The study was conducted in spruce-fir forests (Abieto-Piceetum oxalidozum) surrounding the Middle Ural copper smelter (Revda town, Sverdlovsk Oblast, Russia), which has been operating since 1940. The material was collected in three areas: unpolluted (UP), moderately polluted (MP), and HP areas located 30, 4, and 1–2 km upwind from the smelter, respectively (Supplementary Figure 1A). The areas are as similar as possible in terms of biotope and soil characteristics (Supplementary Table 1); however, they differ strongly in the degree of ecosystem damage caused by pollution (Table 1). To consider forest spatial variability, three sites (2,817–17,153 m² and 100–800 m from each other) were established within each area in September 2017. All autochthonous downed fir (Abies sibirica Ledeb.) and spruce [Picea abies (L.) H. Karst.] logs of the fourth and fifth decay classes [as per Storozhenko (1990); Supplementary Table 2] were considered for sampling. In each of the nine sites, we randomly selected 10 logs of at least 10-cm base diameter that were in contact with the ground and undamaged by large animals.

We measured the length of each sampled log with a laser distance measurer Disto A3 (Leica Geosystems, St. Gallen, Switzerland). The circumference of the top and base (right above the root collar, if present) was measured using a measuring tape with an accuracy of 0.5 cm. Wood resistance to penetration was measured using a dynamic penetrometer (Larjavaara and Muller-Landau, 2010); the time of tree death was determined through dendrochronological cross-dating; and taxonomic identification was performed through the microscopic inspection of resin canals and wood structure [for details, see Dulya et al. (2019)].

On each log, we marked three sampling points more than 1 m apart from each other (Supplementary Figure 1B and Supplementary Table 3). Within a 15-cm distance from the log, CWD-influenced litter subsamples were collected near each point and pooled. From 1.5 to 15 m from these logs and any other CWD, we collected and pooled three subsamples of CWD-uninfluenced litter (the distance between subsamples was equal to the distance between CWD-influenced subsamples). According to the sampling protocol, loose recently fallen plant debris (OL layer) was gently removed from the sampling point, and approximately 10 g of material was collected from the OF (partly decomposed fragmented material) and OH (well-humified) litter layers using a sterile plastic container. At each marked point on the logs, a disk was cut out with a chainsaw, and 10 g of wood fragments was extracted from each disk with bleach-sterilized forceps and pooled in a sterile plastic container. The samples from each microhabitat type were kept cool and then frozen within 2–8 h for genetic analysis. At the same sampling points, composite samples of 45–780 g of wood or litter were collected for chemical analyses. In further statistical analyses, wood, as well as the CWD-influenced and CWD-uninfluenced litter collected from/near the same log, was treated as belonging to the same sampling plot.

In the laboratory, samples intended for chemical analyses were weighed, dried at 65°C for 4 days, and reweighed to determine gravimetric moisture content (MC). Air-dried samples were ground and sieved through 2-mm sieves (MF 10 basic; IKA, Staufen, Germany). Acid-soluble and exchangeable forms of the main metallic pollutants (Cu, Pb, Cd, and Zn) were extracted with 20 ml of 5% HNO₃ and 0.05 N of CaCl₂ solutions, respectively, from 1 g of each sample. After 1 h of shaking and 24-h incubation at room temperature, the extracts were filtered through a paper filter with 8- to 12-mm pore size (Melior XXI, Moscow, Russia). Metal concentrations were measured with an atomic-absorption spectrometer AAS Vario 6 (Analytik Jena, Jena, Germany) following the USEPA Method 7000B (USEPA, 2007). Concentrations of P₂O₅ (extracted with 0.2 N of HCl) were analyzed with a PE-5400v spectrophotometer (Ecros, Saint Petersburg, Russia). The pH of the wood in distilled water extract (1:25) was measured after 5 min shaking using an inoLab 740 ionometer (WTW, Weilheim, Germany). Total Kjeldahl nitrogen content was measured in 0.5 g of forest litter and 1 g of wood using digestion and distillation units DK 20 and UDK 139 (VELP Scientifica, Usmate Velate, Italy). Ash content (fraction of dehydrated sample) was measured after combustion at 600°C for 6 h in a laboratory furnace (PM-16, Elektropribor, Saint Petersburg, Russia).

