Abstract
Soil fungal communities in tropical regions remain poorly understood. In order to increase the knowledge of diversity of soil-inhabiting fungi, we extracted total DNA from top-organic soil collected in seven localities dominated by four major ecosystems in the tropical island of Puerto Rico. In order to comprehensively characterize the fungal community, we PCR-amplified the ITS2 fungal barcode using newly designed degenerated primers and varying annealing temperatures to minimize primer bias. Sequencing results, obtained using Ion Torrent technology, comprised a total of 566,613 sequences after quality filtering. These sequences were clustered into 4,140 molecular operational taxonomic units (MOTUs) after removing low frequency sequences and rarefaction to account for differences in read depth between samples. Our results demonstrate that soil fungal communities in Puerto Rico are structured by ecosystem. Ascomycota, followed by Basidiomycota, dominates the diversity of fungi in soil. Amongst Ascomycota, the recently described soil-inhabiting class Archaeorhizomycetes was present in all the localities and taxa in this class were among the most commonly observed MOTUs. The Basidiomycota community was dominated by soil decomposers and ectomycorrhizal fungi with a distribution strongly affected by local variation to a greater degree than Ascomycota.
- DNA
- Deoxyribonucleic acid
- EMF
- Ectomycorrhizal fungi
- HB
- High abundance
- ITS
- Internal transcribed spacer 2 of the rRNA gene
- LB
- Low abundance
- LSU
- Large subunit of rRNA gene
- MOTU
- Molecular operational taxonomic unit
- NCBI
- National Center for Biotechnology Information
- NF
- National forest
- NGS
- Next generation sequencing
- NMDS
- Non-metric multidimensional scaling
- PCR
- Polymerase chain reaction
- qPCR
- Quantitative PCR
- SF
- State forest
- TSC
- Two step clustering
Introduction
Recent studies using next generation sequencing (NGS) technologies have increased our understanding of the diversity of soil-inhabiting fungi enormously. Most surveys were carried out in the Northern Hemisphere, more specifically in Europe and North America (e.g., Clemmensen et al. 2013; Lentendu et al. 2014; Menkis et al. 2015), however a few investigations targeting soil-inhabiting fungi in tropical regions (e.g., Kemler et al. 2013; Tedersoo et al. 2014). Initial surveys indicate that tropical biomes contain the greatest species richness of soil fungi (McGuire et al. 2013), though across different biomes, soil fungal diversity appears to be similarly structured by abiotic and biotic factors (McGuire et al. 2013; Tedersoo et al. 2014). We do not yet have more extensive sampling from a single tropical region, and more efforts are needed to extend the characterization of tropical soil fungal communities.
The Caribbean tropical island of Puerto Rico exhibits a rich diversity of flora and fauna in just 8,900 km2 (Gannon et al. 2005; Liogier and Martorell 2000). This landmass together with Cuba and Hispaniola form the Greater Antilles of the West Indies, which are all fragments of old continental crust (Ricklefs and Bermingham 2008). Puerto Rico has a complex geographical history due to several periods of separation and rejoining with the other Greater Antilles (Ricklefs and Bermingham 2008). Six major ecosystems have been identified: littoral zone forest, semi-deciduous subtropical dry forest, tropical- and subtropical-moist forest, and subtropical rain forest (Helmer et al. 2002).
Previous studies in Puerto Rico have aimed to characterize fungi adapted to hypersaline environments using classic molecular and culture-based approaches (Burgos-Caraballo et al. 2014; Cantrell and Duval-Perez 2012; Cantrell et al. 2007). More recently, the diversity of soil fungi has been characterized using NGS techniques at three localities in El Yunque National Forest (NF) (Tedersoo et al. 2014). Consequently, soil-inhabiting fungal communities remain largely uncharacterized throughout this island. The aim of this work is to describe the diversity of soil-inhabiting fungi across the major ecosystems in Puerto Rico.
