Trends in NDVI across protected areas in South Africa 

 

by 

 

Blessing George-Kayode 

(2515481) 

 

 

Master of Science 

in 

Resource Conservation Biology 

School of Animal, Plant and Environmental Sciences 

 Faculty of Science, University of the Witwatersrand, Johannesburg, South Africa 

 

 

 

Supervisor: Dr. Jolene Fisher 

 

March 2022 

 



ii 
 

PLAGIARISM DECLARATION 

I understand the penalty that comes with plagiarism, and I declare that is research report titled 

“Trends in NDVI across protected areas in South Africa” was done by myself. I have correctly 

refenced everyone whose work I used in my work. 

 

 

Blessing George-Kayode 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



iii 
 

DEDICATION 

I dedicate this research report to God Almighty, myself, and all the animals and plants that lost 

their existence because of vegetation loss. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



iv 
 

 

 

ACKNOWLEDGEMENTS 

“If I have seen further, it is by standing on the shoulders of Giants” – Sir Isaac Newton 

My profound gratitude goes to Almighty God, the One who made me, the author and Finisher, 

the Alpha and Omega of my life, for His grace, strength, protection, provision and guidance 

throughout the period of this research report and my study as a whole. Without him, this dream 

would not have seen the limelight. 

My sincere appreciation also goes to my supervisor, Dr. Jolene Fisher for her total support 

towards the completion of this work. She has not only supervised this work, but she also taught 

me skills such as GIS and remote sensing that would definitely be useful for me in my career. 

She has indeed brought out the best in me. 

I cannot but appreciate my parents, Mr. George Kayode and Dr. (Mrs.) Aderinsola Eunice 

Kayode who has brought me into this corrupt world uncorrupted. They have really performed all 

their responsibilities as parents, without them after God, I would be nobody. I will forever 

appreciate my one and only sibling and brother, George-Kayode Praise whose encouragements 

has brought me this far. 

 I want to use this opportunity to thank Mr. Ayotunde Ajagunna for his generosity, support, 

encouragements and kindness to me during the cause of my study. His presence in my life made 

this journey less stressful for me. 

Lastly, I am grateful to everyone that has contributed to this study one way or the other; Dr. 

Ayodele Oke, Buster, Thando, Mrs. Temi, Dr Mike, Mr. Abu and Mr. Stanley. I do not take their 

inputs and support for granted. 

 

 

 

 

 



v 
 

ABSTRACT 

Vegetation plays a significant role in the environment as it provides provisioning, supporting, 

and regulating ecosystem goods and services. Vegetation in protected areas in South Africa is 

experiencing significant changes in structure and composition as a result of human activities and 

climate changes. It is imperative that vegetation cover is managed and continuously monitored 

for it to continue to play its role in ecosystems. 

This study aimed at investigating the rate and direction of change in NDVI over a 20-year period 

(2001-2020) inside, and outside, protected areas in South Africa using remote sensing. NDVI is 

used as a proxy for vegetation cover, and Google Earth Engine was used to assess the NDVI 

values for each protected area and its buffers. Grouping the protected areas according to biomes, 

all the biomes of South Africa except the Forest biome were assessed. No protected area was 

grouped under the Forest biome because the protected areas studied were grouped under the 

biome which constituted the largest area. Five protected areas were each studied from Grassland, 

Savanna, Albany Thicket, Nama-Karoo, Succulent Karoo, and Indian Coast biomes. Three and 

two protected areas from the Fynbos and Desert biomes, respectively, were studied. Linear 

regression analysis was done to find the trends in NDVI, and correlation analysis was also done 

to find the relationship of vegetation change between the protected areas and their buffers. The 

result of the analysis shows that protected areas in all the biomes except the Grassland and Indian 

Coast experienced negative trends in NDVI values. This translates to 26 out of the 35 protected 

areas having negative trends in NDVI. Only 10 of the protected areas have more vegetation 

cover than their buffers (5 km and 20 km). The protected areas with positive trends in vegetation 

cover change are gaining vegetation faster or losing vegetation slower than their surroundings. 

Except for the Oviston Nature Reserve and Orange River Mount Nature Reserve, all the 

protected areas have strong correlation values with their respective buffers ranging from 0.600 to 

1.000. More efforts should be focused on protected areas that have negative trends in NDVI as 

well as those with negative trends in vegetation cover change. 

 

 

Keywords: Protected area, Biome, Vegetation cover, Remote sensing, NDVI, South Africa, 

Trends, Google Earth Engine 



vi 
 

 

 

 

TABLE OF CONTENTS 

PLAGIARISM DECLARATION ................................................................................................... ii 

DEDICATION ............................................................................................................................... iii 

ACKNOWLEDGEMENTS ........................................................................................................... iv 

ABSTRACT .................................................................................................................................... v 

TABLE OF CONTENTS ............................................................................................................... vi 

LIST OF FIGURES ..................................................................................................................... viii 

LIST OF TABLES .......................................................................................................................... x 

1. Introduction ............................................................................................................................. 1 

1.1 Problem statement ..................................................................................................................... 2 

1.2 Aims and objectives .................................................................................................................. 3 

2. Literature review ......................................................................................................................... 4 

2.1 Protected areas .......................................................................................................................... 4 

2.2 Biodiversity conservation ......................................................................................................... 5 

2.3 Effect and projection of climate change on vegetation ............................................................. 7 

2.4 Remote sensing and NDVI for monitoring vegetation ............................................................. 8 

3. Methods ................................................................................................................................. 11 

3.1 Study area................................................................................................................................ 11 

3.2 Normalized Difference Vegetation Index (NDVI) ................................................................. 13 

3.3 MODIS .................................................................................................................................... 13 

3.4 Methodology ........................................................................................................................... 14 

3.4.1 Trends in NDVI across Protected areas in South Africa from 2001 to 2021 ................... 14 

3.4.2 Vegetation cover change inside protected areas and their surrounding buffer zones (5 km 

and 20 km) ................................................................................................................................. 15 

4. Results ....................................................................................................................................... 16 

4.1 Trends in NDVI across protected areas in South Africa from 2001 to 2020 .......................... 16 

4.1.1 Grassland Biome .............................................................................................................. 16 

4.1.3 Albany Thicket Biome ..................................................................................................... 19 

4.1.4 Nama-Karoo Biome ......................................................................................................... 21 



vii 
 

4.1.6 Succulent Karoo Biome ................................................................................................... 22 

4.1.7 Fynbos Biome .................................................................................................................. 25 

4.1.8 Desert Biome .................................................................................................................... 26 

4.2 Vegetation cover change inside and outside the protected areas across South Africa from 

2001 to 2020 ................................................................................................................................. 27 

4.2.1 Grassland Biome .............................................................................................................. 28 

4.2.2 Savanna Biome ................................................................................................................. 29 

4.2.3 Albany Thicket Biome ..................................................................................................... 30 

4.2.4 Nama-Karoo Biome ......................................................................................................... 31 

4.2.5 Succulent Karoo Biome ................................................................................................... 32 

4.2.6 Indian Coast Biome .......................................................................................................... 33 

4.2.7 Fynbos .............................................................................................................................. 34 

4.2.8 Desert Biome .................................................................................................................... 34 

4.3 Correlation between Protected Areas in South Africa and their Buffers (5km and 20 km) 

from 2001 to 2020 ......................................................................................................................... 35 

5. Discussion ................................................................................................................................. 37 

6. Conclusion ................................................................................................................................ 43 

References ..................................................................................................................................... 44 

 

 

 

 

 

 

 

 

 

 

 

 

 



viii 
 

 

 

LIST OF FIGURES 

Figure 3.1: Map showing the biomes and protected areas in South Africa that were investigated 

in this study. .................................................................................................................................. 12 

Figure 4.1: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Grassland 

biome; (a) Songimvelo Game Reserve (GR), (b) ......................................................................... 17 

Figure 4.2: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Savanna 

biome; (a) Kahalari Gemsbok National Park (NP), (b) Madikwe Nature Reserve (NR), (c) 

Kruger National Park (NP), (d) Marakele NP, and (e) Pilanesberg Nature Reserve (NR). .......... 18 

Figure 4.3: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Albany 

Thicket biome; (a) Addo Elephant National Park (NP), (b) Great Fish River Nature Reserve 

(NR), (c) Water Meeting Nature Reserve (NR), (d) East London Coast Nature Reserve (NR), and 

(e) Mount Coke State Forest. ........................................................................................................ 20 

Figure 4.4: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Nama-Karoo 

biome; (a) Camdeboo National Park (NP), (b) Mokala National Park (NP), (c) Augerbies Falls 

National Park (NP), (d) Geogap Nature Reserve (NR), and (e) Mountain Zebra National Park 

(NP). .............................................................................................................................................. 22 

Figure 4.5: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Succulent 

Karoo biome; (a) Namaqua National Park (NP), (b) Tankwa Karoo National Park (NP), (c) 

Moedverlooren Nature Reserve (NR), (d) Vaalhoek Nature Reserve (NR), and (e) Akkerendam 

Nature Reserve (NR)..................................................................................................................... 23 

Figure 4.6: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Indian Coast 

biome; (a) Isimangaliso Provincial Nature Reserve (NR), (b) Mkambati Nature Reserve (NR), (c) 



ix 
 

Ngoye Forest Reserve (FR), (d) Richards Bay Game Reserve (GR), and (e) Dwesa-Cwebe 

Wildlife Reserve (WR). ................................................................................................................ 25 

Figure 4.7: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Fynbos 

biome; (a) Rooiberg Nature Reserve (NR), (b) Oorlogskloof Nature Reserve (NR), and (c) West 

Coast National Park (NP).............................................................................................................. 26 

Figure 4.8: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the desert biome; 

(a) Richtersveld National Park (NP), and (b) Orange River Mouth Nature Reserve (NR). ......... 27 



x 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES 

Table 4.1: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Grassland biome including the slope differences between the PA and buffers. .... 28 

Table 4.2: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Savanna biome including the slope differences between the PA and buffers. ....... 29 

Table 4.4: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Nama-Karoo biome including the slope differences between the PA and buffers. 31 



xi 
 

Table 4.5: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Succulent Karoo biome including the slope differences between the PA and 

buffers. .......................................................................................................................................... 32 

Table 4.6: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Indian Coast biome including the slope differences between the PA and buffers. 33 

Table 4.7: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Fynbos biome including the slope differences between the PA and buffers. ......... 34 

Table 4.9: Pearson correlation values derived from the correlation of Mean NDVI values of 

various protected areas South Africa, and their buffers (5 km and 20 km). ................................. 35 

 

 

 

 

 

 

 



1 
 

1. Introduction 

South Africa is known to be one of the world’s most biologically diverse countries, having three 

out of the thirty-six global biodiversity hotspots (Bux et al., 2021). Although the country is 

known for its array of ecosystems and landscapes and high levels of endemic flora and fauna, it 

faces threats from human activities such as land use and conversion (Barger et al., 2018; Skowno 

et al., 2019).  Protected areas are established for the purpose of conserving biodiversity 

especially threatened and endangered species (Small & Sousa, 2016). According to Dudley 

(2008), a protected area is defined as “a clearly defined geographical space, recognized, 

dedicated and managed, through legal or other effective means to achieve the long-term 

conservation of nature with associated ecosystem services and cultural values”. However, 

protected areas face a lot of pressure from human activities, which leads to ecosystem change, 

land use, land cover change, and vegetation change, which is a major cause of biodiversity loss 

both inside and around protected areas (Defries & Nagendra, 2017). As a result, vegetation 

change in and around protected areas is being monitored using remote sensing (Murray et al., 

2018). Remote sensing has become a primary and essential tool for the monitoring, mapping, and 

evaluation of vegetation cover changes especially in protected areas by providing researchers up 

to date satellite information on vegetation cover (Pereira et al., 2013). Many researchers use the 

Normalized Difference Vegetation Index (NDVI) introduced by Tucker (1979) which is one of 

the widely used remotely sensed spectral indices to assess vegetation change at regional and 

global level (Xu et al., 2016). NDVI is sensitive to the photosynthetic activity (Ning et al., 2015) 

of the vegetation in the observed area, the value ranges from -1 to 1 where a value between 0 and 

1 represents vegetation (the higher the value, the higher the greenness), a value between 0 and -1 

represents no vegetation, bare ground, or water.  

