COPYRIGHT AND CITATION CONSIDERATIONS FOR THIS THESIS/ DISSERTATION
o Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
o NonCommercial — You may not use the material for commercial purposes.
o ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.
How to cite this thesis
Surname, Initial(s). (2012) Title of the thesis or dissertation. PhD. (Chemistry)/ M.Sc. (Physics)/ M.A. (Philosophy)/M.Com. (Finance) etc. [Unpublished]: University of Johannesburg. Retrieved from: https://ujdigispace.uj.ac.za (Accessed: Date).
A SPATIAL PRIORITIZATION OF THREATS TO BIODIVERSITY AND CONSERVATION IN THE PROPOSED MAGALIESBERG BIOSPHERE
by
Belinda Anne Cooper
A Minor Dissertation submitted to the Faculty of Science, in partial fulfilment of the academic requirements for the Degree of Magister Artium in Environmental Management
Department of Geography, Environmental Management and Energy Studies, University of Johannesburg
Supervisor: Dr. Clare Kelso Co-Supervisor: Prof. Fethi Ahmed
May 2015
Affidavit: MASTER AND DOCTORAL STUDENTS
TO WHOM IT MAY CONCERN
This serves to confirm that I_ BELINDA ANNE COOPER
Full Name(s) and Surname
ID Number/ Passport___
Student number enrolled for the Qualification
MASTER OF ARTS IN ENVIRONMENTAL MANAGEMENT
in the Faculty of Science
Herewith declare that my academic work is in line with the Plagiarism Policy of the University of Johannesburg with which I am familiar.
I further declare that the work presented in the _MINOR DISSERTATION _(minor dissertation/dissertation/thesis) is authentic and original unless clearly indicated otherwise and in such instances full reference to the source is acknowledged and I do not pretend to receive any credit for such acknowledged quotations, and that there is no copyright infringement in my work. I declare that no unethical research practices were used or material gained through dishonesty. I understand that plagiarism is a serious offence and that should I contravene the Plagiarism Policy notwithstanding signing this affidavit, I may be found guilty of a serious criminal offence (perjury) that would amongst other consequences compel the University of Johannesburg to inform all other tertiary institutions of the offence and to issue a corresponding certificate of reprehensible academic conduct to whomever requests such a certificate from the institution.
Signed at Johannesburg on this 31 of May 2015
Signature Print name _BA COOPER
STAMP: COMMISSIONER OF OATHS
Affidavit certified by a Commissioner of Oaths
This affidavit conforms with the requirements of the JUSTICES OF THE PEACE AND COMMISSIONERS OF OATHS ACT 16 OF 1963 and the applicable Regulations published in the GG GNR 1258 of 21 July 1972; GN 903 of 10 July 1998; GN 109 of 2 February 2001 as amended.
ii Abstract
The research is conducted in the Magaliesberg region of South Africa, in the context of the area being promoted as a Biosphere Reserve in the UNESCO Man and Biosphere Programme. The programme recognises the need to reconcile the conservation of biodiversity with associated cultural value and sustainable socio-economic development, which is demonstrated in designated special areas that contain both unique biodiversity and a gradation of human interventions. Considering the proposed reserves proximity to the economic hub of South Africa - the Gauteng city region – it’s exceptional biodiversity is threatened by extreme modifications to the landscape, by increasing pressures of population, development and extraction which should be incorporated into conservation assessment at a regional scale. Combining the conservation value of an area with an assessment of the particular threats resulting in degradation and loss of habitat, can connect the vulnerability of priority conservation areas to particular threatening processes, and enable the identification of priority regions for targeted conservation action and social upliftment in the biosphere context.
The research is situated as spatial decision support, and presents a systematic regional approach to quantifying and analysing threats and referencing them to the terrestrial biodiversity landscape. The work can be summarised as a conservation and threats assessment that quantifies the exposure (by geographic extent) and intensity (by AHP) of multiple threat criteria to determine priority threats in relation to conservation value. This is achieved using a Spatial Multi Criteria, Analytical Hierarchy Process, approach, whereby relative weights for multiple threat criteria are formalised by expert judgement into a quantitative record, and applied in a weighted linear overlay. First, the spatial variability of several individual threatening processes is assessed in relation to priorities for conservation, after which a composite of priority threats is integrated with priorities for conservation in a single map, and congruent graphic bi-plot. These products can be used to identify combinations of priorities for threat and conservation in priority quadrants that may inform regional spatial planning, and conservation responses.
Each threatening process shows different results for exposure and intensity, with a general pattern for the region emerging, i.e. that threats are concentrated and severe (less exposure, more intensity) or widespread and less severe (more exposure, less intensity). Analysing threats with reference to priority conservation areas showed that factors of habitat decline pervade all priority conservation zones, including formally protected areas with proportionally more exposure in high value conservation areas, compared to habitat loss threats. The distribution of threats of habitat loss in relation to priority conservation areas, indicate a negative linear relationship – they decrease in extent as conservation value increases. Using the priority quadrant approach, results show that 17% of high value conservation land is severely threatened, while 48% is not currently threatened, whereas, 53% of less valuable conservation areas are pervaded by multiple high priority threats. Including criteria for habitat loss and habitat decline and quantifying both extent and intensity of threat, has improved estimations of the vulnerability of biodiversity from a regional perspective, and it is likely that threats to biodiversity in general would have been underestimated had only habitat loss threats been considered for analysis.
iii Acknowledgements
I acknowledge and thank the institutions and individuals who so willingly provided spatial data, and allowed me to use it for this research. My gratitude goes to the experts – the members of MBIG and others - who gave their time and participated in the AHP. The financial assistance of the National Research Foundation (NRF) is also hereby acknowledged.
Many thanks go to my supervisors Dr Clare Kelso and Prof Fethi Ahmed for their valuable insights and helpful comments which have taught me a great deal about research and have improved this dissertation tremendously. Clare, your support and guidance through the process is appreciated.
My love and thanks to my mother for her help and support, and to family and friends, for their encouragement. Special love and heartfelt gratitude go to Dave and Julia for their patience, and for so graciously allowing me the time and space to complete this work.
iv Table of Contents
Affidavit ...... ii
Abstract ...... iii
Acknowledgements ...... iv
Table of Contents ...... v
List of Figures ...... x
List of Tables ...... xii
List of abbreviations ...... xiii
Chapter 1 Introduction ...... 1
1.1 Introduction ...... 1
1.2 Research Motivation ...... 2
1.3 Problem Statement and Hypothesis ...... 5
1.4 Research Aims and Objectives ...... 5
1.5 Report Outline ...... 6
Chapter 2 The proposed Magaliesberg Biosphere Reserve in context...... 8
2.1 Biophysical background ...... 8
2.1.1 Location ...... 8
2.1.2 Geology and Topography ...... 9
2.1.3 Hydrology...... 10
2.1.4 Macro Climate ...... 11
2.1.5 Mountain Habitats (Micro Climate and Ecology) ...... 12
2.1.6 Regional Habitats (Biomes, Bioregions and Vegetation-Types) ...... 14
2.1.7 Species Diversity and Conservation ...... 17
2.1.8 Anthropogenic footprint and heritage ...... 20
2.2 Biosphere Reserves and the Magaliesberg Biosphere Nomination ...... 23
2.2.1 Background to biosphere reserves ...... 23
2.2.2 The beginnings of the proposed MBR ...... 24
v 2.2.3 Current MBR nomination status ...... 24
2.2.4 Current MBR conservation status and proposed zonation ...... 26
2.2.5 The spirit of a biosphere - conservation and sustainability opportunities ...... 26
2.3 Summary ...... 28
Chapter 3 Literature Review –Threats to Biodiversity in the Magaliesberg Region and Spatial Assessment of Threats...... 29
3.1 Introduction ...... 29
3.2 Magaliesberg Regional Threats to Terrestrial Biodiversity ...... 29
3.2.1 Rates of land transformation relevant to the Magaliesberg region ...... 30
3.2.2 Threats associated with habitat loss in the MBR ...... 31
3.2.2.1 Urbanization ------31
3.2.2.2 Mining ------34
3.2.2.3 Cultivation ------35
3.2.3 Threats associated with habitat degradation in the MBR ...... 36
3.2.3.1 Overgrazing ------37
3.2.3.2 Erosion ------38
3.2.3.3 Old lands ------38
3.2.3.4 Fire ------39
3.2.3.5 Alien invasive species ------40
3.2.3.6 Fragmentation ------41
3.2.4 Magaliesberg regional threats synthesis ...... 42
3.3 Assessment of Threats ...... 43
3.3.1 Measures of threat in conservation planning ...... 43
3.3.2 Spatial indicators of vulnerability and threat ...... 45
3.3.3 Integrating GIS, Multi Criteria Analysis and the Analytical Hierarchy Process ...... 47
3.3.4 A review of methodologies that incorporate vulnerability and threat assessment...... 49
3.3.5 From theory to practice - the implementation of priorities ...... 52
3.4 Summary ...... 53 vi Chapter 4 Materials and Methods ...... 54
4.1 Setting the Scene ...... 54
4.1.1 Overview of research objectives and methodology ...... 54
4.1.2 Data acquisition and limitations ...... 55
4.1.3 Software used and general pre-processing ...... 56
4.2 Priority Biodiversity Conservation Value (PCV) ...... 57
4.2.1 Overview of objective and method ...... 57
4.2.2 Data characteristics ...... 58
4.2.3 GIS methods and weightings for layers ...... 58
4.2.4 Layer 1 - Biosphere zones and protected areas ...... 58
4.2.4.1 Data ------58
4.2.4.2 GIS processing and weighting ------59
4.2.5 Layer 2 – Provincial biodiversity assessments ...... 60
4.2.5.1 Data ------60
4.2.5.2 GIS Processing and weighting ------62
4.2.6 Layer 3 – Remaining extent of threatened ecosystems, 2011 ...... 63
4.2.6.1 Data ------63
4.2.6.2 GIS processing and weighting ------64
4.2.7 Layer 4 – Vegetation types of South Africa ...... 64
4.2.7.1 Data ------65
4.2.7.2 GIS processing and weighting ------65
4.2.8 Overlay analysis – Priority conservation value ...... 66
4.3 Priority Biodiversity Threats (PBT) ...... 67
4.3.1 Overview of objective and method ...... 67
4.3.2 Data characteristics ...... 68
4.3.3 Biodiversity threats layers from classified land-cover/ land-use data ...... 69
4.3.3.1 Description of Provincial Land Cover Datasets ------69
vii
4.3.3.2 Land-Cover as the underlying spatial surrogate for threats ------69
4.3.3.3 GIS processing ------70
4.3.3.4 Urbanization Layer ------70
4.3.3.5 Cultivation Layer------71
4.3.3.6 Mining Layer ------72
4.3.3.7 Transformed Open Land Layer ------72
4.3.3.8 Fragmentation Layer ------73
4.3.4 Biodiversity threat layers from other data sources ...... 75
4.3.4.1 Alien Invasive Plants Layer ------75
4.3.4.2 Fire Events Layer ------77
4.3.5 Summary of seven threat layers ...... 79
4.4 The Analytic Hierarchy Process (AHP) Approach ...... 79
4.4.1 AHP threats focus-group meeting and questionnaire ...... 80
4.4.2 Application of the AHP Calc template ...... 83
4.4.3 Applying AHP weightings and sub-criteria scores ...... 85
4.5 Overlay Analysis – Priority Threats to Biodiversity Composite ...... 86
4.6 Incorporating Planning Units for Spatial Analysis ...... 86
4.7 The Synthesis of Priority Conservation Value and Priority Biodiversity Threats ...... 87
4.7.1 Original farm portions cross-classified ...... 87
4.7.2 Pixel-resolution cross-classified map and tables ...... 87
4.8 Summary ...... 88
Chapter 5 Results & Discussion - Spatial dimensions of current land transformation and conservation value ...... 89
5.1 Outline ...... 89
5.2 A spatial synopsis of Priority Conservation Value ...... 89
5.3 The Spatial Extent of Threatening Processes in the MBR ...... 91
5.4 AHP Ranking of Threat Criteria ...... 93 viii 5.5 Combining Quantified Intensity and Exposure of Threats ...... 95
5.6 The Relative Ranking of Sub-Criteria ...... 97
5.7 General Patterns of Exposure of Priority Conservation Value to Individual Threats ...... 98
5.7.1 PCV exposure to habitat loss...... 99
5.7.2 PCV exposure to habitat degradation ...... 102
5.8 Priority Biodiversity Threats Composite (PBT) ...... 106
5.9 Defining Spatial Priorities for Conservation – The Integration of Proirity Conservation Value and Priority Biodiversity Threats ...... 109
5.9.1 Integration by original farm portions ...... 109
5.9.2 Pixel-resolution integration ...... 113
5.10 Summary ...... 117
Chapter 6 Conclusions and Recommendations ...... 119
6.1 Overview of Research Design and Study Context ...... 119
6.2 Research Shortcomings, Achievements and Recommendations...... 120
6.3 Conclusion ...... 123
Reference List ...... 127
Appendices...... 141
Appendix 1 ...... 141
Appendix 2 ...... 143
Appendix 3 ...... 144
Appendix 4 ...... 150
Appendix 5 ...... 151
Appendix 6 ...... 154
ix
List of Figures
Figure 1: Location of the proposed Magaliesberg Biosphere Reserve, South Africa ...... 9
Figure 2: The Magaliesberg, looking west over the Hartebeespoort dam and Moot valley beyond. 12
Figure 3: Layer 4 - Vegetation Types of the MBR ...... 14
Figure 4: Population Density estimated from Spot Building count (Eskom, 2008) and census data (Lightstone, 2010) ...... 22
Figure 5 - Extract from Article 4 of the Statutory Framework of the World Network of Biosphere Reserves (UNESCO, 1996) ...... 25
Figure 6: Wasteland between the Lonmin mine and the Enkaneng informal settlement ...... 35
Figure 7 Examples of different approaches for irreplaceability and vulnerability mapping ...... 51
Figure 8: Flow of Methodological Steps – numbered according to research objectives, ...... 55
Figure 9: Layer 1 Biosphere zones (2012 Nomination) ...... 60
Figure 10: Layer 2 - Critical Biodiversity Areas ...... 63
Figure 11: Layer 3 - Remaining extent of Threatened Ecosystems ...... 64
Figure 12: Priority Conservation Value (PCV) – continuous ranking weighted –sum...... 66
Figure 13: Urbanization Threats Layer ...... 71
Figure 14: Cultivation Threats Layer ...... 71
Figure 15: Mining Threats Layer ...... 72
Figure 16: Transformed open land Threats Layer ...... 73
Figure 17: Fragmentation Threat Layer ...... 74
Figure 18: Merged terrestrial and Riparian invasion rasters, showing equivalent classes but different density and range values...... 76
Figure 19: Alien Plant Invasion Threat Layer ...... 77
Figure 20 - Boolean map of burns for 2011 ...... 78 x Figure 21: Fire Events Threat Layer...... 79
Figure 22: Hierarchical model used in the AHP ...... 80
Figure 23: Priority Threats to Terrestrial Biodiversity (PBT) –continuous ranking from weighted–sum ...... 86
Figure 24: Priority Conservation Value (PCV), showing formally protected areas within the MBR. .. 90
Figure 25: The geographic extent of threats within the MBR ...... 92
Figure 26: The contrast between normalised proportions for AHP weightings and geographic extent of threat criteria...... 95
Figure 27: 3- dimensional plot of the vulnerability of Priority Conservation Value in the MBR ...... 96
Figure 28: The exposure of PCV categories to threat factors ...... 99
Figure 29: Exposure of PCV to sub-criteria threats for habitat loss, in hectares...... 100
Figure 30: The spatial distribution of threats associated with habitat loss across the MBR ...... 102
Figure 31: Exposure per hectare and spatial distribution of habitat degradation threats by sub- criteria, across PCV categories...... 104
Figure 32: Spatial representation of extent and severity of Habitat degradation threats of the MBR...... 106
Figure 33: Categorised Priority Biodiversity Threats - Composite of 7 threat layers ...... 107
Figure 34: Priorities for original farms plotted on two axes...... 111
Figure 35: Original farms prioritized for both conservation value and threat ...... 112
Figure 36: Priority Conservation Value and Priority Biodiversity Threats, ...... 113
Figure 37: Integrated priority maps and graph of conservation and threat ...... 114
xi List of Tables
Table 1 – The Conservation Status and Extent of Vegetation-types in the Magaliesberg Biosphere...... 17
Table 2 - Classified Threatened and Near Threatened Species of the Magaliesberg Region ...... 19
Table 3 - Remaining extent of Threatened Ecosystems in the Biosphere (Source: NBA, 2011) ...... 20
Table 4 - List of eleven woody IAP’s and their listed invader category status ...... 41
Table 5 - Data for PCV map layer...... 58
Table 6 - Scores allocated to the 'zone' field ...... 59
Table 7 - Scoring applied to Critical Biodiversity Areas ...... 62
Table 8 - Criteria used to identify Threatened Ecosystems, and the thresholds applied to determine threat status ...... 63
Table 9 - Scoring applied to Remaining Extent of Thretened Ecosystems ...... 64
Table 10 - Vegetation-Types found in the Magaliesberg Biosphere and their allocated weights ...... 65
Table 11 - Data used for Biodiversity Threat Layers ...... 68
Table 12 - Threat criteria and the AHP scale for comparisons (adapted from Goepel, 2012) ...... 82
Table 13 - Example of sub-criteria within threat criteria ‘transformed open land’, and the impact intensity scale applied to them ...... 82
Table 14 - Biodiversity components and the vulnerability scale used to score them ...... 83
Table 15- AHP relative weightings of threats per participant and consolidated weights ...... 84
Table 16 - Mean scores for intensity of threat applied to classes in each threat layer ...... 85
Table 17 - Matrix of values applied to pixels in each threat layer to represent relative intensity of individual threats...... 97
Table 18 - Detail of sub-criteria classes ranked from highest to lowest by intensity score. Ranking by areal extent is also shown in column 4...... 97
xii List of abbreviations
AHP Analytical Hierarchy Process
AMD Acid mine drainage
ARC ISCW Agricultural Research Council - Institute of Soil Climate and Water
CBA Critical Biodiversity area
COH Cradle of Humankind World Heritage Site
DACE Department of Agriculture Conservation and Environment
DEA Department of Environmental Affairs
DEDECT Department of economic development, environment, conservation, tourism
EIA Environmental impact assessment
EMF Environmental Management framework
ESA Ecological support area
GCRO Gauteng City Region Observatory
GDACE Gauteng Department of Agriculture , Conservation and Environment
GDARD Gauteng department of Agriculture and Rural development
GI Geometrical Interval
HDF Hierarchical Data Format
IAP Invasive Alien Plant
ICC International Co-ordinating Council
IDP Integrated development plan
LCC Land cover classification
MAB Man and Biosphere
MBIG Magaliesberg Biosphere Initiative Group
MBR Magaliesberg Biosphere Reserve
MCA Multi criteria analysis
MMU Minimum mapping unit
MODIS Moderate Resolution Imaging Spectrometer
MPE Magaliesberg Protected Environment
NAIPS National Alien Invasive Plant Survey
NBA National Biodiversity Assessment
xiii
NEMBA National Environmental Management: Biodiversity Act
NEMPAA National Environmental Management: Protected Areas Act
NEMA National Environmental Management Act
NGO Non-Governmental Organisation
NKB Norite koppie bushveld
NPAES National Protected Area Expansion Strategy
NSBA National Spatial Biodiversity Assessment
PA Protected area
PBT Priority Biodiversity Threats
PCV Priority Conservation Value
PGM Platinum group metals
ROI Region of Interest
SANBI South African National Biodiversity Institute
SCP Systematic conservation planning
SDF Spatial development framework
SKEP Succulent Karoo Ecosystem Plan
SPLUMA Spatial planning and land use management Act
SPOT Satellite Pour l’Observation de la Terre
UNEP United Nations Environment Programme
UNESCO United Nations Educational, Scientific and Cultural Organisation
VT Vegetation types
xiv Chapter 1 Introduction
1.1 Introduction
Biodiversity is an umbrella term that is multifaceted. It encompasses the “global variety of species, the genes they contain and the ecosystems that support them” (Ferrier, 2002), which includes aspects related to the structure, functioning and integrity of habitat (Liu & Taylor, 2002). Findings of the Millennium Ecosystem Assessment, (2005) for biodiversity loss, report declines in population sizes and ranges across several taxonomic groups, as well as losses in genetic diversity and a trend towards a more homogenous distribution of species globally. Based on ICUN World Conservation Union criteria, the assessment cites between 10% and 50% of the well-studied higher taxonomic groups of species as currently threatened with extinction. Global patterns of biodiversity loss are consistent in South Africa which is regarded as a ‘mega diverse’ country with exceptional biodiversity assets and ecological infrastructure (Driver, et al., 2012). Yet 40% of South African terrestrial ecosystems are threatened and red-list species assessments show threatened status for one in five mammal species, one in seven frogs and bird species, one in eight plant species and one in twelve reptile and butterfly species (Driver, et al., 2012).
Anthropogenic processes that may spur change in the structure, functioning or composition of biodiversity or ecological processes are considered threats to biodiversity, or drivers of biodiversity loss. Indirect drivers, otherwise referred to as ultimate threats are global scale phenomenon, such as population growth rates, markets and trade, governments and legal frameworks, and customs and behaviours of society. Ultimate threats lead to changes at regional or local scales, which are referred to as direct drivers or proximate threats that generally result in some form of habitat transformation, whether it be outright loss of habitat (i.e. land clearing for infrastructure or agriculture) or a range of modifications of habitat that result in habitat degradation (e.g. soil erosion, damming rivers, invasive spread) (Lambin, et al., 2001; Pressey, et al., 2007; Millenium Ecosystem Assessment, 2005).
One mechanism used to manage population and development pressures globally, is for states, or private initiatives, to set aside demarcated areas for the conservation of remaining biodiversity (Margules & Pressey, 2000). Protected areas and conservation networks provide a mechanism to preserve or at least maintain, intact landscapes, ecosystems and biotic communities, and the resources and services associated with them (Liu & Taylor, 2002).
Child (2004) notes that many developing countries, with high population growth rates have conservation areas with considerable biodiversity, surrounded by dynamic land-uses and demography’s. In addition, recent findings show accelerated population increases at the edges of protected areas, the world over (Southworth, et al., 2006; Moilanen, et al., 2009). This phenomenon suggests that close proximity to protected areas may offer livelihood benefits to marginalised communities, as well as high value real estate and other economic opportunities, such as recreation
1 and tourism development, however not without associated complex problems (Wittmayer & Büscher, 2010) .
The United Nations Educational, Scientific and Cultural Organisation (UNESCO) Man and Biosphere programme (MAB) aims to address some of the challenges faced where populations and protected areas meet, through a world network of biosphere reserves. The reserves are established to deal with the difficulty of how to reconcile the conservation of biodiversity and maintain associated cultural values, while pursuing economic and social development (UNESCO, 1996). Hence, the reserve zonation process and management thereafter is based on tiered conservation and development objectives for three distinct zones (core zone, buffer zone and transition zone), with a view to managing them in a sustainable way to protect and preserve remaining biodiversity while improving the livelihoods of surrounding communities.
According to the Millenium Ecosystem Assessment, (2005), biodiversity losses would be greater presently if it were not for governments, NGO’s, communities and business who take action to conserve biodiversity and promote its sustainable use, through responses such as protected areas and biosphere reserves. Furthermore, the assessment suggests that responses to bioversity loss would have a better chance of sucess if biodiversity conservation and development planning were integrated, especially if conservation responses were reflected in national development policy, or poverty reduction strategies of developing countries.
In South Africa though, despite progressive and robust environmental and Protected Areas legislation (NEMA, 1998; NEMPAA, 2003), biodiversity conservation is still a low priority in many government sectors (Carruthers, 2002; Turpie, 2003), as the social and economic benefits of biodiversity conservation are not always recognised with government primarily focussed on developing the economy and social equity. While development planning is now integrated at municipal, regional and provincial levels, conservation plans have evolved quite separately from development planning mostly at national and provincial level (Turpie & Kleynhans, 2010). These circumstances are challenging for conservation.
1.2 Research Motivation
Considering the requirement that biosphere reserves manage conservation and sustainable development in areas of unique biodiversity value with a “gradation of human interventions” (UNESCO, 1996, p. 16), it is not appropriate to exclude transformed, disturbed or even degraded systems from biosphere borders. Instead the tiered zonation should incorporate these areas so they contribute appropriately towards biosphere goals of conservation, sustainable development and social upliftment (UNESCO, 1996). In this respect the Magaliesberg region is a good candidate for biosphere reserve status because of its extraordinary biodiversity and cultural heritage in close proximity to a large populous and the economic hub of South Africa. However, the region is thus pervaded by dynamic and increasing pressures of population, development and extraction which threaten its ecological integrity and biodiversity. In order to achieve the aforementioned biosphere
2 goals a comprehensive conservation assessment of the region would need to consider these pressures.
Some of the proximal threatening processes that directly impact the terrestrial Magaliesberg environment are habitat fragmentation and degradation, alien species invasions, pollution, population pressure manifested as urban sprawl, over exploitation, deep pit and opencast extraction, injudicious fire regimes, bush encroachment, erosion and commercial cultivation and agriculture. Some of these processes can be expressed spatially, and referenced to the scale, patterns and ecological processes of a landscape, or to priority sites for biodiversity, as a form of biodiversity vulnerability or threat analysis using spatial surrogates, like classified land cover to indicate these threats.
Threats to biodiversity are increasingly being incorporated into conservation planning (Wilson, et al., 2009; Fuller, et al., 2010; Pressey, et al., 2007; Wilson, et al., 2005 b). Spatial conservation planning products (and threat cost-layers that may be associated with them) that cover the Magaliesberg region have been undertaken at the provincial, municipal and local level. Spatial data on individual threats to biodiversity have not been collated for the Magaliesberg region before. Thus the information generated in this research may be used to inform conservation and development planning for the region. Considering the proposed reserves proximity to the economic hub of South Africa – the Gauteng City region – dynamic and increasing pressures of population, development and extraction need to be incorporated into conservation assessment at a regional/ landscape scale.
It is envisaged that a spatial prioritization of proximal threats to terrestrial biodiversity would indicate their spatial variability and the constraints to conservation in respect of the most important sites for biodiversity conservation. The latter has been recently compiled by provincial conservation agencies and together with the biosphere zonation and threatened ecosystem status data, would provide a spatial indication of important biodiversity value and conservation opportunity.
Two methods are increasingly being used to delineate appropriate sites for a wide variety of objectives in the fields of conservation and land-use planning, including threats and vulnerability assessments – Systematic Conservation Prioritization (SCP) (Moilanen, et al., 2009)and Spatial Multiple Criteria Analysis (MCA) (Malczewski, 2010). (Although the latter is also an SCP technique, it has wider applications).
The spatial MCA approach is reported to be adaptable, transparent, relatively easy to implement and can be participatory (Li & Nigh, 2011; Ferrier & Wintle, 2009). It has been applied extensively as a management and planning decision support tool that is able to consider multiple attributes in many conservation-based and land-use applications (Stranger, 2004). Area prioritization specifically, has been applied extensively for the sighting of facilities such as wind-turbines and landfills (Vasiljevic´, et al., 2012), for many aspects of biodiversity conservation such as probability maps of species occurrence (Stranger, 2004; Abbitt, et al., 2000), conservation area networks and zonation for
3 conservation (Geneletti & van Duren, 2008) restoration and reforestation efforts (Orsi & Geneletti, 2010; Zerger, et al., 2011) and threats to biodiversity (Fuller, et al., 2010; Veech, 2003; Vimal, et al., 2012).
Another aspect to consider in SCP is the dynamic nature of landscape structure and function and integrity, influenced by both biophysical and anthropogenic processes (Liu & Taylor, 2002), that may spur change in a landscape. These changes can also be an indication of stress that could affect the ranking of priority sites. In addition, the conservation agenda is ensconced in the economic and political real world dynamic, and conservation outcomes will be influenced by these factors. As a result non-ecological variables such as cost-efficiency, adequacy, threat and vulnerability are receiving increasing attention in spatial conservation planning. These factors can inform conservation targets, the pattern of reserve networks, priority areas and the scheduling of conservation actions (Wilson, et al., 2009).
Information on processes that threaten biodiversity can be applied to assess the relative vulnerability of planning units to these threats (Wilson, et al., 2009). A systematic assessment would consider all the threats affecting an area and their exposure and severity may be estimated by experts. Saaty’s AHP method could be used to formalize expert insight into a quantitative record by weighting the relative intensity of threatening processes of the MBR (Saaty, 1977; O'Connor & Kuyler, 2009). Multiple threat scores could then be combined using MCA techniques, into a composite of threat priorities. If these were to be referenced to planning units, the prioritisation would identify those exposed to the greatest threat, and those that may also contain critical biodiversity.
Typically, spatial conservation assessments use a binary approach with threat cost layers, whereby transformed areas are considered lost to conservation, and are excluded from assessment, particularly if the threat cannot be abated or is too costly to mitigate (Wilson, et al., 2009). Considering that biodiversity includes all ecosystems - managed systems, such as plantations, farms, croplands, pasture and urban open spaces have their own biodiversity (Millenium Ecosystem Assessment, 2005) , and in the context of the Biosphere philosophy, it is precisely these sites that can offer sustainability opportunities for surrounding landowners, businesses, or marginalized communities, with the support of the Biosphere network. It is assumed that in this analysis the transition zone will be a case in point, whereby vulnerable sites are identified, and targeted for sustainable management or monitoring.
The scope of this research therefore includes the quantification of the geographic coverage of multiple threatening processes and the quantification of their relative severity. This is akin to measuring their exposure and intensity - two dimensions of vulnerability identified by Wilson, et al., (2005) that can be assessed spatially. Priority threats could then be integrated with conservation priorities to identify those conservation priorities within the proposed MBR are most vulnerable to these threats. It is hoped that this will establish a baseline from which future threats and vulnerabilities can be measured, mitigated and managed going forward.
4 1.3 Problem Statement and Hypothesis
The proposed Magaliesberg Biosphere borders on the economic hub of South Africa, which increases the population and development pressure on the region. Anthropogenically induced threatening processes, can be seen as variables that lead to land-use-change, habitat loss and habitat degradation. They can stress the ecological integrity of the landscape, and present a challenge for terrestrial biodiversity conservation.
This work is a conservation planning exercise to prioritize sites vulnerable to the threatening processes of land transformation and degradation that impact on terrestrial biodiversity. The intention is to identify current threats that can be represented spatially and rank their severity in relation to one another, by following an AHP MCA approach, and then spatially reference them to the important biodiversity conservation areas in the Magaliesberg Biosphere Reserve (MBR). The methodology design combines techniques and procedures from a number of applications and research in the spatial prioritization, site selection and conservation planning arena over the past two decades.
Addressing the underlying causes of biodiversity loss, i.e. local threatening processes in an area, is necessary for successful implementation of rehabilitation measures and mitigation of further losses. It is hypothesised that identifying vulnerable planning units that contain important biodiversity that are also exposed to a gradation of threats, can inform regional conservation and development decisions to promote sustainable practices in the spirit of a Biosphere Reserve.
It is hoped that the priority conservation value derived from secondary data overlay and the priority biodiversity threats derived by AHP and spatial MCA of secondary data will accurately reflect the current status-quo for the MBR.
1.4 Research Aims and Objectives
The primary aim is a spatial prioritization of threats to the terrestrial biodiversity of the MBR. Then, priority threats are spatially referenced to priority conservation value of the area, as spatial decision support for conservation and sustainable development.
The intention is to use AHP weighted threat criteria (spatial layers of land-use, land-cover and other threatening processes particular to the region) as variables in a GIS multiple criteria decision analysis (MCA), to determine the spatial variability and geographic extent of threats and quantify their relative intensity as stressors which influence the vulnerability of terrestrial biodiversity of the MBR.
The objectives for this aim being:
5 1) The selection and development of criteria (spatial layers) as indicators of biodiversity importance, to reflect spatially, the priority conservation value (PCV) in the MBR. 2) The selection and development of criteria (spatial layers) as indicators of current threatening processes that affect the terrestrial biodiversity of the MBR. 3) The relative weighting of these threat criteria using the Analytic Hierarchical Process (AHP) approach. 4) To input the weighted threat criteria in a spatial multi criteria analysis (MCA) to reflect spatially, priority terrestrial biodiversity threats (PBT) in the MBR. 5) The development of planning units on which to base the prioritization and analysis. 6) A synthesis of the prioritizations where Priority Biodiversity Threats are related to Priority Conservation Value, to expose vulnerable planning units.
A further outcome of this research is a spatial plan for the MBR, by systematic compilation of available biodiversity and threats spatial data from various sources at different spatial scales and timeframes, considering the MBR is situated in two provinces and extends over parts of six municipalities. This will provide a current broad perspective on the regional landscape and a convenient spatial benchmark to inform conservation and sustainable management decisions going forward.
1.5 Report Outline
Chapter one is an overview that situates this research problem in the framework of spatial decision support and threats assessment for conservation planning, and motivates the research design in this context. The hypothesis and objectives of this research are then stated. Following this, Chapter Two provides more detail of the Magaliesberg as a region and the Biosphere context, expanding on the biodiversity resources of the study area, and the current human-environment situation, and the biosphere nomination process. Finally the chapter touches on conservation action and sustainable development opportunities, in relation to Magaliesberg Biosphere objectives.
Chapter Three begins with a literature review on current themes relevant to this study - land transformation rates and relevant threatening processes that affect the regions biodiversity, and conservation planning, spatial multi criteria analysis, the AHP approach and threats and vulnerability assessment. Details are provided on spatial indicators to represent threat, and the types of research problems, methodologies and decision support outcomes that are generated from these indicators. The chapter ends with a look at how spatial conservation products may be applied in land-use planning.
6 Chapter Four describes the methods used to realise the objectives of the research – to generate spatial data on important biodiversity and relevant threat criteria, and the AHP method used to weight threat criteria relatively, and combine them into a priority threat composite. This is followed by the process used to integrate priorities for conservation and threat, into a single priority product. The chapter provides extensive details on processing, challenges and limitations of the methodology and spatial data applied.
Chapter Five discusses results in relation to the research objectives, and in the context of other threats and vulnerability analysis. Finally scenarios for practical management are demonstrated using results.
Chapter Six concludes the study with an overview of research design and study context, followed by a review of work achieved, shortcomings and recommendations and finally a conclusion of the research results.
7 Chapter 2 The proposed Magaliesberg Biosphere Reserve in context
2.1 Biophysical background
Comprehensive and detailed information about the Magaliesberg region and the proposed Biosphere is available in the book entitled The Magaliesberg (Carruthers, 2015), and the Magaliesberg Biosphere Nomination (2012), compiled for the nomination application (Department of Environmental Affairs, 2012). The section below is a description of the study area, with a focus on its habitats, biodiversity and conservation status, as well as background to Biosphere Reserves, the opportunities they present for conservation and sustainable development, and the status of the Magaliesberg Biosphere nomination.
2.1.1 Location
The proposed MBR is 3758,1 km² and extends approximately 120 km west to east across the boundary of the North West and Gauteng Provinces of South Africa, between the cities of Rustenburg in the West and Pretoria in the east. The north-south extent is approximately 40km. The urban edge of north-west Johannesburg roughly coincides with its south eastern extent, and the agricultural hub of Brits is to the north (see Fig 1). The area lies in the temperate latitudes with decimal co-ordinate limits 25.60˚N; 26.10˚S and longitude 27.00˚W and 28.30˚E. An elevation of 1320m above sea level marks the lowest point of the Biosphere, and the highest elevation is 1852m with the highest mountains being 532m from the valley bottom (Carruthers, 2015). The Cradle of Humankind World Heritage Site (COH) and the Magaliesberg Protected Environment (MPE) are located within the proposed biosphere borders. An analysis extent of 4355, 9 km², includes a 2km buffer around the entire MBR.
8
Figure 1: Location of the proposed Magaliesberg Biosphere Reserve, South Africa.
2.1.2 Geology and Topography
About 2000 million years ago the mountain ranges of the Magaliesberg region were formed when massive intrusions of molten magma steadily formed a 65000 km² expanse of solid igneous rock over and between existing sedimentary layers. Under the weight of the magma, the sedimentary layers (made up of a base-layer of Black Reef, overlaid by semi-soluble dolomite and chert which was later topped by alternate layers of quartzite and shale) tilted upwards through the magma at the edges of the igneous intrusion to form a series of semi-concentric mountain ranges of tilted sedimentary layers at the southern perimeter of the igneous mass. Over millions of years of weathering and erosion, resistant quartzite and shale formed the ridges of the Magaliesberg, Witwatersberg and other smaller chains (fig 2) (Carruthers, 2015).
The resistant quartzite formed more gently tilted north-facing slopes, and the deep kloofs engraved in them, are a result of of the erosion of sills and dykes from the igneous intrusion. Geological faulting formed the few poorts that dissect the otherwise continuous ridges, as well as some of the kloofs on the mountain plateau and slopes (Carruthers, 2015). To the north of the Magaliesberg, the periphery of the igneous intrusion, known as the bushveld complex, is made up of mostly norite in this area. It forms flat plains with occasional rocky koppies. Directly south of the Witwatersberg, remains of the Transvaal Sequence form a wide curved band of porous dolomite and chert,
9 characterised by undulating hills dotted with cave formations and small rocky outcrops (Berger, 2006). The dolomite is bordered to the south by the northern edge of the granite dome on which Johannesburg lies.
This unique and varied geology and topography provides a multitude of habitats and ecological niches and micro-climates that support a vast array of biodiversity across the Magaliesberg landscape (Department of Environmental Affairs, 2012; Carruthers, 2015).
2.1.3 Hydrology
The hydrology of the MBR is made up of the following hydrological features : Dams, rivers, streams, waterfalls, springs, small wetlands and groundwater recharge zones. The main tertiary catchment of the region is the Crocodile West and Marico catchment, which drains from the Witwatersrand watershed. The study area is located high up in the catchment, which is urban and agricultural land- use at this point. As a result the nine main rivers of the region carry heavy loads of sediments, litter and nutrients, and all of the large river ecosystems in the Biosphere have been classified as critically endangered or endangered, according to the National Biodiversity Assessment (NBA) 2011, river ecosystem status (Driver, et al., 2012; Biodiversity GIS, 2007), meaning that they have lost all or most of their value as terrestrial botanical habitat (Department of Environmental Affairs, 2012; Gill & Engelbrecht, 2012).
Most of the main rivers flow northwards and converge at one of the few poorts in the Magaliesberg range. Two of these poorts, the Hartebeespoort and Olifantspoort, were dammed in the 1920’s and 1930’s respectively. Hartebeespoort dam is 20km², and the Olifantsnek dam 2.5km² in extent. A fair percentage of both of these dams are silted up, and agricultural and urban runoff present dire problems for water quality (Oberholster & Ashton, 2008).The Buffelspoort dam is less silted and less polluted as it drains the perennial streams and runoff from the mountain (Carruthers, 2015). Mountain streams provide high quality water and are one of the area’s most important natural resources and ecosystem services (Gill & Engelbrecht, 2012).
The topography of an area largely determines where wetlands can potentially form. Generally throughout the Biosphere small wetlands occur in low lying areas along the Moot valley, and along river and stream courses. A series of expansive seasonal grassland wetlands or pans extends southwards from the south-west biosphere buffer area. A wetland study conducted as part of the Magaliesberg Protected Environment, Environmental Management Framework (MPE, EMF) ( DACE, 2007), identified six different types of wetlands associated with the mountains. The largest being a natural wetland occurring high up in the Rustenburg Nature reserve, in a plateau basin of deep alluvial soil between two quartzite ridges. Other smaller mountain wetlands occur where faults cut across drainage lines on the northern slopes. Patches of a specialist vegetative sponge found at these high elevations, absorbs and retains rainwater and runoff (Gill & Engelbrecht, 2012). Seasonal
10 wetlands, pans and wetlands associated with the kloofs and springs have also been identified ( DACE, 2007).
Wetlands are important habitat for many species and provide a filter system for runoff. By retaining and slowly releasing water, they control the rate of runoff, which helps prevent flooding and soil erosion (Jaganath, 2009; Driver, et al., 2012). Areas of dolomite in the south and southwest are noted as important ground-water recharge areas (DACE, 2009)
Although this is a study of threats to terrestrial biodiversity, the hydrological resources of the study area as outlined above, provide examples of ecological functioning and service provision associated with the Magaliesberg Mountains, which forms part of the “ecosystem component” of biodiversity referred to in Section 4.4, that would be associated with valuable biodiversity. On the other hand, the hydrological alteration that has occurred can be considered an historical threat to the biodiversity of low-lying ecosystems (Rebelo, et al., 2011).
2.1.4 Macro Climate
The climate related to the Magaliesberg Biosphere can be considered at the macro and micro level. The macro climate is a function of geographical location, while the micro climate is influenced by local topography, relief and aspect. Here the macro climate is discussed, while the micro climate is referred to in the following section.
The Magaliesberg mountain region is recognised as a climactic transitional area between two bioclimatic zones - a warm temperate climate with dry winters in grassland, and a hot, arid steppe climate, with dry winters in savanna. Temperatures gradually warm to the north with Krugersdorp (just outside of the southern boundary) experiencing an average daily temperature range of ± 14,5˚C to 25,5˚C in January, and 3,5 to 16˚C in July. Pretoria, to the North, records an average daily range of 17.9˚C to 28.6˚C in January and 4.8˚C to 19.7˚C in July. Winters are dry and generally frost-free in the north, while occasional frost occurs in the south, especially in low-lying areas and on the open plains of the south west that are exposed to frontal winds (Department of Environmental Affairs, 2012; SAWS, 1995-2010)
Mean annual precipitation (MAP) in this summer rainfall region (October to April) is generally higher in the southern and eastern parts at an average of 728mm, grading down to an average of 520mm in the north and west. Extreme MAP cycles can see these averages drop to 350mm and rise to 1200mm (Department of Environmental Affairs, 2012; SAWS, 1995-2010).Thunderstorms occur during summer, but occasionally frontal weather systems bring soft soaking rains that can last a few days. Occasional large scale flooding events tend to affect the cultivated floodplains north of the Magaliesberg, in the mid to lower catchment (Department of Environmental Affairs, 2012).
These subtle climactic differences echo the boundary of two distinct South African Biomes – grassland and savanna. The savanna biome is climatically similar to grassland, but with higher 11 minimum temperatures, and savanna generally has a higher Mean Annual Potential Evaporation (MAPE) than grassland (Mucina & Rutherford, 2010 ). In the study area MAPE varies from about 1600mm in the grasslands of the south and increases to around 2000mm in the savannas at the countries northern border (Department of Environmental Affairs, 2012).
The climatic transition and biome interface means that the MBR is represented by species that occur in both biomes that may be more tolerant of variations in climate, as well as species from either the grassland or savanna biome that have adapted to the specific climatic conditions of either. This could indicate that some species found here occur at the limit of their ranges (Carruthers, 2015). The near-natural state of the landscape with large unfragmented patches, variations in altitude and south-facing slopes and kloofs as cool, moist refuge sites, makes this an important region for climate change resilience and adaptability (Driver, et al., 2012).
Figure 2: The Magaliesberg, looking west over the Hartebeespoort dam and Moot valley beyond.
2.1.5 Mountain Habitats (Micro Climate and Ecology)
Few species are restricted to one habitat type only, but each habitat -type is a unique climatic and ecological niche that supports different species assemblages. Pfab in (GDARD, 2011) explains that areas of relief (in Gauteng) are so diverse in topography (slope, aspect, altitude and micro-climate) that they provide unique and varied conditions to spur evolutionary processes (factories to generate biodiversity), and as such are important for climate change adaptation (GDARD, 2011).
