Functional Trait-Based Screening of Zn-Pb Tolerant Wild Plant Species at an Abandoned Mine Site in Gard (France) for Rehabilitation of Mediterranean Metal-Contaminated Soils

The selection of plant species at mine sites is mostly based on metal content in plant parts. Recent works have proposed referring to certain ecological aspects. However, plant traits for plant metal-tolerance still need to be accurately assessed in the field. An abandoned Zn-Pb mine site in Gard (France) offered the opportunity to test a set of ecological criteria. The diversity of micro-habitats was first recorded through floristic relevés and selected categorical and measured plant traits were compared for plant species selection. The floristic composition of the study site consisted in 61 plant species from 31 plant families. This approach enabled us to focus on seven wild plant species naturally growing at the mining site. Their ability to form root symbioses was then observed with a view to phytostabilization management. Four species were considered for phytoextraction: Noccaea caerulescens (J. et C. Presl) FK Meyer, Biscutella laevigata L., Armeria arenaria (Pers.) Schult. and Plantago lanceolata L. The metal content of their aerial and root parts was then determined and compared with that of soil samples collected at the same site. This general approach may lead to the development of a knowledge base for assessment of the ecological restoration trajectory of the site and can help in plant selection for remediation of other metal-rich soils in the Mediterranean area based not only on metal removal but on ecological restoration principles.


Introduction
Heavy metal contaminated soils raise major environmental and human health issues, which may be partially solved by phytoremediation technologies. This cost-effective plant-based approach to remediation takes advantage of the remarkable ability of plants to concentrate trace elements from the environment and to metabolize a large number of molecules in their tissues to reduce their toxicity [1,2]. Consequently, on-site management of heavy metal contaminated soils can be achieved either by using metal hyperaccumulators that are plant species accumulating exceptionally large amounts of heavy metals in their tissues [3] or by using a biocontainment method such as phytostabilization [4].
In most of the cases, plant selection is assessed for commercial phytoremediation (preferentially phytoextraction but not excluding phytostabilization) and is designed on agronomy principles using

Site Location
The study site is located in the town of Rousson in Gard (Figure 1), and more specifically at a place called La Gardie. It is an abandoned mine site which was used for the extraction of zinc. Established in 1876, the Rousson concession (310 ha) included the sites of Landas, Font de Rouve, and Croix de Fauvie (study site). These deposits, located in interlocking limestones, were superficial deposits made up of pieces of shale in ferruginous soil which contained 20 to 40% of zinc. The ore was processed in calamine furnaces. At the time when this small surface mining operation was closed down in 1910, it employed 8 miners and 5 laborers. Today, there are only a few excavations left that spontaneous vegetation colonization is gradually filling up. These red soils, containing wastes from the mine rich in metals, are potentially colonized by a specific flora adapted to metal-rich soils.

Soil Sampling and Metal Analyses
Samples of soil at the mine site were collected randomly from the top 20 cm after litter removal, then sieved to 2 mm and pooled in three composite samples for metal analysis (Cr, Cu, Fe, Mn, Ni, Pb, Zn). In the laboratory, soil composite aliquots were air-dried at room temperature and then ground to pass through a 0.2 mm titanium sieve (RETSCH zm 1000 with tungsten blades) before analyses. Trace and major metal element concentrations were determined on 3 analytical replicates for each of the three soil composite samples. Soils were mineralized in a microwave mineralizer (Milestone Start D) using aqua regia (1/3 HNO3 + 2/3 HCl). The mineralization products were filtered with a 0.45 μm mesh and the metal concentrations were determined by ICP-AES (inductively coupled plasma atomic emission spectroscopy, Spectra 2000, Jobin Yvon Horiba group, Longjumeau, France). Quality controls and accuracy were checked using standard soil reference materials (CRM049-050, from RTC-USA) with accuracies within 100 ± 10%.