 

  



2 
 

1.1 Problem statement 

In protected areas, vegetation plays important roles as it provides essential provisioning, 

supporting, and regulating ecosystem goods and services (Madureira et al., 2015; Stolton et al., 

2015). Though vegetation adapts to new conditions, it is sensitive to changes in atmospheric 

carbon dioxide which makes has made it a major factor in predicting future climate and a 

warning of climate change (Pan et al., 2018). Vegetation in protected areas is experiencing 

significant changes in structure and composition as a result of human activities, and changes in 

climate (De Baan et al., 2013). There has been various research done predicting that the impact 

of climate change on vegetation will exacerbate as the concentration of CO2 in the atmosphere 

increases (Beck et al., 2011). 

To make predictions reliable enough to influence future environmental and conservation policies, 

it is imperative that evaluation of the past is done, and the rate and direction of change is well 

understood (Luo et al., 2011). This improves our chances of predicting climate extremes as well 

as coming up with strategies to mitigate the impact of climate change on the ecosystem (Peters et 

al., 2004). 

Many researchers have used NDVI to examine ecosystem dynamics in protected areas which is 

important for the monitoring, management, and planning of natural resources (Tucker et al., 

2005; Alcaraz-Segura et al., 2008; Nemani et al., 2009; Wang et al., 2012). Despite the use of 

NDVI which is an effective tool to examine vegetation change, most researchers have selected 

specific areas in South Africa for their studies without examining protected areas in all biomes 

across South Africa. For example, Wegmann et al. (2014) used NDVI as a proxy to examine 

vegetation change of several protected areas in Africa. Even though the authors investigated a 

large study area, they were limited to selecting only few protected areas in South Africa. 

In view of the above, this study examined extensively the protected areas in South Africa and 

expanded on the Wegmann et al. (2014) study of vegetation cover changes inside and outside 

protected areas from 2001 - 2020. It is important and necessary for this examination to be done 

because it provides information on the present situation and long-term trend of vegetation change 

in protected areas and biomes across South Africa. The information obtained from this study 

may influence management actions, planning, and monitoring of ecosystems towards 

conservation of biodiversity. 



3 
 

1.2 Aims and objectives 

This project aimed at investigating the rate and direction of change in NDVI over a 20-year 

period (2001-2020) inside and outside protected areas in South Africa 

Specific objectives  

1) To assess the trends in NDVI across protected areas in South Africa from 2001 to 2020.  

2) To evaluate and compare vegetation cover change trends inside protected areas and their 

surrounding buffer zones (5 km and 20 km). 

 

  



4 
 

2. Literature review 

2.1 Protected areas 

Protected areas (PA) over the past few decades have experienced a tremendous expansion that 

has been influenced by the government, NGO, and local communities leading to the rapid growth 

and recognition of protected areas all over the world (Watson et al., 2014). Due to the rapid 

growth, the expectations placed by various stakeholders on protected areas have dramatically 

increased because of the fragmentation of natural ecosystems (Sanderson, 2002). Protected areas 

are now created not just for the conservation of famous seascapes and landscapes, and to protect 

threatened and endangered species, but they are also established to improve the livelihood of the 

local communities surrounding the protected areas, increase the nation’s revenue through tourism 

and play an essential role in species adaptation to the impacts of climate change while mitigating 

them (Dudley & Stolton, 2010). A protected area is an area of land set aside as a strategy for 

biodiversity conservation, sustaining ecosystem services, and protect habitat for endemic and 

endangered species (Margules & Pressey, 2000; Obeng et al., 2021). Even though the area of 

land set aside as a protected area has increased globally to over 13% of the earth surface area 

(UNEP-WDPA, 2019), the threats facing biodiversity as a result of land degradation and 

encroachment have not been alleviated (Hansen & Defries, 2007; Gonzalez-Roglich et al., 2019). 

Protected areas do not occur in isolation so various factors such as land use, species, and 

activities in the landscape surrounding the park influence the functioning of a park (Hansen & 

Defries, 2007). 

Protected areas are not a new development, they have been in existence in the form of sacred 

sites established by local communities (Orsenigo et al., 2019). Nevertheless, protected areas have 

now been modified and developed in this modern age and have rapidly grown from a few 

protected areas at the beginning of the 20th century to over 162 000 national protected areas that 

are legally managed making it about 5.6% of the earth’s area being protected (Watson et al., 

2014). The terrestrial protected areas can be compared to the total land area of South and Central 

America combined and the area of the marine protected areas can also be compared to the 

combination of the Bering Sea, South China sea, Caribbean sea and the Mediterranean sea 

(COP14, 2019).  In addition to the legally protected areas, there are many other protected areas 



5 
 

that are not legally managed but are managed by, for instance, the local communities, 

individuals, religious groups, corporations (COP14, 2019). 

In the early development of modern protected areas, protected areas were established solely for 

the protection of outstanding natural feature, and wildlife (The World Commission on Human 

and Environment Development, 1987). Although many of these protected areas considered or 

incorporated public use while protecting landscape and species in their establishment, they did 

not incorporate tourism until the mid-20th century (UN, 1992). In some developing nations, 

tourism in protected areas boosts the economy of the nation. For instance, Rwanda generates 

about US$200M annually from tourists visits to the Volcanoes National Park, which has been the 

main source of foreign exchange for the country (Aviles-Polanco et al., 2019). Protected areas 

have also been established due to the concerns raised from environmental degradation. These 

concerns have increased the importance attributed to in-situ conservation which has led to the 

expansion of protected areas globally to aid the protection of species and ecosystem from the 

adverse impacts of rapid environmental changes (Watson et al., 2014). Many studies have 

justified the use of protected areas as the major conservation strategy and if well managed, PAs 

are capable of reducing habitat loss (Petraitis, 1989; Janzen, 2004; Kati, 2004; DeFries et al., 

2005; Gavin, 2015), and maintain species populations better compared to other management 

approaches such as the ex-situ conservation approach. 

2.2 Biodiversity conservation 

Over the years, studies have shown that there is a decline in biodiversity which is attributed to 

land use/conversion and habitat fragmentation (Krauss et al., 2010; Li et al., 2016). The loss of 

biodiversity globally is detrimental to the maintenance of environmental function and ecosystem 

services (Butchart et al., 2010). One of the strategies to mitigate the loss of biodiversity is to 

investigate and identify areas important to biodiversity conservation and exert conservation 

efforts to those areas, such as protected areas (Myers, 1988). Threats have an adverse effect on 

protected areas, which can reduce their function in conserving biodiversity, provisioning of 

ecosystem services, and reduce the size of the protected area (Balme et al., 2010). There is a 

correlation between the increase in human activities around the protected area and the extinction 

rate of species which can be exacerbated by illegal activities in the protected areas, increase in 

urban settlement, and climate change (Basommi, 2016). Protected areas do not only conserve 



6 
 

biodiversity, but the local communities also around protected areas benefit from the protected 

areas through programs such as Community Based Natural Resource Management (CBNRM) 

(Hattter & Southworth, 2009). As a result, it is important to monitor and ensure the effectiveness 

of protected areas in the delivery of these services. 

Developing nations, such as South Africa which is one of the world’s most biologically diverse 

countries, have experienced rapid land use and land cover change over the years (Musakwa & 

Wang, 2018), which has a significant effect on the environment (Li et al., 2016) and is caused by 

deforestation and rapid urbanization (Barlow et al., 2016; Richards & Fries, 2016). Depending 

on the degree of change, land cover change can cause a total change in the existing ecosystem or 

a partial change (Levine et al., 2016). Several studies have investigated land use and land cover 

change using vegetation condition as a proxy (Houessou et al., 2013; Zoungrana et al., 2018); 

because it is easy to observe the direction and extent of vegetation cover change regardless of the 

causes of the change. This has led it to be the leading indicator used by scientist to investigate 

land cover change (Reynolds et al., 2007). 

Vegetation plays an essential role in ecosystem functioning and is known to react to changes in 

the environment; thus, it is used to monitor global change. Additionally, vegetation is also 

responsible for land surface energy exchange, biogeochemical and hydrological cycles (Harrison 

& Goni, 2010). Vegetation cover degradation has become a concern because it threatens 

biodiversity, which may result in the degradation of the soil (Yengoh et al., 2015). Land use and 

land cover (LULC) is usually classified to identify vegetation that has undergone conversions 

(Xu et al., 2016). According to Xu et al. (2016), conversion is the total substitution of natural 

vegetation cover by a different land cover, which can be caused by anthropogenic activities such 

as deforestation and climate change. Oftentimes, modification of vegetation types and pattern are 

as a result of the effects of human activities in addition to changes in temperature and rainfall 

conditions on vegetation structure (Adepoju et al., 2019). The vegetation cover, phenology, and 

production changes when the vegetation structure changes as a result of changes in the 

environment. Vegetation cover, which is the amount of vegetation in a specific area has proven 

to be an indicator to monitor environmental changes and also used to differentiate different 

ecosystems (Zhang et al., 2013). 



