Introduction

Soil can be considered essentially as a non-renewable resource by the European Union, for a total of about 400 million hectares. Soil degradation is an acute problem, often resulting from human, industrial and mining activities (JRC 2012). The estimated cost of soil degradation could amount to 38 billion euros per year in the European Union where the area of soils affected by mining activities has been estimated at 0.6 %, compared to a world average of 0.2 %. The remediation of soils contaminated by mining activities is, therefore, a strategic objective for European policies.

Indeed, high concentration of “heavy metals” (HM) in soils of mining areas constitutes a high risk of pollution, also for agricultural activities and villages lying in subsidiary areas (Garcia et al. 2008). In uncontaminated soils, HM concentrations vary in a very wide range (Kabata-Pendias 2001); these toxic elements are known to be adsorbed by clay minerals; phosphates; Fe, Al and Mn oxides and/or hydroxides; and by the organic matter (Kabata-Pendias 2001; Reimann et al. 2003). In mine ores, toxic elements can be incorporated in sulphide minerals such as, for example, As in arsenopyrite and pyrite (Sherman and Randall 2003). Due to the common extractive processes, soils of mining areas are characterized by severe alteration of the pH, absence or very low organic carbon content, loss of structure, extremely high concentrations of toxic elements (obviously related to the mine ores but quite often including the most toxic ones as As, Cd, and Pb in addition to Zn, Cu, Sb, etc.) and, last but not the least, drastic reduction of microbiological diversity (Giller et al. 1998). This means that any recovery activity has two priorities: first of all to reduce the risk of toxic element mobilization and transfer to underground and surface waters and, second, to create the conditions to restore the area to a natural equilibrium. Reactions buffering pH in mine settings play a fundamental role (Jurjovec et al. 2004), but the other fundamental step is the restoration of the microbial diversity.

Recovery could be achieved through natural attenuation which, however, is extremely slow. In addition, environments contaminated by metals generally suffer from low microbial activity; for this reason, any approach of bioremediation can be difficult to be implemented in the abandoned mining areas. On the other hand, studies on mining areas or other “extreme” environments are particularly useful to increase the scarce knowledge still existing about the extent of microbial diversity and cultivability, estimated in the order of 5–10 % and 0.1 %, respectively (Staley and Konopka 1985; Amann et al. 1995; Hugenholtz et al. 1998). The discovery of new microorganisms may increase the huge potential that microorganisms can contribute to the improvement of phytoextraction/stabilization technologies, inherent in their capacity to adapt, to communicate with plants and to accelerate chemical reactions up to 106 (Ortíz-Castro et al. 2009).

The idea of the UMBRELLA project (7FP EU) to progress beyond the current state of knowledge and applications was to use an integrative approach, derived by “geobiological” processes, for soil remediation in mining and other HM-contaminated areas across Europe, coupling microorganisms with plants to reduce risks for the environment and human health with a positive impact on downstream fluvial systems, including international waterways. This approach is based on the identification of appropriate strains of microorganisms native to the site and their use in combination with endemic plants for optimizing processes of phytoextraction or phytostabilization. For this purpose, microbial strains with suitable characteristics, HM tolerant and plant growth promoters (PGP), have been selected to establish microbial consortia to be employed as bioaugmentation agents in phytoremediation experiments. Mycorrhizae contribution was also investigated: mycorrhiza dependence is common among plants selected for phytoremediation of industrial tailings (Turnau et al. 2012); although, the practices commonly used often omit mycorrhiza propagation. Plants generally used in such practices include grasses that can develop without mycorrhiza (facultative mycorrhizal) but whose development would be still dependent on climatic conditions and human care, such as watering, which is an expensive alternative. In such cases, the phytoremediation starts to be sustainable and well progressing when arbuscular mycorrhizal fungi (AMF) appear (Turnau et al. 2012; Ryszka and Turnau 2007). Mycorrhizae were found to improve many aspects of plant life originating from scarce nutrition and low water availability, to reduce stresses from toxic metals, heat and UV and to enhance plant resistance to pathogens.

Here we describe the experience developed at the Italian test site of the abandoned mine of sphalerite and galena in Ingurtosu (Sardinia) to assess the applicability of a “toolbox” developed to establish optimal techniques for coupling microorganisms and plants for the remediation of soils contaminated by mining activities. The experience is part of a set of fieldworks carried out in parallel in mining sites throughout Europe.

The experimental activity on the Ingurtosu test site started with a preliminary characterization including (hydro)geochemistry, heavy metal concentration and their mobility in soil, bioprospecting for microbiology and botany. These data were the base for the development of a toolbox to deliver a microbially assisted phytoremediation process. Euphorbia pithyusa was selected as an endemic pioneer plant to be associated with the bacterial consortium, namely UIC, composed of ten native selected strains, metal-tolerant bacteria and effective producers of PGP compounds. The toolbox was tested in a greenhouse pot experiment to assess if the members associated were mutually compatible under the given specific soil conditions. In Wernitznig et al. (2013), results are shown with a special emphasis on plant physiological parameters and metal interaction, obtaining some positive and some ambiguous results. In this work, we analyse in more detail the positive results for soil microbial activity studied at the community level through Biolog ECOPlates, as a point of an absolute priority for assessing undergoing phytoremediation processes, due to the major role the microbial community plays as indicator of soil quality.

