Potential links between wood‐inhabiting and soil fungal communities: Evidence from high‐throughput sequencing

Abstract Wood‐inhabiting fungi (WIF) are pivotal to wood decomposition, which in turn strongly influences nutrient dynamics in forest soils. However, their dispersal mechanisms remain unclear. We hypothesized that the majority of WIF are soil‐borne. For this reason, the presented research aimed to quantify the contribution of soil as a source and medium for the dispersal of WIF to deadwood using high‐throughput sequencing. We tested effects of tree species (specifically Schima superba and Pinus massoniana) on the percentage of WIF shared between soil and deadwood in a Chinese subtropical forest ecosystem. We also assessed the taxonomic and ecological functional group affiliations of the fungal community shared between soil and deadwood. Our results indicate that soil is a major route for WIF colonization as 12%–15% (depending on the tree species) of soil fungi were simultaneously detected in deadwood. We also demonstrate that tree species (p < 0.01) significantly shapes the composition of the shared soil and deadwood fungal community. The pH of decomposing wood was shown to significantly correspond (p < 0.01) with the shared community of wood‐inhabiting (of both studied tree species) and soil fungi. Furthermore, our data suggest that a wide range of fungal taxonomic (Rozellida, Zygomycota, Ascomycota, and Basidiomycota) and ecological functional groups (saprotrophs, ectomycorrhizal, mycoparasites, and plant pathogens) may use soil as a source and medium for transport to deadwood in subtropical forest ecosystem. While 12%–62% of saprotrophic, ectomycorrhizal, and mycoparasitic WIF may utilize soil to colonize deadwood, only 5% of the detected plant pathogens were detected in both soil and deadwood, implying that these fungi use other dispersal routes. Animal endosymbionts and lichenized WIF were not detected in the soil samples. Future studies should consider assessing the relative contributions of other possible dispersal mechanisms (e.g. wind, water splash, water dispersal, animal dispersal, and mycelial network) in the colonization of deadwood by soil fungi.

son, the presented research aimed to quantify the contribution of soil as a source and medium for the dispersal of WIF to deadwood using high-throughput sequencing. We tested effects of tree species (specifically Schima superba and Pinus massoniana) on the percentage of WIF shared between soil and deadwood in a Chinese subtropical forest ecosystem. We also assessed the taxonomic and ecological functional group affiliations of the fungal community shared between soil and deadwood. Our results indicate that soil is a major route for WIF colonization as 12%-15% (depending on the tree species) of soil fungi were simultaneously detected in deadwood. We also demonstrate that tree species (p < 0.01) significantly shapes the composition of the shared soil and deadwood fungal community. The pH of decomposing wood was shown to significantly correspond (p < 0.01) with the shared community of wood-inhabiting (of both studied tree species) and soil fungi. Furthermore, our data suggest that a wide range of fungal taxonomic (Rozellida, Zygomycota, Ascomycota, and Basidiomycota) and ecological functional groups (saprotrophs, ectomycorrhizal, mycoparasites, and plant pathogens) may use soil as a source and medium for transport to deadwood in subtropical forest ecosystem. While 12%-62% of saprotrophic, ectomycorrhizal, and mycoparasitic WIF may utilize soil to colonize deadwood, only 5% of the detected plant pathogens were detected in both soil and deadwood, implying that these fungi use other dispersal routes. Animal endosymbionts and lichenized WIF were not detected in the soil samples. Future studies should consider assessing the relative contributions of other possible dispersal mechanisms (e.g. wind, water splash, water dispersal, animal dispersal, and mycelial network) in the colonization of deadwood by soil fungi.

K E Y W O R D S
deadwood, fungal dispersal, soil fungi, subtropical forest, wood-inhabiting fungi

