https://orcid.org/0000-0002-7112-3425
ABSTRACT
Spartina spp. are widely distributed salt marsh plants that have a recent history of hybridization and polyploidization. These events have resulted in a heightened tolerance to hydrocarbon contaminants, but the effects of this phenomenon on the rhizosphere microbial communities are unknown. Here, we grew two parental Spartina species, their hybrid and the resulting allopolyploid in salt marsh sediments that were contaminated or not with phenanthrene. The DNA from the rhizosphere soil was extracted and the bacterial 16S rRNA gene was amplified and sequenced, whereas the abundances of the genes encoding for the PAH (polycyclic aromatic hydrocarbon) ring-hydroxylating dioxygenase (RHD) of Gram-negative and Gram-positive bacteria were quantified by real-time PCR. Both the contamination and the plant genotype significantly affected the bacterial communities. In particular, the allopolyploid S. anglica harbored a more diverse bacterial community in its rhizosphere. The interspecific hybrid and the allopolyploid also harbored significantly more copies of the PAH-RHD gene of Gram-negative bacteria in their rhizosphere than the parental species, irrespective of the contamination treatments. Overall, our results are showing that the recent polyploidization events in the Spartina affected its rhizosphere bacterial communities, both under normal and contaminated conditions, possibly increasing its phytoremediation potential.
INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) are a group of ubiquitous organic pollutants that are highly concerning in view of their potentially severe impact on natural ecosystems and public health. The remediation of sites contaminated by these compounds due to human activity, such as oil spill, is mainly carried out through excavation, soil leaching or various techniques based on microbial degradation (Wilson and Jones 1993; Samanta, Singh and Jain 2002). Phytoremediation is an alternative low-cost and environmentally friendly technology that uses plants and their associated microorganisms to remove pollutants from the environment. There are various types of phytoremediation, depending on the contaminant targeted and its fate (Pilon-Smits 2005). For instance, phytoextraction mainly concerns the extraction of inorganic pollutant from the soil and its translocation and accumulation in plant tissues. Volatile compounds can also be extracted from the soil by plants, and after translocation be released into the atmosphere during phytovolatilization.
However, one of the most interesting forms of phytoremediation is rhizodegradation, where plants stimulate a wide variety of root-associated bacteria and fungi (Ghosal et al. 2016; Correa-García et al. 2018) to degrade organic contaminants, often all the way to CO2 (mineralization). Close interactions between plants and soil microorganisms were reported to play a major role in detoxification and metabolization of xenobiotics (Oliveira et al. 2014; El Amrani et al. 2015), and recent advances in ‘omics’ approaches now allow to address the plant–microbe complex in more details (Bell et al. 2014b). Root exudates were demonstrated to have a central role, modifying the microbial PAH degraders community structure (Cebron et al. 2011; Rohrbacher and St-Arnaud 2016) and significantly increasing microbial biomass (Esperschutz et al. 2009). Rhizodegradation of PAH was suggested to be a model for understanding and manipulating plant–microbe interactions (Correa-García et al. 2018).
One area of interest that is understudied is the use of phytoremediation for the cleanup of coastal salt marshes contaminated by hydrocarbons coming from oil spills at sea. In that context, Spartina is a particularly interesting genus, colonizing salt marshes all around the world and providing key ecosystem services, while being particularly tolerant to the effects of oil spills (Lin and Mendelssohn 2012; RamanaRao et al. 2012; Silliman et al. 2012; Lin et al. 2016; Alvarez et al. 2018). This genus is also characterized by numerous hybridization and genome doubling events (polyploidy) (Aïnouche et al. 2009), which are major evolutionary mechanisms for eukaryotes, and most specially for plants (Wendel 2000; Soltis et al. 2009). Recent in vitro work from our group has shown that genome doubling increased tolerance to phenanthrene in S. anglica as compared with its single-genome parents (Cavé-Radet et al. 2019). Indeed, the genomic shock induced by the merging of divergent genomes results in massive shifts in genetic and epigenetic pathways, which may lead to the emergence of new adaptive phenotypes to environmental constraints (Comai 2005). Although previous studies have shown a link between plant genotype and the rhizosphere microbial community structure and gene expression during phytoremediation (Bell et al. 2014a; Yergeau et al. 2018), nothing is known about the effects of genome doubling events on the root-associated microorganisms and their response to PAH contamination.
