1 INTRODUCTION

Boreal forests cover large parts of the northern hemisphere and are persistent carbon (C) sinks of global importance (Pan et al., 2011). Two-thirds of the boreal forest area are managed in some way (Gauthier et al., 2015). Thinning is a common practice applied to increase revenue (Royal Swedish Academy of Agriculture and Forestry, 2015), and due to strong nutrient limitation, nitrogen (N) fertilization is also considered a cost-effective practice to increase forest yields (Nohrstedt, 2001; Saarsalmi & Mälkönen, 2001; Tamm, 1991). In addition, fertilization can increase C stocks below-ground (Högberg et al., 2006; Janssens et al., 2010; Maaroufi et al., 2015), but the underlying mechanisms are still poorly understood (Mayer et al., 2020). In boreal forests, decomposing organic matter accumulates in a purely organic layer (equivalent to the ‘mor layer’ or O-horizon) on top of the mineral soil and accounts for a significant, and dynamic, fraction of below-ground C (Deluca & Boisvenue, 2012). The intuitive explanation that fertilization drives C storage in this organic topsoil by increasing plant litter input is poorly supported (Janssens et al., 2010; Jörgensen et al., 2021a; Kyaschenko et al., 2017b). Instead, increased C stocks following fertilization have primarily been attributed to decreased rates of CO2 release from the soil (Janssens et al., 2010; Maaroufi et al., 2015; Zak et al., 2008).

Trees appear to regulate below-ground processes by altering root traits (Adamczyk et al., 2019; Sun et al., 2018), production of fine roots (Forsmark et al., 2021) and fungal associations (Clemmensen et al., 2013; Kyaschenko et al., 2019), suggesting that several root-related mechanisms interplay to promote C accumulation below-ground. Potentially, increased C stocks in fertilized stands can be explained by a higher C use efficiency of roots and micro-organisms (Forsmark et al., 2021; Manzoni et al., 2017, 2018), resulting in higher C allocation to biomass production and lower C losses from respiration. In boreal forests, soil fungi regulate C dynamics, both by providing source material for soil organic matter (Clemmensen et al., 2013) and as drivers of decomposition (Kyaschenko et al., 2017b; Lindahl et al., 2021). Increased C use efficiency of fungi may elevate mycelial growth and thereby input of fungal necromass into the soil organic C pool. Despite being comparably chemically labile, fungal necromass makes a significant contribution to soil C stocks (Clemmensen et al., 2013) through stabilizing interactions with root tannins (Adamczyk et al., 2019). Root-associated ascomycetes, including ericoid mycorrhizal fungi and root endophytes, but also a variety of other fungi with uncertain ecology, have been highlighted as important contributors of C to the organic topsoil (Clemmensen et al., 2015) and their proliferation would be expected to increase organic matter accumulation. Previous fertilization studies in Scandinavian coniferous forests have observed declines in fungal biomass (Demoling et al., 2008; Högberg, Blasko, et al., 2014), or reductions in fungal biomass only at high N input (Blasko et al., 2013; Maaroufi et al., 2019). However, fertilization stimulated fungal biomass production in strongly N limited, arctic ecosystems (Clemmensen et al., 2006). The response of fungal biomass to fertilization may, thus, vary depending on context and the initial nutrient status at the site.

Being biotrophically dependent on products of photosynthesis, the abundance of root-associated fungi should reflect the above-ground tree biomass (Sterkenburg et al., 2019), and thinning would be expected to reduce C input by roots and associated fungi. However, knowledge about how fungal communities respond to thinning is scarce. In a Mediterranean forest system, fungal biomass decreased after thinning (Collado et al., 2020), whereas another study observed no effects on fungal community composition (Castaño et al., 2018). In a temperate pine stand, fungal biomass 5 years after thinning did not differ from non-thinned stands (Maassen et al., 2006). In boreal Pinus contorta forests, thinning alone did not affect root colonization rates or fungal community composition, but reduced the impact of N fertilization on the composition of the fungal community (Teste et al., 2012).

