Floristic composition of gold and uranium tailings dams , and adja cent polluted areas , on South Africa ’ s deep-level mines

Gold and uranium tailings ( ‘slimes’) dams and the adjacent polluted soils in the deep-level mining regions o f South Africa (Carletonville, Klerksdorp and Welkom) were surveyed for the frequency o f occurrence o f naturally colonizing, actively introduced and persisting plant species. Fifty-six tailings dams with a combined area o f 5864 ha. and a similar area o f tailings-polluted soils, were surveyed between July 1996 and March 1997. During the survey, 376 plant species and subspecies were recorded from the dams and adjacent polluted soils, with an additional 8 6 records obtained between 1998 and 2003 (i.e. a total o f 462 taxa: species and infraspecific species). Overall, the most commonly represented families were the Poaceae (107 species and subspecies), Asteraceae (81). Fabaceae (55) and Anacardiaceae (16). with other families represented by just one to 14 species. Only 60 species were common to all three regions, and o f these 24 had been introduced during rehabilitation attempts. Most o f the species found on tailings were persisters or natural colonizers (53-88%, depending on substrate), with the vast majority being indigenous and perennial taxa (76% and 85% respectively), with semi-woody to woody growth forms (6 6 % being resprouters, forbs, shrubs and trees). Less than 4% o f the naturally-colonizing taxa found during the survey had also been introduced by vegetation practitioners. The majority o f introduced plants were alien herbaceous taxa. The number and frequency o f annuals was only high on recently vegetated sites, whereas annuals were rarely present on old-vegetated and never-vegetated dams. This list includes a wide range o f indigenous plant species that may be suitable for phytoremediation o f tailings dams and polluted soils due to their apparent tolerance o f acid mine drainage and salinity.


INTRODUCTION
Tailings storage facilities (TSF) containing waste rock or milled rock slurry from the gold and uraniummining industry', cover vast areas in South Africa.Gold TSF and 'footprints' (the area o f contaminated soil and residual slimes left behind after re-mining o f the original TSF) cover about 400 km: in the Witwatersrand Basin goldfields alone, comprising about 6 billion tonnes of gold and uranium tailings (Chevrel et al. 2003) Environmental degradation from gold TSF spreads far beyond the w aste deposit sites in the form o f air pollution (Van As et al. 1992;Mizelle et al. 1995), soil pollution (Coetzee 1995;Rosner & van Schalkwyk 2000;Rosner et al. 2001;Witkowski & Weiersbye 1998a) and pollu tion o f streams, rivers, dams and sediments (Funke 1990;Pulles 1992;Hodgson et al. 2001;Naiker et al. 2003;Tutu et al. 2003;Coetzee et al. 2004;Winde et al. 2004a, b, c).Since the TSF for slurry (referred to as slimes dams) are elevated above the natural ground contours and have steep slope angles, they are particularly sus ceptible to erosion (Mizelle et al. 1995;1996).Whereas erosion from agricultural fields may be as high as 10 to 15 tons ha 1 y e a r1, losses from the slopes o f gold slimes dams may exceed 500 tons ha-1 y e a r1 (Blight 1991).Erosion and acid mine drainage from gold slimes dams have severe impacts on nutrient cycling in polluted soils (Witkowski & Weiersbye 1998a).on the regeneration of vegetation (Witkowski & Weiersbye 1998b;Weiersbye & Witkowski 2003) and on the biogeochemical cycling of potentially toxic elements (Weiersbye et al. 1999;Winde et al. 2004a. b;W eiersbye & Cukrowska 2005).
Prior to 1991.South Africa had little legislation spe cifically directed towards environmental protection from mining impacts, although recommendations and statutes existed for the structure and abandonment o f tailings dams (James 1964;James & Mrost 1965; Chamber o f Mines o f South Africa 1968Africa . 1979;;Blight 1969).Mines did not have a legal obligation to prevent dust pollution until the promulgation o f the Atmospheric Pollution Prevention Act 45 (1965), amended in 1973.The Chamber o f Mines Guidelines (1979) recommend that wind and w ater erosion o f dams be controlled by the most practical means possible using the BATNEEC (Best Available Technology Not Entailing Excessive Cost) concept.Erosion control included covering the surface of tailings dams w ith waste rock, or vegetating the tailings; the latter ('grassing') is still considered the most effec tive means o f reducing dust by the industry'.This accept ability is based on the speed w ith w hich the grass cover establishes, rather than on its long-term persistence or effective erosion control.The earliest recorded attempts at rehabilitation (dust control) o f gold tailings dams on the Witw atersrand occurred in 1894.w ith the planting of Ammophila sp.(seed from Kew Gardens.UK), and have been follow ed by a series o f vegetation trials between 1932 and the present day (Thatcher 1979; Weiersbye & Bothalia 36,1 (2006) Witkowski 1998).Slimes dams are inhospitable envi ronments for plant growth, and various combinations o f leaching, liming, fertilization and irrigation are used to facilitate the growth o f a small suite o f herbaceous, mostly pasture species.As predicted by Halliday (1978), the grass cover achieved is temporary and the methods have proven economically and ecologically unsustain able (Thatcher 1979;Weiersbye & Witkowski 1998;Witkowski & Weiersbye 1998a, b).The cost o f grassing the steep (30°-45°) slopes o f tailings dams ranged from R70 000 to R160 000 ha' 1 in the period 1994 to 2002, whereas the cost o f grassing the flat tops o f dams and dam 'footprints' (polluted areas left after the dam itself has been removed for re-processing to recover residual gold) was between R10 000 and R60 000 ha' 1 during the same period (Weiersbye & Witkowski 2002a;Weiersbye et al. 2002).Although grassing and irrigation o f tailings dam slopes significantly abates wind-borne erosion in the short term (Blight 1991), long-term erosion control and containment o f water pollution from gold TSF by grass ing has been unsuccessful (Blight 1991(Blight , 1998;;Weiersbye & Witkowski 1998;Rosner et al. 2001).
South Africa now has some o f the most stringent environmental legislation in the world, with the right The aim o f this study was to undertake a broad-scale survey o f the plant species composition on gold and uranium slimes dams, and slimes-polluted soils, in the deep-level mining regions o f South Africa.The under lying rationale was to assess the feasibility o f a more sustainable ecological engineering and phytoremediation approach to slimes dam rehabilitation, through identify ing a greater suite o f suitable species.The dams surveyed ranged in age (from 9 to 58 years since commissioning), in planted vegetation status (present or absent) and in the time elapsed since planting o f vegetation (from 3 to ± 50 years).The survey objectives were: (i) to provide a systematic list o f indigenous and alien plant species found on slimes dams, and slimes-polluted soils; (ii) to distinguish natural colonizers and persisters (i.e.individ uals present prior to slimes deposition but still surviving despite the new conditions) from species intentionally introduced during vegetating attempts; (iii) to broadly classify species according to functional groups; (iv) to assess the number o f plant species and their frequency on tailings dams differing in vegetation history at each min ing locality; and (v) to assess the number o f plant species and their frequency on each o f the different substrates that together constitute a tailings dam.These substrates were identified in a parallel study and are character ized mainly by differences in slope angle, elevation (i.e.time since slurry deposition), texture, soil organic mat ter, water content, conductivity, pH and redox potential (Witkowski & Weiersbye 1998a).

