Introduction

Highly metal(loid)-contaminated soils are a global issue and lead to the poor and irregular development of vegetation or even the total disappearance of plant cover on affected sites (Gray et al. 2006). Although novel, gentle technologies for plant microbe-based bioremediation (phytoremediation) and in situ immobilization using soil amendments have been developed and widely tested on slightly to moderately contaminated soils (Mench et al. 2009; Vangronsveld et al. 2009), there is little information concerning their performance at extremely high, toxic metal levels. Remedial action at such sites faces a number of problems, including enhanced metal leaching from the root zone or deeper soil layers and problems with crop establishment due to metal toxicity (Nagajyoti et al. 2010; Iqbal et al. 2012). In situ immobilization is only a temporary solution; i.e. contaminants still remain in the environment (Wuana and Okieimen 2011).

Recently, a novel soil remediation option combining enhanced phytoextraction with prior in situ immobilization of metals using immobilization agents has been proposed (Iqbal et al. 2012). Initial metal(loid) immobilization helps reduce the metal(loid) mobility and toxicity to plants, as well as leaching metal(loid)s from the root zone and thus facilitates the development of seedlings or cuttings of the metal-accumulating crops, mainly during the critical initial growth stages. Since the immobilizing agents are expected to limit phytoextraction, it has been suggested they support the phytoextraction process by the subsequent controlled addition of metal-mobilizing agents. Easily biodegraded chelating agents, including naturally occurring substances, come to the fore in enhanced phytoextraction (Iqbal et al. 2012; Ali et al. 2013).

Some studies have tested the feasibility of using acidifying compounds such as elemental sulphur (Wang et al. 2007; Iqbal et al. 2012; Amoakwah et al. 2014; Hoefer et al. 2015). The chemical fate of sulphur in soil (oxidation or reduction) depends on many factors, such as the presence of oxygen and its content (Edwards 1998). The application of elemental sulphur induces slow and steady soil acidification, mainly confined to the rhizosphere, which is considered to limit leaching while rendering metal contaminants available to phytoextraction (Iqbal et al. 2012). Under aerobic conditions, this effect is ascribed to the oxidation of sulphur to sulphuric acid by Thiobacillus bacteria domiciled in the soil matrix and thus a subsequent decrease in soil pH through the release of SO42− and H+ ions (Van Ranst 2006). Under occasional anaerobic conditions, S-oxidizing bacteria may use Mn and Fe oxides as terminal electron acceptors. This leads to a reduction and dissolution of these oxides and greatly contributes to metal desorption (Iqbal et al. 2012). Under anaerobic conditions, S is reduced mainly to hydrogen sulphide and these species can be quickly volatilized, reoxidized to elemental S or precipitated with metals (Edwards 1998). Wenger et al. (2002) described the duration (i.e. a range of 4 weeks to 7 weeks) required for the sulphur-containing amendments to mobilize substantial amounts of the metals investigated. Several authors (Cui et al. 2004; Iqbal et al. 2012; Hoefer et al. 2015; Azeez et al. 2016) have verified the efficiency of elemental sulphur application on metal-tolerant plants, for example maize, sunflower or willow, and confirmed an increased plant uptake of the metals.

Following on from the experiments of Iqbal et al. (2012) and Hoefer et al. (2015), our study tests the feasibility and efficiency of sulphur treatments following immobilization of Cd, Zn and Pb in a highly contaminated soil. Whereas in previous work immobilization was achieved using oxidic (red mud) and silicatic (gravel sludge) additives, the focus here is on carbonaceous materials, namely activated carbon (Břendová et al. 2016; Zemanová et al. 2017), lignite (Havelcová et al. 2009; Uzinger et al. 2014) and vermicompost (Jadia and Fulekar 2008; Hoehne et al. 2016). The aim is metal stabilization/immobilization via processes such as adsorption, cation exchange, surface complexation and/or transformation on the sorbent surface (Kumpiene et al. 2008; Tica et al. 2011), as well as providing nutrients to support plant growth in the contaminated soil.

The objectives of this research are (i) to test the efficiency of the combined immobilization/mobilization technology (i.e. before and after S application) using relatively metal-tolerant maize in pot conditions and (ii) to investigate the solubility of bioavailable nutrients (P, S) and metals (Mn, Fe, Cd, Pb, Zn), maize growth, element uptake by the maize and the resulting remediation factors and efficiencies of the different treatments.