Materials and Methods
Soil samples and total DNA extraction
Three plots were sampled at each of seven localities representing four different ecosystems in Puerto Rico (Fig. 1, Table 1). The top-organic layer of plant debris and rocks was removed in an approximately 1 m2 grid and a table spoon of soil was scooped from each corner and the center of an internal 0.50 m2 grid. Soils were pooled in a sterile plastic bag and manually homogenized for 1 min. Two sub-samples of approximately 0.5 g were added into separate 2.0 mL microtube containing 750 μL of lysis buffer (Xpedition™ Soil/Fecal DNA miniprep, Zymo Research Corporation, Irvine, California, USA). Followed by cell disruption for 30 s using a TerraLyser™ (Zymo Research Corporation). Samples were stored at room temperature and later at 4 °C until DNA extraction was carried out in the laboratory within one month of sampling, following the manufacturer’s protocol. DNA concentration and integrity was verified in 0.8 % agarose gel electrophoresis on 0.5 % Tris Acetate-EDTA buffer (Sigma-Aldrich, St. Louis, Missouri, USA) stained with 1 × GelRed™ (Biotium Inc., Hayward, California, USA). DNA from a pure culture of Neurospora crassa was included as a positive control.
Polymerase Chain Reaction (PCR) and Ion Torrent library preparation
A fragment of the 5.8S, the Internal transcribed spacer 2 (ITS2) and a fragment of the Large Subunit (LSU) of the rRNA genes was amplified using primers gITS7 forward (Ihrmark et al. 2012) and modified ITS4m reverse (5’-TCCTC[C/G][G/C]CTTATTGATATGC-3’), with both primers containing adequate barcode sequences for single-ended amplification (Table 1). The ITS locus is broadly accepted as a taxonomic barcode for Fungi (Schoch et al. 2012) and among ITS1 and ITS2 there are no significant differences in their power of discriminating species between fungal groups (Blaalid et al. 2013). Modifications on the reverse primer ITS4 (White et al. 1990) were included to reduce its known bias against the soil-inhabiting fungal class Archaeorhizomycetes (Schadt and Rosling 2015).
The PCR mixes were comprised of 10 - 20 ng of soil DNA, 1 × SSoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad Laboratories, Hercules, California, USA), and 0.8 nM of each primer in a final volume of 20 μL. PCR amplifications were carried out in a CFR96 Touch™ Real/Time PCR Detection system (Bio-Rad Laboratories) following the protocol, 10 min pre-denaturation at 95 °C, 1 min DNA denaturation at 95 °C, 45 s at three independent annealing temperatures (50, 54 and 58 °C) to reduce primer bias Schmidt et al. (2013), 50 s of extension at 72 °C and 3 min final extension at 72°C. We used a quantitative PCR (qPRC) that allowed us to adjust the number of cycles for each plate between 23 - 27, in order to ensure that the reactions were maintained within linear amplification. To reduce the chance of altering the relative abundance of fungi by biasing against long reads, we used tag primers directly thereby avoiding to an extra nested-PCR run. All reactions were carried out in duplicates, and all six runs were combined before purification using the ZR-96 DNA Clean & Concentrator™-5 (Zymo Research Corporation). DNA concentration was quantified on duplicates using the Quant-iT™ PicoGreen® dsDNA Assay Kit (Life Technologies Corporation, Carlsbad, California, USA) on a TECAN F500 microplate reader and the DNA integrity was checked by electrophoresis in 2 % agarose gel 0.5 × TAE buffer. A sequencing library was prepared by pooling 35 ng DNA from each sample, loaded onto a 318 chip for PGM Ion Torrent sequencing technology (Life Technologies Corporation) and sequenced in the facilities at Uppsala Genome Center (Uppsala University, Sweden).
Assembly and taxonomy of molecular operational taxonomic units (MOTUs)
We obtained a demultiplexed dataset comprising 1,440,213 sequences, with 1,359,657 sequences corresponding to samples and 80,556 to the positive control (Torrent Suite v 4.0.4) (Table 1, Supplementary Fig.S1A). The software mothur v1.33.3 (Schloss et al. 2009) was used for sequence trimming of the FastQ files (Q ≥ 25 in sequence average, minimum length 150 bp) (Table 1, Supplementary Fig. S1B). To improve the accuracy of clustering and assigned taxonomy of the MOTUs we used only the variable ITS2 locus, by trimming the fragments of 5.8S and LSU loci using the software ITSx v1.0.9 (Bengtsson-Palme et al. 2013). After this step, all sequences shorter than 80 bp and non-fungal sequences were removed; consequently our dataset was reduced to 648,535 sequences of fungal ITS2 locus exclusively (Supplementary Fig. 1C).
