Brookes, Pa, Tree Resource Harvesting In Land

Modelling tree resource harvesting on communal land in the Maputaland Centre of Endemism.

By P.A. Brookes

Submitted as part of the requirements for the award of the M.Sc. degree in Conservation Biology at the University of Kent at Canterbury

Durrell Institute of Conservation and Ecology University of Kent at Canterbury September 2004

1

Contents Contents................................................................................................................................................... 2 List of Figures ......................................................................................................................................... 3 Tables ...................................................................................................................................................... 4 Acknowledgements ................................................................................................................................. 5 Abstract ................................................................................................................................................... 6 Introduction ............................................................................................................................................. 7 1.1 Overview .................................................................................................................................... 7 1.2 Systematic conservation planning and design of conservation land-scapes............................... 7 1.3 Setting priorities for identifying plant resource use ................................................................... 9 1.3.1 Resource value and priority plant species.......................................................................... 9 1.3.2 A hierarchical approach to understanding biodiversity and establishing priorities ........... 9 1.3.3 The use of ‘short cuts’ and conceptual filters ...................................................................10 1.3.4 Priority setting related to social, cultural and economic driving forces ...........................11 1.4 South African biodiversity and conservation policies...............................................................12 1.5 Harvesting of plant resources in The Maputaland Centre of Endemism...................................13 1.5.1 Biodiversity and conservation in Maputaland ..................................................................13 1.5.2 Conservation planning and management of biodiversity in Maputaland .........................16 1.5.3 The value of the natural resource base to rural populations in Maputaland .....................18 1.5.4 The fuelwood natural resource base .................................................................................18 1.5.5 The medicinal plants natural resource base ......................................................................19 1.6 Study aims and objectives .........................................................................................................20 2 Study area and methods .................................................................................................................22 2.1 Overview ...................................................................................................................................22 2.2 The Maputaland study area .......................................................................................................22 2.3 Description of study areas .........................................................................................................26 2.3.1 Study area 1 – conservation issues and disturbance threats..............................................26 2.3.2 Study area 2 – conservation issues and disturbance threats..............................................27 2.4 Quantitative resource inventories ..............................................................................................30 2.5 Pre-site evaluation of study areas and determination of priority species ..................................30 2.6 Sampling design and inventory methodology for assessing harvesting impact ........................32 2.7 Modelling ..................................................................................................................................33 3 Results ...........................................................................................................................................39 3.1 Resource harvest inventory, species prioritisation and resource use.........................................39 3.2 Relative resource value, tree abundance and user groups .........................................................41 3.3 Relative resource value, individual tree harvest intensity, harvest impact and resilience.........43 3.4 Harvesting population and landscape level effects and potential for sustainable use ...............48 3.5 Socio-economic results of resource use from the ‘cross-checking’ methods............................48 3.6 Logistic regression modelling of resource use and prediction of harvest impact......................48 3.7 GIS map and graphical representation of resource harvesting risk ...........................................49 4 Discussion......................................................................................................................................55 4.1 Overview ...................................................................................................................................55 4.2 Harvesting and resource value to rural communities of Maputaland ........................................55 4.3 Resource use impact and recommendations..............................................................................58 4.4 Modelling tree resource use in the Maputaland Centre of Endemism.......................................59 5 Conclusions ...................................................................................................................................62 6 References......................................................................................................................................63 Appendices .............................................................................................................................................68

2

List of Figures

Figure 1

Maputaland Centre of Endemism

14

Figure 2

Protected Area Coverage of Maputaland

15

Figure 3

Preliminary Conservation Planning Exercise for Maputaland

17

Figure 4

Ecological Zones and Major Topographical Features of Maputaland

24

Figure 5

Rivers, Lakes, Roads & Towns of Maputaland

25

Figure 6

Study Area 1

28

Figure 7

Study Area 2

29

Figure 8

Landcover Classification Map of Maputaland

31

Figure 9

Photos of Sand Forest Habitat – Study Area 1

35

Figure 10

Photos of Lebombo Woodland Habitat – Study Area 2

36

Figure 11

Photos of Stem Cutting Tree Resource Use

37

Figure 12

Photos of Bark Stripping Tree Resource Use

38

Figure 13a

Stem Cutting – Harvest Intensity of Individual Trees (Study Area 1)

44

Figure 13b

Stem Cutting – Harvest Intensity of Individual Trees (Study Area 2)

45

Figure 13c

Bark Stripping – Harvest Intensity of Individual Trees (Study Area 1)

46

Figure 14

Factors Significant in Modelling Stem Cutting

50

Figure 15

Factors Significant in Modelling Bark Stripping

50

Figure 16

Modelling Stem Cutting Resource Use in Maputaland

51

Figure 17

Modelling Bark Stripping Resource Harvesting in Maputaland

52

Figure 18

Predicted Probability of Resource Harvesting and Distance from Subsistence

53

Figure 19

Site Specific Resource Harvesting and Distance from Subsistence Agriculture

54

3

Tables

Table 1 Priority Tree Species Inventoried Based on ‘Cross-Checking’ Methods

40

Table 2 Relative Resource Value of Priority Species

42

Table 3 Relative Resource Value of Priority Species and Different User Groups

42

Table 4 Opportunity for Sustainable Use Based on Predictors of Resilience to Harvesting

47

Table 5 Results of Multiple Logistic Regression for Stem Cutting Resource Harvesting

49

Table 6 Results of Multiple Logistic Regression for Bark Stripping Resource Harvesting

49

4

Acknowledgements I would like to thank Dr Bob Smith for supervising the project and for giving me the opportunity to work in Maputaland, South Africa. I am grateful for the advice and support that he has given me throughout the project. I owe an enormous gratitude to my field assistant, Derrick T. Tembe, without his knowledge and guidance this project would not have been possible. Go well my friend. In addition, I am indebted to the people of the Tembe and Mathenjwa Tribal Authorities who welcomed me into their communities and allowed me access to their land. I would like to acknowledge the support of the many members of staff at EKZN Wildlife who supported our work and gave advice throughout the duration of the fieldwork. In particular I would like to thank Wayne Matthews (Tembe Elephant Park) for his co-operation in the project. I am grateful for the financial support provided by DICE and the support of the Darwin Initiative. A special thanks are deserved for Julian Eason and Nerissa Chao (MSc students) who aided me in my fieldwork whilst in South Africa. Finally, my family and Jacqueline deserve a special mention. Without their help and support it would not have been possible for me to fulfill an ambition.

5

Abstract Over-harvesting of the plant resource base by human activities is an important factor in contributing to the loss of biodiversity. Systematic conservation planning using modelling can help to inform the decision making process when designing conservation landscapes. The determination of plant species resource use can provide information of biodiversity loss and human impact that can support the planning process. More importantly, because of their cultural and economic importance the sustainable use of plant species can contribute to rural livelihood improvements and the conservation of habitats and ecosystems more generally. The present study has collected inventory data and modelled resource use on communal land in Maputaland, South Africa. Most rural populations pursue subsistence livelihoods and are reliant to some extent on the natural resource base for direct subsistence or indirectly for generation of income. A significant contribution of the study has been the adoption of a robust and practical inventory that prioritises species based on their relative value from the perspective of resource users that is predictive of resource use. The inventory of two sites in the Tembe and Mathenjwa Tribal Authorities found that contrary to expectations there was little evidence of over-harvesting of the resource base, either for subsistence or commercial purposes. Trade in fuelwood and crafts currently provide only limited entrepreneurial and employment opportunities. The majority of resource use, including medicinal plant bark stripping was for subsistence purposes and provides an important resource base for the rural communities. The socio-economic conditions and the wide social acceptance and compliance with regulations are the probable reasons why largescale exploitation is not occurring. Distance from subsistence agriculture and other landscape features were identified as being important predictors of resource use in the modelling and were used for the fine-scale GIS maps. These can be of value in providing recommendations for practical actions and effective land use planning for relevant stakeholders at the appropriate scale. The study would recommend that where high value plant species occur and where there is the potential for socio-economic conditions to rapidly change, that inventories and modelling incorporating methodologies used in the present study are implemented. They can identify effective conservation land use strategies that can be incorporated into the Maputaland conservation planning system. Key words

Maputaland, modelling, tree resource use, communal land

6

Introduction 1.1

Overview

The introduction provides an overview of the biodiversity crisis, the issue of over-harvesting and the role of systematic conservation planning in the design of conservation landscapes. The introduction describes the approaches that are used to identify priority plant species that reflect resource use that can be used in conservation strategies that benefit both biodiversity and rural livelihoods, emphasising the role of sustainable use. The importance of South African biodiversity and its significance to its people are discussed. Finally, the introduction focusses on the Maputaland Centre of Endemism and highlights key biodiversity and conservation issues with reference to the resource use of fuelwood and medicinal plants.

Human activities that result in habitat destruction, introduction of invasive species, pollution and overexploitation are resulting in an increasing loss of biodiversity. The implications of this are considerable and if continued unabated could lead to the loss of ecosystem function, loss of habitats and could undermine rural livelihoods through the degradation of the resource base (Pimm, Jones & Diamond 1988). The United Nations Convention on Biological Diversity was framed on the basis of reconciliation of environmental objectives and the need for social and economic development. The objectives of the convention are the •

Conservation of biodiversity

•

The sustainable use of biological resources

•

The fair and equitable sharing of benefits arising from their use

1.2

Systematic conservation planning and design of conservation land-scapes

There is a limit to the availability of land and resources for biodiversity conservation, but the threat to biodiversity are not uniformly distributed. Threatened areas tend to have high economic value, are transformed into agriculture and urban landscapes or have high human population density and poverty levels that lead to unsustainable resource use (Pressey et al. 1993). Systematic conservation planning is a process that can help to inform the decision making process when designing conservation landscapes. It generally involves mapping the distribution of different conservation features and existing Protected Areas, setting representation targets for each feature, measuring the effectiveness of the present Protected Area system in meeting these targets and using computer-based selection algorithms to identify additional sites (Margules & Pressey 2000). A

7

characteristic of such an approach is that conservation is efficient and designed to achieve targets at minimum costs that maximise the chances of achieving conservation goals. Importantly the process includes steps for the identification of targets and different land-use options through participation of relevant stakeholders, thereby increasing the transparency and support (Margules & Pressey 2000). Developing conservation landscape plans requires fine-scale biodiversity distribution data to delineate boundaries of proposed Protected Areas and other conservation areas. Unfortunately, very little species or population data are available at this scale, even for developed countries. However, it is not necessary to have data on all these biodiversity elements for establishing an effective network of conservation areas. Successful planning exercises have instead used habitat or landscape level conservation features that are known to reflect the distribution of other biodiversity elements and are identified through remote sensing. These data can then be supplemented with information on land ownership, land use, degree of threat and other factors to identify priority areas for conservation (Margules & Pressey 2000). However, remote sensing cannot identify where harvesting of vulnerable species might be occurring. Vegetation cover may not change at all and yet populations of high value, vulnerable species can be disappearing from over-exploitation. High profile examples of this include ‘the silent forests’ that have been affected by the bush meat trade (Cuaron 2000). But present harvesting of many plant species is also considered to be indiscriminate, destructive and unsustainable and is recognised as a serious threat to biodiversity (Cunningham 1988; Dold & Cocks 2002; Geldenhuys 2004). There is a need, therefore to combine large spatial scale analysis with monitoring species at the individual and population level to provide a comprehensive picture when carrying out conservation planning exercises. Modelling the harvesting impact on plant species can provide information on biodiversity loss and human impact. More importantly, modelling plant species resource use can also provide recommendations for strategies that can contribute to both rural livelihood improvements and conservation of biodiversity. The significance of plants, both culturally and economically, can be sufficiently great that sustainable use strategies can lay the foundation for conservation of not only vulnerable species, but of habitats and ecosystems more generally (Plotkin & Famolare 1992; Ticktin 2004). Modelling requires that plant species are chosen that reflect the patterns of resource use and that data collection is robust (Cunningham 2001b). Plant species that assume the highest relative value should be chosen as priority species as they are more likely to reflect harvesting practices. In addition, plant

8

species with the highest relative value are more appropriate for adoption in sustainable use conservation strategies (Cunningham 2001b). 1.3 1.3.1

Setting priorities for identifying plant resource use Resource value and priority plant species

Quantitative data collected through inventories is an essential component for modelling resource use. Inventories ideally should be low cost, robust and incentive-driven (Cunningham 2001a; Mattos, Nepstad, & Viera 1992). Two key considerations need to be addressed in the data collection if it is to be of value for predicting resource use. Firstly, resource value and priority species should be considered from the perspective of local livelihoods and resource users. Secondly, quantitative studies that can be used as a basis for resource management and conservation require long-term studies on marked populations, which are expensive, time consuming and not possible for the hundreds of species that are harvested. Resource management and conservation strategies are often urgently required. This necessitates the use of ‘short cuts’ and ‘conceptual filters’ for choosing priority species that is objective and consistent (Cunningham 2001a). The inventories are designed to measure resource value and how human impact threatens that value (Chapman 1987; Cunningham 1990; Cunningham 2001b). Defining the resource priorities is achieved by following a sequential logical approach to activities (Aumeeruddy-Thomas et al. 1999; Geldenhuys 2004; Tuxill & Nabhan 2001). The approach includes the pre-site collection of relevant information to include wider contexts of conservation, development and economic policy. Importantly it involves the identification of priority local resource issues and stakeholders. The activities involved are aimed at providing recommendations for practical actions in favour of conservation. One appropriate conservation strategy is to adopt the ‘sustainable use’ of plant resources that can benefit both biodiversity and improve rural livelihoods. Therefore species prioritisation should also reflect their potential for sustainable use in management programmes. This requires setting priorities on the basis of ecological principles, in addition to social, cultural and economic driving forces (Cunningham 2001b). 1.3.2

A hierarchical approach to understanding biodiversity and establishing priorities

It is appropriate to adopt a hierarchical approach and nested progression to understanding biodiversity and for identifying priority species (Noss 1990). Inventories often necessitate that human impacts are determined on individual plants first and then considered in terms of plant population dynamics and disturbance to determine community and land scape factors that provide the crucial context of resource use and value (MacNally & Quinn 1998).