Soil C stocks may also be regulated by losses via decomposition. Additions of easily available N through fertilization can alleviate nutritional constraints on saprotrophic litter decomposers, and thereby increase decomposition rates, since higher N availability would increase fungal growth and associated C demand (Boberg et al., 2008). Stimulated decomposition should then increase microbial respiration and decrease C stocks after fertilization. However, field experiments commonly report decreased decomposition following N addition (Janssens et al., 2010). This contradiction may relate to different responses of hydrolytic and oxidative mechanisms of decomposition. In temperate broad-leaf forests, N additions, even at low rates, increased activities of enzymes involved in the hydrolysis of cellulose but decreased activities of oxidative, ligninolytic enzymes (Carreiro et al., 2000; Sinsabaugh et al., 2002). This shift was attributed to a loss of ligninolytic saprotrophs, which have a high potential to oxidize non-hydrolysable and presumably recalcitrant organic matter (Entwistle et al., 2017). Similar shifts may also occur in boreal forests, where additions of N seem to hamper oxidative decomposition of organic matter (Bödeker et al., 2014) and favour cellulolytic fungi relative to ligninolytic species (Talbot & Treseder, 2012). Carbon stocks in the organic topsoil of boreal forests have been found to correlate negatively with the activity of fungal manganese peroxidases (Kyaschenko et al., 2017b) as well as with manganese availability (Stendahl et al., 2017). Organic topsoils consist to a great extent of decomposed and chemically complex organic matter that may primarily serve as a source of N rather than C for decomposers. It is, thus, possible that ‘N-miners’ with oxidative capacity are disfavoured when N-limitation is alleviated by fertilization, and this may, indirectly, hamper C turnover (Craine et al., 2007).

In boreal forests, however, potent saprotrophic decomposers are mainly confined to the uppermost litter layer, especially in nutrient-poor soils, while most of the organic (and mineral) soil is dominated by ectomycorrhizal fungi (Clemmensen et al., 2013; Kyaschenko et al., 2017b; Lindahl et al., 2007). Ectomycorrhizal fungi are considered to be sensitive to N additions (Bahr et al., 2013; Högberg, Yarwood, et al., 2014; Nilsson & Wallander, 2003), possibly due to downregulated C allocation from trees to mycorrhizal symbiosis when inorganic N becomes easily available (Franklin et al., 2014; Högberg et al., 2010; Lilleskov et al., 2019). During their evolution from saprotrophic ancestors, some ectomycorrhizal genera have retained mechanisms for oxidative decomposition and may, thus, be considered as symbiotic N-miners with a capacity to access nutrients in persistent, root-derived organic compounds (Bödeker et al., 2014; Lindahl & Tunlid, 2015; Sterkenburg et al., 2018). By releasing organically bound nutrients and share them with their host trees, such ‘ectomycorrhizal decomposers’ may counteract nutrient retention and maintain tree growth, despite low rates of nutrient mineralization (Baskaran et al., 2017). Notably, species of the genus Cortinarius seem to be central ectomycorrhizal decomposers that, through their production of manganese peroxidases, appear to be involved in the regulation of C stocks in organic topsoils of boreal forests (Bödeker et al., 2014; Lindahl et al., 2021). The genus Cortinarius is considered to be particularly sensitive to N additions (Arnolds, 1991; Lilleskov et al., 2019; Lindahl et al., 2021; Strengbom et al., 2001; van der Linde et al., 2018). While an overall decline in the production of ectomycorrhizal fungal mycelium may reduce below-ground C input, a specific loss of these potent oxidizers may also contribute to increased C stocks in the organic topsoil following N additions.

Effects of phosphorus (P) fertilization on fungal communities and nutrient cycling in boreal forests are not well understood, but alleviated N limitation under elevated N input may shift systems towards P deficiency, with subsequent effects on the structure and functioning of the fungal community. In a Picea abies forest in southern Sweden, subjected to moderate levels of N deposition, both P and N + P fertilization shifted the composition of the fungal community and both fungi and trees showed signs of P limitation (Almeida et al., 2019). In addition, increased levels of extracellular acid phosphatases after N addition have been observed in temperate coniferous forests (Forstner et al., 2019).

