Life in mine tailings: microbial population structure across the bulk soil, rhizosphere, and roots of boreal species colonizing mine tailings in northwestern Québec

Abstract

Purpose

Mining activities have negative effects on soil characteristics and can result in low pH, high heavy metal content, and limited levels of essential nutrients. A tailings storage area located in northwestern Québec showed natural colonization by plants from the adjacent natural environment. The objective of the study was to determine the main edaphic parameters that structured microbial populations associated with the indigenous woody plants that had naturally colonized the site.

Methods

Microbial populations were studied in the bulk soil, the rhizosphere, and inside plant roots using Illumina sequencing, ordination analysis (i.e., redundancy analysis (RDA) and principal coordinates analysis (PCoA)), ternary plotting, and statistical analysis (MANOVA).

Results

The main variables that drove the microbial community patterns were plant species and the tailings pH. Indeed, the main bacterial classes were Gammaproteobacteria and Deltaproteobacteria in both the rhizosphere and root endosphere. Analysis revealed that some dominant operational taxonomic units (e.g., Pseudomonas sp., Acinetobacter sp., and Delftia sp.) were present in increased proportions in roots for each plant species under study. This study also revealed that many of the most abundant fungal genera (e.g., Claussenomyces, Eupenicillium, and Trichoderma) were more abundant in the rhizosphere than in the root endosphere.

Conclusions

This comprehensive study of the microbial community dynamics in the bulk soil, rhizosphere, and root endosphere of boreal trees and shrubs could be beneficial in facilitating the rehabilitation of disturbed ecosystems.

Introduction

In Québec, more than 100 years of mining activities have etched into the province. Since 1890, many metals have been mined in this province, most notably iron, nickel, zinc, copper, and gold, leaving hundreds of abandoned mines sites to be restored with public funds (Ministère de l'Énergie et des Ressources Naturelles (MERN) 2015, Ministère de l'Énergie et des Ressources Naturelles (MERN) 2017, Ministère de l'Énergie et des ressources Naturelles (MERN) 2018). Unfortunately, once exploited, mine sites have many common features that hinder primary succession and slow down the autonomous recovery of disturbed ecosystems. Extreme soil characteristics such as acidic pH, high concentrations of heavy metals, and little or no organic content make it difficult for the establishment of primary successional plants (Sheoran et al. 2010). In this context, it is essential for the plants to associate with soil microorganisms that can promote the growth of the plant, such as PGPR (i.e., plant growth-promoting rhizobacteria), which provide several benefits to plants (Van Der Heijen et al. 2008; Ma et al. 2011). Microbial communities (e.g., bacteria, ectomycorrhizal (ECM) fungi, arbuscular fungi (AM), and dark septate endophytes (DSEs)) living in the plant rhizosphere can also alleviate the phytotoxicity of metals through various mechanisms. They can exclude some metals via restricted membrane permeability or by actively transporting metals outside the cell. They can sequester metals inside the cell and detoxify metals by modifying their speciation (Ma et al. 2011). These microbial communities may also enhance the mobilization and uptake of nutrients by plants (Courty et al. 2010). For their part, plants are able to recruit specific microorganisms to populate the rhizosphere and roots by modulating the secretion of molecules in their root exudates (i.e., sugars, hormones, amino acids, organic acids, etc.). This modulation is in turn influenced by many plant-specific factors, such as species, age and developmental stage, and environmental conditions (Turner et al. 2013). In the context of mine site reclamation, PGPR microorganisms associated with pioneer plants could improve phytoremediation success (Zhuang et al. 2007; Hrynkiewicz and Baum 2011).

The studied mine site, located in the gold-rich region of northwestern Québec, has been naturally colonized by native boreal species (such as alder, birch, and spruce) for over five decades, and it has heterogeneous edaphic characteristics (i.e., plant species, pH, metals). This site is a natural laboratory that enabled us to address a variety of biological inquiries, notably identification of the main drivers for microbial population composition in the bulk soil, rhizosphere, and root endosphere of boreal plant species colonizing the mine tailings.

The main objective of this study was to improve our understanding of plant-microbe systems on acidogenic tailings to assist in restoration efforts. We aimed (1) to determine which environmental factors best explain the microbial communities found in soil, the rhizosphere, and roots; (2) to determine if the microbial population distribution patterns differed between soil, the rhizosphere, and roots for the same plant species; and (3) to determine which microorganisms are significantly more present in soil, the rhizosphere, and roots for a given plant species. Based on the known importance of plant species, pH, and water content which condition the structure of microbial communities in soil, our first hypothesis is that these parameters are the principal drivers of microbial communities in soil in the tailings deposit despite high metal contents (1) (Lauber et al. 2009; Brockett et al. 2012; Hodge and Fitter 2013). Our second hypothesis is that differences in microbial community composition would be observed between plants (this is well-established), but that greater differences in microbial community composition would be observed between root endophytes and the external environments (bulk soil and rhizosphere) of individual plant species due to selective pressure (low pH and metal content) on plants to recruit specific root endophytes to help attenuate environmental stress (2) (Lundberg et al. 2012). Our third hypothesis is that the roots will have the highest relative abundance for the most abundant operational taxonomic units (OTUs) but that the highest diversity (Shannon index) for bacteria and fungi will be found in the rhizosphere (3) (Lundberg et al. 2012; Edwards et al. 2015). A better understanding of the microbial community dynamics between soil, rhizosphere, and roots in a mining context could help us to identify key microorganisms in these recovering environments.