Five broad habitat types associated with the mountain environment are described in Gill & Engelbrecht, (2012): northern slopes, kloofs and streams, crests, cliffs, southern slopes and valleys. A brief description of these habitat types sets the scene for the biodiversity of the region. 12 The east – west orientation of the mountain chains means that radiation is a significant micro climactic factor on the north and south facing slopes. The north gradient, around 20 to 30˚ in the Magaliesberg captures maximum radiation with little retained moisture, which leaves the exposed slopes hot and dry (Carruthers, 2015). Succulent and xerophytic plants are suited to shallow gravelly quartzite soils on higher slopes while trees and shrubs occur on deeper soils down slope (Gill & Engelbrecht, 2012). Deeper soils and runoff from the upper slopes allows for the establishment of trees and shrubs down slope. More than other mountain habitats, the mountain slopes (of the Magaliesberg, Witwatersberg and other chains) are vulnerable to urban and mining development, which forms part of the scope of this study.
The kloofs, have a moderate climate year round, being shielded from direct sunlight and extreme cold weather and winds by their north orientation, sheer-walls and overhanging tree canopy. Most mountain runoff is filtered to these kloofs, contributing to a constant supply of moisture to support forest communities, with some sensitive species only occurring in these kloofs (Gill & Engelbrecht, 2012).
The mountain crests and summit, are generally exposed, dry and windswept. The exposed summit is also prone to frost in winters and is susceptible to lighting strikes. As a result of these factors plant life is generally sparse and stunted, and animal life is limited. The plateau in the Rustenburg nature reserve is an exception, with alluvial soils and wetland features in the basin between a double crest, supporting more biodiversity (Carruthers, 2015).The updrafts of prevailing south winds provide thermals for a number of soaring birds, such as the Cape Vulture, a highlighted threatened species of the region (Whittington-Jones, et al., 2014).
Steep south facing cliffs up to 100m wide, extend almost the entire length of the Magaliesberg. As they are mostly in shade, they remain cool in summer, and reach below freezing temperatures in winter. Moisture is retained in shallow soil pockets on ledges which provide habitat for hardy plants and trees whose roots are secured in deep fissures in the rock face, as well as some birds and animal species (Carruthers, 2015).
Like the cliffs, the 40˚ gradient southern slopes are cool with low evaporation, so moisture is retained. A closed canopy (scarp forest) of mixed bushveld trees is supported by moist shale derived soils, below the line of cliffs. Interspersed with the scarp forest, temporary watercourses drain mountain runoff around and between buttresses of thick grass cover. Carruthers, (2015) mentions frequent burning for green-bite grazing on the south slopes just west of Pretoria has resulted in a reduced density of tree canopy and that grassed buttresses may become susceptible to sheet runoff from overgrazing.
The Moot Valley forms between the southern slopes and the northern slopes of the Witwatersberg range (fig 2). Thick alluvial soils and perennial rivers support mixed woodlands and riverine
13 vegetation. Cultivation, urbanization and alien plant invasions continue to be the main stressors to this habitat.
2.1.6 Regional Habitats (Biomes, Bioregions and Vegetation-Types)
Other habitat types or ecosystems, not necessarily associated with the mountains, are defined on a coarser scale than the mountain habitats described above. These habitats can be associated with plant assemblages which are nested within broader biome and bioregion ecological units. These plant assemblages have been classified as vegetation types (VT) by Mucina and Rutherford (2010) of which there are 435 in South Africa and 15 that occur in the MBR. Vegetation assemblages are commonly used as spatial surrogates for all biodiversity in the absence of fine (point scale) species data (Ferrier, 2002), and has been included as one of the spatial layers to represent biodiversity conservation value in this study. The layer is shown here for ease of reference (fig.3).
Figure 3: Layer 4 - Vegetation Types of the MBR
The fifteen vegetation types occurring in the study area are briefly discussed with reference to the bioregion to which they belong, their extent and current conservation status.
14 Savanna Biome
The Savanna biome covers the Central Bushveld Bioregion, to which seven of the vegetation-types of the Biosphere belong. The bioregion encompasses the plains to the north of the Magaliesberg (Marikana Thornveld and Norite Koppies Bushveld VT), the Moot valley (Moot Plains Bushveld VT) opening out to the western edge of the biosphere (Zeerust Thornveld VT) and on the mountains and ridges across the whole study area (Gold Reef Mountain Bushveld, Gauteng Shale Mountain Bushveld and Andesite Mountain Bushveld VT). These three mountain bushveld vegetation–types cover the mountain habitats described in the previous section, except for the summit habitat, which is a grassland vegetation-type.
Typically this central bushveld bioregion is represented by woody vegetation and a grass dominated herbaceous layer. Depending on local conditions, trees form semi-open to closed thickets or woodlands, and can range from short deciduous bush cover to a medium-tall +5m tree cover of deciduous and evergreen trees. Some vegetation types are dominated by thorny species (Acacia sp). The conservation status and extent of each vegetation type, relative to the biosphere can be found in (Table .1).
Grassland Biome
The grassland bioregion that predominates in the Biosphere is Mesic Highveld Grassland. Characterised by a MAP above 650mm and frost, a thick cover of sourveld grass species dominate in the summer, followed by a dormant winter period. The high diversity of forbs found in grasslands, is what makes grasslands an important biome for species richness (Mucina & Rutherford, 2010 ). Rocky habitats show a diversity of woody species, which occur in the form of scattered shrub groups or solitary small trees. Riverine habitats contain frost tolerant and deciduous woody species.
Five grassland types that are distinguished by geology, soils, elevation topography and rainfall, occur in the biosphere (Carletonville Dolomite Grassland, Egoli Granite Grassland, Rand Highveld Grassland, Waterberg Magaliesberg Summit Sourveld, Soweto Grassland VT) . Notably, Waterberg Magaliesberg Summit Sourveld has a unique set of environmental conditions that support biogeographically important taxa and some endemic floral species (Mucina & Rutherford, 2010 ; Gill & Engelbrecht, 2012). Also, the study area east of the Cradle of Humankind (COH) contains almost all the remaining Egoli Granite Grassland outside of urban Johannesburg. This vegetation type is highly vulnerable to urban expansion (Driver, et al., 2012). Within the biosphere the Diepsloot nature reserve is a provincial nature reserve of this grassland type that is noted as being degraded and dominated by secondary growth. Interestingly though, mixed opinion on the species composition of this grassland type has found that it may be species poor, and typically resembles old fields or secondary grassland in its primary state (Mucina & Rutherford, 2010 ). Refer to Table 1. for the extent and conservation status of Grasslands.
Forest Biome 15 The Northern Afro temperate Forest vegetation type forms part of the Zonal and Intrazonal Forests Bioregion. It is a grouping of inland isolated forest patches of the northern Highveld occurring in low- escarpment kloofs and ridges, usually at an attitude of 1450-1900m above sea level. The Magaliesberg forests mark the western most occurrence of this vegetation type (Mucina & Rutherford, 2010 ).
In the context of the Biosphere these forests are found in the kloofs and valleys along the northern slopes of the Magaliesberg, as described in the previous section. Vegetation is of afromontane origin, and is relatively species poor (Mucina & Rutherford, 2010 ). However, these forests support some biodiversity that does not occur in savanna and grassland systems.
Azonal Vegetation
The Highveld Salt Pans form part of the Inland Saline Vegetation Bioregion. Salt pans are typically found in depressions in plateau landscapes at the grassland-savanna interface in semi-arid areas. One salt pan occurs in the south western corner of the biosphere, making it one of two least dominant vegetation types in the study area. Salt pans change from being freshwater systems in the rainy season, to progressively more saline as evaporation takes place. Biogeographically, it features endemic species and important water bird habitat. Sparse grassland and dwarf shrubland may develop on pan edges, especially under heavy grazing pressure. Wind erosion can be severe, with large winter dust plumes occurring (Mucina & Rutherford, 2010 ).
A significant expanse of Eastern Temperate Freshwater wetland extends southwards from the 2km Biosphere buffer to the south west. Consisting of temporary and stagnant fresh water, these wetlands support aquatic and hygrophilous vegetation in temporarily saturated soils. Endemic highveld herbs are found in this vegetation-type. Threats in the regional context for pans and wetlands are degradation by intensive grazing and alien invasion, as well as wetland draining for cultivation (Mucina & Rutherford, 2010 ).
16 Table 1 – The Conservation Status and Extent of Vegetation-types in the Magaliesberg Biosphere.
(Table compiled using Mucina & Rutherford, 2010 and Driver, et al., 2012)
% state Proportion Conservation Protection % other National MBR extent MBR extent Vegetation Type conserved represented target Level conserved Extent Km² Km² % in MBR ★
Poorly 19% 3.6% - 4128.19 76.10 1.8% 1.58% Zeerust Thornveld Protected
Not Marikana 19% 0.7% 1.5% 2528.70 706.23 27.9% 6.24% Thornveld Protected Not Norite Koppies 24% 0% 3.9% 260.09 10.29 4% 0.03% Bushveld Protected Moderatel 13% Moot Plains 19% - 2900.82 1439.46 49.6% 38.5% Bushveld y (MPE) Protected Gold Reef Well 22.1% 24% 1.2% 2030.98 804.50 39.6% 22.23% Mountain Protected (MPE) Bushveld Gauteng Shale Not 0.4% 24% 1.2% 1025.01 134.35 13.1% 3.2% Mountain Protected (COH) Bushveld Moderatel 6.8% Andesite Mnt. 24% 1.6% 1992.32 156.67 7.9% 4.44% Bushveld y (COH) Protected Carletonville Poorly 1.8% 24% 1.2% 9117.8 511.46 5.6% 12.87% Dolomite Protected (COH) Grassland Not 0.2% 24% - 14513.32 8.01 0.1% 0. 2% Soweto Grassland Protected (KDGR)
Poorly 2.5% Egoli Granite 24% 0.8% 1093.19 324.33 29.7% 7.57% Grassland Protected (DNR) Not Rand Highveld 24% 0.9% - 10261.29 153.07 1.5% 2.78% Grassland Protected
Waterberg Well 27.2% 24% 3.8% 525.87 24.89 4.7% 0.7% Magaliesberg Protected (MPE) Summit Well 28.8% Northern Afro 31% 2.8% 169.74 5.96 3.5% 0.18% temperate Forest Protected (MPE) Not Highveld Salt 24% 0.2% - 1160.87 0.45 < 1% 0.01% Pans Protected
Eastern Poorly Buffer Temperate 24% 4.6% - 556.77 0.20 < 1% Freshwater Protected only wetlands
Abbreviations in brackets refer to protected areas within the Biosphere and 2km buffer- (MPE) Magaliesberg Protected ★
Environment; (COH) Cradle of Humankind; (KDGR) Krugersdorp Game Reserve; (DNR) Diepsloot Nature Reserve;
2.1.7 Species Diversity and Conservation
The heterogeneous geological, topological, climactic, hydrological and soil conditions support a rich variety of fauna and flora, that together constitute the ecosystems present. In addition, the species assemblage also includes some species that occur at the edge of their distribution, in terms of the climatic envelope for species and the location at the grassland/ bushveld ecotone, and embedded forest biome (GDARD, 2011; Mucina & Rutherford, 2010 ). The unique assemblage of floral
17 communities and associated fauna, and the species richness of the Magaliesberg is well recognised, and a biosphere status for the region was conceived due to the large number of species represented in the area, (as opposed to species endemism, of which there are but a few examples) (Carruthers, 2015; DACE, 2007). The area also contains threatened species, which are subjects of further conservation study.
The significant biodiversity of the region has been recorded through scientific investigation for many years. Botanical and zoological collection and hunting expeditions around the Magaliesberg from 1835 onwards into the twentieth century identified and described a number of new species to science. Notably, these include the Sable Antelope and two endemic floras of the mountains (Gill & Engelbrecht, 2012; Carruthers, 2015) . Since then, the Magaliesberg Biosphere Nomination report, 2012, lists a number of biotic research and monitoring outputs, and published articles particular to the region. In addition, there are many floral and faunal publications that cover the Magaliesberg region. Most have a wider geographical range, but some are area specific. A bird checklist lists a total of 434 species occurring in the area (Wesson & Balt, 2014), and Gill and Engelbrecht, (2012), describe over 500 flowering plants in their fieldguide to wild flowers of the Magaliesberg.
Of particular importance for biodiversity is the work done by Carruthers (2015). It provides comprehensive species lists for plant, mammal, bird, reptile, amphibian, and invertebrate taxa, and adds details and anecdotes on a good selection of these. Species assessments done by organisations such as Birdlife South Africa (BLSA), South African Biodiversity Institute (SANBI) and the Animal Demography Unit (ADU) shed light on species occurrence, distributions and their conservation status. BLSA, has identified the Magaliesberg and Witwatersberg areas (more than 62% of the Biosphere area) as Important Bird Areas, meaning that they contain species of global conservation concern and hold large numbers of migratory water birds on a regular basis (Birdlife International, 2013). The ADU has developed a ‘virtual museum’ of biodiversity atlases for major taxa, where occurrence data is collated in quarter degree squares (ADU, 2013). These organisations continue to provide a better understanding of diversity and distribution of species nationally.
Despite these efforts there is still a lack of detailed scientific information on the biodiversity of the region ( DACE, 2007). More investigation is needed, with key research areas being assessments of lesser known species (particularly reptiles and invertebrates), fine-scale biodiversity and ecological surveys of the surrounding landscape to better understand representation and persistence (Margules & Pressey, 2000), and systematic studies to test refugia, migration hypotheses, and climate change effects (Clark, et al., 2011; Contour and Associates, 2013; DACE, 2007)
Biodiversity and/or species assessments may use data on indicator species, or species of special conservation concern, such as IUCN red list threatened species (SANBI, 2010), or Threatened Or Protected Species (TOPS) listed in national legislation (National Environmental Management: Biodiversity Act (NEMBA) (No. 10 of 2004). Threatened species are classified according to their risk of extinction into critically-endangered, endangered or vulnerable categories. Species labelled as
18 near- threatened or of least-concern that are localized endemics or have a restricted area of occupancy are also often assessed for conservation (SANBI, 2010). Endemics particularly, are vulnerable to anthropogenic threats and extinction, and their conservation is the responsibility of the people in the region in which they occur (MacFadyen, 2007). Two provincial conservation plans that used species data as a variable for conservation status (GDARD, 2011; DACE, 2009) were applied in this research to determine Priority Conservation Value (PCV) in the MBR.
Table 2 lists near-threatened and threatened species for some taxonomic groups that occur in the Magaliesberg region. The types of threats responsible for inducing their threatened status, are explored further in Chapter 3.2.
Table 2 - Classified Threatened and Near Threatened Species of the Magaliesberg Region
SPECIES OF CONSERVATION CONCERN IUCN NEMBA MAIN THREATS Plants Aloe peglerae (bergalwyn) (endemic) EN urban expansion , collection for trade, Frithia pulchra (fairy elephants foot) (endemic) RARE Not threatened, Development, overgrazing, invasive Delosperma gautengense VU plants, inappropriate fire management. Urban expansion, erosion alien plant Ceropegia decidua pretoriensis VU invasion. Habenaria barbertoni NT urban expansion Holothrix randii NT urban expansion Mammals habitat destruction, road kills, Atelerix fontalis (Hedgehog) NT PR domestic animal kills, collection Miniopterus schreibersii (Scheiber’s long-fingered bat) NT Roost site disturbance Rhinolophus blasii (horseshoe bat) NT climate change and roost disturbance Myotis tricolor (Temminck's hairy bat) NT climate change and roost disturbance Mellivora capensis (honey badger) NT PR Lutra maculicollis (Spotted-necked otter) NT PR Water extraction from rivers Hyaena brunnea (Brown hyaena) NT PR Panthera pardus ( Leopard) LC VU Reptiles Python sebae natalensis (Southern African python) PR Homoroselaps dorsalis (striped harlequin snake) NT Birds Anthropoides paradiseus (Blue Crane) VU EN habitat loss, poisoning, power lines Gyps coprotheres (Cape Vulture) VU EN Poisoning, power lines Sagittarius serpentarius (Secretary bird) VU NT Power lines,fences Podica senegalensis (African Finfoot) LC VU Habitat loss Circus ranivorus (African Marsh-Harrier) LC VU Habitat loss, degradation Tyto capensis (African Grass-Owl) LC VU Road fatalities, early / unplanned fires Alcedo semitorquata (Half-collared Kingfisher) LC NT Habitat loss, degradation Ciconia nigra (Black Stork) NT VU
19
Butterflies Lepidochrysops rossouwi (roussouws blue) VU Platylesches dolomitica (hilltop hopper) VU Mettisella meninx (marsh sylph) VU
IUCN –Red List, NEMBA – TOPS List : EN-endangered, VU-vulnerable, NT-near threatened ,LC-least concern Source: (SANBI, 2012); (ADU, 2013); (GDARD, 2011); (NEMBA, 2015).
The Cape Vulture (Gyps coprotheres), is the subject of ongoing research and monitoring in the Magaliesberg (Whittington-Jones, et al., 2014) and it has become somewhat of a flagship species for the Magaliesberg. Monitoring of the movements of leopard (Panthera pardus) and brown hyaena (Hyaena brunnea) in the Magaliesberg and COH is a current project (Kuhn, 2014).
It is not only species that are of conservation concern. The South African National Biodiversity Assessment of 2011 (Driver, et al., 2012) identifies threatened ecosystems to inform conservation and development planning. Ecosystem types are classified into critically endangered, endangered, and vulnerable, depending on the proportion of the original extent of each ecosystem type that remains in good ecological condition, and its association with threatened plant species. The threatened ecosystems listed in Table 3, are of significance in the Biosphere study area, and this data is used as a layer to determine PCV in this research.
Table 3 - Remaining extent of Threatened Ecosystems in the Biosphere (Source: NBA, 2011)
Threatened Ecosystem Status Hectares Magaliesberg Pretoria Mountain Bushveld * Critically Endangered 1979 Witwatersberg Pretoria Mountain Bushveld * Critically Endangered 8000 Roodepoort Reef Mountain Bushveld * Critically Endangered 4258 Egoli Granite Grassland Endangered 14470 Witwatersberg Skeerpoort Mountain Bushveld * Endangered 28568 Rand Highveld Grassland Vulnerable 4595 Magaliesberg Hekpoort Mountain Bushveld * Vulnerable 1725 Marikana Thornveld Vulnerable 17100 * These ecosystem types are embedded in some of the classified vegetation types discussed in section 2.1.6.
2.1.8 Anthropogenic footprint and heritage
The Magaliesberg region has had an anthropogenic influence for millennia. The famous early-human fossils preserved in the concrete-like Breccias in cavities in the dolomites of the COH WHS (Berger, 2006) and early stone-age tools and later stone-age etchings found throughout the region, are a testament to the presence of early man and San inhabitants (Carruthers, 2015). Vincent Carruthers’ book The Magaliesberg (part 2) (2015) delves into an interesting historical time-line of socio-political events up to the present. Only the human-environment history that has a bearing on the landscape, and biodiversity of the region is mentioned here.
The beginnings of transformation of the natural state date back to 1600AD with the stone –walled settlements of the Batswana people. They practiced agriculture and metal forging, which came with
20 clearing of grassland for cultivation, woody biomass for fire, building material and other utilitarian goods, and the mining of surface and later, underground ore deposits. By the nineteenth century settlement and trade routes were established, and this time also saw the beginnings of successive invasions/settlement by Pedi, Ndebele and Voortrekker groups. It was also the time of missionary visits and scientific and hunting expeditions, by Voortrekker and European parties, where much of the mega fauna in the region was hunted to extinction (Carruthers, 2015).
Battles were fought throughout the region during the South African war of 1899 – 1902. By this time agriculture was widespread, especially in the Moot valley, and around Brits (notably, tobacco and citrus), but also in the grasslands to the south where wetlands were drained and irrigation canals constructed (Louw, pers.comm., 2010).
The towns of Rustenburg and Pretoria were already established, and the railway between these towns, opened in 1906, serviced the first European exports from the region. The pioneering discoveries of substantial mineral deposits began from 1865, when chrome was discovered north of the Magaliesberg in the bushveld complex, as well as small deposits of other minerals (silver and copper).Gold was discovered in 1875 in the COH around Kromdraai, and limestone for gold processing, was mined in this dolomite region shortly after (Durand, 2010).
About a hundred years ago, the beginnings of political social engineering (apartheid) saw ancestral Tswana and Ndebele territories being declared ‘tribal areas’ in the north while all other land was owned by British and Boer farmers. So called ‘Tribal areas’ later became the homeland of Bophuthatswana, and the source of labour for the platinum mining industry which took off from 1920 (Carruthers, 2002).
The construction of the Hartebeespoort, Olifantsnek and Buffelspoort dams in the 1920’s, changed the hydrology of the region substantially, and a network of irrigation canals increased agricultural output around Brits. Development along the Hartebeespoort dam shore at Schoemansville before the 1960’s was in response to the dam and mountains being used as a weekend recreation and tourist venue for city dwellers (Howard, 1996). As visitor numbers and access to the mountains increased, the first signs of habitat degradation appeared. Erosion, frequent fires, littering and ‘eco- vandalism’ on the mountains was superseded by the environmental stressors of the economic boom of the 1960’s. This time saw the subdivision of farms in favour of business and tourist development rather than farming or conservation. Quartzite and clay quarrying, resulted in massive scarring on the hillsides, and there was regional growth of infrastructure in general (Howard, 1996).
However, even though most of the land was privately owned with the exception of the 4000ha Rustenburg Nature Reserve proclaimed in 1967, many land-owners did opt to conserve their natural areas and large tracts of near-pristine habitat is still evident today (Carruthers, 2002). Eventually, after a concerted press campaign “Save the Magaliesberg”, and the establishment of the Magaliesberg Protection Association, that helped guide legislation, the Magaliesberg was declared a
21 “natural area” in 1978, the first of its kind in South Africa. But, for twelve years there was wrangling over the ownership, management and boundaries of the nature area, until it was finally concluded in 1989, that responsibility for environmental protection rested with the Transvaal province, in communication with a management committee. Post 1994, the Magaliesberg Protected Environment (MPE) is a provincially protected area under the NEMPAA (act 57 of 2003). Protection however is in some cases ineffectual, resulting in illegal and inappropriate developments that go unchallenged by provincial authorities (Carruthers, 2002; Eber, 2005; Fatti, 2013).
The protection of the Magaliesberg is somewhat incongruous with the current character of the area being a growth trajectory for development, because of its central location. There are approximately seven million people living within a 100km radius of the Magaliesberg, but less than 500 000 are estimated to live within the borders of the proposed Biosphere (fig 4) (Eskom, 2008; Lightstone, 2010) .
Figure 4: Population Density estimated from Spot Building count (Eskom, 2008) and census data (Lightstone, 2010).
Thus, the relative remoteness of the area and scenic beauty of the mountains and dam as well as its proximity, has attracted development of country housing estates, tourist venues and associated infrastructure such as the expansion of the Lanseria airport node and the commercial district around the dam. Properties on mountain slopes (with a view), or close to the water, are sought after, which poses a challenge for the conservation of these habitats. The ‘Platinum corridor’ (N4) to the north, connects Pretoria and Rustenburg to Gaborone in the west and Maputo in the east, is earmarked for
22 further economic development in provincial IDP and SDF planning (Department of Environmental Affairs, 2012). Hence important economic development by mining and industry is encouraged, and continues to grow in this area, supported by relatively poor workforces, who live in informal, poorly- serviced settlements. These developments are discussed in more detail in the next section, as threats to the biodiversity of the region.
The current circumstances are challenging for conservation, as the changing land-uses and population growth that is occurring in the Magaliesberg region act as stressors to the ecological integrity of the landscape (Lambin, et al., 2001), including its rich biodiversity and cultural heritage. Carruthers (2015) poses the question of “how to resolve the paradox by which easy access to the Magaliesberg is both its greatest value and its greatest threat” (p 349). This research is an attempt to highlight those areas identified as being of high biodiversity value in relation to areas under threat from anthropogenic stressors. Ultimately, this should contribute a spatial benchmark of current biodiversity value versus threat status, and inform decision making for conservation and development in the future management of the MBR.
2.2 Biosphere Reserves and the Magaliesberg Biosphere Nomination
2.2.1 Background to biosphere reserves
During the early 1970’s the United Nations Educational, Scientific and Cultural Organisation (UNESCO) Man and Biosphere programme (MAB), evolved from its original conservation and research function, into a strategy that would strive for the harmonious integration of remaining natural intact ecosystems with sustainable human activities and development, in the form of an international network of Biosphere Reserves, established in 1976 (Batisse, 2003). Since then, the importance of the multiple functions of Biosphere Reserves (UNESCO, 1996) are realised by the requirement for each reserve to include a formally protected core zone (usually a national park), ‘loosely’ surrounded by defined buffer, and transitional zones, “that should each contribute appropriately to the functions of conservation, sustainable development and scientific understanding”( UNESCO,1996 p.6). Thus, the core protected environment and the buffer and transition zones, should each be utilized appropriately for the resources and ecological services they provide but in a sustainable fashion, so as to protect the core and proximal areas and improve the socio-economic conditions of surrounding communities. In this way biosphere reserves, as opposed to other forms of protected areas can “reconcile the conservation of biodiversity and biological resources with their sustainable use” (UNESCO, 1996, p. 3) by incorporating surrounding land and communities as part of the conservation and development effort.
The main requirements qualifying areas for designation as Biosphere Reserves are that they contain a “mosaic of ecological systems representative of major bio-geographic regions including a gradation of human interventions” (UNESCO, 1996, p. 16). They should be of a suitable size for biodiversity
23 conservation and should “provide opportunities to explore and demonstrate approaches to sustainable development on a regional scale” (UNESCO, 1996, p. 16)
Currently there are 631 UNESCO designated Biosphere Reserves in 119 countries, 14 of which are transboundary sites (UNESCO, 2009-2014). Twenty eight countries in Sub-Saharan Africa are home to 64 Biosphere Reserves, of which South Africa has 6 designated and 3 in various stages of nomination. The Kogelberg Biosphere Reserve was the first to be designated in 1998, followed by Cape West Coast in 2000, Kruger to Canyons and Waterberg in 2001, Cape Winelands in 2007, and Vembe in 2009. These reserves are required to demonstrate the integration of conservation and sustainable development, aided by information sharing and communication between them, and the word-wide network of biosphere reserves (Pool-Stanvliet, 2013).
2.2.2 The beginnings of the proposed MBR
Considering the requirements as stated above, decades of indecisiveness concerning the boundaries and formal protection of the area, and heightened environmental awareness (Carruthers, 2002; Eber, 2005) the Magaliesberg Biosphere Initiative Group (MBIG) realised the Magaliesberg as a region would be an appropriate candidate for Biosphere Reserve status. The proposed biosphere core - the Magaliesberg Protected Environment (MPE), has a unique geology and its varied topography and climate gradients have resulted in a unique mix of biodiversity occurring in the area (Carruthers, 2015). The core would also encompass the Cradle of Humankind World Heritage Site (COH) which has a rich archaeological, paleontological and historical cultural heritage (Carruthers, 2015; Berger, 2006). Aside from this, these two formally protected sites are not isolated, and the patchwork of people and economic activities taking place in their vicinity is steadily increasing, as the region borders on the economic hub of South Africa, providing potential opportunities for sustainable development and biosphere reserve status (Contour and Associates, 2013; Carruthers, 2002).
2.2.3 Current MBR nomination status
After years of preparatory work since 2006 and public participation from 2011, the final nomination documents drawn up on behalf of the North West Department of Economic Development, Environment, Conservation and Tourism (DEDECT) were completed for submission in September 2012. Just before submission the Gauteng Province portion was hurriedly excised from the proposal because the relevant government agencies of Gauteng Province would not endorse the application in time, indicating that the consultation process was inadequate. As a result, only areas within the North-West Province were included in the application as Biosphere Reserve, with a proviso addition “future-expansion-zone” incorporating the Gauteng Province portion, especially to accommodate the COH and private properties in Gauteng that had formally endorsed the concept (Department of Environmental Affairs, 2012). The revised proposal was submitted to national government, and then on to UNESCO’s International Co-ordinating Council (ICC) of the MAB Programme in September
24 2012 (For the purposes of this study this 2012 submission, with future expansion zone in Gauteng, and a 2km buffer around the entire region of interest (ROI) were included for analysis).
At the 25th sitting of the MAB ICC in May 2013, the 2012 application was deferred, in anticipation of two primary concerns being addressed before resubmission. Firstly, to consult with local communities and present a complete biosphere reserve zonation “because the negotiation process to delimit the entire biosphere reserve is still ongoing. Therefore, the zonation is not yet complete and a large piece of buffer and transition zone is lacking in south and southeast, especially around related core area”. Secondly, that the Pelindaba Nuclear facility (conducting civil research in nuclear sciences), located in the biosphere transitional zone, be excluded from the biosphere reserve to keep with a precedent set the previous year, when the proposed Terres de l’Ebre (Spain) biosphere was deferred in a similar case (UNESCO-MAB, 2013).
Following this, the resubmission of 2013 was also deferred at the 26th sitting of the MAB ICC in July 2014. Although it fulfilled the concerns and requirements of the previous sitting, the ICC regarded the zonation pattern of buffer and transition areas unsuitable, and advised “the authorities to resubmit a proposal with an improved zonation pattern with regard to the core areas and buffer zones, in order to fulfil the criteria in the of the World Network of Biosphere Reserves” (UNESCO- MAB , 2014). An extract of the Statutory Framework relating to zonation criteria is shown in figure 5 Under the guidance of the MAB advisory committee, the MBR project steering committee continued to work on endorsements and a continuous buffer surrounding the core zonation pattern, for the 2014 application and submission. In March 2015 the MAB advisory committee recommended the MBR be approved at the next sitting in June 2015 (UNESCO, 2015).
Figure 5 - Extract from Article 4 of the Statutory Framework of the World Network of Biosphere Reserves (UNESCO, 1996). 25 2.2.4 Current MBR conservation status and proposed zonation
The MPE and COH make up the proposed core zone of the Biosphere. Considering the requirement ‘5 (a) in Article 4’, (fig 5), it is important to note that despite the formal protection status of the MPE, as a protected environment promulgated under the National Environmental Management Protected Areas Act (NEMPAA) (No. 57 of 2003) , statutory regulations have yet to be drawn up by the North West Provincial government, and instances of environmental abuse and illegal activities continue (Carruthers, 2002; Eber, 2005; Fatti, 2013) The protection status of the MPE and COH (which is subject to additional world heritage convention laws) does not preclude all development from these areas, and a host of permissions are being sought for developments within these protected environments, which are mostly made up of private properties.
There are a number of small private and public game and nature reserves within the MBR, where the controlling body has a legal mandate to manage the land for conservation objectives. Public protected areas are under the authority of local municipalities and managed by provincial government. Apart from statutorily conserved areas, a number of conservancies, of which there are six in the Biosphere, offer informal protection and stewardship for biodiversity. The conservancies and other private properties that have endorsed the nomination make up the proposed buffer zone. All remaining areas are in the transition zone (or future expansion zone).
In 2008, the National Protected Area Expansion Strategy (NPAES) identified large, intact and unfragmented areas of high importance for biodiversity representation and ecological persistence, suitable for the creation or expansion of large protected areas (NPAES, 2010). Two of these areas fall within the biosphere boundary, the North-West/Gauteng Bushveld and the Vaal Grasslands focus areas. Both areas are in close proximity to existing protected areas and will enhance the regional biodiversity network and corridors.
2.2.5 The spirit of a biosphere - conservation and sustainability opportunities
Since the 1987 call for a global strategy to link economics and the environment through sustainable development, in the World Commission on Environment and Development publication, Our Common Future (WCED, 1987), the term sustainable development has been bandied about and become somewhat of a clichéd phrase. However, despite this, the ways to achieve sustainable development, are widely contested and more challenging to define and implement (Lumley, 2002).
Given the fundamental purpose of biosphere reserves - to integrate man and nature, simply protecting areas to conserve remaining biodiversity is not enough. Additional measures that include socio-economic upliftment by means of low-environmental impact sustainable growth strategies is also appropriate – a harmonizing of ecological services protection and management of biodiversity and sustainable socio-economic development (Kuˇsov'a, et al., 2008).
26 Biosphere zones demarcate areas where appropriate activities are permitted.The core zone is formally protected for conserving biological diversity and monitoring minimally disturbed ecosystems and is limited to use for environmental education, research, and low impact nature based eco- tourism. Buffer zones surround or adjoin the core areas, where activities that are compliant with the Environmental Impact Assessment regulations should occur. Ideally these should be low to medium impact land-uses that maintain biodiversity integrity, such as game and livestock ranching and some tourism (O'Connor & Kuyler, 2009). Although practically, nature based recreation, primary dwellings and new developments mindful of conservation objectives would also be permitted. Flexible transition zones would also include larger tourism developments, cultivated lands, irrigation, orchards, agro-industries, human settlements, support services and infrastructure, mining and industrial development, with an emphisis on cooperative sustainable utilization of the larger transition area to ensure the protection of the natural and heritage resources of the Magaliesberg and enhance this as a benefit to communities (Department of Environmental Affairs, 2012). An example mentioned by Pierce, et al., (2005) is the opportunity for rural migrants living in urban settings adjacent to conservation zones to extend and maintain their indigenous knowledge and biodiversity based traditions through these areas.
In a biosphere context, sustainable tourism is shown to be the most promising factor for local development (Kuˇsov'a, et al., 2008). This can be explained in the way biosphere reserves promote the natural and cultural assets of an area, which potentially attracts international, national and local tourism markets. In a sense this is a commoditisation of natural and cultural capital, which creates advantages and opportunities for sustainable growth (Kuˇsov'a, et al., 2008).
By being given Biosphere status this natural and cultural capital would be protected to varying degrees, and where possible sustainable opportunities are created through restoration and enhancement for conservation, and to improve the sense of place and the tourist experience. Limited resources for often costly restoration efforts can be prioritised through spatial planning methods (Moilanen, et al., 2009), such as invasive alien plant removal (Wannenburgh, 2006), conservation opportunities (biodiversity stewardship) (Knight, et al., 2010) and land reclamation/restoration efforts (Orsi & Geneletti, 2010). Where non-conservation related activities take-place, such as agriculture, mining, and even tourist or residential development, mindful consideration of biodiversity and heritage assets can create socio-economic opportunities, by introducing sustainable practices into industrial and agri-systems and other developments, such as refined production processes, energy and water efficiency, pollution control, appropriate fire and grazing management and green building design and landscaping.
This research can perhaps identify areas most in need of conservation intervention, where important terrestrial biodiversity is most vulnerable to threat and so use spatial conservation planning as a means to aid management and mitigation of these threats, and inform development planning.
27 2.3 Summary
A wealth of bio-physical capital, exceptional in its diversity, is threatened by proximate extreme modifications to the MBR landscape – extensive and rapid urban expansion of country estates, resorts and recreation facilities and expanding mining operations in conjunction with a proliferation of poorly serviced dense informal settlements. The MBR zonation will be an attempt to reduce conflicts in the conservation and development dilemma. There are opportunities for conservation and sustainable development that may help mitigate the vulnerability of biodiversity in each of the biosphere zones, linked to the types of activities taking place in each. Considering the usually sparse resources available to conservation it would be pertinent to prioritize these, taking into account social and economic targets and constraints (Ferrier, 2002; Moilanen, et al., 2009).
28 Chapter 3 Literature Review –Threats to Biodiversity in the
Magaliesberg Region and Spatial Assessment of Threats.
3.1 Introduction
Threats to biodiversity are defined here as human-induced pressures or stressors that impact on ecological, physical and/or evolutionary processes that serve to maintain and generate biodiversity (Rouget, et al., 2003), which ultimately leads to biodiversity loss or deterioration within the ‘bio- spatial’ hierarchy (Soule, 1991). Threats can be classified as ultimate and/or proximate threats (or threatening processes) (Lambin, et al., 2001; Wilson, et al., 2005 b; Pressey, et al., 2007; Soule, 1991). Ultimate threats are indirect factors often with a systemic socio-economic or political origin that are the fundamental cause of biodiversity loss, such as growing human populations, global market forces, unsustainable policies (Soule, 1991; Lambin, et al., 2001) or ineffective application of environmental/conservation legislation (Eber, 2005; Pierce, et al., 2005).
Proximate threats are manifestations of fundamental ultimate threats that affect biodiversity directly at regional or local scales (Pressey, et al., 2007) These direct threats generally result in some form of habitat transformation, whether it be outright loss of habitat (i.e. land clearing for infrastructure or agriculture) or a range of modifications of habitat that result in habitat degradation (e.g. soil erosion, damming rivers, invasive spread).
The previous chapter listed the most obvious pressures per habitat and vegetation type. Usually, data permitting, the spatial extent of proximal pressures, can be mapped and are therefore of interest to spatial conservation planning and are the subject of this dissertation. (Pressey, et al., 2007; Wilson, et al., 2005 b; Noss, 2000). Considering this research is a case study of regional threats, assessed spatially and in relation to specific biodiversity relevant to the MBR, a review is done of relevant threats, in addition to a review of other researcher’s spatial assessments of threats.
This chapter begins with a detailed review of several proximal threats that stress the biological resources of the Magaliesberg region specifically, with a focus on the threats that are assessed spatially in this research. Following the review of threats, the chapter steers towards how threats are analysed spatially, and in relation to spatial surrogates that reflect the regions biodiversity features. The merits and shortcomings of the tools and methods used to do this are reviewed, which include GIS based Multi Criteria Analysis by weighted linear overlay, and several applications that are distinctive in the field of spatial conservation planning (Margules & Pressey, 2000).
3.2 Magaliesberg Regional Threats to Terrestrial Biodiversity
It is well documented in conservation literature that some of the major threats to terrestrial biodiversity are habitat loss and habitat degradation (Erlich, 1988; Driver, et al., 2012; Soule, 1991; Wilcove, et al., 1998). These are broad terms that encompass a range of processes and land-uses
29 that threaten, to varying degrees the composition, structure and functioning of biodiversity at the species and ecosystem level (Goudie & Viles, 1998; Liu & Taylor, 2002).
In South Africa, the 2011 National Biodiversity Assessment compiled by SANBI (Driver, et al., 2012), cites the key pressures on terrestrial ecosystems or drivers of outright habitat loss as cultivation, urban/infrastructure-development, plantation forestry and some forms of mining, while overgrazing, alien plants invasions, inappropriate fire regimes and fragmentation are forms of land and habitat degradation. It also mentions various wastes as pollutants of all media, the impact of terrestrial degradation on aquatic ecosystems and climate change as further stressors to ecosystems. By literature review and expert consensus (see Chapter 4), the threats to terrestrial biodiversity proximal to the MBR study area that are amenable to spatial analysis, because data are available, are: urbanisation and urban sprawl, mining, cultivation, altered fire ecology, fragmentation of habitat, degradation and transformation of open space and the spread of invasive alien woody plants (IAP’s). These specific threats as well as rates of land transformation are reviewed as far as possible in the context of the Magaliesberg region in the following section.
3.2.1 Rates of land transformation relevant to the Magaliesberg region
It is useful to monitor the rates of land transformation in the regional context to gauge current and potential future threats. Mostly this work has been done at a provincial scale, and thus covers a far broader area than the MBR itself. However, figures cited below show general rates of transformation for the North-West (NW) and Gauteng provinces, as well as a localised study measuring rates of change in the Cradle of Humankind World Heritage Site (COH) only. Looking at these rates collectively may provide a perspective for the Magaliesberg, as the region includes a cross section of wilderness areas, productive rural landscapes and urban settings.
Figures for the NW Province reveal that 30% of its extent has been transformed to non-natural land- cover, with agriculture accounting for the majority 73% and urbanisation 24% of transformation. The rate of conversion to non-natural cover from 1994 to 2006 (12 years) has been calculated as approximately 1% of area per annum (DACE, 2009), representing a 12% loss of natural habitat over 12 years.
A similar scenario exists for Gauteng, where the proportion of non-natural land-cover is approximately 43% of the province, with agriculture accounting for 51% of this transformation and urbanisation 38% (GeoTerraImage, 2009). The Gauteng Protected Areas Expansion Strategy Report, 2011, in (Driver, et al., 2012) reports a 13% loss of habitat over 15 years, from 1995 to 2009, representing a rate of conversion for Gauteng of close to 0.9% per annum. At these rates all remaining natural habitat outside of protected areas for these provinces will be converted by around 2050 (Driver, et al., 2012).
30 In contrast to these provincial statistics, an image differencing analysis of the protected COH core area and buffer zone, confirmed a land cover change (not specifically transformation to non-natural) over 10 years (from 1999 to 2009) of approximately 2% for the COH core protected area, and 5% in the buffer zone, (Cooper, 2010), representing a rate of change of 0.2% and 0.5% per annum respectively.
The proportions of transformation and rates of change cited here do not accurately describe those for the entire MBR, but they highlight the rapid land-cover change in both provinces and the slower rate of change in the protected area, and hence the importance of the latter in conserving some remaining intact habitat from undergoing complete transformation.
Some of the more spatially explicit current processes that have accelerated land-use change, and threaten biodiversity of the Magaliesberg region are now reviewed, in the context of the study-area.
3.2.2 Threats associated with habitat loss in the MBR
Overwhelmingly the primary contributor to terrestrial biodiversity loss and destruction of ecosystems is the transformation of habitat through the change of land-use (Lambin, et al., 2001). Resource extraction, urbanization and cultivation are known to be the major contributors to transformation, with impacts ranging from total clearing to fragmentation of natural habitat (Murphy, 1988; Wilcove, et al., 1998; Neke & Du Plessis, 2004; Driver, et al., 2012). As previously mentioned the high levels of economic and population growth in the Magaliesberg region, and the proximity to Gauteng, implies a growing demand for land for infrastructure, low-income and upmarket housing that will continue to put pressure on the MPE core zone and surrounds ( DACE, 2007). There is little quantitative information at local-scales or in urban environments around the effects of land cover change (transformation and degradation) on biodiversity and ecosystem functioning and provisioning, except in biodiversity hotspots such as in the Fynbos and Succulent Karoo biomes (Reyers, et al., 2009; Rouget, et al., 2006; Rouget, et al., 2003; Driver, et al., 2003), and vegetation studies along the urban gradient (Cilliers, et al., 2004). However, addressing the underlying causes of biodiversity loss, i.e. local threatening processes in an area, is necessary for successful implementation of rehabilitation measures and mitigation of further losses.
The factors of habitat loss described hereunder are all spatially represented in this research by merged provincial classified land-cover data (GeoTerraImage, 2009; GeoTerraImage, 2008).
3.2.2.1 Urbanization
Urbanisation is usually a permanent threatening process, eroding local ecosystems and biodiversity (Murphy, 1988). Biodiversity responses to urbanization are affected by changes in habitat and ecological processes as well as change in interactions amongst species, and human-related disturbance to indigenous species (Hansen, et al., 2005; Murphy, 1988).A brief discussion on these changes follows.
31 It is quite evident that there are several types and densities of urban environments that would impact variously on elements of biodiversity. Urban to rural gradient studies investigate the changes to animals and plants along the urban gradient from semi-natural habitat on the urban fringe through to suburbia and inner city environments (McKinney, 2002; Hansen, et al., 2005). All of these urban densities are relevant in the MBR.
McKinney (2002),describes four types of replacement of lost habitat in urban settings, listed below in order of decreasing habitability towards the urban centre, the area where the habitat loss is generally greatest :
Natural remnant vegetation: remaining islands of original vegetation (usually subject to substantial alien plant invasion) Ruderal vegetation: empty lots, abandoned farmland, and other green space that is cleared but not managed Managed vegetation: residential, commercial, and other regularly maintained green spaces Built habitat: buildings and sealed surfaces, such as roads
For this study these descriptions of habitat loss could apply to essentially non-urban areas, such as old agricultural fields and plantations, golf-courses and roads that are scattered throughout the MBR. Thus the effects on biodiversity of these altered, variously managed habitats are relevant in the urban context, and would also be associated with habitat degradation and alteration outside of urban areas.