9

The association between climate, soils, vegetation and landform use makes landscape classification linked with population density a useful tool in setting priorities. These relationships are evident even at a continent wide scale, with moist fertile soils associated with crop growing, dry fertile soils associated with pastoralists and moist less fertile soils associated with cultivation (Bell & McShane 1984). Changes in vegetation occurrence over large areas and long-time scales are possible with aerial photography and use of satellite imagery. They can be useful in understanding land-use patterns and as a planning and predictive tool for conservation programmes as demonstrated in the loss of woodland cover in N Namibia (Marsh & Seely 1992) and rate of forest clearance outside the BwindiImpenetrable National Park (Scott 1995). It is important to link different methods, a process that has been referred to as ‘cross-checking’, as they provide an improved explanation and better prediction of resource use (Cunningham 2001b). For example, participatory mapping techniques have been successfully developed for many rural communities (Poffenberger et al. 1992). They can illustrate the spatial distribution of vegetation types, resources and resource flows of significance and can be translated into topographical maps that give insights into how or why resources are valued (Baker & Mutitjulu Community 1992; Rundstrom 1990; Walsh 1990; Walsh 1993). 1.3.3

The use of ‘short cuts’ and conceptual filters

Coarse conceptual filters are used as ‘short-cuts’ to define ecological groups of plants and for the identification of species that are likely to be more resilient or vulnerable to harvest. Plant growth forms, which represent a sequence from trees through to shrubs and annual herbs, provide a first approximation of vulnerability and therefore value (Raunkiaer 1934). The plant parts that are harvested and the frequency and intensity of harvesting, allied to regeneration characteristics and growth rates are also important predictors of species vulnerability and resilience (van Wyk et al. 1996). The destructive harvest of an individual can seem extreme, but harvesting has to be seen in the perspective of plant populations which in turn need to be viewed in terms of their abundance (Peters 1994). A seemingly low impact at the individual level such as fruit harvesting can have long-term impacts on population of species as it can reduce seed recruitment. In contrast even harvesting of bark and roots that can lead to the destruction of individual trees may have little impact on populations of fast growing, fast reproducing species. Harvesting of Acacia karoo for rope making often kills the trees but at the population level, requirements are easily met by recruitment from the soil seed bank and in addition, human disturbance through agriculture and livestock activities favours Acacia populations (Cunningham 2001a).

10

The use of size-class distributions that measure tree diameter at breast height (DBH) is a practical field method for recording assessment of harvesting impacts, for developing survivor-ship curves and for illustrating the response of plant populations to harvesting. The use of survivor-ship curves has two main weaknesses. First is the assumption that DBH reflects plant age (Tietema 1993; Harper 1977). The second assumption is that the survivorship curve reflects harvesting impact alone, rather than the combination of vegetation disturbance dynamics. Despite these weaknesses, size-class distributions are considered useful predictive tools (Geldenhuys 1992; Cunningham 2001b). Geographic distribution is important in prioritisation of species from a conservation perspective. The highest conservation priority should be given to species with a narrow geographical distribution, restricted habitat and small population size (Cunningham 1991; Pitman et al. 1999; Rabinowitz, Cairns, & Dillon 1986). The extent of decline of a population of restricted range species is used to assign species to the IUCN Red List, and provides a globally accepted setting for priorities. Conservation priorities for many species are currently established in the face of information gaps, through Conservation Assessment and Management Plan workshops (Molur & Walker 1996) as demonstrated in a recent exercise for South African trees (Goldring 2002). It is crucial, however, that local resource values are not overshadowed by national and international conservation priorities (Cunningham 2001a). 1.3.4

Priority setting related to social, cultural and economic driving forces

Identification of priorities that are the focus for conservation is the initial step. An effective conservation policy should identify the stresses to determine how the priorities are threatened, where the sources of these stresses originate and develop practical solutions to reduce or eliminate the threats. Such an approach is more effective in circumstances where natural resource use assumes high importance, where local communities interests can be identified, if strategies are inclusive of diverse uses and there is wide social acceptance of management plans and regulations. Success is more likely where there is an existing integrated conservation and development project with the potential for support. Perhaps the greatest barrier to conservation strategies falls in the domains of social, economic and political considerations (Schopp-Guth & Fremuth 2001). The realisation of this potential is dependent greatly on a number of factors including, not least assured property and access rights for local resource users. Community-based conservation should be considered the most effective management strategy and success should be viewed as the long-term abatement of critical threats and the sustained or enhancement of biodiversity (IIED 1994).

11

1.4

South African biodiversity and conservation policies

South Africa contains an exceptionally rich and diverse array of life forms and ranks as the 3rd most biologically diverse country in the world. This is attributable to its richness of vascular plants (18,000 recorded species), of which 80% are found no where else (Hilton-Taylor 1996). Human activity has been responsible for changing and sculpturing the South African landscape for thousands of years, however, the pace and extent of change has increased rapidly with agricultural and industrial development, a growing human population and unsustainable rates of resource consumption. A comparison between the number of species listed in the Red Data Books between 1980 and 1995 shows an increase of 80%. South Africa has the dubious distinction of containing the highest known concentration of threatened plant and highest extinction estimates for any area in the world (15% of recorded species) (DEAT 1997; Hilton-Taylor 1996). Present day Protected Areas cover 72,000km2 (5.9%) and continue to play a central part in conservation policy. The National Parks have been managed according to the ‘The Yellowstone Model’ where settlement in the parks are prohibited and the use of resources for subsistence and commercial use banned (Muir 1916; Leopold 1949; Stevens 1997). South Africa’s Protected Areas while globally renowned for the contribution they have played in conservation of the national heritage and threatened species are likely to be too small and fragmented to maintain viable populations without costly and intensive management programmes (Goodman 2002). Many now argue that they do not form a holistic land use policy or are contributing to achieving satisfactory conservation outside of Protected Areas. An additional concern is that the Protected Area network has often been accompanied by forced removals and resource dispossession of local communities. The 1997 White Paper on the Conservation and Sustainable Use of South Africa’s Biological Diversity is orientated towards achieving goals of poverty alleviation and rural development, through activities that can act as incentive-driven conservation programmes. Conservation is increasingly expected to contribute to local community’s livelihoods (DEAT 1997). Indigenous forests and woodland resources are generally undervalued in South Africa despite the fact that millions of people rely on them for their livelihood support (Shackleton & Shackleton 2000). Harvesting of natural resources and their sustainable use can contribute to achieving these goals. The Policy for Sustainable Forest Development (DWAF 1996) and the 1998 Forest Act have guide policies that their management should include ‘comprehensive use’ and should contribute to sustainable development (DWAF 1997).

12

1.5 1.5.1

Harvesting of plant resources in The Maputaland Centre of Endemism Biodiversity and conservation in Maputaland

The Maputaland Centre of Endemism (Greater Maputaland) is an area of high conservation value covering approximately 20 000km2 spanning sections of Mozambique, Swaziland and South Africa and forms part of the Maputaland-Pondoland-Albany hotspot (Figure 1) (van Wyk 1994). It is an area of global importance for the conservation of vascular plants and forms one of southern Africa’s most important centres of floristic diversity and endemism (van Wyk1994; World Conservation Union and WWF-UK 1994). It contains an estimated 2500 vascular plants, of these at least 250 species / intraspecific taxa are endemic or near endemic to the region (van Wyk, Everard.D, Midgley, & Gordon1996). It also forms part of an Endemic Bird Area (Stattersfield et al. 1998) and has been identified by WWF as making up one of the 200 most important ecoregions (Olson & Dinerstein 1998). The present study focuses on the South African portion (referred to subsequently as Maputaland) which has an area of 9760km2 and lies between latitude 26.78 and 28.5 degrees South and 31.95 and 32.9 degrees East. The region is the most southerly part of the East Africa coastal plain and many species reach their southern most limits in Maputaland.

Maputaland is found in the NE of the province of KwaZulu-Natal, South Africa. The province is an important region for sub-tropical agriculture and plantation forestry and is the most highly populated in South Africa, home to 8.7 million people, 83% of whom are black (Central Statistical Services 1996). Population growth rate for the region is around 2.4% per annum and is associated with increasing rates of urbanisation, with Durban accounting for 50% of the KwaZulu-Natal population.

Conservation of biodiversity is the responsibility of Ezemvelo KwaZulu-Natal Wildlife (EKZN Wildlife) which is a parastatal body. Protected Area coverage in Maputaland has a combined area of 2482km2 (25.4%) of which 2010 km2 is terrestrial coverage (Figure 2) (Goodman 2002). The existing Protected Areas were mainly established to protect the remaining large mammal populations and recreational fishing sites (Goodman 2002). The majority of the Protected Areas are completely fenced to reduce poaching, livestock encroachment and human wildlife conflicts (Goodman 2002). It has been estimated that 30% of the Ingwavuma and Ubombo population were ‘forced’ to relocate with the establishment of the Ndumu GR, the Coastal Forest Reserve and Tembe Elephant Park. Much of the Protected Area land is subject to land claim by the dispossessed communities and is being investigated by the Restitution of Land Claims Commission.

13

Figure 1

Maputaland Centre of Endemism (Smith 2001)

14

Figure 2

Protected Area Coverage of Maputaland (Adapted from Smith 2001).

15

In the last 30 years EKZN Wildlife has been successful in the development of a ‘wildlife’ industry throughout KwaZulu-Natal, facilitating its wise use and granting full ownership rights and generous incentives. This has resulted in a thriving economic sector and gains in biodiversity. Mkuze GR once isolated and surrounded by hostile land use is now connected to large tracts of private and jointly managed estates. This represents part of a general commitment to involve local communities in biodiversity conservation. EKZN Wildlife has identified a range of strategies, which include the promotion of natural resource management and development of conservation-based entrepreneurial opportunities (Goodman 2002). 1.5.2

Conservation planning and management of biodiversity in Maputaland

The Lubombo Transfrontier Conservation Area under the guidance of the Peace Parks Foundation is developing a conservation planning system for the region that centres on existing reserves and Protected Areas and aims to reconnect wildlife migration patterns (Jones 2004). The South African portion is supported by the Lubombo Spatial Development Initiative to encourage these developments primarily through infrastructure improvements to attract tourists and commercial interests. A preliminary conservation land use plan for Maputaland has been developed to help inform land use decisions (Figure 3) (Smith et al. 2004). Areas of high conservation value were identified after performing a conservation planning exercise that identified biodiversity targets based on original land cover, through workshops involving EKZN Wildlife staff. Land cover elements that were endemic or were at risk of transformation were given higher conservation targets. The main exercise was to identify areas where local communities and private sector could run economically viable and benign conservation projects (http://www.mosaic-conservation.org/maputaland). However, despite the conservation importance and the large amount of the region that has Protected Area status, much of the biodiversity that remains is threatened by anthropogenic factors, related to increasing human population, acquisition of land for subsistence agriculture and commercial developments. Analysis of a landcover classification map that divides Maputaland into five main ecological zones and contains 29 natural and 5 transformed land types showed that more than 3% (0.28% per annum) of natural vegetation was cleared for subsistence agriculture between 1986 and 1998 (Matthews 2001; Smith 2001). The most important factor that determined transformation in Maputaland was explained by the distance to existing agriculture. This acts as a surrogate for human density and occurs at the interface between transformed and pristine areas. It is probable that socioeconomic and cultural factors and not population not per se that are important (Smith 2001).