Experimental Approach for Plant Selection
The main aim is to assess the efficiency of a trait-based approach for phytoremediation plant selection in the field. To achieve this aim, a hierarchical method was designed based on analysis of literature data with experimental measurements to fill the gaps in knowledge, as described in Figure  2.

Floristic Analysis and Habitat Inventory
The study was done during spring in a defined area of 30 m 2 representative of an excavation in the abandoned mining area with spontaneous vegetation colonization. Vertical (herbaceous, shrub and tree strata) and horizontal (global plant cover) structures of plant communities were used to distinguish the different micro-habitats (Figures 1-3). Within each type of micro-habitat,

Soil Sampling and Metal Analyses
Samples of soil at the mine site were collected randomly from the top 20 cm after litter removal, then sieved to 2 mm and pooled in three composite samples for metal analysis (Cr, Cu, Fe, Mn, Ni, Pb, Zn). In the laboratory, soil composite aliquots were air-dried at room temperature and then ground to pass through a 0.2 mm titanium sieve (RETSCH zm 1000 with tungsten blades) before analyses. Trace and major metal element concentrations were determined on 3 analytical replicates for each of the three soil composite samples. Soils were mineralized in a microwave mineralizer (Milestone Start D) using aqua regia (1/3 HNO 3 + 2/3 HCl). The mineralization products were filtered with a 0.45 µm mesh and the metal concentrations were determined by ICP-AES (inductively coupled plasma atomic emission spectroscopy, Spectra 2000, Jobin Yvon Horiba group, Longjumeau, France). Quality controls and accuracy were checked using standard soil reference materials (CRM049-050, from RTC-USA) with accuracies within 100 ± 10%.

Experimental Approach for Plant Selection
The main aim is to assess the efficiency of a trait-based approach for phytoremediation plant selection in the field. To achieve this aim, a hierarchical method was designed based on analysis of literature data with experimental measurements to fill the gaps in knowledge, as described in Figure 2.

Floristic Analysis and Habitat Inventory
The study was done during spring in a defined area of 30 m 2 representative of an excavation in the abandoned mining area with spontaneous vegetation colonization. Vertical (herbaceous, shrub and tree strata) and horizontal (global plant cover) structures of plant communities were used to distinguish the different micro-habitats (Figures 1-3). Within each type of micro-habitat, representative areas characterized by the homogeneity of vegetation were selected for floristic analysis. The micro-habitats were classified based on plant species composition (presence/absence). Finally, a global analysis on 6 relevés was done on the four selected micro-habitats in terms of plant species occurrence.
representative areas characterized by the homogeneity of vegetation were selected for floristic analysis. The micro-habitats were classified based on plant species composition (presence/absence). Finally, a global analysis on 6 relevés was done on the four selected micro-habitats in terms of plant species occurrence.

Figure 2.
Trait-based grid for local plant selection for phytoremediation of Zn-Pb-rich soils from an ecological restoration perspective.

Bibliographical Functional Traits
Functional traits such as type of root, Grime strategy, Raunkiaer biological type, and mycorrhizal status were collected from a review of scientific papers for each of the identified plant species in the field. The functional response traits were selected with the aim of better understanding the influence that these plant species may have on the passive phytoremediation processes.

Root Symbiosis Assessment
Seven plant species were selected, i.e., Armeria arenaria (Pers.) Schult., Biscutella laevigata L., Brachypodium phoenicoides (L.) Roem. & Schult, Trifolium pratense L., Mibora minima (L.) Desv., Noccaea caerulescens (J. et C. Presl) FK Meyer and Thymus vulgaris L. for assessment of their root symbioses in the field. The selection was done after literature review, and only plant species for which there was a lack of knowledge, scarce data or controversial status were chosen. Thin roots were randomly selected in the root system of 3 individuals of each of the selected plant species. The root samples were first rinsed under tap water then with deionized water and stored in alcohol (60%, v/v) at room temperature until proceeding. The occurrence of symbiont colonization was estimated by visual observation of fungal colonization after clearing roots in 10% KOH and staining with lactophenol blue solution, according to Phillips and Hayman [15].