Vegetation, soils and climate
The study was carried out at Anglo American Ltd (subsequently AngloGold, and FreeGold) mines in the Gauteng, North-West and Free State Provinces.Tailings dams situated around Carletonville (Gauteng), Klerksdorp (North-West) and Welkom (Free State) were surveyed for plant species composition.The survey covered ± 12 000 ha, situated within an overall area o f 150 x 100 km over the Upper Witwatersrand Basin, on the West Wits, Vaal and Welkom Reefs (Figure 1).All tailings dams at the West Wits and Elandsrand gold mines (Gauteng), at the Vaal River and Afrikaander gold mines (North-West), and at the Free State Gold, Freddies, Western Holdings, President Brand, President Steyn, Free State Saaiplaas and Free Gold mines (Free State) were included in the survey.Most of the surveyed tailings dams occur within the Grassland Biome (Acocks 1988;Rutherford & Westfall 1994;O 'Connor & Bredenkamp 1997), with one dam in the east occurring in the transition zone between grassland and savanna (Afrikaander Leases in the North-West).The Vaal River and Afrikaander dams are situated on doleritic and sandy soils within the A2 vegetation subdivision (O 'Connor & Bredenkamp 1997) o f the Grassland Biome at an altitude of 1 300 to 1 350 m.The Free State dams occur on clayey to sandy soils within the A2 and B3 subdivisions at 1 300 to 1 400 m.The Carletonville dams are constructed on rocky quartzite, shale and dolomitic soils in the C6+7 subdivisions at 1 600 to 1 650 m.Tailings dams in the Carletonville region are situated within various combinations o f bankenveld, xeric grassland (klipveld) and Acacia karroo savanna (Acocks 1988).The main veld type in the Vaal River and Afrikaander mine areas is a combination of dry transitional Cymbopogon-Themeda veld, with some development o f a mixed grassy false Karoo veld, and dry Cymbopogon-Themeda veld.However, most of the surveyed dams here were surrounded by xeric grassland (klipveld) and Acacia karroo savanna (Bredenkamp & Brown 1995a, b).In the Welkom mine lease area, most dams are surrounded by mesic to seasonally inundated hydromorphic grassland on clays, endorheic saline pans supporting halophytic grasses, sedges and A triplex spe cies, and perennial swamps dominated by Phragmites australis and Tamarix spp.Two tailings dams near Carletonville.three near Klerksdorp, and most o f the Welkom dams are situated on pans, vleis or streams.
The climate o f the region surveyed is highly seasonal and falls within the Austral summer rainfall belt (Schulze 1997).Mean annual precipitation is 662.630 and 604 mm for Carletonville.Klerksdorp and Welkom respec tively, with high inter-annual variability (25-30% ).The regions all experience seasonal extremes o f temperature.Mean daily minima (July) and maxima (January) were 0-2 °C and 25-27.5 °C respectively for Carletonville and Klerksdorp.and < 0 °C and 27.5-30 °C respec tively for Welkom during the study period.Evaporation is 2-2 .5xhigher than rainfall, frosts occur frequently in winter (mean frost days is 150-175) (Schulze 1997), and frequent veld fires occur in winter.Regional land use includes cattle and game fanning (rangelands), maize and sunflower cropping.Wastelands (derelict, degraded lands with little plant cover) and swampy lands inundated by seepage from slimes dams are common within these landscapes.