Materials and methods

Study site and experimental design

The mining site is located in northwestern region of Québec (Canada), known for its gold deposits (Fig. 1a). In this region, the forest is mainly composed of balsam fir (Abies balsamea), black spruce (Picea marinara), white spruce (Picea glauca), paper birch (Betula papyrifera), and Jack pine (Pinus banksiana) (Kneeshaw and Bergeron 1998; Bergeron et al. 2004). This mining site has been less active for the last five decades than it was previously, and natural and heterogeneous vegetation can be observed on all of this tailings deposit, despite the presence of acid mine drainage (generated by mineral sulfur reacting to atmospheric oxygen) and subsequent release of heavy metals from the mineral matrix. The patterns of its plant community development and structure were recently published, describing heavy metal and pH distribution patterns on the site and their influence on plant communities (Tardif et al. 2019). Speckled alder (Alnus incana ssp. rugosa), paper birch, and spruce (Picea sp.) were present in this vegetation (Fig. 1b). For our study, approximately 100 root and soil samples for each species were collected randomly over the mine site at the beginning of September 2015.

Fig. 1

a Maps of meridional Québec with mine site region (black box) (Ministère de l'Énergie et des Ressources Naturelles (MERN), 2013). b Alder, spruce, birch, and mix of species on the mine site

Bulk soil, rhizosphere, and root sampling and preparation for DNA extraction

Plants (approximately 0.5 m in height) were dug with a shovel to harvest the soil around the roots. Soil (50 g) was taken from the root ball (bulk soil). The aboveground and underground parts of the plant were separated into two different Whirl-Pak® bags and kept at 4 °C until they arrived at the laboratory. Most of the soil surrounding the roots was shaken free to preserve only the rhizosphere and the roots, which were shaken in a beaker to collect the attached (i.e., rhizosphere) soil. Before freezing, roots were washed and surface-sterilized for endophyte DNA extraction. The roots were washed for 1 min in 100% ethanol, 1 min in 2.5% bleach, 30 min in fresh 2.5% bleach, and 1 min in 100% ethanol. Roots were then aseptically transferred to sterile Erlenmeyer flasks and rinsed four times with sterile distilled water. An aliquot of water from the final wash was preserved to perform 16S rRNA amplification to confirm sterility (results not shown). Bulk soil, the rhizosphere, and roots were frozen at − 20 °C until DNA extraction.

Environmental parameters analysis

The environmental parameters studied in soils were pH, water content, total N, organic matter, sodium (²³Na), magnesium (²⁴Mg), aluminum (²⁷Al), phosphorus (³¹P), potassium (³⁹K), calcium (⁴⁴Ca), titanium (⁴⁷Ti), vanadium (⁵¹V), chromium (⁵²Cr), manganese (⁵⁵Mn), iron (⁵⁷Fe), cobalt (⁵⁹Co), nickel (⁶⁰Ni), copper (⁶⁵Cu), zinc (⁶⁶Zn), arsenic (⁷⁵As), selenium (⁸²Se), molybdenum (⁹⁵Mo), silver (¹⁰⁷Ag), cadmium (¹¹¹Cd), antimony (¹²¹Sb), barium (¹³⁷Ba), tungsten (¹⁸²W), thallium (²⁰⁵Tl), and lead (²⁰⁶Pb). The results for alder, birch, and spruce bulk soil samples are presented in Supplementary Table 1.

DNA extraction, library preparation, and sequencing

Two grams of frozen bulk soil or the rhizosphere were weighed in tubes containing 1.5 g of silica beads. DNA was extracted, purified, and eluted as described in the PowerSoil® Total RNA Isolation Kit (MO BIO Laboratories, Inc., #12866-25, Carlsbad, CA, USA) and RNA PowerSoil® DNA Elution Accessory Kit (MO BIO Laboratories, Inc., #12867-25, Carlsbad, CA, USA). Root DNA extraction was performed according to the manufacturer’s instructions (PowerLyzer® UltraClean® Tissue & Cells RNA Isolation Kit, MO BIO Laboratories, Inc., #15055-50, Carlsbad, CA, USA). The V4 variable region of the 16S rRNA ribosomal DNA gene sequences (eubacteria and archaea) and the fungal nuclear ribosomal internal transcribed spacer (ITS) region were first amplified. The specific primer sequences (Integrated DNA Technologies, Canada) for these regions are described in Supplementary Table 2. For each region (16S rRNA gene and ITS), four pairs of primers with additional bases inserted between the sequence overhang and the specific sequence (staggered pad) were used proportionally between the samples to create diversity in the sequence reads, thereby improving the quality of sequencing results. For amplification of the 16S rRNA region, a peptide nucleic acid (PNA) clamp was used to prevent amplification of eukaryotic RNA. PNA chloroplast and PNA mitochondrial blockers were used as proposed in Hong et al. (2016). Reverse transcriptase PCR was carried out using the OneStep RT-PCR Kit (Qiagen, #220211, CA, USA).

