3. METHOD OF DERIVING LAND CLASS AND LAND TRANSFORMATION DATA LAYERS 2
4. Description of Land Classes 5
5. DESCRIPTION OF LAND TRANSFORMATION CLASSES 25
6. References 30
ANNEX A: ENVIRONMENTAL DATA FOR LAND CLASSES WITHIN THE GAENP
PLANNING DOMAIN 31
Annex B: Geology and soil data for land classes within the GAENP planning domain 35
Systematic and strategic conservation planning relies on the development of appropriate spatial biodiversity surrogates, and the quantification of the extent to which these surrogates have been transformed from their original, pre-European form. For the expansion of the Greater Addo Elephant National Park (GANP), land classes1 were chosen as the surrogate to represent biodiversity pattern. Dividing the landscape into land classes, which depict homogenous parcels of vegetation, geology and climate, is a widely accepted method of depicting biodiversity pattern, especially in the absence of species data sets that provide reliable presence/absence data.
Existing data layers of pre-European land classes (e.g. national vegetation types developed by Low and Rebelo, 1996) and the extent of current land transformation (e.g. national land cover developed by Thompson, 1997) were at a scale too coarse for the conservation planning purposes of this project. Consequently, new data layers at the appropriate scale needed to be developed through mapping in the field.
This report outlines the Terms of Reference for field mapping, the field approach to mapping the land classes and land transformation within the GANP planning domain. It then briefly describes the GIS procedures used to develop the digital data layers. Lastly, the report outlines in detail the categories of land classes and land transformation distinguished for GANP.
2.TERMS OF REFERENCE
Establish a classification system at the scale of 1:50 000 of the following:
Pre-European extent for primary and secondary BHUs in the GAENP planning domain, that is compatible to the classification system currently being undertaken for the STEP Project.
Land transformation categories in the GAENP planning domain, suitable for conservation planning.
Alien invasive plants
Determine and map at a scale of 1:50 000, through field work:
The present extent of each BHU
The extent of the land transformation categories
Alien invasive plants
Produce a report on BHUs, Land Transformation and Alien Invasive Plants within the planning domain. The report should outline the following:
The diagnostic descriptions of the characteristic flora within each BHU, and the assumptions applied during field mapping of these vegetation types;
A list of main ecological factors which determine the occurrence and functioning of the key ecological factors (soil type, climate etc) and processes (e.g. disturbance regime) which sustain biodiversity, for the vegetation component of each of the primary BHUs that occur in the planning domain;
A description of each land transformation category and sub-category, and the assumptions applied during field mapping of these vegetation types.
This report provides the description for the classification system for the BHUs, Land Transformation and Alien Vegetation classes. It is intended as a reference document for the maps.
3.METHOD OF DERIVING LAND CLASS AND LAND TRANSFORMATION DATA LAYERS
Field mapping of land classes and land transformation was conducted over a three-month period, between April and July 2001. Given the time limitations of the project, the size of the GANP planning domain and the complexity of the vegetation, a scale of 1:50 000 was deemed to be the most appropriate scale of mapping. This is also the scale at which land ownership boundaries are mapped, which serve as the units of selection in the conservation planning exercise, and the units through which land is purchased and the existing Addo National Park is expanded.
A system for defining land classes was devised, based largely on the classification systems developed for the Sub-tropical Thicket Ecosystem Project (STEP), a parallel conservation planning initiative for the Thicket Biome in the Eastern Cape (see the STEP project’s classification system by Jan Vlok, housed at TERU, UPE). The level of transformation for each of the land classes was also mapped, described under two broad themes - urbanisation/agriculture and alien plant infestation. Descriptions for each category of land class and land transformation are detailed in Section Error: Reference source not found.
Hard copy 1:50 000 maps of geo-referenced LandSat TM images (using the spectral band combination Red-Green-Blue), were superimposed with towns, rivers, roads, and farm boundaries. These were printed out and laminated for use in the field. The GANP planning domain covers sixteen 1:50 000 map sheets, as outlined in Table 1.
Table 1. 1:50 000 map sheets for the GANP planning domain
Map Sheet No.
Map Sheet Name
Map Sheet No.
Map Sheet Name
The following steps describe the field approach to mapping land classes and land transformation:
A fixed-wing flight over the study area was conducted initially to serve as an orientation trip. Photograph stills were taken and applied, with limited success, later during the field mapping.
For the area covered by the Thicket Biome within the GANP domain, the STEP project had already undertaken land classification at a 1:100 000 scale. These land class delineations were transferred to the GANP 1:50 000 laminated LandSat TM images to serve as a rough guide for refinement in the field.
With the aid of LandSat TM images of the area, as well as maps of geology, relief and climate, expected pre-European land classes were delineated on the GANP 1:50 000 laminated LandSat TM images.
Land class delineations in Steps 1 and 2 above were tested in the field and refined where necessary. This was undertaken by driving along roads and accessing vantage points, where possible. At the same time, land transformation was mapped. In areas where land was completely transformed, land classes were surmised on the basis of surrounding vegetation, combined with geological and climatic data.
In inaccessible areas (e.g. parts of the Zuurberg and the Alexandria coastal stretches), alternative forms of field verification were sought. The Zuurberg area was flown by fixed-wing aeroplane and photograph stills were taken. Although these proved useful for land classification and, at times, for mapping land transformation, they were ineffective for detecting extent of alien vegetation. Aerial video footage of the Zuurberg and the Alexandria coastal strip were also obtained from the Working for WaterProgramme, in an attempt to survey the alien vegetation in these inaccessible areas. Generally these proved effective, although it was noted that for areas along the Wit Rivier, north east of Kirkwood, mapping of alien vegetation densities were calculated estimates.
Care was taken to ensure that a unique polygon identifier was used for each polygon mapped onto the GANP 1:50 000 laminated LandSat TM images. This polygon identifier was prefixed by the 1:50 000 map sheet number (e.g. 3324BB_1 would indicate the first polygon mapped on the Greystone 1:50 000 map sheet). Field notes describing the land class and land transformation of each polygon were cross-referenced to a unique polygon identifier.
Refined 1:50 000 field delineations of land classes and land transformation were then transferred to acetate (transparency) for capture in the GIS lab. For each map sheet, at least four latitude and longitude coordinates were transferred, to serve as reference tick marks for GIS data capture. Care was taken during transfer to ensure that all mapped polygons were closed and had a unique polygon identifier, which cross-referenced it to field note descriptions of the land class and land transformation. Care was also taken to ensure that polygons spanning map sheets were “edge-matched”.
Mapping accuracy of land classes and land transformation in the field was estimated to be between 50-100 m, except in instances where land transformation followed a fence-line contrast, in which case the farm boundary was used (with an accuracy of approximately 15 m).