Tailings dam construction and composition
The dams were constructed using the paddock system, which involves construction o f peripheral slimes dykes during the day ( "day w alls'), and filling o f the central dam ('night pan') with slimes slurry during the night (Mcphail & Wagner 1987).Excess water is drained away during construction and the construction rate o f the dam is limited by the drying rate o f the day walls.Moisture content is high on the tops and upper slopes o f current dams due to the deposition o f fresh slurry, and decreases with distance down the slope.Moisture content increases sharply again at the base o f the lower slope and in the toepaddock due to seepage from the dam.The slopes o f the dams surveyed ranged between 29° and 35°; these steep slopes result in high erosive losses (Blight 1991).
The tailings are derived from gold and uranium-bear ing conglomerates associated with the sediments o f the Witwatersrand Basin up to 4 000 m below surface.Pvrite is the dominant sulphide in the conglomerates, up to 3% o f the ore mass, with an additional 2% o f other sulphides namely, pyrrhotite.galena, cobaltite, arsenopvrite and chalcopyrite (Anhaesser 1987).The "all sliming' process was introduced in 1921.Slimes particles are cohesionless, predominantly silicaceous and o f the size range associated with clays and silts (Clausen 1973); the depo sition o f such fine particles in an aqueous slurry results in dense compaction and poor aeration in the rooting zone.However, the reactive clay content is negligible, with slimes consisting primarily o f unreactive quartz and pvrophyllite.The virtual absence o f organic matter con tributes further to the negligible cation exchange capac ity.The chemical composition o f the slimes surveyed varied according to the parent substrate, the metallurgical recovery ('slim ing') process used, the composition ot the mined ore and the age o f the deposits (Bosch 1987;Witkowski & Weiersbye 1998a).The main geochemi cal divisions o f slimes deposits in this study occurred between high (up to 5%) and low (< 1%) pyrite dams.Although slimes produced using older processing tech niques have high sulphur contents, slimes derived from the more recent 'Acid Plant' process are lower in sulphur (Bosch 1987).Freshly deposited slimes are grey in col our, saline, moist and alkaline (up to pH 10.0) due to the addition o f liming agents during processing.On exposure to air and water, the oxidation o f pyrite results in the production o f sulphuric acid and ferric hydroxide, with the tailings substrate consequently becoming acidic and yellow.As the substrate acidifies, ferric iron also con tributes to oxidation, and sulphur-utilizing Thiobacillus bacteria that occur in the tailings facilitate further ferrous oxidation (James & Mrost 1965;Bradshaw & Chadwick 1980).On the slopes o f current dams there is conse quently a steep pH and acidity gradient between the top (comprising recent, alkaline deposits) and the base com prising older deposits o f increasing acidity (Witkowski & Weiersbye 1998a).

Site classification
The tailings dams varied in age (from 9 to 58 years since commissioning), in planted vegetation status (veg etated or never vegetated) and in the time elapsed since planting o f vegetation (from 3 to ± 50 years ago).The tailings dams were grouped according to region (Carletonville, Klerksdorp, Welkom), and slopes were classed according to their vegetation history.Vegetation history classes were: (i) recently-vegetated (RV) slopes: amelioration (liming, fertilizing, seeding and irrigation) had ceased >1 to < 4 years previous to 1996; (ii) oldvegetated (OV) slopes: amelioration had ceased > 4 < 50 years previously; and (iii) never-vegetated (NV) slopes: slopes on record as never having been intentionally ameliorated or vegetated.Vegetation records (dates and duration o f planting, method and lists o f species used) were obtained from vegetation contracts stored with the individual mines and from vegetation contractors.Most dams have had some form o f vegetating attempted during the last 50 years.Distinguishing very OV slopes (> 20 years ago) from NV slopes was made difficult both by poor record-keeping prior to the 1980s, and the rapid reversal o f vegetated slopes to eroded, seemingly never-vegetated conditions.This necessitated a forensic approach to determining whether or not a dam had previ ously been vegetated.Previously vegetated areas were identified from old photographs (including aerial survey), and the remains o f old plantings, irrigation pipes and chemical signatures in the slimes as a result o f liming and intensive fertilization (Witkowski & Weiersbye 1998a).All the dams except NV had received similar liming and fertilization regimes.In most cases vegetation had been established using intensive irrigation, and in seven cases, vegetation had been established using dry-land methods.In many cases dam slopes had been subjected to repeated grassing attempts over the years, as each attempt had failed.
Each slope was further subdivided into substrate classes based upon marked differences in physical and chemical properties (Figure 2).From the top of the dam downwards, substrates comprised the flat tops, upper-tomid slopes and berms, mid-to-lower slopes and berms, retaining walls (rock and soil mixed or overlaid with slimes) and toepaddocks (a strip of veld from 20 to 60 m wide surrounding the base o f the dam and bordered by an earthen wall).The toepaddocks are heavily inundated by slimes, strongly acidic and often damp (Witkowski & Weiersbye 1998a).The tops, berms and slopes were further categorized according to whether they comprised younger or recently ameliorated and marginally acidic to alkaline (pH > 6.0) slimes deposits, or older and more acidic (pH < 5.9) slimes deposits.

Sampling methods
Lists o f species planted on each tailings dam, and planting methods, were obtained from vegetation con tracts and assessments archived with Anglo American Mines and from individual contractors, from unpublished theses and reports (including those lodged with indi vidual mines and the Chamber of Mines of South Africa; Thatcher 1979 (and references therein); Wiegenhagen 1996) and from publications (James & Mrost 1965;Wild & Wiltshire 1971;Cresswell 1973;Grove 1974;Clausen 1976;Bradshaw & Chadwick 1980).Two repli cate surveys were carried out on the same dams within a nine-month period, in winter (June to September, 1996;Witkowski & Weiersbye 1996) and subsequently in summer (December 1996to March 1997;Weiersbye & Witkowski 1997).Fifty-six tailings dams comprising 738 different slopes with the same number o f toepaddocks were assessed.Intensive searches were carried out on dams, retaining walls and toepaddocks.These comprised (i) large-scale assessments o f vegetation cover (data not shown, Weiersbye & Witkowski 1998), vegetation struc ture (proportions o f trees and shrubs, forbs and herbs, and grasses) and species presence on all 738 slopes, retaining walls and toepaddocks; and (ii) 254 one hundred metre wide belt transect surveys on a sub-suite o f dams (those which had > 0.5% aerial vegetation cover).Data from these 100 m belt transects were then subdivided for each substrate on the slimes dam (upper, middle and lower slopes, tops, berms and toepaddocks) as these invariably differed in species composition.All species present on transects were identified and the number o f individuals present for most species recorded.Additional records o f species presence only on slimes dams, retaining walls and toepaddocks were obtained from the same mines during 1998 to 2003.
Representative specimens o f each taxon were col lected and pressed.Plants were identified using the keys o f Dyer (1975,1976), Gibbs Russell et al. (1985, 1987, 1991), Venter & Joubert (1985), Coates-Palgrave (1996) and Retief & Herman (1997), and by comparison with specimens at the C.E. Moss and National Herbariums (PRE) o f South Africa.Taxa were named according to Arnold & De Wet (1993).A few taxa (reputedly introduced from Namibia and the Northern Cape) defied identi fication beyond genus.Voucher specimens have been lodged with the C.E. Moss Herbarium, University o f the Witwatersrand.

