Characterisation of alkaline tailings from a lead/zinc mine in South Africa and evaluation of their revegetation potential using five indigenous grass species

Abstract

Tailings from a lead/zinc (Pb/Zn) mine were characterised and their revegetation potential investigated under glasshouse conditions using five grass species with three rates of inorganic fertiliser. The tailings were alkaline with low nutrient concentrations but high total and extractable Zn. The yield of all grass species increased with an increase in fertiliser rate. The yield of Cenchrus ciliaris at the full fertiliser application rate was significantly higher than the other species tested, followed by Digitaria eriantha. Cymbopogon plurinodis was the third-highest-yielding species, whereas yields of Eragrostis superba and Fingeruthia africana were similar. Concentrations of Zn in the foliage tended to be over the reported grass foliage ranges, whereas Pb concentrations were within typical norms. It is recommended that C. ciliaris, D. eriantha and E. superba be used for initial revegetation, with other species used to improve biodiversity after initial cover has been established.

Keywords:

• carbonate-rich
• dolomitic tailings
• indigenous
• low cost
• phytorestoration

Introduction

Tailings from metalliferous mines are generally recognised as being environmentally hazardous because of high concentrations of potentially toxic elements (e.g. Tordoff et al. 2000, Rodriguez et al. 2009, Shi et al. 2011, Santos-Jallath et al. 2012). Lead/zinc (Pb/Zn) mines are no exception, where these, and other elements, may pose serious environmental risks and limit the potential for revegetation (Hossner and Hons 1992, Ye et al. 2000a, 2000b, Lei and Duan 2008, Boussen et al. 2013). Tordoff et al. (2000), Li (2006), Lei and Duan (2008), Shi et al. (2011) and Santos-Jallath et al. (2013) all suggest that establishment of vegetation is the preferred method of tailings stabilisation, where the vegetation can remove, transform or stabilise the waste material and toxins through processes of adsorption, microbial action and stabilisation. This approach is frequently used in conjunction with more traditional methods of contaminant stabilisation, such as liming to increase substrate pH to precipitate toxic metals and reduce acidity associated with the oxidation of sulphide minerals (Simon et al. 2010), the use of phosphate compounds to bind metals (Prasad 2003, Prasad and Freitas 2003, Wang et al. 2008) or various organic amendments (e.g. Ye et al. 1999, Jordan et al. 2008).

Plants used to stabilise contaminated soils and wastes should be easily propagated and established, fast growing, have extensive root systems, and be tolerant or adapted to adverse substrate conditions that include elevated contaminant levels and poor physical and fertility characteristics (Prasad and Freitas 2003). Conesa et al. (2008) suggests that annual species are well suited for this purpose as they tend to have higher rates of adaptation because of shorter life-cycles that promote a larger variety of genotypes in a shorter time-frame. The use of indigenous plants to revegetate mine tailings is often more advantageous than using introduced crop species as indigenous plants may be better adapted to environmental conditions at a site and require less maintenance (Kramer et al. 2000). The use of both organic and non-organic amendments is often used in mine tailing restoration to enhance substrate properties for plant growth and a number of researchers have investigated the potential of various amendments to improve the growth of plants in Pb/Zn mine tailings (e.g. Lan et al. 1998, Ye et al. 2000a, Shu et al. 2002, Chiu et al. 2006, Jordan et al. 2008, Wang et al. 2008, Kabas et al. 2012). However, often the availability of suitable amendments (e.g. compost and sludges) is limited and thus not practicable for use as an amendment for the restoration of mine lands. Under these conditions, inorganic fertilisers are often preferred for the initial and rapid phases of vegetation establishment.

Geologically, Pb and Zn ore deposits often occur together and frequently contain traces of cadmium (Cd), copper (Cu) and other heavy metals (Dudka and Adriano 1997). Both Pb and Cd have no known biological function and as a consequence their accumulation in the environment can cause severe toxicological problems (Gulson et al. 1994, Manz and Castro 1997, Milton et al. 2002). The objective of this study was to characterise the tailings from the Pering Pb/Zn Mine in an arid (carbonate-rich) region of South Africa and to investigate the potential of indigenous grass speciesto revegetate the tailings. A pot experiment was used to evaluate the growth and metal uptake of five grass species grown in tailings material collected from the disposal dam and amended with inorganic fertiliser, under glasshouse conditions. While a range of metals were considered, the focus was primarily on Pb and Zn in this study.

Materials and methods

Site description

The decommissioned Pering Pb/Zn Mine is located in the North West province of South Africa (27°26′ S, 24°16′ E). Mining operations ceased in January 2003 after 17 years of ore extraction. During this time approximately 18 Mt of ore, with an average grade of 0.6% Pb and 3.6% Zn, was mined by opencast methods (Gutzmer 2005) extracting primarily the minerals sphalerite (ZnS) and galena (PbS). In addition, minor traces of pyrite (FeS₂) and chalcopyrite (CuFeS₂) are also reported to have been present in the ore (du Toit 1998). The metal sulphides were separated from the milled ore (typically <0.25 mm) by froth flotation and the fine-grained tailings were discarded as a slurry in the tailings dam. The dam is approximately 522 000 m² with an average depth of 30 m and a volume of 12 100 000 m³. No natural or planted vegetation is present on the tailings dam, though a Eucalyptus sp. windrow had been planted around the base of the dam to reduce wind-blown dust.