| INTRODUC TI ON
Deadwood represents an important carbon (C) pool in global forest ecosystems, contributing approximately 8%, or 73 petagrams of C, to the total carbon stock (Pan et al., 2011). The decomposition of deadwood is crucial to carbon dynamics and nutrient cycling of forest ecosystems (Fukasawa, 2018;Hoppe et al., 2016;Rajala, Peltoniemi, Pennanen, & Mäkipää, 2012). It is a complex ecological process, influenced by diverse factors, such as climate, substrate quality (e.g. C:N ratio, moisture levels, and lignin content) as well as the abundance, composition, and activity of decomposer communities (Fukasawa, 2018;Liu, Schaefer, Qiao, & Liu, 2013;Purahong, Krüger, Buscot, & Wubet, 2016). Wood-inhabiting fungi (WIF) are considered to be the most important class of wood decomposers due to the wood decomposition enzymes, for example, oxidoreductases and hydrolases, that they secrete .
Wind is one of the most common dispersal mechanisms, and may play a significant role in WIF dispersal (Dighton & White, 2005;Jacobsen, Kauserud, Sverdrup-Thygeson, Bjorbaekmo, & Birkemoe, 2017;. Insect dispersal may also play an important role for WIF distribution as several species of saproxylic beetles in temperate forest have been found to be the vectors of many WIF species, including Fomitopsis pinicola, Fomes fomentarius, Trichaptum abietinum, and Trametes versicolor (Jacobsen et al., 2017). Nevertheless, we previously found no significant differences in either WIF richness or community composition between insect-excluded deadwood (of Schima superba and Pinus massoniana) and control material, and direct insect associated fungi, for example, insect parasites and endosymbionts were seldom detected in the deadwood (Purahong et al., 2017). This finding indicates either that insects are less important for WIF distribution in subtropical forests than in temperate forests or that WIF transported by insect vectors can also reach deadwood via other mechanisms. A large proportion of the deadwood in forest ecosystems is located on the forest floor and will thus come into direct contact with soil at some point of the decomposition process (Song, Vail, Sadowsky, & Schilling, 2015). A previous study in boreal forest found that soil contact is significantly correlated with WIF community composition, implying that WIF in this ecosystem use soil as a medium to colonize deadwood, and that soil contact may influence WIF community assemblage, subsequently affecting the wood decomposition process (Rajala et al., 2012).
Hence, fungal transport from soil to deadwood may involve multiple dispersal mechanisms. Following wood colonization, fungal mycelium can grow out into the soil and form a network through cords or rhizomorphs, which, in some species (e.g. Armillaria spp.), can cover areas ranging from several square meters to 1,000 ha (Heaton et al., 2012;Mihail & Bruhn, 2005). Such fungal mycelial networks can expand through soil to colonize new food sources, for example, leaf litter and deadwood, located on the surrounding forest floor (Boddy, Hynes, Bebber, & Fricker, 2009). Fungal spores and mycelial fragments are released from deadwood and can be further transported by wind and water to the soil surface and subsurface (Dighton & White, 2005;Nawaz et al., 2016). Thus, soil can be considered both a source and medium for the transport of wood-inhabiting fungi.
In this study, we had three aims. First, to evaluate the contribution of soil as a source and medium in the colonization of woodinhabiting fungi to deadwood using fungal community datasets derived from deadwood (Purahong et al., 2017) and soil samples (Schuldt et al., 2015) collected from a Chinese subtropical forest ecosystem. Second, to test effects of tree species identity on the percentage of WIF shared between soil and deadwood. Third, to assess the taxonomic and ecological functional group affiliations of the fungal communities found in both deadwood and soil. We hypothesized that a significant proportion of wood-inhabiting fungi from diverse taxonomic and ecological functional groups use soil as both a source and medium for dispersal to deadwood. Tree species identity, as characterized by certain wood physicochemical parameters, was found to be the main factor influencing the taxonomic and functional diversity of soil fungi colonizing deadwood.
The study area had a mean annual temperature of 15.1°C, with a minimum of −6.8°C recorded in January and a maximum of 38.1°C measured in July. The elevation within the area ranged from 251 to 903 m a.s.l. and levels of tree and shrub species richness from 25 to 69 species per 900 m 2 plot. The soil and deadwood samples are suitable for comparison as they were collected from the same plots during the same period (August-September 2012).
Soil fungal communities were sampled from the upper soil layer (0-10 cm) of the 27 forest plots as described by Schuldt et al. (2015). Briefly, eight samples (one of which was collected near the deadwood logs used for WIF community analysis) were collected from each 900 m 2 plot and pooled to obtain composite samples (Schuldt et al., 2015). WIF were collected from the deadwood (diameter = 10 ± 2 cm and length = 25 ± 1 cm) of two tree species, S. superba (family Theaceae) (57 samples) and P. massoniana (family Pinaceae) (58 samples), which had been harvested in the vicinity of the study area, placed among the 27 comparative study plots in August 2010 and allowed to decompose for 2 years (Purahong et al., 2017). Following the decomposition period, two 2-cm thick slices, one from the margin, and one from the center of the sample, were sawed from each deadwood sample, kept frozen at −20°C and transported on dry ice to Germany for further physicochemical and molecular analyses. During the molecular WIF analysis, all of the bark was removed from the deadwood samples before homogenization with liquid nitrogen and a swing mill (Retsch, Haan, Germany).
Microbial DNA was extracted from 1 g portions of the compos-

