ORIGINAL RESEARCH

 

Revision of the North West province, South Africa, vegetation map

 

 

Philip G. Desmet^I; Greer Hawley^II; Anisha Dayaram^{III, IV}; R. John Power^V; Reuben Heydenrych^VI; Catherine M. Dzerefos^VII; Ray Schaller^V; Norbert Hah^VIII; Nancy Job^IX

^IDepartment of Zoology, Nelson Mandela University, P.O. Box 77000, Gqeberha 6031, South Africa
^IIRhodes University, Department of Biochemistry and Microbiology, Biological Sciences Building, Lower University Road, Makhanda 6140, South Africa
^IIIBiodiversity Assessment and Monitoring Division, South African National Biodiversity Institute, Private Bag X7, Claremont, Cape Town 7735, South Africa
^IVRestoration and Conservation Biology Research Group, Centre for African Ecology, School of Animal, Plant & Environmental Sciences, University of the Witwatersrand, 1 Jan Smuts Ave., Braamfontein, Johannesburg, South Africa
^VChief Directorate of Environmental Services, Department of Economic Development, Environment, Conservation & Tourism (DEDECT), North West Provincial Government, Private Bag X 2039, Mahikeng 2735, South Africa
^VI313 Jeremy St, Lynnwood Park, Pretoria 0081, South Africa
^VIIDepartment of Environmental, Water and Earth Sciences, Tshwane University of Technology, Private Bag X680, Pretoria, South Africa
^VIIIDepartment of Biological Sciences, Faculty of Science, Engineering and Agriculture, University of Venda, Private Bag X5050, Thohoyandou 0950, South Africa
^IXFreshwater Biodiversity Programme, South African National Biodiversity Institute, Private Bag X7, Claremont, Cape Town 7735, South Africa

Correspondence

 

 

---

ABSTRACT

BACKGROUND: The vegetation type boundaries in the North West province as they appear in the 2018 National Vegetation Map, for the most part, are derived from agricultural land types that were mapped in the 1980s
Aim & objectivesGiven (1) the importance the National Vegetation Map plays in conservation assessment and planning, as well as environmental planning and decision making; and (2) the map boundary errors reported by users, an update of the provincial vegetation map was considered necessary.
METHODS: A vegetation identification key using high-level environmental parameters (in order of importance: flooding, bioregion, terrain, geology and soil) was developed. This key was used to manually interpret high-resolution colour aerial imagery, together with existing environmental spatial datasets (land types as a proxy for soils, simplified geology and terrain/land form). The existing vegetation type concepts are sound and are mostly retained in this map
RESULTS & CONCLUSION: Changes to the map include: (1) all vegetation boundaries in the province are remapped; (2) Olea Sclerophyllous Forest is proposed as a sub type/community related to the Northern Afrotemperate Forest vegetation type; (3) two existing vegetation types currently not mapped as occurring in the province are brought into the province, namely, Subtropical Alluvial Vegetation and Wa-terberg Mountain Bushveld; and (4) three vegetation units recognised in previous vegetation studies and which are not indicated in the current National Vegetation Map are included here as new vegetation types, namely, Vryburg Thornveld, Morokweng Thornveld and Central Sandy Mountain Bush-veld. The descriptions of all terrestrial vegetation types occurring in the province are also updated and an updated annotated global plant species list for the province is provided. Changes reflected in this vegetation map have been incorporated into the National Vegetation Map Version 2024 beta.

Keywords: North West, vegetation map, classification, ecosystem type, revision.

---

 

 

Introduction

The South African vegetation map is a national scale map of the terrestrial ecosystems found within the country. The current National Vegetation Map, first published in 2006 (Mucina & Rutherford 2006) and updated in 2018 (Dayaram et al. 2019), includes 459 unique vegetation types of which 35 terrestrial and six azonal types occur in the North West province (NW). A beta release of the next version was released in 2024 (SANBI 2006-2024), which now incorporates revisions from work in this paper.

The purpose of the National Vegetation Map is to: a) provide a coarse-filter spatial surrogate that broadly represents biodiversity patterns across the whole country; b) provide ecologically relevant environmental management units; and c) provide ecologically meaningful units that can be used in environmental planning and management (Dayaram et al. 2021). Consequently, it is generally regarded as the national map of terrestrial ecosystems for the country. While itself not formally mandated by law, in practice the National Vegetation Map is regarded as one of South Africa's foundational biodiversity datasets that has an important legislative function, as it informs a number of government environmental and biodiversity planning and management tools, such as maps of Critical Biodiversity Areas and Ecological Support Areas; protected area expansion strategies; and forms a basis for environmental impact assessment. Thus, poorly delineated vegetation types can lead to poor outcomes for conservation planning, land management and planning decisions, and ultimately the loss of biodiversity.

Each vegetation type delineates and describes the parts of the landscape that share similar plant communities that are influenced or determined by shared environmental drivers (Mucina & Rutherford 2006). These maps are essentially models of the natural variation observable in any landscape. They reduce the complexity and continuity of natural landscapes to a set of discrete categories. Irrespective of the methods used to classify landscapes there are invariably mapping errors, especially where the transition between ecosystems is a continuum rather than being marked by a clear boundary and where landscapes have been significantly modified. The National Vegetation Map aims to map the original or historical extent of ecosystems before contemporary settlements, croplands and mining modified landscapes. This is defined as the ecosystems present prior to the advent of permanent European settlement in South Africa circa 350 years ago (Mucina & Rutherford 2006). This is a pragmatic, albeit problematic, threshold as it does not consider the impact of pre-colonial populations on vegetation as significant, whereas it is highly likely that these populations did have extensive and significant impacts on ecosystems (e.g., Sadr 2022). As such, ecosystem classification and mapping can be particularly problematic in highly modified landscapes. In these instances, understanding the key environmental determinants of ecosystems is very important for mapping the original extent of vegetation.

Given the central role that the South African vegetation map plays in land and environmental management and biodiversity conservation, there is an imperative to maintain and update this map to reflect the best available data and emerging knowledge of historical vegetation. In the NW, concerns were raised during the preparation of the North West Biodiversity Sector Plan (NW READ 2015) that the current provincial vegetation map, which is based on the National Vegetation Map published in Mucina and Rutherford (2006), did not accurately reflect observed vegetation patterns. Three important accuracy issues were identified with respect to the 2006 vegetation map:

1. Inaccurate delineation of vegetation type boundaries.

2. Incorrect assignment of areas to a vegetation type class that did not reflect the characteristics of communities observed on the ground.