16

Conservation value High

Low

Protected area Road

20 km Moçambique Swaziland

South Africa

Figure 3

Maputaland

Preliminary Conservation Planning Exercise for Maputaland

The map shows the results of a preliminary conservation planning exercise for the South African section of Maputaland, based on protecting important habitat types and maintaining connectivity between Protected Areas. http://www.mosaic-conservation.org/maputaland RJ Smith. DICE University of Kent at Canterbury.

17

1.5.3

The value of the natural resource base to rural populations in Maputaland

The people of Maputaland have diverse origins, but those of the Tembe-Tonga predominate. Historically the region formed part of the Tembe Kingdom that is now fragmented between South Africa and Mozambique. The poor soils and high levels of biodiversity have heavily influenced the human culture of Maputaland. Most rural populations pursue subsistence livelihoods and are reliant to some extent on natural resource use for direct subsistence or indirectly for generation of income. Plant resource harvesting has attracted an increasing amount of attention due to concerns that relate to rising levels of commercial trade and over-exploitation and the possible benefits of sustainable use management programmes (Cunningham 1990). There is a lack of a clear policy direction and support despite the obvious potential for such. This is no more starkly apparent than the current use of the natural resource base by rural populations for their energy and health requirements. 1.5.4

The fuelwood natural resource base

South Africa produces and consumes 60% of Africa’s electricity supply and yet 80-90% of rural and peri-urban households in Maputaland continue to use fuelwood as their primary energy source (Twine 2002). The national net direct use value of fuelwood after factoring in opportunity costs of labour have been estimated at R2 billion (Williams & Shackleton 2002). At national and regional levels the projected sustainable supply of fuelwood is estimated at 16 million tonnes (wood production from the savannah alone) well above the 9-10 million tonnes apparently required to meet fuelwood needs (Von Maltitz & Scholes 1995). Many villages face increasing shortages, which is largely a function of human population growth and clearance for agriculture (Banks et al. 1996). Quantities used per household range from 0.6-7.7 tonnes per annum and economic values range from no trade to R0.57 per kg (Banks et al. 1996; Gandar 1984). Fuelwood represents a source of livelihood to many. The number of households involved in fuelwood trade is unknown but is likely to fluctuate widely (varied between 7% and 53% in the Limpopo Province) (Shackleton & Shackleton 2000). Determinants of price relate to local wood availability, costs of alternatives and amount of disposable income. The absence of a local price within a rural village appears to be common in areas that are remote or have relatively abundant fuel stocks (Williams & Shackleton2002). Fuelwood collection is associated with a marked selection of species and preferred size classes. Local communities can make use of several dozen species (69 species in KwaJobe in KwaZulu-Natal and 40 species in Bushbuckridge Lowveld in Limpopo Province (Shackleton et al. 1999)). Fuelwood collection usually occurs in the immediate vicinity of the household. Generally live wood is only cut in times of decreasing availability. In many localities demand for fuelwood exceeds supply, which

18

may be associated with decreased frequency of gathering and consumption and increased use of alternatives (crop residues / dung / commercial fuels). 1.5.5

The medicinal plants natural resource base

An estimated 70-80% of people world-wide rely on traditional, largely medicinal plant medicines to meet their primary health care needs (Farnsworth & Soejarto 1991; Penso 1980). In South Africa there are an estimated 27 million users and the trade in medicinal plants forms a multi-million rand ‘hidden’ economy (estimated at R270 million per annum) (Dold & Cocks 2002; Mander 1998). The medicinal plant trade in South Africa and neighbouring countries is centred around KwaZulu-Natal with an estimated 6 million users (Cunningham 1988; Marshall 1998; Williams, Balkwill, & Witkowski 2000). The trade in medicinal plants represents a complex resource management issue in South Africa (Dold & Cocks 2002). In recent years there has been a trend towards commercialisation and shift from subsistence use, which has led to the increased intensity and frequency of harvesting. The factors responsible for this include: •

A rapidly growing and urbanising black population (Cunningham 1988).

•

Development of large urban markets (Durban, Johannesburg)

•

Patient ratio of 1:700-1200 for traditional healers compared to 1:17,400 for medical doctors (Marshall1998)

•

High rate of unemployment in rural areas.

•

Considered a basic requirement for treating certain conditions (Marshal1998).

•

Increasing official and societal recognition (Williams et al 2000)

•

Influx of ‘outsiders’ seeking work

•

International demand for medicinal plant products (Williams et al 2000)

Over 1032 species have been identified as being used in KwaZulu-Natal (Hutchings 1996) with 400 species traded commercially in Durban markets (Cunningham 1988). The species traded differ significantly between study areas, attributable to differences in healing practices, ethnicity and species availability (Dold & Cocks 2002). In South Africa trade is dominated by material with a long shelf life (bark, roots, bulbs, whole plants), with bark accounting for 1/3 of plant products traded in KwaZuluNatal markets (Cunningham 1988; Grace et al. 2003; Mander 1998; Williams et al 2000). The bark of many species are traded although a relatively small number are in high demand and intensively used (Cunningham 1988; Williams et al 2000).

19

Harvesting of medicinal plants was formerly the domain of trained traditional herbalists, medicinal practitioners and diviners. Strict customary practices were respected that regulated plant collection and quantities. There is now a substantial network of commercial gatherers (mainly black women), traditional healers and customers (Mander1998; Williams et al 2000). Present harvesting is considered to be indiscriminate, destructive and unsustainable for many of the popularly traded species and is recognised as a serious threat to biodiversity (Cunningham 1988; Dold & Cocks 2002; Geldenhuys 2004). Demand generates a species-specific trade that has resulted in the local extirpation of several species (Siphonochilus aethiopicus, Warburgia salutaris). In southern Natal forests 51% of Ocotea bullata and 57% of Curtisia dentata had bark removed. In Maputaland all Warburgia salutaris were completely debarked (Cunningham 1988). Communal areas closest to Durban have been the primary focus for harvesting activities (Cooper 1979). As high value species have been depleted commercial harvesting has occurred from more distant and remote areas (Cunningham 1988). In the Durban markets 43% of traders travel from distant localities (200-500+ km) (Mander 1998). There been little policy support for commercial trade in medicinal plants despite their economic and cultural importance. Under development has a significant implication for consumer welfare, market operation and biodiversity (Diederichs, Geldenhuys, & Mitchell 2002).

The White Paper on

Sustainable Forest Development in South Africa (DWAF 1996) and the National Forestry Action Programme (DWAF 1997) recognises fuelwood as the main energy source and that access to sustainable and secure supplies is important for rural households (DME 1998). Despite this agenda, actions and programmes have been based on reduction of consumption rather than investing in sustainable use policies. A lack of clear policy direction relates directly to the inadequacy of supply and demand data and understanding of resource strategies from which to make informed management decisions (Diederichs, Geldenhuys, & Mitchell 2002; Williams & Shackleton 2002). 1.6

Study aims and objectives

Modelling the harvesting of plant species can provide information on the extent of biodiversity loss and human impact on the resource base. More importantly, it can provide recommendations for practical actions in favour of conservation or related objectives that can contribute to both rural livelihood improvements and enhancement of biodiversity. The study’s principle aim was the collection of field data, that is practical, has a scientific basis and at the relevant spatial scales to determine resource use on communal land in the Maputaland ecoregion. Logistic regression modelling and GIS mapping of the field data was incorporated with existing spatial / infrastructure data to model harvested areas.

20

The project’s objectives are •

To develop a list of species to be used in inventories that reflect resource use.

•

To perform an inventory analysis of priority species that measures and maps resource uses.

•

To use logistic regression to model distribution of resource use.

•

The production of ‘fine-scale’ GIS maps that model resource harvesting and can be of value in systematic conservation planning.

21

2 2.1

Study area and methods Overview

The methods provide a general description of Maputaland and site specific description of two study areas chosen for inventory analysis, highlighting the conservation issues and disturbance threats. It then describes the methodologies employed for prioritisation of the species to be included in the inventory, the sampling design for assessing resource value and harvest impact and finally the modelling procedures that can be used to predict resource use. 2.2

The Maputaland study area

Maputaland consists of two major topographical features, the Lebombo Mountains and the Coastal Plain, both of which extend northwards into Mozambique where they are more extensively developed. The mountain range marks the position of the coastline that was formed during the Jurassic period. Since then, the sea level has risen and fallen depositing and eroding a variety of soils, silts and clays (Figure 4). The geology is diverse, consisting of Mesozoic, Tertiary and Quaternary sequences. The Lebombo Mountains are composed of 180 million-year-old Jozini formation rhyolites. At their foot are a series of Cretaceous sediments, which have both riverine and marine origins. Pleistocene sediments form a 50m covering on the Tertiary and Cretaceous rocks and gives rise to the extensive dunes of sand that characterises most of Maputaland. Six major rivers flow through Maputaland. A number of smaller rivers originate in either the Lebombo Mountains or in the coastal plain. All the rivers that flow through the coastal plain are seasonal, usually drying out by mid-winter. There are three natural lake systems, found in the East of the region and there are a number of smaller water bodies, many of which have associated wetlands (Figure 5). Maputaland is subtropical and rainfall is highly seasonal, 77% occurring in the hot summer months. The striking climate feature is the variation in rainfall across the region. Rainfall varies from 11001200mm on the coast, to 600mm on the coastal plain, rising to 800mm on the crest of the Lebombo Mountains. Annual rainfall figures can vary dramatically between years. The current ecosystem in the region is considered to be of recent derivation, many endemic plant taxa comply with the concept of neo-endemics and have evolved to occupy the ecological niches created by the recent formation of the coastal plain. In addition the well-defined climatic and geological conditions have produced a series of ecological zones that contain many distinct habitat types.

22

The region has been neglected for many years and is characterised by high population growth, high levels of poverty and poor economic development. The Ubombo region has increased from a density of 5.2 to 21.4 people per km2 between 1936 and 1990. In 1986 the mean income of the majority of households of Maputaland was US$21 per month and in a recent study in the Tembe Tribal Authority the mean household income was just R493 (US$22) per month. Most households typically have small dry land agricultural plots surrounding their homesteads, which are usually not self sufficient due to the poor sandy soils and lack of water. Minimal cash flows into the communities and 90% of individual household production are consumed within the family unit. Most cash is spent on basic foodstuffs with the remainder on transport, health and school fees (Jones 2004). The primary resource is land and wood used for the construction of traditional homesteads, fuelwood, wild food collection including bushmeat and harvesting of medicinal plants and animals (Cunningham 1985; Jones 2004). There are few formal jobs, some income is generated selling food and trade in natural resources, but most households are extremely reliant on remittances, pensions and child support grants from the government.

23

Figure 4

Ecological Zones and Major Topographical Features of Maputaland (Adapted from Smith 2001).

