South Africa: Restoration of Coastal Dune Forests Following Opencast Mining in Northern KwaZulu-Natal Province

The spreading of topsoil on reshaped dunes; Credit: van Aarde et al. 1998
A 14-year-old rehabilitating stand with well-developed canopy dominated by Acacia karroo (sweet thorn) and emerging secondary species; Credit: van Aarde et al. 1998


The rehabilitation program conducted by Richards Bay Minerals (RBM) of areas exposed to opencast surface mining of sand dunes north of Richards Bay (28º43’S, 32º12’E) on the coast of northern KwaZulu-Natal Province commenced in 1978 and has resulted in the development of a series of known-aged stands of vegetation. By assuming that these spatially separated stands develop along a similar pathway over time, instantaneous sampling should reveal successional changes usually associated with aging and should provide an opportunity to evaluate the success of the rehabilitation. A study conducted between 1992 and 1994 compared relative densities of pioneer and secondary species, species richness, and a similarity index of the herbaceous layer, tree, beetle, millipede, bird, and small-mammal communities of rehabilitating areas of known age with those of 30-year-old unmined forests and unmined forests of unknown age adjacent to the rehabilitating area. Similarly, a study conducted in 1995 examined soils from rehabilitating stands of varying ages and compared them to soils from two disturbed, unmined stands 30 and 58 years of age, and from a mature unmined stand to assess age-related trends in select soil properties. Both studies found that the degree of similarity between unmined stands and rehabilitating stands of different ages increased with increasing regeneration age of rehabilitating stands, suggesting that rehabilitating communities are developing towards the status of unmined communities. It is therefore concluded that RBM’s rehabilitation program is gradually resulting in the reestablishment of a coastal dune forest ecosystem.

Quick Facts

Project Location:
Richards Bay, South Africa, -28.7807276, 32.038285599999995

Geographic Region:

Country or Territory:
South Africa


Coastal, Dune & Upland

Area being restored:
total rehabilitated area not given

Organization Type:
University / Academic Institution


Project Stage:

Start Date:

End Date:

Primary Causes of Degradation

Mining & Resource Extraction, Urbanization, Transportation & Industry

Degradation Description

The coastal dune forests north of Richards Bay have a long history of human presence. Iron-age man has been present since at least 270 AD (Maggs 1976), and Bantu people practicing a slash-and-burn agriculture since the fifteenth century (Bruton et al. 1980). During the nineteenth century, the area was inhabited by the Mbonambi tribe of blacksmiths who made the stabbing spears used by the Zulu king Shaka. Aerial photographs taken during 1937 show highly degraded vegetation owing to the activities of local inhabitants. However, depopulation resulted in the recovery of much of the area through natural succession by 1974 (Weisser & Marques 1982).

In July 1977, Richards Bay Minerals began extracting heavy metals (e.g. zircon, rutile, and ilmenite) from the dunes northeast of Richards Bay via opencast mining. This venture brought renewed degradation to the dune forests, and the mining activities in the area continue at the present time.

Reference Ecosystem Description

Stands of unmined coastal dunes to the north of the present mining operation comprise patches of forest which, based on the inspection of aerial photographs, had been cleared of all vegetation some 30 – 35 years prior to the present study. Some of these patches have been afforested with Pinus elliottii (flash pine) and Eucalyptus grandis (eucalyptus), while others, for unknown reasons, were recolonized by indigenous vegetation, mainly broad-leaved trees such as Canthium inerme (common turkey-berry), Celtis Africana (white stinkwood), Draceana aletriformis (large-leaved dragon tree, dominant in the shrub layer), Clausena anisata (horsewood), Peddiea Africana (poison olive), Deinbollia oblongifolia (dune soap berry), Scutia myrtina (cat-thorn), and Teclea gerrardii (Zulu cherry-orange).

In the rehabilitating areas, early stages of regeneration are dominated by low-growing Acacia karroo shrubs, which develop into woodlands with broad-leaved tree species characteristic of the surrounding indigenous forests establishing themselves after 12 years. In such areas, A. karroo is starting to senesce, and species typical of surrounding unmined areas, such as Sideroxylon inerme (white milkwood), Celtis africana (white stinkwood), Mimusops caffra (coastal red milkwood), Vepris lanceolata (white ironwood), and Trichilia emetica (Natal mahogany) are beginning to occupy the canopy gaps and are common in the undergrowth. On these older stands, the groundcover is dominated by the herb Asystasia gangetica, with the grass Brachiaria chusqueoides also well established (Camp 1990).