Here, we hypothesized that the heightened tolerance of polyploid Spartina species to PAH contaminants observed in vitro also results in an increased abundance of PAH degraders within the root-associated microorganisms. We selected four Spartina species: the parental species Spartina alterniflora and S. maritima, their interspecific sterile hybrid S. x townsendii and the allopolyploid derivative S. anglica. These different Spartina species were grown in salt marsh sediments contaminated or not with phenanthrene, after which the bacterial 16S rRNA gene of the rhizosphere soil was amplified and sequenced, and the PAH ring-hydroxylating dioxygenase (PAH-RHD) genes of Gram-positive (GP) and Gram-negative (GN) bacteria were quantified by real-time quantitative PCR.
MATERIALS AND METHODS
Plant and soil material
Plants were collected in 2016 on their natural habitats along coastlines of France and England. We selected four Spartina species: the parental species Spartina alterniflora (2n = 6x = 62) and S. maritima (2n = 6x = 60), their interspecific sterile hybrid S. x townsendii (2n = 6x = 62) that appeared and emerged at the end of the 19th century and the allopolyploid derivative S. anglica that subsequently appeared following genome doubling (2n = 12x = 120, 122, 124). For the parental species, S. alterniflora was sampled at Le Faou (Roadstead of Brest, France), whereas S. maritima was sampled in Brillac-Sarzeau and Le Hezo (Gulf of Morbihan, France). The homoploid hybrid S. x townsendii, which distribution range does not extend to France was collected in Hythe (Romney Marsh, England), and the allopolyploid S. anglica was sampled at La Guimorais (Saint-Coulomb, France), where bulk sediments (not associated with plants) were also collected. The plant roots and rhizomes were washed abundantly with tap water before their introduction to the new sediment, in order to eliminate foreign plant residues, stones and most of the rhizosphere sediments. After that, all the plants were acclimated in the La Guimorais sediments for three weeks before the start of the experiment. During plant acclimation, the same sediments were used to prepare phenanthrene treated and untreated sediments for the experiment. For this, wet sediments were rinsed several times with tap water and sieved through a 5-mm sieve to reduce salt content and to remove plant residues and stones. We based our phenanthrene sediment contamination level (150 mg kg−1) on the work of Hong et al. (2015), where it was reported that 100 mg kg−1 of phenanthrene was not toxic to S. alterniflora but affected plant–microorganism interactions. Previous studies have shown that the PAH concentration in the sediments of polluted rivers, estuaries or salt marshes around the world ranged from 2 mg kg−1 all the way to 1943 mg kg−1 (Budzinski et al. 1997; Shi et al. 2005; Watts, Ballestero and Gardner 2006; Jiang et al. 2007). We selected a concentration of 150 mg kg−1 of phenanthrene, which represents a non-phytotoxic, moderate exposure in the range of concentrations reported in polluted environments. To contaminate the soil, 300 g of wet sediments was air-dried and then split in two 150 g subsamples. One 150 g subsample was spiked with 100 mM of phenanthrene (1.95 g of phenanthrene diluted in 109.4 mL of absolute ethanol), whereas the other 150 g subsample (control) was spiked with 109.4 mL of absolute ethanol. After total evaporation of ethanol for 1 day, these sediment samples were vigorously mixed by hand with an additional 12.85 kg of wet and rinsed sediments to reach a final concentration of 150 mg phenanthrene kg−1 substrate for the contaminated treatment. Some contaminated and control soils were sampled at this step and kept at −20°C to serve as T0 samples.