Each frozen sample destined for genetic analysis was tightly enclosed with sterile 4 × 4 gauze dressing and lyophilized. About 6–10 g of air-dried samples was homogenized using coffee grinders sterilized with bleach and UV radiation after each sample. DNA was extracted from sterile-weighted 90- to 105 mg of aliquots using NucleoSpin Soil Kit (MACHEREY-NAGEL, Düren, Germany). The homogenization of aliquots in lysis buffer (SL1, 50 μl of Enhancer SX) was conducted using a Precellys 24 homogenizer (Bertin Instruments, Montigny-le-Bretonneux, France) at 4,000 rpm for 30 s.

We amplified the internal transcribed spacer (ITS2) of rRNA using a two-step PCR protocol following the “16S Metagenomic Sequencing Library Preparation” protocol (Part #15044223 Rev. B, Illumina, San Diego, CA, United States). The primary amplification was done using the primers fITS7 (Ihrmark et al., 2012) and ITS4 (White et al., 1990) with added Illumina adapter overhang nucleotide sequences. The amplicons from the primary PCR were diluted 1:10 in sterile, nuclease-free water; and a second PCR was set up to add the Illumina flow cell adaptors and sample-specific Nextera XT (Illumina) indices. PCR was performed using Tersus high-fidelity DNA polymerase (Evrogen, Moscow, Russia). Amplicons were purified and normalized using the SequalPrep Normalization Plate Kit (Thermo Fisher Scientific, Waltham, MA, United States). All samples were pooled in equimolar amounts, and the size distribution of amplicons was assessed by high-throughput capillary gel electrophoresis (Fragment Analyzer System, Agilent Technologies, Santa Clara, CA, United States). Sequencing was performed on an Illumina MiSeq platform in paired-end mode using a MiSeq Reagent Kit v3 (600 cycles) at Evrogen (Moscow, Russia).

Sequencing libraries were demultiplexed using bcl2fastq v2.17.1.14 (Illumina). Primer sequences were trimmed from paired Illumina sequences using cutadapt v.3.1 (Martin, 2011); reads lacking an identifiable primer sequence were discarded. Low-quality reads were filtered out based on the maximum number of expected errors [maxEE > 3; Edgar and Flyvbjerg (2015)] using VSEARCH v.2.15.1 (Rognes et al., 2016). Correction of sequencing errors, paired-end assembly of the forward- and reverse-sequenced reads, and chimera removal using the consensus method were performed with DADA2 v.1.18.0 (Callahan et al., 2016). Denoised sequences were clustered into operational taxonomic units (OTUs) using SWARM v.3.0.0 (Mahé et al., 2021), with the local clustering threshold (d) equal to 5 nucleotides. Average within-OTU sequence similarity was 98.3% (Q1 = 97.7, Q3 = 99.2), which is in range with the empirically derived estimates for species hypotheses based on ITS (Kõljalg et al., 2013; Lücking et al., 2020; Tedersoo et al., 2020).

All denoised sequences were taxonomically identified against the curated reference database UNITE v.8.2 (Nilsson et al., 2019) using the hybrid approach of AMPtk v.1.4.1 (Palmer et al., 2018), as follows. Taxonomic assignments were obtained using two methods: global alignment with VSEARCH and a k-mer-based SINTAX classifier (Edgar, 2016). SINTAX results were used if taxa displayed a sequence similarity of <97% to the reference database. If sequence similarity was >97%, the result with the finest taxonomic annotation was retained. Post-clustering curation of the OTU table was performed using LULU v.0.1.0 (Frøslev et al., 2017). All steps of the bioinformatics analysis workflow were managed with Snakemake v.5.31.1 (Mölder et al., 2021).

Operational taxonomic unit counts with taxonomic assignments and associated sample metadata were handled with phyloseq package v.1.32.0 (McMurdie and Holmes, 2013) of R v.4.0.5 (R Core Team, 2021). Data were visualized with ggplot2 v.3.3.3 (Wickham, 2016). Sankey diagrams were constructed using ggalluvial v.0.12.3 (Brunson, 2017).