The ITS2 fungal sequence dataset was de novo clustered using the software TSC (Jiang et al. 2012) under the following parameters: data type 454 (Ion Torrent and 454 LifeSciences sequencing technology possesses similar sequencing errors (Yang et al. 2013)), single linkage clustering algorithm (recommended for ITS locus in fungi (Lindahl et al. 2013)), sequence identity 0.97 (suitable cutoff for species delimitation in fungi (Blaalid et al. 2013)). We tested three cutoff values for high (HB) and low abundance (LB) sequences (100, 500 and 1000) and obtained 18,961, 18,293 and 18,292 MOTUs respectively for each cutoff. Because cutoffs 500 and 1000 converged to similar total number of MOTUs, we choose 500 for further analysis. Singletons and doubletons were then discarded. The final ITS2 dataset contained 608,300 sequences and 4,378 MOTUs, representing 97.3 % of sequences and 23.9% of MOTUs from the ITS2 dataset, a reduction that allowed for robust analyses (Supplementary Fig. 1D). Predictably, the positive control showed higher retention than the total soil DNA samples, and we found no major differences among samples (Table 1). Ion Torrent sequencing technology can prematurely truncate sequences (Salipante et al. (2014), thus the longest sequences representing each MOTUs were selected using the script pick_rep_set.py in QIIME v1.8.0 (Caporaso et al. 2010). The initial taxonomic assignment of MOTUs was performed using BLASTn 2.2.29+ (Madden 2002) with parameters E-value = 1e-10 and word size = 7, against the reference database 2015-03-02 UNITE + INDS (Koljalg et al. 2013) with only trimmed ITS2 region as described above. Unassigned MOTUs were blasted against the nucleotide database nt at NCBI and sequences of the ITS2 hits were extracted along with their respectively taxonomy using custom scripts deposited at https://github.com/douglasgscofield/add-to-Qiime-DB. For any sequences with taxonomic levels not matching those in the 2015-03-02 UNITE + INDS database, we corrected their taxonomy using Index Fungorum (http://www.indexfungorum.org). We also added Archaeorhizomycetes representative sequences (Menkis et al. 2014).
The final taxonomic identification of each representative sequence was corrected based on 97, 90, 85, 80, 75 and 70 % of sequence identity for assigning MOTUs with names of a species, genus, family, order, class, or phylum, respectively, as proposed by Tedersoo et al. (2014).
The positive control using DNA of N. crassa was used to verify the entire protocol from sampling and tag primer arrangement to clustering and taxonomic assignment. The majority of sequences (95%) in this sample were clustered into a unique MOTU that was identified as N. crassa. The error of 5 % was distributed across 53 MOTUs. Eleven of the MOTUs were unique of the positive control (288 sequences) the rest represents MOTUs also identified in the main dataset and showed variable sequence abundance from 1 – 243 sequences. We concluded that this error originates predominantly from the demultiplexing step and it did not affect the abundance of any particular MOTU significantly (less than 0.1%). The positive control was eliminated from further analysis. The script compute_core_microbiome.py in Qiime was used to account for the difference in sequence depth among samples by the rarefaction method using the lowest number of sequences (11,455) found in one of the samples as recommended by (McMurdie and Holmes 2014). This resulted in a reduction to 4,140 MOTUs. Sequence representatives for each MOTU that satisfied GenBank requirements are available under the accession numbers (KT241043 - KT245134). Sequence data and abundance information per soil sample and locality are available in the Supplementary online File S1.
Statistical analyses
Sequences abundance plots were generated using the library phyloseq (McMurdie and Holmes 2013) in R v3.0.2 (http://www.r-project.org). In order to address differences in MOTUs composition and diversity, rarefaction curves, Chao-1 and Shannon diversity indexes and the non-metric multidimensional scaling (NMDS) based on Bray Curtis distance were computed in the R package ampvis (http://madsalbertsen.github.io/ampvis/). To address the statistically significance of the variables studied here the multivariate analysis of variance Adonis implemented in Qiime was used. Double clustering analysis based on the 100 most abundant MOTUs was carried out using the R script heatmap.2 (http://www.inside-r.org/packages/cran/gplots/docs/heatmap.2). The g-Test of independence was used to calculate the significance of association of MOTUs with locality and ecosystem and corrected by Bonferroni and False Discovery Ration test with 1000 permutations, all as implemented in the Qiime script group_significance.py.