Bibliographical Functional Traits
Functional traits such as type of root, Grime strategy, Raunkiaer biological type, and mycorrhizal status were collected from a review of scientific papers for each of the identified plant species in the field. The functional response traits were selected with the aim of better understanding the influence that these plant species may have on the passive phytoremediation processes. for assessment of their root symbioses in the field. The selection was done after literature review, and only plant species for which there was a lack of knowledge, scarce data or controversial status were chosen. Thin roots were randomly selected in the root system of 3 individuals of each of the selected plant species. The root samples were first rinsed under tap water then with deionized water and stored in alcohol (60%, v/v) at room temperature until proceeding. The occurrence of symbiont colonization was estimated by visual observation of fungal colonization after clearing roots in 10% KOH and staining with lactophenol blue solution, according to Phillips and Hayman [15].

Metal Content in Below-Ground and Above-Ground Plant Parts
Four plant species were selected for metal analysis, i.e., N. caerulescens, B. laevigata, A. arenaria and Plantago lanceolata L. Dried plant samples (root and aerial parts, separately) were ground to pass a 0.2 mm mesh titanium sieve, and three aliquots were analyzed per sample. About 0.5 g of dry matter was digested with the microwave digestion system Milestone start D with a HNO 3 -HCl mixture (volume proportion ratio 2/1). After filtration (0.45 µm), acid digests were analyzed for metal content by ICP-AES (JY 2000 Jobin Yvon Horiba group, Longjumeau, France). Standard plant reference material (DC 73,349) from the China National Analysis Centre for Iron and Steel (NCS), was analyzed as a part of the quality control protocol (accuracies within 100 ± 10%).

Statistical Analysis
Statistical analyses were performed for all data using JMP 12 statistical software (SAS Institute, Cary, NC, USA) using the non-parametric Kruskal-Wallis test because data did not follow a normal distribution pattern. The nonparametric pairwise multiple-comparison Dunn's test was used when the null hypothesis was rejected with the Kruskal-Wallis test.

Soil Contamination
Soils contained high concentrations of zinc (ca.11.7%) and lead (ca.1.7%) as expected in this type of abandoned mining area (Table 1). Moreover, elevated concentrations in Fe, Mn, and Cr were also detected. Zn and Pb soil concentrations were higher than those previously reported in mine tailings in Spain, i.e., highest concentrations ranging between ca. 2360, 7000 and 11,600 mg/kg of Zn in the Alcudia Valley, San José heaps and El Lirio tailing, respectively and ca. 7000 and 10,000 mg/kg of Pb in Belleza tailing and the Alcudia Valley, respectively [16][17][18]. However higher Zn soil content was detected at Mánforas with 135,080 mg/kg [17]. In Portugal, up to 9330 mg/kg of Pb were observed in abandoned Pb mines [19]. Compared to mine sites in France, at the Les Malines and Les Avinières near the present study site, the authors detected lower concentrations in Zn and greater in Pb, i.e., 59,040 mg/kg of Zn and 62,051 mg/kg of Pb [20]. In Poland, in reclaimed soils of Zn-Pb mine wastes, up to 7.57% of Zn and 0.46% of Pb were found [21]. Such heavy metal-rich soils may not be used for agriculture purposes, and the spontaneous plant colonization that occurred after operation ceased was a great opportunity to reduce the transfer of elements into the food web and to reduce the environmental and human exposure.

