Field sites are located in Saucier, Mississippi and Madison, Wisconsin, USA. The Saucier site is located in the Harrison Experimental Forest in southern Mississippi. This geographic region is classified as high decay hazard for wood in service (zone 5-severe) according to both the American Wood Protection Association (AWPA) (Association, 2007) and the Scheffer index (Scheffer, 1971). This soil region is classified as coastal plain and the dominant soil type is sandy clay with surrounding overstory being predominantly loblolly pine (Pinus taeda L.). Untreated southern pine sapwood will typically decay to the point of structural failure after 1 year in ground exposure in this environment. The Saucier, Mississippi test site has been utilized by the United States Department of Agriculture Forest Service (USDA-FS) Forest Products Laboratory (FPL) as a preservative test site since 1940 and contains many of the early testing sites used to develop long-lasting and effective wood protectants. Samples were obtained on April 10, 2014 from 3 test plots at the field site. The mixed plot was a 50 year-old plot containing chromated copper arsenate (CCA), creosote and pentachlorophenol treated posts. The copper plot was a 30 year-old plot containing wooden stakes treated with basic solutions of copper salts. The control samples were obtained from a nearby, unused plot of forest within the Harrison Experimental Forest. Four field replicate soil samples per site-treatment combination were transported to the laboratory and frozen at −30°C for 4 days before processing.

The Madison, WI site has been used by FPL since 1950 for routine testing of experimental and existing preservative systems in a moderate decay hazard (AWPA, zone 2). The soil type at Madison is darker and contains more silt and clay as compared to the Saucier site and is subject to more freeze thaw cycles, in addition to a solid winter freeze. The Madison, WI site is mostly prairie/savannah type ecosystem with continuously decreasing overstory containing mostly mixed hardwoods [red maple (Acer rubrum), mixed oaks (Quercus spp.), and some black locust (Robinia pseudoacacia)]. In comparison to the Saucier site, untreated southern pine sapwood typically fails due to decay within 2 years at this more northern site. Samples were obtained on May 9, 2014 from 3 test plots at the Madison field site. The mixed plot was an approximately 60 year-old plot containing CCA, creosote and pentachlorophenol treated posts approximately 10 cm in diameter. The copper plot was a 40 year-old plot containing 60 cm (2 ft) × 120 cm (4 ft) wooden stakes treated with various copper solutions. The control samples were obtained from a nearby, unused plot of forest within the FPL test site approximately 1 km from the treated plots.

Samples were taken from the soil at both sites using 5 g Terra Core (En Novative Technologies, Dexter, MI, USA) disposable soil samplers. Samples were taken from four locations approximately 10 feet apart within each test plot and 3 plugs were combined within each treatment replicate. Four field replicate soil samples were transported to the laboratory and frozen at −30°C for 3 days before processing.

A 0.25 g aliquot of each hand mixed sample was extracted using the MoBio Power Soil DNA Isolation Kit (Carlsbad, CA, USA) following manufacturer's instructions. The samples were eluted in 100 μl C6 elution buffer and cleaned using the MoBio Powerclean Pro DNA Clean-up Kit following manufacturer's instructions. Samples were eluted in 100 μl DC5 elution buffer then quantified by spectrophotometer and diluted to 10 ng/μl in Tris EDTA pH 8. DNA samples were amplified in triplicate using 25 ng template or water controls and ITS1F and ITS2 primers (De Gannes et al., 2013) with Illumina adapters for the MiSeq platform with 24 unique identifiers on the reverse primers. Phusion Hot Start Flex DNA Polymerase (New England Biolabs, Ipswich, MA, USA) in HF buffer was used according to manufacturer's instructions for PCR with the following program: 4 min at 94°C, followed by 30 cycles of 30 s at 94°C, 60 s at 50°C and 90 s at 72°C and a final extension of 10 min at 72°C. Check gels were run on one of each PCR replicate and 400–500 bp products were confirmed. The three PCR replicates were combined and cleaned using Agencort AMPure XP beads following manufacturer instructions. Each sample was quantified using the Quant-it DNA Assay Kit (high sensitivity, ThermoFisher, Waltham, MA, USA) following the microplate procedure with a Synergy H1 multimodal plate reader (Biotek, Winooski, VT, USA). All samples were then normalized to 10 nM and combined in equal amounts. This pooled sample was submitted to the University of Wisconsin-Madison Biotechnology Center—DNA Sequencing Facility for 250 paired Illumina MiSeq sequencing using paired reads of 250 base pairs.