Results and Discussion
Fungal identification based on MOTUs
We identified a total of 4,140 MOTUs at different taxonomic levels after rarefaction: 559 to species (13.5 %), 965 to genus (23.3 %), 919 to family (22.2%), 785 to order (19.0 %), 354 to class (8.6 %), 548 to phylum (13.2 %) and only 10 MOTUs (0.2 %) remained unclassified but assigned to kingdom fungi. The lack of ITS2 reference sequences of fungi from tropical regions as well as fungal endemism in Puerto Rico are likely reasons for the identification of a high number of MOTUs resolved only to high taxonomic levels such as phylum or class. This phenomenon is common in many recent studies, e.g., in Menkis et al. (2015) 13 of the 30 most common MOTUs were identified only to phylum, and in Toju et al. (2014) 10 of the 25 most common MOTUs were identified only to phylum. There is clearly a great need for more fungal inventories from the tropics, including Puerto Rico, that contain both specimen vouchers – i.e. fruiting bodies, axenic cultures – and their molecular characterization.
Among localities the number of common MOTUs and their percentages are shown in the Table 2. On average, the number of MOTUs shared between pairs of localities was 20.8 %. Specifically, we reported 1,499 MOTUs from El Yunque (Cubuy and El Cacique) while independently Tedersoo et al. (2014) found 1,652 fungal MOTUs in the same forest locality but at different sampling sites (Supplementary Fig. S2) with 264 MOTUs (17.8 %) in common between both studies based on a blast sequence similarity search using 97% similarity against our MOTU dataset. These results support our observation that the large majority of MOTUs are unique to each sampling locality. Interestingly Tedersoo et al. (2014) and our study are independent and employing differing sampling and DNA extraction methodologies yet both captured similar MOTU richness at the same locality, despite high local variation and relatively low MOTU overlap. Field surveys should take into account the degree of spatial separation between sites (McGuire et al. 2013) to increase the amount of expected overlapping species between individual samples in order to capture a more complete set of fungal species.
In terms of taxonomic abundance, Dikarya dominates the diversity of soil-inhabiting fungi in Puerto Rico (Fig. 2A, Supplementary online file S1). MOTUs classified in Ascomycota were the most abundant (2,967 MOTUs; 79.6 % sequences) followed by Basidiomycota (1,022 MOTUs; 17.5 % sequences). Much less abundant fungi included Glomeromycota (206 MOTUs; 1.6 % sequences;), Chytridiomycota (35 MOTUs; 0.2 % sequences), Zygomycota (68 MOTUs; 0.2 % sequences), and Cryptomycota (67 MOTUs; 0.3 % sequences). In accordance with other studies, we observe Dikarya to be the most abundant group as they dominate the diversity of soil fungi in most ecosystems worldwide (Tedersoo et al. 2014; Toju et al. 2014). Species in this group include the principal decomposers of organic matter and many taxa establish symbiotic relationships with plant roots (Smith and Read 2008).
Soil-inhabiting Ascomycota in Puerto Rico
Among Ascomycota the classes with major sequence abundance were Archaeorhizomycetes (20.9 %), Sordariomycetes (13.9 %), Eurotiomycetes (11.4%), Dothideomycetes (10.6 %) and Leotiomycetes (10.6 %) (Figs. 2A & 3). Among ascomycetes, 26 MOTUs were found across all localities: Archaeorhizomyces (4 MOTUs), Pestalotiopsis (2 MOTUs), Aspergillus, Bionectria, Chaetomium, Cladosporium, Cylindrocladium, Haematonectria, Lasiodiplodia, Neurospora, Penicillium, Phialocephala and Talaromyces (1 MOTU each) (Supplementary online File S1).
Focusing on the most abundant class Archaeorhizomycetes, a total number of 50,379 sequences were assigned to the class, these grouped into 190 MOTUs (4.6% of the total MOTUs) (Supplementary online File S1). Only eight of these MOTUs were identified to the species level using environmental sequence previously identified as belonging to the class (Menkis et al. 2014). Probably because available sequence were obtained predominantly from studies performed in non-tropical regions. Our modifications to the traditional ITS4 primer (White et al. 1990) to reduce its bias against Archaeorhizomycetes (Schadt and Rosling 2015) were clearly successful; see Material and Methods for further details. Our results are in agreement with previous observations that the diversity of Archaeorhizomycetes is highest in tropical regions (Tedersoo et al. 2014), however our observed abundances are much higher than in that study resulting from our use of less biased primers.