Diversity of Habitats and Plant Species
A total of 4 micro-habitats were identified after preliminary cartography of the 30 m 2 study site, consisting of (i) grassland colonizing metal-rich soil, (ii) herbaceous colonization of rocky soils, (iii) matorral with shrub dominating and (iv) woody vegetation ( Figure 3). Two supplementary micro-habitats were removed from the inventory. The first concerned an area in which the major part was covered by lichens belonging to 2 species, i.e., Cladonia rengiformis and Cladonia foliacea s.l., beyond the scope of the study. A second area near the road was mostly constituted of ruderals due to the modified substrate of the road foundation.
The floristic composition of the study site consisted in 61 plant species growing wild on these 4 micro-habitat types and identified during the spring period (Table 2). It included 31 botanical families at the period of the relevés. In terms of species diversity, the Poaceae family was the most frequent (8 species), then the Brassicaceae and Rosaceae with 5 species each. Asteraceae occurred with 3 species, followed by Fabaceae, Caryophyllaceae, Rubiaceae and Ranunculaceae (Table 2).

Analysis of Plant Traits Linked with TMM Tolerance
From the literature, different plant traits commonly used in different floristic analyses of spontaneous colonization of mining sites were assessed for all the 61 identified plant species (Table A1). Those categorical traits may help us to understand the distribution pattern of plant species along environmental gradients [22]. However, their local variability needs to be discussed, and continuous traits need to be locally assessed to test the robustness of this approach.

Categorical Trait Analysis of the 61 Identified Plant Species
Considering below-ground parts, 61% of the plant species were characterized by fibrous root systems, 26% by rhizome or tuberous root system and only 13% by slender roots. A hypothesis has been formulated by Sardans and Peñuelas [23] and confirmed by Pierret et al. [24] considering deep roots as able to access water and nutrients in deep layers of soil and/or fractured bedrock, which are unavailable to surface roots. This mechanism will potentially help to maintain higher moisture levels in the upper soil layers and could be a factor explaining the high plant diversity in these dry habitats, despite the water stress due to both Mediterranean conditions and high salt concentration in the water linked to the metalliferous soils.
The mycorrhizal status of most of these 61 plant species has already been described [25][26][27]. Out of all the plant species, 67% were associated with arbuscular endomycorrhizal fungi, ca. 5% with ectomycorrhizal fungi and 8% were reputed to share no mycorrhizal interactions. For ca. 20% of the identified plant species, no information regarding their mycorrhizal status has been reported to the best of our knowledge.
Concerning the preferential type of soil, 44% of the plant species were adapted to dry and rocky soils, ca. 20% specific to calcareous soils, ca. 20% preferred sandy or well-drained soils and 16% were mostly found in disturbed or cultivated soils.
The dominant life form ( Figure 4) were hemicryptophytes with an average percentage of 51%, then therophytes (19%) and phanerophytes (15%). Chamaephytes only represented an average of 9%. Hemicryptophytes and therophytes were found in all of the 6 relevés although the occurrence of phanerophytes varied with none of this life form in the center of the area (open dry grasslands) and 31% in the tree stand (Tables 2 and A1). Geophytes and lianas were only identified in 2 and 3 out of 6 relevés, respectively. been formulated by Sardans and Peñuelas [23] and confirmed by Pierret et al. [24] considering deep roots as able to access water and nutrients in deep layers of soil and/or fractured bedrock, which are unavailable to surface roots. This mechanism will potentially help to maintain higher moisture levels in the upper soil layers and could be a factor explaining the high plant diversity in these dry habitats, despite the water stress due to both Mediterranean conditions and high salt concentration in the water linked to the metalliferous soils.
The mycorrhizal status of most of these 61 plant species has already been described [25][26][27]. Out of all the plant species, 67% were associated with arbuscular endomycorrhizal fungi, ca. 5% with ectomycorrhizal fungi and 8% were reputed to share no mycorrhizal interactions. For ca. 20% of the identified plant species, no information regarding their mycorrhizal status has been reported to the best of our knowledge.
Concerning the preferential type of soil, 44% of the plant species were adapted to dry and rocky soils, ca. 20% specific to calcareous soils, ca. 20% preferred sandy or well-drained soils and 16% were mostly found in disturbed or cultivated soils.
The dominant life form ( Figure 4) were hemicryptophytes with an average percentage of 51%, then therophytes (19%) and phanerophytes (15%). Chamaephytes only represented an average of 9 %. Hemicryptophytes and therophytes were found in all of the 6 relevés although the occurrence of phanerophytes varied with none of this life form in the center of the area (open dry grasslands) and 31% in the tree stand (Table 2 and A1). Geophytes and lianas were only identified in 2 and 3 out of 6 relevés, respectively.