We investigated the interactive effects of fertilization and thinning on C sequestration in both standing biomass and the organic topsoil in Pinus sylvestris forests planted along a latitudinal transect across Sweden and found that N fertilization had a positive effect on C storage both above- and below-ground (Jörgensen et al., 2021a). Carbon sequestration in the organic topsoil was stimulated even further when P was added in combination with N, despite that P addition did not affect tree growth above-ground. The positive effect of N fertilization on C storage was greater at higher latitudes, presumably due to stronger N limitation. Interestingly, reduction of tree biomass and associated litter input by thinning decreased organic topsoil C stocks, but only in N fertilized stands, suggesting that N fertilization increased the importance of litter input relative to decomposition in the regulation of soil C stocks. We also observed declines in soil respiration after N fertilization. Here, we address whether these previously described responses were related to treatment-induced effects on the soil fungal community. We hypothesized that increased C stock following fertilization would be linked to (1a) higher mycelial biomass, an effect that (1b) would be reduced by thinning and/or (2a) decreased abundance of ectomycorrhizal fungi, particularly Cortinarius species and (2b) decreased activity of decomposer enzymes, especially fungal peroxidases. Hypothesis (1) assumes that increased C stock is an effect of increased input of persistent litter (here mycelial necromass), whereas hypothesis (2) assumes that increased C stock primarily is due to downregulated decomposition linked to reduced ectomycorrhizal N-mining.

2 MATERIALS AND METHODS 2.1 Experimental setup

We used a long-term thinning and fertilization experiment at 29 sites distributed across a 1,300 km latitudinal transect (56–67°N) in boreal Scots pine Pinus sylvestris L. dominated forests in Sweden (median Pinus sylvestris proportion of stem volume: 100%, min: 84%). The forests were established by direct seeding or planting, and the experiment was established between 1969 and 1982 when the trees were between 32 and 54 years old. At each site, four different treatments were applied on 0.1 ha plots: ‘No thinning’, ‘Thinning’, ‘Thinning + N’ and ‘Thinning + NP’. At nine of the sites, an additional treatment of ‘No thinning + N’ was implemented. The understorey vegetation depended on the treatments but was dominated by mosses such as Pleurozium schreberi and Hylocomium splendens, and low herbs such as Vaccinium vitis-idaea and Vaccinium myrtillus. No permissions were needed for the field work. For more information about vegetation, see Strengbom et al. (2018). All soils were podzols with distinct organic horizons ranging from 2.5 to 11.7 cm (average: 5.9) in thickness, with C:N ratios of 25–63 (average: 41). The pH ranged from 3.3 to 4.6 with an average of 3.7 was slightly higher in the ‘Thinning + NP’ treatment, and increased with latitude in all treatments (Figure S1; Table S1).

The thinned plots were thinned one to four times during 1974–2015 to maintain a constant basal area of around 18 m2/ha, corresponding to a 20%–25% reduction compared to un-thinned stands. At each thinning, stems were removed while branches and stumps were left in the plot. In the un-thinned stands, trees that had died on site were left. Nitrogen was added as ammonium nitrate (NH4NO3) to fertilized plots at a rate of 100–150 kg/ha every 5th year for the first 25 years and thereafter every 7th year. Phosphorus in the form of superphosphate (CaSO4 + Ca(H2PO4)3) was added at a rate of 100 kg/ha at the start of the experiment and thereafter every 20–21 years. More information about nutrient loads and background N deposition levels are presented in Table S2. The experimental setup is described in more detail in Nilsson et al. (2010).

2.2 Soil sampling, processing and pH

The study used the samples described in Jörgensen et al. (2021a). In the summer of 2016, the organic topsoil was sampled in all 125 plots by pooling 25 soil cores (3 cm in diameter) taken in a grid pattern across each 0.1 ha plot. Soil samples extended down to the organic–mineral soil transition. Living mosses, grasses and roots coarser than 2 mm were removed before pooling. The samples were frozen within 48 hr of sampling and kept at −20°C until analysis. Frozen samples were homogenized, and subsamples were freeze-dried, and ball-milled. Organic topsoil pH was measured on a PHM 93 Reference pH meter (Radiometer, Copenhagen) in a 1:3 weight ratio of fresh soil and H2O.