Waste material collection and preparation

Tailings samples were collected from 10 sites across the top of the tailings dam to a depth of 0.5 m. These samples were bulked (about 500 kg) and homogenised for analysis and for use in the pot experiment. All analyses and the pot experiment were carried out using tailings material that was air-dried, homogenised and milled to pass through a 2 mm sieve. All analyses were conducted in triplicate (unless otherwise indicated) and the mean results reported.

Chemical and physical characteristics

Electrical conductivity (EC) and pH were measured in distilled water with a Radiometer CDM83 electrical conductivity meter and a Radiometer PHM210 pH meter with a standard glass electrode, respectively, using a 10 g tailings:25 ml solution ratio. The pH was also measured in 1 M KCl using the same tailings:solution ratio. Extractable base cations (calcium [Ca], potassium [K], magnesium [Mg] and sodium [Na]) were determined by saturation with NH₄ ⁺ (1 M ammonium acetate, pH 7) and cation exchange capacity (CEC) by subsequent replacement with K⁺ (Soil Classification Working Group 1991). Calcium, K, Mg and Na concentrations were determined by atomic absorption spectrophotometry (AAS; Varian SpectrAA-200) and NH₄ ⁺ concentrations by steam distillation (Bremner and Mulvaney 1982), using a Gerhardt Vapodest 1. Nitrate and ammonia were extracted with 2 M KCl (Maynard and Kalra 1993) and solution concentrations determined colourimetrically using a TRAACS 2000 continuous flow auto-analyser. Total nitrogen (N) was determined by Kjedahl digestion and NH₄ ⁺ by distillation (Bremner and Mulvaney 1982). Plant-available phosphorus (P) was estimated by extracting with AMBIC solution (0.25 M ammonium bicarbonate, pH 8.3) and P was determined colourimetrically (The Non-Affiliated Soil Analysis Work Committee 1990) on a Varian Cary 1E UV-Visible spectrophotometer (UV-Vis). Organic carbon (OC) was determined titrimetrically following potassium dichromate oxidation on <0.5 mm material (Walkley 1947).

Total elemental concentrations, namely silicon (Si), aluminium (Al), iron (Fe), manganese (Mn), Ca, Mg, Na, K, titanium (Ti), P, sulphur (S), arsenic (As), Pb, Zn, Cu, nickel (Ni), chromium (Cr), vanadium (V), barium (Ba) and strontium (Sr), were determined using a Phillips PW1040 X-ray fluorescence spectrometer (XRF) utilising a vacuum sealed X-ray tube set at 50 kV and 50 mA (analysis carriedout by the Discipline of Geology, University of KwaZulu-Natal, Durban). Plant-available Cd, Cu, Cr, Mn, Fe, Ni, Pb and Zn (diethylenetriaminepentaacetic acid [DTPA] extractable, pH 7.3) were determined by the method of Liang and Karamanos (1993). This method was selected as it was developed for use in neutral and alkaline soils and substrates, thus making it suitable for the tailings material. Particle size distribution was determined by the pipette method (Gee and Bauder 1986) after sample treatment with sodium hexametaphosphate and mechanical dispersion by ultrasound.

X-ray diffraction (XRD) of random powders was carried out on a Philips PW1050 diffractometer using monochromated CoKα radiation, from 3° to 75° 2θ with a scanning step of 0.02° at 1° per minute counting interval. The diffraction data were captured by a Sietronics 122D automated microprocessor attached to the X-ray diffractometer. The samples were then analysed qualitatively to determine the major mineralogical components.

Establishment of the pot experiment

Seed of Cenchrus ciliaris L. (perennial), Cymbopogon plurinodis Stapf ex Burtt Davy (perennial), Digitaria eriantha Steud. (perennial), Eragrostis superba Peyr. (weak perennial growing for two to five seasons) and Fingeruthia africana Lehm. (perennial) (all members of the Poaceae) were collected from the Pb/Zn mine site, where all grew in the vicinity of the tailings dam. There was no literature that had specifically tested the use of these species in the reclamation of Pb/Zn tailings, and as such no nutrient and element sufficiency and toxicity limits were found. Germination of D. eriantha was found to be poor, so a commercial cultivar was selected as an alternative. The seed was germinated in commercially available seedling mix in standard nursery trays and grown to three weeks of age before transplanting three seedlings into each pot.