To account for unequal sequencing depths, we rarefied the data to a depth of 7,689 sequences per sample. Thus, three samples with lower number of reads were excluded from the analyses. For the assessment of community alpha diversity, we used OTU richness (S) and effective number of OTUs (Se) in a sample (Jost, 2006). As implemented in metagMisc package v.0.0.4 (Mikryukov, 2018), rarefaction was performed 1,000 times; and S and Se were averaged over the iterations. To avoid index switching artifacts (assignment of reads to incorrect samples during sequencing), prior dissimilarity analysis, we removed the OTUs with a relative abundance of less than 0.25% from each sample (Clavel et al., 2020).

To estimate community dissimilarity, we used Simpson’s index, which is robust to the difference in species richness between communities (Chao et al., 2006). It has an upper limit of one (samples have no OTUs in common) and a lower limit of zero (assemblages are nested – all OTUs from the smaller community are present in a larger community). Ordination was performed with principal coordinates analysis (PCoA) using vegan v.2.5-7 (Oksanen et al., 2020). To examine the differences in community composition, we used non-parametric permutational multivariate analysis of variance [PERMANOVA, 10,000 permutations; Anderson (2001)]. In the test for the difference in community compositions between the substrates, permutations were constrained within an area. The similarity of litter and wood samples in environmental parameters was illustrated with principal component analysis (PCA) using FactoMineR package v.2.4 (Lê et al., 2008).

To compare group means of diversity, similarity, and environmental variables, we used a hierarchical linear model with the substrate, microhabitat type, and pollution area as fixed effects and sampling plot as a random effect (nested within area) to consider the non-independence of individual samples. Concentrations of heavy metals were log-transformed before the analysis. p-Values were adjusted for multiple testing by false discovery rate (FDR) correction. The analysis was performed using lme4 v.1.1-26 (Bates et al., 2015). ANOVA and Tukey’s multiple comparison post hoc test were conducted using the packages car v.3.0-10 (Fox and Weisberg, 2019) and multcomp v.1.4-16 (Hothorn et al., 2008). To reveal pollution-induced changes in OTU richness, we used Cohen’s d effect size (standardized mean difference) and effsize package v.0.8.0 (Torchiano, 2016). Paired Wilcoxon signed-rank test was used to compare CWD-influenced and CWD-uninfluenced litter samples in the numbers of fungal OTUs. Functional annotation of ectomycorrhizal (EcM) taxa was performed using the FungalTraits database v.1.3 (Põlme et al., 2020).

The ratio of sampled fir and spruce CWD within each area mirrored tree stand composition (Table 1 and Supplementary Table 1). Metal concentrations in wood from polluted areas (HP and MP) were significantly higher than in the UP area. CWD from the HP area possessed significantly more P₂O₅ but less moisture and lower moss cover. Herbaceous vegetation, mainly represented by Circaea alpina, Linnaea borealis, Myosotis sylvatica, Oxalis acetosella, and Stellaria holostea, densely covered two-thirds of logs in the UP area but half of the logs in the MP area. Herbs were not observed on the CWD in the HP area, though tree seedlings occupied most logs in both polluted areas (Table 1).

Litter samples of polluted forests were more acidic (ΔpH = 0.99) and had higher metal and P₂O₅ concentrations than UP forests (Table 1). The highest ash content in the UP forest litter illustrates its highest mineralization degree. Within each area, heavy metal concentrations in wood were significantly lower than in forest litter, e.g., by four, 20, and 12 times for Cu in UP, MP, and HP areas, respectively. Wood samples were more acidic, less mineralized, and more saturated with water and possessed six times lower nitrogen content than forest litter (Table 1).

Litter close to CWD was moister, slightly but significantly more acidic, and more saturated with Zn and Cd than the litter collected distant from CWD (F₁,₈₇ > 6.08, p < 0.02). Differences between the substrates and microhabitats are reflected with the samples ordination along two PCA axis in Figure 2.