Data classification
Species, subspecies and varieties found in the 100 m transect survey (n = 327) were categorized according to their % frequency o f occurrence: (i) overall, (ii) in each o f the three regions, (iii) on dams o f different vegetation history classes, and (iv) on each substrate class.The rela tive contribution o f each family (in terms o f component species) was also calculated for each slimes dam sub strate class.
Using the 1998 PRECIS database and vegetation distribution records (Arnold & De Wet 1993;Retief & Herman 1997).all species, subspecies and varieties found during the 1996-1997 survey (n = 376) were cat egorized for each dam substrate and overall according to: (i) whether they were indigenous to South Africa or alien (including naturalized species), and (ii) whether they were indigenous to each o f the three regions surveyed.Species were also grouped on the basis of: (iii) annual or perennial habit; (iv) broad growth habit (shrubs and trees, forbs and perennial herbs, annual herbs, annual and per ennial grasses); and (v) whether they were persisters or naturally colonizing species, or intentionally planted on dams, or combinations.Persisters were generally consid ered to be plants that pre-dated slimes dam construction.These were categorized as old woody plants present on toepaddocks and/or growing through retaining walls, but usually not present on other substrates.Naturally coloniz ing species were those present on other substrates (berms and/or slopes and/or tops), either solely or in addition to being present on toepaddocks and/or retaining walls.All non-woody species present on retaining walls were cat egorized as natural colonizers.Species found subsequent to the survey in 1998 to 2003 (n = 8 6 ) were not included in the frequency analysis or categorization.

Broad species-compositional patterns
A total o f 376 species, subspecies and varieties were recorded during the intensive winter (1996) and summer (1997) surv eys, o f which the frequency o f 327 species and subspecies was recorded using the detailed 100 m transects (Appendix 1).The other (49) species and subspecies are listed in Appendix 2. However, all 376 taxa were included in the general analyses (Tables 1-5).Thirty-six taxa could only be identified to genus level.The number o f species on dams in the three provinces was in the order o f Klerksdorp (216 species in 1 488 ha), Carletonville (168 in 765 ha).Welkom (120 in 3 611 ha) (Appendix 1).The highest number o f species was found on the 100 old-vegetated and 139 never-vegetated slopes (260 and 231 species respectively).The 15 recently-vege tated slopes contained 86 species, o f which 2 1 had been intentionally introduced during grassing, with another 18 weedy annuals and short-lived perennials.
Most species occurred on toepaddocks and retaining walls (287 and 264 species respectively, o f which 246 and 231 respectively were natural colonizers and per sisters), followed by the acidic slopes (149 species of which 106 were natural colonizers) and acidic tops and berms (137 species o f which 103 were natural coloniz ers).Species composition differed markedly between substrates (Appendix 1).Only 32 species (14 o f which had been introduced) were common to more than five substrates, and 161 (16 o f which had been introduced) occurred on just one or two substrates.When considering species common to tops and berms and slopes, only six species were common to both acidic (pH < 6.0) and mar ginally acidic to alkaline substrates (pH > 6.0), whereas for tops and berms only, 12 species were common to both acidic and marginally acidic to alkaline substrates.For slopes, 28 species were common to both acidic and mar ginally acidic to alkaline substrates.Sixty species were common to flat and sloped acidic substrates, and just eight were common to flat and sloped marginally acidic to alkaline substrates.O f the 376 taxa recorded during the survey, only 60 were common to all three mining regions, and o f these.24 were introduced during vegetat ing attempts (Appendix 1).Only 10% o f taxa overall (including < 4% o f those which were natural colonizers and persisters) are known to have been introduced during vegetating attempts (Table 1; Appendix 1).
In addition to the 376 taxa.86 taxa (species, subspe cies and varieties) were recorded from 1998 to 2003 (Appendix 3).but were not included in the general analyses (Tables 1-4) or frequency survey (Appendix \ ).O f the 86 taxa listed in Appendix 3. 40 were introduced species (three were indigenous trees), and 46 natural colonizers and persisters (39 were indigenous).Thus a grand total o f 462 taxa (species, subspecies and varieties) have been identified on gold mine tailings and tailingspolluted soils in the three regions.

Indigenous versus alien species
The survey yielded a total o f 90 alien species (includ ing naturalized species) and 286 species indigenous to the southern African region.Fifty-five alien and 152 indig enous species occurred on tailings dams (slopes, berms and tops), whereas 59 alien and 143 indigenous species occurred on retaining walls and toepaddocks (Appendix 1).Eight alien and 14 indigenous species (of which four had been introduced to dams from other regions) occurred only on dams, and not on polluted soils.Overall, the majority o f species growing on slimes (76.1%), and the vast majority o f natural colonizers and persisters were indigenous to southern Africa (Table 3), with most (91%) normally found in the local province (Table 4).With the exception o f marginally acidic to alkaline slopes (where numbers o f indigenous species only slightly exceeded those o f alien species), the same pattern prevailed on all slime substrates.