Tailings material was placed in 1.9 l plastic pots (with a fine glass-fibre membrane placed over the drainage-holes), lightly tapped a few times and the mass of material determined (3.3 kg). All pots were filled with this mass of tailings (which approximated the tailings bulk density of 1 750 kg m⁻³ reported by Titshall 2007). This bulk density value was used for the calculation of fertiliser application rates (based on a depth of 0.2 m). Fertility analysis indicated that the tailings material had negligible available N and P, with very low K. As no general recommendations are available for the species tested here, rates werebased on recommendations received from the Soil and Analytical Services Division (Department of Agriculture, Cedara) for ryegrass pasture. Though it was expected that this recommendation would be higher than the requirements for natural veld grasses, it was considered suitable in this study to compensate for the low nutritional status of the tailings material. Thus an initial basal fertiliser was applied to each pot consisting of the equivalent of 100 kg N ha⁻¹, 150 kg P ha⁻¹ and 100 kg K ha⁻¹ and referred to as the full fertiliser application rate. In addition, fertiliser was applied at half the full rate and an unfertilised (zero) treatment was included. Given the requirement for a simple approach to improve the establishment of vegetation on the tailings dam, the study was constrained to the use of readily available granular fertiliser. Furthermore, the availability of organic amendments was not considered feasible for the region because of the lack of readily available organic sources. The experiment was arranged in a randomised block design with three replications.

Fertiliser was applied as solutions after planting. Cenchrus ciliaris, D. eriantha and E. superba grew well and were harvested six weeks after planting. The other species grew more slowly and did not have sufficient biomass for plant chemical analysis, thus were allowed to grow for an additional two weeks before harvesting. Plants were harvested by cutting the aboveground foliage 10 mm from the substrate surface. Aboveground biomass was determined after the material was dried in a forced-draft oven at 65 °C for 2 d. The pH (water) of the substrate was measured after harvesting. The tailings material in each pot was air-dried, homogenised and passed through a 2 mm sieve prior to pH analysis.

Analysis of plant and pot material

All aboveground plant material was mechanically milled to pass through a 0.5 mm mesh. The samples were wet digested in reflux tubes using concentrated nitric acid at 130 ± 2 °C for 5–6 h (Slatter 1998). After cooling the digests were filtered into 100 ml volumetric flasks and made up to volume with distilled water. The digests of the grasses were analysed for Cd, Cu, Cr, Fe, Mn, Ni, Pb and Zn by AAS. Given the low yield of some grass species, plant material was bulked and analyses conducted in duplicate, negating the use of statistical comparisons in these instances.

Statistical analysis

Overall differences between means of the yield of foliage of the grass species were compared by analysis of variance (ANOVA; n = 3) using the statistical package GenStat 8.1 (Lawes Agricultural Trust 2005). As there were differences in the growth periods of some of the species tested, a covariate was used to account for the effects of these differences in the ANOVA (Rayner 1967). Where the overall F-statistic for the grass × fertiliser level interaction was found to be significant, means were compared with the least significant difference (LSD) test at the 5% level of significance with GenStat 8.1. Metal concentrations of the foliage were not analysed in this manner because of lack of replication of some treatments, though trends and patterns are discussed. Values were compared to suggested norms reported in the literature. The relationship between the yield of each species and fertiliser application rate was determined by linear regression analysis with GenStat 8.1.

Results and discussion

Chemical and physical characteristics

The tailings material was found to consist of predominantly fine sand and silt, with low clay content (Table 1), though this was expected as the tailings was the by-product of crushing and milling of the rock matrix to about 0.25 mm for the extraction of sulphide ores by froth flotation. However, the high amount of fine sand may lead to compaction problems (Skopp 2000), and was reflected in the high bulk density (1 750 kg m⁻³) of the tailings material (Titshall 2007). Natural settling of the different size fractions in the material during pumping of tailings slurry onto the dam are likely to have contributed to the high bulk density, rather than direct compaction induced through compression or vibration. In addition, the low organic matter content would further contribute to the high bulk density.

Table 1:  Mean and standard deviation (SD; n = 3) chemical and physical characteristics of the tailings material from the Pering Pb/Zn Mine

Characteristic	Mean	SD
PH (H2O)	8.20	0.33
(1 M KCl)	8.55	0.02
Electrical conductivity (25 °C) (mS m^−1)	51.5	1.1
Organic carbon (g 100 g^−1)	0.44	0.05
Extractable P (mg kg^−1)	2.03	0.24
NH4 ^+-N (mg kg^−1)	6.78	0.43
NO3 ^−-N (mg kg^−1)	2.37	0.24
Total N (mg kg^−1)	256	–
Extractable base cations (cmolc kg^−1)
Calcium	4.00	0.23
Magnesium	1.75	0.03
Sodium	0.04	0.04
Potassium	0.02	0
Cation exchange capacity (cmolc kg^−1)	2.26	0.18
Particle size (%)
Coarse sand (0.50–2.00 mm)	0.1	–
Medium sand (0.25–0.50 mm)	5.4	–
Fine sand (0.053–0.25 mm)	79.3	–
Silt (0.002–0.053 mm)	13.3	–
Clay (<0.002 mm)	1.9	–

The tailings material was alkaline and within similar ranges reported by Iavazzo et al. (2012) for Pb/Zn mine wastes from arid parts of Morocco (pH 7.9–8.5). The pH_KCl was higher than the pH_water, which was attributed to the displacement of base cations from the tailings in the salt-buffered solution. The EC (Table 1) was below the limit recognised for saline soils (400 mS m⁻¹) and lower than the value (325 mS m⁻¹) reported by Conesa et al. (2008) for neutral Pb/Zn tailings from southern Spain. The OC content, plant-available nutrients (N and P), extractable base cations and CEC were all low (Table 1), particularly K concentrations. It is suggested here that the OC content measured was a reflection of dissolution of dolomitic carbonates during the acid digestion, rather than carbon derived from organic materials. As such the carbon content is likely tobe an overestimation of true organically derived carbon. This is expected as the tailings would have a low organic carbon content given they were produced from the crushing of dolomitic matrix rock.