Amplicons were visualized on a 2% agarose gel. The samples were purified with AMPure XP magnetic beads (Beckman Coulter Genomics, #A63881, CA, USA), and a second short 8-cycle amplification step was carried out using the Nextera XT index kit (Illumina® #FC-131-1001, set A-B-C and D, CA, USA) to barcode the amplicons according to the protocol in the Illumina “16S Metagenomic Sequencing Library Preparation” guide (part #15044223 Rev. B). Unique codes were added to each sample by amplifying 5 μl of the purified PCR product with 25 μl of KAPA HIFI HotStart Ready Mix, 300 nM each Nextera XT Index Primer (Illumina® Inc., CA, USA), and 10 μl of UltraPure™ DNase/RNase-Free Distilled Water for a total volume of 50 μl. Thermal cycling conditions were as follows: 3 min at 98 °C, 8 cycles of 30 s at 98 °C, 30 s at 55 °C, 30 s at 72 °C, and a final elongation step of 5 min at 72 °C. Indexed amplicons were purified with magnetic beads as described above. Amplicon products were quantified with a Quant-iT™ PicoGreen™ dsDNA Assay Kit following the manufacturer’s instructions (Thermo Fisher Scientific, MA, USA) and then pooled in equimolar amounts. The amplicon pool was verified with a Bioanalyzer 2100 (Agilent Technologies, CA, USA) to confirm amplicon size and to visualize the potential presence of primer or adapter dimers, in which case an additional purification step on the pool with SPRI Select (Beckman Coulter, #B23319, CA, USA) beads was performed. Finally, DNA in the pool was denatured with 0.2 N NaOH, and an internal PhiX DNA control was added at a 5% ratio (Illumina® Inc., #FC-110-3001, CA, USA). Samples were sequenced on a MiSeq instrument (Illumina® Inc., #SY-410-1003, CA, USA) using the MiSeq v2 500 cycle kit (Illumina® Inc., #MS-102-2003, CA, USA) at the National Research Council Canada in Montréal.

Bioinformatics analysis

Sequencing results were analyzed using a previously described methodology (Tremblay et al. 2015; Yergeau et al. 2015). Briefly, reads associated with both the 16S rRNA gene and ITS were separately filtered, assembled, trimmed, and quality controlled.

OTU tables were generated using a three-round clustering strategy. Quality-controlled sequences were dereplicated at 100% identity. These 100% identity clusters were denoised at 99% identity using dnaclust v3 (Ghodsi et al. 2011). Clusters presenting an abundance of three or more reads were scanned for chimeras with UCHIME de novo and reference-based algorithms using the Broad Institute 16S rRNA gene Gold reference (http://microbiomeutil.sourceforge.net). The remaining clusters were clustered at 97% identity (dnaclust) to generate OTUs.

The taxonomic assignment of bacterial and fungal OTU results was performed with the RDP classifier (Bayesian classifier) with a training model constructed from the Greengenes database (version 13.5) (DeSantis et al. 2006; Wang et al. 2007). Fungal OTUs were classified using a training model constructed with the United Database (Kõljalg et al. 2013). The OTU tables were filtered for archaea or fungi and normalized using the edgeR R package/archaea or fungi and normalized using the edgeR R package (Robinson et al. 2010). Sequencing data are available on the NCBI Sequence Read Archive (SRA) portal under accession number PRJNA517646.

Statistical analysis

Redundancy analysis (RDA) was performed to simplify the relationship between an explanatory matrix (x, edaphic parameter data) and a matrix to be explained (y, species data matrix). For the explicative matrix (x), extreme values were determined using the “grubbs.test” function, and each of these values was removed from the dataset. Then, to remove variables that were highly collinear, we calculated the variance inflection factor (VIF) for each and removed those variables with VIF > 5 (Borcard et al. 2011). Subsequently, the “imputeBDLs” function was used for the parametric replacement of rounded zeros. All the compositional data (metals) in matrix (x) were transformed by a centered log ratio by the function “cenLR” to allow their interpretation. We first calculated the log ratio of X to Y and then centered them around the grand mean of these values (Aitchison 2002). Each of the values was scaled to unit variance by dividing by the standard deviation of each. The species data matrix (y) was transformed by a centered log ratio since OTU data are considered compositional data (Fernandes et al. 2014). Subsequently, the 30 most abundant classes and genera were kept for RDA, which was performed using the “rda” function with the significant explanatory variables selected by the “ordiR2step” function. A second RDA was performed on the significant explanatory variables. The significance of the explanatory variable (red arrows) was verified using the “anova.cca” function. The RDAs are presented in type II scaling to allow for a better interpretation of the correlation between matrices (x) and (y) (Legendre and Legendre 2009).

Principal coordinate analysis (PCoA) was performed on normalized OTU tables. A distance matrix was created via the “vedgist” function, and PCoA was performed using the “pcoa” function. Then, multivariate analysis (MANOVA) was carried out with the function “pairwise.perm.manova” (nperm = 999) to determine if microbial and fungal communities were significantly different between bulk soil, the rhizosphere, and roots associated with alder, birch, and spruce. The microbial community profiles of the 20 most abundant bacterial and fungal classes and genera were determined using Shiny software (In-house R script).

Ternary plots were prepared using the “ggtern” function. Each of the OTUs associated with class is presented in Fig. 5, where x, y, and z represent the mean relative abundance of one OTU in bulk soil (x), the rhizosphere (y) and roots (z). Each corner of the triangle represents an environment (bulk soil, the rhizosphere or roots). The colored point nearest to the corner indicates an increased relative abundance of this OTU in this particular environment. For results expressed as a percentage, the absolute value corresponds to 100%.