Project Goals

In accordance with the policy of landowners interacting with the mining company, the rehabilitation program at Richards Bay is directed at establishing an indigenous coastal dune forest on one-third of the area exposed to mining.


The project does not have a monitoring plan.

Description of Project Activities:
Richards Bay Minerals has been extracting heavy metals (zircon, rutile, and ilmenite) from some of the dunes northeast of Richards Bay since July 1977. Here, dune mining is preceded by removal of all surface vegetation, and topsoil is collected for later use in rehabilitation. A dredger, floating on a pond maintained within the dune, sucks up the sand and separates the heavy minerals before they are pumped to a stockpile on land. The remaining sand (> 94% of the original volume) is pumped to the area behind the pond where, after "dewatering," a new dune is formed and subsequently reshaped to resemble the topography of dunes prior to mining (Camp 1990). In accordance with the policy of the landowners, the rehabilitation program at Richards Bay is directed at establishing an indigenous coastal dune forest on one-third of the area exposed to mining. After stabilization of the dune (6-12 months after the removal of vegetation), topsoil collected prior to mining is spread in a 10-15-cm layer over the reshaped dune. A seed mixture comprising Pennisetum americanum (babala grass), Sorghum sp. (sorghum), and Crotalaria juncea (sun hemp), as well as seed from several indigenous species collected prior to mining, is sown on the dune (Camp 1990). Windbreaks (1.5 m high) are erected across the dune, facing the prevailing winds. These structures reduce erosion, wind-induced surface evaporation, and conceivably transpiration from newly germinated seeds. The windbreaks are usually removed after about 3 months. These actions result in the development of a dense cover crop within a month, which may ameliorate the surface microclimate for the germination and subsequent establishment of indigenous species (Lubke et al. 1992). After 6-8 months the cover crop dies off, leaving the dunes covered by dense stands of the indigenous grass Eragrostis curvula (weeping love grass). All vegetation establishment after the initial seeding described above is from the seedbank in the topsoil and from dispersal from surrounding rehabilitating and unmined areas. A management policy of minimum interference is adhered to once the rehabilitation process has been initiated and involves the mechanical control of alien invasive plants (van Aarde et al. 1996a). At the time of the first study, conducted from 1992-1994 to assess the species composition of the rehabilitating stands, the landscape where dune mining took place included a series of regenerating areas varying in age from a few months to 16 years, a belt 100-250 m wide of unmined coastal dune forests fragmented by plantation patches, mainly Eucalyptus saligna (blue gum) and Casuarina equisetifolia (beefwood), along the coast, and a belt forested with Casuarina equisetifolia, which do not invade the rehabilitating areas, on the inland side of regenerating areas. These Casuarina belts range in age from a few months to more than 8 years. The areas inland of the mining operation are relatively densely populated and are characterized by patches of exotic and indigenous vegetation, in most cases continuously disturbed by grazing. The landscape also includes patches of sand dunes stripped of vegetation in preparation for mining, and dunes being reshaped and prepared for rehabilitation. The study was conducted on known-aged stands of rehabilitating areas ranging from one week to 16 years in age, as well as on surrounding unmined forests. Surveys were also conducted on unmined stands 30-35 years old and in a mature forest of unknown age. The analyses presented herein were restricted to data collected during the austral summers from 1992 to 1994. Species richness for various groups of plants and animals was taken as the total number of species recorded during the summer(s) when the surveys were conducted. For the purpose of this study, pioneer species were defined as those having their highest relative density in stands less than 5 years old, except for the herb layer, where stands less than 8 years were included in this category. All other species (excluding the pioneer species) recorded on the unmined stands were designated as secondary species. Relative densities for various groups of plants and animals for each of these two categories were calculated as the sum of the relative densities of all species within the group. Estimation of species-specific relative densities differed between taxa and is described below. --Herbaceous Communities-- The herbaceous layer was defined as all nonwoody vegetation less than 1 m in height. Thirty plots (1,250 m2) were located randomly in each of three different-aged stands within the rehabilitating area. Five to six 1-m2 quadrats were located at each plot and the cover frequency--the percentage of quadrats per plot in which a given species contributed to the cover (Barbour et al. 1987)--was recorded. Relative density for each species was calculated as the fraction of total frequencies for all species within a stand. --Tree Communities-- Tree community surveys were based on point-centered quarter transects (Cottam & Curtis 1956). Four to six transects were located on each of the stands and marked at 10-m intervals. The number of intervals varied according to the length of the transect and ranged between 8 and 14 intervals. Two-thirds of the marks on the transect were randomly selected to be sampled, which ensured that the same individuals were seldom represented in more than one sample. At each mark the nearest distance to a woody canopy species (trees > 2 m), and a woody undercover species (shrubs > 1 > 2 m) was recorded and the species identified. Density was calculated for each canopy and each undercover species by the method of Cottam & Curtis (1956). Total stand density was calculated by summing all species-specific densities. Species-specific relative density is species-specific density presented as a fraction of total density within a stand. --Millipede Communities-- Information required to estimate stand- and species-specific relative densities was collected on fixed-width transects (35 x 6 m) randomly located in each of six different stands as unique sampling areas. To account for potential diurnal changes in activity, only data collected between 06:00 and 11:59 were included in our analysis. During the study period of three weeks in December 1992, approx 3,200 millipedes were collected and classified into morphospecies. Specimens of each morphospecies were identified form reference material kept at the Natural Science Museum in Durban, South Africa. Stand-specific relative densities were calculated as the percentage