24

Figure 5

Rivers, Lakes, Roads & Towns of Maputaland (Adapted from Smith 2001)

25

2.3

Description of study areas

Two study areas for the performance of resource inventories were identified on the basis of their conservation value, concerns regarding over harvesting, the relevance of the natural resources to the rural population and the potential for the adoption of sustainable use strategies. Other factors included the presence of an existing conservation development programme (Lubombo Transfrontier Conservation Area) which is supported by the Lubombo Spatial Development Initiative. There has been no formal quantitative evaluation of natural resource use in the two areas. Resource managers had identified medicinal plant (bark stripping), fuelwood and harvesting of wood for construction and for crafts as urgent resource issues. The extent of resource use for subsistence vs commercial purposes, local vs outsider users, market flows and the impacts on the resource base were unknown. 2.3.1

Study area 1 – conservation issues and disturbance threats

Study area 1 is contained within the Tembe Tribal Authority and straddles the boundary of three wards. It has a total area 1212km2, a population of 31,650 residents (99% Black African origin), population density of 26 people / km2 and contains 4868 households (Figure 6). Study area 1 is comprised of two distinct clearly bound habitat types, sand forest consisting of dry deciduous and semi-deciduous elements which form dense impenetrable stands up to 25m tall and Terminalia woodland characterised by common woody savannah species. The woodland has a well developed understorey represented by Aristida, Pogonarthria, and Perotis grasses (Figure 9). Sand forest forms one of the three distinct types of forest (interlocking canopies) recognised in southern Africa and is more or less restricted to the deep sands of Maputaland and the Maputaland portion of Mozambique (where it is referred to as Licuati Forest) (Low 1996). Sand forest is considered to be relics of coastal dune forests that remained as the coastal plain developed eastwards. Sand forest vegetation harbours the highest plant biodiversity and the highest proportion of endemic species of all habitat types within the Maputaland Centre of Endemism. Of 225 Maputaland Centre endemics, 30 are associated with sand forest and 20 species restricted to it (Matthews et al. 2001). Most of the endemic vertebrate species are likewise restricted to this habitat type (van Rensburg et al. 1999; van Rensburg et al. 2000). Species composition and species diversity declines southwards as tropical elements decrease with changes in climate. Sand forest is a unique and important endemic habitat, 45% of which has already been transformed due to disturbance related to increasing land use for grazing activities, subsistence use and overharvesting of natural resource base. 44% of the remaining sand forest is located within Protected 26

Areas, however, the largest of these remnant patches found in Tembe Elephant Park is suffering disturbance impacts due to the presence of elephants that have recently been contained within the park boundaries (Matthews, van Wyk, van Rooyen, & Botha 2001). Human impact and disturbance on communal land is related to grazing activities, subsistence agriculture and harvesting of the resource base. 2.3.2

Study area 2 – conservation issues and disturbance threats

Study area 2 is contained within a single ward of the Mathenjwa Tribal Authority. It has a total area of 263 km2, a population of 12846 (>99% of Black African origin), population density of 49 people / km2 and contains 2052 households (Figure 7). Study area 2 is of predominantly Lebombo Woodland habitat type, but other habitat types including Lebombo thicket, Lebombo grassland and Lebombo aquatic are present. Lebombo woodland has a low species diversity dominated by Combretum sp. and Acacia sp.(Figure 10). The Usuthu Gorge conservation corridor forms part of the study area and has been earmarked for a community-based conservation initiative that is being developed in the region to establish ecotourism and development opportunities. Land has been donated by the local communities and is being fenced, prior to the introduction of wildlife. Human impact and disturbance is related to grazing activities, subsistence agriculture and harvesting of the resource base.

27

Ndumu

Tembe Elephant Park Makhane Kwa-Ndaba

Embonisweni Mpophomeni Esicabazini

Zama-Zama 5km

Figure 6

Study Area 1

28

Ndumu Game Reserve

Mabona Mpolimpoli

Khume

5km Mbadleni

Figure 7

Study Area 2

29

2.4

Quantitative resource inventories

A sequential approach was adopted that links different methodologies at appropriate scales and incorporates ‘cross-checking’ procedures that identifies resource use. The sequential approach involved, identification of the geographical focus, pre-site collection of relevant information, quantitative studies and modelling resource use. Two key considerations were addressed in the methodology. First was the consideration of resource value from the perspective of local livelihoods. Secondly, time and budget constraints necessitated the use of ‘short cuts’ and conceptual filters for choosing priority species and short-term ‘snap-shot’ inventories. The inventory methodology measured resource value and human impact. The resource objective and inventory methods were decided on in the field (in situ) through preliminary fieldwork. Prioritisation of the focus species for the study was inclusive of ecological, socio-economic and cultural perspectives. The limitations of time and budget constraints, allied to the habitat diversity necessitated the use of short-term ‘snap shot’ surveys. Trees are useful prioritisation species on the basis of their importance as a natural resource use and because they are vulnerable to human impact. Many tree species are slow growing, slow to reproduce, have specific habitat requirements and limited distribution. Two aspects of tree resource harvesting were measured, stem cutting and bark stripping. Resource harvesting of bark stripping and stem cutting can be destructive and remain visible for years and allowed field assessment of damage to individuals and populations. 2.5

Pre-site evaluation of study areas and determination of priority species

‘Cross-checking’ included a literature review of species ecology, resource supply and demand, interviews with key informants, local resource users, resource managers and ‘walk in the woods’ approach (Cunningham 2001b). It included the determination of the primary resource user groups and whether harvesting was for predominantly subsistence or commercial trade. Pre-site evaluation of study area 2 included evaluation of the topographical maps produced from the participatory mapping project which illustrated the spatial distribution of vegetation types, resource areas, resource flows, and landscape features (Chao 2004). This identified the priority species that were inventoried in the two study areas. The resource inventory was performed in two phases: •

Phase 1 utilised regional inventories that showed the distribution and composition of major habitat types and formed the basis for the location of sampling (Figure 8) (Smith 2001).

•

Phase 2 involved the quantitative assessment of harvesting impacts of prioritised species that form the basis for spatial mapping and modelling resource harvesting.

30

Figure 8

Landcover Classification Map of Maputaland (adapted from Smith 2001)

31

2.6

Sampling design and inventory methodology for assessing harvesting impact

The inventory data collected formed a description of the resource base, species composition and intensity of harvest of the priority tree species. Resource users contributed throughout the inventory and provided insights on the technology and tools used for harvesting, the selection criteria for resource use and whether the species showed evidence of regeneration. The inventory of the two study areas was performed over a 6 week period in May and June of 2004. Sampling was performed with a map of the vegetation communities as a basis and with consideration of each resource use option, target species and habitat diversity. The sampling provided a representative cover of the study sites and reliable estimates of resource use and species distribution. Belt transects were chosen as the most appropriate method for sampling (Cunningham 2001a). A stratified sampling strategy was employed based on access via the road network. The first transect was chosen at random and the rest followed a set spacing to ensure that the study areas were sufficiently surveyed. Belt transects had a width of 10m in study area 1 and a width of 5m in study area 2. These differences were based on the density of target tree species at the two study sites. The sampling strategy was designed so that the belt transects followed a compass bearing that travelled at right angles through subsistence agriculture and continued into ‘pristine’ habitat moving away from transformed land. The presence of each target species was mapped (8m resolution) by recording the co-ordinates using a Garmin 12 GPS satellite navigation system (Garmin Corp., Ulathe, KA). Size-class determination of each tree was measured by recording DBH. Only trees with a DBH greater than 10cm was recorded in the inventory. Size-classes were divided into 10cm increments between 10 and 100cm, and thereafter into 101-125cm, 126-150cm, 151-200cm, 201-301cm, 301400cm and >401cm size-classes. The intensity of resource harvesting impact was rated by visual assessment. For bark stripping a four point scale was used based on the proportion of bark removed from the tree trunk below 1.3m (1-10%, 11-30%, 31-50% & 51-100%). A similar 4-point scale was used to assess the percentage of the tree estimated to have been removed by stem cutting (1-10%, 11-50%, 51-90% and 91-100%). Species were ranked according to three criteria, species population abundance, the total number of trees harvested and thirdly as a percentage showing evidence of resource use out of the total population of that species sampled. The third criteria prioritise species with the highest relative resource value from the perspective of resource users. Resource value was also determined for the different user groups for bark stripping and stem cutting. Information on harvest intensity on

32

individual trees was combined with DBH measurements and size-class determinations for the development of survivorship curves to provide an assessment of harvesting impact at the population level. The opportunity for sustainable use of the chosen priority species was assessed based on resource value, ecological considerations and resilience to harvest impact. Species were scored for each of the factors based on the inventory results and ‘cross-checking’ methods. 2.7

Modelling

Logistic regression analysis was used to identify the factors that determine natural resource use in the two study areas and for the production of GIS maps that modelled the spatial pattern of resource harvesting. The model predicts the likelihood that an individual tree will be harvested. Stem cutting and bark stripping resource harvesting was analysed separately, as they reflect different resource user groups. The tested variables were distance to roads, distance to existing subsistence agriculture, elevation, slope, tree species and DBH. To facilitate the analysis all data was imported into ArcView v.3.2 GIS software package (ESRI Inc., Redlands, CA) for manipulation prior to analysis. The road vector files were digitised from Landsat 5 satellite images (Smith 2001). Subsistence agriculture polygons were based on the landcover (30m) classification (Smith 2001). The elevation and slope were based on 90m resolution SRTM data (Smith 2001). Distance from each tree to the road vector and subsistence agriculture polygons was determined using Nearest Distance Extension in ArcView. Data was exported into SPSS v.9 (SPSS Inc. Chicago) statistical software for analysis. The data was analysed using multiple stepwise logistical regression (Forward Wald) to determine whether any of the independent variables affected the probability of resource harvesting with entry and exit variables determined by the Wald statistic with P-values of 0.05 and 0.1 respectively. The relative contribution of variance to the model was determined by the R statistic. Model performance on the testing sets was evaluated by calculating the area under the curve of the receiver operating characteristics (ROC) plots. ROC values range from 0.5 to 1.0. Values above 0.7 indicate a good model fit while those above 0.9 indicate a highly accurate model (Pearce 2000).

33

The resultant model was used to calculate the predicted value for resource harvest. The regression model for calculating the probability of resource harvest was calculated using the following equation

Probability =

e α + β 1x1 + β 2x2 +β 3x3 1 + e α + β 1x1 + β 2x2 +β 3x3

Where α = constant β1 = regression coefficient of the first significant variable β2 = regression coefficient of the first significant variable β3 = regression coefficient of the first significant variable (when applicable) x1 = the first significant variable x2 = the first significant variable x2 = the first significant variable

The model was then transformed into a 25m-resolution risk coverage using the Map Calculator in ArcView. Site-specific resource use and the association with important variables identified in the model was determined by dividing each of the transects into 100m sections in ArcView using the extension Divide Line. The percentage of trees that showed resource use in each transect section was then plotted against significant factors identified in the model.

34

Figure 9

Sand Forest Habitat (Study Area 1)

35

Figure 10

Lebombo Woodland Habitat (Study Area 2)

36

Figure 12

Bark Stripping Resource Use

37

a)

b)

c)

d)

Figure 11

Stem Cutting Tree Resource Use a) Honey Collection b) Stem Cut for Poles c) Poles Waiting for Collection d) Bush Knife Handles 38

3 3.1

Results Resource harvest inventory, species prioritisation and resource use

The methodology included 8 priority species from study area 1 and 5 priority species from study area 2. Table 1 shows the priority tree species inventoried and the species habitat requirements. Additional tree species identified as being used included Hymenocardia ulmoides (fuelwood / poles), Strychnos madagascariensis (fuelwood), Manilkara discolor (fuelwood), Terminalia sericea (fuelwood /poles/

bark used medicinally), Psydrax lucopules (general-purpose timber), Diospyros inhacaensis (poles), Albizia vericolor (bark used medicinally), Albizia forbesii (trade in crafts), Trichilia dregeana (bark

used medicinally). Many other species not documented in the study are used for a variety of purposes. (See Figures 11 and 12 for examples of resource use). A total of 22 separate tansects were walked in study area 1 (total length 40.01km, mean 1.82 km, sampled area of 0.4 km2) and 6 transects in study area 2 (total length 13.6km, mean 2.26km, sampled area 0.14km2). A sample size of 3257 and 2208 trees were recorded in study area 1 and 2 respectively. A high percentage of trees showed no evidence of resource use, 87% (2834) in study area 1 and 89.6% (1979) in study area 2. Of the priority trees inventoried, 9.7% (317) in study area 1 and 10.2% (229) in study area 2 showed signs of stem cutting. The majority of the use was for stem cutting of live wood for collection of fuelwood, poles and general-purpose timber. In study area 1, 0.4% (12) of the focus species had evidence of destructive harvesting for the collection of honey. Only 2.9% (94) in study area 1 and 0.2% (4) in study area 2 had evidence of bark stripping. The majority of bark stripping was related to medicinal use.