7 
 

2.3 Effect and projection of climate change on vegetation 

The South African vegetation is at risk of changes to its structure and composition in response to 

climate change (Ziervogel et al., 2014). According to Von Maltitz et al. (2019), some of the 

South African biomes such as the Forest, Indian ocean coastal belt, Nama karoo, and Fynbos are 

becoming vulnerable due to increasing temperature. The biomes are experiencing significant 

changes like shifting of their boundaries which is caused by climate change (Beck et al., 2011). 

For instance, the Nama-karoo is shrinking in the west and moving northeast into the Grassland 

biome as a result of increasing CO2 and aridification (Midgley et al., 2008). The Grassland 

biome in turn is shrinking due to the increase of the Savanna biome as a result of the spread of 

the trees which are favored by the rise in CO2 and temperature (Midgley & Thuiller, 2011).  It 

was projected by SAEON (2015) that tree density will be on the increase while species diversity 

will be on the decrease due to an increase in temperature and decrease in rainfall as species lose 

their resilience to the effect of drought and heat. The study carried out by Buitenwerf et al. 

(2012), concluded that tree density was increasing majorly due to increasing CO2 in the 

atmosphere, which has changed the ecology of trees and grasses in South Africa. Over the years, 

the adverse effect of climate change on the vegetation in South Africa has been on the increase 

(King et al., 2015). The Savanna biome is being threatened by the rapid increase of woody 

plants, which has the ability to change its state into a thicket (Parr et al., 2012). Prevey et al. 

(2019) proved that there is a link between climate change and plant phenology which is further 

supported by García Criado et al. (2020), whose study testified that there is a relationship 

between climate change and wood encroachment. This is possible because the increase in the 

concentration of CO2 in the atmosphere increases CO2 uptake of plants (Kgope et al., 2010) and 

increases the availability of soil water (Leakey et al., 2009), thereby favoring woody plants. 

Wood encroachment can cause an alteration in the ecosystem, causing biodiversity loss and 

reducing ecosystem services to humans. In addition, it has the potential to negatively impact crop 

production as well as tourism (Honda & Durigan, 2016). Woody encroachment, a term used 

interchangeably with bush encroachment, which occurs when there is an increase in woody 

plants, especially in places where they are usually not found is driven by the increase in 

atmospheric CO2 (Zhang et al., 2019), change in rainfall pattern as well as human activities. Fire 

processes maintain the Savanna ecosystem but human activities altering the fire pattern or 

process by deliberately suppressing fire with construction of fences and roads (Archibald et al., 



8 
 

2013) will enhance the growth of woody plants (Archer et al., 2017) making the Savanna biome 

exceptionally vulnerable to woody encroachment (Stevens et al., 2017). 

With the alarming rate of global warming, biodiversity worth almost $200 billion may be lost by 

2050 (Seidensticker, 2010), including significant loss of several species in the fynbos biome in 

South Africa (Skowno et al., 2021). Engelbrecht et al. (2015) projected that rainfall in South 

Africa would reduce by 30% between 2071 and 2099; likewise, temperature would increase by 

3°C by the end of the century. Furthermore, the projection made by Jury (2013) supports 

Engelbrecht et al. (2015) that there would be an increase in aridity, particularly in the 

Southwestern region of South Africa at the end of the century, which has been mainly due to the 

rise in temperature (Serdeczny et al., 2016). It is also expected that the rate of drought would 

increase by 40% by 2100 (Prudhomme et al., 2014), and this would have an adverse impact on 

vegetation.  South Africa is not only known as a biodiversity hot spot, but it is also known as a 

climate change hot spot (Hoegh-Guldberg et al., 2019).  

2.4 Remote sensing and NDVI for monitoring vegetation 

Remote sensing refers to the process of acquiring information about an object through sensors 

mostly from satellites (Chuvieco et al., 2020). The information is acquired when the 

electromagnetic energy emitted by the observed object is measured (Geller et al., 2017). Remote 

sensing has evolved from simple classifications of land cover (Herbreteau at al., 2007) to 

complex spectral analyses including assessing vegetation condition and crop growth. Now, 

remote sensing is used to monitor vegetation cover, environment, crop growth, and so on. The 

imageries produced from remote sensing are a combination of various bands which produce the 

most appropriate image that is then used for analysis.  Images generated are a good 

representation of what is happening on the ground; an example is the vegetation. This 

information is fundamental to conservation management as remote sensing has the ability to 

view the surface of the earth from various perspectives at different spectral dimensions and 

resolutions (Soberon & Peterson, 2004).   

Over the years, remote sensing has been used to evaluate and analyze vegetation cover change 

over large areas, which is imperative to understand the trends of vegetation change of the area to 

inform environmental policies and decision making (Okin, 2007). Recently, there have been 

advancements in remote sensing technology that have increased the accuracy of satellite images 



9 
 

of vegetation cover. Wegmann et al. (2014) used remote sensing to analyze habitat change within 

and around various protected areas in Africa and their result showed that the rate of vegetation 

change gave a flat or positive trend for most of the protected areas. Evaluation and analyses of 

vegetation cover is only possible and reliable if the appropriate remotely sensed images are 

selected and are converted into the relevant vegetation indices (Barati et al., 2011). Multispectral 

imageries have been used to derive various vegetation indices such as the Normalized 

Differential Vegetation Index (NDVI), the Wide Dynamic Range Vegetation Index (WDRVI), 

and the Enhanced Vegetation Index (EVI) just to mention a few (Jordan, 1969; Huete et al., 

1997; Gitelson, 2004; Xue & Su, 2017). Out of the various vegetation indices available, NDVI is 

popularly known and used to monitor the greenness of an area. It is used to assess and analyze 

vegetation change because of its potential to quickly detect vegetation decline and stress which is 

an advantage in the agriculture sector and land use investigation. NDVI is also preferred most 

time to the other vegetation indices because of its strong correlation with vegetation cover (Xiao 

et al., 2015).  

In the early years of NDVI, the AVHRR was the dominant product used to derive NDVI values 

(James & Kalluri, 1994). Nowadays, Landsat and MODIS (Moderate Resolution Imaging 

Spectroradiometer) products have been used to replace the AVHRR because they have high 

spatial resolutions of 30 m for Landsat and 250 m to 1000 m for MODIS. Hoare and Frost (2004) 

used remote sensing and the NDVI values provided by AVHRR to describe the phenological 

characteristics and seasonal dynamics of all the biomes in southern Africa. Cho and Ramoelo 

(2019) obtained NDVI time series from MODIS product to examine tree-cover change in the 

Savanna and Grassland biomes of South Africa. Their result shows that there has been an 

increase in tree-cover between 2001 and 2018.  

Various studies have shown the benefits of NDVI, Pettorelli et al. (2005) showed that NDVI 

could be used to effectively differentiate between biomes, differentiate dense forests from 

agricultural fields as well as identify seasonal forests and evergreen forests. Furthermore, it is 

effective in the investigation of vegetation properties, biomass, the concentration of chlorophyll 

in leaves, vegetation cover, and plant productivity (Pettorelli et al., 2005; Zhu & Liu, 2015; 

Dutrieux et al., 2015; Tian et al., 2017).  All these qualities of NDVI have resulted in it being 

promoted above other vegetation indices by scientists since the 1970s and these qualities have 



10 
 

informed the decision to enable most remote sensing satellites to capture the necessary 

wavelengths to enable the calculation of the index, including satellites such as Sentinel, Landsat 

and MODIS, produce it at different spatial and temporal resolutions (Shao, 2015). NDVI has 

partly facilitated the use of remote sensing because it simplifies the complexity of vegetation 

properties; nevertheless, it has some significant drawbacks, such as pointed out by Frantz (2019).  

Sensor quality: in the quest to keep producing products that can produce the NDVI values, 

products with different spatial resolutions, band widths and processing have resulted in different 

NDVI values.  

Atmospheric effects: the red wavelength is shorter and stronger compared to the near infrared 

wavelength resulting in Rayleigh scattering which affects NDVI calculation. The problem 

created by Rayleigh scattering can be avoided by using surface reflectance that is 

atmospherically corrected instead of top of the atmosphere (TOA) radiance (Frantz, 2019). 

Saturation ease: NDVI becomes insensitive at a certain high level of biomass or 

environmental changes whereas it is sensitive to the soil, atmosphere and pixel. NDVI values are 

highest when vegetation is at its greenest; and values become saturated when a forest (high 

density of woody vegetation) has a high biomass value (Van Der Meor et al., 2001) causing a 

dense forest to have a higher NDVI value than a green golf course grass. 

 

 

 

 

 

 

 



11 
 

3. Methods 

3.1 Study area 

South Africa is known for biodiversity richness, and concerns have been raised regarding the 

loss of biodiversity. This study was carried out in protected areas across South Africa and the 

biomes of South Africa were used to group the protected areas. It was initially proposed that the 

study would be carried out in five protected areas across each of the nine biomes in South Africa 

but during the data collection, it was observed that some protected areas have overlapping 

biomes. Such protected areas were grouped under the biome which constituted the largest area. 

This arrangement left the forest biome with no protected area and the Fynbos and Desert biomes 

with only three and two protected areas respectively.  

In total, thirty-five protected areas were studied across eight biomes in South Africa which was 

based on the protected area category and the specific type of protected area. The biomes are 

Grassland, Savanna, Albany Thicket, Succulent Karoo, Nama-Karoo, Indian Coastal, Fynbos, 

and Desert and I selected protected areas from the category Formal A which includes specific 

protected area types such as national parks, provincial nature reserves, and forest act protected 

areas as categorized in the South African National Biodiversity Institute, National Biodiversity 

Assessment (NBA) 2011 protected Area layer dataset under the Protected Areas Act (Act 57 of 

2003) and were downloaded from SANBI BGIS. The use of the National Biodiversity 

Assessment (NBA) 2011 file was mainly for selecting the protected areas on ArcMap and also to 

create the map below (Figure 3.1). 

 

 

 

 



12 
 

 

Figure 3.1: Map showing the biomes and protected areas in South Africa that were investigated 

in this study. 

 

 



13 
 

3.2 Normalized Difference Vegetation Index (NDVI)  

Normalized Difference Vegetation Index (NDVI) is a vegetation index used to measure the 

greenness of an area which is achieved when the vegetation reflects the near infrared wavelength 

and absorbs the red wavelength of the electromagnetic spectrum (Maskova et al., 2008). NDVI is 

used in this study because of its simplicity in detecting the status of vegetation growth, 

vegetation cover and phenology (Huang et al., 2021) over a large spatial area. NDVI uses the 

reflected and absorbed wavelength to create a value which signifies the photosynthetic activity 

occurring at a pixel using the formula:  

NDVI =
NIR – Red

NIR + Red
 

The NDVI value is between -1 and 1 where a high NDVI value signifies a high photosynthetic 

activity. The degree of greenness of an area is defined by the NDVI value of the area.  An area 

with a dense vegetation would reflect more near-infrared wavelength than red wavelength 

leading to a high NDVI value, while an area with a sparse or unhealthy vegetation would reflect 

a lesser near-infrared wavelength than the red wavelength which results in a low NDVI.  