The high concentrations of total Ca and Mg (Table 2) reflect the dolomitic mineralogy of the tailings. Total K was also considerably higher than extractable amounts, indicating that it was complexed in a non-labile form. Total Zn concentrations were the highest of the trace metals, which was attributed to the occurrence of residual ZnS in the tailings. Total Pb was considerably lower than the Zn concentrations, which was attributed to the lower ore concentrations and perhaps also the higher PbS extraction efficiencies at the processing plant. The high total sulphur (S) concentration was due to the presence of sulphide minerals not extracted during ore processing and sulphates that formed when residual sulphides had oxidised. Shi et al. (2011) reported total N, P and K values of 30, 370 and 1 820 mg kg⁻¹, respectively and total Zn, Pb and Cu concentrations of 1 328, 1 217 and 163 mg kg⁻¹, respectively, for alkaline Pb/Zn tailings from Fuyang City in China. Conesa et al. (2008) reported total Cu, Zn, Cd and Pb values of 81, 9 130, 33 and 5 310 mg kg⁻¹ for neutral tailings from southern Spain. This range in values highlights the high variability between different tailings materials, but it does suggest that the Pering Pb/Zn tailings were relatively ‘clean’ when compared to other sites across the world.

Table 2:  Total major and minor element composition, determined by X-ray fluorescence analysis, of tailings from the Pering Pb/Zn Mine

Major element composition (% oxide)
SiO2	Al2O3	Fe2O3	MnO	MgO	CaO	Na2O	K2O	TiO2	P2O5	Total
4.87	1.48	4.03	3.66	32.6	49.9	0.26	0.76	0.05	0.02	97.6
Minor element composition (mg kg^−1)
S	As	Pb	Zn	Cu	Ni	Cr	V	Ba	Sr
18564	85	311	2170	6.2	4.4	2.8	5.5	0.8	17.2

The concentrations of DTPA-extractable Pb and Zn were high compared to the other metals measured and, as expected, tended to be higher than typical metal concentrations reported for South African surface soils (Table 3). Given the lack of standardised methods for analysing heavy-metal-contaminated soils (Fernandez et al. 2007, Menzies et al. 2007) it is difficult to compare results reported for soils and other mines. Difficulties arise with respect to the lack of consistent sampling depths, highly variable soil physical and chemical characteristics, and the use of a wide variety of chemical extractants to determine metal concentrations in labile and non-labile fractions. Nonetheless, the data show that only Zn was above the typical concentrations for ‘natural’ soils. In some instances (Cd, Cu and Pb) the metal concentrations exceeded the mean value of the soils, but were within the ranges reported for those metals. Jordan et al. (2008) extracted alkaline tailings with 1 M ammonium acetate to estimate exchangeable metal concentrations and reported values of 672, 8 182, 10.6 and 21.5 mg kg⁻¹ for Pb, Zn, Cu and Cd, respectively. With the exception of Zn, the values reported here were lower than those reported by Jordan et al. (2008). Martínez-Martínez et al. (2013) reported DTPA-extractable Pb and Zn concentrations of alkaline (pH >6) tailings of 65–775 and 497–2 449 mg kg⁻¹, respectively. As with total metal content, the plantavailable content of Pb and Zn of the Pering Pb/Zn tailings is at the lower end of the ranges reported by others for Pb/Zn tailings.

Table 3:  Diethylenetriaminepentaacetic acid (pH 7.3) extractable metal concentrations (±SD; n = 3) in the tailings material from the Pering Pb/Zn Mine and ethylenediaminetetraacetic acid-extractable metal concentrations of surface soils in South Africa. BD = Below detection, ND =not determined

Sample	Element concentration (mg kg^−1)
	Cd	Cr	Cu	Fe	Mn	Ni	Pb	Zn
Tailings	0.31 ± 0.17	BD	6.00 ± 0.12	2.60 ± 0.22	4.08 ± 0.05	0.12 ± 0.06	37.6 ± 11.3	957 ± 110
South African surface soils^a
Mean (n = 4 300)	0.02	0.16	2.87	ND	ND	2.39	3.32	1.66
Minimum	BD	BD	BD	ND	ND	BD	0.02	BD
Maximum	1.46	12.5	67.3	ND	ND	170	109	210
^a Herselman et al. (2005)