Results

The complete dataset used for this study included 1906 OTUs for the 16S rRNA gene region and 889 OTUs for the ITS region. The number of replicates (i.e., individual plants of a given species and type of environment) varied according to availability on site and are listed in Supplementary Table 3. The RDAs were performed using the 30 most abundant taxa composing the bacterial community (fraction representing 37.1%) and the fungal community (fraction representing 81.3%). Canonical RDAs showed that the key environmental parameters driving the presence of the 30 most abundant bacterial and fungal taxa in the three environments (bulk soil, the rhizosphere and root endosphere) are plant species (p_s) and pH, which partly confirms our first hypothesis stating that plant species, pH, and water content would be the key drivers of the microbial community structure (Fig. 2 and Table 1). The ANOVA tests made it possible to determine that the axes “p_s” and “pH” have p values varying between 0.001 and 0.050 (p_s) and between 0.001 and 0.003 (pH) (Table 1). The “p_s” axis partly explains the distribution of microbial communities (bacteria and fungi) in all environments studied except for bacterial communities in the rhizosphere and roots. The “pH” axis partly explains the distribution of bacterial and fungal communities present in all environments studied except bacterial communities in the roots. The bacterial community structure in roots was more closely related to magnesium (Mg) levels in adjacent soil (Fig. 2c and Table 1a).

Fig. 2

a–c Redundancy analysis (RDA) of the effect of soil environmental variables on the composition of bacterial classes in bulk soil (a), the rhizosphere (b), and roots (c). d–f RDA of the effect of soil environmental variables on the composition of fungal classes in bulk soil (d), the rhizosphere (e), and roots (f). Black points represent the 30 most abundant bacterial and fungal taxa. Red arrows indicate soil significant variables. a–c Legend: Ac, Acidobacteria; Aci, Acidimicrobiia; Act, Actinobacteria; Ap, Alphaproteobacteria; B, k_Bacteria; Ba, Bacilli; Bl, Blastocatellia; Bp, Betaproteobacteria; Ca, Candidatus Azambacteria; Ch, Chlamydiae; Cl, Clostridia; Cy, Cytophagia; Dp, Deltaproteobacteria; Fl, Flavobacteria; Ge, Gemmatimonadetes; Gp, Gammaproteobacteria; JG, JG37.AG.4; KD, KD4.96; Kt, Ktedonobacterales; Me, Melainobacteria; O35, OPB35 soil group; Pa, Parcubacteria; Ph, Phycisphaerae; Pl, Planctomycetacia; Pr, Proteobacteria; S2, Acidobacteriaceae_Subgroup 2; S6, Acidobacteriaceae_Subgroup 6; Sa, Saccharibacteria; So, Solibacter; Sp, Spartobacteria; Sp, Sphingobacteriia; TM6, TM6; and Th, Thermophilia. d–f Legend: Ab, Agaricostilbomycetes; Ag, Agaricales; Aga, Agaricomycetes; Am, Arthoniomycetes; Ar, Archaeorhizomycetes; As, Ascomycota; Ba, Basidiomycota; Bl, Blastocladiomycetes; Ch, Chytridiomycetes; Cm, p_Chytridiomycota; Cy, Cystobasidiomycetes; Da, Dacrymycetes; Do, Dothideomycetes; Eu, Eurotiomycetes; Ex, Exobasidiomycetes; F, k_Fungi; Gl, Glomeromycetes; Is, Incertea sedis; Le, Lecanoromycetes; Leo, Leotiomycetes; Mi, Microbotryomycetes; Mo, Monoblepharidomycetes; Or, Orbiliomycetes; Pe, Pezizomycetes; Pu, Pucciniomycetes; Sa, Saccharomycetes; Sc, Schizosaccharomycetes; So, Sordariomycetes; Ta, Taphrinomycetes; Tr, Tremellomycetes; Us, Ustilaginomycetes; and Wa, Wallemiomycetes

Table 1 Statistically significant RDA axes and their p values for bacterial communities (a) and fungal communities (b)

The results shown in Fig. 3 and Table 2 demonstrate that plant species were the main driver of microbial and fungal community structures. Bacterial communities associated with alders differed when root endosphere communities were compared with the communities of the other two environments (the rhizosphere and bulk soil), while in birch, differences were noted between microbial population structure in the bulk soil and the roots. For microbial populations associated with spruce, the population structure differed between bulk soil and the rhizosphere (Table 2c). In addition, fungal populations in alder differed between bulk soil and the root endosphere. Fungal populations associated with birch revealed differences between bulk soil and the two other environments, while in spruce, changes were observed in root endosphere versus bulk soil and the rhizosphere (Table 2d–f).

Fig. 3

a–c Principal coordinates analysis of bacterial communities associated with alder (a), birch (b), and spruce (c) in bulk soil, the rhizosphere, and roots. d–f Principal coordinates analysis of fungal communities associated with alder (d), birch (e), and spruce (f) in bulk soil, the rhizosphere, and roots

Table 2 MANOVA test result (nperm = 999) performed on data presented in Fig. 3. Bacterial community comparisons between bulk soil, the rhizosphere, and roots associated with alder (a), birch (b), and spruce (c). Fungal community comparisons between bulk soil, the rhizosphere, and roots of alder (d), birch (e), and spruce (f)

We also observed that bulk soil, rhizosphere, and roots on this site were dominated by the bacterial classes Gammaproteobacteria (~ 11%), Deltaproteobacteria (~ 10%), Alphaproteobacteria (~ 9%), Planctomycetacia (~ 8%), and Acidobacteria (~ 4%) and the fungal classes “Others” (~ 29%), Agaricomycetes (~ 19%), Leotiomycetes (~ 19%), “Others” (Ascomycota) (~ 12%), and Incertae sedis (~ 4%) (Fig. 4 and Supplementary Table 4).