Studies in urban ecology and biotope mapping (Cilliers, et al., 2004), suggest that from an ecological point of view urban green space as described above, provide essential refuges, dispersal centres and movement corridors for species (Cilliers, et al., 2004), despite that some intensive management (repeated mowing, understory removal) may have negative impacts (Murphy, 1988). Yet, the physical changes that occur towards the city centre, strongly influence ecosystem functioning (e.g. impervious surfaces alter infiltration and runoff) and available habitat for indigenous species, leading to localized extinctions (Murphy, 1988). McKinney (2002), notes that localized extinctions occur especially during the construction phase of developments with the clearing of all vegetation and sometimes topsoil.
Hansen et al (2005) describes the phenomenon of low-density rural residential development (exurban), as being the fastest-growing form of land use in the United States since the 1950’s. The two types of exurban development he refers to includes ‘urban fringe development’ on the periphery of cities, driven by city dwellers wanting more rural lifestyles while still having access to urban jobs and facilities. Second, in rural areas often at the borders of national parks, ‘rural residential development’ is attractive in ‘natural amenities’, and offers good recreational opportunities (Hansen, et al., 2005). Both of these types of development, and their drivers are pertinent and characteristic of the MBR, and although more research still needs to be done on biodiversity response to urban
32 sprawl in different bioregions at local scales, the growing body of literature finds that the pattern and quantity of urban edge development strongly influences both indigenous and alien fauna and flora (Hansen, et al., 2005).
Another form of urban development prevalent at the periphery of urban areas is the growth of formal low-cost housing as well as informal, un-serviced settlements to cater for the influx of migrant workers to the cities of Johannesburg, Pretoria and Rustenburg, as well as the industrial and mining employment opportunities along the platinum highway between Pretoria and Rustenburg. There are approximately 15 informal settlements in the Madibeng district, with a further 20 along the entire Rustenburg platinum belt. A 2011 census of the Marikana ward in this district indicates that 40% of households live in informal dwellings, as opposed to the national average of 15% per ward (Kolver, 2013). This densification at the outskirts of cities contributes to habitat fragmentation and urban sprawl, and is considered to be one of the main threats to biodiversity in the Grassland biome of the North West Province (Cilliers, et al., 2004). In addition, the poverty and lack of services prevalent in these settlements also leads to degradation of habitat, biodiversity and ecosystem services (Hoffman & Ashwell, 2001).
McKinney (2002), and Hansen, et al. (2005) state that a number of taxa specific studies (insects, birds, plants, butterflies, reptiles) show consistent changes in species richness and species composition along the urban rural gradient. Overall, it is difficult to pinpoint which variables affect this. For example some studies indicate that species richness of certain taxa (mammals, birds, some insects and plants) tends to be higher in suburban areas, compared to semi-natural environments, due to environmental heterogeneity, where diverse and productive habitats rich in water and nutrients, occur close together in an urban setting. In addition, introduced ornamentals bear fruit and seed that attract various species, especially in established suburbs where succession has occurred (McKinney, 2002). On the other hand, the same author points out studies where bird diversity is reduced in suburban areas in comparison to semi-rural habitats. In the case of plants in urban areas, the diversity of indigenous species is often reduced as the proportion of alien introduced ornamental and invader species increases toward the urban core, signifying a change in species composition (McKinney, 2002).
Some generalist species adapt well to human impact and they thrive in urban environments by utilising urban habitat niches for hunting, foraging, shelter and nesting. Examples include alien species synonymous with cities, such as brown and black rats, the house sparrow and cockroaches, as well as many hardy annuals and grasses, that may compromise local species composition (McKinney, 2002). Species highly sensitive to human disturbance, such as large mammals, especially predators are mostly extinct from urban environments because they are actively persecuted, and reproduce relatively slowly. Other sensitive species may be niche specialists, or ground nesting species vulnerable to predation by humans and pets. These sensitive species also tend to be threatened by other far reaching transformations that accompany urban sprawl, such as roads, agriculture and other human activities (McKinney, 2002). Introduced animals in urban areas - 33 pet cats and dogs, prey on urban wildlife and can be particularly damaging in rural residential areas, especially those close to protected areas, because of the higher abundance of indigenous species (Hansen, et al., 2005).
3.2.2.2 Mining
The Bushveld Igneous Complex to the north of the Magaliesberg is known to hold the world’s largest platinum group metals (PGM), chromite and vanadium resources (Wells, et al., 2009). Ferrochrome operations occur closest to the mountain range, with the platinum mining belt beyond, both occur within the Marikana Thornveld vegetation type and granite is mined furthest away in the Norite Koppie Bushveld (NKB) (de Bruyn, 2007). A number of aggregate and dolomite/limestone quarries and old-workings talc quarries can be found throughout the MBR (GDACE, 2008). Opencast silica workings on the northern slopes of the Magaliesberg are now encroaching into the MPE (de Bruyn, 2007). It is predicted that much of the area around Rustenburg and Brits, and northwards to Thabazimbi, will be completely transformed by various mining operations in the not too distant future (de Bruyn, 2007; Loubser, 2007).
Mining is essentially a temporary land-use with permanent waste deposits, and depending on the choice of mining methods used the significance of environmental impacts are often site specific (Wells, et al., 2009). Significantly, chemical water pollution in the form of Acid Mine Drainage (AMD) from the oxidisation of pyrite in past gold mining operations around Krugersdorp, on the MBR’s southern border, has affected the quality of surface and ground water resources of the Cradle of Humankind (COH) and upper Crocodile catchment (Wells, et al., 2009; Durand, 2010). The effects on terrestrial biodiversity have yet to be fully determined.
No AMD occurs with the mining of PGM so environmental impacts are mostly in the form of dust and air pollution and habitat clearing. Typically, both underground and opencast mining methods are used after which mined PGM concentrate is dried, smelted and refined, with residue tailings as part of the process (Wells, et al., 2009). Mining operations can impact biodiversity and habitat in various ways. During operations the noise, dust and vibration from rock blasting may affect fauna (Wells, et al., 2009). One of the biggest current problems is excessive dust pollution from numerous tailings and open pits from operating and defunct mines. Dust can be a human health risk with respect to radiation and respiratory diseases, and has a high nuisance impact, lowering the quality of life in surrounding communities. Dust also retards vegetation growth and reduces palatability to animals (GDACE, 2008). While the bigger mining operations are implementing measures to reduce dust and backfill tailings, the smaller mines are not strictly regulated (de Bruyn, 2007). In 2012, the Department of Environmental Affairs gazetted an Air Quality Priority notice for the Bojanala Platinum district municipality, to apply special management interventions to rectify deteriorating air quality. Two affected municipalities Rustenburg and Madibeng, cover part of the MBR (DEA, Notice 495, 2012).
34 de Bruyn, (2007), refers to irreversible habitat clearing for opencast platinum, silica and granite mining and sand and stone quarrying as the biggest threat to biodiversity. Loubser, (2007) cites the practice of ‘boulder hopping’ as an unsustainable form of granite mining that is destroying vast areas of Norite Koppie Bushveld (NKB) vegetation. The process involves mining off the tops of koppies, and dumping waste rubble and non-export quality granite down the hillsides with no regard for rehabilitation.
The indiscriminate issuing of mining permits, without regard for environmental legislation and the lack of government regulation at operational level (e.g. many mines are operating without a water- use licence) are perceived to be major barriers to sustainable mining practice, (de Bruyn, 2007; Loubser, 2007).
A further critical aspect that is undermining the biodiversity and natural resources of the area is burgeoning informal settlements with a lack of service provision from Madibeng (and other municipalities), and the mining houses themselves. The Lonmin Platinum mine reportedly provides accommodation to less than 10% of its 28 000 employees (Kolver, 2013). Many migrant workers live in informal settlements in appalling conditions, with no access to electricity and sanitation or waste disposal (Essa, 2012). Hence, the biodiversity and habitat surrounding these settlements is increasingly vulnerable to degradation (fig 6).
Figure 6: Wasteland between the Lonmin mine and the Enkaneng informal settlement. (Source: The humanitarian news and analysis service, IRIN, photo © Jaspreet Kindra/IRIN)
The N4 Platinum highway marks the northern border of the MBR and while most mining occurs to the north of this highway, its position is critical for harbouring critical habitat (especially remnants of NKB), and for ensuring sustainable practices for operations within its transitional zones.
3.2.2.3 Cultivation
Large water impoundments and hydrological channelling in the past 100years, has promoted intensive agriculture in the region. Historically, fresh produce agriculture (vegetable cultivation, fruit orchards and dairy) would have occurred as close as possible to urban markets, which indicates that
35 arable lowland ecosystems close to urban areas have been modified over the long-term (Rebelo, et al., 2011). Presently, intensive pivot irrigated vegetable cultivation occurs in the Moot Valley and other low-lying areas of the MBR, with orchards towards Brits and large scale dry-land and irrigated cultivation in the grasslands of the south west.
Agricultural practices result in the transformation and degradation of land cover. Hoffman and Ashwell, (2001) refer to the extinction of at least 20 recorded plant species in South Africa due to agriculture - fifteen lost to ploughing for cultivation, three to overgrazing and two to afforestation. In the Rustenburg region a land-cover change study from 1972 to 2002, shows progressive conversion from natural bushveld and grassland to intensive agriculture (Ololade, et al., 2008). However, recently a proliferation of residential housing near Hartebeespoort dam has replaced irrigated croplands, presumably due to market changes - lower economic viability of agriculture relative to higher property values (Eber, 2005).
Agricultural landscapes can be seen as matrix habitats, from large expanses of monoculture through to the increased heterogeneity provided by remnant natural patches tree lines/hedgerows between agricultural fields and farm dams, which still provides variously suitable fragmented habitat for some biodiversity, and possible connectivity for some faunal migration (GDARD, 2011; Norris, 2008). The biodiversity maintained in these modified agricultural mosaics most likely differs from that suited to the original unmodified natural landscape (Norris, 2008), with perhaps only resilient species still occurring.
Despite the recent reduction of area under agriculture (Ololade, et al., 2008), remaining commercial farms are increasingly mechanised and more intensively farmed (Hoffman & Ashwell, 2001), assuming heavy additions of fertilizers and treated seed, and seasonally large expanses of bare soil, which all contribute to further degrading habitat.
3.2.3 Threats associated with habitat degradation in the MBR
Bai, et al., (2008 p.223) define land degradation as long-term loss of productivity (biological and economic) and ecosystem function “caused by disturbances from which land cannot recover unaided.” Hoffman and Ashwell, (2001) refer to land degradation as a combination of soil and veld degradation. The former encompasses wind and water erosion, soil mining, salinization and acidification of soils, water logging and crusting and compaction of the soil surface. The spread of alien invasive species, bush encroachment, deforestation, overgrazing and veld clearing for agriculture and other purposes are listed as types of veld degradation. Other forms of habitat degradation relevant in context, include fragmentation, incorrect fire regimes and many environmental pollutants including noise and light pollution. (Hoffman & Ashwell, 2001). These threatening processes can result in significant changes to vegetation cover, composition and structure, as well as changes to habitat integrity (Liu & Taylor, 2002; O'Connor & Kuyler, 2009) and
36 ecosystem functioning (Reyers, et al., 2009). These changes can also be exacerbated by local environmental conditions and climate (Bai, et al., 2008).
The pattern of land degradation in South Africa today is closely linked to the history of land ownership, land tenure and labour dynamics (Hoffman & Ashwell, 2001) . Bai, et al. (2008), investigated globally the link between ultimate drivers of land degradation such as population pressure, poverty and farming land-use, to measures of change in net primary productivity by remotely sensed normalized difference vegetation index (NDVI). In the case of South Africa they found a positive correlation between land degradation and population density, with a high coincidence of degrading areas within the former apartheid homelands, suggesting more than just rural population density as a variable here. Hoffman and Ashwell (2001) briefly allude to the consequences of migrant labour on degradation in the communal areas (former homelands) by reliance on natural resources, because of rural poverty and poor infrastructure and service delivery.
Hoffman & Ashwell, (2001) report a decreasing veld degradation trend country-wide, which suggests a better awareness of the causes of degradation, and improved application of intervention measures that can lead to a reversal of the problem. Likely factors leading to the degradation and disturbance of open space in the MBR are summarised below. Spatial data for some of these factors made it possible to include degradation as a layer in this research. For each degradation threat it is indicated whether spatial data are available. From the brief discussion on some of the drivers of land degradation below, it is evident that many are interlinked in spurring degradation, for example one stressor, say overgrazing can drive another, say bush encroachment, thus compounding the threatening processes.
3.2.3.1 Overgrazing
Grasslands and savanna grasses are used for pasture, with the raising of livestock on veld being the dominant form of land-use throughout South Africa (Hoffman & Ashwell, 2001). While grassland ecology relies on grazers, rodents birds and insects to distribute seed, stimulate new growth and cycle nutrients through the soil (Van Oudshoorn, 2012), overgrazing by incorrectly managed stock farming in grassland and savanna systems leads to degradation of the veld.
Overgrazing is the repeated use of the grass plant until its reserves become depleted as its roots no longer absorb water and nutrients effectively, leading to the death of the plant (Van Oudshoorn, 2012). Selective grazing of palatable species results in changes in species composition, with an increase in annuals, unpalatable grasses and woody shrubs, such as Stoebe vulgaris (bankrupt bush), until even unpalatable species are grazed and vegetative cover is left denuded. Typically this sparse vegetation cover of overgrazed areas, exposes the soil surface increasing erosion potential and topsoil removal (Van Oudshoorn, 2012; Hoffman & Ashwell, 2001) .
Rouget, et al (2006), states that degradation, largely due to overstocking is the primary cause of biodiversity loss across Southern Africa. Generally, stocking rates in the former apartheid homelands
37 (communal areas) were 1.85 times higher than those of commercial farms, which were stocked at recommended levels (Hoffman & Ashwell, 2001), which led to the communal areas being particularly barren of vegetation and vulnerable to erosion. On the other hand, Hoffman and Ashwell, (2001), found that compositional change to veld was seen to be more of a general problem on commercial farms due to selective grazing.
The MPE EMF ( DACE, 2007) suggests overstocking of game for tourism purposes on portions of land that are unable to support such quantities contributes to overgrazing in the Magaliesberg region. Overgrazing is related to the degradation layer in this research, as a potential cause of degradation detected by satellite land-cover mapping for the region (GeoTerraImage, 2009; GeoTerraImage, 2008) .
3.2.3.2 Erosion
Erosion by wind or water occurs when the soil lacks protective cover such as on bare or tilled lands. Water or wind dislodges and removes particles of soil from the surface, mostly by a fairly uniform removal of topsoil called sheet erosion, but sometimes in a concentrated flow of water forming rills and gullies, landslides and stream bank collapse. Erosion reduces the productivity of the land by reduced water infiltration, soil organic matter, soil nutrients, and soil depth which diminishes the diversity of plants animals and microbes, and threatens the stability of ecosystems (Pimentel & Kounang, 1998). Badly eroded agricultural fields are abandoned, and replaced by converting areas with natural vegetation cover to agriculture, which places more strain on intact biodiversity resources (Pimentel & Kounang, 1998).
The natural rate of erosion in undisturbed vegetative cover is between 0.02 and 0.75 tons per hectare per year, depending on rainfall erosivity, slope and the physical characteristics of the soil, this can increase to 25.7 tons /ha/yr for bare soil and 5.9 tons /ha/yr for rotational mixed cropping (Hoffman & Ashwell, 2001) In general the extent of the Magaliesberg region is classified as med-low erosivity risk according to the type of soils present, and taking slope into account (AGIS, 2007).
Erosion gullies are a symptom of severe erosion. Spatial data for gullies (Mararakanye & Le Roux, 2011) are included as part of the degradation layer in this research.
3.2.3.3 Old lands
Old lands are historical agricultural fields since abandoned, that show various stages of natural or semi-natural re-growth (GeoTerraImage, 2009). Land around the Magaliesberg has been intensively farmed for well over a century (Carruthers, 2015) because of its proximity to markets in Johannesburg and Pretoria. Classified land-cover from satellite images indicates a number of old abandoned agricultural fields indicative of the trend away from agriculture ( DACE, 2007). Hoffmann and Ashwell (2001) list a number of reasons for the decline in total cropland and rangeland area from around 1980 onwards in some provinces, including Gauteng. Likely reasons for the study region being, change of-land-use to urban, mining or conservation, reduced access to institutional
38 support (financial and technical), loss of financial viability due to increased production costs and conversion of marginal cropland to pasture. In the 1990’s government subsidies in the agricultural sector for crop cultivation, were revoked (Neke & Du Plessis, 2004), and with high land values, many of these croplands around the Magaliesberg are left fallow or are converted to residential developments and related infrastructure, or to game farms, as a more conservation oriented alternative (Eber, 2005). Classified old lands are included spatially in this research as a degree of degradation or disturbance threat.
3.2.3.4 Fire
Fire as an ecological factor, is an essential part of the savanna and grassland biomes. Over millions of years fire tolerant and fire dependant species have evolved in the presence of fire (Govender, et al., 2006). On the Highveld, lightning is responsible for about 20% of wild fires, the balance being anthropogenic ignitions. However, fire intensity, pattern and spread in grassland and savanna landscapes is limited by seasonal variations in fuel-load and fuel (biomass) moisture rather than by ignition frequency (the opposite applies for forest biomes) (Scholes, 2012; Govender, et al., 2006).
Research by Govender, et al., (2006) in the KNP savannas indicates that fire intensity is lower for summer than for winter fires, but rises with increased fuel loads in instances where precipitation was high for the preceding two years. Data from the Pilanesberg Game Reserve (typically savanna bushveld close to the Magaliesberg) indicates approximately a third of the area is burned annually, with a larger fraction being burned after a good rainy season (the average being ±600 mm) (Scholes, 2012). These findings indicate a climatological influence on fire intensity and spread. Govender, et al.,( 2006) also showed increased mean fuel loads with increasing post-fire age, to a maximum of 5 years, with a decline in fuel load shown in the sixth year post fire, which could indicate that fire frequency also influences fire intensity.
In savanna vegetation, variation in fire intensity is seen as an essential factor in maintaining the grass/tree matrix. Intense fires result in tree mortality and reduced recruitment to adult sizes (Smith, et al., 2012). Uys, (2004) found large compositional changes (bush encroachment) in grassland plots where fire was excluded for 10+ years.
In terms of biodiversity, the same grassland study found no response to either frequency or season of burn in terms of alpha diversity (species richness). A variety of Beta diversity(landscape scale rate of change in species composition) responses were noted in response to biennial, octennial and no- burn treatments, as opposed to annual fire-break burns. Grassland forbs appeared to be particularly resilient to any fire regime while dominant grasses showed a strong response to season and frequency of burn (Uys, et al., 2004), which could support the suggestion that repeated fire in the wrong season can lead to veld deterioration over the long term (Van Oudshoorn, 2012). Similarly, the results of (Smith, et al., 2012) show a burn interval of 1 to 5 years maintains fire tolerant species and species richness, with perhaps a longer interval required to maintain species with low fire
39 tolerance. The effects of fire exclusion on both tolerant and low-tolerance species, needs further study.
Detrimental effects of fire have been noted with respect to some faunal species, specifically reptiles and in riverine and forest vegetation that is not fire tolerant (Van Oudshoorn, 2012).Traditionally veld management for fire was related to maximising grazing capacity with hot fires applied to combat woody encroachment to pasture (Govender, et al., 2006) and 2-3 year burn cycles to coincide with the flowering cycle of grasses (Scholes, 2012). Fire was also used to remove moribund to improve early grazing (Van Oudshoorn, 2012).
To date however, the effects of fire on all savanna and grassland biodiversity is still poorly understood, and so are the long term effects of changing fire regimes (Smith, et al., 2012). Recent suggestions focus on flexible fire regimes or “pyrodiversity” in these biomes, to encourage a variety of fire-types and a spatial and temporal randomisation of fire regimes over the long-term (Smith, et al., 2012; Scholes, 2012). Spatial fire data is developed as a fire frequency/suppression layer to indicate biodiversity threat in this study.
3.2.3.5 Alien invasive species
Invasive alien species, especially invasive alien plants (IAP’s), are a major problem in the terrestrial environment. The National Alien Invasive Plant Survey 2010 shows a total of 20 million hectares affected by the presence of IAP’s, with 106 new species (mostly naturalised ornamentals) declared invaders in the five years prior to the survey (Kotzé, et al., 2010a; Driver, et al., 2012; Department of Environmental Affairs, 2014). There are two processes of alien plant invasions, expansion and densification, where existing stands of IAP’s spread by dispersal, particularly along seed dispersal vectors, such as paths, roads and seasonal and perennial watercourses, which is then followed by the densification of new colonies (Hoffman & Ashwell, 2001). Most alien species are introduced for agricultural, horticulture or forestry purposes since the 1650s. Many weeds have been introduced accidentally with fodder imported from Argentina for horses during Anglo Boer war (Hoffman & Ashwell, 2001). Problem woody species, such as pine, eucalyptus and poplar (introduced for poles for construction, infrastructure and mining) continue to spread from source plantation, while black wattle (introduced for leather tanning), syringa and jacaranda (introduced ornamentals) have subsequently naturalised (Hoffman & Ashwell, 2001; Department of Environmental Affairs, 2014).
This study uses spatial data from The National Alien Invasive Plant Survey 2010. The survey identifies 27 prominent invasive woody species nationally (Kotzé, et al., 2010), of which 11 occur in the MBR. They are tabled showing listed invader category status in order of prevalence (table 4). Spatial data for occurrence and density (abundance) for the 11 identified woody IAPS in the terrestrial and riverine environment are applied as a threat layer in this study.
40 Table 4 - List of eleven woody IAP’s and their listed invader category status. (source Kotzé, et al., 2010; Department of Environmental Affairs, 2014)
Species Common Name Category Acacia mearnsii/dealbata/baileyana Wattle 2 Eucalyptus spp. Gum 2 Melia azedarach Syringa 3 Populus spp. Poplar 2 Jacaranda mimosifolia Jacaranda 3 Pinus spp Pine 2 Arundo donax Spanish reed 1 Cereus jamacaru Queen of the night 1 Salix babylonica Weeping willow 2 Agave spp. Agave 2 Opuntia spp. Prickly pear 1
Invasive alien plants displace indigenous species, disturb habitats, and disrupt ecosystem functioning, transforming the ecology of the area they inhabit (Driver, et al., 2012). Besides the well known impacts of IAP's on water resources, their impacts on ecosystem functioning and biodiversity change composition and structure by outcompeting indigenous species, leaving them vulnerable to extinction. Alien invaders also affect pollinators and decomposer organisms.
The working for water programme, a well funded initiative established in 1995, has created jobs through natural resource management to clear invasive alien plants and protect vulnerable water resources. The programme had cleared two million hectares by 2007. Efforts are intensive and ongoing as follow ups are needed, but it does not tackle all areas and all problem species. (Wannenburgh, 2006; Driver, et al., 2012)
3.2.3.6 Fragmentation
Human development is a major force behind declining global biodiversity through landscape fragmentation (Fahrig, 2003). Habitat fragmentation is usually defined as a landscape-scale process involving both habitat loss and the breaking apart of habitat. Human land-use practices often result in fragmented patches of remnant vegetation embedded within an agricultural or urban matrix, which is particularly noticeable in lower lying areas of the MBR. Typically, habitat fragmentation leads to decreased remnant patch size, higher edge : interior ratios, and increased patch isolation (Fahrig, 2003), which diminish the chances of a species being represented or persisting in a landscape, because small isolated fragments are less likely to support species due to local extinction, unlike large fragments which not only support species but are the source of dispersal of sub-populations spreading and occupying smaller fragments (Rouget, et al., 2004). In addition, fragmentation may be linked to habitat connectivity or corridor mapping for wildlife movement or climate change adaptation,
41 as fragmented habitats can severely restrict movement and dispersal of species (van der Ree, et al., 2011).
Road network density and pattern affect habitat fragmentation and connectivity (Zhifeng Wu & Cheng, 2013). Besides road mortality impacts on biodiversity, roads, and the fences often associated with them, can hinder access to feeding areas or water points and may present as significant obstacles to migration of some species (van der Ree, et al., 2011). Although no literature on road ecology or fragmentation was found pertaining to the Magaliesberg region, this study applies the road network as a simple spatial indicator of fragmentation threat.
3.2.4 Magaliesberg regional threats synthesis
Other relevant proximal threats often assessed at the species level which can severely impact the biodiversity of the Magaliesberg, are hunting or trapping or even poisoning of fauna -particularly species at higher trophic levels (Kuhn, 2014; Whittington-Jones, et al., 2014)’ or endangered flora harvested for horticulture, such as aloe peglarae, and for traditional medicine markets (Driver, et al., 2012). In addition, excessive harvesting of woody plants for energy needs, especially in rural, peri- urban areas and settlements without electrification, can denude areas and impact biodiversity structure and composition (Hoffman & Ashwell, 2001). Driver, et al., (2012) identify poaching as one of the top two threats affecting formally protected areas.
Some threatening processes impacting the region are without spatial data, however they also contribute to disturbance and degradation of biodiversity, and often stem from another threatening process (Wilcove, et al., 1998), such as those already mentioned. For example, bush encroachment, which alters the structure and composition of habitat by the spread of indigenous woody trees and shrubs into grasslands and the densification of savanna bushveld, is often driven by degraded sparse veld conditions, and fire suppression. Encroachment may be exacerbated by appropriate rainfall conditions, and more so in the absence of browsers (Hoffman & Ashwell, 2001; Ward, 2005).
Pollution is another factor of land degradation that results from poor, unsustainable management of agriculture, industry, urban and mining land-use (Hoffman & Ashwell, 2001). These examples and others mentioned in this section, demonstrate that factors of transformation and degradation are interlinked and may result in complex interactions between multiple threats and natural stressors. Biodiversity’s response to these pressures may vary considerably, and may be immediately apparent, or manifest over time. Some threats can be mitigated and managed, while others lead to changes in biodiversity that are detrimental and irreversible, presenting a problem for sustainable conservation (Goudie & Viles, 1998; Driver, et al., 2012). Thus, the cumulative effects of multiple pressures should also be considered.
42 This research aims to collate spatial data on some of the predominant threats mentioned in this section, to assess the spatial variation of each threat and a threat composite, and in addition quantify the spatial extent and intensity of these threats in a regional context.
3.3 Assessment of Threats
This work is positioned as spatial decision support and can be summarised as a spatial conservation planning exercise that quantifies exposure (by extent) and intensity (by AHP) (Wilson, et al., 2005a) of multiple threat criteria to determine priority threats to biodiversity in a case study. An exploratory and descriptive research design using mixed methods of quantitative analysis is applied to realise the research aims. The research combines several methods used for research and applications in the spatial prioritization, site selection, conservation planning and threats or vulnerability assessment arena over the past two decades. Techniques and tools applied to data and to generate results include, GIS processing and land-cover classification; Spatial Multi Criteria Analysis (MCA), the Analytical Hierarchy Process (AHP), and basic statistics.
This section reviews the literature on spatial prioritization and threats analysis, looking at how threats are assessed spatially and the variety of data used as spatial surrogates to indicate threat in systematic conservation planning and remote sensing applications. Following this is an explanation of how the techniques and tools chosen for this research are integrated and applied, backed up by a technical review of methods used in related research. A brief look at how spatial priorities for conservation and threat can be applied by stakeholders involved in conservation and land-use planning, completes the section.
3.3.1 Measures of threat in conservation planning
Systematic conservation planning (SCP) is a rapidly evolving area of research that is becoming increasingly systematic in the selection of conservation areas by integrating the biodiversity and conservation targets of a region with its socio-economic and development constraints. SCP is compelled by the rate and extent of human population and development growth, and the rate of biodiversity loss and environmental change taking place globally (Moilanen, et al., 2009).
Inroads into incorporating threats and vulnerability in SCP is evident in seminal review papers, which have highlighted the importance of including environmental, social and economic factors rather than solely biological factors, in conservation planning (Margules & Pressey, 2000; Pressey, et al., 2007; Noss, 2000; Wilson, et al., 2005a; Pierce, et al., 2005). At first the focus of spatial conservation prioritization was on the representation and persistence of species, and other biodiversity targets (Margules & Pressey, 2000), then it evolved to identify threat patterns in a spatially explicit manner in relation to biodiversity pattern (Pressey, et al., 1996) and incorporated integrity of habitats, ecosystems and ecological processes (Rouget, et al., 2003; Noss, 2000) particularly through irreplaceability and vulnerability analysis (Pressey & Taffs, 2001; Noss, 2002; Lawler, et al., 2003). Stand-alone threats and vulnerability assessments too, are often referenced to
43 biodiversity features or priorities for conservation planning applications (Neke & Du Plessis, 2004; Reyers, 2004; Rogers, et al., 2010; Vimal, et al., 2012; Weeks, et al., 2013).
More recently attempts have been made to include other dynamic factors into conservation prioritizations, to cater for changes in ultimate drivers of change, the varying behaviour and attitudes of stakeholders as well as the resulting change to biodiversity patterns and processes (Pressey, et al., 2007). Examples are prioritizations for socio-political (Knight, et al., 2010) and economic cost data, restoration efforts (Orsi & Geneletti, 2010) and ecosystem services (Reyers, et al., 2009). Moilanen, et al., (2009) and Knight, et al., (2010) suggest that research is needed and ongoing in spatial conservation techniques to better incorporate dynamic landscapes and threat factors. The effects of compound interacting threats, such as land-cover change with altered hydrology and climate change on ecosystems and biodiversity features is still an under explored aspect of threat and vulnerability analysis in conservation planning.
From a biodiversity perspective, many conservation assessments have focused on the irreplaceability of biodiversity features, using sophisticated algorithms which incorporate spatial context dependence, whereby a sites effectiveness to encompass biodiversity (specified factors or targets) is prioritized in terms of its location, size or network configuration, relative to other locations (Ferrier & Wintle, 2009; Moilanen, et al., 2009).
From a threats perspective, measures of vulnerability spatially define and identify areas or biodiversity features, which are vulnerable to threatening processes. In this context Pressey, et al., (1996, p. 529), define vulnerability as “the likelihood or imminence of biodiversity loss to current or impending threatening processes”. There are no distinctive techniques common to the majority of vulnerability and threats analysis. Wilson, et al., (2005a), categorise threats and vulnerability assessments in terms of the data used to predict current and future vulnerability, and they also identify broad approaches, and combinations of approaches, applied in several studies to integrate data and assess threats or vulnerability. One general approach is informal qualitative or quantitative analysis, such as spatial scoring based techniques, like MCA weighted overlay (Li & Nigh, 2011; Moffett, et al., 2006b) or correlation and rule-based modelling (Veech, 2003; Vimal, et al., 2012; Rouget, et al., 2003). The second approach is more formalized statistical modelling, such as regression tree analysis (Wilson, et al., 2005 b; Rouget, et al., 2003). Alternatively, the complemetarity based algorithms already mentioned, build-in a cost layer, such as transformation threat or economic cost into their priority iterations (Moilanen, et al., 2009).
However, it is evident from the literature that many threats assessments use only one, or surprisingly few vulnerability indicators and these are often arbitrarily quantified. This study is an attempt to bring together several threats (including less common indicators of degradation) as a multiple threat composite, which is formally quantified.
44 The degree of threat or vulnerability can be ascertained by typical environmental risk assessment measures (Aucamp, 2009), and in the SCP context, Wilson, (2005a) uses these to propose three dimensions of vulnerability- exposure, intensity and impact - as an attempt to address some of the challenges and uncertainty with data and measuring, defining and validating the degree of threat or vulnerability. In this research basic exposure and intensity dimensions are applied to demonstrate two ways that threats may be quantified, and shows that using more than one dimension, provides an additional perspective on results.
A spatial approach is applicable to measures of exposure (the probability of a threatening process affecting an area or feature over time, or the sensitivity of areas to threat), examples of the types of spatial data used to indicate exposure are maps of agricultural suitability, soil erodability, or mineral deposits (Wilson, et al., 2005a). In this work the geographic extent of all threat data indicate the exposure dimension. Measures of intensity (the frequency, duration or magnitude of a threatening process), can be mapped spatially in degrees of intensity and can take many forms, for example, the density of IAP’s, road networks or housing, the frequency of fire or the quantity of timber harvest (Wilson, et al., 2005a). The AHP method is applied to measure intensity for all threat criteria in this work. Exposure and intensity can be measured categorically (low, med, high) or as a continuous surface. Impact, which is referred to by Wilson et al (2005a p.530) as “ the effects of a threatening process on particular features”, can only be assessed with reference to a particular biodiversity component or feature (O'Connor & Kuyler, 2009), and is not usually reperesented in a spatial generalisation.
3.3.2 Spatial indicators of vulnerability and threat
The types of threatening processes that are appropriate for spatial analysis are the drivers of habitat transformation and loss, where impacts are concentrated at local or regional scales, such as the spatial extent of urbanisation and agriculture (Rouget, et al., 2003), or invasive plant species spread (Wannenburgh, 2006; Forsyth & van Wilgen, 2009). These land-cover/uses may be derived from processed classified, unclassified or digitized remotely sensed imagery at scales ranging from 1:80 000 (first Landsat images from 1970’s) to 1:10 000 (SPOT 5 for this analysis) or even finer scale data.
Using LCC the pertinent categories of threatening processes (land-uses) can simply be extracted as a mask- layer to represent the spatial extent of transformation, or may be used to determine the percentage of transformation in a given area or planning unit (Li & Nigh, 2011), or within a given vegetation type (Wannenburgh, 2006; Rouget, et al., 2004). LCC can also be used for measuring the extent of habitat fragmentation (Rouget, et al., 2004), or to spatially analyse the impacts of all land cover classes on a particular ecosystem or feature, such as water quality of wetlands (Jaganath, 2009).
45 Besides time-series land-cover and land-use data to indicate proximate threats, there are many other useful general measures of vulnerability. Other ways of expressing current and future population and development pressures on biodiversity include human population demographics at varying scales, such as density mapping by census data (Vimal, et al., 2012), the rate of development by region (Veech, 2003) or by population growth projections (Abbitt, et al., 2000). However, population statistics are often not at an appropriate scale to indicate proximate threats because enumerator districts do not necessarily correspond with conservation planning regions, (as is the case in this study), so other surrogates for proximal anthropogenic pressures are required.
Different monitoring systems may be optimal for different eco-regions, and a time series of observations is almost always required to indicate dynamic change. In tropical and temperate forests and woodlands, burning and logging are major contributors to habitat destruction ( (Fuller, et al., 2010; Orsi & Geneletti, 2010) and the proportion, rate and pattern of deforestation, or biomass burning, detected from remote sensing platforms, are useful indicators of current and potential site vulnerability to population and/or socio-economic and/or development pressure (Rogers, et al., 2010; Pressey & Taffs, 2001; Lambin, et al., 2001). In some conservation assessments the frequency of anthropogenic fire events (Rogers, et al., 2010), or the suppression thereof, have been considered important threat factors for some ecosystems (Noss, 2000). In this research, the threat of fire is indicated by fire frequency and absence data (a time-series burn-date product) derived from an automated algorithm for mapping post-fire burned areas using 500-m MODIS imagery coupled with 1-km MODIS active fire observations (Giglio, et al., 2009).
A useful indicator of land degradation (overgrazing and desertification) in semi-arid eco-regions such as karoo fynbos (Rouget, et al., 2006), as well as savanna and grassland regions are normalised difference vegetation indices (NDVI) which are spectral indicators sensitive to fluctuations in primary productivity associated with climatic fluctuations. Spectral signatures of areas of unnaturally low vegetation cover compared to surroundings (usually a consequence of excessive livestock grazing and browsing) can also be classified by LCC and ancillary methods (GeoTerraImage, 2009). However, Rouget, et al.,(2006) cautions that measuring degradation using coarse scale land-use data is inconsistent and mostly underestimates the extent of degradation, after ground-truth verification found many areas classified as natural that were actually degraded.
There are challenges with mapping invasive alien plant spread (Rouget, et al., 2003), and a lack of fine scale data, particularly for herbaceous invaders that cannot easily be detected by imagery. Some assessments have used climatic/habitat envelope modelling using climatic and topographic variables to predict the likely presence or absence of invasive species (Rouget, et al., 2003; GDARD, 2011).
In areas where intensive agriculture is the primary cause of habitat transformation (in addition to ecosystem disturbance resulting from unsustainable agricultural practices), an alternative to general land-use maps with cultivation-categories (such as irrigated, dry-land, orchards, grains, annual-
46 crops, planted pasture, mixed agriculture, permanent culture), is a spatial agricultural potential indicator for crops or forestry, as a surrogate for current and future threats from cultivation (Reyers, 2004). Similarly, a spatial grazing potential indicator could feasibly be used as a surrogate for pressure from agriculture (Driver, et al., 2003). Neke & Du Plessis, (2004) use agricultural suitability models for crops, forestry and grazing to predict areas vulnerable to future threat.
Spatial data on mining operations and activities may come in the form of point locations, or polygon extent derived from land-use maps. However, this does not capture the range of mining impacts for accurate spatial assessment (Rouget, et al., 2003). The underlying geology and mining potential for an area can indicate potential pressure from mining operations. Buffers around coal seams and mineral deposits are useful for this purpose (Neke & Du Plessis, 2004).
Instead of buffering a given threat factor to indicate impact of human presence, using a distance function in GIS allows threat to decrease in relation to distance away from it. Vimal, et.al. (2012) used a weighted distance function for census data to weight cells to a radius distance of 2km (fine scale) and 50km (coarse scale) from population centres. An alternative to a distance function is the use of a kernel density function, for example, that used by Zhifeng Wu & Cheng, (2013) to quantify the relation between road density and fragmentation. The study found higher road densities (i.e.km/km²) to be positively correlated to increased landscape fragmentation. An alternative to using LCC classes to indicate the threat of urbanization, is to use building density (dots per hectare or cell), from Eskom, (2008) spot-building count data (SANBI, 2009).
Road density has been linked to fragmentation and to several ecological effects of roads, as roads segregate the landscape into smaller patches (Zhifeng Wu & Cheng, 2013; Fahrig, 2003). In the NSBA, (2004) (Rouget, et al., 2004) different land-use types were ranked in terms of connectivity and resistance for a fragmentation index, using roads and LCC. From these data they quantified untransformed habitat patch size and extent, and matrix resistance, which are defined as the degree of ease or difficulty with which species can cross a landscape (Rouget, et al., 2003). This research uses a simple distance function for roads and built-up areas to indicate the threat of fragmentation.
Spatial surrogates to indicate threat or ecosystem status are limited by the types of spatial data available, thus remote sensing is a cost effective, convenient and appropriate as a landscape scale indicator to quantify the general extent of habitat transformation and alteration in a regional context (Noss, 2000). The availability of relatively fine scale provincial LCC data in South Africa is an ideal base-line surrogate for threat, to which other data may be added if available, as is achieved in this research.
3.3.3 Integrating GIS, Multi Criteria Analysis and the Analytical Hierarchy Process
The SCP methods available are generally either complementarity based, or scoring based approaches. The former incorporates spatial context dependence, whereby sites effectiveness to encompass biodiversity (specified factors or targets) is prioritized in terms of its location, size or 47 network configuration, relative to other locations (Ferrier & Wintle, 2009; Wilhere, et al., 2008). In SCP scoring based approaches ( which includes MCA), spatial units are individually assessed independent of one another, and quantified to derive a spatial composite index of conservation priorities, based on multiple criteria to realise an objective (Moilanen, et al., 2009).
A spatial MCA approach is used in this work. Malczewski, (2010) describes the synthesis of GIS and MCA as important for the application of spatial decisions. GIS being a tool for the storage, retreival, manipulation and analysis of spatial and attribute data, with a spatial output, combined with the MCA which provides techniques and proceedures for structuring decision problems. Decision rules define how criteria are combined to form a single composite index and how to best rank alternatives or make choices between them (Malczewski, 2010). Two of the most popular forms of decision rules in GIS based MCA are weighted summation and AHP, which can be related and applied together in the GIS environment (Malczewski, 2010).
“MCA involves the qualitative or quantitative weighting, scoring or ranking of criteria in terms of a single or multiple objectives” (Kwaku Kyem, 2011, p. 490). It is reported to be adaptable, transparent, relatively easy to implement and can be participatory (Li & Nigh, 2011; Ferrier & Wintle, 2009). Where criteria have an explicit geographic component they become spatial data in the form of thematic map layers (environmental, socio-economic, biotic or abiotic, factors or constraints) with computationally straight-forward decision rules, such as binary presence or absence of a particular feature or threat, or ordered threat levels, or a continuous spatial metric such as land-cover. The attributes of each map layer translate to a measure of suitability per unique spatial unit (cadastral boundaries, watersheds, catchments, uniform hexagons or raster pixels), for the objectives stated (Kwaku Kyem, 2011).
The suitability per spatial unit in this research is quantified and ranked using Saaty’s AHP, which is a form of MCA that can be applied spatially (Saaty & Vargas, 2001). The method formalises expert judgement into a quantitative record, based on relative pair-wise comparisons of multiple criteria relative to stated objectives (O'Connor & Kuyler, 2009).The AHP method provides a logical, comprehensive and participatory method to select and weight alternatives and criteria relative to each other and the objectives, by first decomposing the problem in a hierarchical structure and then comparing the criteria in pairs with respect to each level in the hierarchy (Saaty & Vargas, 2001). Expert judgement is applied to evaluate and rank the criteria in relation to the objectives in the hierarchy according to Saaty’s comparison scale (Saaty, 1977). This ranking is used to create a matrix of selected criteria for pair-wise comparisons whereby the eigenvector method computes the relative weight of each factor. A series of aggregations are performed to prioritize each criterion referenced to the objectives, and alternatives with respect to each criterion. Finally the rating of alternatives (spatial units) defines their suitability (Vasiljevic´, et al., 2012).
Variations of MCA and AHP have been applied extensively as a management and planning decision support tool that is able to consider multiple attributes in many conservation-based and
48 land-use applications particularly in land-suitability and site suitability analysis (Chandio, et al., 2012; Vasiljevic´, et al., 2012), regional eco-environmental evaluation (Ying, et al., 2007), protected area planning and conservation zoning (Geneletti & van Duren, 2008; Moffett, et al., 2006b) restoration and forest conservation (Zerger, et al., 2011; Orsi & Geneletti, 2010; Phua & Minowa, 2005), prioritizing alien plant invasion controls (Wannenburgh, 2006; Forsyth & van Wilgen, 2009) and threats to biodiversity (Fuller, et al., 2010; Noss, 2002; O'Connor & Kuyler, 2009; Vimal, et al., 2012).
3.3.4 A review of methodologies that incorporate vulnerability and threat
assessment
Often a single approach is used in a biodiversity site selection or assessment, where multiple criteria (variables or constraints) are selected and summarised in one systematic procedure (MCA). Depending on the different kinds of data, the spatial scales at which the data was collected and the number of data inputs, Rouget, (2004) suggests it may be impractical to use a single approach to summarise biodiversity features and suggests alternative forms of analysis may be more suitable. One alternative mentioned in Marguiles & Pressey (2000), is to introduce select variables or constraints (that influence the biodiversity targets set) as a separate exercise after the initial selection. The types of variables or vulnerability constraints that may be introduced include, amongst others, existing protected areas, expert-preference mappings, irreplaceability, or ecological processes (Driver, et al., 2003) as variables, and, transformed areas (Lawler, et al., 2003; Rouget, et al., 2004), modelled areas of future pressures, (Driver, et al., 2003; Reyers, 2004; Rouget, et al., 2004) and expert derived mapping of constraints (Driver, et al., 2003) for constraints. In some biodiversity assessments, areas that are not expected to contribute to conservation/ biodiversity targets, are not included in the selection process at all, and may be masked out, such as the exclusion mask of all built-up areas applied in C Plan 3.3 (GDARD, 2011).