Sequencing data were processed using the AMPtk v0.4.0 pipeline (Palmer, 2015 -https://github.com/nextgenusfs/amptk). Briefly, overlapping 2 × 250 base pair Illumina MiSeq reads were merged using USEARCH v8.1.1831 (Edgar and Flyvbjerg, 2015), forward and reverse primers were removed from the merged reads, and the reads were trimmed or padded with N's to a set length of 250 base pairs. Because ITS sequences are of variable length they require extra processing steps in comparison to 16S reads that are nearly identical in length. The average length of known ITS1 sequences in public databases is ~250 bp; ranging in size from ~150 bp up to >600 bp, thus we aren't able to recover ^*all^* full length sequences with PE 2 × 250 bp sequencing. And it is recommended to use reads that are truncated to the same length when clustering with UPARSE pipeline, padding/trimming to 250 bp was used to maximize the number of reads passing the quality filters of AMPtk as well as provide enough information to assign taxonomy to OTUs. Processed reads were quality trimmed based on accumulation of expected errors less than 1.0 (Edgar and Flyvbjerg, 2015) and clustered using the UPARSE algorithm using default parameters (singletons removed, 97% OTU radius). An OTU table was generated by mapping the original reads to the OTUs using VSEARCH 1.9.1 (Rognes et al., 2016) and the OTU table was subsequently filtered to eliminate “index-bleed” between samples by setting read counts to zero if the number of reads mapped was less than 0.5% of the sum of all read counts for each OTU. Index-bleed (also called index-crossover, index-hopping, barcode-mismatch, etc.) is a phenomenon in NGS sequencing where a small number of reads are mis-assigned to the wrong sample group during multiplex sequencing (combining multiple samples on a single NGS sequencing run). Thus to filter out low read counts that could have arisen from sample mis-assignment (index-bleed), AMPtk provides a filter to clean an OTU table. Taxonomy was assigned using a combination of UTAX and global alignment (USEARCH Edgar, 2010) to the UNITE v7.0 database (Abarenkov et al., 2010) and non-fungal OTUs were removed prior to downstream data processing. BIOM data from AMPtk pipeline was further analyzed using METACOMET (Wang et al., 2016) and PHINCH (Bik and Pitch Interactive, 2014) to visually compare treatment and sites at each taxonomic level.

All distance-matrix based community analyses were performed on a presence absence (binary) OTU matrix, using the Raup-Crick distance metric as calculated by the raupcrick function in the vegan package (Oksanen et al., 2015) of R (R Core Team, 2013). This metric is robust to common issues in large data sets such as an abundance of zeros, and variation in alpha diversity among samples (Chase et al., 2011). To visualize fungal communities, we performed ordinations using non-parametric multi-dimensional scaling (NMDS), implemented by the metaMDS function in the vegan package (Oksanen et al., 2015) of R (Team, 2016). Permutational multivariate analysis of variance (PERMANOVA) tests were used to test for site and treatment effects. PERMANOVA was calculated by the Adonis function in the vegan package (Oksanen et al., 2015). We also tested for multivariate dispersion among groups using the betadisper function in the vegan package of R. Finally, we performed indicator species analyses to identify specific fungal OTUs associated with each treatment group at each site using the multipatt function in the indicspecies package (Cáceres and Legendre, 2009) in R.

