Preliminary inventory and classification of indigenous afromontane forests on the Blyde River Canyon Nature Reserve, Mpumalanga, South Africa
© Lötter and Beck; licensee BioMed Central Ltd. 2004
Received: 31 January 2004
Accepted: 02 August 2004
Published: 02 August 2004
Mixed evergreen forests form the smallest, most widely distributed and fragmented biome in southern Africa. Within South Africa, 44% of this vegetation type has been transformed. Afromontane forest only covers 0.56 % of South Africa, yet it contains 5.35% of South Africa's plant species. Prior to this investigation of the indigenous forests on the Blyde River Canyon Nature Reserve (BRCNR), very little was known about the size, floristic composition and conservation status of the forest biome conserved within the reserve. We report here an inventory of the forest size, fragmentation, species composition and the basic floristic communities along environmental gradients.
A total of 2111 ha of forest occurs on Blyde River Canyon Nature Reserve. The forest is fragmented, with a total of 60 forest patches recorded, varying from 0.21 ha to 567 ha in size. On average, patch size was 23 ha. Two forest communities – high altitude moist afromontane forest and low altitude dry afromontane forest – are identified. Sub-communities are recognized based on canopy development and slope, respectively. An altitudinal gradient accounts for most of the variation within the forest communities.
BRCNR has a fragmented network of small forest patches that together make up 7.3% of the reserve's surface area. These forest patches host a variety of forest-dependent trees, including some species considered rare, insufficiently known, or listed under the Red Data List of South African Plants. The fragmented nature of the relatively small forest patches accentuates the need for careful fire management and stringent alien plant control.
Mixed evergreen forests form the smallest, most widely distributed and fragmented biome in southern Africa . Originally classified as Undifferentiated Afromontane forest , 44% of this vegetation type within South Africa has been transformed . The forest biome only covers 0.56 % of South Africa , yet it contains 5.35% of South Africa's plant species . These forests have a relatively high species richness of 0.58 species km-2, exceeding the grassland biome with 0.25 species km-2 and lagging only the fynbos with 1.36 species km-2. The forest biome occurring in Mpumalanga is now recognized as Mpumalanga Mistbelt Forest , and it covers only 0.51% of the province's surface area . Cooper  conducted a study on the conservation status of indigenous forests within Transvaal, Natal and the Orange Free State, in which he estimated the Blyde River Canyon Nature Reserve's (BRCNR) forests at 352 ha. Recently, ownership of a 700 ha forest tract bordering on BRCNR at the base of the Drakensberg Escarpment was transferred to the Mpumalanga Parks Board (MPB). Therefore, it was thought that just over 1000 ha of forest occurred on BRCNR.
To begin to understand the nature and distribution of forests in the BRCNR landscape, we need to have an accurate surface area inventory, a species composition list, and an identification and classification of the communities. Standard ecological fieldwork coupled with remotely sensed data and GIS allow for an efficient analysis. Prior to this investigation of the indigenous forests on BRCNR very little was known about the size and floristic composition of the forest biome conserved within the reserve. In comparison, the grassland biome has received much attention on BRCNR with a monitoring program set up to assess the status of the grasslands. We report here an inventory of forest size, fragmentation, species composition and the basic floristic communities along environmental gradients.
Spatial distribution of forests
Flora and forest classification
Moist high-altitude afromontane forest
This was a tall forest community that was heavily logged at the turn of the 20th century. The canopy is still very uneven as the sub-canopy trees have not yet matured or replaced those trees which were originally removed. This forest occurs in the mist-belt along the escarpment at altitudes of between 1450 m and 1700 m. The slopes are steep and generally scattered with large boulders. Preferential species identified for this community include Clivia caulescens, Ochna arborea, Podocarpus latifolius, Rapanea melanophloeos, Vernonia wollastonii and Blechnum attenuatum. Epiphytes are common in these forests. Within this community, two variants are recognized based on canopy development.
Degraded moist high-altitude forest
A very broken and open canopy with a large shrub component characterizes this forest type. A number of species within this community are shade-intolerant and have established themselves under an open canopy. This sub-community has been so over-utilized that the canopy has never been able to close and this community is sometimes dominated by the exotic invader species Acacia mearnsii. Other preferential species include Rubus spp., Myrsine africana and Helichrysum chrysargyrum. This community is only temporary as it will eventually be replaced by climax species during the next seral stage when the canopy closes.
Developing moist high-altitude forest
This forest type has a closed canopy with only a few pioneer, shade-intolerant species, such as Rhus tumulicola, Maesa lanceolata and Acacia mearnsii, found in the canopy. The shrub component is not nearly as dense as the degraded sub-community community. Preferential species include Behnia reticulata, Dovyalis lucida, Olea capensis subsp. macrocarpa, Asplenium rutifolium and Jasminum abyssinicum.