Examples of all-in-one systematic prioritizations are found in studies by Wannenburgh, (2006), and Li and Nigh, (2011), who use a single multi-criteria approach to prioritize planning units to suit their objectives. Vimal, et al., (2012) uses a similar approach to assess the spatial vulnerability of several biodiversity descriptors. In all three studies, criteria values are normalized by expressing them on a scale of 0-1, to allow for comparison of values with different units and ranges, (i.e. area values, density ranges or % coverage of a criterion per planning unit). Next, weights are applied to each criterion, and all criteria (both variables and constraints) are considered together using a GIS weighted linear combination overlay procedure (Malczewski, 2010), to produce consolidated composite map products. In C Plan 3.3 (GDARD, 2011), a number of approaches and analyses were done to develop biodiversity features layers, but the final selection for irreplaceability included all biodiversity features and costs in the C Plan (software) site selection algorithm (more details on C Plan 3.3 follow in Section 4.2.5).
49 An alternative approach to final integration was employed in the National Spatial Biodiversity Assessment 2004 (Rouget, et al., 2004). To determine biodiversity priority, biodiversity factors (habitat-types, species and ecological processes) were processed and analysed into priority areas separately and then combined to produce an overall map of nine significant areas for terrestrial biodiversity that reflect habitat, species and process priorities. Each factor was developed separately because of the different scales of the input data (hence different sizes of planning units), and to avoid averaging different kinds of data, which could potentially hide/mask an area where only a single factor featured (For example a high scoring site for one factor only will not stand apart from a site with an average score for all factors).
In addition to biodiversity priority, eight vulnerability maps of future pressures on biodiversity were assembled using a variety of socio-economic and environmental criteria, to produce two vulnerability products – 1) An average land-use vulnerability score, and average degradation vulnerability score for every 100m x 100m raster cell for the whole country. 2) A map of areas of very high vulnerability to any of the determined future pressures. These vulnerability maps were compared with vegetation- types, species and overall priority areas separately in different analysis, using a measure of the degree to which the priority areas are impacted by vulnerability factors (Rouget, et al., 2004).
In a similar assessment, the Succulent Karoo Ecosystem Plan (SKEP) uses a systematic ecosystem approach (Driver, et al., 2003). Here conservation targets were derived from biodiversity and ecological features layers and inputted into C Plan for a map of irreplaceability (conservation options). This map is interpreted together with spatial information on transformed areas and expected future land-use pressures, producing maps of conservation options in relation to vulnerability across four regions. From these maps broad-scale geographic priorities for conservation action were ultimately identified.
Reyers, (2004), achieves an irreplaceability / vulnerability analysis of quarter degree grid cells in the Limpopo Province by scoring each cell with a value for irreplaceability and vulnerability. Irreplaceability, representing biodiversity targets set for species and vegetation types, taking into account the measure of current threat within each vegetation type, as a proportion of it. Current threats included road buffers and broad land-cover classes of transformed and degraded areas. Irreplaceability of sites was then identified using C Plan software algorithm. Vulnerability was determined as a measure between 0-100 of the average suitability of cells for forestry or dryland cultivation, obtained by overlaying site suitablilty maps on the quarter degree grid cells. Finally irreplaceability and vulnerability scores for each cell were represented on a two dimensional plot, and used to draw maps and identify conservation areas (Reyers, 2004).
A comparative study of the two broad approaches outlined here to represent conservation value and vulnerability constraints in a Mid-Atlantic region of the U.S. was done by Lawler, et al., (2003). The first approach as per the NSBA and SKEP, produced a final product of continuous site rankings for irreplaceability (species occurence in this case) and vulnerability (urbanization, open mines and
50 agriculture), that are simulataneousely displayed in a map using a bivariate colour scheme, representing areas ranging from low irreplaceability and low vulnerability through to high irreplaceability and high vulnerbility (see fig 7a) The second approach as akin to C Plan,3.3 (GDARD, 2011) uses an embedded vulnerability constraint selection algoritm (using C Plan software) with rules to identify the least number of possible sites that include all irreplaceability factors where vulnerability is maximised. This technique produced 27 discrete sites, in their study region that for the most part, met the conditions outlined above (see fig 7b).
The findings of the comparison was that different information is highlighted with each approach. The first being more effective at addressing vulnerability priorities, but at the expense of not necessarily representing all species. The second approach was the opposite, in that it was more adept in selecting for species representation rather than vulnerability.
Figure 7 Examples of different approaches for irreplaceability and vulnerability mapping. 7.a: The simultaneous mapping of separate variables for irreplaceability and vulnerability. 7.b: The discrete sites from the combined variables algorithm approach (Taken from Lawler, et al 2003, p 1769)
Since this research is a detailed study of threats to terrestrial biodiversity in relation to pre- determined valuable conservation areas, the vulnerability of biodiversity, expressed as a continuous ranking is preferred, with products akin to the NSBA (2004) (Rouget, et al., 2004) and SKEP (2003) (Driver, et al., 2003) . Thus, an MCA approach similar to Wannenburgh, (2006), Li and Nigh (2011) and Vimal (2012) is used to develop the biodiversiy and threat/vulnerability composite layers expressed in raster cells, while the final integration and analysis borrows methodology and procedures from the NSBA (2004) SKEP (2003), Lawler (2003), and Vimal (2012).
An additional step in this reseach is the use of the AHP to rank the severity of threats for weighting in the MCA. As previously mentioned the AHP method (or MCA derivatives) has been used for conservation planning and reserve selection (Moffett & Sarkar, 2006a; Moffett, et al., 2006b; Geneletti & van Duren, 2008; Phua & Minowa, 2005), and eco-environmental evaluation (Ying, et al., 2007) but there are few examples where it has been used to analysie vulnerable areas/features or threats to biodiversity (O'Connor & Kuyler, 2009; Jaganath, 2009).
51 3.3.5 From theory to practice - the implementation of priorities
In academic literature there are few examples of theoretical SCP techniques being practically applied to conservation problems which effectively deliver on action (Moilanen, et al., 2009; Pierce, et al., 2005). In non-academic literature, spatial conservation techniques are increasingly being applied and integrated into biodiversity assessments and development planning. South Africa is a forerunner in this aspect (see SANBI, 2007) and research institutions, NGO’s and government sectors continue to work on challenges of application and mainstreaming into integrated spatial planning (Ogunronbi, 2014; Pierce, et al., 2005). While producing a spatial planning product for practical application in the MBR is not an objective of this research, it may be a possible outcome, if integrated priorities for vulnerability and threat are aligned with the MBR management plan, and biosphere zonation. This would provide a status-quo of the patterns of development and conservation value for the regional landscape, and a convenient spatial benchmark to inform conservation and sustainable management decisions for biosphere management.
There is a pressing need to mainstream biodiversity concerns into the policies and practices of municipal departments, and bridge the gap between land-use planning and conservation assessment, in light of the new national framework for Spatial Planning and Land Use Management Act (SPLUMA, 2013), which is a framework aimed at promoting uniformity and consistency in practice and decisions in the spatial planning arena (Ogunronbi, 2014; Pierce, et al., 2005). SPLUMA departs from previous spatial planning legislation in that municipalities will be solely responsible for dealing with land use applications and appeals, while complying with existing environmental legislation (EIA’s), that are authorised at provincial level (Ogunronbi, 2014). Therefore it is important for municipalities to incorporate provincial spatial biodiversity plans, such as CBA’s, into local planning products. Work is ongoing in several South African biosphere reserves to align their spatial planning at the municipal and regional level (Pool-Stanvliet, 2013).
A problem identified for implementation is the variation in capacity at municipal level to deal with biodiversity decisions, especially outside of statutory protected areas (Pierce, et al., 2005; Knight, et al., 2010). Therefore, in addition to planning alignment, spatial planning products should be user- friendly and user-useful for all stakeholders and role-players. With this in mind products should be comprehensible and should make use of cadastral boundaries used in the land-use planning sector to enable easier integration (Pierce, et al., 2005; Driver, et al., 2003).
By combining the conservation value of an area with an assessment of threats, it enables the identification of priorities for conservation action (Pressey, et al., 1996). Priorities of threat and conservation can be integrated as a single map product, with associated simple graphic outputs. This is achieved in this work by combining these variables in a bi-plot, (Pressey & Taffs, 2001; Driver, et al., 2003; Lawler, et al., 2003; Margules & Pressey, 2000) to identify combinations of conservation and threat value. Certain combinations, for example, planning units with high threat status and low conservation value, can be steered towards development, while units of high threat
52 and low conservation value, can be eliminated from conservation priorities (Rouget, et al., 2003). The combined map is also a useful indication of the vulnerability of important biodiversity. The integration of priorities for both threats and conservation would most likely provide a tool for reactionary responses to development in the context of the biosphere reserve, where conservation and sustainable development are encouraged for all zones to varying degrees, and where developments are scrutinized for sustainability and compliance (Department of Environmental Affairs, 2012).
3.4 Summary
Specific threats as well as rates of land transformation have been reviewed in the context of the Magaliesberg region. In general, conservation planning exists to protect and promote the persistence of biodiversity (Margules & Pressey, 2000). Many different systematic conservation planning tools are applied to a range of conservation outcomes, including threats analysys to achieve this. Methods include general additive models and scoring based approaches, to systematically assess threat and vulnerability such as spatial MCA. Not all threats can be represented spatially, but land-cover data and other remotely sensed data that capture concentrated proximal threats are useful for this purpose, and can be related to the biodiversity features of a region, to produce user-friendly integrated priority conservation and threat products and maps to aid spatial decision making for land-use planning.
53 Chapter 4 Materials and Methods
4.1 Setting the Scene
4.1.1 Overview of research objectives and methodology
This work is a conservation planning exercise to prioritize sites vulnerable to the threatening processes of land transformation and degradation that impact on terrestrial biodiversity. The intention is to identify current threats that can be represented spatially and rank their severity in relation to one another, and then spatially reference them to the important biodiversity conservation areas in the Magaliesberg Biosphere Reserve (MBR). The methodology design combines techniques and procedures from a number of applications and research in the spatial prioritization, site selection and conservation planning arena over the past 20 years. These are discussed in a review of related research (Section 3.3.4)
This methodology presents a systematic approach to analysing threats and referencing them to the terrestrial biodiversity landscape. By taking advantage of the availability and quality of existing biodiversity data, areas important for biodiversity conservation are determined using a weighted sum procedure – an overlay of five biodiversity conservation layers, that make up priority conservation value (PCV)(Section 4.2).
Following this, a similar but more detailed analysis is applied to develop the different threats to biodiversity (cost layers) and their relative importance (weighting) as biodiversity stressors (Section 4.3 and Appendices). In addition to seven individual threat- layers, a weighted sum is used to combine and rank them in a composite of priority biodiversity threats (PBT) (Section 4.5), based on a multi criteria analysis (MCA) of the threats (section 4.4). The Analytical Hierarchy Process (AHP) approach was applied to determine the relative severity ranking of each threat for weighting in the MCA (Section 4.5 and Appendices).
Finally, the average PCV and average PBT (vulnerability) is calculated for each planning unit, which is expressed as a continuous priority ranking across the Region of Interest ROI (section 4.5). Thresholds are applied to conservation and threat values in a scatter plot (Lawler, et al., 2003) to identify target clusters for conservation management. Figure 8 is a flow diagram showing the order of objectives and methodological steps, as described in this chapter.
54
1a. Biodiversity conservation layers 1b. PCV - Priority secondary Conservation biodiversity data Value extracted and equally weighted combined to suit ROI sum of 4 5. Planning Units conservation layers PCV and PBT scores 6. Synthesis of PCV and PBT are applied to - FINAL PRODUCTS 2. Biodiversity Threat different scale combining and cross- Layers extracted implementation tabulation of PCV and PTB from landcover and units. 50m pixel for graphic outputs and other secondary resoulution and spatial integration spatial data original farms. 4. PBT - Priority Biodiversity Threats AHP rankings and sub- criteria scores applied 3. AHP weighting and to weighted sum of 7 scoring of threats threat layers focus-group AHP and questionairre on selected threats
Figure 8: Flow of Methodological Steps – numbered according to research objectives,
blue represents methods used and black data outputs.
4.1.2 Data acquisition and limitations
This study makes use of processed secondary datasets in order to minimize duplication and to utilize existing science, by applying these data in another application to generate landscape-specific information.
Having said this, care should be taken to uphold the integrity of the data, its intended use, and the working scale of the data, especially when analysing disparate data in conjunction with each other, in order to derive results. This section describes the challenges and limitations experienced in acquiring and amalgamating the secondary spatial data.
The MBR straddles two provinces, three District Councils and six local municipalities, each using their own set of tools to develop spatial data for planning and conservation purposes at local, regional and district level. Therefore, local conservation data is developed independently per province, municipality or district, in response to specific objectives. This presents a number of challenges when trying to amalgamate multiple-source data for a regional or landscape scale analysis that transverses existing boundary limits, as the data may be mapped at different temporal and spatial scales and developed using different methodologies, rendering it incompatible as a combined unit.
It was found that spatial datasets, especially from local municipal departments are difficult to access and if available, are limited to the extent of the municipal or local boundary. Provincial and national data on the other hand were readily available through specific government departments, research institutes or web-access platforms. Provincial and national data are however generally mapped at a relatively coarser scale than some regional to local analysis.
55 Scale is a crucial element of spatial analysis because it determines the detail at which scientific investigation takes place to inform decision making at different spatial resolutions (Ferrier, 2002), for example, the delineation of biomes at a coarse scale may be a pre-cursor to finer scale study of vegetation types within a biome. The scale at which this analysis takes place is driven and limited by the scale of the datasets used (from 1:10 000 for land-use and to 1:500 000 for fire events in this study). Due to this range in scale being amalgamated into a 1: 50 000 equivalent it is suggested that the final product is a representation of the threat status of the region, and ground-truthing is recommended for fine scale conservation management.
Sometimes temporal differences in the latest available datasets are also a limiting factor for spatial research. GDARD, (2011) points out that in many cases, by the time processed data is made available for use in research, it is already outdated, especially where spatial changes over relatively short time-frames are of consequence, such as rapid rates of land-cover/ land –use change associated with urban sprawl, infrastructure / industrial development, and population migration patterns – all of which are relevant to the proposed Biosphere area. In this study provincial datasets of land-cover derived from images of different ages (2006 to 2009) were mosaiced, to develop some of the threat criteria.
Despite the heterogeneity of the data, and the actual datasets selected for this analysis, it is envisioned that any new data, with the required pre-processing, could be applied to this methodology design, and that as older data becomes obsolete, more current data can replace it. Data sources, mapping scales and acquisition dates for the data used in this research are mentioned in the data tables for each section following.
4.1.3 Software used and general pre-processing
The GIS software used is ESRI product Arc Info Version 10.0, with the Spatial Analyst Extension. The Excel Template AHPcalc Version 16-10-2012, developed by Klaus D. Goepel, Business Performance Management Singapore, was used to calculate AHP weightings for threat criteria.
All input spatial-layers were referenced to GCS WGS 84 decimal degree map units, and transformed where necessary to WGS 1984 datum. For area calculations layers were projected to UTM 35s co- ordinates, central meridian 27.0000 E.
All input data were clipped to the MBR extent, and buffered by 2km to accommodate technical edge effects associated with raster processing and to indicate a zone of influence for threatening processes (Rogers, et al., 2010).
56 4.2 Priority Biodiversity Conservation Value (PCV)
4.2.1 Overview of objective and method
The intention is to produce a discrete spatial dataset to reflect the biodiversity conservation value for all areas in the Magaliesberg Biosphere. Two criteria were applied to determine priority areas of high biodiversity value:
1) The potential of a site to be included in a conservation initiative (whether reservation, restoration or implementing sustainable practice).
2) The potential of a site to contain valuable and important biodiversity elements (as per the principles of representation and persistence (Margules & Pressey, 2000).
The four spatial data layers processed for inclusion in determining PCV met either one of these criteria. In the case of the former, the existing statutory protected area network and other proclaimed protected areas of the region were combined with the delineated proposed Biosphere zones, (the version finalised on 6 September 2012), as required by the Biosphere nomination process (UNESCO, 1996), and to possibly integrate with existing bioregional conservation plans (Pool- Stanvliet, 2013).
Two of the spatial layers met both criteria, the one being the “Remaining extent of threatened eco- systems” layer, of the National Threatened Ecosystems Assessment which forms part of the NBA (2011) and the other, the combined (merged) Provincial ‘Critical Biodiversity Areas’ (CBA’s) layers from the biodiversity assessments of Gauteng Province C Plan Version 3.3, (2011) and North West Province Biodiversity Conservation Assessment, (2008).
The forth input, satisfying the second criteria, included a proportional representation of 15 vegetation types of South Africa (Mucina & Rutherford, 2010 ) that occur in the region as a spatial surrogate for Biodiversity (Ferrier, 2002).
The attributes of each of the four layers were scored to reflect their importance for either criterion. Each layer was apportioned equal weight, and run through a weighted sum.
57 4.2.2 Data characteristics
Table 5, summarises the spatial data used as input for PCV.
Table 5 - Data for PCV map layer.
Data type and Criteria 1 Name Source Custodian spatial resolution Biosphere zonation, 6-09-12, 28- Biosphere Nomination Contour Project Layer 1 Polygon - Vector 08-12, 26-06-12 Documents Managers C.C. Cradle of Humankind and Polygon – Vector GCRO GDARD Gauteng Conservancies 1: 50 000 Formal Protected Areas BGIS website SANBI Polygon -Vector Criteria 2 NWCBA North West Critical Polygon -Vector Layer 2 DEDECT DEDECT Biodiversity Areas, 2008 1: 50 000 Polygon -Vector Gauteng C Plan 3.3, 2011 BGIS website GDARD 1: 40 000 Threatened eco-systems Polygon -Vector Layer 3 BGIS website NBA 2011, SANBI Remaining Extent, 2011 1 : 250 000 Vegetation of S.A. CD Polygon -Vector Layer 4 Vegetation Types, 2006 SANBI Set, 2010 1: 250 000
4.2.3 GIS methods and weightings for layers
Decisions on scoring and weighting of layers is a critical aspect of GIS - MCA, as the results of the analysis depend on the decision maker defining values relative to the decision problem (Saaty, 2005). In essence the process involved in GIS-MCA includes the manipulation of geographic data (input maps), and the decision makers preferences (value judgement), according to specified decision rules, which results in spatially defined decision alternatives (output maps) (Malczewski, 2010). The following section describes each layer in detail and the GIS processing and the derivation of classes and equivalent numerical scores for each.
4.2.4 Layer 1 - Biosphere zones and protected areas
4.2.4.1 Data
The biosphere zonation shapefiles show an officially protected core zone (the MPE) and defined buffer and transitional zones based on endorsements received from Government and land-owners for inclusion into the Biosphere. The zone delineation continues to be adjusted to meet UNESCO requirements (see Section 2.2.3). However, the original preferred zonation was based on incorporating conservation priorities and excluding development pressures (using seven national and provincial data sets in an overlay procedure), as far as possible within non-fragmented easily identifiable boundaries, such as topographical and natural features, road infrastructure and 58 administrative boundaries). More details can be found in the Magaliesberg Biosphere Situational Report, 2011 (Department of Environmental Affairs, 2012).
This study uses the September 2012 zonation which includes the category “Future Expansion” - the Gauteng portion on the Nomination Map, even though the province had been excluded from the 2012 Nomination. It was decided for the purposes of this study to include the Gauteng core (COH) and buffer sites that fell within the future expansion zone, as these will be vital for the success of the biosphere initiative.
4.2.4.2 GIS processing and weighting
Required features from different inputs (Gauteng buffer zones, conservancies and protected areas) were selected and merged and then unioned to the 6-09-2012 Zonation shapefile. A ‘zone’ field was added to the zonation shape-file with categories (core, buffer, transition, future expansion), and Gauteng features assigned to relevant categories and a ROI-buffer category was added. Individual polygons were dissolved to the zone field so only the five multipart categories remained. Cleaning- up of data included edge matching and vertex snapping of polygons to remove gaps of ‘no-data’ and overlapping polygons - which were numerous. The shapefile was rasterised to a 10m grid and the tools boundary clean and expand were applied to remove persistent gaps before aggregating (median value) by a cell factor of 5. The 10m raster was then resampled using the aggregate output as the cell size raster for a 50m x 50m (0.25ha) cell size.
Because Protected Areas (PA’s) are accounted for in the zonation input as well as the Provincial CBA’s they would essentially be scored twice, and receive an overwhelming priority in a weighted overlay. Therefore PA’s were not assigned the maximum score possible in this layer. In order to highlight valuable biodiversity outside of PA’s, scores assigned to conservancies and biosphere buffer zones with a potential for conservation were almost equivalent to PA scores. A new numeric field ‘score’ was added to enable weighted sum and different scoring scenarios to be run with ease. Scores (table 6) range from 1-9 (consistent with the AHP scoring discussed in section 4.4), 9 being the highest potential for conservation, and 1 indicating the least potential. It is assumed that all areas in a Biosphere Reserve and immediate surroundings would be encouraged to promote sustainability and environmental awareness, thus have conservation potential, hence a minimum of 1 not 0.
Table 6 - Scores allocated to the 'zone' field
ZONE Core Buffer Transition Future Exp ROI Buffer
SCORE 7 6 3 2 1
The final conservation zonation input layer is shown in figure 9.
59
Figure 9: Layer 1 Biosphere zones (2012 Nomination)
4.2.5 Layer 2 – Provincial biodiversity assessments
4.2.5.1 Data
Sophisticated complementarity based software such as Marxan and C Plan have been employed for provincial level Biodiversity assessments in Gauteng and the North West Province of South Africa. However the criteria and methodologies used as a basis for each assessment differed, with very different maps resulting (a case in point mentioned in the data limitations section). The major differences between the two datasets for purposes of this study is that a cost layer was not used in the NW assessment and neither were fine filter species data (point locality) used for modelling. A brief summary of each assessment is explained below, followed by an explanation of how the data were combined. The reader is referred to the comprehensive Technical reports for both assessments for more detail.
1) Gauteng Province C Plan Version3.3 Released October 2011
Areas of importance for biodiversity conservation in the Province are presented in hexagonal planning units of 100 hectares, or part thereof. Four categories are distinguished Critical Biodiversity Area (CBA) Irreplaceable sites, CBA Important sites, Ecological support Areas (ESA) and Protected Areas. Certain built-up areas have been excluded.
C Plan software was used to create CBA’s, irreplaceable and important, using an iterative summed irreplaceability approach to select suitable units until all biodiversity targets were reached that occupied the least land area possible.
60 Twelve input layers (factors) were considered, each derived from an involved process that included habitat, species and distribution models and protection targets for red and orange list plants and animals as well as selected indicator species. Fine scale point data were used to develop these biodiversity features. The Environmental and ecological factors considered were near pristine pan clusters, quaternary catchments and a carbon sequestration layer. For Bioclimatic zones environmental parameters overlaid to obtain bioclimatic variables, which were run through a Marxan prioritization to reach a target 10% of the original extent of each zone. Environmental envelopes for 11 identified vegetation communities were used to develop conservation targets for the primary vegetation layer. Lastly a cost layer of threats was determined from Land cover data of built-up and transformed areas, areas under some form of protection or development restriction, and buffered areas for ecological processes and corridors.
The ESA’s represent a number of landscape features used as spatial surrogates to represent areas essential for the maintenance and generation of biodiversity, as per the principle of persistence (Margules & Pressey, 2000). Features used include dolomite, rivers, wetlands, pans, ridges, corridors for climate change and species migration and areas under some form of development restriction (GDARD, 2011).
2) North West Province Critical Biodiversity Areas (NWCBA) (2008)
The data available for the North West Province originates from a wider Provincial Biodiversity Assessment.
Terrestrial and Aquatic features in the landscape that are critical for retaining biodiversity and supporting Ecosystem services and functioning are categorised into CBA 1 (intact species and ecosystems deemed irreplaceable in order to maintain biodiversity targets), CBA 2 (mostly intact and undisturbed ecosystems, with limited losses to biodiversity that do not compromise the meeting of targets), and ESA‘s (moderate disturbance while maintaining ecosystem functionality).
The biodiversity features applied were assigned to a CBA category. Then CBA categories were summarized and unioned for the final terrestrial and aquatic CBA maps. Features included a 1:50 000 scale combined Provincial and National vegetation map to define patches larger than 3ha and 5ha of intact critically endangered ecosystems, and patches of endemic flora < than 10 ha. Expert mapped biodiversity features such as suitable habitat for indicator species and rare or endemic species, important remaining habitat or ecological support and corridor areas. Only areas < 10 000 ha were considered accurate enough to map. Wetlands, pans and priority sub-quaternary catchments made up the aquatic features. Previously identified protected area development corridors (MPE), development nodes and corridor linkages to maintain connectivity in the landscape. Other important terrestrial landscape features were hills and ridges, scenic landscapes and springs. The ESA features include dolomite and groundwater recharge areas, wetland and PA buffers.
61 4.2.5.2 GIS Processing and weighting
Some of the NWCBA features are mapped at a relatively coarse scale intended for broad provincial assessment (e.g. the whole of the Magaliesberg is classed as CBA 2 corridor). This scale is too coarse for this analysis, as well as relative to the Gauteng C Plan 3.3 dataset. Therefore some adjustments were made to the NWCBA field inputs to bring it more in line with the C Plan 3.3. The following coarse scale features were removed from the combined terrestrial and aquatic NWCBA map – CBA Corridors, CBA features, CBA nodes and ESA PA. The finer scale features retained were – CBA corridor links, CBA SA-veg, CBA expert, CBA PA and ESA wetlands and dolomite, resulting in a map more commensurate to the C Plan 3.3 map. The retained features were ‘dissolved’ into ‘multipart’ features, resulting in 55 polygons with unique combinations of categories. Since areas in the NWCBA could belong to more than one category (e.g. CBA 1 and ESA 1) each polygon needed to be assigned to one category. A new field MB_CBA combined was created and populated into CBA 1, CBA 2 , ESA and PA, (in line with C Plan classes) and each polygon was assigned one value, using the highest CBA value given to these fields in the original map. Three exceptions are that ESA 1 and 2, were combined to ESA, and the Expert and Corridor Links fields were assigned an ESA value if not overlapped by any other biodiversity feature, in order to remove incongruent large CBA 1 polygons. The CBA PA’s were assigned to the PA class only if no other feature overlapped it.
No adjustment was made to the C Plan 3.3 data except to dissolve to the C Plan_Area field multipart, add the MB_CBA combined field as per the NWCBA, and populate it with the equivalent C Plan_Area values. The two layers were projected and merged to make a single layer. This was converted to raster and resampled to a 50m cell size, and a numeric score field was added.
The scores for this layer were assigned as follows (0-9) 9 being the highest biodiversity value and 0 representing all areas outside of the CBA categories (blank areas). Protected areas were given the same score as CBA 1 irreplaceable areas as they were categorised as CBA 1 in the NWCBA assessment. The scores and final CBA map follow.
Table 7 - Scoring applied to Critical Biodiversity Areas
AREA TYPE CBA -1 (Irreplaceable) CBA -2(Important) ESA PA
SCORE 9 7 6 9
62
Figure 10: Layer 2 - Critical Biodiversity Areas
4.2.6 Layer 3 – Remaining extent of threatened ecosystems, 2011
4.2.6.1 Data
This is an ecosystem indicator developed in the NSBA 2004 and carried through to the NBA 2011 and is used to assess the threat status of Ecosystem types. Ecosystem types are delineated chiefly from vegetation types but also supplemented with national forest types and finer scale provincial systematic biodiversity plans (Gauteng C Plan Ver. 2 is relevant here). Threatened ecosystem types are classified into critically endangered (CE), endangered (E) and vulnerable (V), depending on the proportion of the original extent of each ecosystem type that remains in good ecological condition, and its association with threatened plant species, relative to three biodiversity thresholds. Ecological condition refers to the intactness of the structure, functioning and condition of ecosystems, and can range from natural, near-natural through to severely modified and is measured chiefly using land- cover data, into three categories – good, fair and poor (Driver, et al., 2012). Table 8, shows the classification of Threatened eco-system types and the biodiversity thresholds applied.
Table 8 - Criteria used to identify Threatened Ecosystems, and the thresholds applied to determine threat status.
STATUS Critically Endangered Endangered Vulnerable CRITERION * Good condition habitat Good condition habitat Ecological Condition / Good condition habitat remaining ≤ biodiversity remaining ≤ BT + 15% (so Irreversible habitat loss remaining ≤ 60 % target, (typically 20%) 35% if BT is 20%) Systematic Biodiversity Very high irreplaceability Very high irreplaceability Very high irreplaceability Plans and high threat and medium threat and low threat *Only criteria applicable to the MBR and buffer have been included.
63 4.2.6.2 GIS processing and weighting
The intention of using this layer is to reflect the NBA 2011 and National list of Threatened Ecosystems (NEMBA, 2011), that occur in the Biosphere. The threatened status of the ecosystems types is reflected - Critically Endangered, Endangered and Vulnerable ecosystems, rather than the ecosystem type itself.
The original shape-file was clipped and dissolved to the Threatened Ecosystem ‘Status’ field, resulting in 3 multipart categories – Critically Endangered, Endangered and Vulnerable areas. The file was projected, rasterized and resampled as per the previous layers.
The following scores (table 9) were applied to categories, taking into consideration the relatively small extent of these remaining areas and their importance to meet biodiversity targets and retain remaining habitat in good ecological condition. The scoring was assigned as described for the CBA layer, (0-9).
Table 9 - Scoring applied to Remaining Extent of Thretened Ecosystems
Vulnerable STATUS Critically Endangered Endangered 7 SCORE 9 8
Figure 11: Layer 3 - Remaining extent of Threatened Ecosystems
4.2.7 Layer 4 – Vegetation types of South Africa
The reasoning for this layer is firstly to highlight vegetation types (surrogates for habitat and eco- systems) that may not be threatened but none the less provides important heterogeneous habitats for biodiversity in the Biosphere. Secondly, to recognise the proportion of the vegetation type that is
64 represented in the ROI in relation to the remaining extent of the vegetation type overall, as an important factor in its conservation.
4.2.7.1 Data
The aim of VEGMAP is to describe and map the vegetation of South Africa using a number of criteria to distinguish heterogeneous vegetation patches (vegetation types) across the landscape. Floristic composition, bio-geographical patterns, physio-geographical and environmental descriptors are the main criteria used. The basic mapped units reflect the spatial complexity and macro-ecology of 435 vegetation types in South Africa (Mucina & Rutherford, 2010 ).
4.2.7.2 GIS processing and weighting
The shape-file was clipped and dissolved to the vegetation –type name, resulting in 15 multipart veg- types that occur in the region. An area field was added and extents in km² were calculated. Then the file was rasterized and resampled to 50m.
Three attributes were applied to score the 15 veg-types- The total area of the veg-type, its transformed percentage (%), and the percentage of the veg-type represented in the Biosphere. The protection status of the veg-type was not included as it mostly correlates with the transformation status in all 15 cases (i.e. A high transformation % for the veg type affords it a high protection status). In Excel 2007, the area transformed of each vegetation type was subtracted from its overall extent to determine the remaining overall extent of the vegetation type in km². Quartiles of the remaining extents guided the allocation of scores for each veg-type from 1 – 9 (A score of 1 for remaining extent ≥ 5000 km² and 9 for ≤ 250 km²). The process was repeated for the percentage of each veg type represented in the ROI (A score of 1 for ≤ 2% representation and a score of 9 for ≥ 40% representation). The two scores per vegetation-type were averaged to a single score (table 10). The final vegetation-type map is shown in (fig.3 p.14).
Table 10 - Vegetation-Types found in the Magaliesberg Biosphere and their allocated weights
Combined Scores for Overall remaining extent and Vegetation Types % represented in the Region of Interest Zeerust Thornveld 2 Marikana Thornveld 6 Norite Koppies Bushveld 6 Moot Plains Bushveld 6 Gold Reef Mountain Bushveld 6 Gauteng Shale Mountain Bushveld 5 Andesite Mountain Bushveld 5 Soweto Highveld Grassland 1 Rand Highveld Grassland 1 Egoli Granite Grassland 7 Waterberg Magaliesberg Summit Sourveld 5 Northern Afrotemperate Forests 6 Highveld Salt Pans 3 Eastern Temperate Freshwater Wetlands 4 65 4.2.8 Overlay analysis – Priority conservation value
None of the overlay tools in Arc Info 10 considers ‘no-data’ for analysis, therefore it was necessary to create cells in the ROI for layers where the criteria do not cover the full extent of the ROI (CBA and Threatened Ecosystems layers). The categories in these two input layers were given preference when combined with a 50m x 50m cell size raster mask for the ROI, with all cells valued at 0. This eliminated blank areas and provided a value for all cells covering the ROI in these two layers. Now, the integer values for each cell in each of the four input layers are commensurate and therefore the layers can simply be combined in an overlay procedure.
The weighted sum tool was the chosen overlay procedure as it maintains the model resolution by not rescaling the scores of cells back to the evaluation scale (normalizing or standardising the scores). Maintaining the model resolution is useful in cases where it is desirable to identify the few top ranking locations, or a specified number of priority sites (ESRI, Inc., 2010). So, in this study the input cells are scored a minimum of 0 and maximum of 9, and there are four layers apportioned equal weight (1), thus the output cells will theoretically range from 0 - 36, and not 0-1 as is the case when normalising the cell values. The bigger range allows for a finer classification of top priority sites.
The results of the weighted sum produced a final PCV map with a continuous value range from 2-31 (fig. 12), which was categorised into 4 classes for tabulations and graphic outputs. This stage marks the completion of objective 1 in the research.
Figure 12: Priority Conservation Value (PCV) – continuous ranking weighted –sum
An A4 version of this map can be found in Appendix 5
66 4.3 Priority Biodiversity Threats (PBT)
4.3.1 Overview of objective and method
The objectives are to produce spatial layers to reflect each selected biodiversity threat factor in the MBR. Then, the relative intensity that each threat poses to biodiversity is determined using the Analytical Hierarchy Process (AHP) method, to produce a ranked spatial composite of threats to biodiversity.
This part of the methodology involved the integration of two distinct steps. First, the development of spatial layers as surrogates to represent threats to terrestrial biodiversity (Fuller, et al., 2010; Reyers, 2004; Lawler, et al., 2003), and second the application of the AHP to quantify (weight) these threats, in relation to one another in the MBR context, to facilitate a spatial multi-criteria-analysis MCA (Moffett, et al., 2006b; Jaganath, 2009).
In order to identify potential threats in the Magaliesberg region, the following aspects were considered in the initial selection of threat factors:
1. A literature review of threats to terrestrial biodiversity and the socio-economic, environmental and ecological characteristics of the region. 2. The availability and resolution of spatial datasets influenced the initial choice for selection, with the obvious requirement that the threat factors be conducive to spatial representation. 3. The most cited threats to terrestrial biodiversity include the destruction and degradation of habitat. These two broad terms were applied as basic criteria in selecting proximal threat factors. 4. The threat factors had to be appropriate for a regional context (hence proximal threats), and also relevant to the study area. 5. The threat factors had to be current threats as this is a baseline study which looks at the spatial relationships of current stressors on the current state of biodiversity.
Seven selected threat factors were introduced at the AHP threats presentation and included as criteria in the AHP threats questionnaire. These were: Urbanisation, Cultivation, Mining, Transformed open space (degradation), Fragmentation, Fire events (altered fire ecology) and Alien Plant invasion.
These threat factors were developed into GIS threat layers, the majority of which were derived from a base of Classified Land-Cover (2009 Gauteng Land-Cover dataset, and North-West Province 2006 Land-Cover project). For some of these layers the base land-cover was supplemented with other polygon feature data that were more detailed, or more current. A distance analysis to National and Major roads (Li & Nigh, 2011) , and transformed areas (Nega, et al., 2010) was done as a spatial proxy for fragmentation. The fire events layer was developed from MODIS burned area product,
67 time-series data (Giglio, et al., 2009), and the Invasive Alien Plant spatial data (Kotzé, et al., 2010) was included unaltered, as a layer.
The AHP was applied to weight, relative to one-another, the seven threat criteria/factors mentioned above. The quantification of threats using the AHP method requires the judgement of expert participants to make pair-wise comparisons for all threat factors (Saaty, 2005).This was achieved by an AHP Questionnaire, the results of which determined the weight and rank for each threat relative to one another as stressors to biodiversity in the region. In addition, a second part to the questionnaire informed the scoring applied to categories of intensity within each threat factor (sub- classes), to reflect the range in intensity of threat posed by each factor. For example, sub-classes of alien plant invasion were based on the density of invasion, and the impact or intensity of invasion at 1% versus 45%, was categorised and scored.
The final products from this objective are seven discrete or continuous scored threat maps, and a MCA, AHP- weighted composite map of PBT. Products can each be analysed in relation to the Priority Biodiversity Conservation areas of the Magaliesberg biosphere.
4.3.2 Data characteristics
Table 11 summarises the spatial data used as input for the biodiversity threats layers. More detail is provided in relevant sections below.
Table 11 - Data used for Biodiversity Threat Layers
Threat Data type and spatial Name and Date Source Custodian Criteria resolution 1; 2; 3; 4; 5 2009 Gauteng Land-cover GCRO GTI (pty) LTD Classified - Raster 10m x 10m 1; 2; 3; 4; 5 2006 North West Land-cover DEDECT DEDECT Classified - Raster 10m x 10m Old abandoned agricultural 2 MPE EMF DEDECT Polygon –Vector fields 3 NW Mines and Quarries MPE EMF DEDECT Polygon – Vector 3 GP Mine Residue Areas GCRO GDARD Polygon – Vector 4 NW & GP GULLIES ARC -ISCW ARC-ISCW Polygon – Vectors NAIPS landscape and Continuous - Rasters 250m x 7 ARC-ISCW ARC-ISCW Riparian 250m MODIS Burned Area Product University of 6 WAMIS Discrete raster 463m x 463m - Burn Date (2000 – 2011) Maryland National and Provincial 5 DEDECT DEDECT Line – Vector Roads (Major) NW 5 Provincial Roads (Major) GP GCRO GDARD Line – Vector
Threat Criteria: 1-urbanisation; 2-cultivation; 3-mining; 4-transformed open space; 5-fragmentation; 6-fire events; 7-Invasive alien plants
68 4.3.3 Biodiversity threats layers from classified land-cover/ land-use data
4.3.3.1 Description of Provincial Land Cover Datasets
Gauteng
This is a detailed land-cover dataset that primarily represents land-cover and land use patterns circa 2009, of the current Gauteng Province, including an overall 2km buffer zone around all boundaries. The land-cover data was generated from 2009 - 2008, 10m resolution multispectral SPOT 5 satellite imagery and fine-scale 2009 aerial photography. The imagery has been processed using digital desk-top mapping procedures, into a thematic 10m x 10m raster dataset of 41 information classes. The land-cover is theoretically suitable for GIS modelling applications of (approximately) 1:30,000 scale or coarser applications, and has a generalised minimum feature mapping unit of 0.05 ha. No statistical map accuracy validation of the 2009 Gauteng Land-Cover dataset was undertaken (GeoTerraImage, 2009).
North West
A detailed, North West Province land-cover dataset making up 40 information classes of land-cover and land-use, derived from late 2005/2006, single-date, 10m resolution SPOT 5 imagery, and Landsat Normalised Difference Vegetation Index (NDVI) comparisons for interpretation of erosion and degraded classes. The land-cover data is in raster 10m x 10m pixel format and is suitable for 1:40 000 scale (or coarser) applications, based on a theoretical 0.1 ha minimum mapping unit for spectrally discernible landscape features. The statistical mapping accuracy for the full legend format is 70.95% (69.35% – 72.56% at 90 percent confidence limits), with a Kappa Index of 69.38. This increased to an accuracy of 80.37% (Kappa Index of 78.57) for the Level-1 legend format of the same dataset (GeoTerraImage, 2008), which is equivalent to the legend format used in this study.
4.3.3.2 Land-Cover as the underlying spatial surrogate for threats
The data selection process began with an initial review of the South African land-cover classes and their hierarchical structure and associations (Thompson, 1996). Both the 2009 Gauteng Land-Cover dataset and the 2006 North-West Province Land-Cover dataset have a legend structure that is consistent with current South African land-cover mapping standards and the internationally accepted FAO Land-Cover Classification System. However, the classes for each dataset were chosen to suit the specific requirements of the respective projects and thus differed from one another (Thompson, pers.comm. 2012). This was particularly limiting where one set used level 1 classification for some land-use categories (e.g. Urban all) and the other, level 2 or 3 classification for the same category (e.g. urban commercial, industrial, roads & tracks) making it impossible to split the combined classes into more specific detail.
Nonetheless, the class descriptions were reviewed (39 for Gauteng and 36 for NW in the ROI) with a view to emphasising biodiversity threat potential in class combinations, in order to aggregate like classes where possible, into a combined level 1 equivalent classification (See Appendix 1).
69 Finally, a basic aggregation of six classes could be used as the underlying input for five of the threat layers: (Urbanisation, Cultivation, Mining, Fragmentation, and Transformed open land). Some of these classes could be re-divided into sub-classes while keeping true to the original classifications. This made it possible to define the range or severity of threat posed to biodiversity, within each threat class. For example, ‘Urbanisation’ could be separated into sub-classes - ‘urban areas’ and ‘smallholdings’ - allowing for a distinction to be made between the severity of threat posed to biodiversity by what is assumed by definition, to be high to medium density urban areas, as opposed to low density urban areas.
Further to this, in some of the threat layers sub-classes were created by adding separate thematic datasets to supplement the underlying land cover classes. For example, vector feature data of erosion gullies were used to create a sub-class of erosion degradation within the transformed open- land layer.
The distinction of sub-classes, and the scoring applied to each, was informed by experts in the AHP threats questionnaire, as described further on in Section 4.4.
4.3.3.3 GIS processing
The ArcGIS Spatial analyst extension and Modelbuilder were used to develop threat layers. Modelbuilder is a convenient user interface and a record of the process – the data inputs, tools used and steps taken in developing layers. It is possible to re-run specific tools, or the entire model with different data inputs, or with different values for existing inputs (ESRI, Inc., 2010).
The land cover datasets for both provinces were clipped and merged with their combined attributes, to result in a single 18 class classified 10m x 10m raster of the MBR with 2km buffer. This was resampled, nearest neighbour to a ± 50m x 50m pixel size (0.25 hectare MMU), which partially takes into account the coarser resolution of some of the other data sets used in the analysis, without becoming too generalised. Additional datasets used are detailed with the layer they pertain too. Weighting applied to each threat layer was informed by the AHP, and ranking of intensity within each sub-class, by associated questionnaire.
4.3.3.4 Urbanization Layer
Datasets used: Classified Land cover For the Urbanisation threats layer all other classes not pertaining to urban land-use were set to ‘0’, while the original codes assigned to each urban class were retained. A new field ‘sub-criteria scores’ was added to rank each urban class into a sub-class based on its’ intensity of threat. There was limited potential to sub-categorize urban classes due to there being no overlap in the original provincial class descriptions (see appendix 1), therefore only two sub-classes resulted: Urban areas and rural centres (medium and high density) Smallholdings, scattered rural (low density).