Fig. 4

a–c Microbial community profiles of the 20 most abundant bacterial classes in bulk soil, the rhizosphere, and roots of alder (a), birch (b), and spruce (c). Letters in parentheses indicate phyla (A, Acidobacteria; Ac, Actinobacteria; B, Bacteroidetes; C, Chlamydiae; Ch, Chloroflexi; F, Firmicutes; P, Proteobacteria; Pl, Planctomycetes; V, Verrucomicrobia). d–f Microbial community profiles of the 20 most abundant fungal classes in bulk soil, the rhizosphere, and roots of alder (d), birch (e), and spruce (f). Letters in parentheses indicate phyla (A, Ascomycota; B, Basidiomycota; C, Chytridiomycota; G, Glomeromycota; Z, Zygomycota)

A ternary plot representing the four most abundant bacterial classes showed that the majority of class-associated OTUs were present in greater proportion in roots than in bulk soil and the rhizosphere (Fig. 5). In contrast, some of the most abundant fungal classes had OTU proportions that were different in different plants for the same fungal class and had more even partitioning of OTU proportions among the three types of environments (bulk soil, the rhizosphere, and roots) (Fig. 6).

Fig. 5

Ternary plots of mean relative abundances (MRAs (%)) of the 4 most abundant bacterial classes. Each point is an OTU whose position is indicative of its MRA in bulk soil (Bs), the rhizosphere (Rh), and roots (Ro) of alder, birch, and spruce

Fig. 6

Ternary plots of mean relative abundances (MRAs (%)) of the 4 most abundant fungal classes (which are not “other”). Each point is an OTU whose position is indicative of its MRA in bulk soil (Bs), the rhizosphere (Rh), and roots (Ro) of alder, birch, and spruce

The OTUs associated with Acinetobacter, Delftia, and Pseudomonas were relatively more abundant in the roots of the plant species studied (Fig. 7a–c). In addition, Actinoplanes, Bryobacter, Candidatus Solibacter, H16, Rhizomicrobium, and Singulisphaera OTUs show a higher relative abundance in birch roots than in bulk soil and the rhizosphere (results not shown). Furthermore, Fig. 7d–f illustrate that Claussenomyces, Eupenicillium, and Trichoderma were less abundant in the roots than in the bulk soil and rhizosphere of all plant species studied. Other species, such as Amphinema sp., Capronia sp., Cladophialophora sp., Cryptotrichosporon sp., Epicoccum sp., Inocybe sp., Mortierella sp., Oidiodendron sp., Schizangiella sp., and Tomentella sp. were present in lower relative abundance in the roots than in the rhizosphere of some plant species (results not shown and Table 3). Finally, Cistella sp. (alder), Hyaloscypha sp. (alder), and Trichocladium sp. (alder and birch) are equally represented in the rhizosphere and roots of different species (shown in parenthesis) (results not shown). Graphical representation of the Shannon index highlighted that the rhizosphere environment presented the highest diversity for bacterial and fungal communities (Fig. 8).

Fig. 7

Relative abundances (%) of various bacterial (a–c) and fungal species (d–f) in bulk soil (Bs), the rhizosphere (Rh), and roots (Ro) of alder, birch, and spruce. Different letters represent a significant difference after unbalanced ANOVA followed by Tukey’s multiple comparison test. Non-parametric Kruskal-Walis test followed by Dunn multicomparison test was done when data transformation was not possible to meet the normality requirement

Table 3 Species found in decreased relative abundance in roots of plant species found on the mine site

Fig. 8

Shannon diversity indices of bacterial (a) and fungal (b) communities found in bulk soil, the rhizosphere, and roots of all species studied. Different letters indicate significant differences after unbalanced ANOVA followed by Tukey’s multiple comparison test

Discussion

Canonical analyses showed that plant species, pH, and Mg were the main drivers of bacterial and fungal communities associated with alder, birch, and spruce in most of the studied environments (bulk soil, the rhizosphere, and roots) (Fig. 2). These results partly confirmed our hypothesis that plant species, pH, and water content might be the key drivers of microbial community structure. Microbial communities are subject to many changes due to biotic factors (such as the presence of plants or other organisms) and many abiotic factors (such as pH, metal concentration, and water content) (Berg and Smalla 2009). Numerous studies have shown that the number of root exudate compounds that can be released by plants depends on factors such as plant type and age, nutritional status, and general homeostatic state (Cavaglieri et al. 2009; Huang et al. 2014). The diversity in root exudates supports a pool of microorganisms with increased biodiversity, contributing to a plant’s adaptability to recruit rhizospheric and endophytic microbial populations according to its needs (Rosenblueth and Martínez-Romero 2006; Cavaglieri et al. 2009). Our findings corroborate that plant species partly explain the 30 most abundant bacterial and fungal taxa (Fig. 2).