The field trial was set up based on the outcome of the positive points of the pot experiment and began in October 2011. The experimental design of the field trial included different treatments used singly or in combination: bioaugmentation with bacterial consortia (BACT), mycorrhizae (MYC) and a commercial mineral amendment (VM) obtained from red mud, a by-product of the bauxite industry which is widely used, after adequate neutralization, for environmental remediation processes due to its metal-trapping capacity (Ma and Feng 2011; Liu et al. 2011). The trapping mechanism is controlled by various phenomena that are different for each metal as a function of its characteristics and of the form in which it is present (speciation in solution, formation of complexes, formation of oxyanions, solubility of its salt) (Brunori et al. 2005). Red mud is characterized by very high alkalinity, and its major constituents are crystalline hematite (Fe2O3), boehmite (γAlO(OH)), quartz (SiO2), sodalite (Na4Al3Si3O12Cl) and gypsum (CaSO4·2H2O), with a minor presence of calcite (CaCO3), whewellite (CaC2O4·H2O) and gibbsite (Al(OH)3). Viromine™ is obtained from the reaction between red mud and sea water, thus reducing the resulting pH from >13 to pH <9 (Virotec International 2013). The environmental applicability of this material according to the Italian legislation was already evaluated (Brunori et al. 2005).

Five months after the beginning of the field trial, an early assessment of the toolbox in field conditions was carried out to decide whether to carry forward the field trial or to go back to the lab for any adjustment of the toolbox.

Materials and methods

Study area

The Ingurtosu mine belongs to the Arburese mining district, located in SW Sardinia (Fig. 1). Exploitation of the Montevecchio–Ingurtosu mineralization was a prime source of Pb and Zn for more than a century. Ore body is a vein system extending in a NNE-SSW direction for at least 12 km. The vein bodies were mostly emplaced within the thermometamorphic aureole of the Variscan intrusion but partly cut the intrusion itself (the so-called radial veins). At the Ingurtosu Pb–Zn mine, as in other mines of the district, galena and sphalerite were the targets of exploitation. Chalcopyrite and pyrite were present in small amounts. Gangue minerals were quartz and siderite. A detailed description of mineral characterization can be found in Medas et al. (2012a, b).

Fig. 1
figure 1

The experimental field at Ingurtosu (Sardinia, Italy) at the initial time. The field is organized in 27 subplots; those in red colour contain the Viromine™ amendment

Full size image

The mine activity ceased in 1968 without any care for the environmental degradation, particularly due to the heavy metal dispersion. The Ingurtosu mine is included in the list of abandoned mine sites to be reclaimed in Sardinia (RAS 2003).

The abandoned mine tailings are drastically influenced by a semi-arid climate with rare but sometimes heavy rain events, scarcity of vegetation and limited groundwater resources (Medas et al. 2012b). Among plants that are able to colonize these sites, the following were observed: Cistus salvifolius, Rosmarinus officinalis, Ranunculus bullatus, Festuca sp., Helichrysum italicum, Ptilostemon casabonae and Euphorbia pithyusa. The last species was chosen for the field experiment as well adapted to the local conditions.

Pot experiment

The experimental design and the methods used in the pot experiment are exhaustively described in Wernitznig et al. (2013). The procedures for the preparation of soil inoculum and analysis of soil microbial activity are the same as described below for the field trial.

Field trial experimental design

The soil was homogenized and levelled with a motor hoe, forming an area of 200 m2, and a channel was then created for surface water drainage. The site was fenced in with wire netting to protect plants from wild animals. An area of about 7.5 m × 22.5 m on the field site was divided into 27 subplots (3 × 9). The side length of each subplot is 2.5 × 2.5 m. For each subplot, the inner part of about 2.0 × 2.0 m has been used, leaving a service strip of about 0.5 m. In the experimental subplots, different treatments were applied to the soil and consisted of BACT, MYC and the commercial amendment Viromine™ (VM), used singly or in combination (Fig. 1). The VM subplots were treated mixing about 10 % w/v of the amendment provided by Virotec Italia Srl in the first 20 cm of soil.

Control plots without treatment (field control, FC) were planted with E. pithyusa, as well as control pots with garden soil (PC) to compare the plant growth. After the partition of the field in subplots and before proceeding with bioaugmentation and plant introduction, three soil samples have been collected from the top layer of each subplot, obtaining a composite sample that was analysed for granulometry, mineralogy, trace element (Pb, Cd, Cu and Zn) content and microbial activity, in order to assess whether the operation of field setup succeeded in establishing an acceptable uniform starting “baseline”.