Illumina sequencing revealed 7,854 fungal OTUs (Supplementary Appendix 1), of which 6,960 (88.6%) were found in litter and 3,908 (49.8%) in wood. The most abundant and diverse phyla were Ascomycota (48.1% of OTUs and 41.4% of reads), Basidiomycota (37.1 and 54.4%), Mortierellomycota (2.3 and 1.5%), and Mucoromycota (1.5 and 0.5%) (Table 2). Other phyla and unclassified fungi comprised 11.0% of OTUs and 1.9% of reads. These five fungal groups were 1.1–3.2 times less diverse in wood than in forest litter.

The total number of OTUs was lower by 17.5% in the MP area and 22.2% in the HP area than in the UP area. The decrease of OTU total number in CWD-uninfluenced forest litter of the MP and HP areas was the strongest of all substrates (22.7 and 33.9%, respectively). OTU richness of four fungal groups declined in polluted areas by 16.0–38.6%, while Mucoromycota increased by 2.6–7.7%.

In CWD-uninfluenced litter, S and Se were lower in MP area (by 18.7 and 31.3%, respectively) and in HP area (by 28.8 and 52.7%, respectively) than in the UP territory (Table 2 and Figure 3). Basidiomycota and Ascomycota S decreased by 4.7–35.3% in polluted areas. Mucoromycota S was stimulated by pollution by 40.5 and 61.9% in MP and HP areas, respectively, while Mortierellomycota gained 3.5% of S in MP area and lost 6.4% in HP area. Other phyla together with unclassified fungi lost 33.0 and 30.2% of S in the MP and HP areas, respectively. Se response to pollution was similar to S for all target groups (Figure 3B). In accordance with these results, S and Se of most target groups in litter were negatively affected by acidity, P content, and Cu content (−0.74 < β < −0.28; Supplementary Table 4). In contrast, Mucoromycota S and Se were positively influenced by litter acidity and Cu concentrations (0.33 < β < 0.49), while Mortierellomycota were not responsive to any of the predictors.

Based on pollution-induced changes of total OTU number in CWD-uninfluenced forest litter, lower-level taxa were tested for their vulnerability to pollution (Figure 4). Glomeromycota and Olpidiomycota lost most OTUs in polluted areas, while Chytridiomycota diversity decreased by approximately 40%. Of species-rich Ascomycota and Basidiomycota taxa, the most vulnerable (i.e., reduced in diversity by more than half) were Botryosphaeriales, Pleosporales, Tubeufiales, Glomerellales, Microascales, Xylariales, Orbiliomycetes, Agaricaceae, Bolbitiaceae, Clavariaceae, Tricholomataceae, Auriculariales, Agaricostilbomycetes, and Pucciniomycetes. Less vulnerable taxa, which lost 50–20% of OTUs in polluted areas, included Capnodiales, Dothideales, Venturiales, Chaetothyriales, Helotiales, Rhytismatales, Thelebolales, Chaetosphaeriales, Diaporthales, Hypocreales, Sordariales, Pezizomycetes, Lecanoromycetes, Inocybaceae, Marasmioid clade, Pluteaceae, Psathyrellaceae, Cantharellales, Sebacinales, Thelephorales, Cystofilobasidiales, Tremellales, Cystobasidiomycetes, and Microbotryomycetes. Conversely, Microthyriales, Saccharomycetes, Entolomataceae, Strophariaceae, Atheliales, Hymenochaetales, and Polyporales lost <18% of OTUs. Mucorales, Umbelopsidales, Eurotiales, Cortinariaceae, Boletales, Russulales, Trechisporales, and Rozellomycota increased in OTU total number by 3.1–54.8% in MP area and by 0.1–48.9% in HP area. The changes of alpha diversity at lower taxonomic levels were quite similar to the changes revealed for the total OTU richness (Supplementary Figure 2).