39

Table 1 Priority Tree Species Innventoried Based on ‘Cross-Checking’ Methods

40

3.2

Relative resource value, tree abundance and user groups

Tree species were ranked according to their abundance, the number of individual trees harvested and importantly, according to the number used out of the total population of that species sampled, a measure of their relative use value (Table 2). In addition, tree species were ranked according to their relative use value for different resource user groups, either stem cutting (Table 3A) or bark-stripping (Table 3B) The population abundance of the species varied. Some species such as Dalium schlechteri (964), Cleistanthus schlechteri (1028) in study area 1 and Combretum zeyheri (1499), Acacia burkei (402) in

study area 2 are abundant and common. Other species, such as Brachylaenia huillensis (204), Newtonia hildebrandtii (270), Balanites maughami (244) and Ptaeroxylon obliquum (217) were less

common. A few species such as Erythrophleum lasianthum (104) were uncommon. The inventory identified species with the highest relative resource value. These included popularly harvested and uncommon species such as Brachylaenia huillensis (21.6%), Erythrophleum lasianthum (26%) and Newtonia hildebrandtii

(18.9%) in study area 1 and Acacia burkei (12.4%) and

Combretum zeyheri (11.1%) in study area 2. Other species, such as Cleistanthus schlechteri (12.1%)

although popularly harvested because they are common species have lower relative use values. Differences in the resource value of tree species for the different user groups are evident. Erythophleum lasianthum has the highest relative value for bark resource users (23.1%) but the lowest

value for stem resource users (2.9%). Similarly Brachylaena huillensis has little resource value to bark users (0%) users and the highest relative value for stem cutting (21.6%).

41

Table 2

Relative Resource Value of Priority Species

Shows priority tree species ranked according to their abundance, the number of individual trees harvested and importantly, according to the percentage showing evidence of resource use out of the total population of that species sampled. The final column prioritises species from the perspective of resource users. (1 = Highest rank).

Tree species

Study area 1 Erythophleum lasianthum Brachylaena huillensis Newtonia hildebrandtii var hildebrandtii Balanites maughami Ptaeroxylon obliquum Cleistanthus schlechteri Albizia adianthifolia Dalium schlechteri Study area 2 Acacia burkei Combretum zeyheri Combretum molle Philonoptera violacea Sclerocarya birrea subsp, caffra

Table 3

Rank population abundance (number sampled )

Rank Resource use resource use as % of the (number used) total species number

Rank Use as % of species abundance

8 (104) 7 (204) 3 (270) 4 (244) 6 (217) 1 (1028) 5 (226) 2 (964 )

7 (27) 4 (44) 3 (51) 5 (39) 6 (27) 1 (124) 8 (26) 2 (85)

26% 21.6% 18.9% 16% 12.4% 12.1% 11.5% 8.8%

1 2 3 4 5 6 7 8

2 (402) 1 (1499) 5 (57) 4 (72) 3 (178)

2 (50) 1 (166) 4= (3) 4= (3) 3 (4)

12.4% 11.1% 5.2% 4.2% 3.9%

1 2 3 4 5

Relative Resource Value of Priority Species and different user groups

Shows priority tree species ranked from the perspective of different resource user groups (1 = Highest Rank). Resource use has been separated into A) Stem cutting & B) Bark stripping resource use.

A) Stem cutting associated with harvesting of fuelwood, general-purpose timber and poles for construction. Tree species Brachylaena huillensis Newtonia hildebrandtii var hildebrandtii Ptaeroxylon obliquum Cleistanthus schlechteri Dalium schlechteri Balanites maughami Albizia adianthifolia Erythophleum lasianthum B)

Resource use as % of the total species number Stem cutting resource use 21.6% 14.4% 12.4% 10.3% 8.2% 4.9% 3.1% 2.9%

Rank 1 2 3 4 5 6 7 8

Bark stripping associated with harvesting for medicinal uses.

Tree species Erythophleum lasianthum Balanites maughami Albizia adianthifolia Newtonia hildebrandtii var hildebrandtii Cleistanthus schlechteri Dalium schlechteri Brachylaena huillensis Ptaeroxylon obliquum

Resource use as % of the total species number Bark stripping resource use 23.1% 11.1% 8% 4.1% 1% 0.4% 0% 0%

42

Rank 1 2 3 4 5 6 7 8

3.3

Relative resource value, individual tree harvest intensity, harvest impact and resilience

The use patterns of the individual trees is related to their resource value (Figure 13a, 13b & 13c). For stem cutting resource use, species with the highest relative resource value (Brachylaenia huillensis, Ptaeroxylon obliquum, Cleistanthus schlechteri, Acacia burkei and Combretum zeyheri) are more

likely to be destructively harvested at the individual level. Individuals of species with lower relative values are less intensively harvested. Use patterns of individual trees are also related to the end product user requirements. Large forest trees such as Newtonia hildebrandtii and Balanites maughami are less likely to be destructively harvested for general-purpose timber uses or fuelwood collection. Species such as Brachylaena huillensis, Ptaeroxylon obliquum are single-stemmed species valued for poles used for building and harvesting

often involved complete destruction of the individual. Some species that are of high value and destructively harvested at the individual level because of the method of harvest have high a resilience to harvesting use patterns. 60% of Acacia burkei trees were heavily harvested (>91% stem cutting), but because harvesting involves the removal of branches rather than cutting the main stem, 87% (27/31) showed complete regrowth. Similarly species with high relative values, that include Erythrophleum lasianthum, Balanites maughami and Albizia adianthifolia are more likely to be destructively harvested at the individual

level for bark stripping uses.

43

100

80

100

Brachylaenia huillensis (n=44) Rank 1

80

%

60

%

60

40

40

20

20

0 1-10%

11-50%

51-90%

0

91-100%

1-10%

Harvest intensity - Stem cutting

Ptaeroxylon obliquum (n=27) Rank 3

80

%

%

60

40

40

20

20

0

91-100%

Cleistanthus schlechteri (n=106) Rank 4

0 1-10%

11-50%

51-90%

91-100%

1-10%

Harvest intensity - Stem cutting

11-50%

51-90%

91-100%

Harvest intensity - Stem cutting

100

100

Dalium schlechteri (n=79) Rank 5

80

%

60

%

60

40

40

20

20

0

Balanites maughami (n=12) Rank 6

0 1-10%

11-50%

51-90%

91-100%

1-10%

Harvest intensity - Stem cutting

11-50%

51-90%

91-100%

Harvest intensity - Stem cutting

100

80

51-90%

100

60

80

11-50%

Harvest intensity - Stem cutting

100

80

Newtonia hildebrandtii (n=39) Rank 2

100

Albizia adianthifolia (n=7) Rank 7

80

%

60

%

60

40

40

20

20

0

0 1-10%

11-50%

51-90%

91-100%

1-10%

Harvest intensity - Stem cutting

Figure 13A)

Erythrophleum lasianthum (n=3) Rank 8

11-50%

51-90%

91-100%

Harvest intensity - Stem cutting

Stem cutting – Harvest Intensity of Individual trees (Study Area 1)

Trees are ranked according to their relative use value. Shows the % of trees and the harvest intensity.

44

100

100

80

Acacia burkei (n=50) Rank 1

80

%

60

%

60

40

40

20

20

0

0

1-10%

11-50%

51-90%

91-100 %

1-10%

Harvest intensity - Stem cutting 100

80

51-90%

91-100 %

100

Combretum molle (n=30 Rank 3

Philonoptera violacea (n=3) Rank 4

80

%

%

60

40

40

20

20

0

0 1-10%

11-50%

51-90%

91-100 %

1-10%

Harvest intensity - Stem cutting

80

11-50%

Harvest intensity - Stem cutting

60

100

Combretum zeyheri (n=165) Rank 2

11-50%

51-90%

91-100 %

Harvest intensity - Stem cutting

Sclerocarya birrea subsp. caffra Rank 5 (n=4)

%

60

40

20

0 1-10%

11-50%

51-90%

91-100 %

Harvest intensity - Stem cutting

Figure 13B)

Stem cutting – Harvest Intensity of Individual trees (Study Area 2)

Trees are ranked according to their relative use value. Shows the % if trees and the harvest intensity

45

100

80

100

Erythrophleum lasianthum (n=24)

80

60

%

%

60

40

40

20

20

0

0

1-10%

11-30%

31 - 50%

51 - 100%

1-10%

Harvest intensity - bark strip

80

11-30%

31 - 50%

51 - 100%

Harvest intensity - bark strip

100

100

80

Albizia adianthifolia (n=18)

60

Newtonia hildebrandtii (n=11)

60

%

%

40

40

20

20

0

0

1-10%

11-30%

31 - 50%

51 - 100%

1-10%

Harvest intensity - bark strip

80

11-30%

31 - 50%

51 - 100%

Harvest intensity - bark strip

100

100

Cleistanthus schlechteri (n=10)

80

Dalium schlechteri (n=4)

%

60

%

60

40

40

20

20

0

0 1-10%

11-30%

31 - 50%

51 - 100%

Harvest intensity - bark strip

Figure 13C)

Balanites maughami (n=27)

1-10%

11-30%

31 - 50%

51 - 100%

Harvest intensity - bark strip

Bark stripping – Harvest Intensity of Individual trees (Study Area 1)

Trees are ranked according to their relative use value. Shows the % of trees and the harvest intensity.

46

Table 4 Opportunity for Sustainable Use Based on Predictors of Resilience to Harvesting

47

3.4

Harvesting population and landscape level effects and potential for sustainable use

Species that have the highest relative value to resource users and are destructively harvested such as Brachylaena huillensis and Erythophleum lasianthum show size-class distribution curves that deviate

from ‘model’ distribution curves (Appendix 2). The majority of the other species have size-class distribution curves that conform to the ‘ideal distribution’ models. The opportunities for sustainable harvest based on predictors of resilience or vulnerability to harvesting and the relative resource value of the species inventoried is shown in Table 4. The table provides a first approximation of the potential for sustainable use of the priority species. 3.5

Socio-economic results of resource use from the ‘cross-checking’ methods

A limited trade in one species (Newtonia hildebrandtii var hildebrandtii) for fuelwood was identified at 19 sites along the road. In addition, a small trade (8-10 small stalls) in local crafts produced from two species (Cleistanthus schlechteri & Albizia forbesii) was occurring along the same road (Figure 6). Resource use in both study areas was restricted to local users and harvesting involved the use of ‘traditional’ implements, either bush knife or small hand saw. Evidence of chain saw use was observed on only two occasions with the cutting of two large Newtonia hildebrandtii var hildebrandtii specimens. 3.6

Logistic regression modelling of resource use and prediction of harvest impact

The initial models for resource harvesting included the variables of distance from subsistence agriculture, distance from road, elevation, slope, tree species and DBH. The final model for stem cutting included distance from subsistence agriculture (P <0.001), elevation (P=0.001), slope (P=0.006) and tree species (P<0.001) (Table 5). Model performance from the ROC value was 0.734, indicating a good model fit. Not unsurprisingly, harvesting is species-specific. Importantly, the model shows that trees that are close to subsistence, at low elevation and on flat ground were more likely to be cut (Figure 14). Distance to road had a small negative contribution to the initial model and was not included in the final model. The final model for bark stripping included distance from subsistence agriculture (P=0.016), DBH (P<0.001) and tree species (P<0.001) (Table 6). Model performance from the ROC value was 0.896, indicating a highly accurate model. Again, harvesting is species-specific. Trees with large DBH that were close to subsistence were more likely to be cut (Figure 15).

48

Table 5

Results of multiple logistic regression showing the factors that significantly determined stem cutting resource harvesting.

Factor

B

Regression coefficient df

Wald

Significance

Subsistence agriculture Elevation Slope Tree Constant

-0.002 -0.017 -0.275

-0.238 -0.091 -0.061 0.153

86.8 14.5 7.6 89.2 8.6

0.0000 0.0001 0.0058 0.0000 0.0034

Table 6

1 1 1 12 1

3.070

Results of multiple logistic regression showing the factors that significantly determined bark stripping resource harvesting.

Factor

B

Regression coefficient df

Wald

Significance

Tree DBH Subsistence agriculture Constant

-1.976 0.009 -0.001 0.427

0.368 0.185 -0.119

50.8 11.3 5.8 0.1

0.0000 0.0008 0.0157 0.7239

3.7

7 1 1 1

GIS map and graphical representation of resource harvesting risk

The bark stripping and stem cutting models were used to calculate the predicted amount of harvesting, which was transformed into the 25m resolution risk coverage maps. Distance from subsistence agriculture, elevation and slope was included in the risk coverage map for stem cutting and distance to subsistence agriculture included for bark stripping. The GIS maps provide a visual representation of the probability of tree harvesting for stem cutting (Figure 16) and bark stripping (Figure 17) for the Maputaland area for the factors included in the model. Figure 18 shows a graphical representation of how the predicted probability of resource harvesting for stem cutting and bark stripping changed with distance from subsistence agriculture, one of the significant factors in the model How the amount of resource use changed with distance from subsistence is shown in Figure 19. A high percentage of trees, 18% for stem cutting and 23% for bark stripping were harvested in the vicinity of subsistence agriculture. This decreased to less than 5% of the trees beyond 1000m. The analysis included resource use of all the trees for stem cutting, but for bark stripping, included the four species with the highest relative value, Erythophleum lasianthum, Balanites maughami, Albizia adianthifolia and Newtonia hildebrandtii var hildebrandtii.