70
Figure 13: Urbanization Threats Layer
4.3.3.5 Cultivation Layer
Datasets used: Classified Land cover
As above, there was limited potential to derive sub-class detail from this layer due to incongruous class descriptions (see appendix 1). All classes not pertaining to cultivation were set to ‘0’, and all remaining cultivation classes were grouped into sub-classes. A new field ‘sub-criteria scores’ was added to rank each cultivation sub-class based on its’ intensity of threat: Commercial cultivation. Small-scale cultivation.
Figure 14: Cultivation Threats Layer
71 4.3.3.6 Mining Layer
Datasets used: Classified Land cover; Gauteng Mine residue areas; MPE mines and quarries
In addition to the land-cover mining classes, a polygon vector of Gauteng mine-residue areas (GDARD, 2011) and one of mines and quarries found in the MPE Environmental Management Framework and Plan, ( DACE, 2007) were clipped, merged together and rasterised and mosaiced to the land-cover mining layer (fig. 15). Due to the volumes and storage of waste in mining operations (see chapter 3.3) it was decided to add all other waste processing and storage land-use categories to the mining layer. The final sub-classes were: Mine residue areas Subsurface mines and infrastructure. Extraction pits, tailings, landfill, sewerage
Figure 15: Mining Threats Layer
4.3.3.7 Transformed Open Land Layer
Datasets used: Mosaiced Land cover; Gully Erosion polygon data for SA; MPE old agricultural fields polygon data.
This layer (fig 16), represents all non-built up and non-cultivated areas that have been transformed or degraded to varying degrees. Land-cover classes include eroded and degraded land, old-lands and recreational and urban open-spaces (see appendix A). In addition shapefiles of provincial gully erosion (Mararakanye & Le Roux, 2011) an old-agricultural fields of the MPE ( DACE, 2007) were clipped, merged and rasterised and then classified and mosaiced to the transformed open-land layer.
Erosion Gully data
72 Gully erosion status per province in South Africa was mapped using high resolution SPOT 5 satellite imagery at a scale of 1:10 000 (panchromatic sharpened images at 2.5 m resolution and multispectral bands merged with panchromatic bands resulting in 5m resolution). An evaluation of automated and semi-automated classification algorithms revealed the spectral complexity of gullies (e.g. some are vegetated, others bare) and the difficulties presented when trying to distinguish them from surrounding areas and other geomorphic features (e.g. river valleys). Hence, to achieve higher accuracies, mapping of all visible gully erosion per province was derived from manual digitizing of imagery (acquisition date 2006 to 2008) by experts, followed by verification through field observation at selected study sites (Mararakanye & Le Roux, 2011).
After the three datasets were combined, a field “sub-criteria scores” was added. Sub-classes included:
Erosion gullies. Degraded and bare land (excluding natural rock). .Plantations, recreational grass, urban open space. Old-lands/ abandoned agricultural fields.
Figure 16: Transformed open land Threats Layer
4.3.3.8 Fragmentation Layer
Datasets used: Mosaiced Land cover; national and provincial main roads
A large proportion of the study area is represented by the combined ‘natural’ classes representing natural or semi-natural land-covers (GeoTerraImage, 2009). These areas needed to be incorporated in the threat analysis as they constitute variously fragmented habitat and as such may be regarded as a threat to biodiversity.
73 A simple Euclidean distance calculation determined the proximity of each pixel to the national and main roads, and all built-up areas. It was then realised that because the land-cover included large, main- road surfaces, the roads data were not necessary, unless national and provincial roads would be weighted (Rouget, et al., 2004), which was decided against. Thus the land-cover distance calculation resulted in a continuous proximity surface for all open-space which formed the fragmentation layer (fig 17). Open-spaces closer to roads or built-up areas were considered more vulnerable, and thus more of a threat to biodiversity than areas further away. The continuous distance surface also gives an indication of the size of fragments and distance between fragments, and highlights any large intact areas relatively far away from main roads and habitation. Although size and distance between fragments were not included in this analysis, pixels in the large intact fragments, distal to roads and built up areas were given a zero sub-class score for fragmentation threat, while pixels in close proximity to the roads and built-up areas were allocated the highest threat scores in the ‘sub-criteria scores’ field.
The 5 class breaks in the continuous proximity surface, based on distance to roads and built-up areas were defined by the ArcGIS geometrical interval classification method, as the natural breaks were not suitable as sub-classes:
0 -90m 91m – 240m 241m – 570m 571m – 1340m 1340m – 3100m
Figure 17: Fragmentation Threat Layer
74 4.3.4 Biodiversity threat layers from other data sources
4.3.4.1 Alien Invasive Plants Layer
The intention of the Invasive Alien Plant (IAP) threat layer is to map the spatial extent and average density/hectare of the 10 woody IAP species recorded for the ROI on a continuous surface. Areas with lower densities of invasion would be allocated lower threat score than those with a high density of invasion.
Datasets used: National Alien Invasive Plants survey (NAIPS) spatial datasets (Kotzé, et al., 2010)
This national project establishes the range and abundance of 30 species groups of well established woody IAP species at the quaternary catchment level, by sampling on an environmentally variable gradient (soil, terrain and climate) the potential occurrence of alien species for South Africa, Lesotho and Swaziland (Kotzé, et al., 2010a). Woody species were the most aerially observable, and the most widespread and abundant of these were captured nationally, using a combination of remote sensing and field survey data (Kotzé, et al., 2010). A separate riparian sample layer was allocated for the strong association between IAP species and riparian zones. This layer was buffered with 500 metres on either side, with the intention for users to reduce this if necessary depending on local conditions. The prescribed working scale for interpreting the NAIPS data is at quaternary catchment level (Kotzé, et al., 2010).
The NAIPS spatial data used in this analysis are:
1) The Landscape Abundance (density) Layer for South Africa (niapslnrsaab2), Average Density (AVDENS) attribute at a 250m x 250m cell resolution. “The Average density (%) is calculated by dividing the accumulated density for all species by the total number of all sample points for a specific unique condition within a quaternary catchment” (Kotzé, et al., 2010). 2) The riparian abundance (density) layer for South Africa, Lesotho and Swaziland (niapsriparab2). As above, the average density (AVEDENS) attribute (250m x 250m cell resolution) along a 1km wide riparian buffer (Kotzé, et al., 2010).
GIS Processing
Both the landscape average density and the riparian average density rasters were extracted to the 50m MB mask raster. Each was then classified to four natural breaks expressed as density % cover. These ranges (0-46% invasion density for landscape and 0-93% invasion density for riparian), were then reclassified based on their natural breaks to four classes for both rasters. The reclassified rasters were merged using the maximum operator to get the maximum value for any overlapping cells (fig 18). This was then projected to the 50m raster mask, for an output cell size of 50m and a ‘sub-criteria scores’ field was added.
75
Figure 18: Merged terrestrial and Riparian invasion rasters, showing equivalent classes but different density and range values.
Two limitations of scale can be highlighted here. Firstly by projecting 250m to 50m, the data remains coarse as the value of each 250m cell is replicated in 25 x 50m cells. Secondly, the riparian invasion is up to 1500m wide in parts, (due to the blocky representation of linear features in raster format). An attempt was made to ‘thin’ the linear feature by reducing the number of cells representing its width to 2 cells wide (500m), but this returned a Boolean result, thereby loosing the classification status of invasion density. Hence the area extent of riparian invasion is exaggerated along some stretches.
The four classes representing average density of invasion for 11 woody species are:
Landscape Riparian
0 – 1.65% 0 – 9% 1.65 – 8.25% 9 – 29% 8.25 – 20% 29 – 55% 20 – 46% 55 – 93%
76
Figure 19: Alien Plant Invasion Threat Layer
4.3.4.2 Fire Events Layer
The purpose of the fire events threat layer is to indicate the absence and frequency of occurrence of fire throughout the ROI, over an 11 year period from June 2000 to September 2011. Then, categories of fire absence or frequency are scored according to the threat posed to the biome type in which the fire events occurred, as fire ecology is closely linked to vegetation (Smith, et al., 2012; Uys, et al., 2004). The layer was achieved through multiple overlays of the data set described below.
Datasets used: MODIS collection 6 burned-area product (WAMIS, 2008-2013), VEGMAP, vegetation types of South Africa (Mucina & Rutherford, 2010 ).
The Moderate Resolution Imaging Spectroradiometer ( MODIS) Collection 6 burned area product is derived from an automated algorithm for mapping post-fire burned areas using 500-m MODIS imagery coupled with 1-km MODIS active fire observations (Giglio, et al., 2009). The combined use of active-fire and reflectance data enables the algorithm to adapt regionally over a wide range of pre- and post-burn conditions and across multiple ecosystems. The researchers estimate the minimum detectable burn size for reliable detection by their algorithm to be on the order of 120 ha (Giglio, et al., 2009).
The MCD64A1(Louis Giglio) Burned Area Product is a set of monthly 463m x 463m pixel size gridded tiles containing per-pixel burning and quality information, and tile-level metadata from June 2000 to September 2011 (WAMIS, 2008-2013). The “burn date” layer defines each pixel by the approximate Julian day of burning from eight days before the beginning of a month to eight days
77 after the end of a month, it also indicates unburned areas, snow, water, or lack of data (Boschetti, et al., 2009).
Although this time-series data is not an official MODIS product, (Frost, pers. comm., 2012), it is distributed in the standard MODIS land format (HDF4). The dataset: MCD64_h20_v11 2000-2011 (burn-date layer) (WAMIS, 2008-2013), was used in this study. The characteristics of VEGMAP are detailed in Section 4.2.7.
GIS processing
Approximately 135 (12 years, ±12 per year) ‘burn-date’ raster layers were imported at a processing extent set to a ROI rectangle polygon, and transformed to GCS-WGS1984 projection and datum, with output rasters in GCS co-ordinates. Each raster represents active fires in time series over a 29- 31 day cycle, and when classified to 4 natural breaks, each break represents 7-8 days of burns. Each raster was reclassified into Boolean maps of “burn = 1” and “no-burn = 0” for each 30 days and ‘combined’ annually which resulted in 12 maps showing the location of annual detected fires (fig.20).
Figure 20 - Boolean map of burns for 2011
Boolean maps were extracted to the MBR extent and a new field ‘year’ was added. These were then ‘combined’ into a single map of annual burns, resulting in 507 combinations of burn years. The ‘year’ fields were then summed to add the burn pixels for each year, resulting in a unique field with number of burns over an 11 year period.
The combined fire raster was reclassified to frequency classes: No fire years; 1-3 fire years; 4-6 fire years and 7-8 fire years.
The biome polygons in VegMap were resampled to the 440m cell size of the fire layer, and classified to type: Savanna (S); Grassland (G); Forest (F) and azonal (A).
The fire and biome rasters were combined, resulting in 12 combinations of fire frequency in the different biome types. These combinations were used as classes for the fire event threat layer. The
78 combined fire/biome raster was resampled to the 50m raster mask, and a ‘sub-criteria score’ field was added. The final step involved applying a mask of built-up and transformed areas to avoid allocation of threat scores for fire absence in these areas.
Classes of fire frequency per biome type ( S= savanna; G= grassland; F= forest; A= azonal) :
S - no fire S - 1-3 fires S - 4-6 fires S – 7-8 fires G - no fire G - 1-3 fires G - 4-6 fires G - 7-8 fires F - no fire F - 1-3 fires F - 4-6 fires A - no fire
Figure 21: Fire Events Threat Layer. Built up and cultivated areas in black are not included in frequency count.
4.3.5 Summary of seven threat layers
The preparation of the seven individual threat layers realises objective 2 in the research. For analysis and discussion further on in this study, the geographic extent of each of the seven threat factors were summarised to the priority conservation categories using zonal statistics tools. For weighting and scoring the relative intensity of threats and their sub-criteria, an additional step is required, the AHP process, which is described in the following section.
4.4 The Analytic Hierarchy Process (AHP) Approach
A defining aspect and objective of this research is to quantify the current threats to terrestrial biodiversity in the Magaliesberg Biosphere context, using the Analytic Hierarchy Process (AHP) method. This method provides a logical, comprehensive and participatory method to select and weight alternatives and criteria relative to each other and the main objective, by first decomposing 79 the problem in an hierarchical structure and then comparing the criteria in pairs with respect to each level in the hierarchy (Saaty & Vargas, 2001).
In this research the first level in the hierarchy has a single objective, (current regional threats to terrestrial biodiversity). The priority value of all biodiversity components (ecosystems, flora, birds, meta fauna and mega fauna) is assumed to be equal to unity (Saaty, 1977). The second hierarchy level has seven objectives (criteria), the current and proximal threat factors, (urbanisation, cultivation, mining, transformed open- land, fragmentation, fire events and alien plant invasions).The priorities of the second level are derived from a matrix of pair wise comparisons with respect to the objective of the first level, bearing in mind the biodiversity components. Thereafter, each raster pixel (alternative) is allocated a priority score based on the priorities of each threat criteria.
Threats to terrestrial biodiversity in the MB (ecosystems; flora; birds; meta fauna; mega fauna)
transformed invasive alien urbanisation cultivation mining fragmentation fire events open land plants
Figure 22: Hierarchical model used in the AHP
This form of multi criteria analysis (MCA) requires the input of experts to subjectively evaluate the threat criteria relative to one another, by pair-wise comparisons in relation to the objective. The qualitative pair-wise scores provided by each expert participant, is then synthesised throughout the hierarchical structure (Saaty & Vargas, 2001) into a relatively objective quantitative ranking for each threat. These results are then applied in the GIS to weight the different threat criteria on a per pixel (50m x 50m) basis.
4.4.1 AHP threats focus-group meeting and questionnaire
AHP is a way of incorporating expert knowledge into quantitative spatial analysis. A pair-wise comparison form and two other related questionnaires were completed by regional experts at an AHP focus-group meeting attended by committee members of the Magaliesberg Biosphere Initiative Group. This meeting was held on the 13-10-2012 with three primary objectives.
1. To introduce the spatial approach to the research, and the AHP as a concept to quantitatively rank the regions threats, and then to explain the AHP questionnaire forms.
80 2. To introduce and discuss the pre-selected threat factors to be analysed, to get feedback as to whether participants concur with the regional threats proposed and their sub- criteria, in order to develop and weight spatial layers. 3. It was anticipated that a hard-copy of the questionnaire forms would be completed by participants during the workshop session.
These objectives were informed by a study done by Forsyth & van Wilgen (2009), who applied a similar approach. They used workshop sessions to define the AHP objective and then select and weight the criteria to achieve it. They then searched for spatial data to represent criteria. However they found they were limited by the choice of datasets available that would spatially define criteria across the study area (Forsyth & van Wilgen, 2009). Hence, in this study it was decided to select criteria based on a review of the literature and availability of data from the outset, before the application of the AHP, and review the choice at the workshop.
After a short presentation and discussion on the questionnaire and regional pressures on biodiversity, the threat factors presented were deemed appropriate. The focus group attendees elected to rather complete the Questionnaire at a later time electronically. Some minor adjustments were made to the forms after feedback, and they were emailed to the 7 workshop attendees, and 13 other considered experts in ecological/environmental sciences or with expert knowledge of the Magaliesberg region. After five weeks, a total of 10 questionnaires were returned completed, reflecting a 50% response rate. Some forms were returned with useful comments. A list of respondents is appended (appendix 2).
The e-mailed questionnaire forms begin with a brief explanation of the research and AHP method, followed by the questionnaire objectives and instructions on how to complete it. The questionnaire forms, including objectives and instructions and four tables described below, make up appendix 3. Table 1 is a description of the threat factors in the context of the spatial data and land-cover classes from which they were derived. These descriptions are used to inform participants in their choices.
Table 2 is the AHP respondent input sheet, adapted from (Goepel, 2012). The exercise involves pair wise comparisons of each threat criteria in relation to a stated objective – in this case:
To establish the relative importance of each threat criteria as a stressor to the existence of biodiversity in the Magaliesberg Region. The judgments elicited from respondents are taken qualitatively and corresponding scale values (1 to 9) are assigned to rank them (Saaty, 1977). A scale of 1-9 was suggested by Saaty, since psychological experiments show that an individual cannot compare more than 7± 2 objects simultaneously, and that using ‘1’ as a unit difference between successive scale values should represent the differences and distinct shades of feeling that people have when making comparisons (Saaty, 1977).
81 Table 12 - Threat criteria and the AHP scale for comparisons (adapted from Goepel, 2012)
THREAT CRITERIA: 1.Urbanization 2.Cultivation 3.Mining
4. Transformed open land 5. Fragmentation 6. Fire Events 7. Invasive alien plants
Intensity of importance Definition Explanation
Two elements contribute equally to the 1 Equal threat objective
Experience and judgment slightly favour one 3 Slightly more threatening element over another
Experience and judgment strongly favour one 5 Somewhat more threatening element over another
One element is favoured very strongly over 7 Strongly more threatening another, it dominance is demonstrated in practice
The evidence favouring one element over 9 Exceedingly more threatening another is of the highest possible order of affirmation
2,4,6,8 can be used to express intermediate values for elements that are very close in importance
Table 3 looks at sub-criteria, the objective being: To establish the range in intensity of impact for each threat. For consistency, the threat intensity scale (Aucamp, 2009; Wilson, et al., 2005a), also ranged from 1-9 (5 categories, odd numbers). This scale is used for the straightforward ranking of sub-criteria within each threat criteria. The mean values across all respondents for each sub-criterion are applied as the scores for sub-criteria within each threat layer.
Table 13 - Example of sub-criteria within threat criteria ‘transformed open land’, and the impact intensity scale applied to them.
Eroded (gullies,dongas)
Degraded (sparse vegetation /bare areas) Threat Criteria 4 Sub-criteria TRANSFORMED Disturbed (old-lands) OPEN- LAND Modified (recreational grass, plantations, urban open space)
Impact Intensity Scale
1 Low Impact 3 Slight Impact 5 Medium Impact 7 High Impact 9 Very High Impact
Table 4, adapted from a similar approach by Jaganath, (2009), looks at the vulnerability of the various biodiversity components to threat criteria, the objective being: To establish the degree to
82 which each threat affects the different components of biodiversity. This measure is similar to Wilson, et al’s, (2005a) impact dimension of vulnerability (see Section 3.3.1). Components of biodiversity that would be variously affected by threat criteria needed to be captured in as few categories as possible. A vulnerability scale also from 1-9 was used to score the biodiversity components against each threat criteria. The results of this table may aid in forming judgements in the pair-wise comparisons and may also be viewed in conjunction with the pair-wise results, but cannot be spatially represented (Wilson, et al., 2005a).
Table 14 - Biodiversity components and the vulnerability scale used to score them.
Compositional (species,populations & communities) component
Flora – Includes vegetation types (e.g. Andesite Mountain Bushveld, riparian zone) & vegetation structure (e.g. trees / large shrubs, grasses / forbs)
Mega fauna – Includes medium to large mammals and reptiles (their distribution assumed to be more affected by fencing and built-up areas)
Meta fauna – Includes small mammals, reptiles, amphibians, (possibly insects and invertebrates) (their distribution assumed not to be affected as much by fencing and built-up areas)
Birds – All species that occur regionally
Ecological component (Structure and functioning) (Liu & Taylor, 2002; Noss, 2000)
Ecosystems – Incorporates ecological structure, functioning and integrity
Vulnerability Scale
1 Not vulnerable 3 Slightly 5 Moderately 7 Vulnerable 9 Extremely vulnerable Vulnerable vulnerable
4.4.2 Application of the AHP Calc template
The populated AHP Calc Excel template Version 16-10-2012 (Goepel, 2012), is attached as Appendix 6. This public domain Excel template synthesises the reciprocal pair-wise comparison matrices for each respondent into ratio scales, from which the principle eigenvector of priorities is derived. The eigenvector shows the dominance of each criterion with respect to every other criterion (Saaty & Vargas, 2001; Saaty, 1977). The template requires that the principle eigenvector weights are expressed as proportions that sum to 100. Therefore each threat criteria is ranked as a percentage of 100 for each participant.
Saaty, (1977) takes into account that human judgement may not be transitive in a mathematical sense (such as A=B and B=C then A=C), which may lead to inconsistent results. For this, a numerical consistency ratio (CR) to check the stability of pair-wise comparison matrices is provided. The CR is the ratio of the consistency index (CI) (calculated from the greatest eigenvalue of the matrix and the order of matrix) and the Random Index (RI) (Saaty’s pre-determined values for matrix 83 orders ≤ 15). It is recommended that the pair-wise comparison matrix be revised should the CR exceed a value of 0.1 (Saaty & Vargas, 2001).
One way in which to synthesize the scores of multiple participants in an AHP is to aggregate the results of each individual to a geometric mean, which results in a consolidated ranking of criteria that is not statistically based on sample size or participants (Stranger, 2004). The AHP Calc template for multiple participants, allows for a maximum of 7 entries. In the summary sheet of the template, the consolidated input matrix for multiple participants was calculated as the geometric mean of all threat criteria weights from the seven input matrices with the lowest CR values out of the 10 participants, to arrive at final weights representative of the extent to which each threat impacts on the regions terrestrial biodiversity.
Some of the CR values for the seven individual participants selected far exceeded 0.1, but the CR value of the consolidated matrix was 0.045, well below that of any single participants’. An improvement in consistency such as this, as explained by P. Fatti (pers.comm., 2013), would be expected when applying a geometric mean across individual scores. Thus, the AHP result with an overall CR of 0.045 is accepted. Table 15, summarizes the AHP weights and consistency scores for all participants and the final consolidated weighting used in the weighted sum procedure.
Table 15- AHP relative weightings of threats per participant and consolidated weights
AHP - Pair wise Comparisons (Relative Weighting) consolidated weights rank Results per participant in percentages Participants : P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P n=7 CRITERIA : urbanization 22 15 7 8 10 6 52 11 9 45 23% urbanization 2 cultivation 10 8 7 6 6 4 16 7 6 8 10% cultivation 4/5 mining 13 50 21 26 52 46 6 6 55 14 31% mining 1 degradation 39 7 7 12 7 7 4 6 6 7 10% degradation 4/5 fragmentation 7 8 7 6 12 6 9 7 5 8 9% fragmentation 6 fire events 3 6 12 4 8 4 4 52 4 6 4% fire events 7 alien invasion 5 6 38 38 5 28 9 11 15 11 13% alien invasion 3
Consistency Ratio % 13.6 25.7 19.4 7.4 24.3 12.3 30.7 37.1 48 9.6 3.4 Consistency Index 0.18 0.339 0.256 0.098 0.32 0.162 0.405 0.489 0.633 0.126 0.045 highest % mining second lowest % fire
84 4.4.3 Applying AHP weightings and sub-criteria scores
This step marks the final requirement to complete objective 3. The mean scores across all participants for threat intensity of sub-criteria were used to indicate levels of intensity (table 16). The intensity scale ranged from 1-9 for sub-criteria, but scores between layers are not relative. Hence a score of 6 for sub-criteria ‘commercial cultivation’ in the cultivation threat layer is not equivalent to the same score for ‘smallholdings’ in the urbanization threats layer. Only when the sub-criteria scores are multiplied by the AHP weight of each layer in the final overlay, do the scores become relative.
Table 16 - Mean scores for intensity of threat applied to classes in each threat layer
Urbanization Classes Mean scores for Intensity of Threat Urban areas and rural centres (medium and high density) 9 Smallholdings, scattered rural (low density). 5 Cultivation Classes Commercial cultivation. 7 Small-scale cultivation. 5 Mining Classes Extraction pits and tailings, landfill sites and sewerage. 9 Subsurface mines and infrastructure. 5 Mine residue areas. 9 Transformed open land classes Erosion gullies. 6 Degraded and bare land (excluding natural rock). 5 Modified (recreational grass, plantations, urban-open) 5 Old-lands/ abandoned agricultural fields 3 Fragmentation classes 0 – 90m 7 91m – 240m 6 241m – 570m 3 571m – 1340m 1 1341m – 3100m 0 Fire event classes Savanna – no fire 4 Savanna – 1-3 fires 0 Savanna – 4-6 fires 0 Savanna – 7-8 fires 6 Grassland – no fire 4 Grassland – 1-3 fires 0 Grassland – 4-6 fires 0 Grassland – 7-8 fires 6 Forest – no fires 0 Forest – 1-3 fires 3 Forest – 4-6 fires 7 Azonal – no fire 0 Alien Plant Invasion classes Landscape Riparian 0 – 1.65% 0 – 9% 0 1.65 – 8.25% 9 – 29% 3 8.25 – 20% 29 – 55% 6 20 – 46% 55 – 93% 8
85 The scale of score values for sub-criteria from 1 to 9 affects the result of the weighted linear combination (overlay) map algebra applied to the threat layers (Malczewski, 2004). The score that is given to a sub-criteria is multiplied by the weight of the threat layer (weights for each threat layer are proportions that sum to 100). Thus in the final overlay procedure, besides a score of 0 (which is applied to all ‘no threat’ pixels) the minimum score for a pixel would be 1 (minimum sub-criteria 1) x minimum layer weight (1)), and the maximum score for a pixel would be 900 (max for a sub-criteria (9) x max layer weight (100)). This range and the distribution of values within it affects the thresholds applied to values and map visualization, so care must be taken with the categorisation of values, as explained in the final step of the methodology.
4.5 Overlay Analysis – Priority Threats to Biodiversity Composite In the following step, the class scores are applied to the ‘sub-criteria score’ field in each of the threat layers and then ArcMap performs a raster math operation to calculate the weighted sum of all the threat layers, according to the AHP defined weights for each threat criteria. The GIS processing for this weighted sum overlay corresponds to that applied to the Priority Biodiversity Conservation Areas overlay (section 4.2.8), to produce the continuous composite of Priority Biodiversity Threats (fig. 23). This PBT Map, as a composite of seven threat criteria completes objective 4.
Figure 23: Priority Threats to Terrestrial Biodiversity (PBT) –continuous ranking from weighted–sum This map is replicated on A4 size in Appendix 5 4.6 Incorporating Planning Units for Spatial Analysis
Sections 4.1 to 4.3 have dealt with the development and weighting of thematic layers and composite priority maps to represent conservation value and terrestrial biodiversity threats in the MBR. The 5th objective of this research is to develop this data into useful products, using planning units as a basis for spatial analysis and to aid applied decision making.
The determination of planning units becomes a function of scale and intended application. The priority map outputs at ± 50m x 50m (0.25 ha) resolution provides for a relatively fine scale analysis, 86 the geographic accuracy of which would be lost if planning units were too large. However, considering the areal extent of the ROI (± 4500km², equating to ±2 million pixels) a larger implementation unit albeit more generalised, would be more appropriate for some processing tasks and for practical applications. For this reason, original farm portions were used to demonstrate an alternative working implementation unit.
4.7 The Synthesis of Priority Conservation Value and Priority Biodiversity Threats
The final objective of this research is to spatially reference the Priority biodiversity threats (PTB) to Priority conservation value (PCV). For cross-tabulations a PCV and PBT score must reflect for each planning unit record. The processing involves converting float rasters to integer, to build attribute tables. For computing purposes scores for both PCV and PTB were classified into 4 intervals, to reduce the number of potential combinations of scores to 16 classes once combined and cross- classified. The resulting maps reveal the spatial relationships between conservation and threat across both planning units as final map products.
4.7.1 Original farm portions cross-classified
To achieve averages of PCV and PBT for each Original farm, incomplete provincial cadastral data on parent farms sourced from the National Geospatial Institute (NGI), were digitally edited to produce polygons of original farms for the study area. Original farms smaller than 50 hectares (including town-lands) were ‘eliminated’ and ‘appended’ to the longest neighbouring boundary, resulting in 329 original farm records. The maps were rasterized and the mean classified pixel scores for PCV and PBT were summarised per farm using Zonal Statistics tools (ESRI, 2012). The two maps were combined, showing a unique PCV and PBT average per farm. Averages were also plotted on a bi-plot.
4.7.2 Pixel-resolution cross-classified map and tables
For pixel resolution cross-classification, attribute tables showed 20 records for PCV scores ranging from 2-31, and 228 records of PBT values ranging from 0-653. To relate each pixel’s PCV and PBT score, the attribute tables needed to relate to one another, hence each pixel had to have its own record to which PCV and PBT scores could be attached. This was achieved by first classifying the scores and converting the rasters to point vectors, which resulted in tables with close to 2 million records each. The tables were joined so that each pixel had a unique combination PCV and PBT score, for input into a scatter-plot and other graphic outputs. The principal cross-classified map is a visual synthesis of both PCV and PBT value per pixel.
87 4.8 Summary
This methodology presents a systematic approach to prioritizing threats to terrestrial biodiversity and referencing them to the priority conservation landscape. Six steps (objectives) were realised to achieve this research aim. The outcome for each objective is summarised as follows:
1. Secondary data that was appropriate to effectively describe biodiversity and conservation value was combined by weighted overlay to produce a map of Priority conservation value (PCV). Four PCV categories are the spatial zones to which threatening processes are geographically referenced for results. 2. Seven spatially appropriate and relevant threatening processes were compiled into spatial layers as criteria to adequately reflect the vulnerability of terrestrial biodiversity. 3. Expert opinion on the relative intensity of threat criteria was formally quantified by AHP method. The mean scores for sub-criteria were applied to reflect the degrees of intensity within each threat criteria. 4. A continuous composite of Priority Biodiversity Threats (PBT) resulted from the weighted sum of the seven threat criteria as weighted by the AHP. The map was classified into four levels of threat, to correspond with the PCV categories. 5. The 50m pixel is the resolution of processing and spatial analysis. A second planning unit, the original farm is applied to demonstrate an alternative implementation unit for practical application. 6. The synthesis of PCV and PBT - priority maps were integrated and cross-tabulated into a single map of priorities and related graphic outputs, for both planning units from which results were generated.
88 Chapter 5 Results & Discussion - Spatial dimensions of current land transformation and conservation value
5.1 Outline
Threats to biodiversity are increasingly being incorporated into conservation planning (Wilson, et al., 2009; Fuller, et al., 2010; Pressey, et al., 2007; Wilson, et al., 2005 b). Spatial conservation planning products (and threat cost-layers that may be associated with them) that cover the Magaliesberg region have been undertaken at the provincial, municipal and local level. Spatial data on individual threats to biodiversity have not been collated for the Magaliesberg region before. Thus the information generated in this research may be used to inform conservation and development planning for the region. Considering the proposed reserves proximity to the economic hub of South Africa – the Gauteng City region – dynamic and increasing pressures of population, development and extraction need to be incorporated into conservation assessment at a regional/ landscape scale.
In this chapter a landscape scale analysis and discussion on the current proximal threats to terrestrial biodiversity and conservation in the MBR follows, based on the outputs of the previous chapter. The section is presented, in the order of the research objectives, focussing first on Priority Biodiversity Conservation Value (PCV), then on individual threatening processes as weighted by the AHP, and then on the priority Biodiversity Threats Composite (PBT) and its spatial relationship with PCV, when the two priority maps are integrated.
In analysing the results, the discussion focuses on exposure (extent) and intensity (AHP weighting) of threatening processes, following Wilson, et al.’s (2005a), discussion on the dimensions of vulnerability (see Section 3.3.1), whereby the vulnerability of PCV categories is measured by their exposure and levels of intensity to each of the seven current threats, and the threat composite. The chapter concludes with a synthesis of the priority products whereby priorities for PCV and PBT are integrated into two planning units to show the spatial variability of priority combinations, and to demonstrate how conservation responses may be assigned to specific priority combinations.
5.2 A spatial synopsis of Priority Conservation Value
In this research the PCV categories are the spatial zones to which threatening processes are geographically referenced. These categories were determined using a weighted sum of four secondary datasets – 1) Zonation map of the MBR; 2) Provincial Critical Biodiversity Areas maps; 3) Vegetation Types of South Africa; 4) Remaining Extent of Threatened Ecosystems for South Africa (see section 4.2). The results of the weighted sum produced a normally distributed range of continuous values for Priority conservation with scores between 2 and 31 (fig 12 p 67).
To generate categorical data a number of rendering options were tried (natural breaks, quartiles, geometric interval, and equal interval) to find the most suitable interval to show levels of
89 conservation priority (ESRI, Inc., 2010). Taking all maps into consideration, a geometrical interval (GI) with four classes was finally applied.
Proportionally, using this interval, the highest value conservation category covers around 19% (838 km²) of the MBR, followed by medium-high 23% (985 km²), medium-low 31% (1340 km²) and lowest conservation value category 27% (1199 km²). These categories are shown in (fig.24), below.
Figure 24: Priority Conservation Value (PCV), showing formally protected areas within the MBR.
The map also shows formally protected areas which make up the biosphere core zone (the 2014 MBR nomination includes both the MPE and COH as core zone). As ‘protected areas’ was a variable used in 3 of the 5 datasets making up the PCV map, they are represented mostly by high and medium PCV categories (scores from 16.4 to 31) .The MPE (to the North) covers the Magaliesberg Mountain chain, and was initially protected for its pristine remaining wilderness of high conservation value (Carruthers, 2015; Carruthers, 2002; DACE, 2007; Eber, 2005), and the COH (to the south) set in mostly primary karst dolomitic grasslands and Witwatersberg mountain habitat, was declared a world heritage site in 1999. Historically both these areas have remained relatively uninhabited, presumably due to their lack of utility for human settlement (Noss, 2000; Rouget, et al., 2003), compared to their surroundings, which have been altered by various anthropogenic activities over many decades. So while these areas are not fully representative of all the biodiversity of the region (see Carruthers, 2015), they represent relatively pristine remaining wilderness in a natural and unfragmented state, and are the protection-focussed Biosphere core zone.
All other medium-high and high PCV areas are particularly important in the context of the biosphere because protected areas or the biosphere core should not be viewed as isolated islands of biodiversity to be conserved amongst a sea of developments (UNESCO, 1996). Rather, being in
90 private tenure and not officially protected, these PCV areas represent very important biodiversity that should be managed through cooperation and stewardship, affording the conservation opportunities for the region (see Section 2.2.5). High PCV outside of the core is mostly found in vegetation-types Egoloi Granite Grassland, Gold-Reef Mountain bushveld, Marikana bushveld and Norite Koppie bushveld (fig, 3 p14). Some of these vegetation types make up the last remaining extent of ecosystems of critically endangered, endangered and vulnerable conservation status, according to the NBA 2011 (Driver, et al., 2012). The conservation and protection status and extent of all the vegetation types of the MBR can be found in table 1.
Intermediate PCV categories, medium-high and medium-low (scores from 11 to 21.7) are spread across the ROI and cover most of the represented vegetation types. An arbitrary distinction between these two categories is that medium-high PCV is loosely correlated to conservancies, i.e. the Biosphere buffer zone (as of Sept 2012). Biosphere guidelines for this zone suggest stricter controls on new activities and cooperation from existing activities, to encourage conservation, sustainable land-use and sustainable practice in general (Contour and Associates, 2013).
The low PCV category with scores below 11, are concentrated around the perimeter of the MBR and its 2km buffer. This approximates the urban edge of the three cities - Johannesburg, Pretoria and Rustenburg (see fig.1 p 8), and noting that 3 of the 5 datasets used in the overlay took account of land-use to distinguish critical and/or intact remaining biodiversity, this pattern of lower biodiversity importance towards the perimeter is to be expected. The pattern is also archetypal of a typical biosphere reserve zonation (UNESCO, 1996). Exceptions to this are the few high PCV sites that extend into small areas of the MBR 2km buffer zone, and the centrally located low PCV areas of the Moot Plains and valley (fig.1 p 8). Low PCV is also distinctive in other vegetation types, Rand Highveld Grassland, Marikana Thornveld and Zeerust Thornveld. Only 1.8% of the latter is represented in the MBR, and this would account for its low PCV status. However it should be noted that if ‘focus areas’ of the National Protected Areas Expansion Strategy (NPAES, 2010) were applied as a criteria for PCV, this area would have achieved a higher PCV status, which highlights the subjectivity of analysis depending on input data, as well as the classification interval applied to the data (Lawler, et al., 2003; Vimal, et al., 2012). The other generally low PCV vegetation types are more affected by threatening processes, which is discussed in the following sections.
5.3 The Spatial Extent of Threatening Processes in the MBR
The seven GIS layers of individual threat factors provide the geographic extent of threats across the MBR, and its PCV zones. Figure 25 shows a ranking of threat factors by their geographic extent as a percentage of the entire MBR extent. Only pixels with sub-criteria scores ≥ 3 were included. Land- uses associated with direct habitat loss - urbanisation, cultivation and mining cover 17% (726km²), 14% (627km²) and 2% (86km²) of the MBR surface area respectively. This indicates that these threatening processes, together with degraded/disturbed veld, are the lowest ranked threatening processes by areal extent in the MBR.
91 On the top end of the rank order for extent, the drivers of habitat degradation – fragmentation, alien invasive plant spread and inappropriate fire regimes, have a wider footprint in the MBR, extending over 80% (3484km²), 42% (1829km²) and 39% (1715km²) of the area respectively. Species invasions and wildfire spread, or the suppression thereof, are not confined by protected area boundaries and thus can dominate as factors of qualitative habitat decline in areas not threatened by direct habitat loss (Noss, 2000). Therefore, in terms of threat exposure by spatial extent in the MBR, the threats associated with land degradation are apparently more widespread than those associated with habitat loss or complete transformation of land-cover, at this time.
The Geographic extent of threats within the MBR Fragmentation 80%
Invasive Alien Plants 42%
Fire Events 39% Urbanization 17%
Cultivation 14%
Degradation 11%
Mining 2%
0 10 20 30 40 50 60 70 80 90
Figure 25: The geographic extent of threats within the MBR.
Percentages indicate the geographic coverage of each threat as a % of the total area of the MBR. For example 42% of the MBR is covered by densities of IAP’s that are considered a threat, while 58% of the area is either clear of IAP’s or populated with densities below threatening levels.
The geographic extent of proximal threats is akin to areal loss of ecosystems, one of the indicators used by Noss et al (1995) to determine their decline (Noss, 2000). Although the percentage extent of threatening processes provides a quantitative indication of risk to biodiversity, it does not provide information about the intensity or severity of this threat. For example, although mining has transformed a relatively small fraction of the MBR, its footprint is concentrated, and impacts to biodiversity are heavily apparent in the north (Ololade, et al., 2008), and other mined areas, through the direct loss of habitat to mining itself, but also habitat decline from dust and light and water pollution, and an increase in population density, with all its related impacts, including formal and informal urban sprawl, pollution and resource harvesting (Ololade, et al., 2008; Howard, 1996). Similar types of impacts would affect the periphery of the MBR which closely approximates the urban-edge in the north-west and south-east particularly (fig.1 p. 8). Hence, the discussion that follows takes a more in depth look at threat exposure and intensity taking into account the AHP weightings as determined by regional experts for threat criteria.
92 The spatial relationships of exposure to threat and intensity of threat within each PCV zone, provides detail on the vulnerability of each zone to particular threatening processes. This is important considering the premise that all PCV zones need to maintain (and in some cases even improve) their existing biodiversity features by active management, and sustainably integrating the human element while limiting risk of deterioration.
5.4 AHP Ranking of Threat Criteria
The geographic coverage of threats and the proximity to roads and built-up areas, indicates current exposure to threat, while AHP weightings for threat criteria and ranking for each sub-criteria, such as variables of IAP density, urban density, frequency of fire and categories of degradation/disturbance used in this research, provide detail on the intensity of threatening process, as a dimension of vulnerability (Wilson, et al., 2005a).
Saaty’s AHP is a form of MCA that can be applied spatially (Moffett & Sarkar, 2006a). By pair-wise comparison, experts qualitatively evaluated the threat criteria in relation to one another, in respect of the following objective : To establish the relative importance of each threat criteria as a stressor to the existence of biodiversity in the MBR. The geometric-mean of weights from 7 participants resulted in a formalised and quantified relative ranking for each threat, with an overall consistency ratio of 0.045 for the AHP, which is accepted. These weights were then applied in a GIS to the different threat criteria on a per pixel (50m x 50m) basis for overlay analysis and per pixel cross- tabulations.
Results of the AHP conducted for this research show the criteria posing the greatest threat to biodiversity of the MBR to be, in ranked order percentage, mining 31%, urbanisation 23%, Alien Invasive Plant spread 13%, cultivation on par with land degradation at 10%, fragmentation 9% and lastly injudicious fire regimes 4% (table 15). In comparison, the top 5 ranks in O'Connor & Kuyler’s (2009) AHP results for impact of land-use on the integrity of moist grassland biodiversity were urban land-use (16%) ahead of timber plantations (14%), dairy farming (13%), irrigated-crops and rural settlements (12%) and dry-land crops (11%). Similarities can be seen in the high rank for urbanisation and medium-high rank for intensive agricultural land-use, while rural settlements and cultivation received similar scores for both studies. A grassland vulnerability study Neke & Du Plessis’, (2004) also estimated high severity of impact scores for the dominant transforming land-uses cultivation, afforestation, urban and mining.
Similarly, a land-cover change analysis from the 1970’s over 3 decades of the Rustenburg region north of the Magaliesberg suggests the high rates of urban expansion due to work opportunities in the mining industry are the cause of significant increase in built-up class, some of which is converted cultivated land. The study also points to changes from grassland and bushveld to cultivated land, because of the development of intensive agriculture in the region (Ololade, et al., 2008). Interestingly, only 0,5% of original 1973 cultivated land, remained in that class by 2002, which indicates the dynamic nature of agricultural process over time, and explains the relatively high
93 proportion of old-lands in the region - a sub-criteria for land degradation threat in this study. The temporal changes from one land-use to the next suggest that such historical investigations of land- cover change would enhance a threats analysis, especially for understanding multiple and compound threats, as is attempted in this study with the threats composite. In terms of compound threats, Wilcove, et al (1998) found in many instances that apparent threat is spawned from another threatening process.
Although Wilcove, et al., (1998) uses very different method to quantify threats to specific groups of imperilled species in the US, their results indicate a similar trend. Coarse scale results were not surprising, revealing habitat loss and degradation to be the most pervasive threat to biodiversity, endangering 85% of imperilled species. Predation or competition from invasive alien species came second, affecting 49% of species, most particularly plants. Similarly in this study IAP species were weighted at 13% threat intensity and ranked 3rd behind only mining and urban land-uses.
At a finer scale, Wilcove, et al, (1998) divided habitat destruction and degradation into 11 categories of threat, to investigate their relative significance. Of these, the following threats were ranked highest as impacting on imperilled species overall: Agriculture (38%), commercial development (35%), water diversion and impoundment (30%), (mostly affecting aquatic species), grazing (22%) (plant impacts) infrastructure development (17%) - particularly roads (15%) and disrupted fire ecology (suppression and increased occurrence) (14%). Surprisingly, mining impacted on 11% of species, (mostly toxic pollutants impacting on aquatic species). Possible reasons for lower effects of mining at a continental or biome scale may be the relatively small footprint of mining activities, compared to urban or agricultural land-uses (Neke & Du Plessis, 2004).
These categories closely resemble threat criteria applied to the AHP, and although the rank order varies slightly from the sub-criteria ranked below, these drivers of habitat loss and destruction can be considered universal. In fact, while vulnerability to threat and the types and intensity of threatening processes vary geographically and in frequency (Wilcove, et al., 1998; Noss, 2000; Vimal, et al., 2012), the literature overwhelming suggests that world- wide, landscapes with high population densities and /or intense resource use, are most at risk from the threatening processes associated with these factors (Noss, 2000; Fuller, et al., 2010; Rogers, et al., 2010; Reyers, 2004; Soule, 1991).
94 5.5 Combining Quantified Intensity and Exposure of Threats
AHP weighted threat factors by Geographic extent of threat proportion factors by proportion
Fire Urbaniza Fragmen Events tion Cultivati tation 4% 8% 9% on 7% Degradat ion Urbaniza Fire Mining Events 1% 10% tion 19% 23%
Invasive Alien Plants 21% Fragmen Cultivati tation Mining on 39% 31% Invasive 10% Degradat Alien ion Plants 5% 13%
Figure 26: The contrast between normalised proportions for AHP weightings and geographic extent of threat criteria.