Materials and methods

Experimental design

The experimental soil was sampled from the topsoil of a Gleyic Fluvisol (FAO 2014), long-term, highly Cd/Pb/Zn-contaminated soil (54 mg Cd kg−1, 3305 mg Pb kg−1, 6172 mg Zn kg−1 as pseudo-total concentrations extracted with aqua regia) from the alluvium of the Litavka River. Anthropogenic soil contamination originates from two main sources: metallurgical industry emission outputs and floods of contaminated water from metallurgical setting pits, the barriers of which have repeatedly been damaged (Borůvka and Vácha 2006). The specific characteristics of the experimental soil are given in Vondráčková et al. (2014).

Four kilograms of air-dried and sieved (≤ 2 mm) soil in 5-L plastic double pots (each inner pot with three holes, 20 cm in diameter and depth) was mixed with nutrients at the rate of 0.1 g N kg−1 (NH4NO3), 0.032 g P kg−1 and 0.08 g K kg−1 (both in the form of K2HPO4) of soil. After 1 month of plant growth, a supplementary application of nutrient solution containing only nitrogen and phosphorous in the same forms and rates as in the initial application was repeated due to visible deficiencies. Another potassium application was omitted to prevent further cation mobilization. The immobilizing agents (applied in two different doses: dose 1 (~ 45 g additive per kg of soil) and dose 2 (~ 90 g additive per kg of soil)), including activated carbon (Erspol Ltd., Czech Republic), lignite (Mibrag GmbH, Germany) and vermicompost (Oekovermes, Germany), were sieved (≤ 2 mm) without any drying and thoroughly mixed with the soil immediately after the first application of nutrient solution. The agents and characteristics were as follows: activated carbon from coconut shells (\( {\mathrm{pH}}_{{\mathrm{H}}_2\mathrm{O}}=7.6 \), 647 g C kg−1, 4.3 g N kg−1, 1.6 g S kg−1, 8.3 mg Zn kg−1), lignite from sedimented organic substances (\( {\mathrm{pH}}_{{\mathrm{H}}_2\mathrm{O}}=3.4 \), 629 g C kg−1, 5.3 g N kg−1, 28.7 g S kg−1, 2.8 mg Zn kg−1) and vermicompost from horse manure (\( {\mathrm{pH}}_{{\mathrm{H}}_2\mathrm{O}}=5.5 \), 247 g C kg−1, 23.3 g N kg−1, 11.2 g S kg−1, 231 mg Zn kg−1). Total Cd and Pb concentrations of the additives were below the limit of detection (< 0.1 mg kg−1 and 2 mg kg−1, respectively).

High-yielding Zea mays (variety Colisee, KWS Saat AG), with a known potential for phytoremediation (Mojiri 2011), was seeded (five individuals per pot) immediately after the initial application of nutrient solution and immobilizing agents. Only 3 maize plants of similar height were left after 10 days of growth. In the vicinity of the roots, 10-cm-long Rhizon samplers (Rhizosphere Research Products, Wageningen, Netherlands) were inserted at a soil depth of 10 cm into individual pots to collect soil pore water (i.e. soil solution).

Nine weeks after sowing, elemental sulphur (Central Chem, Prague, Czech Republic) was applied as a mobilization agent to the soil in all pots at a rate of 0.5 g S kg−1. The S application was very difficult due to its low aqueous solubility. Therefore, a suspension of S and demineralized water was applied via five holes drilled into the uppermost 7 cm of each pot. All soil treatments (untreated control soil with maize plants (C), activated carbon (AC), lignite (Lig) and vermicompost (VC) at dose 1 (~ 45 g additive per kg of soil) and dose 2 (~ 90 g additive per kg of soil)) were replicated 9 times for a total of 63 pots. These high rates were necessary to alleviate the extreme Cd, Pb and Zn concentrations of the Litavka soil.

The plants were grown for 14 weeks in an outdoor university department vegetation hall in Prague–Suchdol under natural temperature and light conditions. Plants were harvested at several intervals (always with three replications)—after 9 weeks (before S application), 12 weeks (19 days after S application) and finally 14 weeks (33 days after S application)—and were separated into roots and shoots. Concurrently, soil samples (approximately 30 g of soil taken from the top 10 cm of soil) were collected at 3 weeks, 9 weeks (before S), 12 weeks (19 days after S) and 14 weeks (33 days after S) after planting and soil solutions after 9 weeks (before S) and 12 weeks (19 days after S) using Rhizons. Soil leachates were collected from the bottom of the outer pot after 14 weeks, i.e. 33 days after S application.