3. Redundant vegetation type descriptions and the existence of new or undescribed vegetation types.

Therefore, the purpose of this project was to resolve these issues in the NW vegetation map using currently available datasets, and to publish the revised vegetation map while aligning to SANBI's guidelines for revisions (Dayaram et al. 2021). Once published, the NW vegetation map can be reviewed for incorporation into the National Vegetation Map.

Whilst the current NW portion of the National Vegetation Map was published in 2006, the origin of the vegetation type boundaries as they currently appear in the map can be traced back to the agricultural land type maps prepared by the Department of Agriculture (Mucina et al. 2006). Land types were originally designed to serve the agricultural industry, and these would be areas with generally uniform climate, terrain and soil patterns (MacVicar et al. 1974).

The vegetation type boundaries as mapped in the 2006 National Vegetation Map are based on a vegetation map for the NW prepared by Bredenkamp and Brown (2003a). Unfortunately, all metadata relating to the development of this map was lost other than a hard-copy version of this map that was subsequently digitised¹. The boundaries in this map, however, appear to be based predominately on the agricultural land type maps for NW. This assumption is supported by a comparison of the boundaries between the Bredenkamp and Brown (2003a) and the land type maps that indicates there is an 80% spatial coincidence of boundaries between the two maps (Figure 1). There has been significant development and refinement in the vegetation type concepts since Acocks (1953, 1975, 1988), but there has been comparatively little refinement of the vegetation type boundaries in the NW. Essentially, the majority of vegetation type boundaries as they appear in the 2006 vegetation map, and subsequent 2018 map, were first mapped sometime in the 1980s. With the current availability of high-resolution aerial imagery, the emergence of desktop Geographic Information System (GIS) mapping technology, as well as the increased importance and use of the vegetation map for site-based environmental management, these have exposed the boundary errors inherent in the 2006 map and have highlighted the need to update the boundary mapping in this map.

The vegetation concepts and descriptions in the NW vegetation map draw on concepts in previous vegetation maps of South Africa (Acocks 1988; Low & Rebelo 1996), as well as the Bredenkamp and Brown (2003a) vegetation map. There are at least 93 published studies or reports relating directly to the vegetation of the NW. This body of literature, however, discusses almost exclusively vegetation concepts at the plant community scale. There are few quantitative studies that explicitly explore the floristic and environmental relationships between phytosociological units at the scale of the vegetation type, and none that attempt to define vegetation types using phytosociological approaches or discuss relationships between phytosociological units and vegetation types.

This is not unexpected as the bulk of the relevant vegetation science literature predates the current vegetation type concepts published in 2006. Post-2006 there has been very little phytosociological research published that covers the northwest region of South Africa. The absence of research directly exploring the relationships between phytosociological units and vegetation types, whilst not unexpected, highlights a very important vegetation science research gap. There is a clear need for more quantitative vegetation science research to develop and refine the current vegetation type concepts at the spatial scale at which they are conceived, as this provides the scientific justification for the concepts which is necessary to affirm the application of vegetation types in legislative and legal processes.

Bredenkamp and Brown (2003b) used a phytosociolog-ical approach to define higher-order vegetation association concepts for the Bankenveld in the Highveld region that are at a similar conceptual scale to vegetation types. Similarly, Winterbach (1998) and Winterbach et al. (2000) defined higher-order vegetation association concepts in the Arid Sweet Bushveld region of the NW to derive units that approach vegetation types. Van der Meulen and Westfall (1979) used agricultural land types as the basis to define and delineate vegetation units. In all these studies the same basic set of environmental elements are associated with these higher-order units, namely, soil (clay vs sandy soils on plains), terrain (plains vs mountains) and geology (quartzite vs igneous).

The spatial extent of individual vegetation studies relevant to the NW varies considerably. Some studies accept the agricultural land types as acceptable vegetation mapping units and conduct phytosociological analyses within these units (Bezuidenhout et al. 1993; Bezuidenhout et al. 1994a, 1994b) or across these units (Van der Meulen & Westfall 1979; Smit 2000). Other studies are conducted at a broader general geographic area (Morris 1976; Bredenkamp et al. 1989; Bezuidenhout & Bredenkamp 1990; Du Preez & Venter 1990a, 1990b; Bezuidenhout et al. 1994c, 1994d), or geological area (Bezuidenhout et al. 1988; Bezuidenhout et al. 1994b), or protected area (Van Zyl 1965; Coetzee 1975; Bredenkamp & Bezuid-enhout 1990; Bredenkamp et al. 1994; Stalmans & De Wet 2003), or even part of a protected area (Brown & Bredenkamp 1994; Brown et al. 1995, 1996).

There are at least 29 published papers or reports that include fine-scale vegetation maps for their respective study areas that are relevant to the NW (Figure 2). Excluded from this list are phytosociological studies that used agricultural land types as the mapping unit rather than generating their own vegetation maps (e.g. Be-zuidenhout et al. 1993; Bezuidenhout et al. 1994a, 1994b; Smit 2000; Van der Meulen & Westfall 1979). Collectively, these fine-scale maps cover 260 000 ha or 2.5% of the province. Despite there being a reasonable wealth of vegetation studies relevant to the region, there is a relative paucity in the extent of published vegetation maps. Added to this is the lack of curation of this information with none of the vegetation maps having spatial data in an accessible data archive.

Whilst the vegetation type concept has been accepted and used in South Africa at least since Acocks (1953), it was only in the 2006 version of the vegetation map that the current vegetation type concept was clearly articulated and defined (Mucina et al. 2006). Despite this major advance in the vegetation map, at least for the vegetation types occurring in the NW, it is not clear in the current delineation and description of the vegetation types what are the environmental variables or factors and species or communities that differentiate one vegetation type from another. These variables are implicit in the 'verbal models' used to define and delineate vegetation types (Mucina et al. 2006). However, a clear functional understanding or description of the differentiating factors between vegetation types is absent in the current descriptions of vegetation types. This is often cited by users of the NW vegetation map as being a limitation to using and interpreting the current map at the site level.