Dry low-altitude afromontane forest
Parts of this forest type were also logged at the turn of the 20th century. This dry forest community occurs just below the escarpment mist-belt, from 1200 m to 1450 m above sea level. Slopes are steep to gentle, with scattered large boulders. Preferential species in this forest community include Brachylaena transvaalensis, Podocarpus falcatus, Adenopodia spicata, Lauridia tetragona and Pteris captoptera. Within this community, two variants are recognized based on degree of slope.
Dry forest on gentle slopes or along drainage lines
This forest type occurs in areas with a gentle slope or along drainage lines. Both the shrub and herb layer densities are high with the increase in soil moisture gained directly from drainage lines or as a result of poor drainage on gentle slopes. Differential species include Acacia ataxacantha, Combretum kraussii, Eugenia natalitia, Gymnosporia mossambicensis and Faurea galpinii.
Dry forest along steep slopes
These low altitude forests occur along steep slopes. Overall canopy density is high with a poorly developed herb layer. This sub-community eventually grades into the moist high-altitude afromontane forest at higher altitudes. Preferential species include Combretum edwardsii, Cryptocarya transvaalensis, Chionanthus peglerae, Xymalos monospora, Oxyanthus speciosus and Prosphytochloa prehensilis.
Asparagus virgatus, Dicliptera clinopodia, Setaria megaphylla and Desmodium repandum are a few of the herbaceous plants that prefer the lower lying forests. Some of the trees include Euclea crispa, Rhamnus prinoides, Combretum kraussii, Myrsine africana, Lauridia tetragona, Protorhus longifolia and Scolopia mundii. Species that occur in the higher lying forests include Oxyanthus speciosus, Xymalos monospora, Rothmannia capensis and Cassipourea malosana. Some of the species that appear to prefer rocky areas include Clivia caulescens and Blechnum attenuatum. On the other hand, Prosphytochloa prehensilis, Rawsonia lucida and Ptaeroxylon obliquum seem to favor areas almost devoid of rocks or boulders. Species that favor low canopy densities, or higher levels of disturbance, are Rubus sp., Blotiella natalensis, Acacia mearnsii and Schefflera umbellifera.
Geldenhuys  states that species diversity within forest patches is determined by patch size and proximity to other forests, which together explain 82% of the species richness in South Africa's forests. The floristic variation within BRCNR's forests can be attributed predominantly to variation in species composition along an altitudinal gradient. Extremes in environmental variables and gradients, associated with a history of over-utilization, have resulted in a floristically and dynamically diverse forest type. Fire exclusion, clearing of old plantation sites, and a history of intensive and selective logging have all contributed to the variation in forest dynamics currently occurring on BRCNR.
The afromontane forests occurring on BRCNR are extremely fragmented, and yet over 2100 ha of forest patches are conserved within its boundaries. A large proportion of forest-dependent tree species are offered protection within BRCNR's borders. Of special interest was the discovery of Jasminum abyssinicum, Combretum edwardsii and Olinia radiata, all of which occurred in relatively large numbers in the forest. Jasminum abyssinicum had a status of Insufficiently Known under the old Transvaal threatened plants programme , and Hilton-Taylor  currently lists this species under the Red Data List of South African Plants. Jasminum abyssinicum occurred in 64% of the sample plots and in both forest communities identified. Combretum edwardsii is similarly listed as Insufficiently Known and also occurs under the Red Data List of South African Plants. Combretum edwardsii occurred in 73% of the sample plots. Pooley  lists O. radiata as very rare in KwaZulu Natal and Transkei and fails to mention its occurrence north of Natal. Palmer & Pitman  describe O. radiata as a rare species restricted to Natal forests from Pondoland to Zululand. This species was surprisingly common in the canopy of BRCNR's forests and occurred in 82% of the sample plots.
Some severely over-utilized forest patches are currently in a state of recovery; however, current and future damage from invasive trees is a threat to this recovery. Some of the forest patches on BRCNR have a forest margin largely comprised of the invasive black wattle (Acacia mearnsii). The flammable nature of this species, compared to natural forest margin species, allows grassland fires to penetrate forests, resulting in a reduction in forest patch size. With a large edge effect resulting from many small forest patches, there is a need for careful fire management and stringent alien plant control.