2.3 Ergosterol extractions and enzyme assays

Fungal biomass was estimated by ergosterol extraction (Nylund & Wallander, 1992) from 0.2 g freeze-dried, finely ground soil. Ergosterol was extracted in 10% KOH dissolved in methanol and transferred to a cyclohexane phase. The cyclohexane was evaporated under N2 gas flow and the ergosterol was resuspended in methanol, after which the extracts were filtered through a 0.45 µm Teflon syringe filter (Millex LCR-4; Millipore) before analysed using high-performance liquid chromatography (HPLC) according to the protocol of Hagenbo et al. (2017). Activities of the hydrolytic enzymes β-1,4-glucosidase, β-1,4-N-acetylglucosaminidase and acid phosphatase were assessed using fluorogenic methylumbelliferyl substrates (Saiya-Cork et al., 2002) in dilute soil suspensions at a concentration of 1 g (fresh weight) soil/L (Kyaschenko et al., 2017a). β-1,4-glucosidases release glucose from oligosaccharides, and play a central function in cellulose decomposition. β-1,4-N-acetylglucosaminidase plays a corresponding role in decomposition of chitin, and is thought to be important in mobilization of organically bound N in boreal forest soils. Acid phosphatases release organically bound phosphate from monoesters (Sinsabaugh et al., 2008). Fluorescence was measured on a luminescence/fluorescence spectrophotometer (LS 50B, PerkinElmer Inc.). Manganese peroxidase (MnP) activity was assayed on soil extracts of 10 g (fresh weight) soil/L using the colorimetric MBTH-DMAB method (Daniel et al., 1994) as described in Kyaschenko et al. (2017b). Absorbance was measured on a SpectraMax Plus 384 microplate reader (Molecular Devices).

2.4 DNA extraction and characterization of the fungal community

DNA was extracted from 2 g of freeze-dried soil with 10 ml 3% CTAB (Clemmensen et al., 2016), precipitated with isopropanol, resuspended in 700 µl SL1 buffer from the NucleoSpin Soil kit (Macherey-Nagel) and cleaned using the kit according to the manufacturer's instructions. PCR amplicons of the ITS2 region were produced using the forward primer gITS2 (Ihrmark et al., 2012) and a 3:1 mixture of the reverse primers ITS4 (White et al., 1990) and ITS4arch (Kyaschenko et al., 2017a) with unique identification tags attached to both primers. 12.5 ng of DNA template was used in the 50 µl reaction (Clemmensen et al., 2016), which was run for 27–31 cycles with Dreamtaq polymerase (Thermo Fischer Scientific) and an annealing temperature of 56°C. Amplicons were cleaned with AMPure (Beckman Coulter Life Sciences) according to the manufacturer's instructions before DNA concentrations were measured on a Qubit fluorimeter (Life Technologies). Amplicons were then pooled into an equimolar mix and cleaned with the E.Z.N.A. Cycle Pure Kit (Omega Bio-Tek). Library preparation and sequencing was conducted by SciLifeLab (NGI) on the PacBio Sequel platform (Pacific Biosciences). The PacBio platform was chosen to minimize biases due to ITS2 length variation (Castaño et al., 2020). Raw sequences were filtered in the bioinformatics pipeline SCATA (https://scata.mykopat.slu.se/, Ihrmark et al., 2012), accepting sequences with length >100 bp, mean quality >20, single base quality >3 and primer sequence similarity >90%. After removal of global singletons, sequences were compared pairwise with USEARCH (Edgar, 2010), and clustered into species hypotheses (Kõljalg et al., 2013; hereafter ‘species’) using single-linkage clustering with a 98.5% similarity threshold. Clusters of plant sequences were removed from further analyses.

The 209 species that accounted for >1% of the total number of sequences in any sample (together representing 85% of the total fungal sequences) were identified by comparisons to the UNITE database (Nilsson et al., 2019) and assigned to ecological guilds according to Clemmensen et al. (2015): root-associated ascomycetes (the genera Archaeorhizomyces, Chaetothyriales, Chloridium, Geomyces, Hyaloscypha, Mollisia, Oidiodendron and Phialocephala; including ericoid mycorrhizal but not ectomycorrhizal fungi), saprotrophic agaricomycetes and ectomycorrhizal fungi (Appendix S1).

2.5 Statistical analyses

Treatment effects on the 209 identified fungal species were evaluated with canonical correspondence analysis (CCA) in CANOCO 5.12 with the four treatments applied at all sites as factorial explanatory variables. The ‘No thinning + N’ treatment (which was implemented at nine sites only), as well as one site with incomplete data, was removed from the dataset before the analysis. Statistical significance of treatments was evaluated by a Monte Carlo test with 999 permutations restricted to within sites only. Fungal community composition was represented graphically using detrended correspondence analysis (DCA).