The National Vegetation Map is mapped at a broad spatial scale of a whole region or landscape. At the site level there will inevitably be boundary errors when using the vegetation map due to the difference between the scale of map production and scale of use. Therefore, users invariably have to interpret on-the-ground observations of vegetation patterns to 'fine scale' the vegetation map and determine the appropriate vegetation type or types occurring at a site. For users of the vegetation map to be able to make this interpretation at the site level, an understanding of the relationship between underlying environmental variables and the delineation of vegetation types is necessary. Having a clear understanding or model for where and why vegetation types occur is essential for the consistent and defensible mapping of vegetation boundaries, and ultimately the integrity of the vegetation type concept. Whilst this thinking is implicit in the current delineation of vegetation types, it is not, however, always made explicit or clear in the current vegetation type descriptions.

Given these observations and limitations of the current National Vegetation Map in the NW, the objectives of this project were to:

1. Draw on the existing vegetation type classification and descriptions to develop an identification key to vegetation types in the NW based on broad environmental variables.

2. Review existing studies, expert inputs and field observations to determine if there are redundant vegetation types (i.e., two vegetation types that can be merged) or undescribed vegetation types that need to be added to the map and, where possible, support proposed changes with numerical data, and use this information to update the current vegetation type descriptions.

3. Using the identification key in conjunction with available environmental spatial data and current high-resolution aerial imagery, remap vegetation type boundaries at a higher spatial resolution.

 

Study area

The NW is located on the African Plateau in central southern Africa on the border between South Africa and southern Botswana. The province is 104 881 km² and measures roughly 550 km (east-west) by 380 km (north-south). It straddles three major physiographic regions: in the west, parts of the Kalahari region, in the northeast, the Bushveld region and in the southeast, the Highveld region. These broad geographic regions are associated with three major drainage systems, namely the Molopo catchment in the Kalahari, Vaal catchment in the High-veld and Limpopo catchment in the Bushveld. The Mol-opo and Vaal systems drain towards the west into the Orange River and ultimately the Atlantic Ocean, whereas the Limpopo system drains to the northeast into the Indian Ocean (Figure 3A).

The median elevation of the NW is 1 271 m (mean 1 263 m, minimum 904 m, maximum 1 817 m). It is a relatively flat to gently undulating landscape punctuated with few and scattered regions of hills or mountains (Figure 3A). The major mountain ranges of the province are to be found in the Northern Bankenveld entailing the Dwars-berg and Rant van Tweedepoort, the Southern Bankenveld entailing the Magaliesberg, Witwatersberg, Enzelsberg and Swartruggens (Partridge et al. 2010), the Pilanesberg, the hilly landscape spanning between Wolmaransstad to Hart-beesfontein known as the Maquassi Hills, the predominantly east-facing low cliffs of the Ghaap Plateau forming a west dipping cuesta on the border between the NW and Northern Cape, and the Vredefort Dome in the southeast bordering Gauteng and Free State provinces. For all these mountain ranges the elevational range between the surrounding plains, valleys and summits rarely exceeds 300 m. The largest altitudinal gradient is located in the western Magaliesberg and Pilanesberg, where the maximum elevational range is approximately 600 m.

The climate of the NW is humid to semi-arid subtropical in character. Rainfall ranges from near 800 mm per annum in the Highveld on the eastern border with Gauteng and decreases to 250 mm in the extreme west of the province. There is a single summer-rainfall season from October through to April. Temperatures are coolest with higher incidence of frost on the Highveld, while the northern savannas are warmest. The Kalahari region has the warmest summer temperatures and the Bushveld region the mildest winters (Figure 4 and 5). Mucina and Rutherford (2006) described the climate of each vegetation type in more detail.

The geology of the region is varied (Figure 6); however, a singular dominant factor influencing vegetation patterns across the province is the widespread presence of Tertiary aeolian Kalahari sand. Outside of the Kalahari region, relic pockets of these sands can be encountered throughout most of the province. In terms of the underlying geology, important rock types with strong influences on vegetation are quartzite-rich sedimentary rocks giving rise to dystrophic sandy soils contrasted with mafic and ultramafic rocks giving rise to base-rich clay soils.

 

Figure 7

 

The flora of the NW is discussed in some detail by Hahn (2013). The flora is characterised by comprising mostly widespread species with very low levels of endemism. There are at least 2 786 species (2 387 indigenous and 399 not indigenous) recorded in the NW (see Supplementary Material 1) with 16 species (0.6%) known to be endemic or near-endemic to the province (Hahn 2013). Five species (44%) within this group of endemic species are associated with dystrophic quartzite geology of the Magaliesberg and Swartruggens regions, which is assigned to the Gold Reef Mountain Bushveld vegetation type (Table 1).

 

Figure 8

 

 

Figure 9

 

Methods

Vegetation mapping

Vegetation type polygons were manually mapped using a heads-up digitising technique (Kennedy 2009). The vegetation types were delineated by interpreting patterns observed in colour aerial imagery overlayed with data layers representing the environmental variables used in the identification key to define vegetation types, namely: (1) land types as a proxy for soils, (2) simplified geology, and (3) terrain. In total 24 spatial datasets were used to inform the mapping process (Table 2).

In addition to the vegetation type identification key that provides a regional-scale framework for interpreting and mapping vegetation types, field observations and published descriptions of landform-vegetation relationships were also used to interpret patterns in aerial imagery at the local scale. Examples of landform-vegetation relationships include catena vegetation sequences or agricultural landtype map descriptions of landform-soil relationships. Different vegetation communities are associated with different landforms, and the landform-vegetation patterns tend to differ between vegetation types.

Vegetation identification key

To map vegetation in a logical and defensible manner it is necessary to have a framework for how vegetation types are classified and related to one another based on vegetation and floristic patterns and underlying environmental variables or determinants of vegetation types. Mucina et al. (2006) describe such a classification framework for how vegetation types in South Africa are circumscribed that forms the basis for how vegetation types are defined and mapped in the current National Vegetation Map. As described in the introduction, in the NW it is often not clear from the existing verbal models describing vegetation types what the defining features are of a vegetation type and what separates one vegetation type from another. Therefore, before any remapping of vegetation type boundaries could be attempted it was necessary to distil from existing vegetation type descriptions, expert inputs, published vegetation studies and field observations what the key environmental determinants are for each vegetation type, and use this information to develop an identification key to the vegetation types being mapped.