According to Geldenhuys , conservation status implies the extent to which populations, species or communities have been modified by the influences of man and the degree to which they might be expected to maintain their genetic diversity and ecological processes in the medium term (10 to 100 years). We see two different aspects to the conservation of the afromontane forest biome in the BRCNR. Firstly, it is the maintenance of the components and critical processes within a forest ecosystem. The disturbed and unstable state of the forest margins are identified as an area requiring further investigation. The effects of the alien tree species (e.g., black wattle) and the destructive burning of forest margins should be of concern to management authorities, as the forest patches are being reduced in size and the impact of edge effects is being amplified on the forest interior. Secondly, it is the maintenance of gene flow between the fragmented forest patches through management of the land surrounding the forest and forest corridors. As the BRCNR forest vegetation is situated along an altitudinal gradient, it therefore seems possible to identify certain forest patches (which may have been harvested in the past, or possibly will in the future) as critical adjuncts for conserved forest patches at the same altitude.
From the evidence we gathered, no "climax" forests exist on BRCNR. Although it is known that BRCNR's forests were utilized, the impact, extent and degree of the utilization are still not quantified. An investigation into the successional and dynamic state of the five largest forest patches is currently underway. Very little of the neighbouring forest on Mariepskop was harvested for its timber, and this forest seems to be the obvious control site for further comparative research.
This study shows that BRCNR has a fragmented network of small forest patches that together make up 7.3% of the reserve's surface area, almost twice the area that was generally known. Two afromontane forest communities are recognized and associated along an altitudinal gradient, one within the moist mist belt and one within drier micro-climates outside the mist belt; further, within each community, variants were recognized based on either available soil moisture or degree of past utilization. These forest patches host a variety of forest-dependent trees, including some species considered rare, insufficiently known, or listed under the Red Data List of South African Plants. The fragmented nature of the relatively small forest patches accentuates the need for careful fire management and stringent alien plant control.
Forest size and fragmentation
All forest patches greater than 0.25 ha in size were mapped on a Geographical Information System (GIS). Forest boundaries were marked on 1:10 000 orthophotos, and the data were digitized onto the Mpumalanga Parks Board's GIS program, SPANS . Total forest size and patch numbers are calculated from the digitized data.
Twenty-two plots (0.04 ha each) were subjectively distributed throughout two of the five largest forest patches occurring on BRCNR (Figure 4), including the Op-de-Berg and Hebronberg forests. Plots covered an altitudinal range from 1240 to 1660 m above sea level and were sampled along environmental gradients, including the factors of slope, surface rockiness, drainage, topography, and disturbance (from logging). Sample relevés included a list of all the species present in a sample plot as well as cover-abundance values for each species, according to the Braun Blanquet cover-abundance scale . A forest flora (see additional file 1 – Appendix A) is compiled from the species composition lists recorded in the 22 relevés.
Data processing and analyses
Eigenvalues produced from the TWINSPAN and DECORANA programs are a measure of the degree of separation in the data. According to Jongman, ter Braak & van Tongeren , low eigenvalues would represent a poor separation of samples and can be regarded as a measure of alpha diversity (species turnover within a plant community). High eigenvalues would represent a strong separation of samples, which can therefore be a measure of beta diversity (species turnover between plant communities).
The 22 relevés were analyzed for a circumscription of possible plant communities. Firstly, a complementary analysis was run using the hierarchical classification program TWINSPAN [22, 23] and the indirect ordination program DECORANA [24, 25]. Complementary analyses of the two different multivariate analysis results provide for an accurate interpretation and description of plant communities . Qualitative cover-abundance values were used for data input instead of quantitative values. Appendix A lists the 167 species recorded in the forest and used in the classification of plant communities. Secondly, the relevés were analyzed for relationships between plant communities and environmental gradients. Canonical correspondence analysis is a direct ordination technique that analyzes and presents such relationships between many species and numerous environmental variables. For this study, we used the program CANOCO . No formal syntaxonomical classification was done in this study. An informal classification was performed with preferential species indicating the names of different vegetation associations.
Specific species responses to environmental variables are a useful way of displaying the impact a certain environmental variable may have on a species. Direct ordinations relate species presence to environmental variables on the basis of species and environmental data from the same set of sample plots . A scatter diagram produced from the CANOCO program depicts the relationship between species and environmental variables. Only those species with a cumulative fit of more than 0.35 are displayed. This ensures that only those species that most positively contributed to the ordination scores in the scatter plot are displayed. Canopy density is used as a surrogate for measuring disturbance, or forest dynamics.
