Abstract
The success of ecosystem restoration following mining could be improved through consideration of the soil microbial community, which forms the foundation of ecosystems. Through sequencing we can assess the response of the microbial community to stresses such as stockpiling, and measure community recovery. We sequenced fungal and bacterial communities associated with intact Banksia woodland reference soils and stockpiled topsoil stored for one to ten years following sand mining in Western Australia. We found that both bacterial and fungal richness declined, but that the fungal community returned to a state similar to reference soils, whereas the bacterial community did not. Notably, Bradyrhizobium was absent from 10-year soils, indicating a possible lack of inoculum available to colonize legumes that are often used for revegetation. Soil fungi and bacteria respond differently to stockpiling and key taxa such as Bradyrhizobium could be lost. In addition, changes in the bacterial community may signal a reduction in plant-available nitrogen and a shift towards more anaerobic conditions consistent with previous studies. These changes in microbial communities support previous findings of reduced plant performance on 10-year stockpiled soils and emphasize the importance of considering soil age during restoration.
Restoration Recap
When the number and abundance of bacterial and fungal species in topsoil stored for future restoration was measured over time, we found that there were important differences from fresh soil in both.
Initially, the number of species increased, but then began to decrease after the first year of storage. There were also changes in the abundances of different bacterial and fungal species, which returned to levels similar to fresh soil in fungi, but not in bacteria. Several important groups, such as Bradyrhizobium, were negatively affected.
The fungi and bacteria found in the soil are important for optimal ecosystem function and therefore restoration success. Practitioners may want to consider preferentially using fresher soil and adding Bradyrhizobium when using legumes as part of a restoration event
Understanding the role of soil microbial communities in systems that undergo restoration is providing useful insights into how both natural as well as restored ecosystems function (Harris 2009). Plants form belowground linkages with a number of bacterial and fungal species that assist them in acquiring nutrients (e.g., nitrogen and phosphorus) and water, and provide protection from pathogens. Both plant and soil communities are important for restoration success (Ohsowski et al. 2012).
Soil condition is a critical component in ecosystem restoration because of its role as physical, biological as well as nutrient foundation for successful plant establishment and colonization (Zhang and Chu 2013, Muñoz-Rojas 2018). For example, soil microbial communities drive important ecosystem functions including soil formation, nutrient cycling and decomposition (Zhang and Chu 2011). Although the living microbial component of soil represents only 0.1–0.3% of total soil volume in most soils, microbes facilitate 90% of soil ecosystem functions (Adhikari and Hartemink 2016). Shifts in microbial community composition following restoration have been associated with changes in ecosystem functions, because microbial communities respond relatively quickly to changes in environmental conditions (Sparling 1992, Emmerling et al. 2000, Izquierdo et al. 2005, Muñoz-Rojas 2018). Therefore, soil microbial community characterization is increasingly used to assess the response of soils to environmental changes (i.e., stress and disturbance) and as an indicator of ecosystem recovery (Costantini 2016, Muñoz-Rojas et al. 2016). Furthermore, soil microbial communities can provide a more accurate representation of short-term recovery in comparison to plant species diversity assessments, which can be misleading and simply reflect the artificial planting schedules in case-by-case restoration schemes, rather than actual successful re-establishment (Harris 2003).
During mining operations, it is a widespread practice to remove and stockpile the topsoil layer for re-spreading in areas targeted for restoration in the future (Muñoz-Rojas et al. 2018). Topsoil quality is an important factor in restoration success (Kardol and Wardle 2010) and topsoil is an important source for seeds, nutrients, and microorganisms (Koch and Samsa 2007). Typically, soil that has been removed during development is transferred to a site in need of restoration within the same or similar ecosystem through a practice known as topsoil transfer (Koch 2007, Koch and Samsa 2007). The availability of donor and recipient sites may not occur simultaneously and in practice, soil from donor sites is often stockpiled until it can be applied to a recipient site. The materials selected for stockpile storage (e.g., soil, subsoil, and overburden) may have detrimental effects on soil microbial community and stored soil quality (Waterhouse et al. 2014). For instance, reduction in primary plant nutrients (e.g., carbon, nitrogen, phosphorus, and potassium) will decrease long-term soil quality (Sheoran et al. 2010). Other effects of stockpiling due to the increase in stratification and compaction which lead to anaerobic conditions include the loss of earthworms (Boyer et al. 2011), decrease in the abundance of mycorrhizae, aerobic bacteria and fungi, and an increase in soil bacterial to fungal ratios (Abdul-Kareem and McRae 1984, Harris et al. 1989).