For the remaining statistical analyses, thinning, N fertilization, P fertilization and latitude were included as explanatory factors in linear mixed-effect models, with site as a random factor using the lmerTest (Kuznetsova et al., 2017) and lme4 (Bates et al., 2015) packages in r version 4.0.3 (R Core Team, 2018). The interaction between N fertilization and thinning, as well as N and latitude, was included initially but removed from the models if not significant (p < 0.05). As the amount of N and P fertilization varied slightly between sites, loads were represented as continuous variables and normalized by the average loads, giving ranges between 0–1.4 for N and 0–1.3 for P. The latitude of the most southern plot was used as a zero reference.

Treatment effects on relative abundance of guilds and were assessed using linear mixed-effects models as described above. Square-root transformation of relative abundances of fungal guilds was necessary to achieve normal distribution of residuals. Within the ectomycorrhizal guild, collective abundances of the three most species-rich genera: Cortinarius, Piloderma and Russula were evaluated relative to the total fungal community and to the ectomycorrhizal community.

A core fungal community was defined as the 40 species that occurred at >1% relative abundance on at least 10 sites (representing 65% of the total number of fungal sequences; Table S3). Treatment effects on the square-root transformed relative abundance of each species in the core fungal community were evaluated as described above, and p-values were adjusted for multiple testing using the correction of Benjamini and Hochberg (1995; Appendix S2).

Treatment effects on ergosterol (mg/m2 and µg/g OM) and enzymatic activities were evaluated in the same way. Manganese peroxidase data were square-root transformed to meet assumptions of normal distribution of residuals. To evaluate correlations with organic topsoil C stocks across treatments, we fitted linear models with either standing tree biomass, ergosterol, relative abundances of different fungal guilds or the ectomycorrhizal genus Cortinarius as explanatory factors and site as a random factor. Data on soil C stocks and tree biomass were collected from Jörgensen et al. (2021a).

3 RESULTS

Mycelial biomass, estimated by ergosterol concentration in the organic topsoil, increased towards the north. Fertilization increased fungal biomass, with a stronger effect at high latitudes. We detected an interaction between N fertilization and thinning with a negative effect of thinning on fungal biomass, but only in N fertilized plots (Figure 1a; Table S4).

Ergosterol concentrations and relative abundances of fungal guilds in the organic topsoil of pine forest plots subjected to thinning and/or fertilization across a latitudinal transect in Sweden. (a) Ergosterol concentration (µg/g OM) in the organic topsoil, relative abundance (% of fungal sequences based on PacBio Sequel sequencing of ITS2) of (b) root-associated ascomycetes, (c) saprotrophic agaricomycetes and (d) ectomycorrhizal fungi. Lines are fitted regression lines from linear models. See Table S4 for statistics

Out of 856,192 sequences, a total number of 575,734 sequences passed quality filtering, and after removal of unique genotypes, 293,068 sequences were clustered into 2,362 global species, whereof the 209 most abundant were identified. N-fertilization changed the composition of the fungal community (conditional effect: p = 0.001), and application of P together with N had an additional effect (conditional effect: p = 0.022; Figure 2). Thinning had no significant effect on fungal community composition. The relative abundance of root-associated ascomycetes was higher in fertilized plots (Figure 1b; Table S4), whereas ectomycorrhizal fungi decreased after fertilization (Figure 1d; Table S4). The relative abundance of the ectomycorrhizal genera Cortinarius and Russula decreased in response to fertilization, both in relation to the total fungal community (Figure 3a,c; Table S5) and in relation to other ectomycorrhizal fungi (Figure 3d,f; Table S5). The genus Piloderma was also negatively affected by N fertilization (Figure 3b) relative to the total fungal community but increased its share of the ectomycorrhizal community (Figure 3f; Table S5). Abundance of saprotrophic agaricomycetes did not differ among treatments, but were generally higher towards the north (Figure 1c; Table S4).