T T G G T C A T T T A G A G G A A G T A A -3′] and ITS4 [5′-T C C T C C G C T T A T T G A T A T G C -3′]) to
amplify the entire fungal internal transcribed spacer (ITS) rRNA region (Gardes & Bruns, 1993;White, Bruns, Lee, & Taylor, 1990). The amplified fragments were then subjected to 454 pyrosequencing and bioinformatics analysis as described in detail by , Purahong et al. (2017) and Schuldt et al. (2015). We did a unidirectional sequencing of the amplicons from the ITS4 end (reverse primer), thus the fungal ITS2 sequences were used for further analysis. The entire ITS region (Tedersoo et al., 2015) and ITS2 is recommended for metabarcoding (Ihrmark et al., 2012;Tedersoo et al., 2015). Fungal ITS OTU representative sequences were first classified against the dynamic version of the UNITE fungal ITS sequence data-  Table S1, Supporting Information. Representative sequences of fungal operational taxonomic units (OTUs) were assigned to ecological functional groups based on sequence similarity (≥90%) using the default parameters of the GAST algorithm (Huse et al., 2008) against the reference dataset (Tedersoo et al., 2014). In addition, FUNGuild was also used to assign the ecological functional groups of WIF (Nguyen et al., 2016). A comparison of the results obtained by these functional group assignment approaches is presented in Table S1, Supporting Information.

| Statistical analysis
In this study, we focused on the fungal OTUs (123) detected in both soil and deadwood samples. As the HTS sequence datasets were processed together, the OTUs present in both datasets refer to the same fungi. The data concerning shared communities that were used for statistical analysis are provided in the Supporting Information (Table S1).
Three-dimensional non-metric multidimensional scaling (3D-NMDS) ordination based on presence/absence data and the Jaccard dissimilarity measure coupled with the envfit function of the vegan package in R were used to investigate and visualize correlations among the factors that influence shared soil-deadwood fungal community composition in P. massoniana and S. superba. 3D-NMDS worked better than a 2D-NMDS for our data, as a result of a lower stress value for the former. We repeated 3D-NMDS coupled with envfit for each tree species to determine the factors that influence shared soil-deadwood fungal community composition in the respective species (P. massoniana and S. superba). The effect of tree species on the shared soil-deadwood fungal composition was analyzed using PERMANOVA based on presence/absence data and the Jaccard dissimilarity measure in the PAST program version 2.17c (Hammer, Harper, & Ryan, 2001