This study builds upon a previously published work by Birnbaum et al. (2017) that assessed plant performance (i.e., plant biomass, specific root length, root diameter) and soil microbial activity (i.e., arbuscular mycorrhizal [AMF] colonization, nodulation rates by N2 fixing bacteria collectively termed rhizobia and nodule biomass) in relation to stockpile age using a common legume (i.e., Acacia saligna) as model plant, a species often utilized in restoration in Western Australia. These authors found that the oldest stockpiled soils (10 years) produced plants with lower biomass but higher mycorrhizal colonization compared with younger stockpiles. Birnbaum et al. (2017) concluded that the negative correlation between host plant biomass and AMF colonization observed in the oldest stockpiled soils may be an effect of microbial communities not accounted for by pot culturing (e.g., presence of fungal pathogens and/or absence of key fungal or bacterial mutualists), so further research is required to describe soil microbial diversity using culture-independent methods, e.g., high throughput sequencing.
Here, we investigated the role of stockpiled soil age on soil bacterial and fungal communities using high throughput sequencing. We hypothesised that bacterial and fungal community composition will be negatively affected by stockpile soil age. We also predicted that the older soils that supported the least biomass production of A. saligna in Birnbaum et al. (2017) would have a higher abundance of pathogenic fungal species and lower abundance of beneficial microbial species. The bacterial and fungal communities from stockpiled soil were compared to those taken from nearby unmined native habitat as a benchmark for restoration success.
Methods
Soil Sampling
Ten stockpiled soil samples ranging in age from one to ten years old at the time of sampling in August 2014 were collected at the Tronox Ltd. (Tronox) Cooljarloo Mineral Sands Mine near Cataby, Western Australia, 170 km north of Perth (30.72° S, 115.42° E) (Birnbaum et al. 2017). Detailed description of the site conditions, vegetation, and stockpiling conditions are in Birnbaum et al. (2017). Briefly, the soils are composed mainly of acidic quartz sands (pH ~ 6, Bassendean sands) overlying clay or lateritic gravel at depths of 1–10 m (Birnbaum et al. 2017). Vegetation at the site is a matrix consisting of five dominant vegetation types: dry heath, dry woodland, wet heath, wet woodland, and wetland (Bradshaw 2015). In this study, sampling was restricted to dry woodland vegetation, which is the most extensive and speciose cover type in the area (Birnbaum et al. 2017). We sampled from five age classes, each represented by three spatially explicit stock piles and one adjacent reference site. Thus, each age class had four experimental units (three stockpiles and one reference site). Reference sites were selected based on mining company records and proximity of a) the closest representative but fully intact vegetation (i.e., unmined and with minimal weed presence) and b) proximity to the appropriate stockpiles (Bradshaw 2015). For each stockpile and reference site, we collected five replicate samples yielding a total of 75 samples from stockpiles and 15 from reference sites. Replicate collection points were spread over stockpiles and reference sites to capture inherent spatial variation in soils, for stockpiles, sampling spanned the length of the top of the stockpile, up to 10 cm deep after scraping away the surface layer and any litter and for woodland, replicates were sampled from a 100-m2 circular area at centre and all cardinal directions. A total of 2 kg of soil was collected for each stockpile. Soil sampling and processing equipment was sterilized with 80% ethanol between piles. Soils were immediately transferred to a cooler in the field before being driven to laboratory facilities at Murdoch University on the same day and stored at 4°C. Soils were sieved through a 2-mm sieve in the laboratory to remove leaves and other coarse material and to homogenize samples. Soils were stored at 4°C and shipped to UBC Okanagan where DNA was extracted from soils using the MoBio PowerMax Soil DNA kit (Qiagen Inc.).