contribution of each species to the total number of millipedes recorded on a given stand. --Beetle Communities-- Beetles (Order Coleoptera) were collected by means of pitfall, flight-intercept, and sweeping methods in stands of known age, sorted to morphospecies, and counted. Relative densities and species richness of all beetles were determined separately for each of the sampling methods because these methods could sample beetle communities differentially. Stand-specific relative densities were calculated as the percentage contribution of each species to the total number of beetles recorded on a given stand for each sampling technique. --Bird Communities-- The information required to estimate stand- and species-specific relative densities, community composition, and species richness was collected along three 425-500-m fixed-line transects placed on each of the stands included in the study, except the stand 1-< 2 years old, where one transect was used. The distance between all transects was at least 200 m, ensuring independence of data collected along each transect. Each transect was surveyed seven times per day. On days when weather conditions did not enable seven surveys, the surveys were completed on the next day. The data for each day were treated as a single replicate for that transect and were collected as described by Burnham et al. (1980). The information included species identity, group size, and perpendicular distance from the line. Stand-specific relative densities (rD) were estimated with the formula rD = rN/rV, where rN = relative number seen and rV = estimated relative visibility of the species. rV was estimated by using mean sighting distance, through a regression formula obtained from 68 species-stand combinations for which enough data were available to estimate density with the computer software DISTANCE (Laake et al. 1993). --Small-mammal Communities-- Small mammals (rodents and shrews) were captured with Sherman livetraps set on permanent trapping grids for five consecutive days and nights (a trapping session) per stand. The traps were checked at dawn, and all individuals captured were marked by toe clipping before release. Population estimates for each trapping session were obtained by the mark-recapture models of Otis et al. (1978) and converted to density by means of the area covered by the trapping grid. The species-specific estimated densities for each stand were used to calculate total stand-specific density by summing all the species-specific densities within a stand. Stand-specific relative densities for each species were then calculated as species-specific densities within a stand expressed as a fraction of stand-specific total density. --Soil Structure-- The second study, conducted between February and October 1995, was aimed at examining the soil properties of four coastal dunes undergoing between 3 and 18 years of rehabilitation following mining and at comparing these with soil properties of two adjacent unmined but disturbed forests of 30 and 58 years old, respectively, as well as with a mature forest of unknown age but at least 100 years old. Because humus, the main constituent of soil organic material (Williams 1987), consists primarily of decomposing plant material that continuously accumulates on the soil surface from which it is broken down and recycled into the top layer of soil, any changes in soil attributes taking place during succession should first be evident in the upper 10 cm. Thus, the upper 10 cm of the soil were sampled in four rehabilitating areas, two unmined but disturbed forests, and a mature forest at Richards Bay, and their soil properties compared. Soil samples (2500 - 3000 g per sampling point) were collected on three sampling occasions. Soils collected on the first two sampling occasions in February and May 1995 were used to determine soil macro characteristics (soil humus or soil organic material), whereas soils collected on the third sampling occasion in October 1995 were used to assess soil chemical characteristics (organic carbon, nitrogen, phosphorus, potassium, sodium, calcium, and magnesium). All sampling was conducted on the sea-facing slopes of the second dune from the coastline. Samples were collected on five randomly selected fixed-width transects (6 x 35 m) located during each of the sampling occasions on each of the stands. Each sample comprised six sub-samples taken at different localities on each of the transects by means of a square corer (10 x 10 x 10 cm). Percentage total organic content (%TotOrg) representing the decomposing plant material for each of the samples is expected as a percentage of the difference in a soil aliquot's weight prior to and after the burning thereof in a crucible over a bunsen burner for 30 minutes (Williams 1987). Percentage organic content representing the decomposing plant material less than 2 mm in particle size (%Org, 2 mm) and greater than 2 mm in particle size (%Org. 2 mm) was determined similarly after aliquots were passed through a 2-mm sieve. To determine pH of the soil, a 10-g sample was mixed with distilled water in a 1:2 ratio of soil (dry weight) to water and the pH of the water recorded with a pH metre. The determination of carbon (C), calcium (Ca), organic nitrogen (N), potassium (K), magnesium (Mg), sodium (Na), and phosphorus (P) concentrations in each of the samples followed the procedures suggested by Moore and Chapman (1986) and The Non-affiliated Soil Analysis Working Committee (1990). Calcium and Mg content was determined by atomic absorption spectroscopy, whereas K and Na were determined by flame emission spectroscopy. Organic carbon was determined by the Walkley-Black method. Nitrogen was determined by the Kjeldahl method of digestion by sulphuric acid. Nitrite is seldom present in detectable amounts (The Non-Affiliated Soil Analysis Working Committee 1990) and therefore was not included in the analysis. Soil fertility was assessed by determining the growth response of Raphanus sativus (radish) seeds planted in aliquots of soil from each sample taken from the upper 10 cm in the four rehabilitating stands, the 30-year disturbed stand, and the mature forest (100 years) and grown in an outdoor planthouse. One hundred aliquots of 300 g each of soil from each stand were placed in 250-ml styrofoam cups, and a randomly selected seed was planted in each of these at 1-cm soil depth, resulting in 100 replicates per stand. Each pot received the same amount of distilled water daily (80 - 100 ml) as needed to prevent drying and was exposed to the same temperature (28ºC) and sunlight. Radishes were allowed to grow for 27 days, after which they were harvested, cleaned by washing, oven-dried at 60ºC for two days, and weighed. Total plant weight, tuber weight (part of plant between first root hair and base of stem), and leaf weight (part of plant above base of stem) were recorded separately.