The field experiment started in October 2011; a few days after having introduced the bacterial and mycorrhizal consortia, endemic E. pithyusa was introduced in 21 subplots. In each subplot, 64 plantlets previously grown for 3 months in the garden soil of the Murgia greenhouse were planted. The remaining six plots were left without plant as negative control (the scheme of field experiment setting is reported in the supplementary material as Table 1S). After transplanting, the field was regularly watered. Five months later, five plantlets with the corresponding soil rhizosphere were collected from each subplot and were transported to the laboratory. The soil samples of the five plantlets from the same plot were mixed, after sieving (0.2 mm), to create a composite soil sample. Microbiological analyses were carried out on fresh soil samples. The fine roots of plants from each plot were stained separately according to standard protocol, and the presence of mycorrhiza was determined.

Plant and soil samples were oven-dried at 50 °C overnight and their heavy metal content analysed after digestion.

Preparation of bacterial and mycorrhizal inocula

Bioaugmentation was performed using the native bacterial consortium (UIC), composed of ten bacterial strains selected among all isolates to have shown the best metabolic characteristics of HM resistance and PGP: UI2-Pseudomonas sp., UI3-Stenotrophomonas maltophilia, UI4-Rhizobium sp., UI6-Niabella sp., UI7-Curtobacterium flaccumfaciens, UI9-Streptomyces violarus, UI18-Streptomyces sp., UI24-Plantibacter sp., UI27-Niabella sp. and UI28-Bacillus cereus. In order to obtain a balanced consortium, the cell concentration of each strain was counted in a Neubauer chamber; then, the different bacterial suspensions were appropriately combined to provide the consortium with ten different strains having a well-defined microbial load. In the field, the consortium suspension was then diluted in tap water to obtain 106 CFU/g of soil corresponding to 108 CFU/m2 and spread on the surface of the soil with a watering can.

The arbuscular mycorrhizal inoculum was produced using trap cultures of fragmented root systems of plant species (Rosmarinus officinalis, Ranunculus bullatus and Carlina corymbosa; in Asphodelus cerasiferus, Ptilostemon casabonae, Cistus salvifolius, Trifolium sp. and H. italicum) and soil containing mycelium and/or spores collected at Sardinian industrial wastes. Each subplot received 1 L of alive or autoclaved inoculum. In each gram of inoculum, about 700 propagules were present (including mycelium, spores, vesicles and roots colonized by AMF mycelium) as estimated on the basis of most probable number (MPN) of propagules according to Trouvelot et al. (1986).

Soil microbial activity

The community level physiological profiles (CLPP) were generated by means of the Biolog® Microstation System 4.2 (Biolog Inc., CA, USA) using ECOPlates. In order to extract the microbial community, aliquots of 30 g of soil were stirred in 0.1 % sodium pyrophosphate solution (1:5 w/v) with glass beads for 2 h. The slurry was settled, and the supernatant was inoculated into Biolog® ECOPlates. The method is described in Sprocati et al. (2013). Kinetic analyses were performed using average well colour development (AWCD) as a parameter that enables to capture an integral fingerprinting of carbon source utilization. AWCD was calculated as the arithmetic mean of the optical density (OD) values of all of the wells in the plate per reading time. Patterns of substrate utilization for each sample were recorded and analysed.

For the estimation of mycorrhizal colonization, the roots were carefully washed with tap water, softened in 10 % KOH for 24 h, washed in water, acidified in 5 % lactic acid in water for 12–24 h, stained with 0.01 % aniline blue in lactic acid (POCH, Poland) for 24 h at room temperature and eventually stored either in lactic acid or in staining solution. The frequency of mycorrhiza (F, %), relative mycorrhizal root length (M, %), intensity of colonization within individual mycorrhizal roots (m, %), relative arbuscular richness (A %) and arbuscule richness in root fragments where the arbuscules were present (a %) were assessed using a light microscope OPTIMUS (Trouvelot et al. 1986). For each treatment, at least 180 1-cm-long root pieces were analysed.

Soil and plant analyses

X-ray diffraction (XRD) patterns were collected using conventional θ-2θ equipment (Panalytical) with Cu Kα wavelength radiation (λ = 1.54060 Å), operating at 40 kV and 40 mA, using the X'Celerator detector.

For soil analysis, aliquots of approximately 0.3 g of soil sample, exactly weighed, were digested with a mixture of 5 mL of 69 % HNO3 (Aristar, BDH), 1 mL of 70 % HClO4 (RPE, FLUKA) and 1 mL of 30 % H2O2 (Aristar, BDH) in the TFM vessels with a microwave system (Milestone 1200 Mega, Italy). The working programme used for microwave digestion was 5 min at 250 W, 10 min at 400 W, 10 min at 600 W and 5 min at 250 W. The resulting solution was completely transferred into a 50-mL volumetric flask and made up to the final volume with ultrapure water.

The elemental composition of plant sample was determined after digestion with HNO3 (ultrapure 69 % HNO3 TMA-HIPERPUR-PLUS, Panreac) and H2O2 (30 % H2O2 ULTREX® II, J.T.Baker) according to Galletti et al. (2003).