Amplicon Preparation
Fungal ITS amplicons were generated using gITS86F (5' GTGARTCATCGARTCTTTGAA) and ITS4r (5' TCCTCCGCTTATTGATATGC) primers in the first round and sequencing adaptor (underlined, underlined and bold) Ion Xpress barcoded (bold) 341F (e.g., 5' CCATCTCATCCCTGCGTGTCTCCGACTCAGCTAAGGTAACGATTACGGGAGGCAGCAG) with P1 (underlined) adapted 806R (5' CCACTACGCCTCCGCTTTCCTCTCTATGGGCAGTCGGTGATGGACTACVSGGGTATCTAAT) primers in the second round. All reactions contained 10 µl of 2 GoTaq Green Master Mix (Promega Corporation, Madison, WI), 2 µL forward primer (1 µM final concentration), 2 µL reverse primer (1 µM final concentration), 1–5 ng of template DNA and water for a total volume of 20 µL.
Bacterial 16S rRNA gene amplicons were generated using a nested primer approach using 27°F (5' AGAGTTTGATCCTGGCTCAG) and 1492R (5' CTACGGCTACCTTGTTACGA) primers in the first round and sequencing adaptor (underlined, underlined and bold) Ion Xpress barcoded (bold) 341F (e.g.,5' CCATCTCATCCCTGCGTGTCTCCGACTCAGCTAAGGTAACGATTACGGGAGGCAGCAG) with P1 (underlined) adapted 806R (5' CCACTACGCCTCCGCTTTCCTCTCTATGGGCAGTCGGTGATGGACTACVSGGGTATCTAAT) primers in the second round. Fungal ITS amplicons were generated using gITS86F (5' GTGARTCATCGARTCTTTGAA) and ITS4r (5' TCCTCCGCTTATTGATATGC) primers in the first round and sequencing adaptor barcoded gITS86F and ITS4r primers in the second round. All reactions contained 10 µL of 2 GoTaq Green Master Mix (Promega Corporation, Madison, WI), 2 µL forward primer (1 µM final concentration), 2 µL reverse primer (1 µM final concentration), 1–5 ng of template DNA and water for a total volume of 20 µL.
The thermocycler program for first round bacterial and fungal amplicon generation consisted of: 95°C for 4 min followed by 25 cycles of 95°C for 30 s, 53.4°C for 45 s and 72°C for 2 min, with a final extension at 72°C for 5 min. For generating sequencing barcode adapted bacterial and fungal amplicons, the thermocycler program was the same except that 20 cycles were used with an annealing temperature of 65°C.
Both bacterial and fungal amplicons were cleaned after each round using an E.Z.N.A. Cycle Pure Kit (Omega Bio-Tek, Norcross, GA), quantified using a Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Mississauga, ON) and visualized on an agarose gel. Template-free and positive controls were included for all reactions. Readers unfamiliar with microbial community analysis can refer to Hart et al. (2015) for a more general description of the approach.
Sequencing
Barcoded amplicons were pooled in equimolar amounts and purified following agarose gel electrophoresis using an E.Z.N.A Gel Extraction Kit (Omega Bio-Tek) and, prior to sequencing on an Ion Torrent PGM (Life Technologies Inc., Carlsbad, CA), the library dilution factor was determined using an Ion Library Quantitation Kit. An Ion PGM Template Hi-Q OT2 400 kit on an Ion OneTouch 2 system was used to prepare DNA for sequencing and enriched Ion Sphere Particles were quantified using an Ion Sphere Quality Control Kit. Sequencing was performed on a 316v2 chip with an Ion PGM Hi-Q View Sequencing Kit.