Analysis of Plant Traits by TMM Tolerance Strategy
The identified plant species were grouped in hyperaccumulators, phytoaccumulators, phytostabilizators or not known in the literature. Using this grid of comparison, the above-cited traits were analyzed ( Figure 6). Even if the number of species considered as hyperaccumulators (2) or as phytoaccumulators (3) is limited in the study field, some traits are typical of the plant strategy. The dominant root system was fibrous for the two hyperaccumulators ( Figure 6a) and deep root system for phytoaccumulators (Figure 6b). No specific root type could be defined for phytostabilizators (Figure 6c). For phytostabilization strategy (43), identified plant species in the study site had mainly rhizomes or tuberous roots (35%, Figure 6c), ca. all of them were known to be endomycorrhized (78%, Figure 6g), the dominant Grime strategy was CS (49%, Figure 6k), and the main life form was hemicryptophyte (46%, Figure 6o). These results are in accordance with many previous studies demonstrating that only a few spontaneous plant species colonizing mine sites may favor heavy metal translocation in the aerial parts and may be useful for phytoextraction. Most of the metal tolerant plant species may accumulate heavy metals in their roots and limit their transfer to the aerial parts, being potentially useful in phytostabilization. Without any human intervention by plant harvesting, the dominant natural process is therefore phytostabilization. Hemicryptophyte life form was strongly present in the three groups (Figure 6m-o). Hemicryptophytes such as biannuals or with thin root systems may limit the phytostabilization potential. On the other hand, the non-perennial aerial parts of hemicryptophytes may also limit the phytoextraction ability. The traits that most discriminated hyperaccumulators from both phytoaccumulators and phytostabilizators were the ability to form arbuscular mycorrhizal (AM) associations (50% non-mycorrhizal for the first and 33% and 78% AM for the other two; Figure 6e, f and g). It is noteworthy that ectomycorrhizal (EC) type was dominant (67%) in phytoaccumulators. It is congruent with the 67% of phanerophyte type. Quercus ilex and Pinus sylvestris are known to be predominantly associated with ectomycorrhizal fungi [26].

Analysis of Plant Traits by TMM Tolerance Strategy
The identified plant species were grouped in hyperaccumulators, phytoaccumulators, phytostabilizators or not known in the literature. Using this grid of comparison, the above-cited traits were analyzed ( Figure 6). Even if the number of species considered as hyperaccumulators (2) or as phytoaccumulators (3) is limited in the study field, some traits are typical of the plant strategy. The dominant root system was fibrous for the two hyperaccumulators ( Figure 6a) and deep root system for phytoaccumulators (Figure 6b). No specific root type could be defined for phytostabilizators (Figure 6c). For phytostabilization strategy (43), identified plant species in the study site had mainly rhizomes or tuberous roots (35%, Figure 6c), ca. all of them were known to be endomycorrhized (78%, Figure 6g), the dominant Grime strategy was CS (49%, Figure 6k), and the main life form was hemicryptophyte (46%, Figure 6o). These results are in accordance with many previous studies demonstrating that only a few spontaneous plant species colonizing mine sites may favor heavy metal translocation in the aerial parts and may be useful for phytoextraction. Most of the metal tolerant plant species may accumulate heavy metals in their roots and limit their transfer to the aerial parts, being potentially useful in phytostabilization. Without any human intervention by plant harvesting, the dominant natural process is therefore phytostabilization. Hemicryptophyte life form was strongly present in the three groups (Figure 6m-o). Hemicryptophytes such as biannuals or with thin root systems may limit the phytostabilization potential. On the other hand, the non-perennial aerial parts of hemicryptophytes may also limit the phytoextraction ability. The traits that most discriminated hyperaccumulators from both phytoaccumulators and phytostabilizators were the ability to form arbuscular mycorrhizal (AM) associations (50% non-mycorrhizal for the first and 33% and 78% AM for the other two; Figure 6e, f and g). It is noteworthy that ectomycorrhizal (EC) type was dominant (67%) in phytoaccumulators. It is congruent with the 67% of phanerophyte type. Quercus ilex and Pinus sylvestris are known to be predominantly associated with ectomycorrhizal fungi [26].  Table A1).