Amendments and soil and plant analysis

The pH (EN 13037 1999) of the amendments was measured, and elemental analysis of C, N and S was undertaken using a Thermo Scientific Flash 2000 NCS Analyser (Germany). Total elemental (Cd, Pb, Zn) concentrations in the amendments were determined by neutron activation analysis (Kubešová and Kučera 2012).

Soil material was collected from each pot, air-dried and analysed for pH in a 1:5 (w/v) suspension of soil and 0.01 M CaCl2 after 1 h of shaking (ISO 10390 2011). Concurrently, the plant-available elements (P, Fe, Mn, Cd, Pb, Zn) and major inorganic (NO3− and SO42−) and organic (oxalate) soil anions were measured. The measurements were performed after extraction with 0.01 M CaCl2 (soil:solution ratio 1:10, 6 h shaking; Tlustoš et al. 1994) and determined using inductively coupled plasma-optical emission spectrometry (ICP-OES) (Varian Vista Pro, Varian, Australia) for elements or by means of an ICS 90 ion chromatograph equipped with IonPac AS14A (Dionex, USA) for anions, respectively. For further details of this method, see Tejnecký et al. (2013). After collection, the soil solutions and leachates were stored in vials, immediately cooled (4 °C) and analysed within 12 h for pH, elemental concentrations and major inorganic and organic anion concentrations using the same procedure as for the soil samples. The soil samples were analysed 3 weeks after maize sowing to determine nutrient and toxic metal bioavailability. To determine dissolved organic carbon (DOC) and N form (total, NO3−, NH4+), analysis was undertaken using a segmented flow analyser (Skalar San++ system, Netherlands). Plant-available and exchangeable fractions of selected elements were analysed using ICP-OES (for P, Cd, Pb and Zn) and flame atomic absorption spectroscopy (FAAS; Varian SpectrAA-280, Australia; for K, Mg) according to the extraction methods of Tlustoš et al. (1994; plant-available fraction) and of Pierzynski (2000; exchangeable fractions, for P).

All plant materials were harvested and dried in an oven at 60 °C to constant weight on sampling dates (i.e. before S application, 19 days and 33 days after S), and the dry weight was recorded before grinding in a stainless steel mill. The roots were washed in an ultrasonic-assisted bath filled with deionized water to remove any soil particles adhering to the roots (Elmasonic S30, Elma Ultrasonic Technology). The total elemental concentrations (P, S, Fe, Mn, Cd, Pb, Zn) in plant parts (roots and shoots) were determined by ICP-OES after microwave-assisted high-pressure acid digestion (65% HNO3:30% H2O2 4:1, Ethos 1, MLS GmbH, Germany). Certified reference material (CTA-OTL-1 oriental tobacco leaves) was used for quality control. The remediation factor (RF, %) was calculated as the ratio between the total metal content in a plant (g per pot) and the total metal content in soil (g per pot) multiplied by 100.

Data analysis

All statistical analyses were performed using Statistica 12.0 software (www.statsoft.com). All data were checked for normality and homogeneity of variance (Shapiro–Wilk and Levene tests). A one-way analysis of variance (ANOVA), followed by a Tukey HSD test (α = 0.05), was used to identify significant differences among treatments before S application and after S application for soil and plant data. The relationship between selected data was analysed by linear regression using Pearson coefficients.