MOTUs classified in Archaeorhizomycetes were found in all the localities (Fig. 2B). Archaeorhizomycetes were most abundant in the wet forests El Yunque (Cubuy) and Maricao, followed by moist forest Guajataca, then dry forests Guanica and Boqueron (Figs. 2B & 3). Only a few reads were classified as Archaeorhizomycetes in samples from the littoral locality Isabela dominated by introduced Casuarina trees. In contrast to other wet localities few reads were classified as Archaeorhizomycetes from the wet forest site El Yunque (El Cacique). We do not have an explanation for the low numbers of Archaeorhizomycetes from the El Cacique site, though this could be due to local soil conditions, or site history as the degree of forest preservation is higher in Cubuy than El Cacique. Current sampling could not resolve whether locality or ecosystem had a greater effect on the distribution of Archaeorhizomycetes in the island; we found that 84 MOTUs differed in abundances among localities (Bonferroni P < 0.01; Supplementary Table S1) while 58 MOTUs differed among ecosystems (Bonferroni P < 0.01; data not shown). This high variation in Archaeorhizomycetes abundance among localities is in accordance with earlier observations of local variation of Archaeorhizomycete (Porter et al. 2008; Rosling et al. 2013). Species abundances have been suggested to be affected by biotic and abiotic factors including type of vegetation, soil horizon, season and pH (Rosling et al. 2013).
Another abundant Ascomycota MOTU was MOTUHB 5, identified as Phialocephala (Vibrisseaceae). This MOTU occurred across all localities (Figs. 2B & 3; Supplementary online File S1), but was particularly abundant in littoral Isabela, which is dominated by introduced Casuarina trees. Species belonging to this genus are soil- and root-inhabiting fungi commonly found in alpine and boreal ecosystems in association with pine roots (Jacobs et al. 2003). High abundance of Phialocephala MOTUHB 5 in this locality could be explained as a result of association with mycorrhizal activity in Casuarina roots (Wang and Qiu 2006). Other studies have also reported Phialocephala as a common root-associated fungi (Bougoure and Cairney 2005) and important decomposers of organic matter in soil in tropical regions (DeAngelis et al. 2013).
Soil-inhabiting Basidiomycota in Puerto Rico
Among Basidiomycota, the class Agaricomycetes was the most diverse with (669 MOTUs; 15.3 % sequences) followed by unclassified Basidiomycota (207 MOTUs, 1.3 %, Fig. 2C). The most diverse families were Agaricaceae, Clavariaceae, Mycenaceae, Thelephoraceae and Psathyrellaceae, most members of which are saprophytic (Supplementary online File S1). Ectomycorrhizal fungi (EMF) classified in the Boletaceae, Inocybaceae, Pluteaceae and Sebacinaceae (108 MOTUs) (Supplementary online File S1) were also detected in Puerto Rico (Fig. 2B). Guajataca was the locality in which basidiomycetes were most abundant (134 MOTUs; 8,103 sequences) followed by Maricao (256 MOTUs; 7,700 sequences). The most abundant MOTU was classified as unidentified Pluteaceae (MOTUHB 31; 3,326 sequences). It was identified at two localities, most abundantly in Guanica (3,044 sequences) with markedly fewer sequences in Guajataca (255 sequences). This pattern of strongly site-specific abundances was observed among the most abundant basidiomycete MOTUs: 15, 1060, 120, 21, 13104, 33 and others (Fig. 3, Supplementary online File S1). The majority of basidiomycete MOTUs had fewer than 20 sequences in total (746 MOTUs, 73.1 %) and no Basidiomycota MOTU was common across all localities nor were any found at all localities, in contrast to Ascomycota (see above).
We identified basidiomycete taxa known to contain EMF fungi, e.g., Inocybaceae, Entolomataceae, Sebacinaceae (Supplementary online File S1). EMF basidiomycetes have been observed in the rhizosphere of native plants on Tenerife in the Canary Islands (Zachow et al. 2009), suggesting that the EMF symbiotic relationship is common in tropical islands as well. The strong local variation we observed in the basidiomycete communities of Puerto Rico has also been observed in other tropical as well as boreal ecosystems (McGuire et al. 2013; Tedersoo et al. 2012). Such strong local variation could be an effect of the presence of clusters of EMF trees that drastically change the local fungal community (McGuire et al. 2013), and also could reflect strong effects of precipitation and temperature on basidiomycete communities (Tedersoo et al. 2012).
In general our results are consistent with suggestions of lower diversity of EMF in tropical ecosystems than in northern temperate ecosystems (Tedersoo et al 2012, 2014). In our study, soil fungal communities were dominated by fungi identified belonging to Ascomycota, in contrast with observations by Tedersoo and co workers (2014) who found greater abundance of Basidiomycota relative to Ascomycota in tropical soils. Clearly more sampling is required throughout the tropics in order to resolve these general patterns of soil fungal communities in tropical soils.