Species characteristic o f particular substrates
The number o f naturally colonizing and persisting spe cies was higher on acidic (pH < 6.0) substrates, in order o f abundance: retaining walls and toepaddocks > slopes and tops and berms > marginally acidic to alkaline (pH > 6.0) substrates (Table 1).Although species number was slightly higher on slopes than on flat substrates (tops & berms), overall vegetation cover was always much higher on flat surfaces (Weiersbye & Witkowski 1998).On marginally acidic to alkaline substrates, a relatively high proportion of introduced taxa were also natural colonizers (28-30% , Table 1).However, once the substrate became more acidic (pH < 6.0), this proportion decreased to 10-23%.Few species (8-11% ) on retaining walls and toepaddocks had been introduced.These two substrates had the highest levels o f vegetation cover and number of species, dominated by indigenous, naturally colonizing and persisting perennial taxa (Appendix 1).

Ecological traits o f species
The vegetation of tailings was dominated by peren nial species (Table 5), with most also characterized by a deciduous habit.The majority of natural colonizers and persisters were perennial plants, whereas the majority of introduced species were annual and short-lived perennials (Table 5, Appendices 1, 2 and 3).Very few indigenous, perennial species had been intentionally introduced to    slimes.Overall.84.6% o f species were perennials versus 15.4% for annuals and short-lived perennials.The vast majority o f species found on slimes overall had semiwoody to woody growth forms: perennial forbs and herbs (4 7 .6%),followed by perennial grasses (18.6%), and shrubs and trees (18.4%), with annual herbs (11.2%) and annual grasses (4.3%) forming minor components (Table 5).A similar pattern was seen on each substrate.The relatively high contribution o f shrubs and trees to the persisting vegetation o f marginally acidic to alkaline tops and berms (± 30%) was due to the low height (depth o f slimes) o f three current (recently commissioned) dams included in the survey.These dams contained large live trees that were rooted in the underlying soil and had sur vived tailings dam construction and inundation by slimes for a number o f years.Islands o f fertility had formed on the slimes under the canopies o f these trees, and contained a number o f herbaceous species that were not found on this substrate under any other conditions.Trees and shrubs were often abundant on the lowest reaches ot dams (base o f lower slope and retaining wall).The large size and/or morphology o f many o f these plants sug gests that they pre-date dam construction, and had grown through the slimes.In contrast, trees and shrubs on the slopes, berms and tops appear to be rooting only within the slimes.
The naturally colonizing and/or persisting taxa com prised mostly woody and semi-woody growth forms, whereas the majority o f introduced taxa were herbaceous forms (Appendix 1).The majority o f naturally colonizing grasses were C4 species, while virtually all the introduced grass taxa were pasture species comprising a mixture of C\ and C4 taxa (Appendix 1; Gibbs Russell et al. 1991).The introduced species mirror the commercial availabil ity of grass seed in South Africa.Most o f these species are intended for intensively managed pasture cultivation on agricultural lands, and not for the rehabilitation of low nutrient, saline and acidic tailings dams.In a previous sur vey o f slimes dam tops in Johannesburg, Thatcher (1979) recorded a total o f 142 species, 94 of which were also recorded in this survey.Thatcher (1979) also tound that the majority of species were natural colonizers and the Asteraceae, Fabaceae and Poaceae were well represented.Woody species diversity was higher on more acidic sites, whereas grasses dominated on less acidic sites.
Few of the 376 species, subspecies and varieties found in the 1996-1997 intensive survey are known to have been intentionally introduced to slimes dams during vegetating attempts over the last 50 years.Some species were introduced to slimes dams in the surveyed region by the Chamber of Mines Vegetation Unit in the 1950s (B.Cook, B. Dawson, J. Easton, pers.comm.)However, it is not known whether existing conspecifics are remnants of the introduced populations, or individuals that have natu rally colonized tailings.Some species exhibiting unusual regional distributions may be remnants o f these attempts by the Chamber o f Mines Vegetation Unit (e.g.Bassia salsoloides colonizing old slimes dams in the Welkom region), whereas other species were collected in remote regions and introduced to tailings dams by mine person nel (e.g.Ruschia spp. in the Welkom region).