Data Analysis
Demultiplexed FASTQ files were generated using Torrent Suite 5.0.5. For both bacterial and fungal amplicons, split_libraries_fastq.py in QIIME (v 1.9.1, Caporaso et al. 2010) was used to merge reads into a single FASTA file (phred quality score Q20, 3 low quality bases read truncation, zero Ns allowed, phred offset of 33). For bacterial amplicons, Qiime was used for open reference operational taxonomic unit (OTU) picking (pick_open_reference_ otus.py) and generating summary plots (summarize_taxa_ through_plots.py). For fungal amplicons, illegal character were removed from the merged FASTA file before using the PIPITS pipeline (Gweon et al. 2015) for dereplication, removal of unique sequences, OTU picking, chimera removal and taxonomy assignment with the pipits_funits and pipits_process commands. Sequences were deposited to the European Nucleotide Archive under the accession number PRJEB29392.
High-level (phylum) distributions of OTUs (bacteria) and phylotopes (fungi) and best possible phylotype identification for both fungal and bacterial OTUs were examined for 95% of all sequences generated. Samples were rarefied to 500 reads/sample for statistical analyses.
Statistical Analyses
The effect of stockpiling on bacterial and fungal communities was assessed using a generalized linear model fitting the evenness and Shannon Diversity data to a Gaussian distribution and the richness data to a Poisson distribution. Differences between the soils sampled (Reference sites, 1, 2, 3, 5 and 10-year old stockpiles) were assessed using multi-response permutation procedure (MRPP) of fungal phylotypes and bacterial OTUs. MRPP, a nonparametric test of the hypothesis of no difference between putative species between soil groups (McCune and Grace 2002). It is a procedure akin to permutational analysis of variance (ANOVA), reporting a test statistic (T), p-value and an agreement statistic (A), which indicates an effect size. As a permutation procedure, this process compares the occurrence of phylotypes/species in each soil group to multiple randomizations of the entire data set, determining whether the pattern observed is likely different to random. An A statistic of greater than 0.3 is considered a significant effect size for ecological data (McCune and Grace 2002). One sample from the bacterial sequences (reference site) failed during sequencing and was not included in our final analysis. All analyses were performed in R (v. 3.4.4, R Core Team 2018).
Results
Diversity of the Bacterial and Fungal Communities
Both the bacterial and fungal communities showed a very similar trend in OTU richness in response to soil stockpiling (Figure 1 and Figure 2). For both communities, richness increased significantly immediately after stockpiling, which declined to reference levels in the five-year-old soils (GLM; Fungi: z = –2.21, p = 0.027; Bacteria: z = –7.20, p = <0.001). However, stockpiling did not affect the evenness (GLM; Fungi: t = 1.74, p = 0.105, Figure S1; Bacteria: t = –0.31, p = 0.762, Figure S2) or the Shannon Diversity (GLM; Fungi: t = 1.16, p = 0.269, Figure S3; Bacteria: t = –1.22, p = 0.246, Figure S4) of either community.
OTU richness of bacterial communities from reference soils and 1, 2, 3, 5 and 10-year stockpile soils. Each point represents the richness from a single sample and the blue line shows the regression.
OTU richness of fungal communities from reference soils and 1, 2, 3, 5 and 10-year stockpile soils.
Bacterial and Fungal Community Structure of Stockpiled Soils
Reference fungal community structure did not differ significantly from the 5- or 10-year stockpiled soils (Figure 3, Supplementary Material Table S1). The most significant differences in fungal community structure were identified for pairwise comparisons between the 1-year soils and all other stockpiled soils. The 1-year soils differed from the reference soils (p = 0.017). A significant difference was also found between the 2-year (p = 0.022) and 10-year soil fungal communities (p = 0.023). Thus, if viewed as a representation of fungal community development over time, the community returns to a reference state within 5 years (p = 0.268).