The total OTU number of EcM fungi in CWD-uninfluenced litter decreased by 6.0 and 14.1% in MP and HP areas, respectively (Table 2); meanwhile, they increased in alpha diversity by 22.2 and 12.7% and in relative abundance by 2.4 and 2.9 times, exceeding 35% of fungal reads in the litter of the HP area. EcM lineages observed in the litter differed in the sensitivity to pollution. The most notable (2.2–84.4 times) abundance increase in both polluted areas was characteristic of /boletus, /pseudotomentella, /endogone, /hygrophorus, /amphinema-tylospora, /suillus-rhizopogon, /laccaria, and /russula-lactarius EcM fungal lineages. The abundance of /paxillus-gyrodon and /piloderma remained unchanged in the MP area and increased in the HP area by 5.1 and 3.0 times, respectively, while /otidea abundance increased in the MP area 10.8-fold and remained unchanged in HP area. The /hydnotrya, /tuber-helvella, /sebacina, /amanita, and /inocybe EcM fungal lineages were relatively non-sensitive to pollution (from 0.7-fold decline to 1.9-fold increase). The most sensitive EcM fungal lineages included /tomentellopsis, /tomentella-thelephora, /tricholoma, /wilcoxina, /rhodoscypha, /clavulina, and /tarzetta, which declined 20–90% in the MP area and 33–100% in the HP area.

The overall OTU richness in the wood from the MP and HP areas was 9.4 and 4.3% lower than in the UP forests. Depending on the fungal group, pollution had no effect or strongly positively affected S and Se in wood (Table 2, Figure 3 and Supplementary Figure 3). As a result, the ratio of alpha diversity between wood and litter samples significantly increased in the gradient of pollution for all fungi, Ascomycota, and Basidiomycota (Figure 5A). The fungal compositional dissimilarity between wood and forest litter decreased in the pollution gradient (Figure 6A and Supplementary Tables 5, 6), though it remained significant in all groups.

Of OTU-rich taxa, total OTU richness of Sordariales, Pleosporales, and Glomeromycota decreased in the wood of polluted areas by more than 50% (Figure 7), whereas Pluteaceae, Tricholomataceae, Thelephorales, Chaetosphaeriales, Venturiales, Orbiliomycetes, and Chytridiomycota decreased by 50–25%. A slightly detrimental effect of pollution on OTU number was observed for Auriculariales, Cantharellales, Polyporales, Saccharomycetes, Capnodiales, Thelebolales, Hypocreales, and Mortierellomycota. Richness of Cortinariaceae, Strophariaceae, Atheliales, Boletales, Hymenochaetales, Russulales, Sebacinales, Trechisporales, Tremellales, Dacrymycetes, Microbotryomycetes, Lecanoromycetes, Chaetothyriales, Eurotiales, Helotiales, and Rozellomycota was increased by pollution.

In all areas, wood harbored 2.5–3.7 times less EcM OTUs than forest litter (Table 2). EcM total OTU number increased in the wood of the MP area by 28.8% and in the HP area by 8.8%. EcM S increased more than twice, while Se remained unchanged, showing higher dominance of communities in polluted areas (Figure 3). This guild increased in the relative abundance in polluted wood (Table 2), leading to a decreased ratio of their abundance between litter and wood from 9.8 in the UP area to 4.0 and 2.3 in the MP and HP areas, respectively. Accordingly, EcM fungal compositional dissimilarity between wood and litter declined with the increasing pollution and became non-significant in polluted areas in contrast to other target groups (Figure 8A). The /boletus, /amphinema-tylospora, /pseudotomentella, /laccaria, /paxillus-gyrodon, /piloderma, and /russula-lactarius lineages, which showed high tolerance to pollution in the litter, also increased in abundance in polluted wood by 1.7–54.3 times (Figure 8C). Moderately tolerant to pollution, /inocybe, /amanita, /sebacina, /otidea, and /cortinarius also increased in polluted wood 2.8- to 83.4-fold. Of EcM fungal lineages negatively responding to polluted litter, /clavulina, /genea-humaria, /pachyphloeus-amylascus, /tomentellopsis, and /tricholoma suffered a 54–100% abundance decrease in polluted wood. Of the pollution-sensitive lineages, /wilcoxina increased in abundance in wood of the MP and HP areas by 10.0 and 8.4 times, respectively, while /tomentella-thelephora abundance was comparable with UP values. Notably, the sensitive lineages increased in abundance in dead wood due to certain pollution-tolerant species, such as Thelephora terrestris and Trichophaea spp.