Generally, alkaline conditions promote sorption of metal cations onto negatively charged colloidal surfaces (Alloway 1990, McBride et al. 1997) and promote precipitation/ adsorption with carbonates (Brady et al. 1999, Schosseler et al. 1999, Lee et al. 2006). The buffered alkaline pH of the DTPA extractant was intended to limit the dissolution of specifically sorbed and mineral-complexed metals, thus better representing the labile component under these conditions. Nonetheless, given that the tailings had a pH >8, it is still possible that the pH 7.3 DTPA solution promoted some minor solid-phase dissolution, releasing complexed metals. The high DTPA-extractable Zn concentrations (44% of total Zn) indicate that this possibly had occurred, though this was not evident for extractable Pb (12% of total Pb). Zhang et al. (2003) and Iavazzo et al. (2011) investigated the characteristics of metals from Pb/Zn mine tailings from a carbonate area using fractionation procedures. Zhang et al. (2003) reported that Pb was primarily associated withcarbonate and oxide fractions and Cu and Zn with sulphide and organic fractions. Iavazzo et al (2011) reported that both Pb and Zn were associated with carbonate minerals (80% and 52%, respectively). The strong association of Pb with carbonates may, in part, explain the lower plant-available fraction extracted in this study. It does, however, suggest that under acidifying conditions, more Pb will be mobilised.

The XRD analysis indicated that the tailings material was crystalline with little amorphous constituents present. It consisted predominantly of dolomite (CaMg(CO₃)₂), with minor amounts of quartz (SiO₂), orthoclase feldspar (KAlSi₃O₈), sphalerite (ZnS) and pyrite (FeS). No galena (PbS) or any other sulphide minerals were detected.

Yield of grass species

The yield of all grass species increased with an increase in fertiliser application rate (Figure 1), which was expected considering the low fertility status of the tailings, suggesting that the use of fertiliser would be important when revegetating the tailings dam. There was a significant (F _10,34 = 3.62; p = 0.002) species × fertiliser interaction. Comparisons of means with LSD_5% showed that the yield of C. ciliaris at the full fertilisation rate was significantly higher (p < 0.05) than all other grass species at any fertilisation level (Figure 1). Similarly, the yield of D. eriantha at the full fertilisation rate was significantly higher (p < 0.05) than all other treatments except for C. ciliaris at the full fertilisation rate. There were no significant differences (p > 0.05) between the yield of D. eriantha and C. ciliaris at the half and zero fertiliser levels. Cymbopogon plurinodis had the third-highest yield at all fertiliser levels, though it grew for eight weeks to achieve this biomass (Figure 1). Eragrostis superba and F. africana gave the lowest yields, but were comparable to one another.

Figure 1:  Mean yield (±SD; n = 3) of foliage biomass of Cenchrus ciliaris (Cc), Digitaria eriantha (De), Eragrostis superba (Es), Fingeruthia africana (Fa) and Cymbopogon plurinodis (Cp) grown in tailings from the Pering Pb/Zn Mine at three levels of fertiliser: Zero = no fertiliser; Half =50 kg N ha⁻¹, 75 kg P ha⁻¹ and 50 kg K ha⁻¹; Full =100 kg N ha⁻¹, 150 kg P ha⁻¹ and 100 kg K ha⁻¹. Different letters indicate a significant difference between means (LSD_5% = 0.52; coefficient of variation = 4.7%)

Positive, linear correlations existed between yield and fertiliser application rate for all grass species tested. In all cases, except for D. eriantha, these linear relationships where highly significant (p < 0.001) and had R ² values greater than 0.8 (Table 4). The lower R ² found for D. eriantha (0.62) was attributed to the high variability in yields between replicates (Figure 1).

Table 4:  Regression relationships between yield of grasses and fertiliser application rate, indicating significance of the regression. FB =Foliage biomass

Grass species	Regression^1	R^2
Cenchrus ciliaris	FB = (2.63 x Fert) + 1.395	0.98***
Cymbopogon plurinodis	FB = (1.394 x Fert) + 0.902	0.91***
Digitaria eriantha	FB = (1.929 x Fert) + 1.575	0.62**
Eragrostis superba	FB = (1.050 x Fert) + 0.507	0.93***
Fingeruthia africana	FB = (1.218 x Fert) + 0.480	0.82***
^1 The regression term Fert refers to the fertiliser application level of 0, 0.5 and 1 (where 0 refers to the Zero fertiliser application rate, 0.5 to the Half fertiliser application rate (50 kg N ha^−1, 75 kg P ha^−1 and 50 kg K ha^−1) and 1 to the Full fertiliser application rate (100 kg N ha^−1, 150 kg P ha^−1 and 100 kg K ha^−1, respectively)
** p < 0.01, *** p < 0.001