- Geldenhuys CJ: Southern African Forests: their biogeography, conservation and utilization. MAB/UNESCO Regional seminar on forests. 1996, Sri LankaGoogle Scholar
- White F: The vegetation of Africa: a descriptive memoir to accompany the UNESCO/AETFAT/UNSO map of Africa. Natural Resources Research 20. 1983, Paris, UNESCOGoogle Scholar
- Lubke R, McKenzie B: Afromontane forest. Vegetation of South Africa, Lesotho and Swaziland. Edited by: Lowe A B and Rebelo A G. 1996, Pretoria, Department of Environmental Affairs & TourismGoogle Scholar
- Geldenhuys CJ, Macdevette DR: Conservation status of coastal and montane evergreen forest. Biotic diversity in southern Africa: concepts and conservation. Edited by: Huntley BJ. 1989, Cape Town, Oxford University PressGoogle Scholar
- Geldenhuys CJ: Requirements for improved and sustainable use of forest biodiversity: examples of multiple use of forests in South Africa. Forum: Biodiversity--Treasures in the world's forests. 1998, GermanyGoogle Scholar
- Von Maltitz G, Mucina L, Geldenhuys CJ, Lawes M, Eeley H, Adie H, Vink D, Fleming G, Bailey C: Classification system for South African indigenous forests: an objective classification for the Department of Water Affairs and Forestry. Environmentek Report, ENV-P-C 2003-17. 2003, Pretoria, CSIRGoogle Scholar
- Lötter MC, Emery AJ, Williamson SD: Forests. Determining the conservation value of land in Mpumalanga. Edited by: Emery AJ, Lötter MC and Williamson SD. 2002, Nelspruit, South Africa, Mpumalanga Parks BoardGoogle Scholar
- Cooper KH: The conservation status of indigenous forests in Transvaal, Natal and OFS, South Africa. 1985, Durban, Wildlife Society of Southern AfricaGoogle Scholar
- Viljoen MJ, Reimold WU: An Introduction to South Africa's Geological and Mining Heritage. 1999, Randburg, South Africa, MintekGoogle Scholar
- Geological Survey of South Africa: Geology Map. 2430 Pilgrim's Rest. 1:250 000. Geological Series. 1986, Pretoria, Government PrinterGoogle Scholar
- Knight RS, Manry DE: Lightning density and burning frequency in South African vegetation. Vegetatio. 1986, 66: 67-76.Google Scholar
- Schmidt E, Lötter M, McCleland W: Trees and shrubs of Mpumalanga and Kruger National Park. 2002, Johannesburg, JacanaGoogle Scholar
- Acocks JPH: Veld types of South Africa. 1988, Pretoria, Botanical Research Institute, Dept. of Agriculture and Water Supply, 3rdGoogle Scholar
- Low AB, Rebelo AG: Vegetation of South Africa, Lesotho and Swaziland. 1996, Pretoria, Department of Environmental Affairs & TourismGoogle Scholar
- Fourie SP: The Transvaal, South Africa, threatened plants programme. Biological Conservation. 1986, 37: 23-42. 10.1016/0006-3207(86)90032-7.View ArticleGoogle Scholar
- Hilton-Taylor C: Red data list of South African plants. 1996, Pretoria, National Botanical InstituteGoogle Scholar
- Pooley Elsa: The complete field guide to the trees of Natal, Zululand and Transkei. 1994, Durban, Natal Flora Publications TrustGoogle Scholar
- Palmer E, Pitman N: Trees of southern Africa. 1972, Cape Town, A.A. BalkemaGoogle Scholar
- SPANS: SPANS GIS. 1997, Ontario, Canada, TYDAC Research Inc.Google Scholar
- Mueller-Dombois D, Ellenberg H: Aims and methods of vegetation ecology. 1974, New York, John Wiley & SonsGoogle Scholar
- Jongman RH, Ter Braak CJF, Van Tongeren FR: Data analysis in comunity and landscape ecology. 1987, Wageningen, The Netherlands, PudocGoogle Scholar
- Hill MO: TWINSPAN: a FORTRAN program for arranging multivariate data in an ordered two-way table by classification of the individuals and the attributes. 1979, New York, Section of Ecology and Systematics, Cornell UniversityGoogle Scholar
- TWINSPAN: TWINSPAN: Two-way indicator species analysis. 1994, Phar Lap Software Inc., 7.0Google Scholar
- Hill MO: DECORANA: a FORTRAN program for detrended correspondence analysis and reciprocal averaging. 1979, New York, Section of Ecology and Systematics, Cornell UniversityGoogle Scholar
- DECORANA: DECORANA: Detrended correspondence analysis. 1994, Phar Lap Software Inc., 7.0Google Scholar
- Kent Martin, Coker Paddy: Vegetation description and analysis: a practical approach. 1992, Boca Raton, Florida, USA, CRC Press, 363-Google Scholar
- CANOCO: CANOCO: Canonical correspondence analysis. 1998, Wageneningen, The Netherlands, Centre for Biometry, 4.0Google Scholar
- ter Braak Cajo J. F.: The analysis of vegetation-environment relationships by canonical correspondence analysis. Vegetatio. 1987, 69: 69-77.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.