Pairwise community structure comparisons between bacteria (top triangle) and fungi (bottom triangle) for all possible soil age comparisons with black squares showing a significant p-value (< 0.05) and a d effect size (A) between 0.05 and 0.3, grey squares represent a significant p-value and 0.15 < A < 0.3 (“somewhat different”) and white squares indicate either no difference, small effect size (A < 0.15), or both.
The bacterial communities did not respond in concert with the fungal communities (Figure 3, Supplementary Material Table S1). Reference soils were different from 1-year, 5-year and 10-year stockpiled soils, but not from 2- and 3-year stockpiled soils. The most significant differences in bacterial communities were found between 10-year soils and 1- (p = 0.009) and 2-year (p = 0.010) stockpiled soils. Also, 3-year and 1-year soils were significantly different (0.014). Comparing each stockpiled soil sequentially to mimic time, there is a shift from reference soils to 1-year soils (p = 0.016), no change between 1-year and 2-year (p = 0.102) and again no difference between 2-year and 3-year (p = 0.059) as well as no difference between 3-year and 5-year soils (p = 0.695). Five and 10-year soils harboured different communities (p = 0.011). Finally, 10-year soils differ somewhat from reference soils (p = 0.004). Please refer to Figure 3 for more details on the nature of these changes.
Fungal Diversity in Stockpiled Soils
While 979 fungal OTUs were identified across all soil samples, those OTUs generating at least 1000 sequences across all samples were retained for analysis, representing 72 OTUs and 95% of all sequences. The samples were dominated by Ascomycota (87.9% of retained sequences), followed by Basidiomycota (9.6%) and Zygomycota (2.5%), with the phylum breakdown maintained between soil stockpile ages and the reference soil (Supplementary Material Table S2).
Of the eleven Classes of fungi identified, all were present in the reference soils and most were also represented in the stockpiled soils of different ages (Figure 4). Notable absences included Microbotryomycetes from the 5-year soils and Ustilaginomycetes from the 2- and 10-year soils. Both of these classes were represented by only one genus: Rhodotorula (Microbotryomycetes), representing oleaginous yeast and Moreaua (Ustilaginomycetes), a genus of smut fungi of Australian sedges (Shivas and Vánky, 2003).
Relative proportions of fungal sequences by class identified from reference soils and 1, 2, 3, 5 and 10-year stockpile soils.
Of the 1,036,659 sequences examined, 31.2% could not be identified past the phylum and/or class level. The best possible identification for OTUs within each class was further examined as relative abundances of sequences for each of the soil stockpile ages and the reference soils (Supplementary Material Figure S5). In the Eurotiomycetes, Cladophialophora was absent from the 10-year soils, Exophiala xenobiotica was absent from the 5-year soils and Auxarthron compactum was absent from the 1- and 2-year soils (Supplementary Material Figure S5a). In the Class Dothidiomycetes, Tubeufiaceae was present in the reference soils and younger stockpiles, but absent in the 5- and 10-year soils (Supplementary Material Figure S5b). Aureobasidium pullulans was absent in the 10-year soils but present in all the others. Notably, Westerdykella nigra was absent from all stockpiled soil and only present in the reference soils. Otherwise, the identified fungi were detected in all soils to varying relative abundances (Supplementary Material Figure S5b). Agaricomycetes were represented by only 3 groups, the dominant being Gymnopilus cyanopalmicola (Supplementary Material Figure S5c). In the Leotiomycetes, Leohumicola was not found in the 1- and 3-year soils (Figure S5d). In the Pezizomycotina, either Thielavia terricola or Bryoria fuscescens dominated, co-occurring only in the reference soils and in the 2-year soils, where Thielavia terricola were always greater (Supplementary Material Figure S5e). In the Sordariomycetes, all species were found in all soils, however, the relative proportions varied, with the 2and 3-year soils hosting relatively more Sordariales sequences than the reference, 1-, 5and 10-year soils (Supplementary Material Figure S5f). Notably, those latter four soils appear to have comparable proportions of Sordariomycete species (Supplementary Material Figure S5f).