All measurements of trace element concentrations were made with a Perkin-Elmer Elan 6100 ICP-MS spectrometer (USA) equipped with a cross-flow nebulizer for the determination of Cd, Cu and Pb in plant samples, while a Perkin-Elmer Optima 2000 DV ICP–OES spectrometer (USA) equipped with a Scott-type spray chamber was used for Zn analysis in plants and for Cd, Cu, Pb and Zn in soil samples.

To ensure adequate quality assurance, the analytical performance of the laboratory was evaluated by analysing certified reference materials (CRM): 141R (trace elements in calcareous loam soils) for soil analysis and NBS 1571 (Orchard leaves) for plant analysis. The results obtained on CRMs always substantially overlap the certified values.

Ultrapure water (18.2 MΩ cm at 25 °C) obtained from a MilliQ Element system (Millipore, France) was used for preparation and dilution of samples and calibrating solutions. Single-element ICP–MS standard solutions, 1,000 mg L−1 in nitric acid, Aristar (VWR), were used for preparation of calibrating solutions. Rhodium ICP–MS standard solution, 1,000 mg L−1 in nitric acid, Aristar (VWR), was used as the internal standard to correct matrix interferences in ICP–MS analysis. Polypropylene Falcon tubes (Blue Max™) were used during sample handling. All labware used in the experiments was soaked in diluted nitric acid overnight and then rinsed with double deionised water.

The pH reaction grade of soil samples was measured following the Italian official methods for soil analyses (D.M. 13/09/99 http://ctntes.arpa.piemonte.it/Bonifiche/Documenti/Norme/13_Set_99.pdf). Total and organic carbon and nitrogen were determined with an elemental analyzer (Leko CNS 2000, Italy).

Both total and bioavailable soil metal concentrations were evaluated in order to determine the influence of the different treatments. The bioavailable metal content was evaluated using both a rapid test developed in our laboratory (Pinto et al. 2010) and the Italian official method for the measurement of plant available fraction with diethylenetriaminepentaacetic acid (DTPA) extraction (D.M. 13/09/99 http://ctntes.arpa.piemonte.it/Bonifiche/Documenti/Norme/13_Set_99.pdf). The rapid test developed in our laboratory can be briefly described as follows: a buffer solution of trisodium citrate and hydroxylamine hydrochloride was used as extractant for a single-step leaching test. The choice of this buffered solution was strictly related to the possibility of directly determining, via titration with dithizone (DZ), the content of Zn, Cu, Pb and Cd, which are among the most representative contaminants in highly mineralized soils. Moreover, the extraction solution is similar, unless for the pH value, to the one used in the revised standardized sequential extraction procedure second step developed by the Community Bureau of Reference (BCR) of the European Community (Rauret et al. 2000). The analysis of bivalent ions through DZ titration was exploited in order to further simplify and quicken the whole procedure; in fact, the extracted HM concentration can be measured, in a few minutes, as moles per liter, without distinguishing among the different considered elements. So far, results have shown that this screening procedure can give useful and rapid information on HM mobility in substantial agreement with the BCR procedure.

Activities of the hydrolytic enzymes (arylesterase according to Renella et al. (2011); acid and alkaline phosphatase according to Tabatabai and Bremner (1969)) were measured in both treated and untreated soils.

Statistical analysis

In order to study the effect of the different factors on the growth of the plants, ANOVA–simultaneous component analysis (ASCA) (Jansen et al. 2005) was used. This technique can be considered as an extension of the classical analysis of variance (ANOVA) for multivariate data that are originating from an experimental design, and it is particularly useful when the significance of the effect of more than one factor on the experimental data has to be evaluated. In particular, ASCA works by partitioning the variance of the experimental data into the individual contributions induced by the effect of controlled factors, usually a treatment or an experimental condition, or of their interactions; the interpretation of each observed significant factor is accomplished by analysing the resulting matrices by simultaneous component analysis (SCA), which is a decomposition method that, for the purposes of ASCA, can be considered practically identical to principal component analysis (PCA).

In particular, a generic experimental data matrix after mean centering, X m , can be decomposed in the contribution due to the effects of all the controlled factors that were varied during the experiments and to their interactions (Eq. 1).

$$ {\mathbf{X}}_m=\mathbf{X}-\mathbf{1}{\mathbf{m}}^T={\mathbf{X}}_A+{\mathbf{X}}_B+{\mathbf{X}}_C+{\mathbf{X}}_{A\times B}+{\mathbf{X}}_{A\times C}+{\mathbf{X}}_{B\times C}+{\mathbf{X}}_{A\times B\times C}+{\mathbf{X}}_{\mathrm{res}} $$
(1)

where X is the original experimental data matrix, m is a row vector containing the overall average values (grand mean) and the different X k are the partitions corresponding to the main effects, binary interactions, ternary interactions and residuals.

In-house written functions running under Matlab© environment (the MathWorks, Natick, MA; version R2010a) were used for preprocessing, calculation and graphical representation of the data.