Experimental design
Both polluted and control sediments were supplemented with sterile vermiculite before the start of the experiment (volume 1/3) for a better substrate breathability. Individual Spartina plants (four species) were then transplanted in pot containing 1 kg of contaminated or control substrates, and rhizomes were carefully placed into rhizobags of 3 cm of diameter made from plastic mesh (1-cm mesh size) inspired from the experimental design of Hong et al. (2015). The rhizobags were filled with the exact same substrate as the remainder of the pot. Rhizobags were used to create a physical separation between the bulk soil and the rhizosphere because of the difficulty to retrieve an intact root system when uprooting Spartina plants from very wet sediments. The rhizosphere was thus defined as the soil inside the rhizobags, where roots and rhizomes were also found. The experiment was replicated three times in a completely randomized design, resulting in a total of 24 pots that were placed in a phytotronic chamber with a light/dark regime of 16/8 h, in an average ambient temperature of 20°C. Pots were watered with 200 mL of half-strength Hoagland's nutrient solution (Hoagland and Arnon 1950) every 5 days. After 60 days of growth, plants were uprooted and the rhizosphere soil (inside the rhizobags) was collected and stored at −20°C until DNA extraction.
Bacterial 16S rRNA gene amplification
Genomic DNA from soil samples was extracted using a MoBio PowerSoil DNA extraction kit (MoBio, Carlsbad, CA). Amplicons were prepared from total DNA by PCR targeting the bacterial 16S rRNA gene (forward: 515F: GTGCCAGCMGCCGCGGTAA and reverse: 806R: GGACTACHVGGGTWTCTAAT) (Caporaso et al. 2012). PCR amplifications were performed in 25 µL final reaction volumes containing final concentrations of 1× KAPA HiFi HotSart ReadyMix (Roche, Laval, Canada), 0.4 mg mL−1 bovine serum albumin, 0.6 µM forward and reverse primers, and 1 µL of DNA. PCR conditions were as follows: 5 min of initial denaturation (95°C), 25 cycles of 30 s denaturation (95°C), 30 s for primer annealing (55°C) and 45 s elongation (72°C), followed by final extension for 10 min at 72°C. PCR amplicons were then purified using AMPure XP beads (Beackman Coulter, Indianapolis, IN). Amplicon indexing was carried out by a second PCR in 25 µL final reaction volumes containing 12.5 µL 2× KAPA HiFi HotSart ReadyMix, 2.5 µL of specific index primers 1 and 2 from the Nextera Index kit and 5 µL of purified amplicon. The second step, PCR conditions, was as follows: 3 min of initial denaturation (95°C), 8 cycles of 30 s denaturation (95°C), 30 s for primer annealing (55°C) and 45 s elongation (72°C), followed by final extension for 5 min at 72°C. Index PCR cleanup was performed using AMPure XP beads. Finally, PCR amplicons were quantified using PicoGreen (Thermo Fisher Scientific, Montréal, Canada), normalized at 1 ng µL−1, pooled together and sent for sequencing on an Illumina MiSeq (paired-end 2 × 250 bp) at the McGill University and Genome Quebec Innovation Center (Montréal, Canada). Raw reads and associated metadata are available through NCBI BioProject accession PRJNA518897 (http://www.ncbi.nlm.nih.gov/bioproject/518897).
Sequencing data processing
Raw sequencing reads were processed using the mothur Illumina MiSeq standard operating procedure (Kozich et al. 2013). Paired-end reads were first assembled and the joined paired-end sequences with ambiguous bases or smaller than 275 bp were excluded. The dataset was dereplicated for faster computation, after which chimera were removed. Unique sequences were aligned using the SILVA nr database (v128) after which the aligned 16S rRNA gene sequences were clustered at 97% sequence similarity, and singletons were removed from the analysis. Sequences were classified with the 16S rRNA PDS reference from Ribosomal Database Project (RDP version 16; 80% cutoff on bootstrap value for confidence taxonomic assignment) and undesirable sequences (chloroplast, mitochondria, unknown, Archaea and Eukaryota) were removed. A total of 4 905 456 sequences were assembled from the bacterial 16S rRNA gene reads, resulting in 3 132 223 clean sequences, corresponding to 121 804 unique sequences that clustered into 28 092 OTUs (operational taxonomic units). Individual samples were represented by 14 671 to 604 114 reads, with a median of 85 052.5 reads.