Oleaginous Yeast
Four classes of fungi were composed of predominately oleaginous yeast (Figure 5): Microbotryomycetes (Rhodotorula), Ustilaginomycetes (Moreaua), Tremellomycetes (Trichosporon, Tremellomycetes sp., Cryptococcus podzolicus, Bullera globospora, Cryptococcus), and Mortierellomycotina (Mortierella alpina and Mortierellaceae). Of these, Trichosporon was absent from 3- and 10-year soils and Rhodotorula was absent from 5-year soils. Oleaginous yeasts comprised 3–4% of sequences in all the soils, except for 10-year soils, where they increased to 10.2%.
Oleaginous yeasts from both phyla Basidiomycota and Zygomycota shown as relative proportions of sequences identified to family/genus of oleaginous yeast found in each stockpile soil age group and reference site soil.
Bacteria
For bacteria, 437 OTUs were identified, many of which were rare. The most abundant 65 OTUs, representing 96.6% of all sequences, were selected for more detailed analysis (Figure 6). Bacterial sequences were dominated by Chloroflexi (54.7%) and Actinobacteria (18.5%), with remaining phyla contributing marginal sequences (Supplementary Material Table S2).
Relative proportions of bacterial phyla identified from reference soils and 1, 2, 3, 5 and 10-year stockpile soils.
Bacterial phyla remained remarkably consistent between the different soil types (Figure 6). Five of the Phyla were represented by a single bacterial phylotype (Cyanobacteria yielded 2 bacterial phylotypes). The candidate phylum WPS-2 was the most abundant phylotype identified of these singles, fluctuating from a high of 5.8% in the reference soils to a low of 1.7% in the 3-year soils.
Whereas both Actinobacteria and Acidobacteria are present in relatively low frequencies, the Phylum Chloroflexi dominates in these soils, accounting for over half of all bacterial sequences recovered (Figure 6). Many identified phylotypes in this group are putative or candidate groups that are known only from direct sequencing efforts and have not been isolated.
The four most dominant bacterial Phyla were Acidobacteria, Actinobacteria, Chloroflexi and Proteobacteria. Within Phylum Acidobacteria, Solibacteriaceae were present across all stockpile soils and reference soils (≥ 10%), whereas in 1-year soils, Solibacteriaceae comprised almost half (i.e., ≥ 45%) (Supplementary Material Figure S6a). Within Phylum Acinobacteria, notably Mycobacterium decreased ~50% in 3- and 10-year soils as compared to 1,2,5-year stockpile soils and reference soils (Supplementary Material Figure S6b). Micrococcaceae was present in all soils, except 10-year soils (Figure S6b). Within Phylum Chloroflexi, Anaerolineae, a strict anaerobe (Yamada et al., 2006) is found exclusively in the reference soils and none of the stockpiled soils (Supplementary Material Figure S6). TK17 and S085 are present in very high proportions, with the highest proportion of TK17 (27% of all sequences) found in the 10-year soils (Supplementary Material Figure S6c). Within Proteobacteria, Acetobactericaea represented the largest proportion across all soils (Supplementary Material Figure S6d). The genus Bradyrhizobium was present in all soils, but absent from the 10-year soils (Supplementary Material Figure S6d). While the Family Bradyrhizobiaceae was present in all the soils, this family encompasses ten genera including Bradyrhizobium. Consequently, Bradyrhizobium may have been recovered from 10-year soils and simply not identified to genus level in those cases.
Discussion
We found that, although both fungal and bacterial richness showed an increase in response to stockpiling between year 0 and year 2, the structure of fungal and bacterial communities responded differently to short- and long-term stockpiling. Fungi and bacteria in soils 1-year post stockpiling were both somewhat different to reference soils, indicating a change in response to the disturbance. Fungi appeared to “recover” to a reference community after five years of stockpiling, whereas bacteria did not.