Results and discussion

Nutrient and toxic metal bioavailability after maize sowing

Three weeks after sowing, the AC and Lig treatments showed the lowest CaCl2-extractable concentrations of DOC, NO3−, NH4+, Mg, K and Zn (Table 1). This can be attributed to the adsorption and/or complexation of cations to the large specific surface of AC and Lig (Qi et al. 2011; Břendová et al. 2016). The highest concentrations of P and K were recorded in the VC2 treatment, which can be explained by the release of nutrients from vermicompost (our measured nutrient contents in vermicompost were 12.5 g K kg−1 and 11 g P kg−1; Uz et al. 2016) and by the release of Pb from dissolved organic matter in the soil (Jordan et al. 1997). According to Pierzynski (2000), 10 mg P kg−1 (Olsen’s method, i.e. extraction with 0.5 M NaHCO3 and then adjustment of the pH to 8.5) can be considered the optimum for plants grown in agricultural soils. The lowest measured extractable P concentration was 23 mg kg−1 in the C and Lig2 treatments but did not indicate a low P availability for plants. However, this result was in contrast to the reddish colour of maize leaves in the C, AC and Lig treatments, indicating P deficiency (Nagajyoti et al. 2010). This inconsistency was probably related to the high level of soil contamination (i.e. the Olsen extraction method can overestimate plant-available P because metal phosphate formation may be blocked by the alkaline pH associated with this method). Higher NO3− concentrations compared to NH4+ supported the nitrification activity, as expected in aerated soil.

Table 1 Dissolved organic carbon (DOC), dissolved organic nitrogen (DON), nutrient and toxic element bioavailability (mg kg−1, mean ± standard error of the mean (SEM), n = 3) in soil 3 weeks after maize sowing
Full size table

Effects of immobilization agents and sulphur on soil pH

The soil pH, measured with both methods (CaCl2 extraction and soil solution obtained by suction cups), followed a similar pattern of changes before S application (Fig. 1a). The \( {\mathrm{pH}}_{{\mathrm{CaCl}}_2}/\mathrm{soil} \) solution increased from 5.5/5.4 in the control to 6.2/6.6 in the AC treatments. This increase can be attributed to the alkalinity of activated carbon and a corresponding liming effect (Břendová et al. 2016). In the remaining treatments (except VC2 in soil solution), soil pH did not change. Before S application, there was no difference between the pH measurement in CaCl2 and soil solution extraction as observed previously by Amoakwah et al. (2013).

Fig. 1
figure 1

Soil solution and extracted CaCl2 pH (mean ± SEM, n = 3) a before S application and b after S application. Significant differences among the treatment and sulphur application means at p < 0.05 are indicated with letters. Soil leachate applies only for day 33. See Table 1 for more details about the treatments

Full size image

In general, only a few changes in \( {\mathrm{pH}}_{{\mathrm{CaCl}}_2} \) were recorded between before and after S application (Fig. 1b). This is supported by the linear regression results for CaCl2-extractable metal concentrations (Cd, Pb, Zn) and pH before and after S application (Cd: before S, r = − 0.62, p < 0.01; 33 days after S, r = − 0.49, p = 0.02; Pb: before S, r = − 0.79, p < 0.01; 33 days after S, r = − 0.49, p = 0.03; Zn: before S, r = − 0.78, p < 0.01; 33 days after S, r = − 0.63, p < 0.01). These results contrast with those of other studies (Iqbal et al. 2012; Hoefer et al. 2015) in which S application had strong effects on the correlation between pH and metals. Therefore, we can speculate that pH did not decrease after S application in the whole profile (pH of soil solution) but only in the top 10 cm of soil (\( {\mathrm{pH}}_{{\mathrm{CaCl}}_2} \)) at the end of the experiment. The S oxidation was limited in most treatments after the application in the pot experiment. However, significant increases in soil solution pH recorded after S application were also related to the absence of an acidifying rhizosphere effect because the last collection was extracted from soil leachate (i.e. bulk soil) and not from soil solution (i.e. root zone). Similar results were observed in the study of Azimzadeh et al. (2016).

Effects of immobilization agents and sulphur on nutrient in soil and toxic metal soil mobility

More substantial changes in CaCl2-extractable nutrients and metals were recorded in soil samples compared to soil solution before and after S application (data for soil solution not shown). A tendency for higher CaCl2-extractable concentrations of P and Fe was recorded in all treatments before S application (Table 2). The highest increase was recorded in VC treatments (i.e. 105% (VC2) and 123% (VC1) increase over control for P and Fe, respectively). Significantly higher CaCl2-extractable Pb was recorded in the VC1 and Lig1 treatments compared to the control. A decrease of the DOC was found in all amendment treatments compared to the control, showing the stability of applied amendments. The highest immobilization for Mn, Cd, Pb and Zn in the AC treatments was caused by an alkaline pH, as demonstrated by the linear regression (Mn: r = − 0.94, p < 0.01; Cd: r = − 0.95, p < 0.01; Pb: r = − 0.91, p < 0.01; Zn: r = − 0.94, p < 0.01). This was probably also caused by the corresponding metal adsorption to the large specific surface of activated carbon (Břendová et al. 2016). Before S application, the oxalate concentrations decreased considerably in all the amended treatments (Table 3). We speculate that the lower oxalate concentration is related to lower Zn and other metal toxicity (i.e. lower CaCl2-extractable metal concentrations from the soil solution) in response to the amendments. Therefore, the root exudation of oxalate, i.e. a potential detoxification mechanism due to the formation of metal complexes, was downregulated. Li et al. (2012) recorded a similar result, i.e. increased exudation of oxalic acids by Chinese cabbage grown in a Zn-polluted soil.