Based on the difference of OTU occurrences between CWD-uninfluenced litter and wood, a few taxa significantly changed substrate preferences in polluted areas. Chytridiomycota, Dothideales (mainly due to Dothioraceae and Dothideaceae), Pezizales, Lecanoromycetes, Phacidiales, Mucoraceae, and Tricholomataceae became more frequent in wood; while Microthyriales, Strigulales, Chaetosphaeriales, Chaetomiaceae, Aspergillaceae, Piskurozymaceae, Boletales, Hydnangiaceae, and Physalacriaceae shifted toward forest litter (Supplementary Table 7).

Fungal alpha diversity was weakly influenced by within-area physicochemical variability of the substrates (Supplementary Table 8). In the UP area, only litter pH significantly negatively affected S of all fungi, Mortierellomycota, and the group “other phyla” (B = −89.8, B = −9.0, B = −19.7, respectively); Mucoromycota diversity negatively responded to litter moisture (B = −4.71 and B = −2.4 for S and Se, respectively) in HP area; and Mortierellomycota S positively responded to N concentration in wood of the MP area (B = 20.4, p < 0.05).

In the UP area, the total number of OTUs close to CWD was 0.5% (17 OTUs) lower than far from CWD due to all groups except for Mortierellomycota and Mucoromycota (Table 2). In polluted areas, CWD-influenced microhabitats’ contribution to fungal diversity in litter increased for most target groups, though remained unchanged for Mortierellomycota and decreased for Mucoromycota. As a result, CWD shadows harbored 2.6 and 7.9% more fungal OTUs in the MP and HP areas, respectively. Notably, the total number of EcM fungal OTUs in polluted areas remained higher in CWD-free litter. However, this difference decreased in the pollution gradient from 8.1% in the UP area to 4.9% in the HP area.

At the level of alpha diversity, CWD-influenced litter samples harbored higher S (Z = 0.32, n = 87, p = 0.003), which was ascribed to all Basidiomycota, Ascomycota, Mortierellomycota, and Mucoromycota (Z > 0.23, n = 87, p < 0.03; Supplementary Table 9). EcM fungal S was comparable between the microhabitats (Z = 0.004, n = 87, p = 0.97). Pollution-induced decrease of S and Se in CWD-influenced litter was less pronounced than in CWD-uninfluenced litter: 13.3% for S and 20.3% for Se in the MP area, and 20.9% for S and 37.4% for Se in the HP area. Therefore, the relative hospitability of CWD-shadowed microhabitats tended to increase in the gradient of pollution (Figure 5B). However, the effect of area type on the diversity ratio was non-significant, formally rejecting our third hypothesis.

In the UP area, most fungal groups (except Mucoromycota) significantly differed in composition between the CWD-influenced and CWD-uninfluenced litter (Figures 6B, 8B, D). In the MP area, only Ascomycota and Mortierellomycota contributed to the compositional differences between the microhabitats. None of the groups differed compositionally between the two microhabitats in the HP area.

At low taxonomic levels, pollution-induced changes of total OTU richness and alpha diversity in CWD-influenced forest litter resembled the patterns observed in CWD-uninfluenced forest litter (Figure 9 and Supplementary Figure 4). The exception was a smaller (compared with CWD-uninfluenced microhabitats) decrease in the diversity of Agaricaceae, Psathyrellaceae, Helotiales, Eurotiomycetes, and Mucorales and primarily wood-associated taxa from Tremellomycetes, Lecanoromycetes, Dothideales, Capnodiales, and Xylariales.

The test for pollution-induced changes in microhabitat preferences of OTUs inhabiting the UP and polluted areas showed that only Crepidotaceae became significantly more frequent in the localities close to CWD in the HP area (Supplementary Table 7).