The fungal community appears to respond to the disturbance and return to a state comparable to the reference soil after 10-years of stockpiling, whereas 1-year soils were consistently different to all other soils. By 5-years, we did not detect a difference in fungal community structure when compared to the reference soils, which remains true with the 10-year soils. The bacterial community of 10-year soils was somewhat different than the reference community, indicating an incomplete recovery. However, the bacterial community did not follow a linear developmental trajectory over time. Bacterial communities may be responding to abiotic factors that we did not include in our study, which could obscure evidence of a linear development, if it exists.
Intriguingly, oleaginous yeasts, taken together, were found to increase in 10-year soils. Oleaginous yeasts are valued in biotechnology applications as a sources of biofuels (Beopoulos et al. 2011). Mortiella alpina is renowned for being particularly productive in that regard (Wang et al. 2018). These fungi tend to be psychrophilic and produce more oil under decreased temperature (Rossi et al. 2009), as well as under reduced nutrient conditions (Blazeck et al. 2014) and can be facultatively anaerobic (Kurakov et al. 2008). An increase to 10% of all sequences (up from 3–4% in all other soils) found in this functional group in 10-year soils could indicate a shift towards fewer available plant nutrients and more anaerobic conditions. An increase in anaerobic conditions during long-term soil stockpiling has been reported in a number of previous studies (Boyer et al. 2011, Abdul-Kareem and McRae 1984, Harris and Birch 1989). Since our samples were taken from the top 10 cm of the profile, it is unlikely that the conditions were anaerobic.
Previous differences observed in the Acacia saligna bioassay (Birnbaum et al. 2017) may be explained by the presence of antagonists, pathogens or conversely by a lack of beneficial microorganisms. Nutrients did not differ significantly between soils (Birnbaum et al. 2017), thus differences observed may also be attributable to microbial sequestration or transformation of nutrients. We found the nodulating nitrogen-fixing bacterium Bradyrhizobium in all soils other than the 10-year stockpiled soils. Bradyrhizobium is a common genus of rhizobia found in A. saligna nodules in Australia (Birnbaum et al. 2016, Marsudi et al. 1999). The Phyla Proteobacteria, where Bradyrhizobium belongs, is most abundant in soils worldwide (Delgado-Baquerizo et al. 2018), yet comprised a proportionally smaller subset of bacteria in our soils. Absence of Bradyrhizobium in 10-year old soils, along with findings from Birnbaum et al. (2017) who reported that A. saligna nodule biomass was smallest when grown in 10-year old soils, provides support that there were little or no Bradyrhizobium in 10-year old soils. This may explain smaller A. saligna biomass in the oldest soils due to lower nitrogen fixing capabilities.
We predicted that a higher number of pathogenic fungal species found in 10-year old soils may explain lower biomass of A. saligna as reported in Birnbaum et al. (2017). We found Phoma sp., a plant pathogen, present in 2- and 3-year soils only. Phoma sp. have been reported to attack legumes, e.g., Medicago trunculata (Tivoli et al. 2006). Consequently, A. saligna may also be susceptible to this plant pathogen as it is also a nodulating Fabaceae forage species that forms arbuscular mycorrhizas. Indeed, Phoma sp. has been isolated previously in Australia (Davidson et al. 2009, Aveskamp et al. 2008). However, Phoma sp. was absent in 10-year old soils and thus it is unlikely to have impacted A. saligna growth in these soils. Notably we found Aureobasidium pullulans, absent only from the 10-year soils, is a yeast that produces a water-soluble glucan gum (Singh et al. 2008). This fungus was originally isolated from legumes and produces indole-3-acetic acid, which is plant growth promoting and strongly inhibits phytopathogenic fungi (Ignatova et al. 2015). The presence of A. pullulans in all soils except the 10–year soil may have contributed to the larger biomass of A. saligna observed in the Birnbaum et al. (2017) experiment, both by reducing pathogen-load and promoting plant growth when grown in younger, 1–5 year old stockpiled soils.
