Shared mycorrhizae but distinct communities of other root-associated microbes on co-occurring native and invasive maples

Study site and root collection

Maple seedlings were collected from the Morgan Arboretum, a forested reserve located in Sainte-Anne-de-Bellevue, Québec, Canada (45.43N, 73.94W). Norway maple trees were planted in the Arboretum in the mid-1950s, and these mature trees were cut and removed from the site in 2008. Although mature trees have been removed, many seedlings are still present in the reserve. In an area containing many Norway and sugar maple seedlings, five parallel 60 m transects (separated by 10 m) were set up. Along each transect, we randomly selected a sampling point. On each sampling date, one Norway maple and one sugar maple seedling of similar height (∼25 cm) were sampled within a radius of 1.5m for each transect. A total of 10 seedlings were collected on each sampling date. Seedlings were selected because entire seedlings could be collected in the field ensuring the exact identification of each root system examined. We repeated this protocol eight times at intervals of two weeks between June and October 2015. A total of 80 seedlings were collected (5 transects ×8 dates ×2 species). Each seedling was removed from the soil with a trowel, leaving soil around the roots to prevent desiccation. Seedlings were then placed in sealable plastic bags and stored in a cooler with ice packs until they were returned to the laboratory for further processing.

Once in the laboratory, roots were carefully removed from the soil, were thoroughly washed with tap water and the small pale feeder roots were isolated. Half of the roots were preserved in Formalin-Acetic Acid-Alcohol (FAA) solution at room temperature for a minimum of 24 h for the morphological analysis and the other half were dried using paper towels and placed in 1.5 ml microtubes and stored at −80 °C for the molecular analysis of the root microbial community. While collecting the roots, approximately 100 ml of soil surrounding the roots of each seedling was collected, air-dried for 48 hrs and stored in paper bags for soil chemical analysis (see Supplemental Information for details of the soil chemical analyses). Seedling collection and root processing until this point occurred on the same day.

DNA Extraction

DNA was extracted from approximately 150 mg of root sample using the PowerSoil DNA Isolation Kit (MOBIO Laboratories, Inc., Carlsbad, CA, USA) with the two following modifications. Firstly, in order to increase cell disruption and increase DNA yield, samples were milled using a MiniBead Beadbeater-16 (BioSpec Products, Bartlesville, OK, USA) for 2 min with three 2.3 mm diameter stainless steel beads (BioSpec Products). After this step, the roots were paste-like in consistency therefore the contents of the PowerBead Tubes tube were added to the tubes containing the crushed roots and the procedure was continued following manufacturer’s instructions. The second modification occurred at the final elution step, where 75 µl of solution C6 solution was added to the spin filter tube instead of the recommended 100 µl, also an effort to increase final DNA concentration. Extracted DNA was stored at −80 °C.

DNA library preparation and sequencing

For bacteria, we amplified the V5-V6 region of the bacterial 16S ribosomal RNA gene using the chloroplast-excluding primers 799F-1115R (799F: AACMGGATTAGATACCCKG; 1115R: AGGGTTGCGCTCGTTG; Chelius & Triplett, 2001; Redford et al., 2010) in order to eliminate contamination by host plant DNA. For the fungal communities, the internal transcribed spacer (ITS) region was amplified using the fungal specific primers ITS-1F (CTTGGTCATTTAGAGGAAGTAA; Gardes & Bruns, 1993) and ITS2 (GCTGCGTTCTTCATCGATGC; White et al., 1990). For the AMF, we used the Glomeromycota specific primer AML2 (GAACCCAAACACTTTGGTTTCC; Lee, Lee & Young, 2008 and the universal eukaryotic primer WANDA (CAGCCGCGGTAATTCCAGCT; Dumbrell et al., 2011). The WANDA and AML2 primer pair amplifies a portion of the small subunit (SSU) RNA gene. We chose this region because it is frequently targeted, and the curated MaarjAM database (Öpik et al., 2010) has been developed based on this region, allowing us to compare our sequences to the largest AMF sequence database to date.