The fungal diversity in forest litter decreased in the gradient of pollution, which corroborates the results of soil fungi research in the vicinity of this and other similar enterprises (Liao and Xie, 2007; Wang et al., 2007; Blagodatskaya et al., 2008; Hui et al., 2012; Bérard et al., 2014, 2016; Chen et al., 2014; Mikryukov and Dulya, 2017, 2018). In addition to heavy metal toxicity, the nearly complete elimination of herbaceous vegetation (Trubina, 2009, 2020) likely played an essential role in the impoverishment of fungal communities, which was exemplified by the substantial loss of biotrophic and saprotrophic species associated with herbaceous plants and their litter. In concordance with previous results (Baldrian, 2010; Torres-Cruz et al., 2018; Mikryukov et al., 2020), eurytopic soil saprobes (e.g., Mortierellomycota, Mucoromycota, Eurotiomycetes, and Helotiales) exhibited high tolerance to soil pollution with heavy metals. Unexpectedly, EcM fungi increased in diversity and doubled in relative abundance in polluted litter. Notably, Agaricomycetes representatives are primarily responsible for this effect, while EcM Pezizomycetes (except for Tuber spp., Hydnotrya spp., and Otidea leporina) are as sensitive to pollution as most saprotrophic representatives of this class. The pollution-induced stimulation of EcM fungi supports previous findings of increased EcM abundance in the upper soil horizons of the studied polluted areas (Veselkin, 2004).

We found subtle but systematic differences between abiotic parameters in CWD-shadowed and CWD-free litter. Subtle differences in pH, microclimate, and the proportion of plant detritus fractions between CWD-shadowed and CWD-free microhabitats were previously shown in coniferous, beech, and eucalypt forests and proved to be influential for mollusks, invertebrates, and microbial biomass (Marra and Edmonds, 1998; Evans et al., 2003; Spears et al., 2003; Kappes et al., 2007; Goldin and Hutchinson, 2015). CWD-shadowed litter harbored a distinct composition and higher alpha diversity of fungal communities. However, these results must be treated with caution because lichenized and wood-decomposing fungi, yeasts known from xylophagous arthropods’ frass, as well as the parasites of lichens, polypores, and myxomycetes, were more frequently identified in CWD-shadowed litter than those far from CWD, implying that mycelium, propagules, and dead remnants raining down from wood have a strong positive influence on nearby soil biodiversity estimates. The direct link between fungal communities in wood and underlying litter was also confirmed by a 20% lower compositional dissimilarity between wood and litter samples collected from the same sampling plot when compared with samples collected from different sampling plots (Supplementary Table 10).

Along with pollution load, the contribution of CWD-shadowed microhabitats to litter fungal diversity increased. Mucoromycota, Mortierellomycota, and most soil saprotrophic orders from Eurotiomycetes and Leotiomycetes possessed higher average numbers of OTUs in the CWD-influenced litter of all areas. With the “fungal” rain from CWD, these pollution-tolerant groups could lead to an overestimation of the capacity of CWD-shadowed litter to maintain fungal gamma and alpha diversity in polluted areas. The expansion of pollution-tolerant EcM and saprotrophic fungi also explains the pollution-induced compositional convergence of communities from CWD-influenced and CWD-uninfluenced microhabitats. Additionally, the leaf and root pathogens and symbionts of herbaceous plants, which were more frequent in CWD-free microsites in UP forests, contributed to the compositional and diversity differences between the microhabitats in those areas. The decline of this group, which was caused by the toxicity-induced elimination of potential hosts in polluted areas (Trubina, 2009, 2020), may have led to the convergence of these two microhabitat types.

Notably, there were two to three times more fungal OTUs in the forest litter than in wood. This proportion corresponds to the results obtained from a similar sampling scheme (Mäkipää et al., 2017). The distinct composition of wood fungal communities and the diversity gap between wood and litter indicate that the unintentional influx of DNA from the outside into a piece of CWD is very limited at this decomposition stage.