Real-time quantitative PCR
The qPCR experiments were conducted on a Stratagene Mx3005P qPCR system (Agilent Technologies, Santa Clara, CA), associated with the corresponding software MxPro Mx3005P (v4.10; Agilent). The qPCR reactions were performed using the primers designed by Cébron et al. (2008). The qPCR reactions were performed in 20 µL total volume containing 1× iTaq universal SYBR® Green reaction mix (Bio-Rad, Hercules, CA) supplemented with 300 µM of each primer and 5 µL of DNA template at a concentration around 5–10 ng µL−1 or water (negative control). The amplifications were carried out following the protocol provided in Cébron et al. (2008) with some modifications. Briefly, the first step consisted of heating to 95°C (5 min) followed by 40 cycles of denaturation at 95°C for 30 s, annealing at either 57°C (PAH-RHD GN) or 54°C (PAH-RHD GP) for 35 s and elongation at 72°C for 75 s, after which SYBR Green fluorescence was measured. At the end of the run, a melting curve analysis was performed where fluorescence was measured at 0.5°C temperature increment every 5 s from 51 to 95°C. Standard curves for the two genes were made from 10-fold dilutions of linearized plasmid containing the gene fragment of interest, cloned from amplified soil DNA (Yergeau et al. 2009).
Statistical analyses
Statistical analyses were performed in R (v 3.5.1). Inversed Simpson index was calculated after rarefaction through random subsampling based on the size of the smallest library (14,671 sequences). Comparisons between α-diversity of bacterial communities, the abundance of the PAH-RHD genes, the relative abundance of phyla and the relative abundance of selected putative hydrocarbon degraders according to the Spartina species and the treatments (polluted or control substrate) were conducted using anova (analysis of variance) with post-hoc Tukey's honestly significant difference (HSD) tests (aov and tukeyHSD functions of the stats package). Prior to anova the normality, and the homoscedasticity of the data to be tested were confirmed using the shapiro.test and bartlett.test of the stats package, respectively. Spearman correlations were calculated using the cor.test of the stats package. Principal coordinate analysis (PCoA) to describe β-diversity was performed on normalized OTU tables using Bray–Curtis dissimilarity index in the vegan (Dixon 2003) and ape (Paradis, Claude and Strimmer 2004) packages. PERMANOVA (permutational multivariate analysis of variance) testing the impact of Spartina species, treatments and their interactions was conducted through 999 permutations. Indicative species among OTUs were investigated using default parameters provided by the indicspecies package (Cáceres and Legendre 2009).
RESULTS
Bacterial alpha diversity and community structure
The presence of phenanthrene led to a general decrease of bacterial alpha diversity in the rhizosphere of all the plant species (two-way anova: F = 12.41, P = 0.002) (Fig. 1A). The two-way anova also highlighted a significant effect of plant species on the bacterial alpha diversity (F = 6.19, P = 0.005), and Tukey's HSD post-hoc tests showed that the rhizosphere of S. anglica was significantly more diverse than the rhizosphere of S. alterniflora and S. maritima (Fig. 1A). The interaction effect contamination × species was not significant for the bacterial alpha diversity in two-way anova tests. PCoA ordinations based on bacterial OTUs showed some level of clustering based on plant species and contamination (Fig. 1B). PERMANOVA tests confirmed that Spartina species (F = 2.09, R2 = 0.22, P < 0.001) and phenanthrene (F = 2.48, R2 = 0.09, P < 0.001) had highly significant influences on the bacterial community structure.
(A) Bacterial diversity (inverse Simpson index) in the rhizosphere of four different Spartina species grown in phenanthrene-contaminated or control sediments. Values annotated by different letters are significantly different according to pairwise t-tests (Welch correction and Bonferroni adjusted, P < 0.05). (B) Principal coordinate analysis (PCoA) for the bacterial communities of the rhizosphere of four different Spartina species grown in phenanthrene-contaminated or control sediments.