Results and discussion

Preliminary biogeochemical characterization of the field trial

Substrate in the test site is representative of the flotation muds discharged in the area until the mine closure in 1968. Grain size and mineralogical composition of the substrate resulted to be substantially constant for all the 27 subplots. Abundance of gravel fraction (<2 mm), deposited during mining activity on top of the mine residues to avoid wind dispersion, has a mean value of 35 ± 5 % in weight. Mean abundance of sand fraction (size ranging between 2 mm and 63 μm) and silt and clay fraction (size <63 μm), all these being residues from the flotation process, was respectively 47 ± 7 % and 17 ± 4 % in weight.

Minerals were detected by XR diffraction in both sand and silt plus clay fractions. Most abundant minerals were quartz and illite, and detectable amounts of sphalerite (ZnS, above the value of ca. 1 % in volume). Other minerals associated to efflorescent salts generated at the flotation mud surfaces under dry conditions were detected by scanning electron microscopy analysis, specifically pyromorphite (Pb5(PO4)3Cl) and calcium sulphate. EDX analysis showed that calcium sulphate contains detectable amounts of Zn, Fe and Pb. Efflorescent salts such as calcium sulphate are indicative of weathering processes occurring. These secondary and highly soluble minerals precipitate at the surface of mine residue banks after strong water mineral interaction and due to the water evaporation. Besides calcium sulphate, efflorescent salts also represent a reservoir of heavy metals that undergo rapid dissolution after watering and/or rain events (Jerz and Rimstidt 2003). These efflorescent salts were also found in roots of plants growing in Naracauli substrate.

Chemical properties of soil at the beginning of the experiments are shown in Table 1. Since the data are the average values of 27 subplots with a relative standard deviation (RSD) of 10–15 %, the field can be considered rather uniform.

Table 1 Soil chemical properties at the beginning of the experiment
Full size table

The HM content was very high, exceeding up to ten times (Zn) the Italian threshold values for soil for industrial use (D.Lgs 152/06 http://94.86.40.85/export/sites/default/archivio/normativa/dlgs_03_04_2006_152.pdf); total carbon, organic carbon and nitrogen contents were very low. The HM mobility, measured with both DTPA and rapid extraction tests, showed that a high fraction of metals can be considered as “bioavailable”; in particular, the HM amount extracted with the rapid test showed a value 10 to 50 times higher than other mining soils (Pinto et al. 2010).

The VM treatment led to an increase of pH (8.5) and an expected decrease (10 %) in HM content due to the diluting effect of VM addition.

Microbial activity in the soil at the beginning of the experiment was very low both as kinetics (untreated soil, US in Fig. 2) and functional diversity (26 %). These data are in accordance with the conditions of the soil, poor in nutrients and organic matter while rich in toxic heavy metals, partially in bioavailable form. Moreover, the basic metabolic fingerprint of the soil microbial community did not show affinity for root exudates, indicating, therefore, that it has not been adapted to the interaction with plants. The area was, indeed, completely barren. The microbial consortium used as bioaugmentation agent was therefore established with autochthonous strains specifically selected to be able to express metabolic functions that could improve the quality of the soil and promote the plant's growth (nitrogen fixation, production of phytohormones and siderophores and mobilization of phosphate). In this way, the metabolic potential needed to facilitate the association between E. pithyusa and the consortium was boosted.

Soil microbial activity, reflecting microbiological processes of soil microorganisms, is a potential indicator of soil quality (Schloter et al. 2003), as plants rely on soil microorganisms to establish and grow. Therefore, if a treatment can establish a good microbial activity in a harsh soil or improve the existing activity, a successful result has been obtained, as the soil quality has been improved. A better quality of the soil, in turn, prepares more favourable conditions for the establishment of a vegetation cover. If the vegetation is able to curb runoff of metals, as well as give strength to the soil and retain water, it plays a major role in environmental remediation (Remon et al. 2005). With this background in mind, the results of soil microbial activity (CLPP) obtained in greenhouse and in field experiments will be analysed.

Assessment of toolbox at greenhouse scale in pot experiment

The test was carried out within a larger experiment, which evaluated the behaviour of six different plants in the Ingurtosu soil, inoculated or not with the native consortium, using commercial soil for cultivated plants as control. Although a few ambiguous results were shown in Wernitznig et al. (2013) with special emphasis on plant physiological parameters, the data set revealed that indigenous E. pithyusa was a well-performing metallophyte species, absorbing elevated amounts in the aerial part of the plant from the original untreated soil, with up to 16.5 mg kg−1 of Cd, 227 mg kg−1 of Cu, 23.1 mg kg−1 of Pb and especially 1,270 mg kg−1 of Zn. Inoculation with the bacterial consortium led not only to a reduction of metal uptake in the aerial part of the plant, but also to a reduced percentage of the exchangeable metals in soil. A positive effect of the bacterial inoculum was observed on germination (100 %) and total survival (80 %) of plants, while in uninoculated soil, plants' germination and survival were 80 % and 40 %, respectively.

An “induced” plant–bacteria association is, however, a complex process that requires time and energy to achieve a balance before showing the real effects. Thus, the signal analysis should not disregard the metabolic processes occurring in the soil before even considering the phenotypic signals. Especially in poor soils influenced by HM, the changes in metabolic activity fingerprint can play an important role (Ortiz-Castro et al. 2009; Ali et al. 2013).