Vegetation type mapping is generally not concerned with mapping plant assemblage boundaries, but rather mapping higher-order spatial scale environmental discontinuities such as aspect, slope, elevation, soil, geology and landform. These are the same variables used to define land types, hence the close historic association between land types and vegetation types. The identification key developed here uses only broad environmental variables to define vegetation units stratified by bioregions or biomes, which represent the major climatic gradients present in the province.

As the first step in remapping the vegetation type boundaries of the NW, a basic identification key to the vegetation types of the province based on mappable environmental variables was developed. This key provided the quantitative framework within which input environmental and imagery datasets could be interpreted and vegetation boundaries mapped in the GIS. The key was based primarily on environmental attributes, but to increase utility for vegetation type identification in the field, broad vegetation structural attribute data was also included in the key. Vegetation structural characteristics are a function of underlying environmental attributes but are not always observable in single observation colour aerial imagery and therefore are not necessarily a reliable variable to use for mapping vegetation.

Species data

Plant species information was collated from existing data sources, as well as from data collected by this project. Data sources include:

1. Herbarium record data from SANBI's POSA database (SANBI 2016).

2. Published vegetation surveys that have been collated and archived in SANBI's National Vegetation Map Database (NVD).

3. Rapid vegetation survey plots and species observations conducted by this project and added to iNaturalist.

A current global species list for the province was created from herbarium record data. The purpose of the global species list was to provide a total flora context for the vegetation survey plot data and also provide a master species list against which to compare and correct plot species data. Data from SANBI's POSA database was obtained via a direct data request. The NW includes all or part of 229 unique quarter degree squares (QDS).

Vegetation survey plot data from most of the phyto-sociological studies that have been undertaken in the province have been collated and archived in the NVD. This is a national database that strives to archive all published vegetation survey data in South Africa. The database currently hosts data for about 58 000 plots. Plots from in and around the NW were extracted from this database for analysis. The purpose of this data was to: (1) inform the important species information in the vegetation type descriptions; and (2) to conduct an ordination analysis to compare the numerical classification plot data versus the current classification of vegetation types.

The purpose of the rapid vegetation survey was to gather species and vegetation type (plant community and dominant species) observation data and photographs of vegetation types over as wide an area as possible in a limited time period. Field work was carried out over two growth seasons (2021/22 and 2022/23). The sampling method relied on noting discernible changes in the vegetation type along a catenal sequence then filling in a prescribed data sheet. A mobile version of the vegetation map was available on the CarryMap application for use in the field. This included both the 2018 version of the national map, as well as an unpublished NW vegetation map created by P Desmet (NW READ 2015). This mobile app allowed for the live tracking of an individual as they move through the landscape and the identification of the existing mapped vegetation type present at a sampling location. Dominant species for a vegetation type were identified and noted and species with ethnobotanical importance, limited distribution or threatened and protected species were photographed and lodged on the iNaturalist App (https://www.inat-uralist.org). Representative photographs of the vegetation type at each site were also uploaded with each species observation, and these were linked to the South African Vegetation Map project in iNaturalist.

Additional vegetation observation data from two previous field campaigns conducted by the authors in 2015 and 2018 were also collated and added to the observation database.

Expert data

Vegetation experts with experience of either mapping or using the provincial vegetation map were also engaged to canvas their opinion on what needed to be changed or updated in the revised map. Input from experts comprised either (1) verbal inputs, (2) relevé data-sets that were not currently in the NVD or, (3) relevant documents or spatial data such as unpublished reports or GIS shapefiles that were not considered in Mucina and Rutherford (2006).

 

Results

Species data

In total 2 985 plots were extracted from the NVD that fall in or within 20 km of the NW (Figure 2). Of this 1 608 (54%) have no accurate georeference, i.e., locality information comprising a description only with no sample point latitude/longitude. For these plots, a geolocation was added based on the nearest town or area that could be determined from the plot locality description data, or failing this, locality clues present in the title of the project or source publication. The sampling density of plots is low. For the 2 985 plots selected from the NW plus 20 km buffer, this is a sampling density of approximately 1 plot per 50 km²; however, only 785 plots fall within the NW equating to a sampling density of approximately 1 plot per 130 km².

In total the NVD dataset contains 28 705 records for 1 610 species (Table 3). This equates to a sampling density of approximately 1 record per 5 km². Note that only genus and species are considered here and no subspe-cific taxa are considered. In contrast to the vegetation survey plot data, the global species list, derived from POSA herbarium record data for the province at the genus and species levels, contains 3 040 taxa of which 407 are not native (Table 4). That means for indigenous species (2 633 taxa) only 61% of species known to occur in the province have been recorded in nearly 3 000 vegetation survey plots.

Vegetation type identification key

An identification key for the vegetation types of the NW (Table 6) was developed based on 15 broad environmental variables grouped into five variable categories (Table 5). The identification key is able to discriminate and identify all 36 terrestrial vegetations types that occur in the province plus the three 'azonal' types associated with hydrologically driven ecosystems.

Summary of changes made to the vegetation map

Changes to the NW vegetation map are summarised according to the potential types of changes described by the National Vegetation Map Committee (Table 7). The changes in vegetation type extents are summarised in Table 8.

Revised vegetation map

The revised vegetation map contains 1 810 polygons compared to the current vegetation map that has 159 polygons (Figure 10). Whilst the vegetation type concepts remain mostly unchanged from Mucina and Rutherford (2006), polygon boundaries have been entirely remapped, and this has resulted in significant changes in extent from most vegetation types (Table 8). The remapping of boundaries is an inevitable product of the much higher resolution mapping informants available to this project, as well as the application of the vegetation type identification key.

Updated descriptions of North West terrestrial vegetation types

The vegetation type descriptions have been updated (Supplementary Material 2: Vegetation type descriptions) to reflect new data available and to better align with the vegetation type identification key (Table 6). Descriptions are based on the original descriptions that appear in Mucina and Rutherford (2006) and where necessary these have been updated based on the inputs presented in Table 9.

 

Discussion

The revised vegetation map is significantly changed mainly with respect to where the boundaries of vegetation types are mapped. Whilst there are some changes proposed to the classification of vegetation types, for the most part, the current vegetation type concepts remain unchanged. The change in the mapping of vegetation type boundaries is an inevitable result of: (1) a clearer understanding of vegetation determinants (i.e., vegetation type classification framework or identification key); and more importantly, (2) the much-increased resolution and availability of mapping informants. It is very important to note that the significant change in mapped boundaries does not suggest or imply in any way that the current vegetation type concepts are invalid.