Bacterial community composition
The relative abundance of various dominant bacterial phyla in the rhizosphere of Spartina varied between the different plant species and between the contaminated and non-contaminated soils (Fig. 2). These shifts were not significant for some phyla, such as the Acidobacteria, the Actinobacteria and the Bacteroidetes. In contrast, plant species significantly influenced the rhizosphere relative abundance of Chloroflexi (F = 6.21, P = 0.05, significantly higher in the rhizosphere of S. anglica as compared with S. alterniflora and S. x townsendii in Tukey's HSD tests), Planctomycetes (F = 8.86, P = 0.00108, significantly higher in the rhizosphere of S. anglica as compared with S. alterniflora and S. x townsendii, and in the rhizosphere of S. maritima vs S. alterniflora) and total Proteobacteria (F= 4.54, P = 0.02, significantly lower in the rhizosphere of S. anglica as compared with S. alterniflora and S. x townsendii). A significant effect of contamination was also observed for the Chloroflexi (F= 20.93, P < 0.001), total Proteobacteria (F= 14.23, P = 0.002), Verrucomicrobia (F = 4.85, P = 0.43) and Gammaproteobacteria (F = 11.39, P = 0.004). The genera Sphingobacterium, Acinetobacter, Nocardia, Pseudomonas, Mycobacterium, Burkholderia, Bacillus, Sphingomonas, Rhodococcus, Paenibacillus, Massilia,Alcanivoraxand Cycloclasticus were singled out as putative hydrocarbon degraders based on a survey of the available literature. The summed relative abundance of all these genera varied between around 2% to over 3% of all reads and was significantly higher in the rhizosphere of plants growing in contaminated soil (F = 8.43, P = 0.01, Fig. 3), but showed no significant differences between plant species. For the individual genera, the relative abundance of Paenibacillus, Bacillus, Burkholderia, Acinetobacter, Alcanivorax, Rhodococcus and Pseudomonas did not vary significantly between the plant species and treatments. However, the relative abundances of Sphingomonas (F = 9.85, P < 0.001, significantly higher in the rhizosphere of S. x townsendii as compared with all other species), Sphingobacterium (F = 5.41, P = 0.009, significantly higher in the rhizosphere of S. maritima as compared with all other species), Nocardia (F = 7.56, P = 0.002; significantly higher in the rhizosphere of S. anglica as compared with all other species) showed significant differences between the rhizosphere of the different plant species. The relative abundances of Massilia (F = 17.54, P < 0.001), Cycloclasticus (F = 4.96, P = 0.04) and Mycobacterium (F = 7.49, P = 0.01) were significantly affected by the phenanthrene treatment, with significantly higher relative abundance in the rhizosphere of plants growing in the contaminated soil (Fig. 3). In addition, the interaction between contamination and plant species was significant for Sphingobacterium (F = 4.73, P = 0.001). Indicator species analyses confirmed some of the trends observed in anova tests, as it identified bacterial OTUs related to Massilia and Cycloclasticus as the OTUs with the highest indicator power for contaminated rhizospheres among all OTUs (Table 1; Table S1, Supporting Information). The relative abundance of Massilia was significantly and negatively correlated to bacterial alpha diversity (inverse Simpson index) (rs = −0.48, P = 0.017). Apart from a positive correlation for Rhodococcus (rs = 0.41, P = 0.047), the other genera were not significantly correlated to bacterial alpha diversity.
Relative abundance of the most represented bacterial phyla in the rhizosphere of four different Spartina species grown in phenanthrene-contaminated or control sediments.
Relative abundance of putative PAH-degrading bacteria genera in the rhizosphere of four different Spartina species grown in phenanthrene-contaminated or control sediments.