For this study, the already calculated NDVI time series from 2001 to 2020 generated by MODIS 

(MOD13A2, https://lpdaac.usgs.gov/products/mod13a2v006/) were used.  

3.3 MODIS 

In this study, the MOD13A2 Version 6 Terra (https://lpdaac.usgs.gov/products/mod13a2v006/) 

Vegetation Indices was used. This dataset provides the NDVI of a pixel at a spatial resolution of 

1000 m (Didan, 2015) which is sufficient for the monitoring of the vegetation in the selected 

protected areas which are at regional level. The MOD13A2 has two primary vegetation layers, 

which are the Normalized Difference Vegetation index (NDVI), and the Enhanced Vegetation 

Index (EVI) but this study will be focusing on NDVI. It is a series of 16-day composites, where 

the best/clearest pixel value is chosen from a composite of 16 days’ worth of images. This 

dataset has the following bands: NDVI, EVI, two detailed QA (quality assurance), and reflective 

bands 1 (red), 2 (near-infrared), 3 (blue) and 7 (mid-infrared). The MOD13A2 dataset was used 

as opposed MOD13A1 version 6, which has a spatial resolution of about 500 m because the 

https://lpdaac.usgs.gov/products/mod13a2v006/


14 
 

MOD13A2 version 6 is available from 2000 to date while MOD13A1 version 6 is only available 

from 2011. 

3.4 Methodology 

3.4.1 Trends in NDVI across Protected areas in South Africa from 2001 to 2021 

For this study, the Google Earth Engine (GEE) was utilized. GEE is a platform that houses 

various satellite images and information, it also functions as a tool that helps scientist to analyze 

the data obtained from satellites. The GEE Code Editor was used to process the mean NDVI 

values of each of the protected areas and their buffer zones (5 km, and 20 km). The buffer zones 

were randomly selected, and the kilometers are enough/adequate to be able to compare the 

different vegetation cover change. The following datasets were imported into GEE Code Editor: 

a) Large Scale International Boundary (LSIB) from the National Geospatial-Intelligence Agency 

(NGA) whose function is to select the country of interest (South Africa), b) The World Database 

on Protected Areas (WDPA) whose purpose is to select protected areas in the country of interest, 

and MOD13A2 Terra Vegetation Indices which has the calculated NDVI values. Thereafter, a 

point of interest on a protected area on the map was picked for processing. In this coder, the 

JavaScript language was used to compile the code script in order to generate the mean NDVI 

values for each of the protected areas and their buffers. It is important to clarify that the mean 

NDVI values generated on the GEE were a composite of 16 days such that 2 mean NDVI values 

were calculated for each month. A total of 35 protected areas from eight biomes were selected: 

five protected areas from six biomes (Grassland, Savanna, Albany thicket, Succulent karoo, 

Nama-Karoo, and Indian coast) each, and three and two protected areas from two different 

biomes (Fynbos and Desert respectively). Mean NDVI values for each of the 20 years (2001 – 

2020) were calculated in GEE for each of the protected areas and their respective buffer zones (5 

km and 20 km). A time series was then generated for each of the protected areas and their 

respective buffer zones. 

The NDVI time-series was exported as csv into Microsoft Excel software to calculate annual 

mean NDVI values, to plot simple linear regression trendlines of the values, and also to derive 

slope values of the trendlines which were then used to understand the trends in vegetation cover 

change across protected areas. In addition, correlation analysis was done with Statistical Package 

for Social Sciences (SPSS) software to generate the Pearson correlation coefficient (r) which was 



15 
 

used to analyze the relationship between the trend in the mean annual NDVI in the protected 

areas and their buffer zones. Pearson correlation coefficient (r) values range from 0 to 1 and the 

higher the value the higher the relationship between the protected areas and their buffers.  

3.4.2 Vegetation cover change inside protected areas and their surrounding buffer zones (5 

km and 20 km) 

The slope values were used to detect the differences in the NDVI linear regression trendlines of 

the protected areas and the surrounding buffers (Anderson et al., 2005). Areas with negative 

slope values shows a decreasing vegetation cover and areas with positive slope values shows that 

there is an increase in vegetation cover. In a situation where the slope value is 0, it indicates that 

there is no change in vegetation cover in that area over a certain period of time. 

To detect the vegetation cover change inside and outside the protected areas, the slope values of 

the NDVI trendlines the protected areas and buffer zones were examined. There is a positive 

vegetation cover change when the slope value of the protected area is greater than the slope 

values of the buffer zones (5 km and 20 km), indicating that vegetation loss is at a higher rate in 

the buffer zones than inside the protected area or vegetation is increasing at a faster rate inside 

the protected area than the buffer zones. A negative vegetation cover change shows that the slope 

values in the buffer zones are greater than the slope value of the protected area which indicates 

that vegetation loss is at a higher rate inside the protected areas than the buffer zones or 

vegetation cover is increasing in the buffer zones at a faster rate than inside the protected areas. 

There is no vegetation cover difference when the slope value of the protected area is the same 

with the slope values of the buffers and this shows that rate of vegetation gain or loss is the same 

inside the protected area and in the buffer zones. 

 

  



16 
 

4. Results  

4.1 Trends in NDVI across protected areas in South Africa from 2001 to 2020 

This subsection presents the time series charts and outlines the trends in the Mean NDVI values 

of the selected protected areas and buffers (5 km and 20 km) from 2001 to 2020 for each of the 

biomes; Grassland, Savanna, Albany Thicket, Succulent Karoo, Nama-Karoo, Indian Coast, 

Fynbos, and Desert. All protected areas in six out of the eight biomes covered have negative trends 

in mean NDVI. The Grassland biome and the Nama-karoo biome have three to four of the 

protected areas with positive trend in NDVI.  

4.1.1 Grassland Biome 

On the average, there is a positive trend in NDVI for the Grassland biome from 2001 to 2020 as 

shown in Figure 4.1. All the protected areas in this biome have positive trends in Mean NDVI 

except the Golden Gate Highland NP which has a negative trend. Strong correlation exists for all 

parks within the Grassland biome between mean NDVI within the park, and its respective buffer 

zones with ‘r’ values ranging from 0.962 to 0.994 (5 km buffer), and 0.881 to 0.974 (20 km 

buffer) except for Oviston Nature Reserve with r values of 0.23 (5 km buffer) and 0.57 (20 km 

buffer) in Table 4.1. Ukahlamba-Drakensberg Park (UDP) and Golden Gate Highlands National 

Park (GGHNP) have higher ranges of mean NDVI values (0.51 - 0.52; 0.45 – 0.43) than their 5 

km buffers (0.49 - 0.50; 0.44 – 0.43) 20 km buffers (0.47 – 0.48; 0.43 – 0.42) whereas Oviston 

Nature Reserve has lower mean NDVI values (0.120 – 0.135) than its 5 km buffer (0.20 – 0.22) 

and 20 km buffer (0.23 – 0.235). The range of mean NDVI values inside Songimvelo Game 

Reserve is similar to its 5 km buffer (0.56 – 0.57), and lower than its 20 km buffer (0.58 – 0.59). 

In Motlase Crayon Nature Reserve, the range of mean NDVI values (0.58 -0.60) is lower than 5 

km buffer (0.62 – 0.64) but higher than 20 km buffer (0.57 – 0.59). 



17 
 

 

Figure 4.1: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Grassland 

biome; (a) Songimvelo Game Reserve (GR), (b) Golden Gate Highlands National Park 

(GGHNP), (c) Motlase Crayon Nature Reserve (NR) (d) Ukahlamba-Drakensberg Park (UDP), 

and (e) Oviston Nature Reserve (NR). 

4.1.2 Savanna Biome 

All protected areas in the Savanna biome have negative trends in mean NDVI throughout the 

study period (2001-2020) (Figure 4.2). The protected areas also have strong correlation with 

their respective buffers ranging from 0.985 to 1.000 (5 km) and 0.931 to 0.996 (20 km) (Table 

4.1). The range of mean NDVI values inside Pilanesberg Nature Reserve (0.44 – 0.42) is higher 

0.36

0.38

0.4

0.42

0.44

0.46

0.48

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

GGH NP

0.44

0.46

0.48

0.5

0.52

0.54

1995 2000 2005 2010 2015 2020 2025
M

e
an

 N
D

V
I

YEAR

Ukahlamba-Drakensberg Park

0.52

0.54

0.56

0.58

0.6

0.62

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Songimvelo GR

PA 5 km buffers 20 km buffer

Linear (PA) Linear (5 km buffers) Linear (20 km buffer)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Oviston NR

0.5
0.52
0.54
0.56
0.58

0.6
0.62
0.64
0.66

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Motlase Crayon NR 

(a) (b)

(d)(c)

(e)



18 
 

than its 5 km buffer (0.410 – 0.390) and 20 km buffer (0.405 – 0.39) while that of the Kruger 

National Park (0.43 – 0.40) is lower than its 5 km buffer (0.43 – 0.41) and 20 km buffer (0.44 – 

0.42). Kahalari Gemsbok National Park and Marakele National Park have close mean NDVI 

with their buffers (0.20 – 0.19 and 0.45 – 0.43), while the range of mean NDVI values of 

Madikwe Nature Reserve is the same with its 5 km (0.43 – 0.38) and higher than the 20 km 

buffer (0.42 – 0.37). 

 

Figure 4.2: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Savanna 

biome; (a) Kahalari Gemsbok National Park (NP), (b) Madikwe Nature Reserve (NR), (c) 

Kruger National Park (NP), (d) Marakele NP, and (e) Pilanesberg Nature Reserve (NR). 

0.15

0.2

0.25

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Kahalari Gemsbok NP

PA 5 km buffer 20 km buffer
Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

0.25

0.3

0.35

0.4

0.45

0.5

0.55

1995 2000 2005 2010 2015 2020 2025
M

e
an

 N
D

V
I

YEAR

Madikwe NR

0.3

0.35

0.4

0.45

0.5

0.55

0.6

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Kruger NP

0.35

0.4

0.45

0.5

0.55

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Marakele NP

0.3

0.35

0.4

0.45

0.5

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Pilanesberg NR

(b)(a)

(c)

(e)

(d)



19 
 

4.1.3 Albany Thicket Biome 

The protected areas and their buffers (5 km and 20 km) analyzed in this biome all have negative 

trends in mean NDVI over the 20 years period except the Water Meeting Nature Reserve where 

the mean NDVI values inside the PA and the 5 km buffer have positive trends and the 20 km 

buffer has negative trend (Figure 4.3). All the PAs in this biome have strong correlation with 

their respective biomes with ‘r’ values ranging from 0.932 to 0.996 (5 km) and 0.910 to 0.992 

(20 km) (Table 4.1). The parks have higher ranges of mean NDVI values than their buffers 

except for the Addo Elephant National Park with the range of mean NDVI values (0.535 – 0.500) 

inside the park slightly lower than 5 km buffer (0.540 – 0.505) and higher than 20 km buffer 

(0.50 – 0.46). 