The vegetation identification key is important for informing the current vegetation map revision. It also serves a far greater purpose beyond just this vegetation map revision. Firstly, it enables users of the vegetation map to clearly understand how vegetation is assigned to different vegetation types and therefore users can apply the classification framework to mapping vegetation at finer spatial scales. When mapping vegetation at the provincial scale there are time and budget constraints limiting the amount of detail that can be mapped relative to what can be observed in the informants. It is not practically possible to manually map vegetation at infinitely fine scales over large regions. Therefore, there are inherent boundary or misclassification errors in the final map product due to mapping scale. Using the vegetation type identification key it is possible for users of the map to apply the classification framework at a fine or local spatial scale to improve mapping accuracy or interpretation for specific purposes, for example, fine-scale vegetation mapping for environmental impact assessments.

Secondly, the identification key can illuminate inconsistencies in the current vegetation type classification and thus identify where vegetation types could be split, aggregated or new ones defined. For example, one such inconsistency highlighted with this project relates to the definition and mapping of Central Sandy Bushveld. This vegetation type contains both plains and mountain habitat, as well as several major geological rock types (granite and quartzite/sandstone). It is likely that applying a similar vegetation type classification as used here to elsewhere in South Africa will identify inconsistencies in the definition and mapping of vegetation types.

A stated objective of this project was to conduct a quantitative floristic analysis to find support for the vegetation type concepts using the available relevé database. This objective was not achieved within the allocated project time period. Quantitative floristic analysis to validate the vegetation type concepts used in the National Vegetation Map is a major research gap that should be addressed not only for the NW, but also more broadly in South Africa. These analyses should be earmarked as a future research priority.

There is some support in the literature for the vegetation type concepts as framed in the vegetation type identification key. For example, the Bredenkamp and Brown (2003b) analysis of the Bankenveld area supports the mountain vs plains vegetation distinction, as well as separation of grasslands based on moisture availability, soil texture and depth. Within the mountain category, vegetation units are separated based on aspect and elevation rather than geological rock type. This does suggest that within the current vegetation type classification there will be a necessary and pragmatic tradeoff between ecological units that are easy to map and identify (viz. discrete mountains with similar geology) versus phytosociologically correct units that are more complex to map (viz. aspect and elevation gradients). Similarly, the Winterbach (1998) and Winterbach et al. (2000) analysis of the Arid Sweet Bushveld region also supports the major environmental divisions associated with higher-order vegetation associations, namely, clay vs sandy soils on plains, and plains vs mountains. Both these studies suggest that it is highly likely that the current vegetation type concepts can be supported and further refined through quantitative floristic analysis.

A deficit of observations on the iNaturalist app for the NW was noted. iNaturalist is a very accessible and practical tool for collecting and identifying biodiversity information. Two training workshops were held with DEDECT officials to introduce them to the potential of the iNatu-ralist and the Carrymap apps. Within the province the iNaturalist app could have future applications for gathering biodiversity data, monitoring environmental compliance; to improve decision making in the EIA process; and to monitor the distribution of invasive alien species. Within the context of this study, iNaturalist proved very useful for capturing and linking field observations to a national database. Species observations in iNaturalist were uploaded together with context photographs of the vegetation type and linked to an iNaturalist National Vegetation Map project managed by SANBI. This project is using iNaturalist to collect representative photographs of all South African vegetation types.

An interesting observation with regards the species data is the large disparity between observations collected via survey plots versus herbarium records. Nearly 40% of the province's flora has never been recorded in a vegetation survey plot. This observation can be partly due to the fact that surveying flora for vegetation analysis (i.e., releves) tends to under report or omit uncommon and rare species. This observation can also be due to under sampling of the province for vegetation analysis. The very low sampling density of releves; the tendency for samples to be clumped rather than uniformly distributed; and the disparity in species records between herbarium versus plot data would suggest that from a vegetation description and analysis perspective that the NW is significantly under sampled. As highlighted above there is still a need for further floristic surveys and analysis to better understand and describe our vegetation types.

This revision of the NW vegetation map has focused on the terrestrial ecosystems of the province and therefore the descriptions of 'azonal' ecosystems are not updated here. Consideration of these ecosystems is, however, central to the mapping process, as well as understanding of terrestrial ecosystems. In the mapping process these ecosystems are generally always mapped first as they are often the easiest units to identify, more importantly they provide a concrete starting point for interpreting the input data in relation to the identification key and ultimately understanding vegetation/landscape patterns. The NW has for the most part relatively flat landscapes that support wide floodplain/alluvial ecosystems and extensive endorheic pan ecosystems. The province also straddles three major biogeograph-ic regions that influence the vegetation composition of these ecosystems. Therefore, there is a great extent and diversity of azonal systems, and it has been necessary to map the larger occurrences of these ecosystems to have more consistent environmental and floristic definitions of terrestrial ecosystems.

In the revised vegetation map the extent of azon-al ecosystems has been significantly extended from the 198 000 ha or 2% in the 2006 vegetation map to 618 000 ha or 6% of the province in the present map. Mapping has focused on azonal ecosystems associated with drainage lines and there has been no attempt to map endorheic pan systems except where these are associated with drainage lines.

It must be noted that during this project there was extensive discussion amongst stakeholders of the appropriateness of the term 'azonal'. In the NW context these ecosystems include all ecosystem types where the occasional occurrence of surface water and waterlogged soils is amongst the primary environmental determinants of ecosystem structure, function and definition. The term azonal could apply to any ecosystem with limited extent or that occurs widely across the landscape as a distinct feature within other ecosystems. The term is also discriminatory towards aquatic/ wetland ecosystems as azonal can imply ecosystems of lesser importance or status. In the terrestrial realm collective terms such as grassland or savanna are used to group ecosystems. These terms are broadly descriptive of the nature of the contained ecosystems. Conversely, the term azonal in the context of the NW vegetation map does not convey the very important fact that the contained ecosystems are all determined and driven by water and hydrological processes. Whilst the use of the term 'azonal' is retained here, it is highly recommended that a new collective term for ecosystems driven by water be sought that is accepted by other terrestrial and aquatic/wetland ecologists. For the three azonal ecosystems considered in the NW vegetation map, the term 'alluvial' ecosystems would be a much more appropriate descriptive name.