Measured Trait Analysis on Selected Plant Species
Mycorrhizal status of 7 selected plant species: an efficient tool for plant species selection for phytostabilization strategy?
N. caerulescens (Figure 7a) and B. laevigata (not shown) were not mycorrhized at this site. However, there is no consensus regarding the mycorrhizal status of B. laevigata and N. caerulescens, long considered not to be mycorrhized; some authors detected AM associations with both species [26][27][28].
A total of 20% of root length colonization by AM fungi was detected for A. arenaria (Figure 7b) although no AM colonization was revealed for M. minima (not shown). No previous data regarding the mycorrhizal status of A. arenaria and M. minima have been published. To the best of our knowledge, this is the first report of AM association with A. arenaria. With regard to M. minima, the smallest Poaceae occurring in France, it is also the first report concerning its mycorrhizal status. Its very short life cycle (few months) could be linked to a lack of AM association.
Wang and Qiu [26] and Pawlowska et al. [27] gave discordant results for T. pratense. At the present site, AM mycelia were abundant, and many spores were detected in roots of T. pratense with an overall colonization ca. 80% of root length (Figure 7c).
B. phoenicoides mycorrhizal status has previously been analyzed [29]. However, data are scarce. In our study, this species appeared to be endomycorrhized (Figure 7d). T. vulgaris was the only Lamiaceae identified at our study site and since Lamiaceae are usually good candidates for phytostabilization, we endeavored to confirm its mycorrhizal status. This perennial developed AM association and a dense web of mycelia was observed with many vesicles (Figure 7e).    Table A1).

Measured Trait Analysis on Selected Plant Species
Mycorrhizal status of 7 selected plant species: an efficient tool for plant species selection for phytostabilization strategy?
N. caerulescens (Figure 7a) and B. laevigata (not shown) were not mycorrhized at this site. However, there is no consensus regarding the mycorrhizal status of B. laevigata and N. caerulescens, long considered not to be mycorrhized; some authors detected AM associations with both species [26][27][28].
N. caerulescens (Figure 7a) and B. laevigata (not shown) were not mycorrhized at this site. However, there is no consensus regarding the mycorrhizal status of B. laevigata and N. caerulescens, long considered not to be mycorrhized; some authors detected AM associations with both species [26][27][28].
A total of 20% of root length colonization by AM fungi was detected for A. arenaria (Figure 7b) although no AM colonization was revealed for M. minima (not shown). No previous data regarding the mycorrhizal status of A. arenaria and M. minima have been published. To the best of our knowledge, this is the first report of AM association with A. arenaria. With regard to M. minima, the smallest Poaceae occurring in France, it is also the first report concerning its mycorrhizal status. Its very short life cycle (few months) could be linked to a lack of AM association.
The occurrence of AM symbioses is a strong advantage for phytostabilization but not specific to this type of process. Mycorrhizal associations may favor TMM extraction by enhancing metal mobilization [30]. Therefore, this trait alone would not be a good criterium for plant species selection for the purpose of phytostabilization strategy. Plant colonization at mining sites may be favored by AM fungi, the latter, which may lower metal toxicity and improve nutrient availability for plants. However, in a Zn-, Cd-, Pb-, and Cupolluted field study, no evidence for an effect of AM symbioses has been found on plant metal uptake [31]. Therefore, with regard to a phytoextraction strategy, the authors suggest not channeling efforts exclusively towards reinforcing AM symbiosis. In a recent review paper dealing with AM associations at mining sites, it was observed that more than 80% of the plant species from metallic mines were endomycorrhized suggesting adaptive strategies in coevolved fungal strains and plant  Wang and Qiu [26] and Pawlowska et al. [27] gave discordant results for T. pratense. At the present site, AM mycelia were abundant, and many spores were detected in roots of T. pratense with an overall colonization ca. 80% of root length (Figure 7c).
B. phoenicoides mycorrhizal status has previously been analyzed [29]. However, data are scarce. In our study, this species appeared to be endomycorrhized (Figure 7d). T. vulgaris was the only Lamiaceae identified at our study site and since Lamiaceae are usually good candidates for phytostabilization, we endeavored to confirm its mycorrhizal status. This perennial developed AM association and a dense web of mycelia was observed with many vesicles (Figure 7e).
The occurrence of AM symbioses is a strong advantage for phytostabilization but not specific to this type of process. Mycorrhizal associations may favor TMM extraction by enhancing metal mobilization [30]. Therefore, this trait alone would not be a good criterium for plant species selection for the purpose of phytostabilization strategy.
Plant colonization at mining sites may be favored by AM fungi, the latter, which may lower metal toxicity and improve nutrient availability for plants. However, in a Zn-, Cd-, Pb-, and Cu-polluted field study, no evidence for an effect of AM symbioses has been found on plant metal uptake [31]. Therefore, with regard to a phytoextraction strategy, the authors suggest not channeling efforts exclusively towards reinforcing AM symbiosis. In a recent review paper dealing with AM associations at mining sites, it was observed that more than 80% of the plant species from metallic mines were endomycorrhized suggesting adaptive strategies in coevolved fungal strains and plant species [32]. Rehabilitation of metal-rich soils without metal removal may be achieved by selecting plant species with their co-evolved mycorrhizal symbionts.