 



20 
 

 

Figure 4.3: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Albany 

Thicket biome; (a) Addo Elephant National Park (NP), (b) Great Fish River Nature Reserve 

(NR), (c) Water Meeting Nature Reserve (NR), (d) East London Coast Nature Reserve (NR), and 

(e) Mount Coke State Forest. 

 

 

 

0.4

0.45

0.5

0.55

0.6

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Addo Elephant NP

PA 5 km buffer 20 km buffer

Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

0.35

0.4

0.45

0.5

0.55

0.6

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Great Fish River NR 

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Water Meeting NR 

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I
YEAR

East London Coast NR 

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

Year

Mount Coke State Forest 

(d)

(b)

(c)

(e)

(a)



21 
 

4.1.4 Nama-Karoo Biome 

The protected areas and their buffers shown in Figure 4.4 displays negative trends except 

Mountain Zebra National Park which has a slight negative trend while the buffers have positive 

trends in mean NDVI. The parks have strong correlation values with their buffers ranging from 

0.949 to 0.997 (5 km) and 0.897 to 0.978 (20 km) (Table 4.1). Mountain Zebra National Park 

and Augerbies Falls National Park range of mean NDVI values (0.31 – 0.30; 0.15 – 0.13) are 

lower than their respective 5 km buffers (0.31- 0.32; 0.16 – 0.15) and 20 km buffers (0.32 – 0.33; 

0.170 – 0.155). The range of mean NDVI values inside Camdeboo National Park (0.35 – 0.33) is 

higher than the mean NDVI values of the buffers (0.32 – 0.30; 0.31 – 0.29). For Mokala National 

Park, the range of mean NDVI values inside the park (0.30 – 0.27) is higher than the buffers at 

the beginning of the study but declined rapidly towards the end of the study with values the same 

as those of the 5 km buffer (0.29 – 0.27) but lower than the mean NDVI values of 20 km buffer 

(0.280 – 0.275). Geogap Nature Reserve and its buffers approximately have the same mean 

NDVI values (0.18 – 0.16) giving them the same trends. 



22 
 

 

Figure 4.4: Time series charts showing trendlines of mean NDVI values from 2001 to 2020 (20 

years) for the protected areas and their respective buffers (5 km and 20 km) in the Nama-Karoo 

biome; (a) Camdeboo National Park (NP), (b) Mokala National Park (NP), (c) Augerbies Falls 

National Park (NP), (d) Geogap Nature Reserve (NR), and (e) Mountain Zebra National Park 

(NP). 

4.1.6 Succulent Karoo Biome 

In the Succulent Karoo biome, all the protected areas and their buffers have negative trends in 

mean NDVI (Figure 4.5). They also all have strong correlation values with their buffers with ‘r’ 

values ranging from 0.859 to 0.998 (5 km) and 0.885 to 0.984 (20 km) (Table 4.1). The ranges of 

mean NDVI in Tankwa Karoo National Park, Moedverlooren Nature Reserve, and Vaalhoek 

Nature Reserve (0.155 – 0.145; 0.170 – 0.165; 0.30 – 0.28) are lower than their 5 km buffers 

0.2

0.25

0.3

0.35

0.4

0.45

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Camdeboo NP

PA 5 km buffers 20 km buffer
Linear (PA) Linear (5 km buffers) Linear (20 km buffer)

0.2
0.22
0.24
0.26
0.28

0.3
0.32
0.34
0.36

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Mokala NP

0.1

0.12

0.14

0.16

0.18

0.2

0.22

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Augerbies Falls NP 

0.12
0.13
0.14
0.15
0.16
0.17
0.18
0.19

0.2
0.21

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Geogap NR 

0.24

0.26

0.28

0.3

0.32

0.34

0.36

0.38

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Mountain Zebra NP (e)

(a) (b)

(c) (d)



23 
 

(0.17 – 0.15; 0.180 – 0.175; 0.35 – 0.33) and 20 km buffers (0.175 – 0.160; 0.205 – 0.175; 0.35 – 

0.30) while that of Akkerendam Nature Reserve is higher than its buffers (0.23 – 0.20; 0.22 – 

0.18). At the beginning of the study time, Namaqua National Park (0.235 – 0.215) and its buffers 

(0.240 – 0.215; 0.245 - 0.215) had slightly different mean NDVI values but by the end of the 

study they had the same values. 

 

 

Figure 4.5: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Succulent 

Karoo biome; (a) Namaqua National Park (NP), (b) Tankwa Karoo National Park (NP), (c) 

Moedverlooren Nature Reserve (NR), (d) Vaalhoek Nature Reserve (NR), and (e) Akkerendam 

Nature Reserve (NR). 

0.2

0.25

0.3

0.35

0.4

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Vaalhoek NR 

0.19

0.21

0.23

0.25

0.27

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Namaqua NP 

PA 5 km buffer 20 km buffer
Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

0.1

0.12

0.14

0.16

0.18

0.2

0.22

1995 2000 2005 2010 2015 2020 2025
M

e
an

 N
D

V
I

YEAR

Tankwa Karoo NP  

0.14
0.15
0.16
0.17
0.18
0.19

0.2
0.21
0.22
0.23
0.24

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Moedverloren NR 

0.14

0.16

0.18

0.2

0.22

0.24

0.26

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Akkerendam NR (e)

(d)(c)

(b)(a)



24 
 

4.1.6 Indian Coast Biome 

The PAs in this biome have positive trends in mean NDVI except for Mkambati Nature reserve 

with negative trends and Isimangaliso Provincial Nature Reserve with its 20 km buffer having 

negative trends and 5 km buffer has a positive trend (Figure 4.6). The parks all have a strong 

correlation with their buffers with ‘r’ values ranging from 0.600 to 0.996 (5 km) and 0.637 to 

0.992 (20 km) (Table 4.1). The ranges of Mean NDVI values inside Ngoye Forest Reserve and 

Dwesa-Cwebe Wildlife Reserve (0.74 – 0.78; 0.77 – 0.80) are higher than their 5 km buffer (0.63 

-0.67; 0.69 – 0.72) and their 20 km buffers (0.62 – 0.64; 0.65 – 0.68) while Isimangaliso 

Provincial Nature Reserve (0.57 – 0.56) is lower than its buffers (0.585 – 0.590; 0.585 -0.580). 

Mkambati Nature Reserve and Richards Bay Game Reserve mean NDVI values are 

approximately the same with their respective 5 km buffers (0.54 – 0.50; 0.57 – 0.59) but they are 

higher and lower respectively with their 20 km buffers (0.50 – 0.46; 0.64 – 0.65). 



25 
 

 

Figure 4.6: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Indian Coast 

biome; (a) Isimangaliso Provincial Nature Reserve (NR), (b) Mkambati Nature Reserve (NR), (c) 

Ngoye Forest Reserve (FR), (d) Richards Bay Game Reserve (GR), and (e) Dwesa-Cwebe 

Wildlife Reserve (WR). 

4.1.7 Fynbos Biome 

In the Fynbos biome, Rooiberg Nature Reserve, and Oorlogskloof Nature Reserve with their 

buffers have negative trends in mean NDVI values while West Coast National Park has a 

positive trend in mean NDVI (Figure 4.7). All the PAs have strong correlation values with their 

respective biomes ranging from 0.876 to 0.979 (5 km) and 0.845 to 0.940 (20 km) (Table 4.1). 

The ranges of mean NDVI values inside Oorlogskloof Nature Reserve, and West Coast National 

Park (0.35 – 0.34; 0.42 – 0.44) are lower than their respective 5 km buffers’ (0.30 - 0.27; 0.4 – 

0.5

0.55

0.6

0.65

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Isimangaliso Provincial NR 

PA 5 km buffer 20 km buffer

Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

0.35

0.4

0.45

0.5

0.55

0.6

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Mkambati NR

0.5
0.55

0.6
0.65

0.7
0.75

0.8

0.85

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Ngoye FR

0.55

0.6

0.65

0.7

0.75

0.8

0.85

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Dwesa-Cwebe Wildlife Reserve

0.5

0.55

0.6

0.65

0.7

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Richards Bay GR

(a)

(c)

(b)

(e)

(d)



26 
 

0.42) and 20 km buffers’ (0.25 - 0.23; 0.380 – 0.385). At the beginning of the study time, the 

range of mean NDVI values inside Rooiberg Nature Reserve (0.38 – 0.30) is higher than its 20 

km buffer (0.36 – 0.32) but lower than its 5 km buffer (0.41 – 0.33), and at the end of the study 

time, the values went below its buffers’. 

 

Figure 4.7: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the Fynbos 

biome; (a) Rooiberg Nature Reserve (NR), (b) Oorlogskloof Nature Reserve (NR), and (c) West 

Coast National Park (NP). 

4.1.8 Desert Biome 

Both protected areas and their respective buffers have negative trend in mean NDVI in the 

Desert biome (Figure 4.8). The correlation between Orange River Mouth Nature Reserve and its 

5km buffer is strong (0.706) and weak with its 20 km buffer (0.395) while the correlation 

between Richtersveld National Park and its buffers is strong r=0.994 (5 km) and r=0.982 (20 km) 

(Table 4.9). The range of mean NDVI inside the Orange River Mouth Nature Reserve (0.18 – 

016) is higher than its buffers which are relatively the same (0.1 – 0.09), and the range of mean 

NDVI values in Richtersveld National Park (0.145 – 0.120) is slightly lower than its buffers 

which are also relatively the same (0.148 – 0.125). 

0.3

0.35

0.4

0.45

0.5

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

Year

West Coast NP

0.15

0.2

0.25

0.3

0.35

0.4

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

Year

Oorlogskloof NR

0.2

0.25

0.3

0.35

0.4

0.45

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

Year

Rooiberg NR

PA 5 km buffer 20 km buffer

Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

(b)(a)

(c)



27 
 

 

Figure 4.8: Time series charts showing trendlines of mean NDVI values of the protected areas 

and their respective buffers (5 km and 20 km) from 2001 to 2020 (20 years) in the desert biome; 

(a) Richtersveld National Park (NP), and (b) Orange River Mouth Nature Reserve (NR). 