DISCUSSION
This survey found a surprisingly high number ot plant species growing on tailings and tailings-polluted soils.Earlier (pre-1980) attempts at vegetating slimes utilized a number o f woody, alien species such as Australian acacias (wattles), eucalypts and tamarix in addition to herbaceous legumes and pasture grasses (Thatcher 1979), whereas more recent attempts utilized herbaceous (pasture) species and cultivars.On the basis of old pho tographs, many of these planted trees still survive on the slopes and tops o f tailings dams.Although the contribu tion of natural colonizers and persisters to cover could be substantial on the flatter surfaces of dams (tops, berms, toepaddocks).the contribution to cover on slopes was extremely low and individual plants were transient due to the high rates of erosion (Weiersbye & Witkowski 1998).The lower number of species found on marginally acidic to alkaline (pH > 6.0) tops, berms and slopes could be due to some influence o f pH.but also to the much younger   1 4 0 )  2 .3 (6 )  14 4 (38)  48 0 (1 3 8 )  3.1 (9)  17.0(49)  4 7 .6 (1 7 9  age o f these substrates in comparison to acidic (pH < 6.0; i.e. older and more oxidized) substrates, the low number o f species introduced to recently ameliorated slopes by contractors, and the fact that many dams were current dams, with slurry still being deposited on the top.As a consequence, the tops o f current dams only supported reeds and sedges {Phragmites sp. and Cyperaceae), if any vegetation.Marginally acidic to alkaline substrates are also usually further and higher (in altitude) removed from seed sources (i.e.surrounding veld) than the older, acidic substrates on the lower slopes o f dams.In addition, vegetating efforts were seldom undertaken on slimes o f more recent genesis as the Chamber o f Mines vegetation guidelines (1979) recommended a dormant period for slimes dams prior to leaching and grassing.
According to species distributional databases, the pat tern o f species number found on slimes dams is similar to that for the provinces as a whole, with the North-West and Gauteng having the highest number o f species, and the Free State the lowest (Arnold & De Wet 1993;Retief & Herman 1997).At the local scale, many Free State dams occur within a degraded agricultural and semi-industrial setting, with the only natural seed source emanating from wastelands, hydromorphic grasslands, perennial swamps dominated by Phragmites australis, and alkaline pans.This landscape context limits the diversity and availabil ity of natural colonizers.Most dams in the North-West and Gauteng region were in close proximity to natural, albeit degraded veld and private (mine) nature reserves, and this environmental setting provides a wider diversity o f suitable species for natural colonization.
O f the species identified, < 5% had been actively intro duced during grassing.Despite the high species diversity o f natural colonizers and persisters, most individuals were detected on toepaddocks, retaining walls and on the flatter surfaces o f the dams (berms and tops), with the actual contribution to cover on slopes being extremely low.In contrast, the number o f species introduced during vegetating efforts was extremely low.despite the high cover achieved on recently grassed dam slopes.However, most introduced species are herbaceous and weedy, and both cover and number o f species declines rapidly once liming, fertilization and supplemental watering has ceased.Watering occurs either in the form o f irrigation on dormant dams (slimes are no longer deposited) or water from slurry deposition on current dams.Less than five species remained on any particular tailings slope by four years after amelioration had ceased (Weiersbye & Witkowski 1998).Although we recorded an additional 14 introduced pasture species in 1998-2003 from the same slimes dam slopes surveyed in 1996-1997, these plants occurred on dams that were undergoing grassing.These same species had not been detected in the previous survey on post-amelioration grassed sites (despite having been originally planted), which suggests that they lack persistence.
Although introduced species on dams were predomi nantly pasture grasses, the naturally colonizing species were predominantly perennials with woody and semiwoody growth forms.In the case o f tailings dams in the Carletonville and Klerksdorp regions, the predominance o f indigenous, woody growth forms as natural colonizers and persisters is expected as these growth forms are com mon in the localities surrounding the dams (Arnold & De Wet 1993;Bredenkamp & Brown 1995a.b;Retief & Herman 1997).However, tailings dams in the Free State are surrounded by hydromorphic grasslands, degraded wetlands and alkaline pans, with woody and semi-woody plants largely restricted to low, dry rocky outcrops (Fuls et al. 1992(Fuls et al. , 1993;;Malan et al. 1998).The colonization o f tailings in this latter region by semi-woody/woodv species despite their restricted availability strongly sug gests that these are suitable growth forms for tailings dam rehabilitation.
The dominance o f Poaceae, Asteraceae, Fabaceae and Anacardiaceae on acid slimes suggests an inherent toler ance to the prevailing conditions in species o f these fami lies.These results are further reinforced by seed biology studies (Witkowski & Weiersbye 1998b;Weiersbye & Witkowski 2002b, 2003) and by plant growth and wateruse trials in acid slimes and AMD conditions in which hard-seeded legumes and Rhus spp.perform especially well (Weiersbye et al. 1998;Dharamraj et al. 1999;Dye et al. 2005).More recently, the use of AFLP analysis has demonstrated that there is genetic evidence for local adaptation o f some woody species to the slimes-polluted soils around tailings dams (Angus 2005).Some species examined during this surv ey show ed no signs o f physio logical stress despite growth on slimes, e.g.Tamarix spp., Acacia spp.. Lessertia spp.(= Sutherlandia spp.), Rhus spp.. Asparagus spp.and perennial Eragrostis spp.Seed production and seed viability levels in these taxa approaches that o f conspecifics grow ing in unpolluted veld, and seed production and viability in Asparagus spp., woody legumes and Rhus spp. on tailings is high, with seedlings establishing around parent plants.In con trast, seed production and viability in most grasses and Asteraceae growing on tailings is low' and regeneration on tailings would therefore be dependent on seed dis persal from beyond the dam (Witkowski & Weiersbye 1998b).A parallel survey found that the majority of plants persisting on tailings were infected by arbuscular mycorrhizal (AM) fungi (Straker et al. 2006a. b).In addi tion.plant growth experiments have demonstrated that slimes-tolerant AM fungi, and.for most indigenous hardseeded legumes, compatible tolerant rhizobia.contribute significantly to host plant survival and growth in acidic slimes (Weiersbye et al. 1998;Straker et al. 2006c).
The dominant plant functional growth forms (i.e.w oody and semi-w oody perennials, resprouters) o f slimes dams are typical of stressful environments (Grime 1979), w hereas many o f the grass and forb species are character istic o f nutrient-poor (especially nitrogen), low competi tion environments in the Grassland Biome (O 'C onnor & Bredenkamp 1997).The distribution o f species on par ticular tailings substrates appears to be associated with known physiological tolerances to moisture and nutrient availability regimes.For example, species o f Eragrostis and Sporobolus that w ere prev alent on the and substrates typical o f old v egetated and never vegetated tailings can tolerate dehydration of foliage to the point o f air dryness (G aff 1971;G aff & Ellis 1974) (Roux 1969), and these species were abundant only on the low nutrient substrates typical o f the oldest vegetated and never vegetated sites.Populations o f S. vulgaris and similar species present on never vegetated sites that were subsequently grassed and fertilized have died out.
Tailings solution extracts have extremely high con ductivity and salinity (Witkowski & Weiersbye 1998a).Halophytic plants such as Tamarix spp.and A triplex spp.naturally colonize, and may form dense cover, on the moist and marginally acidic to alkaline slopes and tops o f tailings dams.However, there is no analogous natural environment to the combination o f acidity, salin ity, high heavy metal availability and low macronutrient (N, P, K, Ca and Mg) availability that is found on the lower reaches o f dams.Plants that naturally colonize gold tailings, or persist on tailings-polluted substrates are therefore exhibiting a remarkable combination of adaptive or constitutive physiological tolerances and are being subjected to massive selection pressures (Bradshaw 1952;1970;Mehary 1994).For example, the majority o f species growing on slimes dams and polluted sub strates in the goldfields have depressed seed production and viability levels in comparison to conspecifics on unpolluted substrates (Weiersbye & Witkowski 2002b).However, those species that occur frequently on slimes and slimes-polluted soils are also the species that main tain regeneration potential on these same substrates (Witkowski & Weiersbye 1998).Tree species such as Acacia karroo, A. hereroensis, A. hebeclada and Rhus lancea maintain relatively high levels o f seed viability and germination despite elevated inorganic contents, and high frequencies o f seedling abnormalities (Weiersbye & Witkowski 2003).Rhus lancea is also capable o f vig orous growth in acid mine drainage, maintaining high evapotranspiration rates (Dye et al. 2005).Salinity and acid-tolerant land-races o f the grass Cynodon dactylon, and local ecotypes o f Hyparrhenia hirta are now used in slimes dam rehabilitation (B.Dawson, EMPR Services, pers.comm.).Apparently healthy C. dactylon growing in gold tailings can contain 30 mg g' 1 dry mass o f iron in the root epidermis and cortex, and 3 mg g 1 dry mass o f uranium within the root stele, with even higher Fe and U concentrations associated with arbuscular mycorrhizal structures (Weiersbye et al. 1999).Some indigenous, naturally-colonizing or persisting species that were frequently encountered on slimes dams, have also been recorded from gold tailings in Zimbabwe and Botswana (Wild 1974a, b), and from andalusite, asbestos, gold, platinum and base metal tailings in the Limpopo and Mpumulanga Provinces (I.M. Weiersbye, K. Balkwill & E.T.F.Witkowski unpublished).Landraces that persist and colonize gold tailings and acid mine drainage-polluted soils can be expected to have phytoremediation potential for the gold mining industry in South Africa.