In contrast to the AHP weightings, the geographic extent of threats show very different normalised proportions for threat criteria (fig.26). While it is not the objective of this research to compare these, some inferences can be made about threats to a region when exposure and intensity are viewed in combination, such as visually on a map. For example, Neke & Du Plessis, (2004), found very similar results to this research, whereby mining and urban land-use received maximum scores for negative impact on grasslands, however they are localized impacts that only account for 1.9% and 0.3% of former grassland area in their study, a very small percentage compared to exposure by cultivation. In conservation planning assessment of the vulnerability of priority conservation areas and the regions threats viewed in combination and in relation to Wilson et al’s., (2005a pg 530) ‘Three dimensions of vulnerability’ plot, provides an alternative perspective.
95
D
Figure 27: 3- dimensional plot of the vulnerability of Priority Conservation Value in the MBR (Notice that only exposure and intensity are plotted on the X and Y axis. Vulnerability is not plotted on the impact scale - Z axis). “Dimensions of vulnerability” in biodiversity threat analysis A = Threat of urbanization and mining; B = Threat of fragmentation and fire; C = Threat of IAP; D = Threat of degradation and cultivation (adapted from Wilson, et al. 2005a)
A mock-up of this plot (fig.27) shows that the incongruent proportions for direct habitat loss threats of mining and urbanization would feature highly on the intensity scale but low for current exposure (A), while the opposite would be true for injudicious fire events and fragmentation (B), threats that have higher exposure but are less intense. A different scenario is evident for more analogous proportions for IAP threat (C), and degradation and cultivation threats (D) where exposure and intensity are represented more equally. All threats would vary along both intensity and exposure axis, depending on sub-criteria values, and the impact of threat on biodiversity features (not analysed here) would vary along the positive/negative intensity plane only (Wilson, et al., 2005a).
For example focussing on impacts of invasive alien plants to rare bird species in a protected area, the occurrence and density of A.donax, may compete with local phragmites species, reducing preferred habitat for some reed nesting birds (-ve, -ve impact), whereas a large stand of poplar, while impacting water resources, may provide nesting habitat for raptors (-ve, +ve impact).
Quantified results for exposure and intensity of threat are important to correctly assess vulnerability of biodiversity features. Results are scale dependant, and are useful in a regional context in this research. In the following discussion, the results of the geographic extent of threats and the AHP weightings for intensity are amalgamated as far as possible. First, the spatial variability of individual threatening processes is assessed in relation to PCV categories, so that strategies can target particular areas for conservation. Results of the threat composite and synthesised products are useful in this regard.
96 5.6 The Relative Ranking of Sub-Criteria
The scores for sub-criteria in this research provide finer detail of the severity of threat posed within each threat category. Their ranking becomes relative when scores are multiplied by the criteria AHP weights before overlaying in the GIS. A graphic representation of quantified scores for the relative intensity of sub-criteria threats is shown in (table 17). Quartile thresholds have been applied to indicate a gradation of threat values, to aid the discussion (Lawler, et al., 2003; Vimal, et al., 2012).
Table 17 - Matrix of values applied to pixels in each threat layer to represent relative intensity of individual threats.
AHP Threat Criteria Weight (%) Sub-Criteria Scores 3 4 5 6 7 8 9 Mining 31 155 279 Urbanisation 23 115 207 Invasive Alien Plants 13 39 78 104 Cultivation 10 50 70 Transformed open land 10 30 50 60 Fragmentation 9 27 54 63 Fire Events 4 12 16 24 28 PRIORITY BIODIVERSITY THREAT CATEGORIES 0-29 Low threat 54-91 medium-high threat 29-54 medium-low threat 91-279 high threat
Table 18 - Detail of sub-criteria classes ranked from highest to lowest by intensity score. Ranking by areal extent is also shown in column 4.
Rank Relative Km² Km² Rank Threat Sub- Order Score Extent Order Criteria Criteria
1 279 67 13 Mining Categories: mines surface extraction, tailings, dumps, urban land-fill urbaniza Categories: high/med density residential, commercial, infrastructure, rural 2 207 512 8 tion centres 3 155 19 15 Mining Categories: mines sub-surface & all mining infrastructure urbaniza Categories: small-holdings, scattered rural, 4 115 215 11 tion low-density residential estates Density: invasion of woody species 20-46% terrestrial, 5 104 204 12 IAP's 55-93% riparian 6 78 690 6 IAP's Density: invasion of woody species 8-20% terrestrial, 30-55% riparian Cultivati 7 70 605 7 Category: all types of large-scale commercial cultivation on Fragme 8 63 1522 2 Proximity: within 240m of built-up areas and image-defined-roads ntation Transfor 9 60 10 17 med Category: soil erosion gullies open Fragme Proximity: between 240m-570m from built-up areas and image-defined- 10 54 960 4 ntation roads Cultivati 11 50 22 14 Category: all types of small scale cultivation on Transfor Categories: degraded land, bare areas, recreational grass, urban open- 11 50 226 10 med areas open 12 39 935 5 IAP's Density: invasion of woody species 2-8% terrestrial, 9-30% riparian Transfor 13 30 239 9 Category: old agricultural fields med
97
open Fire 14 28 2 19 Frequency: 4-6 fires in 12 years in afromontane forest areas events Fragme Proximity: between 570m-1.3km from built-up areas and image-defined- 15 27 1003 3 ntation roads Fire 16 24 13 16 Frequency: 7-8+ fires in 12 years, in savanna and grassland events Fire 17 16 1696 1 Frequency: no fire in 12 years fire suppression in savanna and grassland events Fire 18 12 4 18 Frequency: 1-3 fires in 12 years in afromontane forest areas events
Of the threats responsible for direct habitat loss all mining categories and both urban classes account for the 4 highest ranked threats. Commercial cultivation falls within the medium-high threat category but ranks below two IAP invasion classes accounting for invasion densities above 8% terrestrial and 30% riparian (med-high threat) and densities above 20% terrestrial and 55% riparian (high threat). The only other categories above the 50th percentile are fragmentation effects within a 240m proximity of built-up areas and roads, and erosion gullies as a category of transformed-open land, both considered to be medium-high threats, and factors of habitat decline. All other categories of transformed open land, small-scale cultivation, injudicious fire events and the lowest IAP density and fragmentation proximity classes are associated with habitat degradation and decline, and fall below the 50th percentile for threat intensity. They are considered to be medium-low and low threat processes in the MBR currently.
5.7 General Patterns of Exposure of Priority Conservation Value to Individual Threats
In conservation assessment and in the context of biosphere management it is worthwhile to spatially assess the vulnerability of priority conservation areas to different threat factors (Rouget, et al., 2003; Pressey, et al., 1996). The exposure of each PCV category to historical and current threat factors shows some interesting correlations. Fig. 28 indicates that high PCV is less exposed to threats overall, while low PCV is more exposed to threat. Also, it is evident that threats associated with habitat loss affect conservation categories from high to low at increasing ratios. Where low PCV is comprised 28% cultivated and 34% urban land, high PCV is only exposed to 2% and 6% of the same land-uses respectively. The threats associated with habitat loss also make up a smaller proportion of the overall threat exposure per category. Conversely, threats associated with habitat degradation are more evenly distributed across all PCV categories, but proportionately per category, there is more exposure to degradation threats, albeit less severe in high PCV categories. Notably, higher PCV is most exposed to the threat of alien invasive plants and imprudent fire events, with a higher percentage of coverage of these threats than lower PCV.
98
Proportions of threat coverage by PCV category
39 48 62 high PCV 126 6 1. Mining 2. Cultivation 49 46 74 medium-high PCV 1 8 8 12 3. Urbanisation 4. Disturbed/degraded med-low PCV 14 14 13 44 42 83 2 5. Fire Events 6. Invasive Alien Plants low PCV 28 34 12 26 35 93 3 7. Fragmentation Percent coverage of threat factors
Figure 28: The exposure of PCV categories to threat factors. Habitat-loss threats, numbered 1-3; Habitat-degradation threats numbered 4-7.
5.7.1 PCV exposure to habitat loss
The spatial distribution of the higher intensity threats associated with direct habitat loss (Noss, 2000), derived from discrete land-cover classes in provincial LCC maps, is shown in fig. 30. Besides the limitation in finding common classes to represent various intensities of threat across provincial datasets, the classes are quantified by extent at relatively high resolution, and it is useful to understand the pattern of development in a regional context (Vimal, et al., 2012).
In the case of mining, considered the most threatening activity, by intensity to the biodiversity of the MBR, fig.30 shows PGM mining operations concentrated to the north of the Magaliesberg and a number of surface mines and quarries, and mine residue areas along the slopes of the Witwatersberg Range, and close to the urban-edges of Johannesburg and Pretoria, where aggregate quarries and borrow-pits have been established close to urban markets to reduce transport costs (GDACE, 2008). Only 1% of high and med-high PCV is exposed to mining activity, while 2% and 3% of med-low and low conservation categories respectively, are exposed. The graph in fig.29 shows details of PCV exposure to different mining activities. The majority of activity in high PCV areas is open excavations and waste sites. Considering the findings of experts by AHP, that mining is the most threatening process by intensity, it is recommended that approvals for future mining applications in the MBR take cognisance of biosphere zone restrictions on activities as well as PCV status to keep exposure to mining in check. This highlights the importance of the contribution of expert opinion in the AHP to identify severe concentrated threats.
With surface mining complete destruction of the surface always occurs, which affects surface-water, soil, fauna and flora and complete loss of visual integrity during operations (GDACE, 2008). Environmental impacts from the types of mining that occurs in the region have been outlined in Section 3.2.2. Many of these impacts can be partly mitigated by legislated mine closure rehabilitation
99 and/or restoration plans, which may result in the threat of some operations being temporary. However close management and monitoring during operations is critical to ensure that as far as possible undisturbed adjacent areas are not impacted by mining activities, and that soil erosion by storm water runoff and that compaction in- situ and of stockpiled topsoil is prevented (GDACE, 2008). Without post-closure interventions, the footprint of older abandoned mine residue areas may remain heavily polluted, and often succumb to ecological succession by mostly invasive species. The case of the latter is also true for old abandoned agricultural fields (McKinney, 2002).
2500 MINING 35000 URBANISATION 35000 CULTIVATION subsurface/inf 30000 30000 2000 rastructure Smallholdings mine residue small scale 25000 ,scattered 25000 1500 extraction rural 20000 commercial, 20000 medium & pits/tailings 15000 large scale 1000 15000 high density 10000 10000 urban 500 5000 5000 0 0 0
Figure 29: Exposure of PCV to sub-criteria threats for habitat loss, in hectares.
The top legend entry represents the least severe threat, and the bottom entry the most severe.
Urban landscapes are perhaps the most threatening to biodiversity by their permanence and tendency to expand into urban sprawl (McKinney, 2002). This is clearly evident in patterns of urban spread in a northern trajectory from NW Johannesburg and around Hartebeespoort dam (Cooper, 2010; Long & Hoogendoorn, 2013), and can be seen on the map fig 30. Noss (2000), remarks that in developed regions, areas near water (coastal areas, shorelines of rivers, lakes and dams) are likely to be inundated by human activities, and high population densities. The landscape around Hartebeespoort dam has been modified for agriculture, tourism and recreation ( DACE, 2007; Eber, 2005) and more recently a proliferation of medium to low density rural residential housing in aesthetic surroundings – exurban development (Hansen, et al., 2005; Long & Hoogendoorn, 2013). Although urban environments predominate in low PCV, covering >31000ha (fig. 29),the preferred siting of tourist facilities and residences in appealing surroundings, near nature and with good views, should be regarded as a likely future threat for the higher value conservation areas. A case in point being the recent imprudent siting of a tourist venue development, Kgaswane lodge, in a provincial protected environment (Fatti, 2013). Paradoxically biosphere guidelines promote tourism and out- door recreation as low-impact sustainable development, suitable for biosphere reserves (Department of Environmental Affairs, 2012; UNESCO, 1996). With this in mind, to ensure their sustainability, sensitivity is required with regards to siting facilities and certain recreational pursuits such as hiking and off-road vehicles that can threaten localised endemic and endangered plants animals, and habitat ( DACE, 2007; Wilcove, et al., 1998).
100 Two urban sub-criteria, ranked 2nd (medium to high density urban) and 4th (low density urban) are applied to represent the intensity of urbanization threat in the MBR. The latter comprises large expanses of urban-edge mixed development, characteristically exurban development, smallholdings, intensive agriculture – animal feedlots and batteries and greenhouse productions (see appendix A), many of which are interspersed with remnant semi-natural to natural tracts, while the med-high density urban consists of industrial factories, commercial centres formal residential as well as high density informal and under-serviced residential areas. This land-use mosaic forms part of the varied rural-urban gradient of biodiversity habitability described by McKinney, (2002), the impacts of which are briefly discussed in section 3.2.2.
Currently 6% of high, and 8% of med-high PCV area, is exposed to urbanization (fig. 28). Urban sub- criteria are almost equally represented in these zones, low density urban covering >5500ha and med-high density >7000ha (fig 29).
The distribution of cultivation activities, (fig. 30) in the MBR clearly shows large-scale mostly grain mono-cropping in former grasslands to the SW, and on a smaller scale irrigated/pivot cropping in the proximity of rivers and dams. These areas of low elevation and low topographic relief generally create wetland habitats, which are typically very important biodiversity features (Driver, et al., 2012). Often these areas have fertile soils and tend to be highly modified historically (Noss, 2000), as is evident by the swathe of cultivation occurring along the Moot valley.
The findings by Rebelo, et al., (2011), suggest the impact of agriculture and urbanisation on biodiversity in the city of Cape Town is marked in low-lying areas where drainage and infilling of seasonal wetlands for developments and lowlands receiving input from storm water runoff has transformed the low-lying landscape to a degree that the historical extent of wetland types is unknown (Rebelo, et al., 2011). Similarly in the MBR, anecdotal evidence of wetland drainage for cultivation in the southern sections of the COH during the early 1900’s (Louw, pers. comm. 2010), as well as the altered hydrology of the region (see section 2.1.3), and historical and current land-use patterns, suggest major change and impact particularly on lowland and low relief ecosystems of the MBR.
The biodiversity maintained in these modified agricultural mosaics most likely differs from that suited to the original occurring natural landscape (Norris, 2008). However, cultivation activities achieve a lower score for threat intensity than other habit loss processes, and this may be explained in that agricultural landscapes can be seen as matrix habitats, from large expanses of monoculture (presumed to have higher fertilizer and pesticide inputs) through to the increased habitat heterogeneity provided by remnant natural patches, small farm dams and rows of natural or planted vegetation between agricultural fields, which still provides variously suitable habitat for some biodiversity (GDARD, 2011; Norris, 2008). The best available sub-criteria to grade cultivation reflect this matrix habitat, with commercial cultivation ranking 7th and small-scale cultivation 11th for terrestrial biodiversity threat. Fig. 29 shows that high PCV categories are relatively free of cultivation
101 activity, likely due to their predominant mountainous topography and the absence of surface water and poor soils in the dolomitic COH.
Figure 30: The spatial distribution of threats associated with habitat loss across the MBR. The threats showing relative weights are overlaid to PCV. Distribution patterns for mining and cultivation distinctly follow patterns of geology and topography, with concentrated mining in the bushveld complex to the north, and cultivation along valleys and plains on either side of the Magaliesberg chain, and the flat grasslands in the south-west.
5.7.2 PCV exposure to habitat degradation
In this study an attempt is made to quantify factors of qualitative habitat decline as the threats associated with habitat degradation, especially pertinent in areas not threatened by direct habitat loss, such as protected areas. It can be more challenging to map and quantify some of the factors leading to habitat degradation compared to those related to habitat loss, (see Rouget, et al., 2006), and the literature shows they are seldom incorporated into vulnerability and threats analysis. Fig.31 shows the exposure extent and severity ranking for habitat degradation threats, correlations can be made between these factors in relation to PCV (Noss, 2000).
Habitat fragmentation has the largest spatial coverage, in that 80% of the MBR area occurs within 1,3km of a built-up area or road in the proximity analysis, and is hence considered threatened to some degree (fig.17). The entire built-up footprint considered highly fragmented, accounts for some of this extent, (see the spike in low PCV fig 29). As roads were not scored according to their type, any national, main or minor roads classified as ‘urban’ in the LCC has equal proximity bearing, and as such, fragmentation has been overestimated for smaller quieter thoroughfares. Rouget, et al., (2004) quantified untransformed habitat patch size and extent as part of a fragmentation index, thus perhaps for this discussion it is more telling to note that 20% of the MBR is further than 1.3km away from a public road or built-up area and is not fragmented, providing wide expanses of ‘wilderness’. 23% of the area is more than 570m away from built up areas or any roads, and accounts for one of 102 the lowest ranked threats by intensity. These non-fragmented areas of low or no threat, are concentrated in high PCV zones and cover 1704 km², extending beyond the mountains and karst dolomite of the protected areas to include smaller expanses of Egoli Granite Grassland and Zeerust Thornveld vegetation types. Within fragmented areas (<570m to roads/built areas), exposure consistently increases as conservation value decreases (fig.31 and fig.32).
A factor not considered for fragmentation is that some of this extent is traversed by farm boundary fence lines and game fences, which do restrict movement for some free-ranging faunal species. In this study a surrogate for fragmentation was applied in this study, where patch shape, edge effects, distance between patches and other variables were not considered (Fahrig, 2003; Rouget, et al., 2004). However, the proximity analysis applied does give an indication of patch size, and quantifies exposure to direct proximal threats, particularly roads, which are a severe threat to many faunal species, as indicated by road-kill statistics (Wilcove, et al., 1998; van der Ree, et al., 2011).In addition, increased runoff, pollution, fence-lines and maintenance associated with roads, may induce localized changes in habitat structure and function, and can affect the composition of nearby flora (van der Ree, et al., 2011).
Alien invasive plants threaten biodiversity (Noss, 2000; Wilcove, et al., 1998) and are considered important compositional indicators for biodiversity integrity of moist grasslands by O'Connor & Kuyler (2009). The IAP threat layer was based on terrestrial and riparian species abundance (Kotzé, et al., 2010), a collective density to indicate threat, rather than the location and distribution of species themselves. The data also shows invasions are clearly more abundant and denser in riparian corridors than in terrestrial landscapes (see fig.18). Of the 11 invasive woody species recorded in the MBR, two are exclusively riparian invaders. Others occur in both terrestrial and riparian habitats throughout the MBR, with some species predominant in the warmer climate to the north. After the AHP, the class range 20-46% terrestrial invasion density and 56-93% riparian invasion density ranked 5th as the highest priority threat to biodiversity after mining and urbanisation. The mid-range density class ranked 6th as a medium-high threat, while the lowest density class range (2-8% terrestrial invasion density and 9-30% riparian) ranked 12th out of a possible 18, as a medium-low threat. In the AHP, invasive alien plants are considered the priority threat to biodiversity of all threats associated with habitat degradation in the MBR.
High PCV zones have the highest exposure to dense invasions. Fig. 31 shows a positive linear relationship between high PCV having the greatest exposure (48% coverage) to the densest IAP invasion class. This percentage steadily drops as PCV decreases, reaching 35% coverage in the lowest PCV zone, likely due to predominance of urban land-cover. These figures indicate that the presence of IAP’s is not restricted by protected area boundaries and that their occurrence, being fairly widespread across PCV zones and throughout the MBR, would have wide-ranging impacts on the composition, structure and functioning of biodiversity within it (Forsyth & van Wilgen, 2009; Neke
103 & Du Plessis, 2004), and as such provides a number of potential ongoing conservation opportunities in all biosphere zones.
The absence of IAP’s through the dolomite and summit grasslands is noticeable in the data because isolated woody clumps that typify these landscapes are presumably smaller than the minimum mapping unit of analysis, so have not been detected. Herbaceous invaders of Gauteng grasslands, such as pom-pom weed (Campuloclinium macrocephalum) are more likely to spread and persist in grassland environments, with potentially detrimental biodiversity impacts (Driver, et al., 2012). Although O'Connor & Kuyler’s (2009) AHP found greater negative impacts on composition from woody, rather than herbaceous species in South Africa’s moist grassland sub-biome, mapping the spread of herbaceous invaders would be a worthwhile addition to threats analysis in the MBR, were data available.
FRAGMENTATION 241m570m--570m1340m from IAP DENSITY roads/built-up 2-8% T & 9-30% R 50000 30000 45000 91m-240m 9-20% T & 31-55% R 40000 25000 0-90m 21-46% T & 56-93% R 35000 20000 30000 15000 25000 20000 10000
15000 5000
10000 0 5000 0 low PCV med-low PCV med-high high PCV PCV
TRANSFORMED OPEN LAND FIRE forest 1-3 fires 10000 100000 9000 old-lands 10000 savanna fire 8000 degraded/bare absence 7000 1000 recreational grassland fire 6000 grass absence 100 5000 gullies 4000 savanna/grass 7-8 10 fires 3000 2000 1 forest 4-6 fires 1000 0 low PCV med-low med-high high PCV PCV PCV
Figure 31: Exposure per hectare and spatial distribution of habitat degradation threats by sub-criteria, across PCV categories. (The hectare scale for fire is logarithmic).
104 The purpose of the transformed open land layer is to show the varying degrees of alteration and degradation of open space from its pristine natural state. Non built-up areas such as recreational planted grassland, urban open space, old timber plantations and old agricultural fields have been variously transformed, disturbed and managed, but are not necessarily classifiable as degraded land. Typically these are anomalous disturbed areas where planted and self- seeded invasive and exotic herbaceous and/or woody species occur and often dominate. Alternatively they may be managed, planted and/or altered to the extent that they have lost much of their natural character. Moderate or severe degradation and erosion gullies on the other hand, refers to areas of low vegetation cover compared to surroundings – usually a consequence of high densities of livestock or game grazing and browsing (GeoTerraImage, 2009; Hoffman & Ashwell, 2001). Together degraded and eroded areas and transformed open land make up a significant proportion of open space in the MBR and is thus a valuable indicator of vulnerability and threat (fig 31).
The layer achieved an equivalent AHP weighting to cultivation (10%), with the most severe threat class (medium-high threat) being erosion gullies, followed by degraded and bare areas, recreational planted grass, urban-open space, timber plantations and lastly old-agricultural fields all falling in the medium-low threat category. The Magaliesberg region currently has very few erosion gullies and a small proportion of degraded and artificially bare areas, few of which occur in high PCV. Of all degradation threats, high PCV is most prone to transformation from recreational planted grass (>2000 hectares) likely in the form of golf courses and sports/ recreation fields associated with residential and tourist facilities, most of which occur in med-low PCV. The extent of old agricultural fields peaks in the medium-high PCV zone and are concentrated around the perimeter of the MPE (fig 31), This is a recent trend away from farming due to changes in property values, and the lower economic viability of agriculture, and when the boundary of the MPE was proclaimed - originally a nature area in 1977 and later a reduced-size MPE in 1989 (Carruthers, 2015; DACE, 2007).
Anthropogenic or imprudent fire events were ranked least severe threat to biodiversity by the expert panel. In some habitats fires operate as spatiotemporal threats – where negative impact is followed by recovery, resulting in a decrease in threat intensity depending on response (Pressey, et al., 2007). In this instance, all 4 sub-criteria fell in the bottom 5 ranking for intensity of threat, while 3 sub-criteria related to high fire frequency also have very low extents. There are however some small isolated patches where frequent burns occur in certain high PCV in the forests and summit grasslands, and in Egoli Granite Grassland, but the extent of forest patches is smaller than the MMU for burn data, so the assumption that fire is occurring repeatedly in forest is unreliable. On the other hand the absence of fire in savanna and grassland, related to fire suppression, is the top ranked threat in terms of extent > 1600 Km², and this figure does not include built-up areas. As discussed in section 3.2.3, fire suppression changes the grass/tree matrix, resulting in bush thickening in savanna and bush encroachment in grassland. However, annual firebreaks and the road network would curb fire-spread in many cases, which would decrease the size of burns and it is therefore likely that the 105 number of smaller fires(< 21 hectares) is underestimated due to the burn detection algorithm (Giglio, et al., 2009).Despite the low severity ranking of fire as a threat to biodiversity, the fire layer, indicating burn frequency, is useful for fire management where fire anomalies particularly in high PCV can be investigated, better-managed and monitored. Further useful spatial information on season of burn from the MODIS data could enhance research and management of fire as well.
The combined patterns of degradation are shown in fig 32, which confirms how widespread these threats are in comparison to the more concentrated habitat loss threats (fig 30).
Figure 322: Spatial representation of extent and severity of Habitat degradation threats of the MBR. Threats show relative weights for sub-criteria. The most severe degradation threat occurs along river courses where IAP’s dominate. White areas are not threatened and represent mostly remote summit grasslands and other remote grassland with regular fires that are not disturbed nor invaded with woody species.
5.8 Priority Biodiversity Threats Composite (PBT)
The PBT composite (fig 33) is the spatial integration of quantified exposure and intensity of all threat processes and combines by weighted overlay, the 7 AHP-weighted threat layers that have thus far been described in detail in preceding sections. The results of the weighted sum produced a bimodal, highly positively skewed distribution of values for priority threats with a range between 0 and 653, with a large number of low values and a small number of values over 350. The bimodal distribution is a result of the bulk of records concentrated below the mean value of 122 and the urban/fragmentation combined threat forming the second peak.
Classifying the data range into suitable nominal intervals to grade threat was required for graphic information and cross tabulations. The literature on irreplaceability and vulnerability does not emphasize thresholds used to differentiate high and low irreplaceability and vulnerability value in a spatial application. Thus, assigning thresholds to the range of values was informed by methods used
106 by Rouget, et al., (2004) and Wannenburgh, (2006) - a simple clustering into categories using natural breaks in the data, and Vimal, (2012) who applied quintile breaks.
However, in this case, with a highly positively skewed distribution, the categorisation and display of scores suited a geometrical interval (GI), as natural breaks and quintile breaks were not suitable visually. The GI classifier suits continuous data and combines characteristics of other classification types to highlight change in both middle and extreme values in a dataset, to produce “visually appealing and cartographically comprehensive” results (ESRI, Inc., 2010).
A four class classification was used for effective map display but also to minimise the number of potential combinations when the PCV and PTB maps were combined. The classification resulted in a large range of threat records falling in the high threat category, thus potentially masking individual highest scoring threats (Rouget, et al., 2004). As threat layers were not spatially mutually exclusive, a technical issue with Weighted Linear Overlay (WLO) is that it is unclear from final scores alone which multiple criteria inputs are contributing to the outcome (Moilanen, et al., 2009). Also, with WLO criteria cannot be combined intelligently to identify interactions between them, and how variations in threat are influenced by changes in each criteria value, as can be achieved with other linear and nonlinear statistical modelling methods (Wilson, et al., 2005a) . For these reasons it was necessary to retain scores before classification in the composite map and to refer to individual threat layers in the description of results, in addition to the composite.
This said, the classified PTB composite highest threat category covers 1429 km² about 33% of the MBR, the 2 intermediate threat categories, 1715 km² (medium-high threat 31% and medium-low threat 8%) and the lowest threat category covers 1216km² (±28%).
Figure 33: Categorised Priority Biodiversity Threats - Composite of 7 threat layers protected areas (MPE, COH) overlaid.
107 All scores from 149-653 (intense orange/red on the map) are considered high threat. This category would not only include the top-ranked individual threats – urbanisation and mining, but also a composite of threat factors, because of the weighted sum of criteria that are not spatially mutually exclusive. For example the highest scoring fragmentation threat class (within, or proximal to urban areas) and urban threats would overlap, and their scores summed, resulting in higher scores for these pixels in the threats composite. Likewise, if ‘dense IAP invasion’ and ‘high fire frequency in forest’ classes are compounded, the area of overlap would be classified as high threat, as opposed to an intermediate or low-threat respectively, if these threat factors were separated.
However overlaying areas in the composite to individual threats, shows in general that highest threat values coincide with urban and mining land-use, where human population numbers and activities are increased, seen on the map around the periphery of the MBR -in the metropolis of Pretoria, Rustenburg and Krugersdorp in north-west Johannesburg, and the peri-urban and industrial outskirts of these cities. More centrally on this map, high threat values are also a result of high population densities located around the Hartebeespoort dam as well as the densely populated informal settlements of Majakineng in Madibeng. High threat is also concentrated along river courses throughout the region, because of the relatively high AHP ranking for dense IAP’s.
Intermediate threats (yellow and light green) with a class interval from 70 to 148, represent much of the cultivated land as well as compounded degradation threats (fire, terrestrial alien invasions, degraded open spaces and fragmentation). The class occurs throughout the MBR, and commonly shows a spatial pattern of being buffered between high and low threat areas, which could be influenced by the decrease in threat away from built-up areas in the fragmentation layer.
The map clearly shows that areas of low vulnerability to human pressures, i.e. low-threat threat scores of 0 to 69 predominate in protected areas (MPE and COH) – occur in mountainous terrain and the karst dolomites. These areas have historically been underutilized by humans and have thus incurred little opportunity cost for protection under the NEMPAA. Portions of Egoli Granite Grasslands and Zeerust Thornveld share these characteristics and also show low threat values. Therefore it is not surprising that they are earmarked for future protection under the National Protected Areas Expansion Strategy for precisely these reasons (NPAES, 2010). A few small isolated sites around the MBR periphery are also indicated as low threat. These areas are mostly high relief ridges that are seemingly free from proximate human pressures currently.
The composite threat map is unique in that it captures a quantified measure for overlapping threats by incorporating factors of land degradation, that are not usually considered in most biodiversity assessments, with the commonly used first- level classified land-cover data of transformation which does not grade vulnerability/threat. Even though the composite is a spatial generalisation, because data of different scales was included, it provides a more detailed picture of threatening processes of varying intensity that can affect a landscape simultaneously. It also identifies areas that are exposed to multiple threats. Thus, the composite map demonstrates that spatial variation in exposure to
108 different threatening processes of varying intensity can be mapped quantitatively, using a MCA AHP approach.
However, there are limitations and uncertainties inherent in threats assessment, (Pressey, et al., 2007; Moilanen, et al., 2009) as threats do change over time, so combining multiple threats should proceed with caution. For example, Rouget et al (2004) questions whether an area vulnerable to multiple threats is more threatened than an area vulnerable to a single perhaps more severe threat. Furthermore, because of the variation in response to a single, multiple and/or compound threats by different biodiversity features or different locations, at the ecosystem, community or species level (Goudie & Viles, 1998), this spatial assessment provides a landscape scale framework of current threat condition for multiple threats, but could be updated at any time given new data for specific biodiversity components, or to predict future threats.
5.9 Defining Spatial Priorities for Conservation – The Integration of Proirity Conservation Value and Priority Biodiversity Threats
A comprehensive assessment for conservation opportunities and development planning in a biosphere context may reflect a combination of priority biodiversity and priority threats to biodiversity. The PBT composite map and PCV map are combined and cross-tabulated into two final synthesised priority products for this research, so that threats can be spatially referenced to important biodiversity and conservation opportunity.
The results and discussion for objective 1 to objective 4 thus far, have been generated at 50m pixel resolution, which is carried through to a final PCV / PBT integrated map and graphic. Results have not been generalized or filtered, and this is intended to show the nexus of conservation and threat at the finest scale possible. Yet for certain practical applications, such as stakeholder engagements (Driver, et al., 2003), a more generalised implementation unit may be more suitable. This is demonstrated for objective 5 as a second integrated map using original farm portions as planning units.
From these maps it is possible to view priorities for conservation and threats simultaneously, thus summarising the information on conservation and threat into one visual product. They are intended for practical broad scale planning and application of conservation management, and a geographic framework to identify conservation/sustainability opportunities for the area in a variety of forums. The products were inspired by the NSBA (2004) “Average composite vulnerability product” (Rouget, et al., 2004) and the SKEP “Framework for Action” (Driver, et al., 2003) that was intended for regional practical application in conservation management in the Karoo.
5.9.1 Integration by original farm portions
An alternative to pixel based analysis as applied, is the original farm, which is also demonstrated as an implementation unit in this research. Spatial planning decisions requiring landowners consent or
109 cooperation are often determined using farm portions as the spatial planning unit because cadastral boundaries are generally easier and quicker to work with than natural features such as rivers or catchments, or arbitrary units such as pixels, 100ha plots or hexagons (Pierce, et al., 2005). Indeed, one of the requirements for tiered zonation in the MBR application is that private land owners endorse the Biosphere concept, so individual farm portions are very relevant.
Recent spatial conservation prioritization highlights the need to incorporate socio-political data on the commitment and capacity of private land-owners and government agents towards conservation opportunity, because these entities generally have the final say on land-use, and decisions on conservation action (Knight, et al., 2010). Cadastral boundaries may be useful units for social assessment and are also widely applied in spatial development planning, thus it would be practical to base prioritizations on units of tenure (see Li & Nigh, 2011). However, many land portions in the peri-urban MBR landscape are typically small, and as such are not conducive to spatial analysis using coarser scale threat data. Also, Rouget, et al. (2004) suggests land parcels as implementation units may antagonise land owners, but are useful as a backdrop.
For reasons above the parent farm portion as implementation unit is demonstrated here. It provides a framework for quick visual referencing and a more practical management friendly option for identifying conservation priorities between 329 units across the MBR, although summarising conservation and threat averages to a coarser scale resulted in the spatial location of pixel scores being lost (Driver, et al., 2003).
Quantified prioritizations for conservation and threat per original farm can be shown in graphic form. The scatter plot of original farms (fig 34) below, illustrates a feature of many irreplaceability and vulnerability analysis (Margules & Pressey, 2000; Lawler, et al., 2003; Pressey & Taffs, 2001; Noss, 2002). In planning conservation action, farms are grouped into quadrants, based on their clustering, to determine different management responses to single and multiple threats. Identifying combinations of conservation and threat value per farm (or planning unit) can quickly eliminate areas of high threat and low conservation value from conservation priorities, and instead include these areas in development planning (Rouget, et al., 2003).
110
2 1
4 3
Figure 34: Priorities for original farms plotted on two axes.
The Y axis plots the average conservation value of all original farms in the MBR, and the X axis plots the average threat value (composite of 7 threats) to which these farms are exposed. Priority quadrants indicate clusters of farms with various combinations of conservation value and level of threat (Adapted from Margules & Pressey, 2000; Driver, et al., 2003; Lawler, et al., 2003)
Bearing in mind each original farm is made up of one or many private land parcels, and that pixel scores have been classified and averaged per original farm, only a preliminary indication may be made on this basis. Farms located in quadrant 1 may receive a priority ‘red-flag’ conservation response, while farms in quadrant 3 would not be prioritised for conservation action. Responses to Quadrant 2 and quadrant 4 are less obvious, and may be indicated by MBR zone-related conservation goals, according to the management plan for whichever biosphere zone the farm falls under.
The cross classification can be visualised in a map by representing different variations of conservation and threat per farm, using a simple colour scheme of low, intermediate and high values for conservation and threat ( fig 35).
111
CON high
O SERVATI
N
low
THREAT low high
Figure 35: Original farms prioritized for both conservation value and threat.
Intermediate interval categories (medium-low and medium-high) are consolidated, providing a simplified (see fig 36 below) colour reference to distinguish priorities.
The map is a spatial summary of original farms average priority conservation value in relation to their average vulnerability. While the location of pixel scores are lost in the generalisation, (for example the high PCV mountain slopes and crests east of Hartebeespoort dam and extending into the city of Pretoria) ,the scale of farms and the spatial focus to identify responses to priorities is easy to understand and use in an applied setting. The colour classification is a simplification of the one used in fig. 36 below, and is conducive for a clear visual reference that can be referenced to the scatter plot as follows: lime green represents farms in the top left of quadrant 2, dark brown shades represent farms in quadrant 1, white shades indicate farms in quadrant 3 and deep orange quadrant 4. Less intense shades of the above and mustard yellow are intermediate values for both conservation and threat, and indicate farms closer to the intersecting lines of the scatter graph. In viewing the scatter plot, map and attribute table together the top ranked priorities per quadrant can be identified and easily cross-referenced spatially.
It is clear from the map that farms in the south east (grasslands) are on average, of high conservation priority and the most vulnerable to proximal threats (quadrant 1). Farms in the far west (dense bushveld) are clearly not vulnerable to proximal threats, but their conservation value is low (quadrant 3). Farms of low average conservation value and high threat (quadrant 4) are located in or close to the cities of Pretoria, Rustenburg and Johannesburg while farms with low average threats and high conservation value(quadrant 2), are concentrated more centrally and correspond quite closely to the COH and MPE boundaries.
112 5.9.2 Pixel-resolution integration
This section focuses on the final product of this research the integrated maps and cross tabulation of PCV and PBT on a pixel by pixel basis, in order to relate priority threats to important biodiversity at the finest scale possible. Class limits were made simply for purpose of the study to identify high priority sites and spatial variability, using the 4 class GI classified intervals used consistently for all priority maps. It was decided that the resulting scatter plot (appendix 4) produced from the pixel cross-tabulation, was not visually effective because of the large number of records (over 2 million) and it was also not practical or feasible to identify and cross-reference individual pixels that occur in specific sectors of the plot, (as was achieved with the original farm quadrants). Instead, the number of pixels occurring within specific sectors, is converted to km2 and translated into a graphic where the extent of each priority combination can be related to the corresponding map using the same colour scheme, again providing a spatial and graphic cross-reference to identify conservation opportunities and concerns. The map (fig 36) and graphic (fig 37) are the final outputs of this research, and are presented together.
high
401 402 403 404
CONSERVATION
302 302 303 304
201 202 203 204
low
101 102 103 104 low THREAT high
Figure 36: Priority Conservation Value and Priority Biodiversity Threats, The map is displayed using geometrical intervals. For each pixel or planning unit a priority ranking was based on four levels of conservation value 1- low, 2- med low, 3- med high, 4- high, combined with four levels of biodiversity threat 100- low, 200- med low, 300- med high and 400- high. A bi-variate colour scheme (Lawler, et al., 2003) indicates the gradation of conservation value and threat severity, as well as showing all combinations of them. The intensity of Lime-green increases as PCV increases in the absence of threats. The intensity of orange increases with higher threat severity as PCV decreases. Black indicates high PCV and high threats, with other darker shades indicating high threat. White and light hues indicate low values for both conservation and threat. The map is repeated in Appendix 5. 113
Geographic extent of priority threat categories falling within priority conservation zones
3 4 KM² 633,4 700,0 2
600,0 413,6 500,0 351,3 445,7 352,6 424,6 400,0 107,9 404,2 43,7 300,0 118,0 314,0 200,0 225,4 100,0 93,5 200,8 1 86,9 0,0 2 146,0 1 3 2 4 3 4 1
Priority Conservation Q3 1/1 Q3 2/1 Q4 3/1 Q4 4/1 LOW Q3 1/2 Q3 2/2 Q4 3/2 Q4 4/2 MED-LOW
Q2 1/3 Q2 2/3 Q1 3/3 Q1 4/3 MED-HIGH Q2 1/4 Q2 2/4 Q1 3/4 Q1 4/4 HIGH LOW MED-LOW MED-HIGH HIGH Priority Threat
Figure 37: Integrated priority maps and graph of conservation and threat Products are based on the cross-tabulation of 4 classes of PCV and 4 levels of PBT grouped into priority quadrants. The graphic shows the areal extent (Km ²) of the sixteen priority combinations. Colours reflect those of the map above.
The results and maps have shown thus far that all seven threat factors considered, and the combined threat composite, vary spatially in extent and intensity. The variability in the distribution of threats can inform conservation planning about where risks of transformation from different threats are likely to occur. Fig 36 also shows the pattern and spatial relationship between priority conservation and priority threats at the smallest scale possible, to show the location of conservation/development interfaces. The accompanying graphic (fig 37) gives an overall indication
114 of the extent to which priority conservation is vulnerable to threat. To demonstrate a practical application of results here, different combinations of threat and conservation are grouped into priority quadrants, as the framework to initiate conservation responses. In the spirit of a biosphere, most conservation responses would be more or less applicable to all quadrants, while development restrictions would apply to specific quadrants, as indicated for different zones in the biosphere management plan (Contour and Associates, 2013). This would be an example of using spatial planning as a reactionary prioritization approach, to steer development away from valuable biodiversity, in favour of areas that enable sustainable development (Rouget, et al., 2003).
In general, the graphic results of cross tabulations shows that values for threat are lower in high priority conservation areas and that threat values are maximised and concentrated in low priority conservation areas. Results per priority quadrant and some examples of the types of conservation responses most relevant to each are outlined below. The quadrant approach demonstrates the value of spatial planning, for decisions at the conservation and development nexus, where the patterns of development and biodiversity value are an important consideration to ensure long-term sustainability of the area.
Quadrant 1 (high, med-high PCV/ high, med-high PBT)
The total area of the MBR including a 2km buffer is 4361.6 Km 2. Q1 shows areas that are of high and medium-high conservation value that have a high or medium high threat status.Q1 accounts for 20.3% of pixels (887 Km2). Within this quadrant only 3.3% of pixels (146 Km2) are highest priorities for both conservation and threat (Q1 4/4).These pixels are concentrated in the south-east of the study area - an area that continues to be impacted by urban expansion and peripheral activities/processes. Smaller concentrations occur to the north of the Magaliesberg in areas vulnerable to PCM mining. A few Q1 4/4 pixels occur in the COH and around the fringes of the MPE. The threat here may be attributed in part, to dense IAP invasions. None of the secondary data used to determine important areas for biodiversity picked up on the high threat status of the important sites in Q1 4/4, and therefore they are considered top priority in this research.
Q1 is arguably the most important for urgent conservation management and monitoring, and probably requires stricter conservation control to assure future biodiversity conservation and to reduce current threats (Vimal, et al., 2012). However, if after site investigation it is found that transformation cannot be abated, for example existing residential dwellings, business premises, mines or roads within these high PCV zones, then two responses might apply. First, to provide expert information and support on how to maintain remaining biodiversity and improve sustainability in households or operations, for example biodiversity stewardship programmes, recommendations for cleaner production and compliance, and roads impact mitigation. Alternatively it might be worthwhile to abandon conservation efforts in these top priority areas that may be totally transformed, and rather concentrate resources on other less threatened but high priority PCV sites (Wannenburgh, 2006), such as those in Q2.
115 Quadrant 2 (high, med-high PCV/ low, med-low PBT)
Quadrant 2 Q2 covers 937 Km2, or 21.5% of pixels. It shows areas of high and medium-high conservation value, with low and medium-low threat status. Maximum values for conservation occur in this quadrant with 404 km2 of prime conservation land that is not currently threatened. Management responses for a site within or outside of the biosphere core area in this quadrant might point towards a conservation response of ‘least suitable for future development’, which should be considered in the land-use planning process. In this zone EIA applications would be carefully scrutinized, but unpreventable designated developments might be managed to mitigate vulnerability as far as possible, even if this entails difficulties and extra expense (Wilson, et al., 2009). It might be necessary to make additional resources and expertise available through biosphere management to monitor EMP’s, or for example, implement a strategy to work with developers to reduce fragmentation and maintain or expand biodiversity corridors. Considering regional population and development pressures, areas of conservation worth outside of protected areas in Q2, may be prioritised and continually monitored to assess vulnerability, as development dynamics and hence conservation priorities would change over time (Noss, 2002).Given that this zone is not yet impacted by development, it is a zone of potentially low opportunity cost for MAB conservation objectives, which may suit sustainable low-impact activities such as education and outreach, as well as demonstration sites for biodiversity stewardship as this quadrant would include land in conservancies.