The forward and reverse primers used in the PCR consisted of the target primer, a unique twelve-nucleotide barcode, and an Illumina adaptor (5′ –Illumina adaptor –12 nucleotide barcode –target primer). Reactions were performed using 0.5 unit of Phusion Hot Start II High-Fidelity DNA polymerase and accompanying 5X Phusion HF Buffer (Thermo Scientific) with one µL of undiluted root DNA for bacteria and fungi but a 1/10 DNA dilution was used for the AMF. The reactions were carried out in a final volume of 25 µl in the presence of 200 µM dNTPs, 200 pmols of each primer, and 0.75 µl of DMSO. The reaction was run on an Eppendorf MasterCycler thermocycler with a 30 s initial denaturation step at 98 °C and 35 cycles of 98 °C for 15 s, annealing at 64 °C for 30 s, and extension at 72 °C for 30 s, with a final extension at 72 °C for 10 min. Controls with no DNA and positives with target DNA were run with every series of amplifications to test for the presence of contaminants and ensure PCR reaction mix was correct. The resulting PCR products were loaded on a 1% agarose gel to verify each PCR run. All amplified samples were cleaned and normalized using the Invitrogen Sequalprep PCR Cleanup and Normalization Kit. Multiplexed amplicon libraries for each of the 3 groups were prepared by mixing equimolar concentrations of DNA, and each library was quantified using Qubit dsDNA High Sensitivity kit. Each DNA library was sequenced using an Illumina MiSeq Sequencer with the 2 × 300 bp paired end platform using the Illunima MiSeq Reagent Kit v3.

Bioinformatics

Illumina adapters were trimmed from the raw sequences using BBDuk (BBtools package, https://jgi.doe.gov/data-and-tools/bbtools/), and the trimmed paired-end sequences were assembled with PEAR (Zhang et al., 2014). Sequences were demultiplexed using default settings in QIIME (v. 1.9.1) (Caporaso et al., 2010) with the exception of allowing for two primer mismatches. Chimeric sequences were identified using the Usearch algorithm and then eliminated. We then binned the remaining sequences into operational taxonomic units (OTUs) at a 97% sequence similarity cutoff. Taxonomic identity of each OTU was determined with QIIME using the uclust algorithm and the Greengenes database ver 13_8 (DeSantis et al., 2006) for bacteria; RDP classifier and the UNITE database ver 12_11 (Kõljalg et al., 2005) for fungi, and the uclust algorithm and MaarjAM database for AMF (Öpik et al., 2010).

Root staining and quantification of AMF and Dark Septate Endophyte colonization in maple roots

The preserved roots were removed from the FAA solution and placed separately in OmniSette^® Tissue-Teks^® (Fisher Scientific). Samples were then cleared by autoclaving for 20 min in 10% KOH. The autoclave step was repeated two times, changing the KOH solution each time. If the KOH solution was a dark brown color after the second autoclave step the process was repeated a third time in order to ensure the roots would be sufficiently cleared for morphological analysis. Following the KOH step, roots were gently rinsed with tap water and bleached with 35% hydrogen peroxide for 30 min. Samples were then rinsed again with tap water and acidified in 15% HCl for 15 min. Finally, roots were stained in 0.1% chlorazol black E at 90 °C for 12 min (Brundrett, Piché & Peterson, 1984). After staining, the samples were allowed to destain in a 50% glycerin solution overnight. The roots were then mounted on slides in glycerine jelly and squashed with a cover slip (Widden, 2001). Root fungi were examined using a Leica DM6000 light microscope at a magnification of 400x. The colonization rate for each root sample was quantified using the magnified grid-intersect method (McGonigle et al., 1990). Each intersect was evaluated for the following AMF structures: hyphae, arbuscules and vesicles, for a total of 100 intersects per root sample. For each fungal structure, the colonization rate was determined by counting the total number of intersects in which it was present. Total AMF colonization rates were determined by counting all intersects in which at least one AMF structure was present. When the stained roots were observed under the microscope, the dark septate endophytic (DSE) fungi were clearly evident and easily distinguishable from the AMF, therefore at each intersect the presence of these fungi were also noted and quantified.