Top 10 strongest indicator bacterial OTUs for the phenanthrene-contaminated Spartina rhizospheres.
| OTU | stat | P-value | Taxonomy |
|---|---|---|---|
| Otu00188 | 0.998 | 0.001 | Massilia |
| Otu00140 | 0.996 | 0.001 | Cycloclasticus |
| Otu00370 | 0.991 | 0.001 | Alcaligenaceae unclassified |
| Otu00552 | 0.989 | 0.049 | Flavobacteriaceae unclassified |
| Otu02955 | 0.960 | 0.002 | Salinirepens |
| Otu01642 | 0.954 | 0.006 | Sphingobium |
| Otu02528 | 0.954 | 0.005 | Alcaligenaceae unclassified |
| Otu01089 | 0.941 | 0.005 | Peredibacter |
| Otu04988 | 0.921 | 0.008 | Unassigned |
| Otu03519 | 0.896 | 0.006 | Gammaproteobacteria unclassified |
| OTU | stat | P-value | Taxonomy |
|---|---|---|---|
| Otu00188 | 0.998 | 0.001 | Massilia |
| Otu00140 | 0.996 | 0.001 | Cycloclasticus |
| Otu00370 | 0.991 | 0.001 | Alcaligenaceae unclassified |
| Otu00552 | 0.989 | 0.049 | Flavobacteriaceae unclassified |
| Otu02955 | 0.960 | 0.002 | Salinirepens |
| Otu01642 | 0.954 | 0.006 | Sphingobium |
| Otu02528 | 0.954 | 0.005 | Alcaligenaceae unclassified |
| Otu01089 | 0.941 | 0.005 | Peredibacter |
| Otu04988 | 0.921 | 0.008 | Unassigned |
| Otu03519 | 0.896 | 0.006 | Gammaproteobacteria unclassified |
Top 10 strongest indicator bacterial OTUs for the phenanthrene-contaminated Spartina rhizospheres.
| OTU | stat | P-value | Taxonomy |
|---|---|---|---|
| Otu00188 | 0.998 | 0.001 | Massilia |
| Otu00140 | 0.996 | 0.001 | Cycloclasticus |
| Otu00370 | 0.991 | 0.001 | Alcaligenaceae unclassified |
| Otu00552 | 0.989 | 0.049 | Flavobacteriaceae unclassified |
| Otu02955 | 0.960 | 0.002 | Salinirepens |
| Otu01642 | 0.954 | 0.006 | Sphingobium |
| Otu02528 | 0.954 | 0.005 | Alcaligenaceae unclassified |
| Otu01089 | 0.941 | 0.005 | Peredibacter |
| Otu04988 | 0.921 | 0.008 | Unassigned |
| Otu03519 | 0.896 | 0.006 | Gammaproteobacteria unclassified |
| OTU | stat | P-value | Taxonomy |
|---|---|---|---|
| Otu00188 | 0.998 | 0.001 | Massilia |
| Otu00140 | 0.996 | 0.001 | Cycloclasticus |
| Otu00370 | 0.991 | 0.001 | Alcaligenaceae unclassified |
| Otu00552 | 0.989 | 0.049 | Flavobacteriaceae unclassified |
| Otu02955 | 0.960 | 0.002 | Salinirepens |
| Otu01642 | 0.954 | 0.006 | Sphingobium |
| Otu02528 | 0.954 | 0.005 | Alcaligenaceae unclassified |
| Otu01089 | 0.941 | 0.005 | Peredibacter |
| Otu04988 | 0.921 | 0.008 | Unassigned |
| Otu03519 | 0.896 | 0.006 | Gammaproteobacteria unclassified |
Abundance of PAH ring-hydroxylating dioxygenases
The abundance of the gene encoding for the alpha subunit of the PAH-RHD of Gram-positive bacteria was not significantly different between the rhizosphere of the four Spartina species and was not affected by sediment contamination, nor by their interaction (Fig. 4A). In contrast, the abundance of the Gram-negative version of the same gene showed significant differences between the four Spartina species (F = 63.2, P = 4.31 × 10−9), with the allopolyploid S. anglica and the hybrid S. x townsendii harboring on average 354% more copies of the gene in their rhizospheres than the parental species (Fig. 4B). Contamination and the contamination × species interaction did not have any significant effect on the abundance of the Gram-negative version of the PAH-RHD.
Abundance of PAH-RHD alpha subunit genes for (A)Gram-positive and (B) Gram-negative bacteria in the rhizosphere of four different Spartina species grown in phenanthrene-contaminated or control sediments. Values annotated by different letters are significantly different according to Tukey's HSD test at P < 0.05.