Table 2 CaCl2-extractable element concentration (mg kg−1, mean ± SEM, n = 3) for the different treatments and periods
Full size table
Table 3 CaCl2-extractable inorganic and organic anion concentration (mg kg−1, mean ± SEM, n = 3) for the different treatments and periods
Full size table

As mentioned above, only a few pH and metal concentration changes were recorded in CaCl2 extracts after S application (Fig. 1b, Table 2). We can speculate that the S oxidation and therefore the metal mobilization (Fe, Mn, Cd, Pb, Zn) in the C, AC and VC treatments were inhibited. Generally, S oxidation is performed by numerous microbial species (Hagedorn 2010) within a wide pH range of 1.9 to 8.5, but it is also affected by various other environmental conditions. The inhomogeneity of S application in the pots, the S rate (higher S rate, i.e. 1.0 g kg−1, with a more pronounced pH response and metal solubilization; Mohnke 2014; Hoefer et al. 2015), the use of organic immobilization agents (lowering the redox potential) and/or more time for observation are possible factors limiting S oxidation.

The CaCl2-extractable Mn, Cd, Pb, Zn and sulphate concentrations increased and \( {\mathrm{pH}}_{{\mathrm{CaCl}}_2} \) decreased after S application in investigated treatments in the last sampling period (Fig. 1b, Tables 2 and 3). This may be also related to the S forms of amendments: inorganic forms easily accessed by microbes (mainly sulphate and pyritic S) and organic forms (bound to carbon and esters) greatly depend on the coal integrity, i.e. structure, and the S release from ester sulphates using extracellular enzymatic hydrolysis (Ruamsap and Akaracharanya 2002; Hagedorn 2010; Blum et al. 2013). It could also be linked to its preparation: the milling of amendments before mixing with soil could have increased the microbial availability of the organic S form since its structure was destroyed (Ruamsap and Akaracharanya 2002). Blum et al. (2013) showed that microbial S transformation in soil is greatly affected by the composition of different S forms of the source material (i.e. amendments and soil). Therefore, we speculate that easily accessible S forms in Lig and, to a lesser extent, in other amendments functioned as a driver for S oxidation and thus stimulated S-oxidizing bacteria, probably before S was applied to the soil. This could explain the tendency for increasing CaCl2-extractable sulphate concentrations in the AC2, Lig and VC treatments, and no notable change was determined in the control after S addition (Table 3).

A tendency for decreasing CaCl2-extractable nitrate concentrations was only recorded in the C and AC2 treatments after S application, which did not correspond with low biomass yield and is in line with the findings of Hoefer et al. (2015). Hailegnaw et al. (2019) also found immobilization of soil nitrates due to decline of negatively charged groups at a high rate of high temperature–produced biochar. In contrast, the CaCl2-extractable nitrate concentration in the AC1, Lig and VC treatments increased after S application despite higher plant uptake due to the higher biomass production compared to the C and AC treatments. The VC and Lig applications may have influenced the N mineralization, leading to a sustained release from organic compounds. This agrees with the results of Murugan and Swarnam (2013), who found slow but steady N mineralization in vermicompost-treated soil. The oxalate concentrations significantly increased in the control treatment after S application, probably due to increasing root exudation of organic acids.