The soil of the Ingurtosu mine used in the greenhouse experiment was characterized by a quantitatively weak metabolic activity (Table 2) while showing a functional diversity of 74 %, the microbial community being capable of using a fairly wide spectrum of substrates. With the introduction of E. pithyusa, the metabolic activity changed during the experimental time for both soil conditions (non-inoculated and inoculated with the bacterial consortium UIC), although in different ways, both as quantity and quality. Namely, in the absence of inoculum, the metabolic activity tended to zero and lost 65 % of the initial functional diversity, while the presence of inoculum sustained the metabolic activity and the loss of functional diversity was negligible if compared with the original soil. The metabolic pattern changed during the pot experiment: the final pattern of inoculated soil showed that the whole community acquired some new metabolic competences, losing a few others. Among the six acquired competences, four were for root exudates: l-asparagine, l-phenylalanine, l-threonine and d-xylose.

Table 2 Metabolic profiles of the microbial communities from uninoculated and inoculated Ingurtosu soil at the beginning (T0) and at the end (15 weeks) of the greenhouse experiment: pattern of substrates utilization, the AWCD and functional diversity
Full size table

In summary, during E. pithyusa germination and growth in the pot experiment, the soil without bioaugmentation has almost completely lost its natural microbial activity, while bioaugmentation with the PGP bacterial consortium UIC allowed the soil to retain a metabolic activity with a high functional diversity. These data are indicators of the metabolic processes that are switched on or off in the soil–plant system and suggest that the microbial community is playing a key role in the possibility of establishing a positive association between the microbial consortium and E. pithyusa in this soil. Above all, the changing of functional diversity indicates that the microbial inoculum is shaping the whole microbial community to support its own growth and to interact with E. pithyusa, in the presence of high concentrations of metals and low concentration of nutrients. Newly acquired metabolic competences include the ability to use root exudates, indicating that the microbial community is adapting to communicate with the plant (Ortiz-Castro et al. 2009). On the other hand it is evident that the composition and concentration of the native microbial community in the presence of E. pithyusa cannot support its survival without bioaugmentation.

These results confirmed the effectiveness of the approach for selecting the inoculum for bioaugmentation, namely the use of a tailor-made consortium selected by ecological criteria, which exploits the metabolic potential of the native microbial community, increasing the competences needed to interact with plants (PGP) and to tolerate HM pollution (Sprocati et al. 2012). On this basis, the field trial was set up.

Early assessment of effects exerted by the toolbox in a field trial

Soil microbial activity

Microbial activity in the soil at the beginning of the experiment was very low. The bioaugmentation with the bacterial consortium produced, soon after the inoculum, an increase in soil activity (Fig. 2). VM addition produced a further decrease of microbial activity, while in the combined treatment (BACT + VM), bacteria slightly mitigated the negative effect of the VM.

Fig. 2
figure 2

Kinetic analysis of the metabolic profile at community level at the beginning of the experiment. Treatments are indicated by the symbols: solid circle, untreated soil (US); solid square, bacteria (BACT); solid triangle, Viromine™ (VM); solid diamond, bacteria and Viromine™ (BACT + VM)

Full size image

Five months after the beginning of the experiment, the soil microbial activity was increased throughout all subplots but in a different degree for the different treatments (Table 3). The overall increase of metabolic activity was clearly attributable to the presence of E. pithyusa plants introduced in each subplot. Data show that the metabolic activity, expressed as AWCD, is higher in the treatments with the bacterial consortium. Anyway, while the parameter AWCD gives a cumulative measure of the metabolic activity, the analysis of the metabolic pattern developed by the microbial community provides more complete information, since it allows to obtain a measure of the functional diversity of the soil (Preston-Mafham et al. 2002) which, in turn, reflects the diversity of the microbial community. In Table 3, we observe that the sole presence of E. pithyusa (FC) has produced in the soil rhizosphere conditions that allowed the development of a microbial flora much richer than before the introduction of the plants (US), increasing the functional diversity by 26 to 81 %. The greater functional diversity (94 %,) however, is always observed where bioaugmentation with the microbial consortium was performed (BACT).

Table 3 Differences in the metabolic profiles of soil microbial communities at the beginning and after 5 months of field experiment
Full size table

Even in the treatments with the VM which had originally a negative effect on microbial activity, the metabolic profile of soil gained a high functional diversity, both in the presence or absence of the bacterial inoculum (87 and 84 %, respectively) as well combined with MYC (84 %). This finding could be related with the increase in pH caused by the addition of VM. The treatments MYC and BACT + MYC produced the lowest values of functional diversity, 74 and 77 %, respectively. Also in this case, the bacterial inoculum slightly improved the value. The hypothesis to explain this result is that mycorrhizal activity in soil escapes the method of investigation specifically aimed at bacterial community. This issue needs therefore to be better investigated.