Metal Content in 4 Plant Species for Phytoextraction Strategy
In agreement with the literature, N. caerulescens was the best accumulating species out of the 4 selected for zinc and lead (Table 3), and it is defined as a facultative Zn-hyperaccumulator [3]. This species also accumulated high levels of Fe in its aerial parts. Furthermore, B. laevigata and A. arenaria seemed to be suitable as Zn-phytoaccumulators with more than 1000 mg/kg (dry matter, DM) of zinc accumulated in their aerial parts. Those species have already been identified as valuable temperate zone phytoaccumulators of Zn and Cd. P. lanceolata, in a lesser way, appeared as a good Zn phytoaccumulators. Zn content up to 430 mg/kg in shoots of P. lanceolata was previously reported in a mining area of Southern Poland [33] although 946 mg/kg were detected in P. lanceolata aerial parts in the present study. All these four plant species are also potentially good candidates in Mediterranean areas. Although average soil Pb concentration was high (Table 1), this element was moderately accumulated in the aerial parts of N. caerulescens (ca. 496 mg/kg) and ranged from 48 to 75 mg/kg in the aerial parts of the three other plant species. Apart from Fe, Zn and Pb were the studied elements most translocated to the aerial parts of the four plant species. However, Pb, as a non-essential element, may be transferred to aerial parts via transporters of other elements essential to plants. This type of elemental competition between nutrients and toxic metals is most of the time antagonistic. However, the present results show concomitant translocation of Zn and Pb. Our results demonstrated similar results for N. caerulescens, A. arenaria, B. laevigata and P. lanceolata. N. caerulescens is sometimes considered as a Zn-hyperaccumulator and Pb co-accumulator [34]. A previous work on N. caerulescens also shows a negative correlation between Zn content in shoot parts and Ca and Mg concentrations in shoot parts due to competition between these different cations [13]. Synergistic and antagonistic interactions for element absorption in plants is a challenging field of research and there is still a need for more knowledge. However, these preliminary results showed the interest of these plant species in Zn and Pb-rich soils.  [17,19,20,27,[35][36][37] enabling a comparison of occurrence of plant species even though these studies were conducted under varying biogeographical conditions. Out of the seven relevés from the cited literature, Rubus ulmifolius Schott and P. lanceolata were the most frequently identified species on these sites. B. leavigata, D. glomerata and F. ovina occurred in 4 out of the 7 relevés in France, Poland and/or Portugal. N. caerulescens only occurred in the 2 relevés from France. Hemicryptophytes were dominant in most of the studies [20,38] including in the present study. No systematic selection could be made solely on the basis of plant traits, but common features are shared by the spontaneous vegetation of the European and Mediterranean abandoned Zn-mine sites that it might be useful to identify with a view to the ecological restoration of Zn-rich soils.
It is worth noticing that, in many studies, the occurrence of certain exotic phanerophytes was recurrent, such as Eucalyptus globulus [19], but these species are not to be encouraged in ecological restoration processes; it would be better to focus on native plant species.
On the basis of all the information collected through the diverse relevés, it appeared that some ubiquitous Zn-tolerant plant species in Mediterranean and European areas are potentially good candidates for the first stages of ecological restoration of Zn-Pb-rich soils. Previous restoration programs in Poland have used D. glomerata and T. pratense, also identified at the present study site. However, this was done with selected cultivars and not seedlings of wild origin [39]. This previous integrative study has suggested various potential lines of research including endomycorrhizal inoculation before transplantation with selected fungal strains and also selection of xerothermic plant species to cope with water stress. Their feedback shows how important the development of rhizosphere microorganisms consortia is for successful site restoration [39]. AM colonization was also a dominant trait as previously reported [27]. However, local plant associations should be favored and the geographical situation needs to be taken into account. Moreover, more than the below-ground parts of the plants, rhizosphere micro-organisms are potentially the key factor for plant colonization in these metal-rich soils following the mycorrhizal fungal diversity-ecosystem function concept of Hazard and Johnson [40]. Plant functioning (nutrition, stress-tolerance, etc.) is intrinsically dependent on associated micro-organisms [41].