| Soil is an important route of WIF dispersal
Our results demonstrate that soil is an important route for the colonization of soil fungi to deadwood in Chinese subtropical forest ecosystem. A taxonomically diverse array of fungi with various ecological functional groups was detected in soil and deadwood samples (Table 1; Figure A1). We detected a total of 123 fungal OTUs in both soil and deadwood samples. This finding suggests that at least 12% of the total WIF community (997 detected OTUs, of which 12% and 15% were detected in S. superba and P. massoniana, respectively) use soil as a source and transport medium to deadwood. This proportion of fungal OTUs shared between soil and deadwood identified in the Chinese subtropical forest is consistent with a previously reported proportion in a temperate forest (~10%), but much lower than the reported proportion in a boreal forest (~50%) (Mäkipää et al., 2017;van der Wal, klein Gunnewiek, & de Boer, 2017). This may be related to differences in biomes and/or the role of deadwood in each biome (Mäkipää et al., 2017;Purahong et al., 2017;van der Wal et al., 2017).
Interestingly, despite similarities in fungal community composition in the soil and minimum distance between the deadwood of the two tree species (S. superba and P. massoniana) in each experimental plot, numbers of fungal OTUs detected in soil and deadwood of the two species differed (Table 1). P. massoniana harbored most of the OTUs (94%) shared between soil and wood-inhabiting fungi; among these shared OTUs, 54 (46.5%) were detected in P. massoniana but not S. superba wood. In contrast, S. superba harbored 69 OTUs (56% of the total shared community), most of which were also detected in P. massoniana (62 OTUs, 90%), as only seven (10%) WIF OTUs were specifically associated with this tree species.
Furthermore, the two species differed significantly in terms of WIF community transported via the soil (PERMANOVA, F = 8.80 (presence/absence data, Jaccard dissimilarity measure), p < 0.001, 999 permutations, Figure 1). Spatial distance is one of the most important drivers of fungal community assembly, and this factor could have affected the validity of the experiment (Peršoh, 2015;Purahong, Krüger, et al., 2016;Talbot et al., 2014). However, we minimized the effect of spatial distance by spacing deadwood samples from the two tree species at a set minimum distance in each experimental plot. Thus, our results clearly indicate that tree species identity as defined by wood physicochemical properties including initial N content (highly correlated with initial C content), pH of decomposed wood, initial C:N ratio, and initial total lignin content, strongly influences the WIF community transported from soil to deadwood ( Figure 1). The pH of the decomposed wood samples was significantly correlated with the shared soil and deadwood fungal community in both S. superba and P. massoniana (Table 2). This is consistent with findings that under controlled conditions WIF species can significantly change the pH of colonized deadwood after 2-4 weeks (Humar, Petrič, & Pohleven, 2001). pH changes in decomposing wood can influence subsequent fungal communities as the optimal pH for fungal growth and reproduction rates is species-specific (Hoppe et al., 2016;Purahong, Krüger, et al., 2016;Yamanaka, 2003).

| Relative proportion of the fungal phyla, classes, and families transported from soil to deadwood
Our data suggest that the two dominant fungal phyla in deadwood-Ascomycota and Basidiomycota (Hoppe et al., 2016;Purahong, Wubet, Krüger, & Buscot, 2018;Rajala et al., 2012)may both be potentially dispersed to deadwood via the soil in Chinese subtropical forest ecosystem. This pattern was consistent in both studied tree species (Ascomycetes and Basidiomycetes accounting for 53%-59% and 36%-42% of the total shared community, respectively) ( Figure A1). In addition, Zygomycota and Rozellida fungi may be dispersed via soil. The Ascomycota identified in S. superba and P. massoniana deadwood mainly be- Marasmiaceae, and Thelephoraceae). At fine taxonomic resolution, that is, genus and OTU levels, the two tree species differed greatly in terms of WIF community composition ( Figure A1). Notably, although the same pool of fungal OTUs was present in the soil of all TA B L E 1 Numbers of total, specific, and shared wood-inhabiting fungal OTUs (separated according to ecological functional groups) identified in soil and Pinus massoniana and Schima superba deadwood samples  experimental plots, specific sub-pools of these OTUs successfully colonized deadwood of each tree species (Figure 1), in accordance with previous findings (Hoppe et al., 2016). We conclude that fungi from the phyla Ascomycota and Basidiomycota potentially use the soil as a source and transport medium to colonize deadwood. We also provide evidence that other fungal phyla which are less frequently detected in deadwood, such as Zygomycota and Rozellida, may use soil as a medium for dispersal to deadwood.

| Proportion of WIF functional groups dispersed via the soil medium
Diverse functional groups of fungi are present in deadwood Purahong et al., 2017), but not all of them are transported to deadwood via soil. We expected to find that free-living fungal functional groups that can inhabit either soil or detritus spheres (such as saprotrophic fungi) can more readily use soil as a means to colonize deadwood than mycoparasites, endophytes (fungal endophytes and animal endosymbionts), or plant pathogens. This is because mycoparasites, endophytes, and plant pathogens have complex lifestyles, requiring not only fungal propagules, but also suitable hosts, to be present in or on the deadwood following dispersal. However, we found that saprotrophs, ectomycorrhiza, mycoparasites, and plant pathogens are all potentially transported via soil (  (Table 1). A recent study also found that a high proportion of ectomycorrhizal fungi (85%) in boreal forest use soil as a source and medium for transport to deadwood (Mäkipää et al., 2017). However, for Thelephorales (e.g. Tomentella), insects may also play a large role in their dispersal (Lilleskov & Bruns, 2005). It should be noted that F I G U R E 1 Three-dimensional non-metric multidimensional scaling (3D-NMDS) ordinations of wood-inhabiting fungal community composition in Pinus massoniana (green) and Schima superba (red) deadwood (calculated using data for the 123 fungal OTUs detected in both deadwood and soil samples) based on presence/absence data and the Jaccard dissimilarity measure. The NMDS ordination was fitted to factors describing wood physicochemical properties (significant factors p < 0.01 are shown as vectors, with statistical values presented in the table). PERMANOVA using presence/absence data and the Jaccard dissimilarity measure was used to test the effect of tree species on wood-inhabiting fungal community composition (statistical significance is based on 999 permutations) although the role of ectomycorrhizal fungi in deadwood decomposition remains unclear, there is increasing evidence that certain ectomycorrhizal fungi may be facultative saprotrophs (Lindahl & Tunlid, 2015;Rajala et al., 2012;Rajala, Tuomivirta, Pennanen, & Mäkipää, 2015).
The most commonly detected WIF saprotrophs in our samples in-