CONCLUSIONS
O f the 376 species found in the intensive survey, only 60 were common to all three regions, and o f these 24 had been introduced during rehabilitation attempts.Most of the species found on tailings were persisters or natural colonizers (53-88% , depending on substrate), with the vast majority being indigenous and perennial taxa (76% and 85% respectively) with semi-woody to woody growth forms (66% being resprouters, forbs, shrubs and trees).The present rehabilitation aims o f mine management (i.e. the requirement for rapid green cover) forces vegeta tion contractors to expend massive effort and expense in modifying tailings dams to become temporarily suitable substrates for high basal cover pastures.In contrast, those species actually persisting on, and naturally colonizing tailings and tailings-polluted soils, are non-pasture spe cies, o f which < 5% have also been introduced during vegetating efforts.Naturally colonizing and persisting species are predominantly indigenous perennials com prised o f resprouting, semi-woody and woody plants and C4 tussock grasses, which, by virtue o f the comparatively longer life-span o f individuals and apparent tolerance to native slimes conditions are more likely to assist in the establishment o f self-sustaining cover and rehabilitation o f gold tailings.Finally, only multi-stemmed/shrubby or ground-covering woody plants and other growth forms with high basal cover are suitable for planting on the slopes of the dams due to erosion foci that may develop around large, single-stemmed-trees.However, plantings o f shrubs and trees are suitable for the berms and tops o f slimes dams, providing that planting densities are optimized in order to survive on incoming rainfall alone in an increasingly arid environment as the dam dries out.Woody plantings on the berms and tops o f dams could minimize recharge o f the phreatic surface within the dam, and thus limit the potential for seepage, as well as lower ing the risk of fire that is ever-present in highveld winter grasslands, provide large canopies in order to abate wind and dust generation, and facilitate nutrient cycling proc esses and 'safe-sites' for seedling establishment more effectively than grasses (Cresswell 1973).The results of this broad-scale survey show conclusively that the reha bilitation industry needs to pay much greater attention to the use o f indigenous plant species and growth forms on TSF that have a higher probability o f contributing to sus tainable cover, dust control and hydrological containment than the currently used pasture species.
ing AngloGold personnel for their assistance: Nico Theunissen, Clive Taylor, Neville Green, Chris Olivier, Gunther Wiegenhagen, Piet Van der Grijp, Bob Freeman.Harry de Jonge, Des Bell, Abri Groenewald, Gesie Weingerl and Steve Bullock.Brian Dawson, Brian Cook, Les Reyneke, Piet van Deventer, other vegeta tion contractors, AngloGold, FreeGold (Harmony) and the Chamber o f Mines o f South Africa are thanked for access to vegetation contracts and records.Les Brown, an anonymous referee, and the editor provided useful comments on a previous draft o f the paper.
•a .a« r.   i^irÍÊ H H h h h * * *Taxa that were intentionally introduced at some stage during slimes dam vegetating attempts prior t o survey i n 1996.^Taxa that were naturally present on slimes during this survey, and were also introduced t o gold slimes dam trials subsequent t o this survey (between 1996 and 2000).
, and contain an estimated 430 000 tons o f uranium (Council for Geosciences 1998; Winde 2004a.b. c) and approxi mately 30 million tonnes o f sulphur (Witkowski & Weiersbye 1998).The volume o f waste generated by mining in South Africa increases at the rate o f 315 m il lion tonnes per annum, mostly in the form o f tailings, o f which 105 million tonnes per annum is generated by the gold mining industry on the Witwatersrand Basin alone, at the rate o f 200 000 tonnes o f waste per ton o f gold (Department o f Tourism, Economic and Environmental Affairs.2002; Chamber o f Mines o f South Africa, 2004).
to a healthy environment elevated to a basic human right in the Constitution o f South Africa (Act No. 108 o f 1996).A number o f Acts stress the responsibility o f industry to prevent environmental damage, and pro vide for the prosecution o f polluters.These include the Environment Conservation Act (ECA) No. 73 o f 1989 and ECA Amendment Act 50 o f 2003, the Conservation o f Agricultural Resources Act No. 43 o f 1983 and amendments o f 2001, the National Environmental Management Act 107 o f 1998 and amendments, the National Water Act 36 o f 1998, the National Nuclear Regulator Act 47 o f 1999, the National Environmental Management: Biodiversity Act No. 10 o f 2004, the National Environmental Management: Air Quality Act No. 39 o f 2004, the Minerals and Petroleum Resources Development Act 28 o f 2002, the National Environmental Management: Protected Areas Act No. 57 o f 2003, and the National Environmental Management Amendment Act No. 46 o f 2003, which facilitated the 'Green Scorpions' unit to investigate environmental offences.In addition, the new regulatory framework for water usage renders industry liable for the cost o f water used, and polluted, as a result o f operations and rehabilitation under the new Waste Discharge Charge System o f the Department of Water Affairs and Forestry.This environmental legisla tion means that novel, sustainable and cost-effective methods o f containing pollution from tailings dams, have to be established.