 

4.2 Vegetation cover change inside and outside the protected areas across South Africa 

from 2001 to 2020 

In this section, the slope values of mean NDVI versus the years of the protected areas and their 

respective buffers (5 km and 20 km) which were calculated during the linear regression analysis 

are presented. To understand the vegetation cover change of each of the protected areas, the 

differences in the slope values of the protected areas and their buffers were calculated (i.e.: 5 km 

buffer minus PA, 20 km minus PA) and are presented in the Tables below (Table 4.1 – 4.8).  

Slope differences with negative values means that there is a positive trend in vegetation cover 

change which indicates that the slope value of the PA is higher than that of the buffer which 

means that the PA is gaining vegetation faster or losing vegetation slower than its surrounding. 

Slope difference with positive values means there is a negative trend in vegetation cover change, 

and it shows that the slope value of the PA is lower than that of the buffer and this indicates that 

the vegetation cover inside the PA is increasing at a slower rate or declining at a faster rate than 

its surrounding. Having a slope difference value of 0 means that there is a flat trend in vegetation 

cover change which also means that the slope value of the PA is the same as that of the buffer.    

0.1

0.12

0.14

0.16

0.18

0.2

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Richtersveld NP 

PA 5 km buffer 20 km buffer

Linear (PA) Linear (5 km buffer) Linear (20 km buffer)

0

0.05

0.1

0.15

0.2

0.25

1995 2000 2005 2010 2015 2020 2025

M
e

an
 N

D
V

I

YEAR

Orange River Mouth NR (a) (b)



28 
 

4.2.1 Grassland Biome 

Ukahlamba- Drakensberg Park, Oviston Nature Reserve, and Motlase Crayon Nature Reserve 

have positive trends in vegetation cover change for both buffers as shown in Table 4.1. These 

parks experienced an increase in vegetation from 2001 to 2020 because the slope values inside 

the parks and around the parks are positive. Golden Gates Highlands National Park has a 

negative trend in vegetation cover change for both buffers. Songimvelo Game Reserve has a 

positive trend in vegetation cover change for 5 km buffer and a negative trend in vegetation 

cover change for 20 km buffer. 

Table 4.1: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Grassland biome including the slope differences between the PA and buffers.    

 

 

 
Protected Area (PA) 

 

Slope values 

 

 

Inside 

the PA 

5 km 

buffer 

20 km 

buffer 

Difference 

in slope (5 

km buffer 

– PA) 

Difference 

in slope 

(20 km 

buffer – 

PA) 

1 Golden Gates Highlands 

National Park -0.0013 -0.001 -0.001 0.0003 0.0003 

2 Ukahlamba- Drakensberg 

Park 0.0004 0.0002 -0.000008 -0.0002 -0.00041 

3 Songimvelo Game 

Reserve 0.0003 0.00004 0.0005 -0.00026 0.0002 

4 Oviston Nature Reserve 0.0015 0.0009 0.0006 -0.0006 -0.0009 

5 Motlase Crayon Nature 

Reserve 0.0008 0.0007 0.0006 -0.0001 -0.0002 



29 
 

 

4.2.2 Savanna Biome 

Kruger National Park and Madikwe Game Reserve have negative trends with the 5 km and 20 

km buffer. Marakele National Park have a positive vegetation cover change trends for both the 5 

km and 20 km buffers. Kalahari Gemsbok National Park has a flat trend with the 5 km buffer and 

a negative trend with the 20 km buffer. Pilanesberg Game Reserve has a positive trend for the 5 

km buffer and a negative trend for the 20 km buffer (Table 4.2). 

Table 4.2: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Savanna biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Kruger National Park -0.0027 -0.0026 -0.0022 0.0001 0.0005 

2 Kalahari Gemsbok 

National Park -0.0001 -0.0001 -0.00003 0 0.00007 

3 Marakele National Park -0.0006 -0.0008 -0.001 -0.0002 -0.0004 

4 Madikwe Game Reserve -0.0029 -0.0026 -0.0022 0.0003 0.0007 

5 Pilanesberg Game 

Reserve -0.0009 -0.001 -0.0007 -0.0001 0.0002 

 

 

 

 

 



30 
 

4.2.3 Albany Thicket Biome 

Addo Elephant National Park, Great Fish River Nature Reserve, and Mount Coke State Forest 

have experienced a negative vegetation cover change for 5 km buffer and positive vegetation 

cover change for 20 km buffer (Table 4.3). For East London Coast Nature Reserve, there is a 

positive trend in vegetation cover change for both 5 km and 20 km buffers. Water meeting 

Nature Reserve has a positive trend for 5 km buffer and a flat trend for 20 km buffer. 

Table 4.3: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Albany thicket biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Addo Elephant National 

Park -0.0019 -0.0018 -0.002 0.0001 -0.0001 

2 Great Fish River Nature 

Reserve -0.0014 -0.0011 -0.0017 0.0003 -0.0003 

3 Water meeting Nature 

Reserve 0.0002 -0.0002 0.0002 -0.0004 0 

4 East London Coast Nature 

Reserve -0.0004 -0.0009 -0.0005 -0.0005 -0.0001 

5 Mount coke State Forest -0.0004 0.0001 -0.0006 0.0005 -0.0002 

 

 

 

 



31 
 

4.2.4 Nama-Karoo Biome 

Mokala National Park, Geogap National Park, and Mountain Zebra National Park all experienced 

negative vegetation cover change for 5 km and 20 km buffers as shown in Table 4.4. Camdeboo 

National Park has positive trends for both buffers and Augrabies Falls National Park has a 

negative trend for 5 km and a flat trend for 20 km. 

Table 4.4: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Nama-Karoo biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km buffer 20 km 

buffer 

Change 

in slope 

(5 km 

buffer – 

PA) 

Change 

in slope 

(20 km – 

PA) 

1 Camdeboo National Park -0.0011 -0.0012 -0.0014 -0.0001 -0.0003 

2 Mokala National Park -0.0016 -0.0011 -0.0003 0.0005 0.0013 

 Augrabies Falls National 

Park -0.0009 -0.0008 -0.0009 0.0001 0 

4 Geogap National Park -0.0012 -0.0011 -0.0009 0.0001 0.0003 

5 Mountain Zebra National 

Park -0.00004 0.0003 0.0006 0.00034 0.00064 

 

 

 

 

 



32 
 

4.2.5 Succulent Karoo Biome 

Moedverloren Nature Reserve, Vaalhoel Nature Reserve, and Akkerendam Nature Reserve have 

positive vegetation cover change trend (Table 4.5). Although the slope values inside the PAs are 

negative, they are not as low as those of the buffers. This shows that the rate of vegetation loss 

inside the PAs is not as rapid as their surroundings. Tankwa Karoo National Park and Namaqua 

National Park has a flat trend with the 5 km and 20 km buffer respectively and a negative trend 

with 20 km and 5 km buffer respectively.  

Table 4.5: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Succulent Karoo biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Namaqua National Park -0.0013 -0.0011 -0.0013 0.0002 0 

2 Tankwa Karoo National 

Park -0.0007 -0.0007 -0.0006 0 0.0001 

3 Moedverloren Nature 

Reserve -0.0002 -0.0004 -0.0006 -0.0002 -0.0004 

4 Vaalhoel Nature Reserve -0.001 -0.0012 -0.0023 -0.0002 -0.0013 

 

 

 

 

 

 



33 
 

4.2.6 Indian Coast Biome 

All PAs experienced a negative and positive trend in vegetation cover change for 5 km and 20 

km buffers respectively except Ngoye Forest Reserve which has flat trend for 5 km buffer and 

Dwesa-Cwebe Nature Reserve which has negative trend for 20 km instead (Table 4.6). 

Table 4.6: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Indian Coast biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Richards Bay Game 

Reserve 0.0006 0.0012 0.0003 0.0006 -0.0003 

2 Ngoye Forest Reserve 0.0017 0.0017 0.0007 0 -0.001 

3 Isimangaliso Provincial 

Nature Reserve -0.000006 0.0002 -0.0002 0.000206 -0.00019 

4 Mkambati Nature Reserve -0.0019 -0.0017 -0.002 0.0002 -0.0001 

5 Dwesa-Cwebe 0.0008 0.0012 0.001 0.0004 0.0002 

 

 

 

 

 



34 
 

4.2.7 Fynbos 

West Coast National Park and Rooiberg Nature Reserve have positive and negative trends 

respectively while Oorlogskloof Nature Reserve has a positive trend for 5 km buffer and flat 

trend in 20 km buffer (Table 4.7). 

Table 4.7: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Fynbos biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Oorlogskloof -0.0009 -0.0011 -0.0009 -0.0002 0 

2 Rooiberg Nature Reserve -0.0044 -0.004 -0.0025 0.0004 0.0019 

3 West Coast National Park 0.0012 0.001 0.0003 -0.0002 -0.0009 

 

4.2.8 Desert Biome 

All protected areas in the biome experienced a negative trend in vegetation cover change for both 

buffers (Table 4.8). The PAs experienced vegetation loss at a faster rate inside than their 

surroundings. 

Table 4.8: Slope values of selected protected areas (PA) and their respective buffers (5 km and 

20 km) in the Desert biome including the slope differences between the PA and buffers.    

  

Protected Area 

 

Slope values 

Inside the 

PA 

5 km 

buffer 

20 km 

buffer 

Change 

in slope (5 

km buffer 

– PA) 

Change 

in slope 

(20 km – 

PA) 

1 Richtersveld National Park -0.0013 -0.001 -0.0012 0.0003 0.0001 

2 Orange River Mount -0.0007 -0.0004 -0.0005 0.0003 0.0002 



35 
 

4.3 Correlation between Protected Areas in South Africa and their Buffers (5km and 20 

km) from 2001 to 2020 

From the analysis in Table 4.9 below, it can be observed that majority of the protected areas have 

high correlation values with their respective buffers. Kruger National Park has the highest 

correlation values of 1.000 (5 km buffer) and 0.996 (20 km) while Oviston Nature Reserve has the 

least correlation values of 0.571 (5 km buffer) and 0.234 (20 km buffer).  

Table 4.9: Pearson correlation values derived from the correlation of Mean NDVI values of 

various protected areas South Africa, and their buffers (5 km and 20 km). 