49

A).

B).

C). 1.30

116

500

115

400

1.25

200

1.20 slope

elevation (m)

distance (m)

114

300

113

1.15 112

100

1.10

111

110

0 No use

Figure 14

1.05 No use

Stem cutting

Stem cutting

No use

Stem cutting

Factors Significant in Modelling Stem Cutting

A).

B).

500

150

400

120

300

90

DBH (cm)

distance (m)

Shows the mean and standard error of the factors that were significant in the model for pre stem cutting resource use. A) Distance from subsistence agriculture B) Elevation C) Slope.

200

100

60

30

0

0 No use

Figure 15

Bark strip

No use

Bark strip

Factors Significant in Modelling Bark stripping

Shows the mean and standard error of the factors that were significant in the model for predicting bark stripping resource use. A) Distance from subsistence agriculture B) DBH.

50

Probability of Stem Cutting

Figure 16

Modelling Stem Cutting Resource Use in Maputaland

51

Figure 17

Modelling Bark Stripping Resource Harvesting in Maputaland

52

Bark strip

0.7

Stem cutting

Probability of tree harvest

0.6 0.5 0.4 0.3 0.2 0.1 0 0

500

1000

1500

2000

Distance from subsistence agriculture (m)

Figure 18

Predicted probability of Resource Harvesting and Distance from Subsistence Agriculture

53

2500

3000

A) 30

Stem cutting

% of trees harvested

Log. (Stem cutting) y = -4.2155Ln(x) + 32.783 R2 = 0.8638

20

10

0 0

200

400

600

800

1000

1200

1400

1600

Distance from subsistence agriculture (m)

B) Bark strip

30

% of trees harvested

Log. (Bark strip) y = -5.5323Ln(x) + 43.239 R2 = 0.6796

20

10

0 0

200

400

600

800

1000

1200

1400

1600

Distance from subsistence agriculture (m)

Figure 19

Site Specific Resource Harvesting and Distance from Subsistence Agriculture

A) Shows the resource use at study areas 1 and 2 for stem cutting and distance from subsistence agriculture. B) Shows the resource use at study site 1 of four tree species with the highest relative value for bark stripping (Erythrophleum lasianthum, Balanites maughami, Albizia adianthifolia, & Newtonia hildebrandtii) and distance from subsistence agriculture.

54

4 4.1

Discussion Overview

Inventories and models of the spatial distribution of plant species resource use can provide important information for determining biodiversity loss and human impact that is not possible with remote sensing techniques. In addition, if the appropriate species are chosen, they can also provide recommendations for practical actions and effective land-use planning that can contribute to both rural livelihood improvements and biodiversity conservation. The sustainable use of plants species, because of their cultural and economic value can lay the foundations for the conservation of not only the vulnerable species of concern, but also of habitats and ecosystems more generally. The study objective was to collect quantitative data on resource use at the appropriate scales that can be of significance to resource users, local communities, resource managers and policy makers. A sequential approach was adopted that identified the geographical focus and the priority species based on awareness of wider conservation issues and concerns related to over harvesting. A significant contribution of the present study has been the development of a robust and practical inventory that utilises priority species based on their relative value from the perspective of users. This gave a description of how the resource base was being utilised and formed the basis for spatial mapping. An important output has been the development of a ‘fine-scale’ GIS map to model natural resource harvesting that can provide recommendations for practical actions and effective land use planning. 4.2

Harvesting and resource value to rural communities of Maputaland

Quantitative studies of tree resource harvesting can be used as an appropriate measure for inventories to assess resource use in rural communities. Rural communities value tree resources and their biology and geographical distribution make them vulnerable to harvesting impact. Short-term ‘snap-shot’ inventories can be employed, as harvesting impact can remain visible for long periods. This is an important consideration. Other plant resources, for example bulbs or whole plants are heavily harvested, but their growth form and method of harvest require long-term monitoring of populations to determine the impact of resource use (Raunkiaer 1934; Cunningham 2001b). An additional significant factor is that different tree resource uses and users can be readily identified from harvesting practices. Contrary to expectations and most other reports, there was little evidence of over-harvesting of the plant resource base either for subsistence use or for commercial purposes in the study areas. Trade in fuelwood and crafts currently provide only limited entrepreneurial and employment opportunities to a few homesteads. The majority of use was for subsistence purposes and provides an important resource base for rural communities.

55

Of the focal species 10% had evidence of stem cutting and a diverse range of harvesting activities was identified, including use of wood for general-purpose timber, poles for construction, fuelwood and harvesting of honey. Not unsurprisingly, resource use was species-specific. Importantly, however, from a conservation perspective, there was evidence that individuals of species that are popularly harvested and have the highest relative resource value are more likely to be destructively harvested. Harvesting intensity and the impact this has on individual trees is related to a number of factors. Important considerations are the end product requirement of the resource user and the biology and resilience of the target species. Some resource use has little impact whereas others are detrimental and destructive. For example, harvesting for poles has little impact on multi-stemmed trees such as Dalium schlechteri, but single-stemmed species such as Brachylaena huillensis and Ptaeroxylon obliquum are

likely to be destructively harvested. Stem cutting of other species (Newtonia hildebrandtii

and

Balanites maughami) for general-purpose timber and fuelwood uses was often more benign at the

individual plant level. Harvesting impact is also fundamentally related to plant resilience. Acacia burkei branches were heavily harvested but species readily resprouted, and resource use appeared to

have little impact. The final ecological consideration of harvesting relates to the impact at the population level and how this can contribute to landscape level effects. Harvesting impacts of different size-classes can provide useful insights. Species that have the highest relative value to resource users and are destructively harvested such as Brachylaena huillensis and Erythophleum lasianthum show size-class distribution curves that deviate from ‘model’ distribution curves. This may be indicative that harvesting is having detrimental effects at the population level. Such species would have a higher priority for conservation strategies and for the implementation of management programmes. However, determination of impacts at the population level require long-term studies of permanent plots or comparison with appropriate control groups that are not available for the species in this study. Fuelwood collection of live trees was not an apparent problem at the study sites. Fuelwood has been identified as being an important primary energy source for rural communities (Twine 2002; Williams & Shackleton 2002). Many rural communities face increasing shortages, related to intensity of use, largely a function of human population growth and clearance for agriculture (Banks, Griffin, Shackleton, Shackleton, & Mavrandonis 1996). The majority of fuelwood collection at the study areas was for subsistence use and a large number of species were regularly used confirming previous studies (Shackleton, Netshiluvhi, Shackleton, Geach, Ballance, & Fairbanks 1999). There was an absence of substantial commercial trade, which is apparently not uncommon in areas that are remote or where there are relatively abundant fuel stocks (Williams & Shackleton 2002).

56

However, a small-scale local trade in fuelwood of Newtonia hildebrandtii var hildebrandtii was identified at site 1. Trade was restricted to 19 outlets, located next to homesteads situated along a main road. Trade volume appeared to be low and trade in other species was not apparent. In addition, a small craft trade was also in evidence along the same road with 8-10 temporary stalls again located in close vicinity to homesteads. Two species of trees were identified as being used for the production of crafts, mainly trays, Cleistanthus schlechteri and Albizia forbesii. The trade in fuelwood and crafts currently represent small entrepreneurial and employment opportunities for a limited number of households, presumably reliant on the small number of passing tourists. The use of Newtonia hildebrandtii var hildebrandtii

and Cleistanthus schlechteri appeared sustainable, however,

confirmation of this requires long-term monitoring studies. Bark stripping of the 13 priority species in the two study areas was low, less than 3% in study area 1 and less than 0.5% in study area 2. This was despite the fact that 12 of the 13 species have been recorded as being used for medicinal purposes (Goldring 2002; Grace, Prendergast, Jager, & van Staden 2003; Mander 1998). Four species, (Erythropleum lasianthum (23%), Balanites maughami (11%), Albizia adianthifolia (8%) and Newtonia hildebrandtii (4%)), however, were more heavily harvested. These higher value species were more likely to be destructively bark stripped (Cunningham1990). Three of the species are regularly traded in the Durban markets (Grace, Prendergast, Jager, & van Staden 2003; Mander 1998). There was evidence of bark stripping of other species included in the inventory, but at much lower frequency and intensity. Five individuals of three other species (Terminalia sericea, Albizia versicolor and Trichilia dregeana), not included in the inventory were recorded as being bark stripped. The vulnerability of the tree species in the study to bark stripping is not known. Bark regeneration after removal is not a common response for most species, although some species may show complete regrowth after ring barking (Warburgia salutaris) or complete bark removal (Prunus africana). Experimental studies or long-term monitoring are required to determine the outcome of bark stripping resource-harvesting practices on the species studied (Geldenhuys 1999). Bark stripping appeared to be for subsistence by local herbalists and resource users. Again, contrary to expectations there was little evidence of trade and no identified market flows. The findings are in agreement with a recent report that found no signs of over exploitation of medicinal plants near the village of Mnqobokazi, Maputaland (Trygger 2003). Present harvesting was not considered to be indiscriminate, destructive and unsustainable as has been documented for other areas (Cooper 1979; Cunningham 1988; Dold & Cocks 2002; Geldenhuys 2004).

57

4.3

Resource use impact and recommendations

There may be several explanations as to why large-scale overexploitation or commercial harvesting of the natural resource base was not taking place. The lack of market flow may be related to distance to markets or to the fact that a trade network was not well developed in the area. Value is related to supply and demand and for many uses, substitute species or alternatives can be readily found. This was evident in the study areas for wood used for poles, general-purpose timber and fuelwood. For medicinal plants, although many species are commercially traded only a few are in high demand (Cunningham 1988; Williams, Balkwill, & Witkowski 2000). There appeared to be few opportunities for trade. There is little disposable income due to the high poverty levels. Perhaps of more significance is the fact that most of the rural population pursue subsistence livelihoods and are reliant to some extent on the natural resource base, which forms an important part of their culture. Harvesting practices appeared to rely on traditional knowledge of species biology that respected customary practices. Community leaders and local resource users implement traditional management practices and conservation measures and there appeared to be a wide social acceptance and compliance with regulations with recognition of ownership, access rights and responsibilities. It cannot be ruled out that over-harvesting in isolated patches might occur or that exploitation of high value species has resulted in their local extinction. This has in fact already been documented for one species (Warburgia salutaris) which is now only found in adjacent Protected Areas (Mander 1997). The present study cannot ascertain whether harvesting has resulted in changes in species population dynamics or landscape effects. However, some high value species were being more intensively and destructively harvested at the individual level. Long-term monitoring studies or experimental procedures are required to determine whether use of these species can be considered sustainable. However, the presence of high value species and factors that relate to an increasing human population, that include urbanisation, high rates of unemployment, influx of outsiders, increased road access and entrepreneurial opportunities could result in an increase in harvesting pressure. If associated with a loss of access rights and diminishing of regulations that enforce conservation practices this could result in over-exploitation and loss of the resource base. Increased demand for fuelwood and trade in crafts of canopy dominant keystone species such as Newtonia hildebrandtii var hildebrandtii and Cleistanthus schlechteri could have significant affects on the fragile endemic sand forest habitat.

Commercial gatherers are prepared to travel increasing distance from the major markets for gathering medicinal plant products and will often speculatively harvest species if there is a potential for trade (Cunningham 1988; Mander 1998). Increased harvesting pressure of slow growing, slow reproducing,

58

uncommon species with limited geographical distribution such as Erythropleum lasianthum, Albizia adianthifolia and Balanites maughami are likely to result in a rapid decline of vulnerable species and

loss of associated biodiversity. Changes in harvesting procedures of species like Acacia burkei could lead to unsustainable use and environmental degradation The study would recommend that while over-exploitation of the resource base is not apparent at the present time, the presence of high value species and the potential for socio-economic conditions to change rapidly requires the implementation of conservation strategies. Inventory of high value species should form an important component for management plans. There are important information gaps and uncertainties for most of the factors for determining sustainable use. However, the prioritisation of species and identifying the opportunities for sustainable harvest based on predictors of resilience or vulnerability can lay the foundation for the instigation of monitoring and adaptive management of priority species. The process should be participatory, building on traditional knowledge, local institutions and should be incentive-driven (IIED 1994). A recent example of such a development is the formation of the Sizamimphilo Association, which comprises a core group of bark harvesters that trade in the Durban market. They have implemented a sustainable harvesting management plan that contributes to the recovery of the forest and allows sustainable harvesting in the uMzimkulu District (Diederichs, Geldenhuys, & Mitchell 2002).