Twenty four soil samples were dried at 105°C for 2 days, then immediately capped and transferred to a desiccator. Samples were weighed in a dry room to 4 decimal places and directly into a tared Teflon digestion vessel-Anton Paar (Ashland, VA) HVT50 vessels. Two milliliters of 70% HNO₃ (Sigma-Aldrich (Milwaukee, WI) ACS-grade) was added, using an acid-resistant bottle-top pipette, to each digestion vessel. Samples were pre-digested for 15 min. at room temperature. In a fume hood, 5 ml 18 MΩ H₂O was added, and further digested for 10 min, then vessels were sealed. Digested solutions were cooled, then filtered through glass microfiber filter (Whatman 934-AH) into a 50-ml volumetric flask and brought to volume using 18 MΩ H₂O.

A Horiba (Edison, NJ) ULTIMA II high resolution spectrometer was used for the ICP-OES analysis. The instrument was equipped with a solid state, water cooled 40 MHz radio frequency source, a Czerny Turner monochrometer with 1 meter focal length, nitrogen purged optical bench at 6 L/min, holographic grating (2400 grooves/mm) with resolution <5 pm in the 120–320 nm range (1st order) and <10 pm in the 320–800 nm range (2nd order), and a vertical torch with radial viewing. Power level was 1,000 W. Gas flows were: Plasma gas 12 L/min, sheath gas 0.2 L/min, nebulizer gas 0.28 L/min. Sample introduction used a Conical U-series concentric glass nebulizer (1 ml/min) and a Tracey spray chamber with helix connection and TruFlow sample monitor (Glass Expansion, Pocasset, MA).

Standards were prepared from 1,000 ppm stock solutions without serial dilution. Standard curves were obtained using external standards for Ca 396.847, Ca 422.673, Cr 205.552, Cr 267.716, Cu 224.700, Cu 324.750, Cu 327.396, K 766.455, Mg 279.079, Mg 280.270, Mg 285.213, and P178.229 nm. Each standard curve was checked using a separate calibration solution (Inorganic Ventures (Christiansburg, VA) ICP-MS 71A multi-element standard), then the best line chosen for quantitation of samples. The calibrations, each with a correlation coefficient r² > 0.9999, were checked at the start, middle and end of the ICP sequence, with pass falling within 10% of target. Ten replicate readings of a blank were also run at the start, middle and end of the ICP sequence, then used to determine Limit of Quantitation (LOQ). Acquisition was done in Max mode peak shape, with 0.5 s integration time and 3 replicate readings averaged for the reported measurement. A 20/15 um slit combination was used for all lines. The instrument electronics were turned on ~30 min. prior to start in order to minimize instrumental drift. Prior to each measurement, the sample intake was rinsed 20 s with 5% HNO3 (v/v) at high pump speed, followed by a 20 s rinse with analyte solution at high pump speed, then a 20 s plasma equilibration period at normal pump speed. Prior to calibration, a profile acquisition was taken to visually inspect peak shape and manually set background correction points at peak base.

In order to compare fungal species composition between sites and treatments, PERMANOVA was used to analyze the taxonomy data. Due to large differences in community composition between the MS and WI sites, data sets were analyzed separately.

Sites were significantly different from each other with respect to fungal species composition (P < 0.0001) (Figure 1). Chemical analysis showed richer nutrient profiles in the MS site and lower pH of soil (Table 1). Highly correlated OTUs that contributed to the differences between sites along each NMDS axis were compiled in PERMANOVA and the five OTUs with the highest and lowest degrees of correlation for each site and axis of the NMDS ordination are listed in Table 2. These represent outliers from the shared core fungal groups within the total pool of OTUs and help to define differences between the sites.

Significant treatment and dispersion effects were seen in MS (PERMANOVA on treatment r-squared = 0.96, p < 0.0001, F = 125 Mississippi betadisper F = 61.871 p < 0.001) and WI (PERMANOVA on treatment r-squared = 0.39, p = 0.09, F = 2.85 betadisper F = 7.98, p < 0.001). 2-Dimensional results from the Non-metric Multidimensional Scaling (NMDS) analysis are presented in Figure 2 showing treatment differences in both MS (Figure 2A) and WI (Figure 2B).