Sample plot of a detrended correspondence analysis (DCA) based on the composition of the fungal community in the organic topsoil of pine forests plots subjected to thinning and/or fertilization across a latitudinal transect in Sweden. The analysis is based on PacBio Sequel sequencing of ITS2 amplicons, and encompasses 209 species that accounted for ≥1% of the fungal sequences in any sample. Latitude was included as a supplementary variable in the analysis. DCA axis 1 explains 12.1% and axis 2 explains 5.2% of the total variation

Relative abundances of ectomycorrhizal fungal genera in the organic topsoil of pine forest plots subjected to thinning and/or fertilization across a latitudinal transect in Sweden. Relative abundance (% or ITS2 sequences) of (a) Cortinarius, (b) Piloderma and (c) Russula in the total fungal community and (d), (e) and (f) among all ectomycorrhizal fungi (% of ectomycorrhizal (ECM) ITS2 sequences). Lines are fitted regression lines from linear models. See Table S5 for statistics

Of the core fungal community, 12 species declined in relative abundance after N fertilization, including the ectomycorrhizal Cortinarius acutus, Russula decolorans and Suillus variegatus, one ericoid mycorrhizal species in the order Chaetothyriales and the ericoid mycorrhizal fungus Hyaloscypha hepaticicola. The latter also increased its abundance in response to thinning. Phosphorus fertilization in combination with N had a negative impact on an unidentified Leotiomycete. Two Archaeorhizomyces species were both positively affected by N (Figure 4; Table S6).

Relative abundance of species in the fungal community of organic horizon of pine plots subjected to different combinations of thinning and fertilization. Relative abundance (% of fungal reads from PacBio Sequel sequencing of ITS2 amplicons) of 14 species with a statistically significant response (p ≤ 0.05) to any of the treatments. Abbreviations in parentheses denote functional guild: ECM = ectomycorrhiza, ERM = ericoid mycorrhiza, RA = root-associated ascomycete, no letters = unknown guild. Bars are mean abundance and error bars denote SE. See Table S6 for statistics

We did not detect any significant treatment effect on manganese peroxidase activity (Figure 5a; Table S7). β-1,4-glucosidase activity declined towards the north in non-fertilized stands, but a positive interaction between N fertilization and latitude counteracted the latitudinal decline in fertilized plots (Figure 5b; Table S7). β-1,4-N-acetylglucosaminidase activity was unaffected by fertilization but positively affected by thinning (Figure 5c; Table S7). Acid phosphatase activity was higher in thinned stands and in N-fertilized stands but lower in the combined N and P fertilization treatment (Figure 5d; Table S7).

Enzyme activities in the organic topsoil of pine forest plots subjected to different combinations of thinning and fertilization along a latitudinal transect in Sweden. Enzyme activities of (a) manganese peroxidase, (b) β-1,4-glucosidase, (c) β-1,4-N-acetylglucosaminidase and (d) acid phosphatase. Lines are fitted regression lines from linear models. See Table S7 for statistics

We tested the links between organic topsoil C stocks and potential fungal regulatory factors to elucidate the strongest correlations across plots and treatments. Ergosterol concentration (Figure 6a) and β-1,4-glucosidase activity (Figure S2) were positively correlated with organic topsoil C stocks, whereas Cortinarius relative abundance was negatively correlated with C stocks (Figure 6f).

Correlations between organic topsoil C stocks and potential regulatory factors across pine forests plots subjected to different combinations of thinning and fertilization along a latitudinal transect in Sweden. Partial residuals, after accounting for site-specific random effects, of C in the organic topsoil (g/m2) plotted against (a) standing biomass of trees (tonnes), (b) ergosterol (µg/g OM), or square root transformed relative abundances () of: (c) root-associated ascomycetes, (d) saprotrophic agaricomycetes, (e) ectomycorrhizal fungi and (f) Cortinarius species. Significant effects (p ≤ 0.05), evaluated by linear models, are highlighted in bold text in the figure. N = 124 4 DISCUSSION

We found that fertilization changed the composition of root-associated fungal communities, both at the guild and species level (Figures 1 and 4). Interestingly, community change was particularly accentuated when P was added together with N (Figure 2). Changes in root-associated fungal communities after fertilization are thought to depend largely on reduced allocation of tree C below-ground (Högberg et al., 2003). In boreal forests, plant N and P are primarily derived from soil organic stocks via decomposition driven by ectomycorrhizal fungi. This mycorrhizal mining for nutrients is driven by allocation of plant photosynthates to fungal symbionts in response to nutrient shortage (Lindahl & Tunlid, 2015). Phosphorus limitation and a sustained need to mobilize P from organic sources may maintain plant C allocation to fungi also in N fertilized forests (Nylund & Wallander, 1992; Wallander et al., 2011), whereas combined supply of both resources would be expected to lead to more dramatic changes in the symbiosis (Suz et al., 2021).