Ecological Outcomes Achieved

Eliminate existing threats to the ecosystem:
The rehabilitating and unmined stands are characterized by a progressive age-related increase in heterogeneity, with the earliest stages of recovery appearing as grasslands, which are soon transformed into an almost impenetrable shrub layer dominated by Acacia karroo. Stands 5-8 years old were still dominated by A. karroo, which had attained a height of 1.5-3 meters. Older stands (8-11 years) were still dominated by A. karroo up to 8 m high and exhibited a ground cover dominated by the indigenous grass Digitaria diversinervis (finger grass). The 11-16-year-old stands comprise a well-developed woodland canopy, and, although dominated by A. karroo, patches of treefalls resulting from senescence were dominated by a variety of secondary dune-forest tree species at heights up to 5 m. The 35-year-old stands appeared as stands with a relatively greater diversity than the oldest rehabilitating stands, with A. karroo less dominant in the canopy layer. The unmined forest of unknown age included in the present study was characterized by a canopy 12-15 m high comprising a variety of tree species. The canopy had several gaps resulting from tree falls. --Species Richness-- In general, for all but the mammals, the number of species increased with increasing age of rehabilitating stands. Furthermore, with the exception of the mammals and herbaceous layer, the unmined stands harbored more species per group than the mined rehabilitating stands. In the case of the trees and millipedes, increased species richness with stand age was characterized by the addition of species. In the case of the beetles and birds, however, the increase in species richness was characterized by both species addition and replacement. --Relative Densities-- For all the groups the general trend was a decrease in the relative densities of pioneer species as stand age increased. Relative densities for pioneer species for all the groups were also considerably lower on unmined stands than on the mined rehabilitating stands. This suggests that, although increasing age of stands is not always associated with species replacement, it is associated with a decrease in the relative densities of those species first to colonize areas disturbed by mining. --Similarity Indices-- Similarity between unmined stands and rehabilitating stands of different ages increased with increasing age of rehabilitating stands. This suggests that all these rehabilitating communities, in terms of species composition and relative densities, are developing towards the unmined communities. --Soil Composition-- Soil organic material, percentage organic carbon and concentrations of sodium, potassium, magnesium, calcium, and nitrogen all increased with an increase in habitat regeneration age. Concentrations of most of these elements were lower than those recorded on the 58-year-old unmined and mature unmined stands; however, multivariate analyses suggest that the similarity of these values for rehabilitating stands and unmined stands is increasing in direct proportion with regeneration age. This suggested age-related increase in soil fertility is supported by the radish growth trial, which showed an increase in plant biomass with soil age.