Azonal ecosystems have been grouped into three existing ecosystem types, each associated with the three major river catchments/bioregions of the province. They are:

1. AZa 5 Highveld Alluvial Vegetation in the Highveld/ Vaal River catchment including AzF3 Eastern Temperate Freshwater Wetlands.

2. AZi 3 Southern Kalahari Mekgacha in the Kalahari/ Molopo River catchment.

3. AZa 7 Subtropical Alluvial Vegetation in the Bush-veld/Crocodile River catchment.

Notable fluvial landscapes of the province include:

1. Kgomo-Kgomo/Tswaing area has extensive grassland floodplains (AZa 7 Subtropical Alluvial Vegetation) associated with several rivers flowing northwards out of Gauteng into the Pienaars/Moretele River and includes the Apies, Tshwane and Kutswane rivers. These floodplain ecosystems are unique within the NW and possibly within the Bushveld Bioregion. The only other area in South Africa with similar floodplain ecosystems is the Nylsvlei in Limpopo. Unfortunately, these ecosystems are being heavily impacted by sprawling peri-urban and rural settlements. This area is in great need of conservation action, as well as wetland rehabilitation.

2. The Senegalia galpinii riparian gallery forest on the Crocodile River, where the Moretele River enters, is one of the most iconic AZa 7 Subtropical Alluvial Vegetation riparian communities in the province.

3. The Mooi River catchment above the Klerkskraal Dam is possibly the largest and most intact mesic fluvial system on the western Highveld (mapped as AZa 5 Highveld Alluvial Vegetation in this map and as Temperate Freshwater Wetlands in the 2006 vegetation map), and as such should receive greater conservation focus, as it is essentially the last remaining intact grassland catchment landscape on the western Highveld.

4. Very extensive grassland floodplain systems (AZa 5 Highveld Alluvial Vegetation) are mostly associated with Gh 14 Western Highveld Sandy Grassland. These fluvial systems cover nearly 250 000 ha. Distinct features of these fluvial systems are the presence of surface calcrete; the general lack of well-defined river channels; and are often associated with networks of pans indicative of palaeo-river channels (mapped as AZi 10 Highveld Salt Pans in the SA Vegetation Map); and the singular dominance of the tree Searsia lancea.

5. A defining environmental characteristic of AZi 3 Southern Kalahari Mekgacha is the presence of surface calcrete indicating the 'riverbed'. A unique vegetation feature associated with this calcrete not identified by previous authors is the abundance of species with Nama Karoo affinities such as Pentzia incana (Asteraceae), Ruschia griquensis, Ruschia semidentata, Ruschia spinosa (Aizoaceae) and various Zygophyllaceae and Acanthaceae. This is the only vegetation type within the Kalahari bioregion where succulent taxa are encountered in any abundance and represents the only major incursion of Nama Karoo biome affinities into the NW and the Savanna biome.

Much of south-central NW has been ploughed for crop production and this makes observing vegetation type boundaries on the ground, or at least where they used to occur, almost impossible to observe. This is particularly difficult across the transitions of Highveld grassland types that are the primary target of cultivation. Whilst these boundaries are not clearly observable today, having a clear vegetation-environment model does make predicting where these boundaries are likely to be much easier. This reinforces the importance of having such a model for mapping and describing vegetation types, and it would be beneficial if this model could be extended to include all vegetation types in South Africa.

 

Conclusions

The revised vegetation map of the NW is a significant improvement on the 2018 National Vegetation Map and has been incorporated into the current NVM 2024 beta release. Firstly, the vegetation type classification model based on five broad environmental variables (flooding, bioregion, terrain, geology and soil) provides a consistent and explicit framework for understanding the distribution and hence mapping vegetation types. Whilst agricultural land types remain a good proxy for mapping vegetation, breaking these units down in their underlying environmental determinants (soil and topography) and mapping these provide better proxies for mapping vegetation. Secondly, the abundance of high-resolution remote sensing products, relative to the 1980s when land types were mapped, means that vegetation type boundary accuracy is significantly improved. Thirdly, although a quantitative phytosociological analysis was not able to be completed, based on the description of existing vegetation types, the available literature, stakeholder inputs and field observations, the current vegetation type concepts are valid units. It was necessary, however, to elevate three previously described vegetation concepts as new vegetation types to accommodate observable vegetation patterns in the landscape and also to align with the vegetation classification model.

It is recommended that the existing azonal vegetation type category be replaced with the term alluvial for the three azonal vegetation types in the NW where occasional flooding or waterlogging is a primary determinant of vegetation. This alluvial vegetation type unit also contains the majority of wetlands in the province.

This will not be the last word on the mapping of vegetation in the NW. Despite the wealth of phytosociological literature available for the province, there are still major gaps in our descriptive vegetation science knowledge in the province. Also, there is no research at all that relates phytosociological vegetation concepts to the modern South African vegetation type concepts, and the vegetation classification model or framework developed here might provide the basis for developing a similar framework for the entire country. Given the importance of vegetations types in environmental policy, planning and decision making, having clear, consistent and defensible environmental definitions for vegetation types will help practitioners identify and map vegetation types on the ground.

 

Acknowledgements

This project was entirely funded by the North West Provincial Government Department of Economic Development, Environment, Conservation and Tourism (Project number DEDECT 06/2021). The South African National Biodiversity Institute (SANBI) provided extensive technical support to the project, as well as facilitated access to data and contributed to fieldtrip expenses. We are grateful to the members of the project steering committee for their support and inputs on the project design and execution, namely, Tharina Bo-shoff, Adriaan van Straaten, Aluwani Tshiila, Malefyane Mosadi, Doug Macfarlane, Leo Quayle, Ryan Kok and Willem Boshoff. Many individuals provided valuable technical inputs, datasets or comments on the vegetation map, namely, Andrew Skowno, Nacelle Collins, Lorraine Mills, George Bredenkamp, Jacobus Smit, Marc Stalmans, Tony de Castro, Noel van Rooyen, Naas Grové, David Hoare and Leslie Brown. The members of the National Vegetation Map Committee who reviewed and commented on the draft manuscript are thanked for their insights: Tony Rebelo, Debbie Jewitt, Andrew Skowno and Kagiso Mogajane.