Soil pH may directly and indirectly modulate the microbial population in the soil. A high concentration of H⁺ associated with an acidic pH could directly disrupt intracellular pH in a microbial population without any “acidic pH” adaptations (i.e., reverse of membrane potential, cell impermeability to protons, or enzymes or chemicals capable of binding or using protons) (Booth 1985; Baker-Austin and Dopson 2007; Krulwich et al. 2011). In addition, an acidic pH can indirectly influence microbial populations through a modification of the solubility and therefore bioavailability of certain metals (Rieuwerts et al. 1998; Bolan et al. 2014). Solubility modification of certain elements, whether essential (e.g., Ca, Co, Cr, Cu, Fe, K, Mg, Mn) or nonessential (e.g., Al, As, Cd, Pb) to microbial metabolism, can be toxic for organisms that do not possess mechanisms to alleviate the oxidative stress caused by metals (Hall 2002; Hossain et al. 2012). Microbial populations will, however, adapt to survive in environments with harsh conditions such as fluctuations in pH and high metal concentrations (Bruins et al. 2000; Nwuche and Ugoji 2008). In the present study, the results shown in Supplementary Table 1 indicate a wide range of pH values in the tailings deposit (2.84 to 8.90). Rousk et al. (2010) and Liu et al. (2014) also reported that pH could greatly affect bacterial species abundance and diversity in soil. In this context, it is logical to hypothesize that depending on individual plant location, the role of pH was paramount in structuring microbial communities via its influence on the bioavailability of specific metals and metalloids.

The tailings deposit we studied showed high spatial variability in Mg concentrations (44.62 to 5247.96 μg_metal.g_dryweight^-1) (Supplementary Table 1). Shaul (2002) stated that optimal Mg concentrations in the leaves of healthy plants should range from 83 to 415 μg_metal.g_dryweight^-1. However, our previous study of the mine site recorded Mg foliar levels that spanned a much wider range (< 1 to 5399 μg_metal.g_dryweight^-1) for alder, birch, and spruce (unpublished results). This is noteworthy since Mg is an essential element for plant photosynthesis and acts as a cofactor and modulator for more than 300 enzymes (e.g., RNA polymerase and ATPase) (Shaul 2002; Verbruggen and Hermans 2013). In addition, Mg²⁺ is the most prevalent cation found in prokaryotic cells, with a concentration of approximately 15–25 mM (Moomaw and Maguire 2008). In these cells, Mg²⁺ is a cofactor of ATP and numerous enzymes, promoting the stability of the membrane and representing a structural element of the ribosome (Moomaw and Maguire 2008; Ramesh and Winkler 2010). However, how the Mg²⁺ concentration in soil could lead to Mg²⁺ imbalance in microbial cells remains poorly known. Therefore, the spatial variability in Mg concentrations and heterogeneous pH values affect the bioavailability of Mg in the tailings deposit and could lead to deficiency or/and toxicity in plant communities, leading to physiological stress for plants (i.e., a decrease in photosynthetic rates) (Huber and Jones 2013; Verbruggen and Hermans 2013; Guo et al. 2016). As previously stated, the physiological state of plants is one of the factors governing the selection of rhizosphere and root endophytic organisms (Rosenblueth and Martínez-Romero 2006; Hartmann et al. 2009). The great variations in Mg concentration at the mine tailings site could explain why this parameter was found to be partly responsible for the bacterial community structure in roots (Fig. 2c and Supplementary Table 1).

The rhizosphere is defined as the thin layer of soil firmly attached to plant roots. It is considered a different area from the bulk soil because the roots, present only in the rhizosphere, have a biological, chemical, and physical influence on the soil through root growth, nutrient and water uptake, respiration, and rhizodeposition (via root exudates). The rhizosphere is rich in root exudates composed of sugars, organic acids, proteins, and hormones and can attract a plethora of diverse microorganisms to meet the needs of the plant (Broeckling et al. 2008, Huang et al. 2014; York et al. 2016). Indeed, Berendsen et al. (2012) mentioned that plants could shape their rhizomicrobiome by secreting specific root exudates to draw microorganisms to their rhizosphere. The attracted microorganisms could be beneficial for the plant by facilitating nutrient uptake (i.e., phosphorus and nitrogen) and preventing pathogen colonization (Berendsen et al. 2012), among other functions. The abundance and diversity of the molecules found in the rhizosphere could explain why the highest alpha diversity (Shannon index) was observed in the rhizosphere compared to the two other environments, bulk soil and the root endosphere (Fig. 8). In addition, PCoA showed that most of the major differences in bacterial and fungal community structures were observed between bulk soil and the two other studied environments, the rhizosphere and root endosphere. This refutes the second hypothesis, which proposed that the most important differences in microbial communities would be observed between the roots and other two environments (Fig. 3 and Table 2). This contradiction could be explained by the fact that the combination of nutrients found at the plant-soil interface consisted of a much wider variety of molecules excreted by the roots than that available in the bulk soil (Gaiero et al. 2013; Turner et al. 2013).