Factors limiting recovery of the ecosystem:
Based on the analysis of data gathered during these studies, it follows that rehabilitation is occurring and that this may result in the reestablishment of a coastal dune forest ecosystem. However, rehabilitation affected by succession depends on the availability of species sources from which colonization can take place. In the case of the Richards Bay mining operation, the layout of the mining path results in such refuges being present in the form of a relatively narrow unmined seaward strip along the mining path, as well as fragments of relatively undisturbed forests ahead of and behind the mining path. On the landward side, rehabilitating areas are bordered by intensely disturbed areas of plantations and densely inhabited areas. Due to these contextual factors, the rate at which natural succession can proceed is somewhat limited. Nitrogen levels in the soil at rehabilitating sites represent another factor influencing the rate of recovery. Despite the age-related increase in nitrogen concentrations, absolute values on mined stands were much lower than those on unmined stands and those recorded in other biomes (Tilman 1987; Stock et al. 1995). Even though the unmined stands had much higher levels of nitrogen than the rehabilitating stands, the levels of nitrogen in the unmined stands are not considered high. Low levels of nitrogen in these stands are therefore not surprising, particularly when considering that dynamic models predict that it can take 100 years for nitrogen concentrations to change from low to high on nutrient-poor soil (Tilman 1985). The relatively high growth rate of Acacia karroo, which dominates all the mined rehabilitating stands, could conceivably affect nutrient turnover (Vitousek & Walker 1989; Stock & Allsopp 1992). Like other Acacia species, A. karroo has the ability to fix nitrogen (Barnes & Fagg 1995), and thus to help enrich surface soils. Therefore, with its ability to grow at low nitrogen levels, A. karroo may be an important component driving the vegetation development in the rehabilitating stands at Richards Bay, influencing changes in soil characteristics in tandem with those tree species that are becoming established in openings within 18-year-old A. karroo woodland (van Dyk 1997).

Socio-Economic & Community Outcomes Achieved

Key Lessons Learned

The success of this rehabilitation program may be assessed in terms of structural (e.g., species richness and species diversity) and functional (e.g., cycling and fixation of energy and minerals) similarities of rehabilitating areas with those of unmined areas. Because the rehabilitation program was initiated relatively recently, differences are likely to exist between the oldest rehabilitating stand and unmined areas. The evaluation of the rehabilitation program thus has to rely on: (1) the trends of community development there being similar to those recorded on disturbed but unmined areas; or (2) the temporal trends of community variables on rehabilitating stands either forming or having a high likelihood of forming a continuum with values recorded on unmined areas–increasing similarity with unmined areas.