4.4

Modelling tree resource use in the Maputaland Centre of Endemism

Strategic planning that identifies conservation priorities is an important component of efficient and effective conservation planning. Remote sensing of habitats and landscape level features that reflect the distribution of other biodiversity can be used in systematic planning. It cannot however, identify over harvesting of high value species or adequately identify where sustainable use strategies can effectively contribute to conservation strategies (Smith 2001). A comprehensive picture is provided by combining large spatial scale analysis with the monitoring of species at the individual and population level. The present study has used the inventory data collected to develop a ‘fine scale’ GIS map that models tree resource harvesting in Maputaland Centre of Endemism. Logistic regression modelling identified that a number of factors were important in predicting resource use. The model performance from the ROC value was 0.734 for stem cutting indicating a good model fit and 0.896 for bark cutting indicating a highly accurate model. For both bark stripping (P<0.001) and stem cutting (P<0.001), not unsurprisingly, tree species was an important factor reflecting speciesspecific selection by resource users. For bark stripping, DBH was an important factor (P<0.001), with larger size-classes of trees more likely to be stripped.

59

Distance from subsistence agriculture was an important predictor for both stem cutting (P<0.001) and bark stripping (P<0.001). This acts as a surrogate for human density and occurs at the interface between transformed and pristine habitat. The majority of the population at the two study sites resides in households that are situated on transformed subsistence land. Most households typically have small dry land agricultural plots surrounding their homesteads and pursue subsistence livelihoods reliant on the surrounding natural resource base. In addition, slope (P=0.006) and elevation (P<0.001) was found to be an important variable for stem cutting. Harvesting of resources, particularly fuelwood and timber for construction purposes is difficult in areas that are elevated and have steep slopes. These areas will be avoided when more suitable sites are available. The distance from an access road was found not to contribute significantly to the model for predicting resource use. Distance from road was probably not significant as most of Maputaland is made accessible by a network of unplanned roads (Smith 2001). The maps provide an important tool for the visual representation of information on resource use. They can inform of areas where resource harvesting is likely to be important, where there is the potential for over harvesting and where sustainable use strategies can contribute to conservation and improve livelihoods. Harvesting frequency and intensity, however, are related to many factors that have not been included in the model. These include ecological and biological considerations such as vegetation habitat type and species composition. It is likely that socio-economic consideration and not populations per se that are important in determining site-specific harvesting practices (Woodroffe 2000). This could include factors such as land-ownership and access, land use and socio-economic factors such as poverty and employment levels. It is difficult to quantitate some of these factors for modelling requirements. It may not be necessary to do so, as much of the communal land in Maputaland is homogeneous. It would be of value to identify those communal areas where resource access rights, responsibilities and regulations have been eroded. These areas are more likely to be over-harvested. It is also important to perform additional inventories, a larger database can be used to develop more robust models. However, information gaps and biological uncertainties require the adoption of adaptive management practices that monitor and determine the extent and impact of harvesting that are site-specific. Identification of factors and relating them to harvesting practices provides information that can be of value to resource users. In the present study the amount of stem cutting and bark stripping resource use and how this changes with distance from subsistence agriculture was determined. 18% of trees were stem cut and 23% (of the high value species) were bark stripped in the vicinity of subsistence agriculture. This was reduced to less than 5% of trees at distances greater than 1000m. This observation is important, as the immediate impression is that close to subsistence, unsustainable

60

resource use is occurring. Indeed, most access to communal land requires transit across subsistence agriculture, magnifying this impression. It is important that consideration of the harvest impact should be viewed at the population and landscape level in a quantitative manner using management plans. Management requires that targets are set and management actions can be measured. Modelling risk factors and analysis of quantitative data can be used to evaluate the success of policies and to provide recommendations for adaptive management and future actions. The study has developed a predictive model that can identify areas where resource harvesting is likely to occur. It has incorporated robust methodologies that prioritise species from the perspective of resource users. These outputs are of value to policy makers, resource managers, local communities and resource users alike. Inventories of species and models of resource use can be used to determine the extent of harvesting impact on the resource base that can be used for assessment of biodiversity loss. More importantly, modelling of appropriate priority species can provide recommendations for practical actions and effective land-use planning that can contribute to both rural livelihood improvements and conservation of biodiversity. The significance, both culturally and economically of some plant species can be sufficiently great that sustainable use strategies can lay the foundation for conservation of not only vulnerable species, but of habitats and ecosystems more generally.

61

5

Conclusions

Inventories and models that map plant resource use can provide important information for determining biodiversity loss and human impact that is not possible with remote sensing techniques. In addition, if the appropriate species are chosen, they can also provide recommendations for practical actions and effective land-use planning that can contribute to both rural livelihood improvements and biodiversity conservation. A significant contribution of the present study has been the development of a robust and practical inventory that utilises priority species based on their relative value from the perspective of users. This gave a description of how the resource base was being utilised and formed the basis for spatial mapping. An important output has been the development of a ‘fine-scale’ GIS map to model natural resource harvesting that can provide recommendations for practical actions and effective land use planning. The inventory found that contrary to expectations there was little evidence of over harvesting of the resource base either for subsistence or commercial purposes. Trade in fuelwood and crafts currently provides only limited entrepreneurial and employment opportunities. The majority of use, including medicinal plant harvesting, was for subsistence purposes and provided an important resource base for rural communities. The current socio-economic conditions and the wide social acceptance and compliance with regulations that recognise ownership, access rights and responsibilities are the probable explanations for the reasons why large-scale exploitation is not presently occurring in the study areas inventoried. Distance from subsistence agriculture and other land scape features were identified as being important predictors of resource use and were used for the production of ‘fine scale’ GIS maps. These can be of value in providing recommendations for practical actions and effective land use planning for relevant stakeholders at the appropriate scale. The study would recommend that where high value plant species occur, especially in vulnerable habitats and in situations where socio-economic conditions can change rapidly, that inventories and modelling, incorporating methodologies used in the present study, should be implemented. The process should be participatory, building on traditional knowledge, local institutions and should be incentive-driven. These can lead to the establishment of conservation strategies of not only vulnerable plant species but of habitats and ecosystems more generally.

62

6

References

Aumeeruddy-Thomas, Y. A., Saigal, S., Kapoor, N. & Cunningham, A. B. 1999. Joint management in the making - reflections and experiences. People and Plants Working Paper no.7. Division of Ecological Sciences, United Nations Educational, Scientific and Cultural Organisation, Paris, France. Baker & Mutitjulu Community. 1992. Comparing two views of the landscape: aboriginal ecological knowledge and modern scientific knowledge. Rangeland Journal 14: 174-189. Banks, D. I., Griffin, N., Shackleton, C. M., Shackleton, S. E. & Mavrandonis, J. 1996. Wood supply and demand around two rural settlements in a semi-arid savannah, South Africa. Biomass and Bioenergy 11: 319-331. Bell, R. H. V. & McShane, T. O. 1984. Landscape classification. In Conservation and Wildlife Management in Africa. US Peace Corps, Washington, D.C., pp 95-106. Central Statistical Services. 1996. Central Statistics. Republic of South Africa. Chao, N. 2004. Local Community perceptions to the establishment of a community conservation area in Usuthu Gorge, South Africa. MSc. Dissertation, Durrell Institute of Conservation & Ecology, University of Kent at Canterbury. Chapman, M. 1987. Traditional political structure and conservation in Oceania. Ambio 16: 201-205. Cooper, K. H. 1979. An investigation of the herbal medicine trade in Natal/KwaZulu. University of Natal, Institute of Natural Resources, (INR Investigational Report 29), Pietersmaritzburg. Cuaron, A. D. 2000. A global perspective on habitat disturbance and tropical rainforests mammals. Conservation Biology 14: 1574-1579. Cunningham, A. B. 1985. The resource value of indigenous plants to rural people in a low agricultural potential area. PhD, University of Cape Town, Cape Town. Cunningham, A. B. 1988. An Investigation of the Herbal Medicine Trade in Natal/KwaZulu. Institute of Natural Resources, University of Natal, Pietersmaritzburg, South Africa. Investigational Report No 29. Cunningham, A. B. 1990. African medicinal plants: setting priorities at the interface between conservation and primary health care. Report for the WWF Project 3331. Institute of Natural Resources. Pietersmaritzburg, South Africa. Cunningham, A. B. 1991. Development of a conservation policy on commercially exploited medicinal plants: a case study from southern Africa. In Conservation of Medicinal Plants. Cambridge University Press, Cambridge, pp. 337-358. Cunningham, A. B. 2001a. Measuring Individual Plants and Assessing Harvesting Impacts. In Applied Ethnobotany: People, Wild Plant Use & Conservation. Earthscan Publications Ltd, London and Sterling, VA, pp. 96-143. Cunningham, A. B. 2001b. Settlement, commercialisation and change. In Applied Ethnobotany: people, wild plant use & conservation. Earthscan Publications Ltd, London and Sterling, VA, pp. 6095.

63

DEAT 1997. White Paper on the Conservation and Sustainable Use of Biological Diversity. Department of Environment Affairs and Tourism, Pretoria. Diederichs, N., Geldenhuys, C. J. & Mitchell, D. 2002. The first legal harvesters of protected medicinal plants in South Africa. Science in Africa. DME 1998. White Paper on the Energy Policy of the Republic of South Africa. Department of Minerals and Energy (DME), Pretoria. Dold, A. P. & Cocks, M. L. 2002. The trade in medicinal plants in the Eastern Cape Province, South Africa. South African Journal of Science 98: 589-597. DWAF 1996. Sustainable Forest Development in South Africa: the Policy of the Government of National Unity. Department of Water Affairs and Forestry, Pretoria. DWAF 1997. South Africa's National Forestry Action Programme. Department of Water Affairs and Forestry, Pretoria. Farnsworth, N. R. & Soejarto, D. D. 1991. Global importance of medicinal plants. In The Conservation of medicinal plants. Cambridge University Press, Cambridge, pp. 25-51. Gandar, M. V. 1984. Wood as a source of fuel in South Africa. Southern African Forestry Journal 1: 9. Geldenhuys, C. J. 1992. The use of diameter distributions in sustained use management of forests: examples from Southern Africa. Paper presented at the SAREC/Zimbabwe Forestry Commission Symposium: Ecology and Management of Indigenous Forests in Southern Africa. Geldenhuys, C. J. 2004. Bark harvesting for traditional medicine: from illegal resource degradation to participatory management. Scandinavian Journal of Forest Science 22: 1. Goldring, J. 2002. Workshop Proceedings: Revision of the National List of Protected Trees as per Section 12, National Forests Act of 1998. Department of Water Affairs & Forestry, Directorate: Forest Regulation (Scientific Services), Pretoria. Goodman, P. S. 2002. A rapid assessment of terrestrial protected area management effectiveness in KwaZulu-Natal, South Africa, KwaZulu-Natal Wildlife. Grace, O. M., Prendergast, H. D. V., Jager, A. K. & van Staden, J. 2003. Bark medicines used in traditional healthcare in KwaZulu-Natal, South Africa: An inventory. South African Journal of Botany 69: 301-363. Harper, J. L. 1977. Population Biology of Plants Academic Press, New York. Hilton-Taylor, C. 1996. Red Data List of southern African plants. Pretoria, National Botanical Institute. Stretlitzia 4: 117. Hutchings, A. 1996. Zulu medicinal plants: an inventory. University of Natal Press, Pietersmaritzburg. IIED 1994. Whose Eden? An overview of community approaches to wildlife management. International Institute for Environment and Development, London. Jones, J. L. 2004. Transboundary Conservation in Southern Africa: Exploring conflict between local resource access and conservation. Centre for Environmental Studies, University of Pretoria, Pretoria.