Quadrant 3 (low, med-low PCV/ low, med-low PBT)
Q3 covers areas of low to medium-low conservation priority, with low to medium low threat. This is the smallest quadrant by area covering 621 Km2, or 14% of pixels. Ground truthing may reveal why sites in Q3 have low conservation status, however low PCV scores may be attributed to a non- threatened vegetation-type or ecosystem, or a vegetation type that is well represented outside of the MBR. In the context of the biosphere the low threat status of Q3 may be managed to remain that way, with low-key sustainable development. This would maintain a low opportunity cost for future conservation or protection of these areas, in the event that the conservation status of areas outside of the MBR change and areas within the biosphere are then required to meet biodiversity targets. For example focus areas of Zeerust Thornveld and Egoli Granite Grassland in the NPAES (NPAES, 2010). Other conservation opportunities exist in Q3, such as extending less fragmented corridors away from areas of high relief to provide a wider range of climate envelopes and topographic types to ensure resilience to future threats, like climate change. In this study lower value threats are important indicators of degrees of land degradation. They should not be ignored as there may be potential to mitigate these threats, or to improve biodiversity in these areas at lower cost.
116 Quadrant 4 (low, med-low PCV/ high med-high PBT)
This is the quadrant where threats are maximised (in intensity and extent) in low and medium-low conservation priority areas. About 44% of pixels, (1917 Km2) make up Q4, mostly transformed by intensive agriculture, industry and residential development, or a combination of habitat degradation threats, for example, dense invasions close to urban boundaries. Future planning responses might be to densify development in Q4, because the zone is already largely developed where some threats cannot be abated. Concentrating future development in these areas may prevent urban sprawl and further fragmentation. For example, with changes in land values, it would be preferable to convert old lands close to urban areas, rather than virgin ground into a new township for development.
Biosphere principles can be applied, and conservation options are still available in these areas even though they have been designated as having low conservation value, because sometimes opportunities for sustainability are preferable closer to where people live and work, to uplift and benefit surrounding communities. Guidelines for managing the biosphere transition zone, which corresponds spatially to Q4, would support investment in sustainability initiatives for this reason (Contour and Associates, 2013; UNESCO, 1996). A range of interventions can be employed to improve urban ecology, such as removing IAP’s, planting endemic trees that are scarce or threatened, reducing the use of poisons and pesticides and reducing pollution and waste outputs. However, it could also be argued that because vulnerability is high in these areas, conservation resources may be more worthwhile in less transformed (Q3) or higher priority conservation areas (Q1) (Vimal, et al., 2012).
Keeping the quadrant specific responses in mind, inevitably solutions to problems and mitigating threats should be integrated as part of a systematic solution. Solutions are based on a thorough understanding of the complexity of threats to biodiversity features and the environment in general (Goudie & Viles, 1998) and also the socio-political milieu and attitude of land-owners. Therefore, identifying vulnerability and conservation opportunities as has been demonstrated in this study is only the first step, and responses following from the quadrant framework, should be determined on a case by case basis. This quadrant approach highlights the ways in which the spatial analysis can be applied in practical ways to regions with different conservation potential. This is a strategy inherent in the MAB programme.
5.10 Summary
After describing the spatial variation and pattern of Priority Conservation Value (PCV) in relation to formal protected areas, the chapter focuses on regional scale results and ensuing discussion on the exposure and intensity (Wilson, et al., 2005a) of seven current and direct threats to biodiversity and important conservation of the MBR. The extent of threats is quantified geographically, while their intensity is determined by Analytical Hierarchy Process (AHP) (Saaty, 2005), a method used to
117 formalize expert insight into a quantitative record of the relative intensity of threatening processes affecting the region.
The seven threats varied spatially across the MBR landscape, and each threat has different results for exposure and intensity. Results are discussed in the context of other results in the threats and vulnerability literature, and discussion replicates the literature review by distinguishing between the threats associated with habitat loss and those of habitat degradation.
The threats composite (PBT) was then related to PCV to provide insight on the vulnerability of important biodiversity to particular threatening processes. Spatial relationships between priorities for conservation and those for threat were in general unsurprising, revealing that values for threat are lower in high priority conservation areas and that threat values are maximised and concentrated in low priority conservation areas, with some exceptions. Analysing threats with reference to PCV zones also showed that qualitative losses to biodiversity and ecosystem health can pervade protected areas, and that high PCV, is located outside of the protected area network and is vulnerable to threat.
Finally the chapter concludes by illustrating how integrated priority maps and graphics, at different scales can be applied in conservation management and land-use planning for the MBR, with biosphere reserve principles in mind.
118 Chapter 6 Conclusions and Recommendations
6.1 Overview of Research Design and Study Context
Although the research is undertaken in the field of environmental management it is multidisciplinary and relates closely to the specialty of conservation biology. More specifically, study areas that have influenced this research and/or the data used in it, are listed in order of relevance from the general to the specific: remote sensing and image processing; land-cover classification and change; biodiversity, ecosystem and conservation assessment; spatial conservation planning; GIS processing; basic statistical analysis, Multi Criteria Analysis and AHP; spatial prioritization; and threats or vulnerability assessment.
An exploratory and descriptive research design using mixed methods of quantitative analysis was applied to realise the research aims. The methodology combines techniques and procedures from a number of applications and research in the spatial prioritization, site selection and conservation planning arena over the past two decades.
The work was positioned as spatial decision support and can be summarised as a spatial conservation planning exercise that quantifies exposure (by extent) and intensity (by AHP) (Wilson, et al., 2005a) of multiple threat criteria to determine priority threats to biodiversity in a case study. Results were derived from spatially integrating and relating priority threats with priorities for conservation, which established the current vulnerability of the latter to the processes of habitat loss and land degradation and change (Pressey & Taffs, 2001; Rouget, et al., 2003; Noss, 2002; Vimal, et al., 2012). The research design presented a systematic approach to quantifying and analysing threats and referencing them to the terrestrial biodiversity landscape.
This research was carried out in the Magaliesberg region of South Africa, in the context of the area being promoted as a biosphere reserve in the UNESCO MAB programme, which recognises the need to reconcile the conservation of biodiversity and maintain associated cultural values, while pursuing economic and social development (UNESCO, 1996).The reserve area extends 4361.6 km2, its main feature being a quartzite and shale mountain chain, the Magaliesberg. The region is unique in that it lies at the interface of two important and extensive biomes in South Africa, and has varied topographical and climatological gradients, making it home to a rich biodiversity. Its history and cultural heritage is also exceptional.
Notwithstanding past human influence, extensive modifications to landscape have occurred in the last fifty years. As the surrounding cities expand, migration to water-side and country estates as primary residences has boomed, together with holiday homes and tourist resorts as the area continues to be recognised for its scenery and recreational utility. Basic housing and informal settlements have also mushroomed at the periphery of growing cities and in response to an expanding mining industry that attracts a migrant workforce. Extensive hydrological diversion and
119 impoundments has seen an increase in intensive farming operations, but the recent past has seen conversions to game farming and urban land uses, particularly in the central areas close to the mountains and dams. Considering the proposed reserves close proximity to the economic hub of South Africa – the Gauteng City region – these dynamic and increasing pressures of population, development and extraction might be considered in the management of the biosphere and in a comprehensive conservation assessment for the region.
6.2 Research Shortcomings, Achievements and Recommendations
Spatial Conservation assessments have increasingly incorporating aspects other than biodiversity criteria, such as amongst others, vulnerability and threat as factors to consider when determining conservation priorities and resolving spatial conservation problems in a changing world (Moilanen, et al., 2009). In many studies a combination of approaches are applied to integrate data and assess threats or vulnerability. One general approach (Wilson, et al., 2005a) is informal qualitative or quantitative analysis, such as spatial scoring based techniques, like MCA weighted overlay (Li & Nigh, 2011; Moffett, et al., 2006b) or correlation and rule-based modelling (Veech, 2003; Vimal, et al., 2012; Rouget, et al., 2003). The second approach is more formalized statistical modelling, such as regression tree analysis (Wilson, et al., 2005 b; Rouget, et al., 2003). Alternatively, complemetarity based algorithms, incorporate cost factors such as threats or vulnerability into conservation planning outcomes (Moilanen, et al., 2009).
Yet, despite many advances in the spatial prioritization space, Wilson et al’s, (2005a) review of vulnerability assessment indicates it remains a difficult task, with most approaches experiencing problems related to assumptions about data and uncertainties inherent in predictive analysis and around multiple criteria and aggregation of data, not to mention the lack of validation of methods and results. Knight, et al., (2010) maintains that mapping vulnerability is still “overly simplistic”, compared to other “reactive” factors such as economic-cost, and ecological variables, where approaches have advanced rapidly.
In terms of this research, the attempt to prioritise conservation opportunities and threats to biodiversity as a comprehensive assessment with meaningful results proved to be complex. However, the inclusion of multiple threats and combining formally quantified scores for these threats, using a basic additive model, was a simple and robust method to assess the spatial patterns of single and composite threats, in relation to biodiversity which has proved to be the first comprehensive spatial assessment for the entire MBR region.
Despite having managed to address the objectives stated in Chapter One using a defensible and systematic research design, challenges and limitations were encountered with data and methods. These have been referred to throughout the text at relevant sections where appropriate. A few shortcomings are reiterated here specifically to allow for improvements to be made in future research.
120 A number of assumptions have been made regarding many aspects of this work. The single level hierarchy AHP assumes all components of biodiversity are equally vulnerable to a given threat. This is clearly simplified as sensitivity of ecosystems and different biodiversity features to human pressure tends to vary (Noss, 2000; Goudie & Viles, 1998). Therefore, a more robust AHP investigating select biodiversity components or features would assess threat impacts more comprehensively. Alternatively, a similar AHP to O'Connor & Kuyler’s, (2009) could investigate landscape elements of different habitat-types within the MBR as indicators of the relative impact of threat factors on biodiversity integrity. Data permitting, these investigations would have extensive scopes, if their results could be applied spatially. In a similar fashion, the study could have been strengthened with an extended AHP hierarchy which included the sub-criteria threats as the preceding order in the hierarchy, as these would then have been ranked by AHP instead of multiplying their mean scores by the relative rank of the threat criteria, as was the case in the method used.
It is assumed that the applied threat criteria are accurate in extent, and that they accurately reflect the vulnerability of terrestrial biodiversity currently. However, there is usually always some uncertainty with image classification that must be borne in mind. Also, it is acknowledged that in the absence spatial data to account for other threats, (for example resource harvesting and trapping, the application of poisons, herbaceous and faunal invasions, bush encroachment), only a portion of potential vulnerability is addressed and mapped.
In order to achieve this multi criteria analysis it was necessary to assume that similar values for combined threat scores in different areas are equivalent in intensity, in the absence of statistical modelling to address intelligent aggregations of different criteria. Wilson, et al., (2005a) suggests that formal quantitative modelling statistical analysis would be an improvement in this regard, for example, multiple threat interactions, or how variations of multiple threats and environmental indicators could influence and predict vulnerability.
Thresholds used to grade threat intensity and conservation worth into classes are arbitrary. The step was necessary to achieve the integration of data, but the continuous gradation of scores was lost, thus it becomes important to consider the range in each class when thresholds are fixed. Class intervals were made simply for purpose of the study. First, to contain the number of possible combinations of threat and conservation priorities, and second to indicate spatial variability effectively on a map, given the skewed distribution of PBT data. Thus, a different priority combination may apply to a given planning unit if thresholds were varied. There is not much literature to guide threshold limits in a spatial context, and a review of current practice would be worthwhile.
The research relies on existing spatial data to produce layers for threat and biodiversity, and most but not all secondary data were assessed for accuracy. Also the amalgamation of disparate spatial information (generated using different methods and at different scales) has likely obscured patterns and affected accuracy to some extent, for example overestimation of coverage of some factors, notably fire and IAP invasion extent due to their coarse scale compared to classified land-cover data,
121 and provincial biodiversity assessment that were mapped using different methods and at different scales to each other and to the coarse resolution of the biosphere zonation, which was ultimately demarcated by physical and cadastral boundaries and roads. One of the strengths of this research is the application of multiple criteria which would reduce the effect of an error within an individual dataset. The research achieves accuracy at the regional scale, for conservation intervention.
The research as it stands should be considered a snapshot, a status-quo of current threat status and conservation condition, as it is limited to extent of threatening processes in historical and current space, and not the likelihood or probability of an area being exposed to a threatening process over time, using probability models or statistics to indicate future vulnerability (Reyers, 2004; Rouget, et al., 2003; Wilson, et al., 2005 b). An interesting spatial study to predict future vulnerability would be to geographically reference mining-permit applications, or EIA applications for developments to PCV zones of the MBR.
Original farm priorities for conservation and threat are illustrative here. Individual properties would be a useful unit of assessment as transformation mostly takes place at this scale, with a land-owner committing to a particular land-use (O'Connor & Kuyler, 2009). For this reason if socio-economic data and attitude assessment, towards stewardship, for example, were linked at the property level, a good indication of conservation opportunity may emerge (Knight, et al., 2010) and would be recommended to support the systematic conservation planning of the MBR.
The types of data used to indicate or predict vulnerability or threat to biodiversity range from past and present land-use/land-cover coverage, atlases of resources or land-use potential, and environmental, ecological, social or demographic variables. It is evident from the literature that many threats assessments use only one, or surprisingly few vulnerability indicators and these are often arbitrarily quantified. Also, integrated biodiversity or conservation assessments that incorporate threat as a cost layer, usually only include variables of quantitative habitat loss, and not qualitative factors of land degradation (Noss, 2000). This research has succeeded to collate spatial data on several predominant threats for the region (including less common indicators of degradation) in a multiple threat composite, and to assess the spatial variation of each threat and a threat composite in relation to conservation priorities. In addition the spatial extent and intensity of these threats have been formally quantified by experts in a regional context, for the case study. Basic exposure and intensity dimensions are applied to demonstrate two ways that threats may be quantified, and shows that using more than one dimension, provides an additional perspective for regional results.
Often cost variables are binary in that an area is either lost to conservation or not. In this research threatening processes are quantified more thoroughly by incorporating the key element – intensity, in addition to quantifying geographical extent which provides more detail on levels of threat, than many other threat cost layers used in SCP. In the same way a binary approach is sometimes applied to protected areas as ‘not threatened’. Analysing threats with reference to PCV zones, has shown the
122 general spatial variability of vulnerability of biodiversity of the region and highlighted that qualitative losses to biodiversity and ecosystem health has pervaded protected areas.
Each threat has different results for exposure and intensity, and it is interesting to note the general pattern that has emerged for the region, i.e. that threats are generally concentrated and severe (less exposure, more intensity), or widespread and less severe (more exposure, less intensity). This can be explained as the distribution and exposure of habitat degradation threats being more widespread than those of habitat loss, (which tend to be concentrated in specific areas).Including criteria on qualitative declines in habitat, and measuring both extent and intensity of threat has improved estimations of vulnerability from a regional perspective.
Furthermore, although habitat degradation represents qualitative decline in habitat they are largely spurred by habitat loss processes, for example clearing for development causes compounding effects, not only the loss of biodiversity on site, but resulting increases in fragmentation effects alien invasions and altered land management such as fire suppression and water channelization, that contribute to degradation (Goudie & Viles, 1998).It is likely that threats would have been underestimated by considering quantitative habitat loss factors only.
The systematic steps used in this work are repeatable and given new or updated data, models can be rerun. The steps are also replicable in other contexts, data dependant. The availability of relatively fine scale provincial LCC data in South Africa is an ideal base-line surrogate for threat, to which other data may be added if available, as is demonstrated in this research. In some instances detailed work on single threat can inform management and monitoring. In this case the fire layer provides new information for the area in an accessible form for use by various stakeholders, which. is useful for fire management particularly where fire anomalies occur in high PCV.
6.3 Conclusion
This study has attempted to identify patterns of threat in relation to priorities for conservation. The distribution of PCV is archetypal of a biosphere zonation with centrally located important biodiversity core areas–relatively pristine remaining wilderness in a natural and unfragmented state (about 20% of the area) - surrounded by a gradation of human interventions.
As a dimension of vulnerability of priority conservation (Wilson, et al., 2005a) the geographic coverage of threats and the proximity to roads and built-up areas, indicates current exposure to threat,. Results for exposure show direct habitat loss - urbanisation, cultivation and mining cover 17% (726km²), 14% (627km²) and 2% (86km²) of the MBR surface area respectively. The distributions of these threats in relation to priority conservation indicate a negative linear relationship - they decrease in extent as PCV increases. For example, where low PCV is comprised 28% cultivated and 34% urban land, high PCV is only exposed to 2% and 6% of the same land-uses respectively.
123 The key drivers of habitat degradation are fragmentation, alien invasive plant spread and inappropriate fire regimes extending over 80% (3484km²), 42% (1829km²) and 39% (1715km²) of the area respectively. These are less concentrated but have a wider footprint than direct loss factors and their distribution patterns in relation to conservation are more variable. Notable there are positive correlations between extent of the most dense IAP invasion class and high conservation areas, as well as greater extents of fire suppression in grasslands in high conservation areas. Degradation threats pervade all high conservation zones, with proportionately more exposure in high PCV categories, compared to direct loss. Together these degradation threats with disturbed land factors make up a significant proportion of open space in the MBR and are thus a valuable indicator of vulnerability and threat.
The intensity of a threat, being the second dimension of vulnerability, includes levels of IAP density and fragmentation and categories of injudicious fire events, urbanisation, mining, cultivation and degradation/disturbance used in this research. Intensity is quantified as AHP relative weightings for threat criteria (Saaty, 1977) multiplied by sub-criteria scores.
The consistency index of the AHP was within accepted limits, with the relative weights for intensity being , mining 31%, urbanisation 23%, Alien Invasive Plant spread 13%, cultivation on par with land degradation at 10%, fragmentation 9% and lastly injudicious fire regimes 4%.Weights were multiplied by sub-threat criteria scores and the top 50th percentile of entries, considered to be high and medium-high threat, included all mining and urban classes, followed by the highest two IAP invasion categories, large scale cultivation, erosion gullies and fragmentation within 240 m of built up areas. All other categories of transformed open land, small-scale cultivation, injudicious fire events and the lowest IAP density and fragmentation were ranked below the 50th percentile for threat intensity. They are considered to be medium-low and low threat processes in the MBR currently.
Results are obviously location specific, but reiterate what is confirmed in the literature that habitat loss in in landscapes with high population densities and intense resource extraction, followed by spread of alien invasive plants are the greatest threats to biodiversity (Fuller, et al., 2010; Wilcove, et al., 1998; Noss, 2000; Rouget, et al., 2003; Rogers, et al., 2010).
The PBT composite is the spatial integration of quantified exposure and intensity of all threat processes. This map is synthesised with PCV into one map (fig 37), so that priority threats can be spatially referenced to priorities for biodiversity and conservation opportunity. The spatial relationships of exposure to threat and intensity of threat within each PCV zone, provides detail on the vulnerability of each zone to particular threatening processes. This is important considering the premise that all PCV zones need to maintain (and in some cases even improve) their existing biodiversity features by active management, and sustainably integrating the human element while limiting risk of deterioration.
124 Accordingly, the framework to initiate conservation responses is based on the vulnerability of priority conservation. Different combinations of PBT and PCV are grouped into priority quadrants (see fig, 37). The most extreme combinations are highlighted here to conclude results. Pixels with high scores for both conservation and threat (Q1 4/4) cover 146Km2, which means that 17% of the high PCV zone is severely threatened. None of the secondary data used to determine important areas for biodiversity picked up on the high threat status of the important sites in Q1 4/4, and therefore they are considered top priority in this research. Maximum values for PCV occur in quadrant 2, specifically (Q2 1/4), with 404 km2 of prime conservation land that is not currently threatened, making up 48% of the zone. Conversely only 108km2 or 9% of low PCV is not threatened (Q3 1/1), while 633km2 or 53% is pervaded by high threats. Intermediate combinations of PBT and PCV are are important indicators of land degradation, that have not been taken into account in most biodiversity assessments. To conclude, results clearly show that values for threat are lower in high priority conservation areas that are largely formally protected, and that threat values are maximised and concentrated in low priority conservation areas, indicating a growing disparity between vulnerability to habitat loss and formal protection. In urban landscapes, conservation is less achievable because the cost of protection involves high opportunity costs (Pierce, et al., 2005). This becomes a problem when not all remaining biodiversity is represented in existing protected areas.
Priority quadrants may help to steer unsustainable development away from high PCV and also direct conservations efforts to the most suitable sites (Pierce, et al., 2005). Maps represent a current baseline of biodiversity value and the processes that threaten it, in the Magaliesberg regional landscape, close to the time that it may be nominated as a biosphere reserve. In the absence of a spatial plan for the entire biosphere that is aligned to bioregional plans (Pool-Stanvliet, 2013), these products can assist to integrate strategic conservation planning into decision making for development applications in the context of the new SPLUMA, NEMA and EIA legislation, and may be viewed in conjunction with existing municipal Spatial Development Plans, Integrated Development Plans, Environmental Management Frameworks, bioregional plans, and more specifically the management plan developed for the MBR (Contour and Associates, 2013).
A limitation of all threats and vulnerability spatial approaches is difficulty in considering ultimate threats because these factors are highly dynamic and they modify local ecosystems, so vulnerability of biodiversity remains somewhat unpredictable. Examples of ultimate threats are population migration, global markets and trends, government policies and institutional capacity and climate change. (Wilcove, et al., 1998; Wilson, et al., 2005 b; Pressey, et al., 2007; Noss, 2000). In particular, through the process of this research institutional incapacity has been cited as one of the biggest challenges to conservation in South Africa, for two reasons (Turpie, 2003; DACE, 2009; Pierce, et al., 2005; Knight, et al., 2010). First, a lack of awareness of the importance of biodiversity for economic and social sustainability at the municipal level (Pierce, et al., 2005), and second, that government emphasises development needs, rather than biodiversity and the environment, because of the high levels of poverty and unemployment in South Africa (Turpie, 2003).
125 To this end systematic conservation assessment in the spirit of a biosphere reserve should identify ‘conservation opportunity’, and include willing and able institutions and individuals in spatial priorities for conservation (Knight, et al., 2010). It is hoped that the biosphere management plan, complimented by this spatial prioritization of threats and conservation can support the UNESCO Man And Biosphere programme to mobilise institutions and individuals for mutual benefit from conservation opportunities in this wonderfully diverse Magaliesberg Biosphere Reserve.
126 Reference List
DACE, 2007. Magaliesberg protected environment: Environmental Framework and Management Plan, Mmabatho: North West Province Department of Agriculture, Conservation and Environment.
Abbitt, R., Scott, J. & Wilcove, D., 2000. The geography of vulnerability: incorporating species geography and human development patterns into conservation planning. Biological Conservation, 96(2), pp. 169-175.
ADU, 2013. Animal Demography Unit - Virtual Museum. [Online] Available at: http://vmus.adu.org.za/vm_sp_list.php [Accessed 29 01 2013].
AGIS, 2007. Agricultural Geo-Referenced Information System. [Online] Available at: http://www.agis.agric.za/agismap_atlas/AtlasViewer [Accessed 24 04 2012].
Aucamp, P., 2009. Environmental Impact Assessment : A practical guide for the discerning practitioner. Pretoria: Van Schaik.
Bai, Z., Dent, D., Olsson, L. & Schaepman, M., 2008. Proxy global assessment of land degradation. Soil Use and Management, Volume 24, pp. 223-234.
Batisse, M., 2003. Developing and focussing the Biosphere Reserve concept. In: B. Thukur, ed. Perspectives in resource management in developing countries. New Delhi: Concept Publishing Company, pp. 160-177.
Berger, L. H. B. B., 2006. A guide to Sterkfontein, the Cradle of Humankind. Cape Town: Struik Publishers.
Biodiversity GIS, 2007. National spatial Biodiversity Assessment 2004 : River Ecosystem status. [Online] Available at: http://bgis.sanbi.org/NSBA/riverStatus.asp [Accessed 10 7 2012].
Birdlife International, 2013. Important Bird Areas Factsheet: Magaliesberg and Witwatersberg. [Online] Available at: http://www.birdlife.org/datazone/sitefactsheet.php?id=7093 [Accessed 29 01 2013].
127 Boschetti, L., Roy, D. & Hoffmann, A., 2009. MODIS Collection 5 Burned Area Product - MCD45 - Users guide Ver. 2.0, Maryland: University of Maryland.
Burgess, N., D’Amico Hales, J., Ricketts, T. & Dinerstein, E., 2006. Factoring species, non- species values and threats into biodiversity prioritisation across the ecoregions of Africa and its islands. Biological Conservation, Volume 127, pp. 383-401.
Carruthers, J., 2002. From ‘land’ to ‘place’: communities and conservation in the Magaliesberg area, South Africa. African Historical Review, 34(1), pp. 72-103.
Carruthers, V., 2015. The Magaliesberg. Biosphere ed. Pretoria: Protea Book House.
Chandio, I., Matori, A.N.B.,Wan Yusof, K.B., Talpur, M.A., Balogun, A., Lawal, D.U., 2012. GIS- based analytic hierarchy process as a multicriteria decision analysis instrument: a review. Arab Journal Of Geoscience. DOI 10.1007/s12517-012-0568-8
Child, G., 2004. Growth of Modern Nature Conservation in South Africa. . In: B. Child, ed. Parks in Transition: Biodiversity, rural development, and the bottom line.. London.: Earthscan.
Cilliers, S. S., Muller, N. & Drewes, E., 2004. Overview on urban nature conservation: situation in the western-grassland biome of South Africa. Urban Forestry & Urban Greening, Volume 3, pp. 49-62.
Clark, V., Barker, N. & Mucina, L., 2011. The Great Escarpment of southern Africa : a new frontier for biodiversity exploration. Biodiversity and Conservation, 20(12), pp. 2543-2561.
Contour and Associates, 2013. Magaliesburg Biosphere- Proposed Management Plan 9 Sept.2013, Mmabatho: North West Department of Economic Development, Environment ,Conservation and Tourism.
Cooper, B., 2010. Assessing the viability of using traditional Maximum Likelihood Classification and Change Detection methods for land-cover analysis - A case study of the Cradle of Humankind, World Heritage Site using multispectral imagery from 1999 and 2009. University of Johannesburg: [Unpublished Research].
DACE, 2009. North West Province Biodiversity Conservation Assessment Technical Report, Version 1.2, Mmbatho: North West Department of Agriculture, Conservation, Environment and Rural Development. de Bruyn, C., 2007. Mining, An Industry Overview : Engagement with the platinum, ferrochrome and other mining industries in the North West 1998 to 2007. African Wildlife, 61(2).
128 DEA, Notice 495, 2012. RSA National Environmental Management : Air Quality Act, 2004: N0 39 OF 2004 : Declaration of the Waterberg National Priority Area , No.35435, Pretoria: Government Gazette.
Department of Environmental Affairs, 2012. Magaliesberg Biosphere Nomination, Mmabatho: North West Department of development,environment,conservation and tourism.
Department of Environmental Affairs, 2014. National Environmental Management: Biodiversity Act NO 10 of 2004: Alien and Invasive Species List, 2014: No. 37886, Pretoria: Government Gazette.
Driver, A., Desmet, P., Rouget, M., Cowling, R., Maze, S., 2003. Succulent Karoo Ecosystem Plan : Biodiversity Component Technical Reprt, Cape Town: Botanical Society of South Africa.
Driver, A., Sink, K.J., Nel, J.N., Holness, S., 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, Pretoria: South African National Biodiversity Institute and Department of Environmental Affairs.
Durand, J. M. F., 2010. The threat of mine effluentto the UNESCO status of the Cradle of Humankind World Heritage Site. The Journal for Transdisciplinary Research in Southern Africa, July, 6(1), pp. 73-92.
Eber, S., 2005. Climbing the mountain : urban development in the Magaliesberg protected natural environment. The South African Journal of Environmental Law and Policy, 12(1).
Erlich, P., 1988. The Loss of Diversity - Causes and Consequences. In: E. Wilson, ed. Biodiversity. Washington DC: National Academy Press, pp. 21-28.
Eskom, 2008. SPOT Building Count 2008, Eskom Dwelling Inventory, Johannesburg: Eskom/CSIR Collaboration.
ESRI, Inc., 2010. ArcGIS Desktop Help Library. Redlands, California: ESRI.
ESRI, 2012. ARCInfo 10.0. Redlands, California: ESRI.
Essa, A., 2012. Aljazeera indepth feature : South Africa miners complain of 'living hell'. [Online] Available at: http://www.aljazeera.com/indepth/features/2012/09/201298131558961628.html [Accessed 18 11 2013].
Fahrig, L., 2003. Effects of habitat fragmentation on Biodiversity.. Annual Review of Ecology, Evolution and Systematics, Volume 34, pp. 487-515.
129 Fatti, P., 2013. Statement on the Kgaswane Country Lodge Magaliesberg Protected Environment, Final Report, Johannesburg: Magaliesberg Protection Association.
Fatti, P., 12 November 2012, Emeritus Professor, School of Statistics & Actuarial Science, University of the Witwatersrand, Informal Interview with Belinda Cooper.
Ferrier, S., 2002. Mapping Spatial Pattern in Biodiversity for Regional Conservation Planning: Where to from Here?. Systematic Biology, 51(2), pp. 331-363.
Ferrier, S. & Wintle, B., 2009. Quantitative approaches to Spatial Conservation Prioritization : Matching the solution to the need. In: A. Moilanen, K. Wilson & H. Possingham, eds. Spatial Conservation Prioritization - Quantitative methods and computational tools. New York: Oxford University Press, pp. 1-15.
Forsyth, G. L. M. D. & van Wilgen, B., 2009. Prioritising quaternary catchments for invasive alien plant control within the Fynbos and Karoo biomes of the Western Cape province, Stellenbosch: CSIR Natural Resources and the Environment.
Frost,P., 26 June 2012, Remote Sensing Specialist CSIR-Meraka Institute, email communication with Belinda Cooper.
Fuller, D., Meijaard, E., Christy, L. & Jessup, T., 2010. Spatial assessment of threats to biodiversity within East Kalimantan, Indonesia. Applied Geography, Volume 30, pp. 416-425.
GDACE, 2008. Gauteng Department of Agriculture, Conservation and Environment: Mining and Environmental Impact Guide, Johannesburg: Gauteng Department of Agriculture, Conservation and Environment.
GDARD, 2011. Technical Report : Gauteng Conservation Plan Version 3.3, Johannesburg: Gauteng Department of Agriculture and Rural Development .
Geneletti, D. & van Duren, I., 2008. Protected area zoning for conservation and use: A combination of spatial multicriteria and multiobjective evaluation. Landscape and Urban Planning, Volume 85, pp. 97-110.
GeoTerraImage, 2008. North West Province 2006 Land-Cover Project - Summary MetaData and End-User Notes, Pretoria: GeoterraImage (Pty) Ltd, South Africa.
GeoTerraImage, 2009. Gauteng land-cover dataset - Summary Report & Metadata., Pretoria: GeoTerraImage (GTI) Pty Ltd.
Giglio, L., Loboda, T., Roy, D.P., Quayle, B., Justice, C.O., 2009. An active-fire based burned area mapping algorithm for the MODIS sensor. Remote Sensing of Environment , Volume 113, pp. 408-420.
130 Gill, K. & Engelbrecht, A., 2012. Wild Flowers of the Magaliesberg, Sandton, Johannesburg: Kevin Gill.
Goepel, K., 2012. AHPcalc excel template. [Online] Available at: http://bpmsg.com [Accessed 10 2012].
Goudie, A. & Viles, H., 1998. The Earth Transformed : An introduction to Human Impacts on the Environment. Oxford: Blackwell Publishers.
Govender, N., Trollope, W. & Von Wilgen, B., 2006. The effect of fire season, fire frequency, rainfall and management on fire intensity in savanna vegetation in South Africa.. Journal of Applied Ecology, Volume 43, pp. 748-758.
Hansen, A. J., Knight, R L, Marsluff, J M, Powell, S, Brown, K, Gude, H, Jones, K., 2005. Effects of Exurban Development on Biodiversity: Patterns, Mechanisms, and Research Needs. Ecological Applications, 15(6), pp. 1893-1905.
Hoffman, T. & Ashwell, A., 2001. Nature Divided: Land Degradation in South Africa. Capre Town: University of Cape Town Press.
Howard, B., 1996. The Magaliesberg: Under Threat?. African Wildlife, 50(6), pp. 15-16.
Jaganath, C., 2009. The Cumulative Effect of Wetland Degradation on Water Quality at a Landscape Scale, Durban: Unpublished Masters Thesis.
Knight, A., Cowling, R., Difford, M. & Campbell, B., 2010. Mapping human and social dimensions of conservation opportuniy for the scheduling of conservation opportunity on private land. Conservation Biology, 24(5), pp. 1348-1358.
Kolver, L., 2013. Mining Weekly : SA mining challenges about far more than Marikana, migrant labour. [Online] Available at: http://www.miningweekly.com/article/sa-mining-challenges-about-far-more-than- marikana-migrant-labour-manuel-2013-08-30 [Accessed 13 11 2013].
Kotzé, J., Beukes, B., Van den Berg, E. & Newby, T., 2010a. National Invasive Alien Plant Survey Report Number: GW/A/2010/21, Pretoria: Agricultural Research Council - Institute for Soil, Climate and Water.
Kotzé, J., Beukes, B., Van den Berg, E. & Newby, T., 2010. National Invasive Alien Plant Survey – DATA SET, Pretoria: Agricultural Research Council: Institute for Soil, Climate and Water, (CD ROM).
131 Kuˇsov'a, D., T¡e¡sitel, J., Mat¡ejka, K. & Bartoˇs, M., 2008. Biosphere reserves—An attempt to form sustainable landscapes A case study of three biosphere reserves in the Czech Republic. Landscape and Urban Planning , Volume 84, pp. 38-51.
Kuhn, B., 2014. A preliminary assessment of the carnivore community outside Johannesburg, South Africa : short communication. South African Journal of Wildlife Research, 44(1), pp. 95-98.
Kwaku Kyem, P., 2011. Natural Resource Conflicts, Their Managment and GIS Applications. In: T. Nyerges, H. Couclelis & R. McMaster, eds. The Sage Handbook of GIS and Society. London: Sage Publications limited.
Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A, Bruce, J.W., Coomes, O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J., Leemans, R., Lin, X., Morano, E.F., Mortimore, M., Ramakrishnan, P.S., Richards, J., 2001. The causes of land- use and land-cover change: moving beyond the myths. Global Environmental Change, Volume 11, pp. 261-269.
Lawler, J., White, D. & Master, L., 2003. Integrating Representation and Vulnerability: Two Approaches for Prioritizing Areas for Conservation. Ecological Applications, December, 13(6), pp. 1762-1772.
Lightstone, 2010. Demprokey: Polygon files Census 2001 and updated 2010 income and demographics per EA (Enumerator Area from Census 2001) and 2011 ward boundaries., Johannesburg: Lightstone.
Liu, J. & Taylor, W., 2002. Coupling landscaape ecology with natural resource management: Paradigm shifts and new approaches. In: J. Liu & W. Taylor, eds. Integrating Landscape Ecology into Natural Resource Management. Cambridge: Cambridge University Press, pp. 1-20.
Li, Y. & Nigh, T., 2011. GIS-based prioritization of private land parcels for biodiversity conservation:A case study from the Current and Eleven Point Conservation OpportunityAreas, Missouri. Applied Geography 31 (2011) 98e107, Volume 31, pp. 98-107.
Long, D. & Hoogendoorn, G., 2013. Second Home Owner Perceptions of Their Environmental Impacts: The Case of Hartbeespoort. Urban Forum.
Loubser, A., 2007. Granite Mining. African Wildlife, 61(2).
Louw, C., 6 December 2010, Former Kromdraai Resident and Land Owner, Informal Interview with Belinda Cooper.
Lumley, S., 2002. Sustainability and Development in Less Developed Countries - Immolating the Future?. England: Ashgate Publishing.
132 MacFadyen, D., 2007. Gautengs Biodiversity - The time to conserve is now. African Wildlife, 61(2).
Malczewski, J., 2004. GIS-based land-use suitability analysis: a critical overview. Progress in Planning, Volume 2, pp. 3-65.
Malczewski, J., 2010. Multiple Criteria Decision Analysis and Geographic Information Systems. In: M. Ehrgott, ed. Trends in Multiple Criteria Decision Analysis, International Series in Operations Research and Management Science 142. s.l.:Springer Science+Business Media, pp. 369-395.
Mararakanye, N. & Le Roux, J., 2011. Manually Digitizing of Gully Erosion in South Africa Using High Resolution SPOT 5 Satellite Imagery at 1: 10 000 Scale, Pretoria: Department of Agriculture, Forestry and Fisheries.
Margules, C. & Pressey, R., 2000. Systematic conservation planning. Nature, Volume 405, pp. 243-253.
McKinney, M., 2002. Urbanization, Biodiversity and Conservation. BioScience, 52(10), pp. 883- 890.
Millenium Ecosystem Assessment, 2005. Ecosystems and Human Well-being: Biodiversity Synthesis., Washington DC: World Resources Institute.
Moffett, A., Dyer, J. & Sarkar, S., 2006b. Integrating biodiversity representation with multiple criteria in North-Central Namibia using non-dominated alternatives and a modified analytic hierarchy process. Biological Conservation, Volume 126, pp. 181-191.
Moffett, A. & Sarkar, S., 2006a. Incorporating multiple criteria into the design of conservation area networks: a minireview with recommendations. Diversity and Distributions, Volume 12, pp. 125-137.
Moilanen, A., Possingham, H. & Polasky, S., 2009. A mathematical classification of Conservation Prioritization Problems. In: A. Moilanen, K. Wilson & H. Possingham, eds. Spatial Conservation Prioritization - Quantitative methods and computational tools. New York: Oxford University Press, pp. 28-41.
Moilanen, A., Wilson, K. & Possingham, H. eds., 2009. Spatial Conservation Prioritization - Quantitative methods and computational tools. New York: Oxford University Press.
Mucina, L. & Rutherford, M. eds., 2010 . The Vegetation of South Africa, Lesotho and Swaziland. (CD Set) ed. Pretoria: Strelitzia 19, South African National Biodiversity Institute.
133 Murphy, D., 1988. Challenges to Biological diversity in urban areas. In: Wilson.E.O., ed. Biodiversity. Washington DC: National Academy Press.
Nega, T., Fu, W. & Vrtis, G., 2010. Open space index: a GIS-based tool for assessing human penetration. Local Environment, 15(8), pp. 743-759.
Neke, K. & Du Plessis, M., 2004. The Threat of Transformation : Quantifying the Vulnerability of Grasslands in South Africa. Conservation Biology, 18(2), pp. 466-477.
NEMA, 1998. RSA National Environmental Management Act: NO 107 OF 1998. Pretoria: Government Printer.
NEMBA, 2011. RSA National Environmental Management: Biodiversity Act (1 0/2004): National list of ecosystems that are threatened and in need of protection : No.35435, Pretoria: Government Gazette.
NEMBA, 2015. National Environmental Management: Biodibersity Act (Act 10 of 2004): Threatened or Protected Species Regulations, Pretoria: Government Gazette.
NEMPAA, 2003. RSA National Environemtal Management : Protected Areas Act : no 57 of 2003. Pretoria(Cape Town): Government Gazette.
Norris, K., 2008. Agriculture and biodiversity conservation: opportunity knocks. Conservation Letters, Volume 1, pp. 2-11.
Noss, R., 2000. High-risk ecosystems as foci for considering biodiversity and ecological integrity in ecological risk assessments. Environmental Science and Policy, Volume 3, pp. 321-332.
Noss, R., 2002. A Multicriteria Assessment of the Irreplaceability and Vulnerability of Sites in the Greater Yellowstone Ecosystem. Conservation Biology, 16(4), pp. 895-907.
NPAES, 2010. National Protected Areas Expansion Strategy for South Africa 2008, Pretoria: Government of South Africa.
Oberholster, P. J. & Ashton, P. J., 2008. State of the Nation report: An overview of the current status of water quality and eutrophication in South African rivers and reservoirs., Pretoria: Parliamentary Grant Deliverable. Council for Scientific and Industrial Research (CSIR).
O'Connor, T. & Kuyler, P., 2009. Impact of land use on the biodiversity integrity of the moist sub- biome of the grassland biome, South Africa. Journal of Environmental Management, Volume 90, pp. 384-395.
134 Ogunronbi, S., 2014. Power Point Presentation: Spatial Planning and Land Use Management Act : 2013, Pretoria: Rural Development & Land Reform : Branch Spatial Planning & Land Use Management.
Ololade, O., Annegarn, H., Limpitlaw, D. & Kneen, M., 2008. Land-Use/Cover mapping and change detection in the Rustenburg Mining Region using Landsat images.
Orsi, F. & Geneletti, D., 2010. Identifying priority areas for Forest Landscape Restoration in Chiapas (Mexico):An operational approach combining ecological and socioeconomic criteria. Landscape and Urban Planning, Volume 94, pp. 20-30.
Phua, M. & Minowa, M., 2005. A GIS-based multi-criteria decision making approach to forest conservation planning at a landscape scale:a case study in the Kinabalu Area, Sabah, Malaysia. Landscape and Urban Planning , Volume 71, pp. 207-222.
Pierce, S.M., Cowling, R.M., Knight, A.T., Lombard, A.T., Rouget, M., Wolf, T., 2005. Systematic conservation planning products for land-use planning: Interpretation for implementation. Biological Conservation, Volume 125, pp. 441-458.
Pimentel, D. & Kounang, N., 1998. Ecology of soil erosion in ecosystems. Ecosystems, Volume 1, pp. 416-426.
Pool-Stanvliet, R., 2013. A history of the UNESCO Man and the Biosphere Programme in South Africa. South African Journal of Science, 109(9/10), pp. 1-6.
Pressey, R., Cabeza, M., Watts, M.E, Cowling, R.M., Wilson, K.A., 2007. Conservation planning in a changing World. Trends in Ecology and Evolution, 22(11), pp. 583-593.
Pressey, R., Ferrier, L.S., Hager, T.C., Woods, C.A., Tully, S.L., Weinman, K.M., 1996. How well protected are them forests of north eastern New South Wales? Analyses of forest environments in relation to formal protection measures, land tenure and vulnerability to clearing.. Forest Ecology and Management, Volume 85, pp. 311-333.
Pressey, R. & Taffs, K., 2001. Scheduling conservation action in production landscapes:priority areas in western New South Wales defined by irreplaceability and vulnerability to vegetation loss. Biological Conservation, Volume 100, pp. 355-376.
Rebelo, A., Holmes, P., Dorse, C. & Wood, J., 2011. Impacts of urbanization in a biodiversity hotspot: Conservation challenges in Metropolitan Cape Town. South African Journal of Botany, Volume 77, pp. 20-35.
135 Reyers, B., 2004. Incorporating anthropogenic threats into evaluations of regional biodiversity and prioritisation of conservation areas in the Limpopo Province, South Africa. Biological Conservation, Volume 118, pp. 521-531.
Reyers, B., O’Farrell, P.J., Cowling, R.M., Egoh, B.N., Le Maitre, D.C., Vlok, J.H., 2009. Ecosystem Services, Land-Cover Change, and Stakeholders: Finding a Sustainable Foothold for a Semiarid Biodiversity Hotspot. Ecology and Society , 14(1).
Rogers, H., Glewa, L., Honzákb, M. & Hudson, M., 2010. Prioritizing key biodiversity areas in Madagascar by including data on human pressure and ecosystem services. Landscape and Urban Planning , Volume 96, pp. 48-56.
Rouget, M., Cowling, R., Vlok, J. & Thompson, M., 2006. Getting the biodiversity intactness index right: the importance of habitat degradation data. Global Change Biology, Volume 12, pp. 2032-2036.
Rouget, M., Reyers, B., Jonas, Z., Desmet, P., Driver, A., Maze, K., Egoh, B., Cowling, R.M., 2004. South African National Spatial Biodiversity Assessment 2004:Technical Report. Volume 1: Terrestrial Component., Pretoria: South African National Biodiversity Institute.