We found that bacterial community structure patterns did not follow those of the fungal community, nor did the bacteria detected in our soils follow patterns described for common soil communities elsewhere (Delgado-Baquerizo et al. 2018), which typically include much higher relative proportions of Proteobacteria, Acidobacteria and Actinobacteria. More than half of our bacterial sequences were Chloroflexi, with most of those bacteria falling into putative, candidate phylotypes that are not well described in terms of their ecological function. Rather, many of these phylotypes are known only from sequencing efforts. Although we found representatives of both Acidobacteria and Actinobacteria, these were present in low relative abundance in our soils. Furthermore, both Plantomycetes and Verrucomicrobia were conspicuously absent from our dataset even though these are common soil phyla (Delgado-Baquerizo et al. 2018).
Chloroflexi is a large phylum representing diverse organisms without shared morphological, physiological, or biochemical characteristics (Gupta et al. 2013). Well-known classes originate from studies of activated wastewater sludge (Gupta et al. 2013) and these are absent from our sequences. Some bacteria in this phylum are thermophilic, can be anoxygenic or anaerobic and possibly photoautotrophic (Gupta et al. 2013). The only photoautotrophic family in our dataset was Kouleothrixaceae, which was not highly abundant and absent from both 1- and 10-year soils. Anaerolineae, present exclusively in the reference soils, are anaerobic and highly abundant in deeper layers of hypersaline microbial mats in Australia (Wong et al. 2015), where these are studied as analogs of stomatolites, thought to be some of the first primordial ecosystems created on earth 3.5 billion years ago (Schneider et al. 2013, Wong et al. 2015). Interestingly, these habitats are common in Western Australia and the high abundance of Chloroflexi in the Banksia woodland could reflect the presence of primordial bacterial lineages present in these soils. Indeed, the majority of the bacterial sequences found in this study fall into putative, candidate phylotypes, the ecological relevance of which can only be surmised by association with known taxa described here. Banskia woodland might therefore provide an excellent opportunity to characterize potentially very important but little known bacterial phylotypes.
It is important to consider the role of revegetation on stockpile microbial community development. In the current study, we did not collect detailed information on plant communities. However, previous research investigated the seedbank of these stockpiles (Bradshaw 2015) which can provide more insight into plant responses across the chronosequence. Bradshaw (2015) showed acacia seeds (A. stenophylla) only in the 2004 reference community. Considering the prodigious seed production capacity of Acacia spp. It is unlikely that they had colonized our 2004 stockpile and may explain the lack of Bradyrhizum in these soils. In general, Bradshaw (2015) showed that plant species richness among sites and stockpiles were highly similar so it is likely that soil microbial compositional changes were driven by soil abiotic conditions, rather than differences in plant communities.
In conclusion, our results have demonstrated that bacterial and fungal communities show different responses to short- and long-term stockpiling. We demonstrated that fungal communities respond initially to the disturbance, however return to a state comparable to the reference soil after 10–years of stockpiling. Bacterial communities in 10–year soils, on the other hand, were somewhat different to the reference community, indicating an incomplete recovery. Unexpectedly, we found a lower proportion of Proteobacteria, especially Bradyrhizobium, a nitrogen-fixing bacterium commonly found in legume nodules which may explain the lower plant and nodule biomass in A. saligna found in an experiment using the same soils (Birnbaum et al. 2017). For a successful restoration outcome using legumes, soil inoculum containing suitable nitrogen fixing bacteria should be added. Finally, our finding that Banksia woodlands soils are unique in terms of their bacterial communities containing ancient lineages is an important discovery as it points to ecosystem qualities that warrant further exploration. Whether the observed changes in stockpile composition affect subsequent restoration practices is yet unknown. Changes during storage should be followed to ascertain whether they impact the restoration of Banskia woodlands.
Acknowledgements
We would like to thank Willa Veber and Laura Bradshaw for help with collecting the soils. We thank Tronox Ltd. for financial support and staff for administrative and logistical support.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.