Thirteen indicator species were found in the mixed preservative MS plot (Table 3); of those, 6 were highly correlated (corr Val = 1). Twenty-three indicator species were found in the mixed preservative WI plot (Table 4); of those, 8 were highly correlated (corr Val = 1). A total of 14 indicator species were identified with the copper treatments in MS (Table 5), three of which were highly correlated with copper treatments (Corr_val = 1) and the remainder were still correlated, but to a lesser extent (Corr_val = 0.894). A total of 7 indicator species were identified for copper exposed soils in WI (Table 6). Two species (Orbillia sp. and Rhytidhysteron rufulum) had the highest degree of association (Corr_val = 1), while the remaining species were less correlated (Corr_value = 0.894–0.0866). Twenty-five indicator species were found in the unexposed MS plot (Table 7); of those, 6 were highly correlated (corr Val = 1). A total of four indicator species were identified for the WI unexposed soil (Table 8).

Of the total 6,668 OTUs, 1,330 (20%) of the OTUs were classified using Funguild (Nguyen et al., 2016), which assigns functional guild information to taxonomic sequence data.

Unsurprisingly, the fungal soil communities differed between our two field sites that were separated by 995 miles and 13° of latitude. Overall, a greater number of OTUs were found in the WI site, presumably due to the more neutral soil pH and richer soil nutrient composition (see Table 1). Differences between treatments were also noted for both sites with greater difference between treatments in the harsher decay hazard climate (MS); treatments were also different in WI but with much greater variability.

Indicator species were determined for each site treatment combination and notable observations were the low diversity of wood saprobes present in the soil, which did not appear to be influenced by treatment history but did differ between test sites. Several ectomycorrhizal fungi were found to be associated with both mixed and Cu preservatives exposure, which agrees with previous literature (Meharg and Cairney, 2000; Fomina et al., 2005; Colpaert, 2008), suggesting that ectomycorrhizal fungi are able to persist and might be beneficial remediators for preservative exposed soils, and that long-term preservative exposure has little effect on the abundance and prevalence of this group of fungi.

Thirteen indicator species were found in the mixed preservative MS plot (Table 2); of those, 6 were highly correlated (corr Val = 1). Cenococcum geophilum is a mycorrhizal species associated with a diverse list of hosts with wide distribution and is routinely identified from soil, based on morphology of colonized roots (Douhan and Rizzo, 2005). Herpotrichiellaceae is a family of loculoascomycetes with black yeast anamorphs (Untereiner et al., 1995). Ceriporiopsis sp. include lignin degrading white-rot fungi studied for enzymes that can be used in biopulping (Ferraz et al., 2003). Neofusicoccum grevillea is a pathogen on Frivillea aurea causing a leaf spot (Sakalidis et al., 2013). The genus Neopaxillus is in the crepidotaceae family in the agaricales, N. dominicanus is a new species found in the Dominican Republic in 2011 (Vizzini et al., 2012).

Twenty-three indicator species were found in the mixed preservative WI plot (Table 3); of those, 8 were highly correlated (corr Val = 1). Of the highly correlated, Clavicorona taxophila is a saprotroph in the Clavariaceae family including mushroom-forming fungi with Clavicorona producing coralloid sporocarps (Birkebak et al., 2013). It is a small coral fungus that is currently red-listed in the UK as threatened (Evans et al., 2006). Nothojafnea cryptotricha is a mycorrhiza of eucalypts (Warcup, 1990). Archaeorhizomyces borealis is a pine associated species of the soil-inhabiting genus commonly isolated from environmental samples (Menkis et al., 2014). They constitute a significant component of the rhizosphere in fungal DNA community analyses, comprising up to one third of the total fungal community (Porter et al., 2008). Ganoderma lucidum belongs to a genus of white rot fungi that are considered one of the most important medicinal fungi worldwide and is an ingredient in many health products supposedly having anti-cancer, anti-aging, and antimicrobial functions (Paterson, 2006). Podospora intestinacea is a coprophilous fungus commonly isolated form dung and produces a perithecial resting structure that can persist in soil or dung (Watling and Richardson, 2010). Flagelloscypha citrispora is a basidiomycete that occurs on rotten logs and stumps (Reid, 1964). Species of the Coniothyrium genus have been used as biocontrol for Sclerotinia as they are mycoparasites of mycelium and sclerotia; this treatment has been used in sunflower production (Whipps and Gerlagh, 1992). Operculomyces laminatus is a chytrid formerly known as Rhizophlyctis harderi (Powell et al., 2011).