Moreover, root endosphere acquisition is driven by three main factors: (1) soil parameters, (2) plant factors (root exudates) that allow colonization and compatibility, and (3) microbial factors that influence the viability of microorganisms inside the roots (Gaiero et al. 2013). Once inside the roots, microorganisms could have many functions, such as enhancing the acquisition of mineral nutriments by solubilizing them (e.g., phosphate, potassium and nitrogen), stimulating plant growth through production of growth factors (among these auxins and cytokinin), stimulating root growth through production of 1-aminocyclopropane-1-carboxylic acid (ACC) deaminase enzymes, and inducing systemic resistance against plant pathogens (Badri et al. 2009; Ma et al. 2016). Our results showed that alder and birch root endospheres had different bacterial populations than bulk soil and the rhizosphere (Fig. 3 and Table 2). Some studies have shown that the root endophytic compartment presents microorganisms distinct from those of the rhizosphere and bulk soil. Edwards et al. (2015) showed that the phyla inside Arabidopsis roots (endophytic compartment) were less diverse than those found in the rhizosphere. Furthermore, Gottel et al. (2011) demonstrated that in Populus deltoides, the microorganism profile in the root endophytic compartment was different from that in the rhizosphere. It was also suggested that endophytic microorganisms possess genes that are different from those found in microorganisms present in the rhizosphere, which could explain the former’s endophytic behavior (Santoyo et al. 2016). The root endosphere is a highly selective environment and allows the penetration of only specific beneficial microorganisms (i.e., rapid contact with plant tissues) versus microbes found in the rhizosphere (Lundberg et al. 2012; Edwards et al. 2015; Santoyo et al. 2016). These observations could explain the significant difference in microbial populations found in roots and the other soil compartments studied, the rhizosphere and bulk soil (Fig. 3 and Table 2).

Limited research has been conducted on microbial communities across bulk soil, the rhizosphere, and tree roots. However, Gottel et al. (2011) presented recent advances in the study of microbial communities across different environments (the rhizosphere and root endosphere) of Populus deltoides. They showed that the rhizosphere is dominated by Acidobacteria (31%) and Alphaproteobacteria (30%), while the root endosphere is dominated by Gammaproteobacteria (54%) and Alphaproteobacteria (23%). In addition, the study also pointed out that a single Pseudomonas-like OTU represents 34% of root endophyte sequences. Our study demonstrated that the rhizosphere of alder, birch, and roots differ from those of P. deltoides reported in Gottel et al. (2011). The rhizospheres were composed of 8.19%, 10.80%, and 9.22% Gammaproteobacteria and 9.07%, 9.09%, and 9.16% Deltaproteobacteria, respectively. These bacterial classes were the most abundant classes found in each of the studied plant species. In addition, the root endospheres of alder, birch, and spruce were also mostly populated by Gammaproteobacteria, with relative abundances of 13.29%, 15.96%, and 19.28%, respectively. Deltaproteobacteria are also the second most abundant bacterial class in alder, birch, and spruce root endospheres, with relative abundances of 9.46%, 9.26%, and 8.45% (Fig. 4a-c and Supplementary Table 4b), respectively. In fact, it was shown in the literature that plant species and environmental parameters are strong drivers of microbial population composition in the rhizosphere and roots, which could explain the differences between our study and those of Gottel et al. (2011) and other scientists (Hartmann et al. 2011; Philippot et al. 2013).

To be more specific, ternary plots showed that OTUs associated with the most abundant bacterial classes seem to have a higher relative abundance in roots than in bulk soil and the rhizosphere (Fig. 6). At the species level, Acinetobacter sp., Delftia sp., and Pseudomonas sp. showed a higher relative abundance in plant roots than in the rhizosphere and bulk soil for a given plant species (Fig. 7a, b and c). Izumi (2011) reported that the endophytic bacterium Acinetobacter sp. is more represented in forest tree roots than in agricultural crops. Indeed, the genus Pseudomonas is one of the best root colonizers and represents one of the major groups (with Bacillus sp.) of endophytic communities in trees (Izumi 2011). Pseudomonas sp. and Acinetobacter sp. have PGPR activity in the rhizosphere of diverse plants, including nonsymbiotic biological fixation of nitrogen (Sivasakthi et al. 2014). More specifically, Patten and Glick (2002) showed that inactivation of the ipdc gene encoding a key enzyme involved with indoleacetic acid (IAA) in Pseudomonas putida leads to the development of fewer roots than the wild type, which suggests that IAA secreted by P. putida plays a major role in host root development. Moreover, the fluorescent microbe Pseudomonas could secrete 2,4-DAPG (2,4-diacetylphloroglucinol), which could enhance the defense of plant roots against phytopathogens (Weller et al. 2007). In addition, Acinetobacter sp., an organism living in plant tissues, including roots, promotes plant growth through IAA production (Li et al. 2012). Bacteria belonging to the genus Delftia have been poorly identified in plant roots; however, Roesch et al. (2007) reported the presence of Delftia spp. in roots of field-grown maize. Moreover, Han et al. (2005) described Delftia tsuruhatensis HR4 as a diazotroph organism with potential biocontrol activity against phytopathogens. Within the context of mine site reclamation, it has already been shown that endophytic microorganisms are major players in plant adaptation to environmental stress. The increased relative abundance of Pseudomonas sp., Acinetobacter sp., and Delftia sp. observed in roots could stimulate plant growth in a disturbed environment, such as the studied mine site (Rosenblueth and Martínez-Romero 2006; Weyens et al. 2009; Li et al. 2012).