 Biome Protected Area (PA) Correlation values between mean 

NDVI 

PA and 5 km PA and 20 km 

1 Grassland Golden Gates Highlands National 

Park 

0.962 0.881 

2 Ukahlamba- Drakensberg Park 0.962 0.881 

3 Songimvelo Game Reserve 0.994 0.974 

4 Oviston Nature Reserve 0.571 0.234 

5 Motlase Crayon Nature Reserve 0.974 0.946 

6 Savanna Kruger National Park 1.000 0.996 

7 Kalahari Gemsbok National Park 0.998 0.984 

8 Marakele National Park 0.985 0.931 

9 Madikwe Game Reseve 0.998 0.989 

10 Pilanesberg Game Reserve 0.994 0.982 

11 Albany Thicket Addo Elephant National Park 0.996 0.992 

12 Great Fish River Nature Reserve 0.993 0.977 

13 Water meeting Nature Reserve 0.958 0.924 

14 East London Coast Nature Reserve 0.973 0.967 

15 Mount coke State Forest 0.932 0.910 



36 
 

16 Succulent Karoo  Namaqua National Park 0.998 0.984 

17 Tankwa Karoo National Park 0.960 0.953 

18 Moedverloren Nature Reserve 0.859 0.885 

19 Vaalhoel Nature Reserve 0.987 0.963 

20 Akkerendam Nature Reserve 0.970 0.950 

21 Nama-Karoo Camdeboo National Park 0.992 0.975 

22 Mokala National Park 0.993 0.948 

23 Augrabies Falls National Park 0.997 0.978 

24 Geogap National Park 0.997 0.988 

25 Mountain Zebra National Park 0.949 0.897 

26 Indian Coast Richards Bay Game Reserve 0.600 0.637 

27 Ngoye Forest Reserve 0.855 0.738 

28 Isimangaliso Provincial Nature 

Reserve 

0.977 0.938 

29 Mkambati Nature Reserve 0.996 0.992 

30 Dwesa-Cwebe Nature Reserve 0.968 0.923 

31 Fynbos Oorlogskloof Nature Reserve 0.979 0.940 

32 Rooiberg Nature Reserve 0.876 0.779 

33 West Coast National Park 0.897 0.845 

34 Desert  Richtersveld National Park 0.994 0.982 

35 Orange River Mount Nature 

Reserve 

0.706 0.395 

 

 

 



37 
 

5. Discussion  

Various researchers have studied vegetation cover change of biomes across South Africa using 

remote sensing and have predicted significant changes due to climate change and increasing 

atmospheric CO2 (Okin 2007; Wegmann et al., 2014; Prevey et al., 2019). This study is aimed at 

investigating the rate and direction of change in NDVI from 2001 to 2020 inside and around 

protected areas in South Africa. The NDVI is used as a proxy to monitor vegetation cover 

change in the protected areas. 

In this study, eight biomes of South Africa were examined and all protected areas I investigated 

in six of the biomes experienced negative trends in NDVI values from 2001 to 2020 (Savanna, 

Albany thicket, Nana-Karoo, Succulent karoo, Fynbos and Desert). The two biomes that 

experienced positive trends in NDVI values are the Grassland and Indian Coast biomes. 

The Grassland biome of South Africa has experienced significant changes in vegetation cover 

over the years. According to the research carried out by Masubelele et al. (2015), there has been 

an evident increase in tall shrub and grass cover. As shown in Figure 4.1 (a, c-e), there is an 

increase in NDVI values with positive trends, which translates to an increase in grasses and tall 

shrubs. It has been reported that the Grassland biome is one of the least protected biomes in 

South Africa (Jewitt, 2018), and it has also been predicted that there would be an increase in 

aridification of the biome as a result of an increase in CO2 and temperature and decrease in 

rainfall thereby leading to reduced vegetation (Driver et al., 2012). This prediction contradicts 

the data gathered and results obtained as four out of five protected areas examined had a positive 

trend in NDVI (Figure 4.2) which means that there has been an increase in vegetation cover. The 

increase in vegetation in the Grassland biome could be because of an increase in atmospheric 

CO2, which supports the findings of Higgins and Scheiter's (2012), adaptive Dynamic Global 

Vegetation Model (aDGVM). The result of the model shows that increased atmospheric CO2 

alongside increased temperatures favors grasses and tall shrubs. The aDGVM is supported by 

Morgan et al. (2011) and Leakey et al. (2012) experiments on grasslands in the United States. 

Their experiments proved that increased atmospheric CO2 and temperature resulting in improved 

soil moisture conditions is an advantage for grasses leading to their increase (Bond, 2008). These 

conditions have been proven to exacerbate bush encroachment in grassland (Duker et al., 2015). 

On average, the trendlines of the mean NDVI values inside the protected areas in the Grassland 



38 
 

biome have steeper positive slope lines than their buffers (Table 4.1), which means that there is a 

faster rate of change and increase in vegetation cover inside the protected areas than their 

surroundings. The increase in vegetation cover may be attributed to an effective management 

plan in some protected areas such as culling of animals, fire management, and antipoaching 

activities (Mishra et al., 2015). In addition, the buffers of the protected areas may be 

experiencing a slower rate of vegetation growth because of human activities such as agriculture, 

animal grazing and land use change as observed by Jones et al. (2018). 

The majority of protected areas in the Savanna and Albany Thicket biomes experienced negative 

trends in NDVI, as shown in Figure 4.2. This could be because of the extreme and severe 

drought which lasted from 2015 to 2018 in South Africa. The drought significantly impacted 

South Africa's vegetation as vegetation rapidly declined due to lack or inadequate rainfall. 

According to Scholes and Walker (1993), the availability of water in the savanna is an important 

factor that shapes the structure of the ecosystem. Although there are other factors like movement 

of water, soil condition, temperature, and plant type, that influence the functions and structure of 

the savanna ecology, availability of water is the dominant factor for the savanna ecology because 

savannas are semi-arid, and they rely on seasonal rainfall to function well. A faster rate of 

change occurred and in a negative direction inside most of the protected areas compared to their 

buffers in the Savanna biome (Table 4.2). This means that although vegetation is on the decline 

in these protected areas, it is declining faster inside the protected areas than in their surroundings.  

In the Albany thicket, most protected areas have a faster rate and positive trends in vegetation 

cover change compared with their 20 km buffers than 5 km buffers (Table 4.3) which may have 

resulted from human activities occurring 20 km away from the protected areas. Activities such as 

construction of buildings and roads, overexploitation of natural resources, and mining can result 

to fragmentation and land degradation, pollution, introduction of invasive species and wildlife 

diseases (Schulze et al., 2018).  

The Protected areas in the Nama-Karoo and Succulent Karoo biomes experienced negative 

NDVI trends (Figure 4.4 and 4.5). Maestre at al. (2012) study supports this outcome and it is 

reported by the authors that the Karoo is experiencing enormous land degradation. A decrease in 

vegetation cover has been predicted to continue. It is expected that the area of the Karoos would 

increase in response to the projected increase in aridity as a result of the increasing impact of 



39 
 

climate change and anthropogenic activities, especially overgrazing (Schlaepfer et al., 2017). 

This expectation agrees with the projections made by Engelbrecht and Engelbrecht (2016) that 

the Karoos aridity would increase, thereby expanding desert-like environments. Most of the 

Succulent Karoo biome's protected areas have positive trends in vegetation cover change 

between inside and around the protected areas (Table 4.5), unlike the Nama-Karoo biome (Table 

4.4). These positive trends could be as a result of the increase in land set aside for conservation 

in the Succulent Karoo biome (Timm Hoffman et al., 2018) because the biome is widely known 

as a biodiversity hotspot which made it to be globally recognized (Myers et al., 2000) which is 

not the case with the Nama-Karoo. 

The Indian Coast biome shows a positive trend in NDVI values. Three out of the five protected 

areas studied experienced positive trends alongside their buffers (Figure 4.6c-e). This shows that 

there had been an increase in vegetation cover during the study time. Most likely, the vegetation 

in the protected areas and buffers have access to water services provided by the nearest ocean 

and may have also adapted to the saltwater from the ocean (Mucina et al., 2006), making them 

less affected by the intense drought in South Africa between 2015 and 2018. While most of the 

protected areas have negative trends in vegetation cover change as compared with their 

respective 5 km buffers, they have positive trends in vegetation cover change compared with 

their 20 km buffers (Table 4.6). These trends explain that the vegetation in the protected areas is 

doing better than in their respective 20 km buffers but not as good as their 5 km buffers. 

Anthropogenic activities tend to happen far from protected areas which explains why these 

protected areas and their 5 km buffers have better vegetation cover than their 20 km buffers. 

In the Fynbos, two out of the three protected areas experienced negative trends in mean NDVI 

values. Likewise, the two protected areas studied in the Desert biome also have negative trends 

in their mean NDVI values. The decline in vegetation might be as a result of the negative 

impacts of the extreme drought, climate change, and anthropogenic activities these biomes face. 

In this study, 74% of the studied protected areas and their buffer had negative trends in NDVI, 

which means that vegetation cover has reduced inside and outside these protected areas, which 

has, in turn, reduced biodiversity from 2001 to 2020. Jones et al. (2018) explained that protected 

areas face significant threats that can reduce their function in conserving biodiversity, providing 

ecosystem services, and reducing the size of the protected areas. One of the threats highlighted 



40 
 

by King et al. (2015) is climate change, and its adverse effect on vegetation in South Africa has 

been on the increase. Protected areas also face threats from humans because these areas were 

formally their homelands which supported their livelihood (Garnett et al., 2018), and the increase 

in the human population and urbanization has significant effects on the biodiversity inside and 

outside protected areas (Richards & Friess, 2016; Barlow et al., 2016). Basommi (2016) 

elaborated in his study that there is a correlation between the increase in human activities around 

the protected areas (buffer) and the extinction rate of species. These activities are poaching, 

deforestation, land conversion, and urban settlement. The primary purpose of the buffer area is to 

give additional protection to the protected area and reduce conflicts between the local people 

surrounding the protected areas and the protected area by putting less restriction on the resources 

in the buffers (Hattter & Southworth, 2009). 

On the other hand, Wells and Brandon (1992) disagree that providing benefits that are limited 

through the buffer zone to the people will change their attitude towards protected areas. The 

results of this study may support Wells and Brandon (1992) because most of the protected areas 

studied have lesser vegetation cover inside than their surroundings. According to McDonald et 

al. (2009), the average distance between protected areas and urban settlement is less than 50 km. 

This distance is reducing drastically, putting pressure on the protected areas. Guneralp and Seto 

(2013) predicted that between 2000 and 2030, urban settlements in a 50 km buffer of protected 

areas would be multiplied by three. 

Nevertheless, some protected areas such as Marakele National Park, West Coast National Park, 

Motlase Crayon Nature Reserve and so on are not facing pressure from their surroundings. As a 

matter of fact, vegetation cover change is faster and in a positive direction in these protected 

areas. This may be because of some favorable practices carried out by the management of the 

protected areas, which improves vegetation growth. Such practices include but are not limited to 

the culling of animals, fire management, and antipoaching activities (Mishra et al., 2015). 

Culling of animals for example may have and impact of vegetation abundance. This brings a 

balance between the animals and available vegetation. 