Rouget, M., Richardson, D.M., Cowling, R.M., Lloyd, J.W., Lombard, A.T., 2003. Current patterns of habitat transformation and future threats to biodiversity in terrestrial ecosystems of the Cape Floristic Region, South Africa.. Biological Conservation, Volume 112, pp. 63-85.
Saaty, T., 1977. A Scaling Method for Priorities in Hierarchical Structures. Journal of Mathematical Psychology, Volume 15, pp. 234-281.
Saaty, T., 2005. Making and Validating complex decisions with the AHP/ANP. Journal of Systems Science and Systems Engineering, 14(1), pp. 1-36.
Saaty, T. & Vargas, L., 2001. The Seven Pillars of the Analytic Hierarchy Process. In: Models Methods Concepts and Applicationsof the Analytic Hierarchy Process (International Series in Operations Research and Management Science). s.l.:Springer, pp. 27-46.
SANBI, 2009. Updating National Land Cover, Pretoria: South African National Biodiversity Institute.
SANBI, 2010. Threatened Species: A guide to Red Lists and their use in conservation. , Pretoria: South African National Biodiversity Institute.
SANBI, 2012. South African National Biodiversity Institute: Threatened Species Programme. National Assessment: Red List of South African Plants version 2012.1.. [Online]
136 Available at: http://redlist.sanbi.org/species. [Accessed 7 02 2013].
SAWS, 1995-2010. South African Weather Service : 1st Weather. [Online] Available at: http://www.1stweather.com/ [Accessed 28 2 2015].
Scholes, B., 2012. Fire Management in the Magaliesberg : Presentation. Magaliesberg Protected Environment AGM, CSIR.
Smith, M.D., van Wilgen, B., Burns, C.E., Govender, N., Potgieter, A.L.F., Andelman, S., Biggs, H.C., Botha, J., Trollope, W.S., 2012. Long-term effects of fire frequency and season on herbaceous vegetation in savannas of the Kruger National Park, South Africa. Journal of Plant Ecology, pp. 1-13.
Soule, M., 1991. Conservation: Tactics for a constant crisis. Science, Volume 253, pp. 744-750.
Southworth, J., Nagendra, H. & Munroe, D., 2006. Introduction to the special issue: Are parks working? Exploring human–environment tradeoffs in protected area conservation. Applied Geography, Volume 26, pp. 87-95.
Stranger, M., 2004. The Integration of Spatial Analysis Techniques and Decision Support Systems for Natural Resource Management, PhD Dissertation: University of West Virginia.
Thompson, M., 1996. A standard land-cover classification scheme for remote-sensing applications in South Africa. South African Jourmal of Science, Jan, Volume Vol 92, pp. 34-43.
Thompson, M., June 2012, Professional in remote sensing, image processing, image interpretation, Director at GeoterraImage, telephone interview with Belinda Cooper.
Turpie, J., 2003. The existence value of biodiversity in South Africa:How interest, experience, knowledge, income and perceived level of threat infuence local willingness to pay.. Ecological Economics, Volume 46, pp. 199-216.
Turpie, J. & Kleynhans, M., 2010. Wetland Health and Importance Research Programme: Wetland valuation. Vol IV: A protocol for the quantification and valuation of wetland ecosystem services. WRC Report No. TT 443/09, Pretoria: Water Research Commission.
UNESCO, 1996. Biosphere reserves: The Seville Strategy and the statutory framework of the world network, Paris: UNESCO.
UNESCO, 2009-2014. UNESCO-Media Services : New Man and the Biosphere Reserves in 2014. [Online]
137 Available at: http://www.unesco.org/new/en/media-services/multimedia/photos/mab-2014/ [Accessed 18 10 2014].
UNESCO-MAB, 2015. 21st meeting of the International Advisory Committee for Biosphere Reserves, 2-5 February Report. Paris: UNESCO
UNESCO-MAB , 2014. International Co-ordinating Council of the Man and the Biosphere (MAB) Programme Twenty-sixth session. [Online] Available at: http://www.unesco.org/new/fileadmin/MULTIMEDIA/HQ/SC/pdf/SC14-CONF-226- 15-MAB-ICC_Final_Report_en_8-7-2014-v2.pdf [Accessed 18 10 2014].
UNESCO-MAB, 2013. International Co-ordinating Council of the Man And Biosphere (MAB) Programme. Twenty-fifth Session. [Online] Available at: http://www.unesco.org/new/fileadmin/MULTIMEDIA/HQ/SC/pdf/SC-13-CONF-225- 11_Final_Report_en.pdf [Accessed 15 7 2013].
Uys, R., Bond, W. & Everson, T., 2004. The effect of different fire regimes on plant diversity in southern African grasslands. Biological Conservation , Volume 118, pp. 489-499. van der Ree, R., Jaeger, J. A. G., van der Grift, E. A. & Clevenger, A. P., 2011. Effects of Roads and traffic on wildlife populations and landscape function: road ecology is moving towards larger scales.. Ecology and Society, 16(1), p. art 48 (guest editorial).
Van Oudshoorn, F., 2012. Guide to Grasses of southern Africa. 3rd ed. Pretoria: Briza Publications.
Vasiljevic´, T., Srdjevic´, Z., Bajc¡etic´, R. & Miloradov, M., 2012. GIS and the Analytic Hierarchy Process for Regional Landfill Site Selection in Transitional Countries: A Case Study From Serbia. Environmental Management, Volume 49, pp. 445-458.
Veech, J., 2003. Incorporating socioeconomic factors into the analysis of biodiversity hotspots. Applied Geography, Volume 23, pp. 73-88.
Vimal, R., Pluvine, P., Sacca, C., Mazagol, P.O., Etlicher, B., Thompson, J.D., 2012. Exploring spatial patterns of vulnerability for diverse biodiversity descriptors in regional conservation planning. Journal of Environmental Management, Volume 95, pp. 9-16.
WAMIS, 2008-2013. Wide Area Monitoring Information System Long-Term Time Series. [Online] Available at: http://wamis.meraka.org.za/products/long-term-time-series [Accessed 11 06 2012].
138 Wannenburgh, A., 2006. A Systematic Spatial Prioritization for Invasive Alien Plant Control in South Africa, Cape Town: University of Cape Town. Masters Dissertation.
Ward, D., 2005. Do we understand the causes of bush encroachment in African savannas?. African Journal of Range Forage Science, Volume 22, pp. 101-105.
WCED, 1987. World Commission on Environment and Development: Our Common Future. UK: Oxford University Press.
Weeks, E., Overton, J. & Walker, S., 2013. Estimating patterns of vulnerability in a changing landscape: a case study of New Zealand's indigenous grasslands.. Environmental Conservation, 40(1), pp. 84-95.
Wells, J., Van Meurs, L.H., Rabie, M.A., Joubert, G.F., Moir, F., Russell, J., 2009. Terrestrial Minerals. In: H. Strydom & N. King, eds. Fuggle and Rabie's Environmental Management in South Africa, 2nd Edition. Cape Town: Juta Law, pp. 513 - 578.
Wesson & Balt, 2014. Birds of the Magaliesberg. Pretoria: Brits Hartebeespoort Rotary Club.
Whittington-Jones, C., Wolter, K. & West, S., 2014. Monitoring of Cape Vulture (Gyps coprotheres) breeding colonies in the Magaliesberg,South Africa 2007-2009. Vulture News, Volume 60.
Wilcove, D., Rothstein, D., Dubow, J., Phillips, A., Losos, E., 1998. Quantifying Threats to Imperiled Species in the United States. Bioscience, 48(8), pp. 607-615.
Wilhere, G., Goering, M. & Wang, H., 2008. Average optimacity: An index to guide site prioritization for Biodiversity conservation. Biological Conservation, Volume 141, pp. 770-781.
Wilson, K., Cabeza, A. & Klein, K., 2009. Fundamental Concepts of Spatial Conservation Prioritization. In: New York: Oxford University Press, pp. 16-27.
Wilson, K., Newton, A., Echeverrý, C., Weston, C., Bergman, M., 2005 b. A vulnerability analysis of the temperate forests of south central Chile. Biological Conservation, Volume 122, pp. 9-21.
Wilson, K., Pressey, R., Newton, A., Bergman, M., Possingham, H., Weston, C., 2005a. Measuring and Incorporating Vulnerability into conservation planning. Environmental Management, 35(5), pp. 527-543.
Wittmayer, J. & Büscher, B., 2010. Conserving Conflict? Transfrontier Conservation, Development Discourses and Local Conflict Between South Africa and Lesotho. Human Ecology, Volume 38, pp. 763-773.
139 Ying, X., Guang-Ming, Z., Gui-Qiu, C., Lin, T., Ke-Lin, W., Dao-You, H., 2007. Combining AHP with GIS in synthetic evaluation of eco-environment quality—A case study of Hunan Province, China. Ecological Modelling, Volume 209, pp. 97-109.
Zerger, A., Warren, G., Hill, P., Robertson, D., Weidemann, A., Lawton, K., 2011. Multi-criteria assessment for linking regional conservation planning and farm-scale actions. Environmental Modelling and Software, Volume 26, pp. 103-110.
Zhifeng Wu, C. & Cheng, J., 2013. Using kernel density estimation to assess the spatial pattern of road density and its impact on landscape fragmentation. International Journal of Geographical Information Science, 27(2), pp. 222-230.
140 Appendices
Appendix 1
APPENDIX : A MAGALIESBERG REGION LAND COVER LEGEND
Merging of Gauteng Land Cover 2009, and North West Land Cover 2006. Aggregation to basic Land Cover showing original Gauteng and North-West legends. Re-combinations of classes from both Land-cover sets to emphisise Biodiversity Threat Potential in class combinations.
Recoded & Aggregated Categories Level I equivalent Level 2 equivalent Gauteng LandCover Original Level 3 N.W. LandCover Original Level 3 To determine types of threat To determine severity within type Legend 39 Classes in ROI Legend 35 classes in ROI I.D. I.D. I.D. I.D. 1 Natural / untransformed 1 natural/untransformed 4 dense trees/bush 1 closed tree and bush 40 woodland/open bush 2 open closed tree & bush 39 wooded grassland 3 open bush 5 grassland 7 grassland 25 rocky grass matrix 8 sparse grassland 37 wetland non pan 10 wetlands vegetated 38 wetland pans 11 wetlands dry pans 9 natural bare rock 14 natural bare rock 4 sparse bush
2 water bodies 2 water bodies 10 natural water 9 water natural 7 man-made water 18 water artificial
3 Cultivated land 3 cultivated land 0 cultivated crops 24 cultivated perm irr orchard 1 cultivated other 25 cultivated annual dryland comm 2 cultivated pasture 27 cultivated irrigated annual grain 28 cultivated irrigated pivot 26 cultivated annual dryland subs
4 cultivated smallholdings 26 smallholdings - cultivated 31 smallholding cultivated
4 transformed non-built up 5 oldlands 13 oldlands- (topo)dense trees/bush 29 cultivated old fields (Degraded, eroded and 14 oldlands - (topo)grasslands disturbed land) 17 oldlands (topo)-woodland/openbush 19 oldlands- dense trees/bush 22 oldlands - wooded grassland 23 oldlands -woodland/openbush 6 plantations woodlots 24 plantation/woodlots 17 plantations and woodlots 7 Managed urban open space 34 urban grassland 15 planted grass sport 35 urban trees 16 planted grass golf 36 urban woodland 32 sports and recreation grassland 8 degraded land 3 degraded 21 disturbed/degraded 7 smallholdings - degraded 18 oldlands - degraded 12 oldlands -(topo) degraded 9 non vegetated bare 11 non-vegetated bare 20 erosion all types 15 oldlands - (topo)non vegetated/bare 21 oldlands - non vegetated/bare 10 cattle camps 6 intensive cattle camps 23 cattle camps
141
5 Built-up areas 11 smallholdings/low density 28 smalholdings - dense trees/ bush 30 smallholdings plots (Peri urban, rural , urban) 29 smallholdings - grassland 30 smallholdings - wooded grassland 31 smallholdings - woodland/open bush 12 scattered rural/low density 34 scattered rural all 13 animal batteries/ low density 36 animal batteries 14 urban/med to high density 33 urban 32 urban all 33 industry and commerce all 35 roads and tracks 37 greenhouses 6 Mining 15 mines and quarries 8 mines 38 mines extraction and tailings (mines,quarries, 16 subsurface and infra 39 mines subsurface & infrastructure waste dumps) 17 sewrage 19 water sewerage 18 landfill 22 landfill sites The 18 x Level 2 equivalent Land Cover Categories are used in this research
142 Appendix 2
LIST OF QUESTIONNAIRE RESPONDENTS
Name : Anthony Duigan Field of Expertise : Chair: Rhenosterspruit Conservancy Contact : 083 253 4576 & MBIG Committee [email protected] Name : Jason Sampson Field of Expertise : Environmental Rehabilitation Contact : 082-975-3990 Name : John Wesson Field of Expertise : Conservation, Parks and Contact : 083-444-7649 Recreation, Landscape planning, [email protected] Hortuculture Name : Vincent Carruthers Field of Expertise : Environmentalist Contact : [email protected] 082-411-8033 Name : Andrew Cauldwell Field of Expertise : Biodiversity Consultant Contact : 082-044-4413 Name : L. Paul Fatti Field of Expertise : Statistician, with a long-term concern Contact : 083- 266- 1532 with the Magaliesberg Name : Rob Millar Field of Expertise : Xanadu Conservancy Custodian Contact : 082-459-7091
Name : Morné Brits Field of Expertise : Botany / ecology / EIA Contact : 074-245-6359 Name : Maryna Storie Field of Expertise : GIS Land managment and disaster Contact : [email protected] risk assessment Name : Clare Kelso Field of Expertise : environmental ethics, climatology, Contact : [email protected] historical geographies
143
Appendix 3 A PAIRWISE COMPARISON OF THREATS TO BIODIVERSITY IN THE PROPOSED MAGALIESBERG BIOSPHERE BY ANALYTIC HIERARCHY PROCESS (AHP)
Contents Introduction pg. 1 Questionnaire Instructions pg. 2 Table 1: Descriptions pg. 3 Table 2: AHP Questionnaire pg. 4 Table 3: Intensity of Threat pg. 5 Table 4: Vulnerability to Threat pg. 6 A brief Introduction to the study
The nature of the research is spatial decision support for conservation and sustainable development, where the primary focus is to define areas within the Magaliesberg Biosphere under pressure from direct or indirect anthropogenic stressors that may be threatening the existing biodiversity resources of the region.
Central to regional conservation planning is how to decide on which areas to concentrate efforts and funding for biodiversity preservation and remediation, taking into account social and economic constraints. By using spatial layers as proxies for threats, a priority map of threatened sites in relation to critical biodiversity areas and the biosphere zones, can inform conservation action and promote sustainable practices in the spirit of a Biosphere.
It is well documented in conservation literature that some of the major threats to terrestrial biodiversity are habitat loss and habitat degradation. These are broad terms that encompass a range of processes and land-uses that threaten, to varying degrees the structure and functioning of biodiversity at the species and ecosystem level.
An explanation of the AHP method
An aspect of this research is to quantify these threats to biodiversity in the Magaliesberg Biosphere context, using the AHP (Analytic Hierarchy Process) method, which is a form of Multi criteria analysis requiring the input of experts to quantify these threats.
The method is straightforward and involves pair wise comparisons of each threat in relation to clear objectives– in this case the degree to which each threat, relative to each other, affects biodiversity, and the different components of terrestrial biodiversity. A resulting matrix of subjective pair-wise scores provided by each expert participant is then mathematically processed into a fairly objective, quantitative relative- ranking for each threat. The consistency of the pair-wise comparison is checked by a numerical consistency index, after-which the scores are used in the GIS to weight the different threat layers to make up a composite map of biodiversity pressures.
144 Questionnaire Instructions
OBJECTIVE: To establish the degree to which each threat criteria, relative to one another, affects biodiversity in general, and the different components of biodiversity.
CRITERIA: Seven threats to biodiversity that can be spatially represented and scaled. Please refer to Table 1 (on pg. 3) for a list and descriptions.
Part one
OBJECTIVE: To establish the relative importance of each threat criteria as a stressor to the existence of biodiversity in the Magaliesberg Region.
STEP 1: Consider the threat criteria described in Table 1. Make pair-wise comparisons between the different criteria, by comparing each factor to every other factor in terms of its threat to biodiversity and assign a comparative score to each. Please follow the instructions in Table 2 (on pg. 4).
Part two
OBJECTIVE: To establish the range in intensity of impact within each threat criteria.
STEP 2: Some of the threat criteria are sub-categorised to scale the intensity of threat within each criteria. Please assign a score for the sub-categories listed in Table 3 (on pg. 5). You may add comments or reasons.
Part three
OBJECTIVE: To establish the degree to which each threat affects the different components of biodiversity.
STEP 3: Consider the threat criteria described in Table 1 and to the best of your knowledge assign a ‘vulnerability to threats score’ for each biodiversity component in Table 4 (on pg.6). You may add comments or reasons.
Thank you for your participation
Name : Field of Contact : Expertise :
145
THREAT CRITERIA SPATIAL LAYERS DESCRIPTION / DEFINITION OF THREAT CRITERIA AND MAP LAYERS A Habitat loss pressures 1 Built-Up Areas/ High Ranges from total transformation of original land cover- a predominantly impervious, non vegetated surface to and Medium Density medium density formal settlements in urban and rural centres. Includes commercial, agricultural, industrial and Urban residential structures and major roads within urban and rural areas. 1 URBANIZATION 2 High Density Informal High density informal residential settlements generally not serviced with piped sanitation, waste removal or Settlements storm-water drainage. 3 Smallholdings/ Low Peri-urban and scattered rural low density structures. Surrounding open land may be in a natural or semi- Density Built-Up natural state or used for small scale agriculture. Includes plots, scattered villages, and country estates. 1 Cultivated Land Distinct field boundaries of permanent or seasonal agricultural production. Includes irrigated, rain-fed, commercial and subsistence cultivation of annual grain crops, vegetables, planted pasture and orchards. 2 CULTIVATION Fertilizers and pesticides may be applied to soil and crops, and lands in temporary production often lie fallow for some months. 1 Mines, quarries and All surface infrastructure and non-vegetated surfaces associated with mining. Extraction pits including open waste facilities cast, borrow pits, sand and rock quarries, mine tailings, slimes dams, waste and storage dumps, sewage ponds. 3 MINING Typically associated with possible point-source environmental risk from leachate, spills, dust, air pollution, and water pollution as well as the transformation of natural vegetation and geomorphological elements. B Habitat degradation pressures 4 1 Degraded/eroded land Artificially induced non-vegetated bare areas exhibiting low vegetation cover compared to surrounding natural or semi-natural areas. Typically negligible grass, tree or shrub cover. Erosion Gullies, dongas and sheet erosions rills may feature. Degraded lands often result from overgrazing of livestock and/or excessive harvesting of wood TRANSFORMED 2 Non-Built Up Areas/ resources and are prone to soil loss by erosion processes and/or bush encroachment LAND Disturbed Open Space Non-built-up areas that have been transformed and disturbed to varying degrees, but not yet classifiable as degraded. Typically old-lands not managed with anomalous grassy areas and/or woody invasion. Also open spaces within and close to urban areas that are variously managed and where planted and self- seeded invasive and exotic species occur and often dominate. Recreational planted grass is included. FRAGMENTATION OF 1 Natural , Semi-natural The size of untransformed patches and distances between them, as well as the road network density in these 5 OPEN SPACE Untransformed Areas patches. Fragmentation affects range, connectivity, dispersal and persistence of biodiversity. 1 Extent and Frequency The frequency of fires (>25 hectares) in the grassland and savanna biomes and forest patches, determined over 6 FIRE EVENTS Of Fire Events an eleven year period from 2000 – 2011. Excessive recurrent fire or the absence of fire has unique repercussions for the composition, structure and ecology of each vegetation type. ALIEN PLANT 1 Invasive Alien Plant The abundance (occurrence and density) of Invasive Alien plant species measured in 6 hectare grids (terrestrial 7 INVASIONS Density and riparian zones included). Eleven woody species that occur in the region are mapped.
146
Table 2 – AHP PAIR-WISE COMPARISONS
Objective: To establish the relative importance of each criteria as a threat to the existence of biodiversity in the Magaliesberg Region. Please compare the importance of the elements in relation to the above objective and fill in the table: Which criterion in each pair is more threatening to biodiversity, A or B, and how much more severe is the threat (Use the scale 1-9 as given below). Please use the light green spaces for your choices. Element 1 Criterion 1 URBANIZATION ( scale: low, medium, high density) 2 Criterion 2 CULTIVATION (scale: commercial, small-scale ) 3 Criterion 3 MINING (scale: deep-level, open-cast, tailings, dumps) 4 Criterion 4 TRANSFORMED LAND (scale: eroded, degraded, disturbed, modified) 5 Criterion 5 FRAGMENTATION OF OPEN SPACE ( scale: size, proximity) 6 Criterion 6 FIRE EVENTS (scale: absence, infrequent, excessive) 7 Criterion 7 ALIEN PLANT INVASIONS (low, medium, high density)
Which criteria is more of a threat to biodiversity, and by how much? More of a threat (A Intensity A B or B) (1-9)
1 Cultivation 2 Mining with 3 Transformed land Criterion 1 4 Fragmentation URBANIZATION 5 Fire Events 6 Alien Invasion
compared
7 1 Mining 2 Transformed land 3 Criterion 2 Fragmentation Vs. 4 CULTIVATION Fire Events 5 Alien Invasion 6 1 Transformed land 2 Fragmentation Criterion 3 3 Vs. Fire Events MINING 4 Alien Invasion 5 1 Fragmentation Criterion 4 2 Fire Events TRANSFORMED Vs. 3 Alien Invasion LAND 4 1 Fire Events Criterion 5 2 Vs. Alien Invasion FRAGMENTATION 3 1 Criterion 6 FIRE Alien Invasion 2 EVENTS Vs.
Intensity of importance Definition Explanation Two elements contribute equally to the 1 Equal threat objective Experience and judgment slightly favour one 3 Slightly more threatening element over another Experience and judgment strongly favour 5 Somewhat more threatening one element over another One element is favoured very strongly over 7 Strongly more threatening another, it dominance is demonstrated in practice The evidence favouring one element over 9 Exceedingly more threatening another is of the highest possible order of affirmation 2,4,6,8 can be used to express intermediate values for elements that are very close in importance
147
Table 3 SUB – CRITERIA Objective: To establish the range in intensity of impact within each threat criteria
Low density (peri-urban and rural)
Criteria 1 Density scale Medium density (rural centres, residential) URBANIZATION High density (informal) High density (formal)
Criteria 2 Large scale (commercial) Scaled by type CULTIVATION Small scale (subsistence) Deep- level mining (platinum) Criteria 3 Scaled by type Open- Cast rock and sand quarries MINING Mine Tailings, waste dumps, separation ponds Eroded (gullies,dongas)
Criteria 4 Degraded (sparse vegetation /bare areas) Intensity scale TRANSFORMED LAND Disturbed (old-lands) Modified (recreational grass, urban open space) One Fire event in forest or riparian zone
Repeat fire events in forest or riparian zone Criteria 6 Frequency scale Absence of fire in grassland/savanna FIRE EVENTS Excessive repeat fires in grassland/savanna Repeat fire in the “wrong” season High density (20-45%) Criteria 7 Density Scale Medium density (5-20%) ALIEN SPECIES INVASIONS Low density (1-5%)
Impact Intensity Scale
1 Low Impact 3 Slight Impact 5 Medium Impact 7 High Impact 9 Very High Impact
148 Table 4 – Vulnerability of Biodiversity Components to Threat Criteria
OBJECTIVE: To establish the degree to which each threat affects the different components of biodiversity.
Biodiversity Terrestrial Flora Meta Fauna Mega Fauna Birds Ecosystems Components Criteria 1. Urbanization
Criteria 2. Cultivation
Criteria 3. Mining
Criteria 4. Transformed land
Criteria 5. Fragmentation
Criteria 6. Fire Events
Criteria 7. Alien Plant Invasions
Vulnerability Scale 1 Not vulnerable 3 Slightly vulnerable 5 Moderately Vulnerable 7 Vulnerable 9 Extremely vulnerable
The End, Thank-you !
149 Appendix 4
Pixel Resolution Bi-Plot
69 149
HIGH
PRIORITY 22
MEDIUM
CONSERVATION
16
11
LOW
LOW MEDIUM HIGH PRIORITY THREAT
Scatter Plot of pixel resolution Priority Conservation Value and Priority Biodiversity Threats. Red lines indicate low, intermediate and high class thresholds and blue lines indicate lower and upper mid-range limits. These thresholds are used to identify clusters to target for conservation action (Margules & Pressey, 2000).
150
Appendix 5
151
Rustenburg
hreat Priorities
N s 0-9
653
0 5 1 0 2 0 3 0 40 Johannesburg Low 0 ••:::::i-c::=-•••• c::===:::::i•••••Ki lometers
152 INTEGRATED PRIORITIES FOR CONSERVATION {C) & THREAT {T)
low C/lo 201 301 high C/low T 0 D D D 401
102 D202 D302 D 402
403 103 D203 D303 D
304 high C/high T 404
N
A 0 5 10 20 30 40 -=--= ---======---•Kilometers
153 Appendix 6
154 Klaus Goepel AHP 25/11/13
AHP Analytic Hierarchy Process (multiple inputs) K. Goepel Version 16.10.2012 http://bpmsg.com Only input data in the light green fields and worksheets! n= 7 Number of criteria (3 to 8)
N= 7 Number of Participants (1 to 7)
p= 0 selected Participant (0=consol.) 2 7 Consolidated
sheet '8x8'! Input Fields (green)
Objective To establish which pressures are most threatening to the existence of biodiversity
Author B. Cooper Date 21/12/12
Table Element Comment Weights 1 urbanization Comment 1 23% 2 cultivation Comment 2 10% 3 mining Comment 3 31% 4 transformed land 10%
5 fragmentation 9%
6 fire events 4%
7 alien invasion 13% 8 -
Eigenvalue lambda 7.269 3.4% Consistency Ratio CR
invasion
events normalized principal
Matrix Eigenvector urbanization cultivation mining transformed land fragmentation alien 0 fire
urbanization 1 4 2/3 1 5/6 2 3/5 5 2 1 23.3%
cultivation 1/4 1 1/3 4/5 1 3/8 1 2/3 1 1/3 1 9.5%
mining 1 1/2 3 1/9 1 3 5/9 3 1/2 5 4/5 2 1 30.7%
transformed 5/9 1 1/4 2/7 1 1 3 1/6 1/2 1 land 9.8%
fragmentation 2/5 5/7 2/7 1 1/9 1 4 3/5 7/8 1 9.4%
fire events 1/5 3/5 1/6 1/3 2/9 1 1/4 1 4.3%
alien invasion 1/2 3/4 1/2 2 1 1/7 4 1 1 13.0%
0 1 1 1 1 1 1 1 1
http://bpmsg.com AHPcalc versionA 16.10.12.xlsx-Summary http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 1 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: ac A B Important (1-9) Date: 21/12/12 1 cultivation a 2
2 mining a 3 A 3 transformed land b 2 B 4 urbanization fragmentation a 4 5 fire events a 8
6 compared with alien invasion a 5
7 1 mining b 3
2 transformed land b 5
3 fragmentation a 4
cultivation red with 4 fire events a 8
5 alien invasion a 5 6 compa
1 transformed land b 4 2 fragmentation a 4 3 mining fire events a 7 4 alien invasion a 5 comp. with 5
1 fragmentation a 4 2 fire events a 9 transformed land 3 alien invasion a 5
4 comp. with
1 fire events a 5
2 fragmentation vs alien invasion a 3 3 1 alien invasion b 5 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance by Klaus Goepel AHPcalc versionA 16.10.12.xlsx-Input1 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 2 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: vc A B Important (1-9) Date: 21/12/12 1 cultivation a 3
2 mining b 5 A 3 transformed land a 3 B 4 urbanization fragmentation a 5 5 fire events a 7
6 compared with alien invasion a 5
7 1 mining b 5
2 transformed land B 1
3 fragmentation B 1
cultivation red with 4 fire events b 5
5 alien invasion a 3 6 compa
1 transformed land a 7 2 fragmentation a 7 3 mining fire events a 9 4 alien invasion a 9 comp. with 5
1 fragmentation a 1 2 fire events a 5 transformed land 3 alien invasion b 1
4 comp. with
1 fire events a 5
2 fragmentation vs alien invasion a 5 3 1 alien invasion b 7 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input2 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 3 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: mb A B Important (1-9) Date: 21/12/12 1 cultivation a 6
2 mining a 7 A 3 transformed land a 9 B 4 urbanization fragmentation a 6 5 fire events a 8
6 compared with alien invasion a 7
7 1 mining a 6
2 transformed land a 8
3 fragmentation a 6
cultivation red with 4 fire events a 8
5 alien invasion a 3 6 compa
1 transformed land a 8 2 fragmentation b 3 3 mining fire events a 8 4 alien invasion b 6 comp. with 5
1 fragmentation b 8 2 fire events b 1 transformed land 3 alien invasion b 8
4 comp. with
1 fire events a 8
2 fragmentation vs alien invasion a 1 3 1 alien invasion b 7 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input3 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 4 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: pf A B Important (1-9) Date: 21/12/12 1 cultivation a 3
2 mining b 5 A 3 transformed land b 3 B 4 urbanization fragmentation a 2 5 fire events a 4
6 compared with alien invasion b 4
7 1 mining b 6
2 transformed land b 4
3 fragmentation a 2
cultivation red with 4 fire events a 2
5 alien invasion b 5 6 compa
1 transformed land a 4 2 fragmentation a 5 3 mining fire events a 7 4 alien invasion b 2 comp. with 5
1 fragmentation a 3 2 fire events a 4 transformed land 3 alien invasion b 3
4 comp. with
1 fire events a 3
2 fragmentation vs alien invasion b 5 3 1 alien invasion b 7 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input4 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 5 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: ad A B Important (1-9) Date: 21/12/12 1 cultivation a 9
2 mining b 7 A 3 transformed land a 3 B 4 urbanization fragmentation b 3 5 fire events a 1
6 compared with alien invasion a 5
7 1 mining b 7
2 transformed land b 1
3 fragmentation b 5
cultivation red with 4 fire events b 7
5 alien invasion a 5 6 compa
1 transformed land a 7 2 fragmentation a 7 3 mining fire events a 7 4 alien invasion a 7 comp. with 5
1 fragmentation b 3 2 fire events a 3 transformed land 3 alien invasion a 1
4 comp. with
1 fire events a 3
2 fragmentation vs alien invasion a 3 3 1 alien invasion a 3 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input5 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 6 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: ms A B Important (1-9) Date: 21/12/12 1 cultivation a 4
2 mining b 9 A 3 transformed land a 1 B 4 urbanization fragmentation a 2 5 fire events a 6
6 compared with alien invasion b 8
7 1 mining b 9
2 transformed land b 2
3 fragmentation a 1
cultivation red with 4 fire events a 3
5 alien invasion b 7 6 compa
1 transformed land a 6 2 fragmentation a 7 3 mining fire events a 9 4 alien invasion a 2 comp. with 5
1 fragmentation a 1 2 fire events a 6 transformed land 3 alien invasion b 7
4 comp. with
1 fire events a 8
2 fragmentation vs alien invasion b 8 3 1 alien invasion b 9 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input6 http://bpmsg.com AHP 25/11/13
AHP Analytic Hierarchy Process n= 7 Input 7 Objective To establish which pressures are most threatening to the existence of Only input data in the light green fields! Please compare the importance of the elements in relation to the above objective and fill in the table: Which element in each pair is more important, A or B, and how much more important is it. (Use the scale 1-9 as given below)
Element Comment 1 urbanization Comment 1 2 cultivation Comment 2 3 mining Comment 3 4 transformed land 0 5 fragmentation 0 6 fire events 0 7 alien invasion 0 8 0 0 Element More Intensity Name: ck A B Important (1-9) Date: 21/12/12 1 cultivation a 5
2 mining a 5 A 3 transformed land a 5 B 4 urbanization fragmentation a 5 5 fire events a 7
6 compared with alien invasion a 5
7 1 mining b 3
2 transformed land a 1
3 fragmentation a 1
cultivation red with 4 fire events a 3
5 alien invasion a 1 6 compa
1 transformed land a 3 2 fragmentation a 3 3 mining fire events b 1 4 alien invasion a 3 comp. with 5
1 fragmentation a 1 2 fire events b 1 transformed land 3 alien invasion b 3
4 comp. with
1 fire events a 3
2 fragmentation vs alien invasion b 3 3 1 alien invasion b 3 fire events vs 2 1 vs
Intensity of importance Definition Explanation 1 Equal importance Two elements contribute equally to the objective Experience and judgment slightly favor one element over 3 Moderate importance another Experience and judgment strongly favor one element over 5 Strong Importance another One element is favored very strongly over another, it 7 Very strong importance dominance is demonstrated in practice The evidence favoring one element over another is of the 9 Extreme importance highest possible order of affirmation
2,4,6,8 can be used to express intermediate values, 1.1, 1.2, etc. for elements that are very close in importance
AHPcalc versionA 16.10.12.xlsx-Input7 http://bpmsg.com AHP 25/11/13 Multiple Input Sheet AHP Analytic Hierarchy Process K. Goepel Multiple Input
1 7 = k number of participants geometric mean: k 7 = n number of criteria b ij ( a 1 ij a 2 ij a k ij )
Consolidated ac 21/12/12 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 1 4.1 0.7 1.8 2.6 5.0 2.0 1.0 1 1 2 3 1/2 4 8 5 1 2 1/4 1 0.3 0.8 1.4 1.6 1.3 1.0 2 1/2 1 1/3 1/5 4 8 5 1 3 1 1/2 3 1/9 1 3.5 3.5 5.8 2.1 1.0 3 1/3 3 1 1/4 4 7 5 1 4 5/9 1 1/4 2/7 1 0.9 3.2 0.5 1.0 4 2 5 4 1 4 9 5 1 5 2/5 5/7 2/7 1 1/9 1 4.6 0.9 1.0 5 1/4 1/4 1/4 1/4 1 5 3 1 6 1/5 3/5 1/6 1/3 2/9 1 0.3 1.0 6 1/8 1/8 1/7 1/9 1/5 1 1/5 1 7 1/2 3/4 1/2 2 1 1/7 4 1 1.0 7 1/5 1/5 1/5 1/5 1/3 5 1 1 8 1 1 1 1 1 1 1 1 8 1 1 1 1 1 1 1 1
vc Date mb 21/12/12 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 1 3 1/5 3 5 7 5 1 1 1 6 7 9 6 8 7 1 2 1/3 1 1/5 1 1 1/5 3 1 2 1/6 1 6 8 6 8 3 1 3 5 5 1 7 7 9 9 1 3 1/7 1/6 1 8 1/3 8 1/6 1 4 1/3 1 1/7 1 1 5 1 1 4 1/9 1/8 1/8 1 1/8 1 1/8 1 5 1/5 1 1/7 1 1 5 5 1 5 1/6 1/6 3 8 1 8 1 1 6 1/7 5 1/9 1/5 1/5 1 1/7 1 6 1/8 1/8 1/8 1 1/8 1 1/7 1 7 1/5 1/3 1/9 1 1/5 7 1 1 7 1/7 1/3 6 8 1 7 1 1 8 1 1 1 1 1 1 1 1 8 1 1 1 1 1 1 1 1
pf 21/12/12 ad 21/12/12 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 1 3 1/5 1/3 2 4 1/4 1 1 1 9 1/7 3 1/3 1 5 1 2 1/3 1 1/6 1/4 2 2 1/5 1 2 1/9 1 1/7 1 1/5 1/7 5 1 3 5 6 1 4 5 7 1/2 1 3 7 7 1 7 7 7 7 1 4 3 4 1/4 1 3 4 1/3 1 4 1/3 1 1/7 1 1/3 3 1 1 5 1/2 1/2 1/5 1/3 1 3 1/5 1 5 3 5 1/7 3 1 3 3 1 6 1/4 1/2 1/7 1/4 1/3 1 1/7 1 6 1 7 1/7 1/3 1/3 1 3 1 7 4 5 2 3 5 7 1 1 7 1/5 1/5 1/7 1 1/3 1/3 1 1 8 1 1 1 1 1 1 1 1 8 1 1 1 1 1 1 1 1
ms 21/12/12 ck 21/12/12 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 1 4 1/9 1 2 6 1/8 1 1 1 5 5 5 5 7 5 1 2 1/4 1 1/9 1/2 1 3 1/7 1 2 1/5 1 1/3 1 1 3 1 1 3 9 9 1 6 7 9 2 1 3 1/5 3 1 3 3 1 3 1 4 1 2 1/6 1 1 6 1/7 1 4 1/5 1 1/3 1 1 1 1/3 1 5 1/2 1 1/7 1 1 8 1/8 1 5 1/5 1 1/3 1 1 3 1/3 1 6 1/6 1/3 1/9 1/6 1/8 1 1/9 1 6 1/7 1/3 1 1 1/3 1 1/3 1 7 8 7 1/2 7 8 9 1 1 7 1/5 1 1/3 3 3 3 1 1 8 1 1 1 1 1 1 1 1 8 1 1 1 1 1 1 1 1
9 of 15 AHPcalc versionA 16.10.12.xlsx-multInp http://bpmsg.com AHP 25/11/13 Multiple Input Sheet
10 of 15 AHPcalc versionA 16.10.12.xlsx-multInp http://bpmsg.com AHP
AHP Analytic Hierarchy Process (8x8 Matrix) 7
KPI1 KPI2 KPI3 KPI4 KPI5 KPI6 KPI7 KPI8
KPI1 1 4 2/3 1 5/6 2 3/5 5 2 1 KPI2 1/4 1 1/3 4/5 1 3/8 1 2/3 1 1/3 1 KPI3 1 1/2 3 1/9 1 3 5/9 3 1/2 5 4/5 2 1 KPI4 5/9 1 1/4 2/7 1 1 3 1/6 1/2 1 KPI5 0.3848 0.7239 0.2831 1.1041 1 4 3/5 7/8 1 KPI6 0.2011 0.6071 0.1723 0.3151 0.2177 1 1/4 1 KPI7 0.4953 0.7666 0.4854 1.9329 1.1504 3.966 1 1 KPI8 1 1 1 1 1 1 1 1 sum (col) 4 1/3 11 4/7 3 2/9 10 1/2 10 4/5 25 1/6 8 7 normalized principal Eigenvector Normalization 1 2 3 4 5 6 7 8 normalized matrix 1st 6th iteration
1 KPI1 0.23 0.35 0.21 0.17 0.24 0.20 0.25 0.00 24% 23%
2 KPI2 0.06 0.09 0.10 0.08 0.13 0.07 0.16 0.00 10% 10%
3 KPI3 0.34 0.27 0.31 0.34 0.33 0.23 0.26 0.00 30% 31%
4 KPI4 0.13 0.11 0.09 0.10 0.08 0.13 0.06 0.00 10% 10%
5 KPI5 0.09 0.06 0.09 0.10 0.09 0.18 0.11 0.00 10% 9%
6 KPI6 0.05 0.05 0.05 0.03 0.02 0.04 0.03 0.00 4% 4%
7 KPI7 0.11 0.07 0.15 0.18 0.11 0.16 0.12 0.00 13% 13%
8 KPI8 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0% 0%
Lambda 101% 1.1027 0.9891 1.0264 1.0188 1.0768 1.0433 0 7.269 principal E n 7 CI 0.045 1.32 CR 3.4% Consisten Random Index n 1 2 3 4 5 6 7 8 RI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41
0.2253 0.2299 0.234 0.2317 0.2376 0.2307 0.2445 0 23.34% -3.6E-03 0.094 0.0846 0.0972 0.1023 0.0941 0.1009 0.0949 0 9.54% -8.5E-04 1st 0.3097 0.3131 0.3037 0.3043 0.3084 0.3082 0.3052 0 30.75% 1.2E-02 0.0974 0.1041 0.0967 0.0931 0.0981 0.0947 0.0991 0 9.76% -1.2E-03 0.0959 0.0944 0.0956 0.0948 0.0908 0.0964 0.0911 0 9.41% -9.8E-03 0.0427 0.044 0.0428 0.0419 0.0439 0.0389 0.0432 0 4.25% 3.4E-03 0.1351 0.1298 0.13 0.1319 0.1271 0.1302 0.122 0 12.94% 3.6E-04 0 0 0 0 0 0 0 0 0.00% 0.0E+00 0 23.29% 0.233 0.233 0.2329 0.2329 0.2328 0.2329 0.2327 -4.8E-04 0.0953 0.0955 0.0953 0.0952 0.0953 0.0952 0.0953 0 9.53% -1.2E-04 2nd 0.3068 0.3068 0.3069 0.3069 0.3069 0.3069 0.3069 0 30.69% -6.2E-04 0.0976 0.0975 0.0976 0.0977 0.0976 0.0976 0.0976 0 9.76% -3.3E-05 0.0944 0.0945 0.0944 0.0944 0.0945 0.0944 0.0945 0 9.45% 3.3E-04 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0 4.28% 3.0E-04 0.13 0.13 0.13 0.13 0.1301 0.13 0.1302 0 13.01% 6.1E-04 0 0 0 0 0 0 0 0 0.00% 0.0E+00 23.29% 0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0 1.1E-05
AHPcalc versionA 16.10.12.xlsx-8x8 http://bpmsg.com AHP
0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0 9.53% 1.5E-06 3rd 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0 30.69% -1.0E-07 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0 9.76% 2.9E-07 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0 9.45% -4.0E-06 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0 4.28% -6.5E-07 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0 13.00% -8.1E-06 0 0 0 0 0 0 0 0 0.00% 0.0E+00
0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0 23.29% 3.5E-09 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0 9.53% 4.1E-10 4th 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0 30.69% -9.8E-10 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0 9.76% 1.3E-10 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0 9.45% -9.1E-10 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0 4.28% 1.3E-10 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0 13.00% -2.3E-09 0 0 0 0 0 0 0 0 0.00% 0.0E+00
0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0 23.29% 5.0E-16 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0 9.53% 1.2E-16 5th 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0 30.69% 0.0E+00 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0 9.76% 1.2E-16 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0 9.45% 0.0E+00 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0 4.28% 6.2E-17 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0 13.00% 0.0E+00 0 0 0 0 0 0 0 0 0.00% 0.0E+00
0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0.2329 0 23.29% 4.7E-16 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0.0953 0 9.53% 1.9E-16 6th 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0.3069 0 30.69% 7.2E-16 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0.0976 0 9.76% 1.9E-16 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0.0945 0 9.45% 1.9E-16 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0.0428 0 4.28% 6.9E-17 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0 13.00% 2.8E-16 0 0 0 0 0 0 0 0 0.00% 0.0E+00
AHPcalc versionA 16.10.12.xlsx-8x8 http://bpmsg.com AHP
igenvalue cy
AHPcalc versionA 16.10.12.xlsx-8x8 http://bpmsg.com
This work is licensed under the Creative Commons Attribution-Noncommercial 3.0 Singapore Lic (You need to give credit to the author) To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/3.0/sg/ or send a letter to Creative Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA. Date Comment 05/03/12 1st draft based on AHPcalc vers 27.8.11; allows input from max. 7 participants 07/04/12 added selection of participants (p), "0"= consolidated 09/05/12 correction in sheet 8x8 to select CR depending on the number of criteria n 22/05/12 some minor corrections in the input sheets 16/10/12 Correction to select the result of individual participants in the summary sheet. Result was not correct selecting p =4 to 7. (Thanks for feedback from Pascal)
http://bpmsg.com
ense.