Element concentrations in foliage

The concentrations of Cr, Cd and Ni were all below the analytical detection limits indicating that these metals are unlikely to be a cause for concern. They are, therefore, excluded from further discussion. There were no clear trends in Cu uptake by any of the grasses with respect to fertiliser application (Table 5). In some instances (C. ciliaris, C. plurinodis, D. eriantha and E. superba), Cu concentrations increased with increasing fertiliser application. Furthermore, some treatments had very high variability (e.g. C. ciliaris at the half fertiliser level). Nonetheless, concentrations measured were generally within the typical range reported for turfgrass and other plants (Table 5). Iron uptake was also variable with no consistent pattern evident for any of the grass species. Again, some treatments had high variability between replicates. Generally, the concentrations were within the adequacy range reported for turfgrass (Table 5). Fingeruthia africana was the only species that had Fe concentrations well below the recommended sufficiency range (Table 5), possibly resulting in the low yield of this species. For all grass species, except F. africana, there was an increase in Mn uptake with increasing fertiliser applications. In the case of F. africana at the full fertiliser application rate, Mn concentration of the foliage was higher (41.22 mg kg⁻¹) than at the half and zero fertiliser application rates, which were similar (31.75 and 31.24 mg kg⁻¹, respectively). Typically, the concentrations were within or near the adequacy range reported for turfgrass and other plants (Table 5).

Table 5:  Mean concentrations (±SD; n = 3) of Cu, Fe, Mn, Zn and Pb in the foliage of five grass species grown in tailings from the Pering Pb/Zn Mine with three levels of fertiliser application. The adequacy range for Zn is reported for turfgrass (Bennett 1993), ‘normal’ and ‘toxic’ ranges for Cu, Zn and Pb in vegetables (Conesa et al. 2009), and ‘normal’ and ‘contaminated’ ranges of Cu, Fe, Mn, Zn and Pb are reported for plants (Ross 1994). The pH (water) of the tailings material at the end of the pot experiment is also given. BD = Below detection

Species	Fertiliser^a	Cu (mg kg^−1)	Fe (mg kg^−1)	(mg kg^−1)	Pb (mg kg^−1)	Zn (mg kg^−1)	Final pH (H2O)
Cenchrus ciliaris	Zero	6.16	34.74	75.71	BD	167 ± 12	7.60 ± 0.02
	Half	13.13 ± 4.81	54.82 ± 22.70	110.66 ± 29.26	2.16 ± 0.58	278 ± 31	7.55 ± 0.03
	Full	9.71 ±1.28	207.89 ± 223.56	169.67 ± 18.90	7.46 ± 2.41	381 ± 33	7.48 ± 0.11
Cymbopogon plurinodis	Zero	5.69 ± 0.37	75.76 ± 1.93	32.50 ± 5.48	4.20 ± 0.96	56.7 ± 41.4	7.41 ± 0.23
	Half	6.47 ± 13.44	31.53 ± 10.05	34.46 ± 4.45	1.82 ± 0.91	336 ± 89	7.40 ± 0.15
	Full	6.96 ± 0.31	54.97 ± 8.54	49.65 ± 17.65	1.60 ± 0.66	621 ± 48	7.44 ± 0.06
Digitaria eriantha	Zero	6.62 ± 031	53.84 ± 3.88	37.41 ± 6.79	11.6 ± 4.0	312 ± 12	7.51 ± 0.03
	Half	7.94 ± 1.27	54.41 ± 1.82	51.30 ± 4.59	6.41 ± 6.81	561 ± 192	7.56 ± 0.02
	Full	9.08 ± 0.47	55.47 ± 2.71	64.09 ± 3.89	1.49 ± 1.29	918 ± 112	7.42 ± 0.08
Eragrostis superba	Zero^b	6.56 ± 0.58	69.14 ± 6.92	88.90 ± 18.97	5.22	565	7.57 ± 0.06
	Half	7.55 ± 1.34	69.46 ± 29.18	112.06 ± 18.35	9.50 ± 1.98	989 ± 136	7.56 ± 0.10
	Full	10.21 ± 0.62	88.60 ± 9.46	126.08 ± 29.84	13.1 ± 2.9	1 105±92	7.54 ± 0.11
Fingeruthia africana	Zero	5.98	54.74	44.64	9.71 ± 1.51	348 ± 20	7.62 ± 0.06
	Half	7.63 ± 0.64	60.74 ± 7.35	83.39 ± 7.35	2.54 ± 0.33	576 ± 136	7.50 ± 0.04
	Full	8.84 ± 0.36	75.79 ± 8.42	116.79 ± 8.42	2.06 ± 0.86	1 153 ± 74	7.48 ± 0.06
Bennett (1993)		5-20	35-100	25-150	–	–
Conesa et al. (2009)	'Normal'	3-20	–	–	2-5	15-150
	'Toxic'	25-40	–	–	–	500-1 500
Ross (1994)	'Normal'	4-15	–	15-1 000	0.1-10	8-400
	'Contaminated'	20-100	–	300-500	30-300	100-400
^a Zero, half and full refer to the fertiliser application rate, where full is the equivalent of 100 kg N ha^−1, 150 kg P ha^−1 and 100 kg K ha^−1
^b Due to low yield, not all replicates could be analysed, thus replicates were bulked and analysed in duplicate. The control treatment of Eragrostis superba only comprised a single analysis, thus the SD from the mean could not be calculated

Considering the tailings material was derived from Pb and Zn ore processing, it was anticipated that these elements may be elevated in plant foliage. Extractable values of Zn (Table 2) indicated moderately high availability of this element in the tailings, despite the high pH of the material. This is reflected in the high concentrations measured in the foliage of almost all grass species at all fertiliser application rates. All concentrations were above the adequacy range reported for turfgrass (Table 5). For all grass speciesthere also were marked increases in Zn concentrations with increasing fertiliser application. In the case of E. superba and F. africana, the concentrations exceeded 1 000 mg kg⁻¹ at the full fertiliser application rate, possibly due to a concentrating effect because of the low yield of these grass species.