From the results (Table 4.1 – 4.8), it is deduced that 29% of the protected areas have a faster rate 

of vegetation cover change in a positive direction than both of their respective buffers. 40% and 

49% of the protected areas studied have a faster rate of vegetation cover change in the positive 



41 
 

direction than their 5 km and 20 km buffers respectively. The percentages (40% and 49%) 

explain that vegetation is reducing more in the 20 km buffer than in the 5 km buffers. This 

difference is expected since 20 km away from the protected areas is closer to human settlements 

and exposes the area to human activities like agriculture, animal grazing, deforestation, and 

development. The study carried out by De La Fuente et al. (2020) shows that there are more 

infrastructures in areas surrounding protected areas than inside the protected areas. Although 

they studied 10 km away from the protected areas, their results can be related to what is 

happening in the 20 km buffers of this study. 

Even though, according to De La Fuente et al. (2020), there are more infrastructures happening 

in areas surrounding the protected areas, the overall result of this study shows that vegetation is 

reducing faster in most of the protected areas than in their surroundings, especially the 5 km 

buffers. The faster rate of decline may be because the protected area experiences some level of 

development, especially for tourist satisfaction. Another reason may be the uncontrolled 

population growth of animals, which will negatively impact the protected area (Guldermond & 

van Aarde, 2008). As the population of animals in the park increases, since they directly or 

indirectly feed on plants, vegetation cover rapidly reduces if they are not given enough time to 

replenish (Young et al., 2009).   

The correlation analysis done between the protected areas and their buffers (Table 4.9) shows 

that there are strong correlation values between them except for the Oviston Nature Reserve with 

correlation values of 0.571 (5 km) and 0.234 (20 km) and Orange River Mount Nature Reserve 

with a correlation value of 0.395 (20 km) (Table 4.9). It is assumed that protected areas with 

strong correlation values show that what is happening inside the protected area is also happening 

in the buffers. The protected areas with weak correlation values have little to no relationship 

between what is happening in the protected area and the buffers. 

As earlier pointed out, this study expanded on the vegetation cover change study of Wegmann et 

al. (2014). While they examined protected areas across Africa, this study focused on protected 

areas across South Africa. The main aim of their study was to assess the habitat change of 

species in protected areas across Africa. An important method used was to get information on 

land cover change of the protected areas over time. In doing so, vegetation cover change of the 

protected areas was assessed from 1982 to 2006 with remote sensing and NDVI. They studied 



42 
 

the vegetation cover change inside and outside the protected areas using 17 km, 31 km, and 84 

km buffers. Their results show that most of the protected areas in Africa, particularly in South 

Africa, have negative trends in vegetation cover change, i.e., vegetation is increasing or reducing 

slower in the buffer zones compared to within the protected areas. Wegmann et al. (2014) result 

is similar to the result of this study because most of the protected areas examined have negative 

trends in vegetation cover change. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



43 
 

6. Conclusion 

Vegetation structure and composition in protected areas is faced with significant threats due to 

climate change and anthropogenic activities. Although various studies have been done to monitor 

vegetation cover change in protected areas and their buffers with the aid of remote sensing and 

NDVI, none has examined protected areas in different biomes across South Africa. This study 

extensively examined the trends in NDVI across protected areas with their buffers (5 km and 20 

km) in South Africa. It expanded on the Wegmann et al. (2014) study of vegetation cover 

changes inside and outside protected areas from 2001 to 2020. This study was done with remote 

sensing in Google Earth Engine, and NDVI was used as a proxy to examine the vegetation cover 

change of the protected areas and their buffers (5 km and 20 km). A total of 35 protected areas 

were studied; it is reported that 26 out of the 35 protected areas have negative trends in NDVI, 

and 10 of the protected areas have a positive trend in vegetation cover change when compared to 

their buffers (5 km and 20 km).  

The results of this study should be carefully used because even though NDVI is used as a proxy 

for vegetation and this study looked at NDVI trends to examine vegetation cover change, it is not 

entirely sufficient to determine the vegetation cover change of an area. Other variables that can 

influence vegetation cover change such as temperature, rainfall, soil condition, and human 

population must be examined in further research since it is not in the scope of this study. 

However, the NDVI trends give us an idea of how well vegetation in protected areas across 

South Africa is doing. Managers of protected areas can utilize the result of this study to guide 

them to areas and biomes that are vulnerable to the impacts of climate change and anthropogenic 

activities, thereby needing urgent attention. Managers of the protected areas with negative 

vegetation cover change should be aware that there is at least a condition in the protected area 

that is not favorable thereby practices like fire management, culling of animals, and antipoaching 

activities must be effectively done.  

 

  



44 
 

References 

Adepoju, K., Adelabu, S., & Fashae, O. (2019). Vegetation response to recent trends in climate 

and landuse dynamics in a typical humid and dry tropical region under global change. Advances 

in Meteorology, 5(4), 536-800. 

Alcaraz-Segura, D., Baldi, G., Durante, P., & Garbulsky, M. F. (2008). Analysis of the temporal 

dynamics of NDVI in protected areas: three case studies at different spatial, temporal and 

management scales. Ecosystems, 17 (3), 678-948. 

Anderson, M. C., Norman, J. M., Kustas, W. P., Li, F., Prueger, J. H., & Mecikalski, J. R. 

(2005). Effects of vegetation clumping on two–source model estimates of surface energy fluxes 

from an agricultural landscape during SMACEX. Journal of Hydrometeorology, 6(6), 892-909. 

Archer, S. R., Andersen, E. M., Predick, K. I., Schwinning, S., Steidl, R. J., & Woods, S. R. 

(2017). Woody plant encroachment: Causes and consequences. In D. Briske (Ed.), Rangeland 

systems (pp. 25–84). Cham, Switzerland: Springer. 

Archibald, S., Lehmann, C. E., Gómez-Dans, J. L., & Bradstock, R. A. (2013). Defining 

pyromes and global syndromes of fire regimes. Proceedings of the National Academy of 

Sciences, 110(16), 6442-6447. 

Avilés-Polanco, G., Jefferson, D. J., Almendarez-Hernández, M. A., & Beltrán-Morales, L. F. 

(2019). Factors that explain the utilization of the nagoya protocol framework for access and 

benefit sharing. Sustainability, 11(20), 5550. 

Balme, G. A., Slotow, R., & Hunter, L. T. B. (2010). Edge effects and the impact of non-

protected areas in carnivore conservation: Leopards in the Phinda-Mkhuze complex, South 

Africa. Animal Conservation, 13, 315–323. 

Barati, S., Rayegani, B., Saati, M., Sharifi, A., & Nasri, M. (2011). Comparison the accuracies of 

different spectral indices for estimation of vegetation cover fraction in sparse vegetated 

areas. The Egyptian Journal of Remote Sensing and Space Science, 14(1), 49-56. 

Barger, N., Barger, T. A., Sankaran, M., & Meyfroidt, P. (2018). Direct and indirect drivers of 

land degradation and restoration (No. UCL-Université Catholique de Louvain). p. 198–314.  



45 
 

Barlow, J., Lennox, G. D., Ferreira, J., Berenguer, E., Lees, A. C., Mac Nally, R., ... & Gardner, 

T. A. (2016). Anthropogenic disturbance in tropical forests can double biodiversity loss from 

deforestation. Nature, 535(7610), 144-147. 

Basommi, L. P., Guan, Q. F., Cheng, D. D., & Singh, S. K. (2016). Dynamics of land use change 

in a mining area: a case study of Nadowli District, Ghana. Journal of Mountain Science, 13(4), 

633-642. 

Beck, P. S., Juday, G. P., Alix, C., Barber, V. A., Winslow, S. E., Sousa, E. E., ... & Goetz, S. J. 

(2011). Changes in forest productivity across Alaska consistent with biome shift. Ecology 

letters, 14(4), 373-379. 

Bond, W.J. (2008). What limits trees in C4 grasslands and savannas? Annual Review of Ecology, 

Evolution and Systematics 39: 641–659. 

Buitenwerf, R., Bond, W. J., Stevens, N., & Trollope, W. (2012). Increased tree densities in 

South African savannas:> 50 years of data suggests CO 2 as a driver. Global Change 

Biology, 18(2), 675-684. 

Butchart, S. H. M., Walpole, M., Collen, B., Van Strien, A., Scharlemann, J. P. W., Almond, R. 

E. A., & Watson, R. (2010). Global biodiversity: Indicators of recent declines. Science, 

328(5982), 1164–1168.  

Bux, Q., P. Anderson, & P. J. O'Farrell. (2021). Understanding the local biodiversity and open 

space strategies in two South African cities. Ecology and Society, 26(3):4. 

Cho, M. A., & Ramoelo, A. (2019). Optimal dates for assessing long-term changes in tree-cover 

in the semi-arid biomes of South Africa using MODIS NDVI time series (2001–

2018). International Journal of Applied Earth Observation and Geoinformation, 81, 27-36. 

Chuvieco, E., Aguado, I., Salas, J., García, M., Yebra, M., & Oliva, P. (2020). Satellite remote 

sensing contributions to wildland fire science and management. Current Forestry Reports, 6(2), 

81-96. 

COP14, S. S. F. (2019). Convention on International Trade in Endangered Species of Wild 

Fauna and Flora _ CITES CoP18, 1-22. 



46 
 

De Baan, L., Alkemade, R., & Koellner, T. (2013). Land use impacts on biodiversity in LCA: a 

global approach. The International Journal of Life Cycle Assessment, 18(6), 1216-1230. 

De La Fuente, B., Bertzky, B., Delli, G., Mandrici, A., Conti, M., Florczyk, A. J., ... & Dubois, 

G. (2020). Built-up areas within and around protected areas: Global patterns and 40-year 

trends. Global Ecology and Conservation, 24, 01291. 

DeFries, R., Hansen, A., Newton, A. C., & Hansen, M. C. (2005). Increasing isolation of 

protected areas in tropical forests over the past twenty years. Ecological applications, 15(1), 19-

26. 

Defries, R., & Nagendra, H. (2017). Ecosystem management as a wicked problem. Science 

80:265–270. 

Didan, K. (2015). MOD13A2 MODIS/Terra Vegetation Indices 16-Day L3 Global 1km SIN Grid 

V006 [Data set]. NASA EOSDIS Land Processes DAAC. Accessed 2021-08-31 from 

https://doi.org/10.5067/MODIS/MOD13A2.006. 

Driver, A., Sink, K.J., Nel, J.L., Holness, S.H., Van Niekerk, L., Daniels, F., Jonas, Z., Majiedt, 

P.A., Harris, L. & Maze, K. (2012). National biodiversity assessment 2011: an assessment of 

South Africa’s biodiversity and ecosystems. Synthesis report. South African National 

Biodiversity Institute & Department of Environmental Affairs, Pretoria, ZA. 

Dudley, N. (2008). Guidelines for Applying Protected Area Management Categories. 

www.iucn.org. 

Dudley, N., & Stolton, S. (Eds.). (2010). Arguments for pro