| Links between wood-inhabiting and soil fungal communities
There are four hypotheses for a shared occurrence of fungal OTUs between the deadwood and the soil (Mäkipää et al., 2017  were present in both the soil and the wood. In our study we were not able to quantify the relative important of each hypothesis but we can assess the overall contribution of soil as source and medium for transport of wood-inhabiting fungi to deadwood. We expected the number of fungal OTUs migrating from the initial deadwood to the soil to be low, as we did not detect any of the fungal endophytes in the fungal community shared between soil and deadwood. Fungal OTUs that use soil as source and/or medium to colonize deadwood could follow any of the four migratory patterns, because they must be transported to, survive in, and/or colonize the soil. As we removed the bark from deadwood samples before homogenization and DNA extraction, fungi attached to the bark that had not penetrated the inner tissue of the deadwood (i.e. soil-or wind-dispersed fungal spores that had attached to the bark) should have been excluded from our analysis. Mäkipää et al. (2017) studied on soil-and deadwood-inhabiting fungal communities in boreal forests and found that these communities interact along the decay gradient of Norway spruce logs, and a relatively high proportion of the total fungal community is present in both soil and deadwood. Mäkipää et al. (2017) also found that WIF locally influence the soil-inhabiting fungal communities at all decay stages because certain WIF only occur in the soil under specific decaying logs, but it is impossible to determine how many soil fungal OTUs colonized the deadwood due to the study's experimental set-up. To enable this, a long-term, time-series analysis (from initial decay to late decay stages) of the fungal OTUs present in both soil and deadwood would be required.

| Potential biases of the fungal datasets
In this study the sequence data were generated with the pyrosequencing sequencing technology (no longer available). However, we achieved substantial numbers of sequence reads per sample to reasonably infer the fungal diversity in soil and deadwood. The TA B L E 2 Goodness-of-fit statistics (r 2 ) for factors fitted to the three-dimensional non-metric multidimensional scaling (3D-NMDS) ordinations of wood-inhabiting fungal community composition in Pinus massoniana and Schima superba deadwood (calculated using data for 116 and 69 fungal OTUs shared between soil samples and deadwood of P. massoniana and S. superba, respectively) based on presence/absence data and the Jaccard dissimilarity measure sequencing depths of 10,000 and 3,077 sequences per sample were used for the soil and deadwood samples, respectively (Purahong et al., 2017;Schuldt et al., 2015). The primer used (ITS1F and ITS4) represent genetic markers known to carry possible bias toward amplification of basidiomycetes and ascomycetes, respectively (Bellemain et al., 2010). We thus, carefully interpreted the results from this study as with the current primer set we may not amplify the total taxa of Zygomycota and other fungal phyla.

ACK N OWLED G M ENTS
Our work was partly funded by the DFG-German Research Foundation funded projects WU 614/1-3 and BU 941/12-3. We thank the administration of the Gutianshan National Nature Reserve for granting research permission, and gratefully acknowledge the support provided by Bernhard Schmid, Keping Ma, Guillaume Lentendu, Xiaojuan Liu, Yang Bo, Chen Lin, and the entire BEF-China team. We thank Ali Nawaz for reading manuscript.

CO N FLI C T O F I NTE R E S T S
None declared.

AUTH O R CO NTR I B UTI O N S
T.W. and W.P. conceived the study, W.P. and T.W. analyzed the data and wrote the manuscript, K.P., C.W., T.W., H.B., and F.B. setup the field experiments. F.B., H.B., C.W., and K.P. contributed data and edited the manuscript.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
The