Lead concentrations were variable among the grass species, but overall foliage concentrations tended to be low. The Pb concentrations of C. ciliaris and E. superba increased with increasing fertiliser application rate, whereas decreases were observed for the other grass species, possibly because of acidification of the tailings through nitrification of the N fertiliser and also root nutrient exchanges processes. The pH (water) of the tailings material at the end of the pot experiment (Table 5) showed that for all treatments and species the pH had decreased from the starting pH (Table 1) by between 0.6 and 0.8. However, only C. ciliaris and F. africana showed a decrease in final pH as fertiliser rate increased. It is thus possible that for C. ciliaris the acidification affect of the N fertiliser resulted in increased solubility and consequently uptake of Pb by the plants. Acidification will result in the solubilisation of metal–carbonate complexes and desorption of metal cations from exchange sites. In the case of E. superba the improved growth due to fertilisation may have resulted in an increased uptake of Pb by this species. The other species may exhibit better exclusion mechanisms, though this has never been investigated for these species.

The macronutrient concentrations of the tailings indicated that N, P and K are likely to be growth-limiting elements. The improved growth of all species due to fertiliser addition supported this suggestion (Figure 1). Zinc concentrations were very high in some grass species, and may have contributed to suboptimal yields, though no visual symptoms of toxicity were evident. The high Zn content of the plants was directly attributed to the high availability of Zn in the tailings. Excess Zn can result in stunted root and shoot growth, and chlorosis of new leaves (Farago 1994, Kabata-Pendias 2001). However, most plant species show tolerance to high Zn concentrations (Kabata-Pendias 2001). Although Zn concentrations were over the upper range suggested by Bennett (1993) when grown in the tailings (Table 5), this author reported that some grasses have not shown toxicity symptoms with Zn concentrations of up to 3 000 mg kg⁻¹. In addition, the grass species used in this study (except D. eriantha) were collected from the vicinity of the tailings and were possibly adapted to high Zn levels.

Ye et al. (1999) reported mean Zn and Pb concentrations of 454 and 40 mg kg⁻¹, respectively, in the foliage of the grass Agropyron elongatum grown in acidic Pb/Zn mine tailings. Addition of fertiliser led to increases in the foliage concentrations of Pb and Zn (490 and 59 g kg⁻¹, respectively). The addition of fertiliser had a similar effect on Zn concentrations in the grasses tested in the present study, though in some instances the increases were substantially greater than those reported by Ye et al. (1999). The same was not true for Pb concentrations, however, as the response was more variable in the grasses investigated. Inall cases though, Pb concentrations were lower than those reported by Ye et al. (1999). The higher Zn concentration found here was likely due to the much higher amounts of available Zn (DTPA-extractable) in the tailings (957 mg kg⁻¹) compared to 41 mg kg⁻¹ in the tailings studied by Ye et al. (1999). The reverse was true for Pb, with 38 mg kg⁻¹ DTPA-extractable Pb in the present study compared to 154 mg Pb kg⁻¹ in the study of Ye et al. (1999).

More recently, Conesa et al. (2008) investigated the uptake of Cu, Zn, Cd and Pb by two native grass species (Piptatherum miliaceum and Lygeum spartum) grown in pure Pb/Zn tailings or tailings-contaminated soil from southern Spain under greenhouse conditions. The authors found that L. spartum accumulated about 4, 190, 11 and 10 mg kg⁻¹ of Cu, Zn, Cd and Pb, respectively, whereas P. miliaceum accumulated in the order of 40, 900, 10 and 150 mg kg⁻¹ of Cu, Zn, Cd and Pb, respectively. This range of values covers the range reported for the grasses tested in the present study. Santos-Jallath et al. (2013) measured metal concentrations in nine vegetation species growing naturally on acid Pb/Zn mine tailings in Mexico, of which the grass C. ciliaris was one of the species studied. For this species they reported foliage concentrations of Cd, Pb, Cu and Zn of 3.2, 36.7, 6.6 and 232 mg kg⁻¹, respectively. These were similar to the values measured here for C. ciliaris, except for Pb, which was lower in the present study than in the study by Santos-Jallath et al. (2013). The strongly alkaline, carbonate-rich nature of the tailings used in this study is likely to have immobilised the Pb, thus lowering Pb uptake by the plant. Generally, trace metal concentrations decrease with an increase in soil pH (Gray et al. 1998, Mordvedt 2000), this being attributed to increased sorption onto negative exchange sites, formation of insoluble metal hydrous oxides and metal–carbonate complexes. However, in the present study the use of grass species occurring in the vicinity of the mine were preferred as these species are adapted to the climatic and alkaline soil conditions of the area.