A total of 14 indicator species were identified with the copper exposure in MS (Table 4). The highest correlations between species and copper treatment were for Hygrocybe virginiana, Anthostomella leucospermi, and Cladophialophora chaetospira.

A total of 7 indicator species were identified for copper exposed soils in WI (Table 5). Orbillia sp. is an operculate ascomycete belonging to the family Orbilliaceae which are often nematophagous (Ahrén et al., 1998); members of this genus have been isolated from forest soils in Argentina (Allegrucci et al., 2009). It produces an orange cup shaped fruiting body and has been observed on copper treated field stakes from our Wisconsin field site. Efforts are currently underway to confirm these fruiting bodies as being Orbillia spp. Rhytidhysteron rufulum is a saprophyte or weak parasite on a variety of plants most commonly associated with pan-tropical environments (Murillo et al., 2009).

Twenty-five indicator species were found in the unexposed MS plot (Table 6); of those, 6 were highly correlated (corr Val = 1). Arthrinium arundinis is a plant pathogen reported to cause kernel blight of barley (Crous and Groenewald, 2013). Cryptococcus cuniculi was previously described as an anamorphic yeast that was isolated from rabbit feces and is now considered a basidiomycetous yeast (Findley et al., 2009). Oliveonia pauxilla is a holobasidiomycete fungus and a saprobe on dead leaf litter (Warcup and Talbot, 1962). Craterocolla cerasi is a non-culturable fungus in the sebacinaceae group of the Hymenomycetidae family with fungi involved in various mycorrhizal associations (Weiss et al., 2004). Flagelloscypha citrispora is a wood saprobe also found in the mixed preservative plot in WI described above.

A total of four indicator species were determined for the WI unexposed soil (Table 7). A group containing Neotropical ascomycetes (Coniochaetales sp.) was correlated (0.866), and has a related species, Coniochaetales lignaria, that has been evaluated for its ability to break down furans and other phenolics that are the result of lignocellulose pretreatment for hydrolysis (López et al., 2004). Septoglomus sp. was also found to be correlated with unexposed soils in WI. The Septoglomus genus contains mostly mycorrhizal species that are widely distributed (Rydlová et al., 2015). This isolate clustered with Septoglomus constrictum, which has been isolated in other studies associated with heavy metal contaminated mine spoil (Shetty et al., 1994). Geotrichum candidum was also correlated (0.866) with untreated WI soils and has been widely studied for its production of lipases (Sugihara et al., 1990), phenol oxidases (Assas et al., 2000) and a host of other enzymes that can effectively break down a wide range of chemical classes (Sun et al., 2008). Ophiostoma piliferum is a sapstain fungus associated with bark beetles (Klepzig, 1998). The overall low diversity of the unexposed WI soils could be attributed to habitat. The samples were taken from a grassy prairie area with virtually no canopy and very little coarse woody debris—a driver of fungal diversity (Zellweger et al., 2015)—in the vicinity.

Relative proportions of fungal guilds were generally conserved across sites and preservative exposures. Pie charts showing proportions of guilds classified for each treatment group are shown in Figure 3. Wood saprobes compromised a relatively small portion of the total fungi detected in each sample group, likely owing to the fact that isolations where made from soil and not wood. Higher proportions of ECM fungi were noted at the MS site, but this is likely due to the predominant pine overstory which is heavily dependent on ECM fungi for growth and survival (Svenson et al., 1991) and predominately lower pH soils.

Analysis at the functional guild level showed that wood saprobes also make up a relatively small proportion of the total soil fungal community (7.29% of the OTUs characterized) and that saprophytic fungi, many of which are capable of inducing soft rot fungal degradation patterns in wood, comprise a large portion. A total of 67 OTUs were classified as animal pathogens, which include Cryptococcus, Beauvaria, Metarhizium, Fusarium, Cordyceps, and others. Two additional OTUs were classified as animal symbionts, both belonging to the genus Rhodotorula. A total of 233 OTUs were classified as Arbuscular Mycorrhizae mostly contained in the genus Glomus. A total of 41 OTUs were classified as dung associated saprotrophs mostly Penicillium, Chaetomium, or Podospora.