It is important to note that the stands sampled during these studies do not represent a complete successional sere, but that the rehabilitating stands are representative of relatively early stages (0-18 years) of community development. The unmined areas 30-35 years old and 58 years old have been selected to represent a later seral stage, and the other unmined area (Zulti North), a relatively older coastal dune forest, to represent a mature coastal dune forest (Weisser 1987). The analyses presented here show definite increasing trends in species richness for the trees, beetles, millipedes, and birds. The lack of a similar trend for rodents and shrews may be ascribed to the low species richness of these communities, and a low sampling effort of unmined areas may account for the relatively few herbaceous species found here. Other studies, using different indices of diversity and species richness, on woody plants (Lubke et al. 1992; Mentis & Ellery 1994), millipedes (van Aarde et al. 1996), bettles (Vogt 1993), ants (Majer & de Kock 1992), birds (Kritzinger 1996), and rodents (Ferreira 1993; Ferreira & van Aarde 1996) of the area showed increasing species diversity.

The definitions of pioneer and secondary species used in this study make it inevitable that a decrease in pioneers and an increase in secondary species will be observed with an increase in habitat regeneration age. These trends, however, serve as an illustration of unidirectional changes in species composition, as can occur during ecological succession. Species composition for the different stands differed considerably and resulted from the addition and/or replacement of species. It, thus, seems that ecological succession is driving the development of the communities from relatively simple (e.g., communities dominated by relatively few pioneer species) to relatively complex (e.g., communities dominated by several secondary species).

As already mentioned, the unmined forest 30-35 years old represents a later seral stage of coastal dune forest succession, while the unmined area at Zulti North depicts a “mature” coastal dune forest of unknown age (Weisser 1987). In terms of the present investigation, these two areas may be considered as controls to which the rehabilitating stands have been compared. The analyses of patterns of similarity for all the studied taxa indicate that an increase in rehabilitating stand age is associated with an increase in the similarity of that stand with the unmined stand. Thus, although the characteristics of the rehabilitating communities differ from those of the unmined stands, it appears that all the communities in rehabilitating stands are in the process of developing characteristics increasingly similar to those of communities in unmined stands.

The difference in species richness between the rehabilitating stands and the unmined stands most likely results from the absence and/or rarity of some forest specialists in rehabilitating stands. With many species typical of unmined areas already present in the oldest rehabilitating stands, however, it seems as if conditions within these areas are changing to facilitate the establishment of communities with characteristics similar to those of unmined areas, with a possible subsequent colonization by forest specialists. In the case of the vegetation, this probably results from changes in light intensity and soil characteristics. Colonization of older rehabilitating areas by plant species typical of adjacent unmined areas is facilitated by fruit-eating birds and vervet monkeys foraging in unmined as well as rehabilitating areas (Foord et al. 1994). Colonization of rehabilitating stands by animal taxa through species addition and replacement probably results from habitat requirements being fulfilled through structural development of the vegetation.

The results of the soil study indicate that nitrogen, carbon, calcium, magnesium, and potassium concentrations are all changing directionally with regeneration age, suggesting that the availability of these elements in rehabilitating stands is increasing with time. This unidirectional quantitative change in soil characteristics supports the notion that soil properties are developing on this sere of coastal dune forest and that soil fertility is increasing with age. Thus, with time following disturbance, and without further disturbance, soil properties of rehabilitating areas may approach those of surrounding unmined forests in the Richards Bay area. Indeed, the concentrations of elements other than potassium on the 58-year-old stand were very similar to those recorded on the mature stand, suggesting that full recovery of these elements may occur within a relatively short time.

In conclusion, the results obtained in these studies show that there are still differences between the oldest rehabilitating stand and the unmined coastal forest, but that the similarity between rehabilitating stands and the unmined forest is increasing with the regeneration age of the unmined stand. The criteria employed herein therefore lead to the conclusion that rehabilitation of community structure has initially been successful but is not yet complete.

Sources and Amounts of Funding

The rehabilitation program is conducted by Richards Bay Minerals as part of its standard operating procedures. Financial and logistical support for the two comparative studies described herein was provided by Richards Bay Minerals, the Foundation for Research Development, the University of Pretoria, and the Department of Trade and Industry.

Other Resources

van Aarde, R.J., A-M. Smit and A.S. Claassens. 1998. Soil characteristics of rehabilitating and unmined coastal dunes at Richards Bay, KwaZulu-Natal, South Africa. Restoration Ecology 6(1):102-110.

van Aarde, et al. 1996. An evaluation of habitat rehabilitation on coastal dune forests in northern KwaZulu-Natal, South Africa. Restoration Ecology 4(4):334-345.

Primary Contact

Organizational Contact