Conclusions
This work enabled the implementation of a plant database consisting of above-ground and below-ground plant traits of plant species able to grow on Zn-Pb rich soils. This preliminary work may be a useful tool for plant selection in rehabilitation programs for Zn-Pb-rich soils. It also highlighted the need for more information regarding below-ground traits and rhizosphere microbial consortia.
The trait-based analysis provided a basis for drawing a general picture of plant communities in a Mediterranean abandoned Zn-Pb mine site. Below-ground traits appeared as important features for phytostabilization. Ectomycorrhizae were dominant in the Zn-phytoaccumulators species and AM, in the phytostabilizators. Together, these plant strategies may favor fungal interaction diversity and enhance the sustainability of the plant-fungal communities.
The four plant species selected for phytoextraction, i.e., N. caerulescens, B. laevigata, A. arenaria and P. lanceolata showed interesting Pb and Zn accumulation capacities in their aerial parts. However, these plant species are of reduced biomass and a phytoextraction process with those species would not be efficient. Those plant species are however interesting in the early plant succession stage after mines are abandoned. Among the identified potential phytoaccumulators, Quercus ilex and Pinus sylvestris, both phanerophytes, are long-term colonizers and persistent plant species. However, management strategies based on tree plantation would create more litter with reduced undergrowth. With long-term Zn potential accumulation in the aerial parts, there is a need for further knowledge regarding the risk of Zn transfer into the food web.
Plant communities at the mine site mostly favored a passive phytostabilization that is maintained over time by seasonal turnover of therophytes and persistence of belowground parts of hemicryptophytes and geophytes and both belowground and aboveground parts of chamaephytes and phanerophytes. Only few plant species were potentially able to accumulate Zn and Pb in the aerial parts. Moreover, a phytoextraction process requires human intervention by removal of metal-rich aerial parts. Consequently, even if both strategies occurred in these plant communities, the overall trend is the immobilization of heavy metals in the soils and root systems.