64

Leopold, A. 1949. A Sand County Almanac and Sketches Here and There. Oxford University Press, New York. Low, A. B. 1996. Vegetation of South Africa, Lesotho and Swaziland. DEAT, Pretoria. MacNally, R. & Quinn, G. P. 1998. Symposium introduction: the importance of scale in ecology. Australian Journal of Ecology 23: 1-7. Mander, J., Quinn, N. & Mander, M. 1997. Trade in wildlife medicinals in South Africa. Institute of Natural Resources Investigational Report 154. Institute of Natural Resources, Pietersmaritzburg. Mander, M. 1998. Marketing of medicinal plants in South Africa: A case study in KwaZulu-Natal. Food and Agricultural Organisation of the United Nations, Rome, Italy. Margules, C. R. & Pressey, R. L. 2000. Systematic conservation planning. Nature 405: 243-253. Marsh, A. C. & Seely, M. K. 1992. Oshanas: sustaining people, environment and development in Central Owambo, Namibia. Typoprint, Windhoek, Namibia. Marshall, N. T. 1998. Searching for a cure: Conservation of medicinal wildlife resources in East and Southern Africa. TRAFFIC International. Matthews, W. S., van Wyk, A. E., van Rooyen, N. & Botha, G. A. 2001. Vegetation of the Tembe Elephant Park, Maputaland, South Africa. South African Journal of Botany 67: 573-594. Mattos, M., Nepstad, D. C. & Viera, I. C. M. 1992. Cartilha sobremapeamento de area, cubagem de maderia e inventario florestal. Woods Hole Research Centre, Belem, Brazil. Molur, S. & Walker, S. 1996. Conservation Assessment and Management Plan (CAMP II) for Selected Species of Medicinal Plants of Southern India. Coimbatore, India. Muir, J. 1916. A thousand mile walk to the gulf. Houghton Mifflin, Boston, MA. Noss, R. F. 1990. Indicators for monitoring biodiversity: a hierarchical approach. Conservation Biology 4: 355-364. Olson, D. M. & Dinerstein, E. 1998. The Global 200: a representation approach to conserving the earth's most biologically valuable ecoregions. Conservation Biology 12: 502-515. Pearce, J. & Ferrier, S. 2000. Evaluating the predictive performance of habitat models developed using logistic regression. Ecological Modelling 133: 225-245. Penso, G. 1980. The role of WHO in the selection and characterisation of medicinal plants (vegetable drugs). Journal of Ethnopharmacology 2 183-188. Peters, C. M. 1994. Sustainable Harvest of Non--timber Forest Plant Resources in Tropical Moist forest: an ecological primer. Biodiversity Support Programme, Washington, DC. Picton, M. A. & Putz, F. E. 1989. African Textiles British Museum Publications, London. Pimm, S.L., Jones, H.L., & Diamond, J. 1988. On the risk of extinction. American Naturalist. 132: 757-785. Pitman, N. C. A., Terborgh, J., Silman, M. R. & Nunez, P. 1999. Tree species distributions in an upper Amazonian forest. Ecology 80: 2651-2661.

65

Plotkin, M. & Famolare, L. 1992. Sustainable Harvest and Marketing of rain forest Products. Island Press, Washington, DC. Poffenberger, M. B., McGean, M. B., Ravindranath, N. H. & Gadgil, M. 1992. Field Methods Manual, volume 1: diagnostic tools for supporting joint forest management systems. Joint Forest Management Support Programme, Society for Promotion of Wasteland Development, New Delhi, India. Pooley, E. 1993. The complete field guide to trees of Natal, Zululand and Tanskei. Natal Floral Publications Trust, Cape Town. Pressey, R. L., Humphries, C. J., Margules, C. R., Vanewright, R. I. & Williams, P. H. 1993. Beyond opportunism - Key principles for systematic reserve selection. Trends in Ecology & Evolution 8: 124128. Rabinowitz, D. S., Cairns, S. & Dillon, T. 1986. Seven forms of rarity and their frequency in the flora of the British Isles. In Conservation Biology: the science of scarcity and diversity. Sinauer Associates, Sunderland, Massachusetts, pp. 182-204. Raunkiaer, C. 1934. Life forms of Plants and Statistical Plant Geography Clarendon Press, Oxford. Rundstrom, R. A. 1990. A cultural interpretation of Inuit map accuracy. The Geographical Review 80: 155-168. Schopp-Guth, A., & Fremuth, W. 2001. Sustainable use of medicinal plants and nature conservation in the Prespa National Park area, Albania. Medicinal Plant Conservation 7: 5-8. Scott, P. J. 1995. Progress report on multiple-use management to Uganda national Parks. Fort Portal, Uganda, WWF Rwenzori Mountains Conservation and Development Project. Shackleton, C. M. 2001. Re-examining local and market orientated use of wild plant species for the conservation of biodiversity. Environmental Conservation 28: 270-278. Shackleton, C. M., Netshiluvhi, T. R., Shackleton, S. E., Geach, B. S., Ballance, A. & Fairbanks, D. F. K. 1999. Direct use values of woodland resources from three rural villages. Rep. No. ENV-P-198120, CSIR, Pretoria. Shackleton, C. M. & Shackleton, S. E. 2000. Direct use values of secondary resources harvested from communal savannahs in the Bushbuckridge Lowveld, South Africa. Journal of Tropical Forest Products 26: 28-47. Smith, R. J. 2001. Designing an integrated protected area network for Maputand, South Africa, PhD, Durrell Institute of Conservation & Ecology, University of Kent at Canterbury. Smith, R. J., Goodman, P. S., Matthews, W. S. & Leader-Williams, N. 2004. Systematic conservation land-use planning: a review of perceived problems and actual benefits, illustrated with a case study from Maputaland, South Africa. Oryx (in press). Stattersfield, J. M., Crosby, M. J., Long, A. J. & et al. 1998. Endemic bird areas of the world: priorities for biodiversity conservation. Conservation Series 7. Birdlife International. Stevens, S. 1997. The Legacy of Yellowstone. In Conservation Through Cultural Survival: Indigenous Peoples and Protected Areas, Island Press, Washington.

66

Ticktin, T. 2004. The ecological implication of harvesting non-timber forest products. Ecology 41: 1121. Tietema, T. 1993. Biomass determination of fuelwood trees and bushes of Botswana, Southern Africa. Forest Ecology and Management 60: 257-269. Trygger, S. 2003. Local extraction and use of medicinal plants - a case study from Mkuze Wetland, South Africa. Minor Field Studies No.257, Swedish University of Agricultural Sciences, SLU Communications, Uppsala. Tuxill, H. & Nabhan, G. P. 2001. People, plants and protected areas. Earthscan, London. Twine, W. 2002. The fuelwood problem: are we reaching a crisis point? Environment Today 2: 20-21. van Rensburg, B. J., Chown, S. L., van Jaarsveld, A. S. & McGeoch, M. A. 2000. Spatial variation and biogeography of sand forest avian assemblages in South Africa. Journal of Biogeography 27: 13851401. van Rensburg, B. J., McGeoch, M. A., Chown, S. L. & van Jaarsveld, A. S. 1999. Conservation of heterogeneity among dung beetles in the Maputand Centre of Endemism, South Africa. Biological Conservation 88: 145-153. van Wyk, A. E. 1994. Maputaland Pondoland region. In Centres of plant diversity: a guide and strategy for their conservation. Oxford University Press, Oxford, pp. 227-235. van Wyk, G. F., Everard.D, A., Midgley, J. J., & Gordon, I. G. 1996. Classification and dynamics of a southern African subtropical coastal lowland forest. South African Journal of Botany 62: 133-142. van Wyk, B. & van Wyk,P. 1997. Field Guide to Trees of Southern Africa. Struik Publishers, Cape Town. Von Maltitz, G. P. & Scholes, R. J. 1995. The burning of fuelwood in South Africa: when is it sustainable. Environmental Monitoring and Assessment 38: 243-251. Walsh, F. J. 1990. An ecological study of traditional use of "country": Martu in the Great and Little Sandy deserts, Western Australia'. Proceedings of the Ecological Society of Australia 16: 23-37. Walsh, F. J. 1993. The relevance of some aspects of Aboriginal subsistence to the management of national parks; with reference to Martu people of the Western Desert. In Aboriginal involvement in parks and protected areas. Australian Institute of Aboriginal and Torres Strait islander Studies, Canberra, pp. 75-97. Williams, A. & Shackleton, C. M. 2002. Fuelwood use in South Africa: Where to in the 21st Century? Southern African Forestry Journal 196: 1-7. Williams, V. L., Balkwill, K. & Witkowski, E. T. F. 2000. Unravelling the commercial market for medicinal plants and plant parts on the Witwatersrand, South Africa. Economic Botany 54: 310-327. Woodroffe, R. 2000. Predators and people: using human densities to interpret declines of large carnivores. Animal Conservation. 3: 165-173. World Conservation Union and WWF-UK. 1994. Centres of plant diversity: a guide and strategy for their conservation.

67

Appendices Appendix 1

Priority tree species used in the inventories Species

Common English name

Albizia adianthifolia Balanites maughami Brachylaenia huillensis Cleistanthus schlechteri Dalium schlechteri Erythrophleum lasianthum Newtonia hildebrandtii Ptaeroxylon obliquum Acacia burkei Combretum zeyheri Combretum molle Philonoptera violacea Sclerocarya birrea subsp. caffra

Flat Crown Torchwood Silver Oak False Tamboti Zulu Podberry Swazi Ordeal Lebombo Wattle Sneezewood Black Monkey Thorn Large Fruited Bushwillow Velvet Bushwillow Apple Leaf Marula

68

Appendix 2 Size-Class Distributions of Tree Species Inventoried

Three ideal types of size-class distribution are recognised for trees in uneven-aged mixed species forests. The reverse J-shaped (negative exponential) eg. Albizia adianthifolia and Dalium schlechteri. The Uni-modal (flat curve) eg Newtonia hildebrandtii and Acacia burkei and the Bell-shaped curve is the third type eg Combretum zeyheri. Species with the highest relative value to resource user and are destructively harvested such as Brachylaenia huillensis and Erythrophleum lasianthum show size class distribution curves that deviate from ‘model’ distribution curves. This may be indicative that harvesting is having detrimental affects at the population level. Such species would have a higher conservation priority. The graphs show the size-class distribution of the species included in the inventories, determined from the DBH. The Black Bars indicate the number of trees that have had human resource impact

Acacia burkei 50 45 40

30 25 20 15 10

69

>400

301-400

201-300

151-200

126-150

DBH (cm)

101-125

81-90

71-80

61-70

51-60

41-50

31-40

21-30

0

91-100

5 10-20

No. of trees

35

70

2

0

DBH (cm) >400

4

>400

6 301-400

8

301-400

10 201-300

12

201-300

14 151-200

16

151-200

18 126-150

Philonoptera violacea

126-150

DBH (cm) 101-125

91-100

81-90

71-80

61-70

51-60

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

>400

301-400

201-300

151-200

126-150

101-125

DBH (cm)

101-125

91-100

81-90

71-80

61-70

51-60

41-50

80

41-50

31-40

21-30

10-20

No. of trees

No. of trees 14

31-40

21-30

10-20

No. of trees

. Erythrophleum lasianthum

12

10 8

6

4

2

0

Brachylaenia huillensis

70

60

50

40

30

20

10

0

71

DBH (cm) >400

0

>400

5 301-400

10

301-400

15 201-300

20

201-300

25 151-200

30

151-200

35 126-150

Newtonia hildebrandtii

126-150

DBH (cm) 101-125

91-100

81-90

71-80

61-70

51-60

41-50

31-40

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

>400

301-400

201-300

151-200

126-150

101-125

DBH (cm)

101-125

91-100

81-90

71-80

61-70

51-60

41-50

31-40

40 21-30

10-20

No. of trees 40

21-30

10-20

No. of trees

No. of trees 140

Cleistanthus schlechteri

120

100 80

60

40

20

0

Albizia adianthifolia

35

30

25

20

15

10

5

0

72

10

5

0

DBH (cm) >400

15

>400

20 301-400

25

301-400

30 201-300

35

201-300

40 151-200

45

151-200

Balanites maughami 126-150

50

126-150

DBH (cm) 101-125

91-100

81-90

71-80

61-70

51-60

41-50

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

>400

301-400

201-300

151-200

126-150

101-125

DBH (cm)

101-125

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

No. of trees 60

31-40

21-30

10-20

No. of trees

No. of trees 140

120

Dalium schlechteri

100

80

60

40

20

0

Ptaeroxylon obliquum

50

40

30

20

10

0

73 301-400 >400

>400

DBH (cm)

301-400

0 201-300

5

201-300

10 151-200

15

151-200

20 126-150

Combretum molle

126-150

DBH (cm) 101-125

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

No. of trees

91-100

81-90

71-80

61-70

51-60

41-50

31-40

21-30

10-20

>400

301-400

201-300

151-200

126-150

101-125

DBH (cm)

101-125

91-100

81-90

71-80

61-70

51-60

25

41-50

31-40

21-30

10-20

No. of trees

No. of trees 300

Combretum zeyheri

250

200

150

100

50

0

25

Sclerocarya birrea subsp. caffra

20

15

10

5

0

74

75