In each treatment, the resulting microbial community has also developed a greater affinity for the root exudates, increasing from one to eight substrates metabolized, out of nine available in the microplate (Table 3). Only 2-hydroxybenzoic acid was not used by any community, while γ-hydroxybutyric acid was only metabolized by the community developed in the treatment BACT. An increased affinity for root exudates is indicating that communication between bacteria and plants has been established. E. pithyusa has certainly played a major role in this, but we can say that bioaugmentation with the consortium further improved the affinity of the community for plant exudates by showing that they are acting in concert as a toolbox.

The improvement of the spectrum of substrates metabolized by the soil microbial community is an indirect index of the biodiversity of the community: the more varied is the metabolic potential, the more populations are present and the community is richer and more versatile (Sprocati et al. 2013). So, we can say that E. pithyusa was able to positively influence the microbial community in the rhizosphere and that bioaugmentation with the consortium further improved the metabolic potential of the community. As a consequence of the enrichment, the microbial community is more resilient, allowing the plant–soil system to better overcome stress and critical conditions.

Some of these findings do not match with some outcomes of the pot experiment, where it was concluded that the originally present microbial community could not support its survival in the presence of E. pithyusa without bioaugmentation. Actually, those results could suggest, unlike the field, that E. pithyusa had a negative effect on the native microbial community, not being supported by the PGP capacity of the consortium.

The different results between the pot and field experiments are attributable to various factors and, not least, to the diversity of experimental systems, the pot being, unlike the field, a closed system, which prevents exchanges with other systems and then the communication with neighbouring components. Other findings derived from the pot experiment are instead reflected in the field, such as the development of new metabolic pathways including root exudates.

Both experiments confirmed the effectiveness of the bioaugmentation approach, which exploits the metabolic potential of the native microbial community, by increasing the concentration of those microorganisms with targeted metabolic competences that are necessary for the interaction with plants (PGP) and able to interact with heavy metals (tailor-made consortium).

Effect of treatments on plants

Five months after the beginning of the experiment, no significant differences in plant survival were observed among the FC, BACT, MYC and MYC + BACT treatments, where over 90 % of plants were still growing. The presence of VM reduced the plant survival to 60 %, but this negative effect was mitigated by the bacterial inoculum: in the BACT + VM subplots, the plant survival was about 80 %. The high survival rate, together with the physiological parameters (chlorophyll, flavonoids and nitrogen balance index, data not shown) that were comparable with those of PC, confirms the healthy condition of the plants.

This result is encouraging, both for the choice of the plant species and the selected bacteria, and it was not obvious, given that in a field experiment performed in a neighbour mining area with similar characteristics (chemical properties and climatic conditions) using different plant species (Pistacia lentiscus and Scrophularia canina), a survival of less than 50 % was observed after 5 months in the soil without treatment (Bacchetta et al. 2012).

The analysis of fine roots separated from collected plants has shown that all the plants were mycorrhizal and the roots were found to be 100 % mycorrhizal for the whole root length, despite the fact that mycorrhizal inoculum was introduced only in some plots. The plots were established in a place where no plants were present for years. AMF are obligatory biotrophic (Smith and Read 2008) and cannot survive in the absence of host plants for a longer time.

Different explanations could be given about the origin of mycorrhizal fungi in plants from plots that were not inoculated. The simplest explanation is the following: the plantlets were cultivated in garden soil before introduction in the experimental plot (in fact, the plants cultivated in the pots containing garden soil and maintained as positive control in the field had well-developed mycorrhiza). So, plants were already mycorrhizal prior to the experiment. However, fungal morphotype diversity was found, and the higher diversity was found in inoculated plots. In addition, in MYC plots, there were more abundant fine roots with heavily developed arbuscules.

Plants were analysed for their metal content (Fig. 3), and results show that E. pithyusa is naturally able to uptake HM (except for Cu) also in the absence of any treatment (FC). The capacity to absorb Cd in all the conditions, the Zn uptake only in BACT and MYC, and Pb in MYC and VM may be due to specific mechanisms of the plant and to different HM bioavailabilities. No significant differences were found in Cu uptake.

Fig. 3
figure 3

Heavy metal average content (mg·kg−1) in the shoots of plants grown on different treatments. FC field control, BACT bacteria, MYC mycorrhiza, VM Viromine™, BACT + VM bacteria and Viromine™, MYC + VM mycorrhiza and Viromine™, PC pot control. N = 3 subplots per condition; p < 0.05 and p < 0.01 compared to PC are marked with a single asterisk and double asterisk, respectively (Student's test)

Full size image

Statistical analysis

The complete set of soil and plant parameters monitored in the different treatments was analysed by the multivariate ASCA. When considered as a whole, the experiments did not form a complete full factorial design, but a specific three-factor subset did, so data were analysed accordingly. Firstly, the effect of field position, MYC and BACT were examined by considering three levels for the position and two levels each for the other two factors (present/absent). The corresponding 12 experiments formed a full factorial design that could then be examined using ASCA. When ASCA was applied to the resulting data set, none of the main effect of the factors and of their interactions was evaluated as significant at p = 0.05. On the other hand, when the same analysis was repeated on the 12 experiment full factorial designs resulting by selecting the experiments where position, VM and MYC, or position, VM and BACT were present, a significant effect of VM was observed. Moreover, in both cases, also the interaction of VM with the other additive (MYC or BACT, respectively) resulted to be statistically meaningful.