DISCUSSION
Recently, we have shown enhanced in vitro tolerance to phenanthrene in the allopolyploid in S. anglica as compared with the parental species (Cavé-Radet et al. 2019). The hypothesis behind the present pot study was that this increased tolerance of the allopolyploid S. anglica to hydrocarbon contamination would also result in an increased abundance of PAH degraders within its root-associated microorganisms. The results presented here are coherent with this hypothesis, as we found significant differences between the different pot-grown Spartina species in terms of their bacterial community composition, structure and diversity, and, more importantly, in the abundance of PAH-RHD genes in their rhizosphere. The allopolyploid S. anglica harbored a more diverse bacterial community, composed of relatively more Nocardia, Chloroflexi and Planctomycetes and less Proteobacteria in its rhizosphere as compared with its diploid parents and their hybrid. In addition, the allopolyploid (S. anglica) and the hybrid (S. x townsendii) had significantly more PAH-RHD genes related to Gram-negative bacteria in their rhizosphere than the parental species, suggesting a heightened capacity for PAH degradation by the rhizosphere microbial community. Previous studies from our group using willows have shown that the phylogeny of the host plant significantly influenced the fungal community composition under high levels of contaminant (Bell et al. 2014a). We also showed that the metatranscriptomic response of the rhizosphere microbial communities to contamination varied between different willow genotypes, and that this response was mirrored in the growth of the genotypes in contaminated soil (Yergeau et al. 2018). However, this is the first time, to our knowledge, that functional and taxonomical differences between the rhizosphere bacterial communities of recently naturally speciated plants are reported in the context of soil/sediment contamination.
The rhizosphere bacterial communities of the four pot-grown Spartina species tested showed significant responses to the presence of the contaminant. In the presence of PAH, salt marsh plants were reported to harbor different microbial communities, favoring hydrocarbon-degrading microorganisms (Ribeiro et al. 2011). Similarly, we found here an increase in the relative abundance of putative PAH degraders, and more specifically for the genera Mycobacterium, Cycloclasticus and Massilia. These genera represent large bacterial groups that are able to metabolize and degrade PAH (Kanaly and Harayama 2000; Johnsen, Wick and Harms 2005) and are consistent with previous results about the PAH-degrading bacteria associated to Spartina in salt marshes (Daane et al. 2001; Launen et al. 2008). Interestingly, Cycloclasticus spp. were often reported as one of the predominant PAH degraders in seawater (Yergeau et al. 2015b; Tremblay et al. 2017), whereas Mycobacterium are typical soil PAH degraders (Hennessee and Li 2016) and Massilia were found to degrade PAHs in soils or associated to roots (Liu et al. 2014; Wang et al. 2016). This diversity of putative PAH degraders associated with the rhizosphere of Spartina might be linked to the nature of its habitat, at the interface of terrestrial, plant and marine ecosystems. These bacterial genera probably contain candidates of interest for salt marsh remediation, but their potential would have to be confirmed.
However, these taxonomical shifts in the relative abundance of these putative microbial hydrocarbon degraders were not reflected in the abundance of the PAH-RHD genes. Indeed, contamination with phenanthrene did not shift the abundance of PAH-RHD genes in the rhizosphere of the pot-grown Spartina, both for the Gram-positive and Gram-negative variants of the gene. Previous studies had alluded to a similar phenomenon, where genes and microorganisms related to the degradation of hydrocarbon contaminants were already enriched in the rhizosphere of plants growing in non-contaminated soil (Daane et al. 2001). These hydrocarbon degrading genes were also expressed by microbial isolates in response to root exudates (Mark et al. 2005; Matilla et al. 2007; Ramachandran et al. 2011), and found in the rhizosphere metatranscriptome of willows growing in not-contaminated soils (Yergeau et al. 2014; Yergeau et al. 2018). One of the explanations put forward was the presence of secondary metabolites in the rhizosphere environment that were structurally analog to hydrocarbon contaminants (Singer, Crowley and Thompson 2003), resulting in an enrichment and activation of PAH degradation genes even in the absence of exogenous PAH. Alternatively, the significant enrichment of PAH-RHD GN genes in the rhizospheres of the hybrid and the allopolyploid could be due to a larger total bacterial community as recently reported for Solidago polyploids (Wu et al. 2019). Regardless of the underlying mechanism, the higher abundance of PAH-RHD genes in the rhizosphere of the hybrid and the allopolyploid suggest a heightened capacity for contaminant tolerance and degradation. Of course, this higher abundance of PAH-RHD genes does not mean that they were actively expressed, and a study based on the quantification of transcripts is needed to complete this observation.