 

References

Acocks, J.PH., 1953, 'Veld types of South Africa', Memoirs of the Botanical Survey of South Africa 28, 1-192.         [ Links ]

Acocks, J.PH., 1975, 'Veld types of South Africa', 2nd ed., Memoirs of the Botanical Survey of South Africa 40, 1-128.         [ Links ]

Acocks, J.PH., 1988, 'Veld types of South Africa', 3rd ed., Memoirs of the Botanical Survey of South Africa 57, 1-146.         [ Links ]

Bezuidenhout, H., 1988, "n Plantekologiese studie van die Mooirivieropvanggebied, Transvaal', MSc thesis, Potchefstroom University [University of the North West], South Africa.         [ Links ]

Bezuidenhout, H. & Bredenkamp, G.J., 1990, 'A reconnais sance survey of the vegetation of the dolomitic region in the Potchefstroom-Ventersdorp-Randfontein area, South Africa', Phytocoenologia 18, 387-403, https://doi.org/10.1127/phyto/18/1990/387.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Elsenbroek, J.H., 1988, 'The vegetation of the alkali granite and border ing quartzite in the Vredefort Dome north-west of Parys', South African Journal of Science and Technology 7, 4-9, https://doi.org/10.4102/satnt.v7i1.892.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Theron, G.K., 1993, 'The vegetation of the Bd and Ea land types in the grass land of the western Transvaal, South Africa', South African Journal of Botany 59, 319-331, https://doi.org/10.1016/S0254-6299(16)30735-9.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Theron, G.K., 1994a, 'A Braun-Blanquet reclassification of the CymbopogonThemeda grassland in the Lichtenburg area, south-western Transvaal', South African Journal of Botany 60, 306-314, https://doi.org/10.1016/S0254-6299(16)30584-1.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Theron, G.K., 1994b, 'Phytosociological classes of the western Transvaal grassland, South Africa', Koedoe 37, 1-18, https://doi.org/10.4102/koedoe.v37i1.322.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Theron, G.K., 1994c, 'A classification of the vegetation of the western Transvaal dolomite and chert grassland, South Africa', South African Journal of Botany 60, 152-161, http://dx.doi.org/10.1016/S0254-6299(16)30626-3.         [ Links ]

Bezuidenhout, H., Bredenkamp, G.J. & Theron, G.K., 1994d, 'Syntaxonomy of the vegetation of the Fb land type in the western Transvaal grassland, South Africa', South African Journal of Botany 60, 72-81, https://doi.org/10.1016/S0254-6299(16)30663-9.         [ Links ]

Bredenkamp, G.J. & Bezuidenhout, H., 1990, 'The phytosociology of the Faan Meintjes Nature Reserve in the western Transvaal grassland, South Africa', South African Journal of Botany 56, 54-64, https://doi.org/10.1016/S0254-6299(16)31111-5.         [ Links ]

Bredenkamp, G.J., Bezuidenhout, H., Joubert, H. & Naude, C., 1994, 'The vegetation of the Boskop Dam Nature Reserve, Potchefstroom', Koedoe 37, 19-33, https://doi.org/10.4102/koedoe.v37i1.323.         [ Links ]

Bredenkamp, G.J. & Brown, L.R., 2003a, 'Habitat Types of North-West Province', in 'North West Province Biodiver sity Site Inventory and Database Development', Technical Report, Strategic Environmental Focus (Pty) Ltd, Pretoria.         [ Links ]

Bredenkamp, G.J. & Brown, L.R., 2003b, 'A reappraisal of Acocks' Bankenveld: origin and diversity of vegetation types', South African Journal of Botany 69, 7-26, https://doi.org/10.1016/S0254-6299(15)30357-4.         [ Links ]

Bredenkamp, G.J., Joubert, A.F. & Bezuidenhout, H., 1989, 'A reconnaissance survey of the vegetation of the plains of the Potchefstroom-Fochville-Parys area', South African Journal of Botany 55, 199-206, https://doi.org/10.1016/S0254-6299(16)31208-X.         [ Links ]

Brown, L.R., 1997, 'A plant ecological study and wildlife man agement plan of the Borakalalo Nature Reserve, North West Province', PhD thesis, University of Pretoria, South Africa, http://hdl.handle.net/2263/82472.         [ Links ]

Brown, L.R & Bredenkamp, G.J., 1994, 'The phytosociology of the southern section of Borakalalo Nature Reserve, South Africa', Koedoe 37, 59-72, https://doi.org/10.4102/koedoe.v37i2.337.         [ Links ]

Brown, L.R., Bredenkamp, G.J. & Van Rooyen, N., 1995, 'The phytosociology of the western section of Borakalalo Nature Reserve', Koedoe 38, 49-64, http://dx.doi.org/10.4102/koedoe.v38i2.314.         [ Links ]

Brown, L.R., Bredenkamp, G.J. & Van Rooyen, N., 1996, 'The phytosociology of the northern section of the Borakalalo Nature Reserve', Koedoe 39, 9-24, https://doi.org/10.4102/koedoe.v39i1.279.         [ Links ]

Coetzee, B.J., 1975, 'A phytosociological classification of the Rustenburg Nature Reserve', Bothalia 11, 561-580, https://doi.org/10.4102/abc.v11i4.1502.         [ Links ]

Dayaram, A., Harris, L.R., Grobler, B.A., Van der Merwe, S., Rebelo, A.G., Ward Powrie, L., Vlok, J.H.J., Desmet, P.G., Qabaqaba, M., Hlahane, K.M., & Skowno, A.L., 2019, 'Vegetation Map of South Africa, Lesotho and Swaziland 2018: A description of changes since 2006', Bothalia 49, a2452, https://doi.org/10.4102/abc.v49i1.2452.         [ Links ]

Dayaram, A., Skowno, A.L., Driver, A., Sink, K., Van Deventer, H., Smith-Adao, L., Van Niekerk, L., Harris, L.R., Job, N. & Nel, J.L., 2021, The South African National Ecosystem Clas sification System Handbook, 1^st ed., South African National Biodiversity Institute, Pretoria, South Africa, https://opus.sanbi.org/jspui/handle/20.500.12143/7150.         [ Links ]