In a study examining the effect of fertilisation on the phytoextraction potential of two Rumex acetosa accessions (metal and non-metal tolerant) grown in soils contaminated by Pb/Zn mining activites, Barrutia et al. (2009) reported that fertilisation increased the ability of the metaltolerant accession to take up Pb and Zn. Although none of the species investigated in this study were considered to be hyperaccumulator plants, it was also found that fertilisation increased the uptake of Zn, though not necessarily other metals. Given the risk associated with the introduction of heavy metals into grazing stock if metal uptake is high, an increase in metal uptake under fertilisation suggests that high rates of fertilisation need to be avoided.

The results for C. ciliaris and D. eriantha suggest that these two species would be highly suitable for revegetation of the tailings and require further field testing. Their high yield and tolerance of the substrate conditions are favourable attributes that are likely to promote the rapid establishment of a cover, as even with no fertiliser these two species had higher yields than the other grass species tested. Eragrostis superba is reported to be a fast-growing species that establishes readily (van Oudtshoorn 2002). In the present experiment, this species established very easily and observations while seed collecting in the field suggested that this species also had high seed-producing potential. Furthermore, most of the seed (>90%; LWT unpublished data) collected from the site was viable suggesting high propagation potential. This is a favourable trait for rapid revegetation of derelict sites. Whereas the other two species (F. africana and C. plurinodis) collected from the mine site produced higher yields than E. superba, they were observed to be slower growing under pot conditions than the other species tested. These species may be more suitable as long-term additions to improve species diversity at the sites. Both F. africana and C. plurinodis are considered to be highly unpalatable (van Oudtshoorn 2002), thus the addition of these species may reduce the movement of grazing animals onto the tailings dam, reducing the risk of development of paths that may increase the potential for erosion during the seasonal rainstorm events.

Generally, Pb concentrations in the foliage were not considered high in the grasses, but the high Zn concentrations may be cause for concern. Chaney (1983, cited by Ye et al. 1999) noted that excessive dietary Zn (300–1 000 mg kg⁻¹) and Pb (30 mg kg⁻¹) can be toxic to sheep and cattle. The greatest risk will be when animals are allowed to graze on vegetation grown on the tailings material, rather than in the surrounding areas, and this should be limited as far as possible. However, a risk not investigated here is the consequence of wind-dispersed tailings dust across the landscape on the nutritional quality of surrounding grazing land.

Given that this was a pot study, a cautionary note on extrapolating our findings to field conditions is necessary. De Vries (1980) argues that the results from pot experiments cannot be extrapolated to the field unless the pot experiment dealt with growing factors that dominated all other factors. The pot dimensions (volume of substrate the pot can hold), the preparation of the substrate, watering regime and glasshouse environmental conditions are all factors that are reported to affect the response of plants grown under glasshouse conditions (e.g. de Vries 1980, Ray and Sinclair 1998, Friesl et al. 2006, Passioura 2006) and the predetermined environmental parameters of pot experiments seldom represent the true range of field conditions, thus restricting the value of the results. Nonetheless, pot experiments are useful to improve our understanding of plant behaviour under preselected conditions (de Vries 1980, Friesl et al. 2006). This will, however, require calibration under field conditions.

Conclusions

The findings of this study showed that Zn and, to a lesser extent, Pb were the only metals likely to be of concern in the tailings material. Other metals measured tended to have low total and extractable concentrations and, as such, were not considered to be a problem. Low macronutrient concentrations were anticipated to be a potential plant growth limiting factor, but could be overcome with fertilisation.

The outcome of the pot experiment conducted suggested that all the species tested here could potentially be used to establish a vegetation cover on the tailings dam. The higher biomass of C. ciliaris and D. eriantha are consideredfavourable to increase vegetation biomass and thus increase the organic carbon status of the tailings. Although E. superba had lower biomass production, the high viability of its seeds makes it a favourable species for initial site establishment and succession. The other species may have limited value for site establishment because of apparent slow growth, but they may be useful for increasing vegetation diversity in the future. Apart from Zn, there was limited evidence for potential metal toxicity in the plants. Lead concentrations were within ?normal' plant ranges, because of both low concentrations in the tailings materials and the low mobility of Fe in the dolomitic matrix. Zinc concentrations frequently exceeded guideline limits and could potentially pose a toxicity risk, both to the plant and also animals grazing these plants. However, evidence from the literature suggests that plants grown in pots tend to take up higher levels of elements than under field conditions, suggesting that the levels reported here may be an overestimation of what can be expected in the field.

Further testing under field conditions is suggested. Fertilisation is likely to be key in ensuring early establishment, and the performance of the species tested here needs to be evaluated under natural conditions.

Acknowledgements

The authors acknowledge funding from the BHP-Billiton Johannesburg Technology Centre and would like to thank, in particular, Mrs Ritva Muhlbauer and the staff of the Pering Pb/Zn Mine for their assistance.