Considering the low saturation of wood with heavy metals and its high water-holding capacity, we predicted that the impoverishment of fungal communities in CWD would be lower than in HP forest litter. Conversely, the fungal alpha diversity in wood from the moderately polluted area was similar to that of the unpolluted area and increased in HP forests, leading to a higher proportion of forest floor fungal diversity being supported by CWD. One of the primary factors of wood dwellers’ welfare is the availability of woody debris as forage or habitat (Heilmann-Clausen and Christensen, 2003; Bässler et al., 2010; Josefsson et al., 2010; Runnel and Lõhmus, 2017; Runnel et al., 2021). Therefore, the increased tree dieback and treefall caused by pollution close to the smelter (Vorobeichik et al., 2014) could promote fungal diversity in wood. This effect can hardly be neutralized by the observed levels of toxicity in wood. According to metal-tolerance tests, fungi can withstand more than 25 μg/g of water-soluble copper salts (Englander and Corden, 1971; Colpaert and Van Assche, 1987; Daniel and Nilsson, 1988; Woodward and De Groot, 1999). According to the results of chemical analyses, in the heavily polluted area, only one-third of wood samples with 26.6–87.6 μg/g of exchangeable copper forms (which include water-soluble forms) could exceed this threshold, while 90% of forest litter samples contained 28.0–422.0 μg/g of exchangeable copper compounds.

The pollution-induced convergence of fungal diversity and composition between forest litter and wood provide indirect evidence of CWD sheltering soil fungi in polluted areas. However, in addition to typical wood dwellers, the taxa that did not decrease (or even increased) in diversity and relative abundance in polluted wood were the eurytopic soil saprobes and EcM fungi that tolerate polluted soil. Decayed wood can hardly be vital for these groups close to the smelter.

By contrast, saprotrophic Agaricomycetes suffered from a heavy pollution-induced diversity decline in forest litter (as was shown in this and previous works for southern taiga and pre-forest-steppe pine-birch forests; Mikryukov et al., 2020). Therefore, the revealed significant repartition of surviving Tricholomataceae OTUs between wood and forest litter may indicate that CWD can serve as one of the last harbors for this vulnerable group in polluted environments. We focused on saprotrophic members of this taxon and, by using publicly available data on fruiting body observations, classified 88 OTUs annotated to the species level by substrate specialization into terricolous, wood-dwelling, and wood-terricolous species (fruiting on soil, wood, or both substrates, respectively; Supplementary Table 11). The former and latter groups significantly repartitioned toward the wood in the heavily polluted area (Supplementary Figure 5) due to the disproportionate decrease in occurrence within polluted litter and wood. This indicates a sheltering function of CWD in inhospitable environments, even for species considered terricolous. Thus, a peculiar delimitation of species based on substrate requirements provided the opportunity to discern the importance of CWD for non-wood-dwelling fungi. It would be informative to perform a similar analysis of multispecies groups, which repartitioned toward wood under pollution effects (i.e., Pezizomycetes and Lecanoromycetes).

Some of these findings contribute to knowledge about the ecological requirements of poorly studied fungal groups. For instance, we registered Archaeorhizomycetes in a large number of CWD pieces (including in the HP area), which supports their single observation in wood by Rajala et al. (2011) and verifies that Archaeorhizomycetes tend to inhabit acidic and organic-rich substrates (Pinto-Figueroa et al., 2019). Many Mortierella OTUs were abundant and frequent in wood. Of these, Mortierella gemmifera, Mortierella pulchella, and Mortierella humilis have already been found in well-decayed Norway spruce CWD (Baldrian et al., 2016). Here, we show that for Mortierella fimbricystis, M. gemmifera, Mortierella jenkinii, and M. pulchella, wood can be preferential and thus a potentially important substrate in dark coniferous forests (Supplementary Figure 6).

This work has two obvious limitations regarding replication. First, our work was performed in the vicinity of a single copper smelter and may not be indicative of other heavy metal pollution sources in boreal and temperate forests. However, the patterns of forest litter fungal response to industrial pollution revealed in our previous work in the southern taiga near the Middle Ural copper smelter and birch-pine forests near the Karabash copper smelter (Southern Urals, pre-forest-steppe subzone) are largely similar (Mikryukov et al., 2020). Second, our study plots were geographically aggregated to the source of pollution by distance, which limits their statistical independence and adds an area effect that confounds the pollution effect.