FIGURE 1 .
FIGURE 1.-Sketch map showing the location o f the study sites (dark grey shading), in relation to the Witwatersrand Basin (light grev shading) (after Anhauser 1987).
FIGURE 2.-Aspect o f a gold slimes dam divided into the substrate classes.

*
Taxa that were intentionally introduced at some stage during slimes dam vegetating attempts prior to survey in 1996.^Taxa that were naturally present on slimes during this survey, and were also introduced to gold slimes dam trials subsequent to this survey (between 1996 and 2000).APPENDIX I P e r c e n ta g e frequency of species and subspecies within each substrate, vegetation class and region lor plants found i n the 100 m transects.Although 327 species and subspecies were identified, for the pur poses of this frequency analysis the 5 Tamarix spp.and putative hybrids were combined due to difficulty i n telling them apart in the field (i.e.n vO sO OC IT) OC Tf sC «/*< sC: C;; vC> <N 00j ; rn rfr O C Tt rn O C O ' o o oc o r*d o d r i d f N o c ^d ' d r i d ^oc --o c rr -p r i r i r i d d r i h- of species and subspecies within each substrate, vegetation class and region for plants found i n the 100 m transects.Although 327 species and subspecies were id e n tifie d , for the p u r poses of this frequency analysis the 5 Tamarix spp.and putative hybrids were combined due to d iffic u lty i n telling them apart in the field (i.e.n = 323) cont.

APPENDIX 1 .
-______ Percentage frequency of species and subspecies within each substrate, vegetation class and region for plants found i n the 100 m transects.Although 327 species and subspecies were identified, tor the pur poses of this frequency analysis the 5 Tamarix spp.and putative hybrids were combined due to difficulty in telling them apart i n the field (i.e.n

•
were intentionally introduced at some stage during slimes dam vegetating attempts prior to survey i n 1996.+Taxa that were naturally present on slimes during this survey, and were also introduced to gold slimes dam r i v "i -r i r i d © o rrn o r -o o o r -o © O ^C rn O vC O O riO sO O O --Percentage frequency of species and subspecies within each substrate, vegetation class and region for plants found i n the 100 m transects.Although 327 species and subspecies were identified, for the pur poses of this frequency analysis the 5 Tamarix spp.and putative hybrids were combined due to difficulty in telling them apart i n the field (i.e.n Taxa that were intentionally intnxluced at some stage during slimes dam vegetating attempts prior to survey in 1996.*Taxa that were naturally present on slimes during this survey, and were also introduced to gold slimes dam tnals subsequent to this survey (between 1996 and 2000).APPENDIX I .-Percentagefrequency of species and subspecies within each substrate, vegetation class and region for plants found in the 100 m transects.Although 327 species and subspecies were identified, for the pur poses of this frequency analysis the 5 Tamarix spp.and putative hybrids were combined due to difficulty i n telling them apart i n the field (i.e.n

TABLE 5 .-Categories o f plants found on gold slimes dams. Values are percentages, and values in parentheses are no. taxa (species, subspecies and varieties) found
o f environments such as iron-age kraal sites(Blackmore et al. 1990).C. ciliaris was only prevalent on recently vegetated sites, with few individu als surviving the transition to old vegetated sites and the concomitant decline in nutrient availability.In contrast, Stoebe vulgaris and some dominant indigenous grasses on tailings are inhibited by high nitrogen availability . The grass Cenchrus ciliaris is tolerant o f high nitrogen and phosphorus availability (O 'Connor & Bredenkamp 1997) and is an indicator o f nutrient enrichment in natural ecosystems.109being characteristic