Another study performed on bacterial and fungal communities in P. deltoides conducted by Shakya et al. (2013) showed that both the rhizosphere and root endosphere are dominated by Ascomycota (52%), Basidiomycota (26.9%), and Chytridiomycota (7.8%). Our results are in agreement with those presented by Shakya et al. (2013), where the rhizosphere of alder, birch, and spruce are dominated by Ascomycota with relative abundances of 33.18%, 34.55%, and 31.96%, respectively, and Basidiomycota with relative abundances in alder, birch, and spruce of 29.41%, 26.48%, and 29.45%. In addition, root endospheres of alder, birch, and spruce were also dominated by Ascomycota, with relative abundances of 37.17%, 38.62%, and 29.68%. Basidiomycota is the second most abundant phylum in the root endospheres of alder, birch, and spruce, with relative abundances of 19.02%, 18.58%, and 18.02% (Fig. 4d–f and Supplementary Table 5a). As observed in bacterial communities, fungal community composition is driven by plant species and environmental parameters (Hartmann et al. 2011; Gaiero et al. 2013). These soil environmental parameters partially drive the global physiological state of the plant, which ultimately results in specific selection of rhizosphere and root endosphere fungi by selective root exudates (Hartmann et al. 2011). The context of the mining site could explain the divergences in fungal communities (Ascomycota) observed in plants growing on the mine site (such as alder, birch, and spruce) versus communities in P. deltoides from the study reported by Shakya et al. (2013).

Ternary plots highlighted that fungal OTUs associated with the four most abundant classes identified (Agaricomycetes, Leotiomycetes, Incertae sedis, and Sordariomycetes) had relative abundances well distributed across the various environments in all of the plant species represented in the study (Fig. 6). More specifically, OTUs identified as Cistella sp. (alder), Hyaloscypha sp. (alder), and Trichocladium sp. (alder and birch) associated with the Pezizomycotina subdivision were equally represented in the rhizosphere and roots of some of the plant species (in parenthesis), as suggested by Gottel et al. (2011) (results not shown). Indeed, many fungal species, such as Claussenomyces sp., Eupenicillium sp. and Trichoderma sp., presented a decreased relative abundance in roots for all studied species (Fig. 6). In addition, Table 3 presents genera that were less abundant than the previously mentioned genera in the roots of alder, birch, and/or spruce. The colonization of roots by fungal endophytes leads to various benefits for both mutualistic partners, such as induction of defense against pathogens by release of a myriad of metabolites, enhanced resistance to abiotic stress (i.e., drought and heavy metals), and increased capture of essential nutrients (i.e., nitrogen and phosphorus) (Porras-Alfaro and Bayman 2011; Fortin et al. 2015). The promotion of plant growth by mycorrhizal fungi is possible via an extraradical network that develops in the rhizosphere. Among others, this network is responsible for uptake of nutrients and water that will ultimately be transferred to the plant via the intraradical network (hyphae in host root tissues) (Bonfante and Genre 2010). This extraradical network also greatly extends into the rhizosphere and bulk soil to allow high nutrient and water absorption, which could explain why many of the 30 most abundant fungal species studied are present in lower relative abundance in the roots than in the rhizosphere of some tree species (Table 3) (Grünig et al. 2011; Fortin et al. 2015).

Conclusions

This study revealed that in an acidogenic mine tailings deposit, plant species and soil pH were the key environmental drivers that structured plant-associated microbial communities in the three compartments studied: bulk soil, the rhizosphere, and the root endosphere. The plant species was the most influential parameter in almost all of the compartments studied. We also observed that for a given plant species, microbial community structure varied considerably between bulk soil, rhizosphere, and root endosphere compartments (PCoA). The highest biodiversity indices were found in the rhizosphere, corroborating the rhizosphere effect, whereby the greatest variety and most abundant supply of nutrients supports a larger and more diverse microbial community than that found in other compartments. Our results also revealed that the most abundant bacterial classes were Gammaproteobacteria and Deltaproteobacteria in both the rhizosphere and root endosphere. We highlighted the fact that plant species drive microbial composition in the rhizosphere and root endosphere, which could explain the differences between the results of the study conducted on P. deltoides and the results presented in this paper (Gottel et al. 2011).

The results indicated that bacteria such as Acinetobacter sp., Delftia sp., and Pseudomonas sp. consistently dominated the root endosphere (increased relative abundance), although they were not very abundant in the rhizosphere or in bulk soil. In addition, we found that most of the fungal taxa were associated with Ascomycota and Basidiomycota, which is in agreement with results of other published studies on P. deltoides. The divergences in the relative abundance of the Ascomycota phylum in this study versus that in the study published by Shakya et al. (2013) could be explained by influences specific to the environment studied on the rhizosphere and root fungal species selection by plants. Thus, many fungal endophytes (such as Claussenomyces sp., Eupenicillium sp., and Trichoderma sp.) are less represented in the roots of species selected for this study than in the roots in their study.

The selectivity in fungal endophyte recruitment by alder, birch, and spruce is supported by the observation of a lower relative abundance for some fungal species (Claussenomyces sp., Eupenicillium sp., and Trichoderma sp.) in root endosphere than in adjacent soil. The potentially important extension of the extraradical network into the rhizosphere compared to intraradical networks in the root endosphere could account for the decreased relative abundance observed for many fungal species. This study sheds light on the biotic and abiotic environmental drivers that shape microbial communities associated with alder, birch, and spruce in the context of an acidogenic mine tailings deposit. To our knowledge, this is the first study on microbial community structures across boreal plant species and compartments (roots, rhizosphere, bulk soil) in a mine setting in Québec. This stepping stone is significant in beginning to elucidate the interactions between plant species and their abiotic environment using approaches such as transcriptomics. Identifying the mechanisms involved in plant-microbe-geosphere interactions will contribute to developing more effective, tailored phytotechnologies for the ecological restoration of disturbed sites.