Effects of immobilization agents and sulphur on maize growth

Maize biomass increased for the treatments in the order of C ≈ AC ≈ Lig < VC after S application (Fig. 2a). This is consistent with Tury et al. (2008), who reported positive effects of compost on plant growth in contaminated soils. After S application, the final maize biomass was also significantly increased in the Lig2 treatment (Fig. 2b). Despite presenting the lowest concentrations of CaCl2-extractable Cd, Pb and Zn (Table 2), the AC treatments showed a substantially lower (up to 14 times) biomass production than the VC treatments. The NO3− immobilization was only pronounced in the AC2 treatment. Therefore, we can conclude that the VC treatment—and to a lesser extent the Lig application—showed the trend of steady release of some nutrients, while in the AC treatment, sorption mechanisms limited the availability of nutrients, during the whole experimental period and thus limited plant growth.

Fig. 2
figure 2

Maize biomass (dry weight, DW) measured a before S application and b after S application. Significant differences among the treatment means at p < 0.05 are indicated with letters. Error bars indicate the standard error of the mean (n = 3). See Table 1 for more details about the treatments

Full size image

Effects of immobilization agents and sulphur on total metal uptake and toxic metal remediation factors

Large amounts of metals (Mn, Fe, Cd, Pb, Zn) were taken up by the maize roots due to the high metal concentrations (both total and CaCl2-extractable) in the Litavka soil (Table 4). Before S application, majority of treatments did not show statistical differences in the root concentrations, higher values recorded for the VC1 treatment; could be caused by some impurity coming from soil; and were not determined in other samplings. Only the mobile elements like Zn and, up to some extent, Cd decreased root concentrations in treated soils before S application, presenting a stronger effect at treatments with higher rates of amendments. Generally, metal (Fe, Mn, Cd, Pb, Zn) concentrations in the maize shoots were largest in the control, and partly at AC1 treatment with low biomass production, and lowest in the treatments with high biomass production before and after S application (Fig. 2, Table 5). The opposite trend was found for roots: high biomass yield corresponded with high Fe, Mn and Pb concentrations, because of relatively high availability, and root uptake, but limited translocation to aboveground tissues (Siedlecka 1995). Hence, we conclude that biomass production caused a dilution effect for the respective metal concentrations only in the maize shoots. Vondráčková et al. (2014) recorded similar results. Another reason could be related to an increase in biomass over time after S application. Moreover, the plant membrane selectivity during stress due to toxicity or deficiency can deteriorate and lead to an uncontrolled uptake of toxic metals (Hall 2002). Pourrut et al. (2011) described the breakdown of the Casparian stripe under toxicity stress. This could explain the excessive uptake of less mobile Pb and other metals (Cd, Zn, Fe and Mn) into the shoots in the control (Table 5). There were no considerable changes in root or shoot concentrations of elements after S application.

Table 4 Element concentrations in maize roots (mg kg−1, mean ± SEM, n = 3) for the different treatments and periods
Full size table
Table 5 Element concentrations in maize shoots (mg kg−1, mean ± SEM, n = 3) for the different treatments and periods
Full size table

In general, distinguishing between toxicity symptoms and deficiencies was rather difficult throughout the experiment. Certainly, a combination of both occurred. We can speculate that P and K deficiencies might have reduced plant growth at treatments with high C/N ratio. The high CaCl2-extractable concentrations of Cd and Zn might have induced plant stress and a poor performance (i.e. tapering leaves and a reduction in the leaf area) in the C treatment. Other authors have recorded similar symptoms (Reichmann 2002; White 2012; Sherameti and Varma 2015). These findings can be supported by comparing the concentrations of nutrients and metals measured in the maize shoots with the sufficiency ranges recommended in the literature (White 2012). The Cd, Pb and Zn concentrations in the maize shoots in all treatments were often higher than the toxicity range (5–10 mg Cd kg−1; 10–20 mg Pb kg−1; 100–700 mg Zn kg−1). The concentrations in the control were always the highest. The content of Fe and Mn in shoots lay within the sufficiency range (50–250 mg Fe kg−1; 10–20 mg Mn kg−1) in both VC treatments, but within the toxicity range (> 500 mg Fe kg−1; 200–5300 mg Mn kg−1) in the control. The content of P in the shoots indicates a deficiency for the plants in all treatments (sufficiency range 3000–5000 mg kg−1; Marschner 1995).