The advantage of using ASCA for assessing the multivariate contribution of controlled factors on data resulting from an experimental design is that, once the original variance is partitioned, interpretation of the observed effects can be interpreted by using PCA on the effect matrices (Smilde et al. 2005). In particular, in the present study, since the only contributions identified as significant were those from VM and its interaction with either MYC or BACT, PCA was applied to the two matrices X vm + X vm+bact (resulting from the analysis of the experiments where VM, BACT and position were controlled) and X vm + X vm+myc (resulting from the analysis of the experiments where VM, MYC and position were controlled).

When the data corresponding to the main effect of VM and from its interaction with BACT were analysed by PCA, the first two components, as expected due to the design, explained 100 % of the total variance. As evident from Fig. 4, where the corresponding scores are reported, the first principal component accounts for the main effect of VM and for that part of the interaction between VM and bacteria which modulates the main effect only in magnitude but does not change its direction. On the other hand, the second component accounts for the effect of the interaction that manifests also as a change in the direction of the VM-induced variation due to the presence or absence of BACT. It should be stressed that this second effect is only a modulation on the main effect of VM evidenced in the first PC.

Fig. 4
figure 4

SCA on the matrix corresponding to the main effect of Viromine™ (vm) and its interaction with the bacteria (bact): projection of the design point onto the first (PC1) and second (PC2) component

Full size image

Inspection of the loadings (supplementary data, Fig. 1S) allows interpreting the observed variation in terms of the variables involved. In particular, as far as the first PC is concerned, it can be stated that the main effect of VM is to induce the increase of pH, organic carbon, total carbon content in the soil and the capacity of metabolizing xylose and glucose-1-phosphate and increasing the AWCD, while at the same time resulting in a decrease of metal mobility measured with both methods (DZ and DTPA). In particular, the reduction of the content of DTPA-extractable Cd and Zn corresponds to a decrease of Cd and Zn content in plants. These data confirm a positive correlation between the two extraction methods applied to the soil and a correspondence with the Zn and Cd plant uptake.

GABA and glycogen give negative contribution. On the other hand, as already stated, the second component mostly accounts for the modulation of the main effect of VM due to the interaction with bacteria. With respect to their value due to the main effect of VM, DTPA-extractable Cu and Cd, phenyl ethylamine and AWCD are higher when both treatments are present, while pH, total carbon, d-galactonic acid lactone, glycyl glutamic acid, glucose-1-phosphate, acid phosphatase and aryl esterase are higher when only one between BACT and VM is considered. The increase of the soil enzymatic activity could be reasonably considered an effect of the decrease of the mobility and, consequently, of the bioavailability of toxic elements due to the addition of the mineral amendment (Koo et al. 2012). The inhibition of the enzymatic activity in a soil is in fact generally attributable to the presence of toxic elements free to bind to enzymes or to the complex enzyme–substrate (Renella et al. 2011). So, higher enzymatic activity in a soil corresponds to a lower toxic potential of the soil, that is, higher enzymatic activity should be intrinsically related to the “health” of the plants living in the soil and to microbial diversity.

The same analysis was repeated on the matrix containing the data for the main effect of VM and its interaction with MYC. Also in this case, all the variation was captured in the first two components, whose scores looked exactly the same as those already shown in Fig. 4. Indeed, the variables contributing the most to the model were almost identical to the VM/BACT data (supplementary materials, Figs. 2S and 3S).

Conclusions

The data set collected from both the greenhouse experiment and the 5-month field trial has identified a number of positive elements that suggest that endemic E. pithyusa and bacterial consortium UIC are compatible and able to establish an association in the soil of the Ingurtosu site. The introduction of E. pithyusa clearly increased soil microbial activity even in the absence of bioaugmentation; nevertheless, the presence of the bacterial inoculum further increased soil quality, expanding the functional diversity and especially the affinity of microbial community towards root exudates showing to act in concert as a toolbox.

E. pithyusa proved to be a well-performing metallophyte species, which is able to absorb, in the aerial part of the plant, Cd, Pb and Zn. The presence of Viromine™ led to a reduction of metal bioavailability especially for Zn and Cd with the corresponding reduction of their uptake in the aerial part of the plant.

In the field trial, after 5 months, 90 % of the plants survived and bacterial inoculum survived as well, despite the cold season. Multivariate analysis supports the conclusion that the proposed toolbox, composed of endemic E. pithyusa and the native UI consortium, can be established in the soil of the mining area of Ingurtosu.

In conclusion, a positive assessment of the applicability of the toolbox designed to be used for microbially assisted phytoremediation can be stated, and therefore, the field trial can be carried forward.

During the field trial maintenance, the effectiveness of the toolbox in limiting the dispersion of metals from the soil to the water system (phytoextraction or phytostabilization) can be tracked, and the convenience of extending the toolbox to other botanical species in order to improve plant coverage will be assessed.