Du Preez, PJ. & Venter, H.J.T., 1990a, 'The phytosociology of the woody vegetation in the southern part of the Vredefort Dome Area. Part I: Communities of the plains, riverbanks and islands', South African Journal of Botany 56, 631-636, https://doi.org/10.1016/S0254-6299(16)30998-X.         [ Links ]

Du Preez, PJ. & Venter, H.J.T., 1990b, 'The phytosociology of the woody vegetation in the southern part of the Vredefort Dome Area. Part II: Communities of the hills', South African Journal of Botany 56, 637-644, http://dx.doi.org/10.1016/S0254-6299(16)30999-1.         [ Links ]

Hahn, N., 2013, 'Rare, endangered and endemic flora of the North West Province', Unpublished Report, Department of Economic Development, Environment, Conservation and Tourism, North West Provincial Government, https://www.researchgate.net/publication/318673797_Rare_en-dangered_and_endemic_flora_of_the_North_West_Province_Unpublished_Report_to_the_Department_of_Eco-nomic_Development_Conservation_and_Tourism_North_ West_Provincial_Government.         [ Links ]

Kennedy, M., 2009, Introducing Geographic Information Sys tems with ArcGIS: A Workbook Approach to Learning GIS, John Wiley & Sons, New Jersey, USA.         [ Links ]

Low, A.B. & Rebelo, A.G. (eds.), 1996, 'Vegetation of South Af rica, Lesotho and Swaziland. A companion to the Vegetation Map of South Africa, Lesotho and Swaziland', Department of Environmental Affairs and Tourism, Pretoria, South Africa.         [ Links ]

Macvicar, C.N., Scotney, D.M., Skinner, T.E., Niehaus, H.S. & Laubser, J.M., 1974, 'A classification of land (climate, terrain form, soil) primarily for rainfed agriculture', South African Journal of Agriculture Extension 3, 21-24.         [ Links ]

Morris, J.W., 1976, 'Automatic classification of the highveld grassland of Lichtenburg, south-eastern Transvaal', Bothalia 12, 267-292, https://doi.org/10.4102/abc.v12i2.1419.         [ Links ]

Mucina, L. & Rutherford, M.C. (eds.), 2006, 'The vegetation of South Africa, Lesotho and Swaziland', Strelitzia 19, South African National Biodiversity Institute, Pretoria, https://archive.org/details/vegetationofsout19muci.         [ Links ]

Mucina, L., Rutherford, M.C. & Powrie, L.W., 2006, 'The Logic of the map: Approaches and Procedures', in L. Mucina & M.C. Rutherford (eds.), The vegetation of South Africa, Lesotho and Swaziland, pp. 12-29, South African National Biodiversity Institute, Pretoria, http://bgis.sanbi.org/Projects/Detail/209.         [ Links ]

North West Department of Rural, Environment and Agri cultural Development (NW READ), 2015, 'North West Biodiversity Sector Plan', Report, North West Provincial Government, Mahikeng, http://bgis.sanbi.org/Projects/Detail/179.         [ Links ]

Partridge, T.C., Dollar, E.S.J., Moolman, J. & Dollar, L.H., 2010, 'The geomorphic provinces of South Africa, Le sotho and Swaziland: A physiographic subdivision for earth and environmental scientists', Transactions of the Royal Society of South Africa 65: 1-47, http://dx.doi.org/10.1080/00359191003652033.         [ Links ]

Sadr, K., 2022, 'The Stone Towers of Kweneng in Gauteng Province', South African Archaeological Bulletin 77, 115 126, https://www.academia.edu/96516934/THE_STONE_TOWERS_OF_KWENENG_IN_GAUTENG_PROVINCE.         [ Links ]

Smit, J.H.L., 2000, 'Phytosociology and veld management of the eastern Kalahari thornveld', MSc thesis, University of Pretoria, South Africa, http://hdl.handle.net/2263/27293.         [ Links ]

South African National Biodiversity Institute (SANBI), 2016, 'Botanical Database of Southern Africa (BODATSA)', [data-set], http://posa.sanbi.org.         [ Links ]

South African National Biodiversity Institute (SANBI), 2023, 'The VEGMAP Project Handbook: A guide for users and contributors. Version 2.', Pretoria, South Africa. https://opus.sanbi.org/items/715b7d50-b3ed-4dd9-b3c9-c2e-6a8e8e57b.         [ Links ]

South African National Biodiversity Institute (SANBI), 2006 2024, 'The National Vegetation Map Version 2024.' [dataset], https://bgis.sanbi.org/vegmap.         [ Links ]

Stalmans, M. & De Wet, F., 2003, 'Soils and vegetation of the Madikwe and Pilanesberg expansion areas in the Heritage Park, North West Province', Report, International Conser vation Services and EnviroPulse, August 2003.         [ Links ]

Van der Meulen, F. & Westtfall R.H., 1979, 'A vegetation map of the western Transvaal Bushveld', Bothalia 12, 731-735, https://doi.org/10.4102/abc.v12i4.1445.         [ Links ]

Van Zyl, J.H.M., 1965, 'The vegetation of the S.A. Lombard Nature Reserve and its utilisation by certain antelope', Zoologica Africana 1, 55-71, https://doi.org/10.1080/00445096.1965.11447299.         [ Links ]

Wikipedia contributors, 2024, 'Geographic information sys tem', Wikipedia The Free Encyclopedia, accessed 20 May 2024, https://en.wikipedia.org/w/index.php?title=Geo-graphic_information_system&oldid=1221791368.         [ Links ]

Winterbach, R., 1998, 'A phytosociological synthesis of Aca cia tortilis communities in the north-western savanna of South Africa', MSc thesis, University of Pretoria, South Africa, http://hdl.handle.net/2263/83194.         [ Links ]

Winterbach, R., Bredenkamp, G.J., Deutschländer, M.S. & Mucina, L., 2000, Preliminary syntaxonomic scheme of vegetation classes for the Central Bushveld of South Africa', in PS. White, L. Mucina, J.S. Leps & E. van der Maarel(eds.), Proceedings of 41st I.A.V.S. Symposium, Uppsala, pp. 123-127, http://hdl.handle.nety20.500.11937/8283.         [ Links ]

 

 

Correspondence:
Philip G. Desmet
e-mail: drphil@ecosolgis.com

Submitted: 24 October 2023
Accepted: 12 April 2024
Published: 11 October 2024

 

 

Supplementary Data

The supplementary data is available in pdf: [Supplementary data]