoptera violaceae (Klotzsch) Schrire 
 Schotia brachypetala Sond. 
 Schotia capitata Bolle 
 Senegalia burkei (Benth.) Kyal. & Boatwr. 
 Senegalia nigrescens (Oliv.) P.J.H.Hurter 
 Senegalia schweinfurthii (Brenan & Exell) Seigler & Ebinger 
 Senegalia senegal (L.) Britton 
 Vachellia exuvialis (I.Verd.) Kyal. & Boatwr. 
 Vachellia grandicornuta (Gerstner) Seigler & Ebinger 
 Vachellia robusta (Burch.) Kyal. & Boatwr. 
 Vachellia xanthophloea (Benth.) P.J.H.Hurter 
Flacourtiaceae Scolopia zeyheri (Nees) Harv. 
Loganiaceae Strychnos madagascariensis Poir. 
 Strychnos spinose Lam. 
118 
Meliaceae Trichilia emetica Vahl  
Menispermaceae Cocculus hirsutus (L.) Diels 
Moraceae Ficus sycomorus L. 
Ochnaceae Ochna inermis (Forssk.) Schweinf. 
Olacaceae Ximenia caffra Sond. 
Oleaceae Jasminum fluminense Vell. 
Phyllanthaceae Phyllanthus reticulatus Poir. 
Rhamnaceae Berchemia discolor (Klotzsch) Hemsl. 
 Ziziphus mucronata Willd. 
Rubiaceae Canthium ciliatum (Klotzsch) Kuntze 
 Gardenia volkensii K.Schum. 
 Pavetta catophylla K.Schum. 
 Pavetta lanceolate Eckl. 
 Plectroniella armata (K.Schum.) Robyns 
 Pyrostria hystrix (Bremek.) Bridson 
 Tricalysia junodi (Schinz) Brenan 
 Vangueria infausta Burch. 
Rutaceae Teclea pilosa (Engl.) I.Verd. 
 Zanthoxylum capense (Thunb.) Harv. 
Salvodoraceae Azima tetracantha Lam. 
Sapindaceae Pappea capensis Eckl. & Zeyh. 
Sapotaceae Manikara mochisia (Baker) Dubard 
Tiliaceae Grewia bicolor Juss. 
 Grewia flava DC. 
 Grewia flavescens Juss. 
 Grewia hexamita Burret 
 Grewia villosa Willd. 
Verbenaceae Clerodendrum glabrum  
 Lippia javanica (Burm.f.) Spreng. 
Vitaceae Cissus cornifolia (Baker) Planch. 
 Rhoicissus revoili Planch. 
 Rhoicissus tridentata (L.f.) Wild & R.B.Drumm. 
 
119 
Appendix B: Dominant family changes between 2002 and 2015 in 
response to fire 
 
 
Figure B-1: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the riparian zone of the control treatment. 
120 
 
Figure B-2: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the crest zone of the control treatment. 
121 
 
 
Figure B-3: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the riparian zone of the partially fenced treatment. 
122 
 
 
Figure B-4: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the sodic zone of the partially fenced treatment. 
123 
 
 
Figure B-5: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the crest zone of the partially fenced treatment. 
124 
 
 
Figure B-6: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the riparian zone of the fully fenced treatment. 
125 
 
 
Figure B-7: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the sodic zone of the fully fenced treatment. 
126 
 
 
Figure B-8: Top three dominant woody families (expressed as frequency in %) and species per 
fire treatment per year (i.e. 2002 and 2015) for both seedling and established communities in 
the crest zone of the fully fenced treatment.