Statistical analysis

Prior to the analyses of the microbial sequence data, in order to eliminate potentially spurious OTUs caused by PCR error or sequencing artifacts, all OTUs with fewer than 10 occurrences were removed and the number of sequences per microbial community was rarefied based on the number of sequences present in the sample with the lowest number sequences in order to maintain equal sampling depth across samples. To illustrate differences in the structure of bacterial and fungal communities, non-metric multidimensional scaling (NMDS) analysis based on Bray–Curtis similarities at the OTU level (using Hellinger transformation) was used, and the significance of the observed differences were determined by PERMANOVA (distance-based permutational multivariable analysis of variance, Anderson, 2001) using 999 permutations. To identify which OTUs differed in abundance between maple species we used ANCOM (Analysis of composition of microbiomes; (Mandal et al., 2015) implemented in R, with the multcorr2 parameter. ANCOM was used because of its power and accuracy in the detection of differential abundance of OTUs and for its low false detection rate (Weiss et al., 2017). For this analysis, unrarefied community datasets are used, and all OTUs with fewer than 10 sequence reads were removed. We also classified the functional guild of the 25 most common fungal OTUs using FUNGuild (Nguyen et al., 2016). These 25 OTUs included all OTUs that had >0.5% relative abundance for each maple species. In brief, FUNGuild uses the given OTU taxonomy assigned file to compare it to its database to attempt to assign an ecological guild to each OTU. For each match a confidence rating is also given with highly probable being absolutely certain, probable being fairly certain and the lowest rating being possible, meaning the identification is suspected but not highly supported.

We calculated richness of all three microbe groups from the OTU community matrices and performed t-tests to test for differences between species. The overall effects of species, date and interaction between species and date on AMF and DSE colonization were determined with two-way ANOVAs. Two-way ANOVAs were also used to determine if there were any significant differences in soil chemical parameters with species and sampling date. When data did not satisfy the Shapiro–Wilk test for normality a log transformation was performed. All statistical analyses and visualization were performed in R (R Core Team, 2013) using the following packages: ggplot2 (Wickham, 2009), picante (Kembel et al., 2010), vegan (Oksanen et al., 2017), and ANCOM (Mandal et al., 2015).

The metadata and sequences files have been deposited in Figshare (see DNA Deposition and Data Availability sections). Raw sequence reads have also been deposited in the NCBI Sequence Read Archive under accession number PRJNA553535.

Taxonomic identification of soil microbial communities associated with the roots of Norway and sugar maples

Due to a contamination with one of the forward primers used, all 10 samples from the June sampling date were removed from the analysis and we did not obtain any sequences from one Norway maple sampled on July 11th. From the remaining 69 samples, we obtained a total of 207,280 bacterial sequences after quality trimming and the removal of chimeras. Once all OTUs with fewer than 10 sequences were removed, and the dataset was rarefied to 1,000 sequences per sample, a total of 66 samples and 1077 OTUs remained and were used in the analyses. A total of 18 bacterial phyla were detected from the maple roots, and these 18 phyla were further classified into 47 bacterial orders with 12 having an abundance of over 1% (Fig. 1A). Approximately 16% of the OTUs were not identified to order.

A total 256 389 fungal (ITS) sequences were obtained after quality trimming and the removal of chimeras. When all OTUs with fewer than 10 sequences were removed, and the dataset was rarefied to 1,000 sequences per sample, a total of 75 samples were included in the analyses and the remaining sequences clustered into 474 OTUs. The majority of OTUs were Basidiomycota 50.2%, followed by Ascomycota 31.1%, Glomeromycota 4.3%, Zygomycota 1.7% and 12.7% were unidentified at phylum level. The OTUs were classified into 40 orders, with 11 having a relative abundance of over 1% (Fig. 1B).

Using the AMF specific primers, we obtained sequences from 79 of the 80 samples. A total of 166 403 sequences were obtained after quality trimming and the removal of chimeras. Once all OTUs with fewer than 10 sequences were removed, and the dataset was rarefied 800, a total of 76 samples were included in the analyses and the remaining sequences clustered into 199 OTUs. The relative abundance of the AMF OTUs classified to family level was similar for both host species (Fig. 1C).

Quantification of differences in the composition of soil microbial communities associated with the roots of Norway and sugar maples

We found that the composition of bacterial and fungal communities associated with the roots of Norway and sugar maple were distinct from each other, while the AMF communities were not distinguishable based on plant host (Fig. 2, Table 1). Host species explained 3.6% of the variation in root bacterial and 8.4% of the variation for the fungal community structure. Sampling date explained 3.6% of the total variation in the bacterial communities and had no significant effect on the variation in the fungal communities (Table 1).

We also found significant differences in the relative abundance of particular OTUs between the native and invasive species. Many OTUs from each microbial group were found in both maple species, with plant hosts sharing 95.4%, 65.6% and 83% of all bacterial, fungal and AMF OTUs, respectively. However, when we compared OTU abundance between species using ANCOM (Mandal et al., 2015), we found three bacterial and 10 fungal (ITS) OTUs whose abundances differed significantly between the sugar and Norway maples (P < 0.05, Fig. 3). All three bacterial OTUs and nine of the 10 fungal OTUs had significantly higher abundance in sugar maple compared to Norway maple (Fig. 3). The ten fungal OTUs that varied significantly in abundance between maple species, represent 12.1% of the relative sequence abundance in Norway maple and 39.7% of the sequences in sugar maple. The nine fungal OTUs that were found in significantly greater abundances in sugar maple accounted for 37% of its total OTU relative abundance compared to only 6.6% in Norway maple. The three differentially abundant bacterial OTUs represent 7.2% and 15.0% of the relative sequences in Norway maple and sugar maple, respectively.