The discrepancy between the qPCR of functional genes and the relative abundance of putative hydrocarbon degraders could be explained with two reasons: (i) unknown microbial groups could harbor these genes, since part of soil microbial diversity is still unexplored; and (ii) it is possible that not all members of the genera singled out harbored PAH-RHD degradation genes, as intra-genus and intra-species variability in that trait is expected. Recent studies showed that bacterial isolates that are undistinguishable based on their 16S rRNA gene harbored different gene pools, and had different ecological behaviors and functional capacity (Irshad and Yergeau 2018; Lopes et al. 2018). These results are suggesting that one should exert caution when trying to extrapolate functional information from amplicon sequencing data.
As bacteria are thought to be the major players in organic contaminant degradation during rhizoremediation (Bell et al. 2014a,b; El Amrani et al. 2015; Correa-García et al. 2018), and that recent plant–microbe metatranscriptomic studies confirmed that the hydrocarbon degradation genes expressed in the root-rhizosphere environment were mostly linked to bacteria (Gonzalez et al. 2018; Yergeau et al. 2018), it is interesting to note the higher bacterial diversity in the rhizosphere of the allopolyploid S. anglica as compared with the parental Spartina species. Upon a contamination event, the plant-associated microbial diversity could have a central importance, as higher diversity generally results in functional redundancy, which would enable the microbial communities to cope with a wider variety of environmental conditions while still providing essential services to the plant. Previous studies from our group have shown that the initial soil diversity explained better the difference in willow growth under highly contaminated conditions than diversity at the end of the experiment (Yergeau et al. 2015a). It was further suggested in that study that restoring microbial diversity of degraded environments could be the key for successful phytoremediation. This potential link between the higher rhizosphere diversity and the increased resilience of S. anglica to contamination is intriguing and warrants further research.
Even though we acclimatized the plants in the sediments for three weeks, we cannot exclude the possibility that the differences in the microbial communities in the rhizosphere of S. anglica were because the bulk soil used for the experiment was from a site where S. anglica predominantly grew. This might also have affected indirectly the microbial communities through shifts in the physiology of the parental plants and of the hybrid that were adapted to the conditions found in other sediments. However, some commonalities in the responses of the allopolyploid and of the hybrid suggest that the effect of the soil origin on some of the results presented here was not necessarily that strong.
We described here for the first time the root-associated bacterial community of four pot-grown Spartina species in a context of allopolyploidization and PAH contamination. Significant differences were observed between the plant species and between contaminated and control rhizospheres. The most salient differences were the higher abundance of Gram-negative PAH-RHD genes and the higher bacterial diversity in the rhizosphere of the allopolyploid as compared with the parental species, which could be a major factor contributing to the increased tolerance of S. anglica to contaminant stress. The underlying cause for these shifts could be related to changes in the root exudation patterns following hybridization, but more work would be necessary to test this hypothesis.
FUNDING
AC-R was supported by an Erasmus+ mobility scholarship from the European Commission. This work was supported by the Ministère de l'Enseignement Supérieur et de la Recherche, the CNRS (Centre National de la Recherche Scientifique), the Observatoire des Sciences et de l'Univers de Rennes (OSUR) and a NSERC (Natural Science and Engineering Research Council of Canada) Discovery Grant (2014-05274 to EY). This research was enabled in part by support provided by Calcul Québec (www.calculquebec.ca) and Compute Canada (www.computecanada.ca).
Conflicts of interest
None declared.