The highest total Cd and Zn uptake in the maize shoots was recorded in the VC treatments (Fig. 3a, c) and increased phytoremediation efficiency by 400% for Cd and 100% for Zn in comparison with the control. Phytoremediation efficiency could be substantially increased in the VC and, to a lesser extent, in the Lig treatments, since the total Cd and Zn uptake in the maize shoots was higher than that in the control. While the total Cd and Zn uptake by the maize shoots from the VC treatments was not significantly affected by S application, the content in the Lig2 treatment increased significantly, but the values were more than 40% (for Zn) to 70% (for Cd) lower than those in the VC treatments. The phytoremediation efficiency in VC treatments was mostly driven by higher shoot biomass. The Cd and Zn concentration in shoot decreased after S application. The VC treatments did not show higher nutrient availability than the other treatments (Table 3). Therefore, the higher Cd and Zn removal is only related to the high shoot growth, very likely due to the effect of vermicompost on plant growth. Alternatively, a high metal-accumulating plant (i.e. willow instead of relatively metal-tolerant maize) could be used in further experiments to enable larger amounts of Cd, Pb and Zn to be taken up without suffering from toxicity (Mohnke 2014).

Fig. 3
figure 3

a–c Total metal content (biomass multiplied by the metal concentration, mg pot−1) in maize shoots after S application. Significant differences among the treatment means at p < 0.05 are indicated with letters. Error bars indicate the standard error of the mean (n = 3). See Table 1 for more details about the treatments

Full size image

Before S application, the remediation factors for Cd, Pb and Zn increased considerably in the VC treatments by approximately three, five and four times, respectively (Fig. 4a–c). For Cd and Zn, this can be explained by higher total biomass, as confirmed by linear regression (Cd: r = 0.94, p < 0.01; Zn: r = 0.94, p < 0.01). In contrast, the higher remediation factor for Pb can be better explained by higher Pb concentrations in the root (r = 0.89, p < 0.01) than by higher biomass. Moreover, after S application, the remediation factors also increased in the Lig treatment for Cd and Zn. The increased removal of toxic metals is due to the higher plant growth.

Fig. 4
figure 4

Proportion of a Cd, b Pb and c Zn removed from the soil by maize biomass before S application (treatment) and after S application (sulphur application) periods as indicated by the remediation factor (RF, %). Significant differences among the treatment and sulphur application means at p < 0.05 are indicated with letters. Error bars indicate the standard error of the mean (n = 3). See Table 1 for more details about the treatments

Full size image

Conclusions

Before S application, AC, Lig and VC applications reduced the CaCl2-extractable Cd and Zn concentrations, while the concentrations of Pb increased in the Lig1 and VC1 treatments compared to the control. After S application, the metal concentrations (Fe, Mn, Cd, Pb, Zn) in the soil solution and the CaCl2 extraction in the C, AC and VC treatments were not significantly affected by S application. Similarly, the Cd, Pb and Zn concentrations in the maize shoots did not significantly increase in the respective treatments. Several processes (heterogeneous S application or the use of organic agents; Edwards 1998) might have decreased the redox potential and thus inhibited S oxidation. The CaCl2-extractable concentrations of Mn, Cd, Pb, Zn and sulphate increased significantly, and the pH decreased after S addition in investigated treatments in the last sampling period. We suppose that easily accessible S forms in amendments functioned as a driver for microbial S oxidation, probably even before S was applied to soil. During the whole experimental period, the VC treatment and, to a lesser extent, the Lig treatment showed the trend of steady release of some nutrients, while in AC treatment, sorption mechanisms limited the availability of nutrients. Consequently, the highest biomass production and the lowest toxicity and deficiency symptoms were recorded in the VC and Lig treatments, while the AC treatments produced even less biomass than the control. Severe toxicity symptoms were found over the duration of the experiment there. The maize biomass production at the final harvest was about five times higher in the Lig2 treatment and fourteen times higher in the VC treatment compared to the control. The remediation factors for Cd, Pb and Zn increased mainly in the VC treatments because of higher total biomass (for Cd and Zn) and higher metal concentration in the roots (for Pb).

In future research, redox potential changes in the plant rhizosphere should be regularly monitored to test the possibility of microbial sulphate reduction when using Litavka soil and organic amendments. Using different plants (hyperaccumulators or fast-growing, metal-accumulating trees such as Salix smithiana) that are able to take up large amounts of toxic metals in highly contaminated soils might also be productive. The subsequent S application should be improved to guarantee a more homogeneous distribution in the soil, as well as a higher S application rate. Alternatively, longer observation times could be used to obtain statistical significance.