Only 8 out of the 25 most commonly occurring OTUs could be classified to a functional group using FUNGuild, with the remaining left unidentified. These 25 OTUs made up 74% of the relative OTU abundance in Norway maple and 82% in sugar maple. Only two OTU functional guild classifications were ranked as highly probable and both were categorized as symbiotic ectomycorrhizal fungi. Four OTUs were ranked as probable and two possible. These six OTUs were either classified as solely saprotrophs or another category which included saprotrophs, symbionts and pathogens. The top two fungal OTUs with higher abundances in sugar maple (Fig. 3, Didymosphaeriaceae and Entoloma sp), were classified as saprotroph-symbiont-pathogen with Didymosphaeriaceae having the ‘probable’ rating and Entoloma sp. ‘possible’. These two OTUs made up 30.8% and 5.2% of the OTU relative abundance in sugar and Norway maple, respectively.

Quantification of differences in the richness of soil microbial communities on the roots of Norway and sugar maples

Bacterial OTU richness was significantly higher in Norway maple (288.23 ± SE 4.87) compared to sugar maple (251.74 ± SE 5.29, t-test; P < 0.001). The same pattern was seen with the fungi, with Norway maple having an OTU richness of 54.72 ± SE 2.76, and sugar maple 45.97 ± SE 2.36 (t-test; P = 0.02). AMF OTU richness did not differ between species. Norway maple had an OTU richness of 28.87 ± SE 1.58 and sugar maple AMF OTU richness was 27.94 ± SE 2.08 (t-test; P = 0.72).

Quantification of differences in colonization rates of AMF and DSE on the roots of Norway and sugar maples

Overall, there was no significant effect of maple species on total AMF colonization (ANOVA; F_1,64 = 0.24, P = 0.62). Total AMF colonization rates were equally high in both Norway maple and sugar maples with an average of 81.1% and 80%, respectively. However, date (F_7,64 = 22.09, P < 0.001) and the interaction between species and date (F_7,64 = 2.94, P = 0.01) was significant.

There was no significant effect of maple species on the abundance of arbuscules (ANOVA; F_1,64 = 0.206, P = 0.651) and vesicles (F_1,64 = 0.029, P = 0.865). The abundance of vesicles was low with an average of 2% and 2.1% for the Norway and sugar maples, respectively (Fig. 4A), while the abundance of arbuscules was similar and quite high in the Norway and sugar maples with an average abundance of 77.9% and 76.6%, respectively (Fig. 4B). However, the effect of date on the abundance of arbuscules and vesicles was significant (F_7,64 = 29.96, P = 0.006 and F_7,64 = 9.16, P < 0.001, respectively). There was also a significant interaction effect between maple species and date on the abundance of arbuscules (F_7,64 = 3.152, P = 0.006) but not for vesicle abundance (F_7,79 = 1.28, P = 0.274). No differences in vesicle and arbuscule colonization levels were detected between maples species at any of the eight sampling dates (Tukey post-hoc test, P > 0.05).

There was a significant effect of maple species and date on the colonization of dark septate endophytes (ANOVA; species: F_1,64 = 25.56, P < 0.001; date: F_7,64 = 5.86, P < 0.001), but the interaction term was not significant (F_7,64 = 2.10, P = 0.06). The average rate of colonization of dark septate endophytes in the Norway maple and sugar maple was 16.5% and 33.4%, respectively (Fig. 4C). DSE were significantly more abundant in the sugar maple compared to Norway maple on August 28 and September 13 (Tukey post-hoc test, P < 0.05, Fig. 4C).

Chemical analysis of the soil surrounding the Norway and Sugar maple roots

There were no significant differences in any of the measured soil chemical parameters among species and sampling dates (ANOVA tests; all P > 0.05). The only significant effect observed was the interaction term of species and date with C/N ratio (P = 0.04). C/N ratio decreased with time for Norway maple but slightly increased for sugar maple. ANOVA results and chemical analyses of the soil data set can be found in the Supplemental Information.