A total of 147 OTUs were classified as Ectomycorrhizal fungi and included the genera Lactarius, Russula, Amanita, Tuber and several other commonly found fleshy terrestrial fungi. Several representatives within this group were found to be indicator species in the community analysis, Lactarius chrysorrheus-(mixed preservative associated in WI), Tuber sp. (Cu associated in WI), and Leccinum talamanaceae (Cu associated in MS). Ectomycorrhizal fungi have been found to possess abilities to biotransform and sequester toxins, such as arsenic in the form of arsenobetaine (Nearing et al., 2015) and other persistent heavy metals contaminating soil (Khan et al., 2000). Previous works to bio-augment and remediate treated wood waste and soil have revolved primarily around preservative tolerant species of polyporoid fungi (De Groot and Woodward, 1999; Illman and Yang, 2004) and have not fully explored the roles of these terrestrial fungi as potential remediators. Ectomycorrhizal fungi have been studied for their role in remediating heavy metal contaminated soil and several species that are known to be metal tolerant were detected in our study (Cenococcum geophilum, Amanita spp., and Pisolothus tinctorus). An important exception would be H. virginiana, which was found to be highly associated with Cu treatments in MS, but is actually classified as an undefined saprotroph since Hygrocybe species are considered to be neither saprotrophic nor mycorrhizal (Seitzman et al., 2010) but symbionts of mosses. A total of 17 OTUs were classified as endophytes and included many common plant pathogens (Phialophora, Cladosporium, Periconia macrospinosa). Representatives of this guild were found as indicator species in both unexposed (Herpotrichiellaceae sp.), mixed preservative exposed (P. macrospinosa) and Cu exposed soils (Phialophora europaea). A total of 590 OTUs were classified as saprophytes and included an extremely wide range of species from the entire spectrum of fungal taxonomy (lower fungi, such as Mortierella to higher basidiomycetes, such as Peniophorella pallida and Lycoperdon perlatum).

A total of 97 OTUs were classified as wood saprobes, which would include all of the typical wood decay fungi. Table 8 shows a breakdown of those fungi classified as wood saprobes by Funguild and their decay traits (soft, brown or white) with their occurrence across treatments.

A total of 14 brown rots, 38 soft rots and 37 white rots were identified and characterized from the soil samples (an additional 8 OTUs, containing 4 species of Psathyrella, Crepidotus sp. and Simocybe spp., were not characterized according to decay traits). Overall, a relatively low number of known wood decay fungi were obtained from our sampling, but these results will serve as important baseline data as we begin to focus on fungi that colonize and persist in solid wood and also highlights the importance of saprophytic fungi in the decay process. At both sites, higher abundance of wood rot fungi was noted for the exposed areas compared to unexposed areas. A relatively small number of wood saprobes were recovered in our sampling. Many of these species are well studied wood rot fungi or at least the genera, except for Trechispora spp. which is more common on decaying wood debris on the forest floor than solid wood. The less recognized species could be more difficult to culture or less likely to show fruiting in the field or even occur in later stages of decay. Our results could lead to more studies on these lesser emphasized wood rot fungi, especially in the preservative tolerance field. Proximity to remaining field stakes in the plot was not taken into consideration in this study, but could be a significant source of variation in these types of studies. Future studies to address these effects would provide much needed information on whether or not nutrient availability (i.e., untreated wood stakes) stimulates below ground fungal activity. Future studies will include more detailed information on wood preservative residues and breakdown products in the soil and would be an obvious next step in correlating long term preservative breakdown with fungal community composition. Another important consideration is whether present soil fungi directly correlate to wood decay. Continued studies will investigate transfer of soil fungi into untreated pine stakes and eventually preservative treated pine in field exposure.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.