Soil-inhabiting Glomeromycota in Puerto Rico
We found Glomeromycota to be most abundant in Isabela (46 MOTUs; 1,281 sequences), the littoral Casuarina site, and in contrast to nearly all other localities, its abundance was consistent among samples (Fig. 2B). Similar to the ascomycete Phialocephala, this may reflect an association with Casuarina roots, with the consistency of abundance due to the local ubiquity of Casuarina. Whether our results reflect the actual abundance of Glomeromycota at this and other sites is unclear, given the known difficulties in amplifying ITS regions from this group of fungi in general (Hart et al. 2015).
Soil-inhabiting fungal community diversity across localities and ecosystems
The MOTU accumulation curves indicate high variation in number of MOTUs per individual sample (200 – 800 MOTUs in total, Supplementary Fig. S2) and localities (900 – 1200 MOTUs in total, Fig. 5). The observed MOTU accumulation curves continued to accumulate total alpha diversity, while both the Chao-1 richness and Shannon diversity measures reached or approached a plateau in the majority of localities (Fig. 4) as well as samples (Supplementary Fig. S2). This indicates that we have sampled the very large majority of soil fungal diversity at our localities, while some rare taxa likely remains undetected.
Based on the curves for observed MOTU and Chao-1 richness, the moist Guajataca has the highest richness of soil fungal MOTUs and El Yunque (El Cacique) the lowest (Fig. 4). The low fungal richness of El Yunque El Cacique is reflected in our results and discussion of specific taxonomic groups above, and remains a puzzling observation.
The Shannon curve, which takes into account abundance and evenness, indicates that all protected localities (national and state forests) have higher diversity of soil fungi than Isabela, which has low plant diversity and is dominated by introduced Casuarina trees (Fig. 4). The low abundance and diversity of soil fungi at Isabela is also reflected in the analysis of most common MOTUs (Fig. 3 & Supplementary Fig. S3). Above we have outlined some reasons for particular taxonomic representation of soil fungi at Isabela, because of the dominance of exotic plant species Casuarina. Whether lower diversity at this site also reflects more general effects of anthropogenic activities on soil fungal communities in Puerto Rico or other tropical areas will require specific sampling efforts.
We used non-metric multidimensional scaling (NMDS) of the 21 samples to identify how fungal community composition differed with locality and ecosystem (Fig. 5). This analysis showed that the soil fungal communities are distinct depending on both the locality (r2 = 0.844, P < 0.01) as well as the ecosystem (r2 = 0.781, P < 0.01) (Fig. 5), reflecting the MOTU-specific analysis above and confirmed using clustering analysis (Fig. 3). This same pattern was also recovered analyzing the Ascomycota and Basidiomycota communities separately; (Ascomycota, locality: r2 = 0.831, P < 0.01; ecosystem: r2 = 0.7794, P < 0.01), Basidiomycota, locality: r2 = 0.796, P < 0.01; ecosystem: r2 = 0.532, P < 0.02; Supplementary Fig. S2). More detailed locality information than we gathered would be required to separate the effects of specific climatic and edaphic conditions, including any effect of covariance of soil fungal communities with soil calcium content (Tedersoo et al. 2014).
Conclusion
Soil fungal communities in Puerto Rico are organized similarly to other mature tropical and temperate soil fungal communities, with the Ascomycota dominating, followed by the Basidiomycota. In particular, we have shown that Archaeorhizomycetes is one of the most abundant classes in many localities; this was possible only after addressing known primer and amplification biases. Soil fungal community structure also varies significantly among localities and ecosystems, with just a handful (26) MOTUs present at all seven localities, all common soil decomposer ascomycetes. Basidiomycetes were mostly saprophytic and they had a more local distribution compared to Ascomycetes. MOTU accumulation analysis showed that we have identified the majority of fungal taxa present in each sample and locality. This reflects our comprehensive methodology, which includes in situ DNA extraction, the use of deep sequencing output obtained from the Ion Torrent platform, and increasing the breadth of taxonomic identification in our bioinformatic pipeline.
Acknowledgements
This work was supported by the Swedish Research Council VR and the Carl Trygger Foundation for Scientific Research. We thank Stefan Bertilsson, Department of Limnology, Uppsala University, for providing access to the CFR96 Touch™ Real/Time PCR Detection system. We acknowledge the Departamento de Recursos Naturales y Ambientales of Puerto Rico for granting a collecting permit (2014-IC-082) to MJC.