In this study, we particularly investigated potential fungal mechanisms that may explain previously reported changes in organic topsoil C stocks following 40 years of fertilization and thinning in Swedish Pinus sylvestris forests (Jörgensen et al., 2021a). Taken together, we found that increased soil C storage after N fertilization correlated with both increased fungal biomass and losses of mycorrhizal fungal taxa that have previously been identified as important decomposers in boreal forest (Lindahl et al., 2021). However, we found no support for the hypothesis that this fertilization-induced effect was associated with a direct decrease in manganese peroxidase activity after fertilization.

Increased N availability following fertilization may ameliorate N-limitation of trees and associated fungi, increasing their C use efficiency and growth. Indeed, concordant with hypothesis (1a), ergosterol concentration in the organic topsoil was elevated in fertilized plots, with a stronger effect towards high latitudes (Figure 1a; Table S4). The stronger response to fertilization in the north suggests intensified nutrient limitation of the fungal community in harsher climates, potentially in combination with higher rates of N deposition towards the south (Karlsson et al., 2019) that render forests less responsive to additional input of N. Our observation stands in contrast to previous studies in Scandinavian coniferous forests, which observed no, or negative, responses of fungal biomass to fertilization (Blasko et al., 2013; Demoling et al., 2008; Högberg, Blasko, et al., 2014; Maaroufi et al., 2019). It is possible that the Pinus sylvestris forests in this study were characterized by particularly low nutrient availability, and thus responded in a similar manner as the arctic tundra systems presented in Clemmensen et al. (2006). Higher mycelial biomass without increased soil respiration (even decreased respiration in non-thinned, fertilized stands; Jörgensen et al., 2021a) supports increased C use efficiency of fungi and more rapid mycelial growth after fertilization (Manzoni et al., 2018). However, the observed increases in standing fungal biomass may also be due to slower turnover rates of fungal mycelium instead of increased production (Hagenbo et al., 2017). Slow biomass turnover may also contribute to lower soil respiration by increasing apparent C use efficiency at the community level (Manzoni et al., 2018). We also found that root-associated ascomycetes proliferated in fertilized plots (Figure 1b), probably related to higher root biomass (Forsmark et al., 2021). In particular, two Archaeorhizomyces species responded positively to fertilization (Figure 4). One of these was highly abundant, on average accounting for 14% of the fungal sequences, and thus likely to have been largely responsible for the increase in fungal biomass after fertilization. Very little is known about the ecology of this only recently discovered genus (Rosling et al., 2011), but we postulate that Archaeorhizomyces species associate with roots and increase their C use efficiency and growth at high N availability. In addition, in line with hypothesis (1b), we observed a negative effect of thinning on fungal biomass, but only in fertilized stands. This supports a positive link between mycelial production and tree biomass in N fertilized systems and a plausible transition from N limitation to C limitation of the fungal collective after fertilization.

To evaluate tentative mechanisms explaining the effect on C stocks, we studied overarching correlations between C stocks and potential regulators across treatments. Fungal biomass correlated positively with organic topsoil C stocks (Figure 6b), and it seems plausible to attribute part of the treatment-induced gain in C stock to higher input of fungal biomass, according to hypotheses (1a and b). In particular, Archaeorhizomycetes may have contributed to C accumulation below-ground, in line with the idea that root-associated ascomycetes are an important source of persistent organic material (Clemmensen et al., 2013, 2015, 2021). Together with associated mycelium, roots may contribute to formation of stable organic matter due to their high content of tannins (Adamczyk et al., 2019), and a recent study in boreal forest suggested that roots contribute to increased C stocks after fertilization by increased C use efficiency and production (Forsmark et al., 2021). We do not have any data on fine-root productivity in our sites, but it is possible that increased root production also contributed the increased C stocks in our study.

Decline of saprotrophic agaricomycetes with ligninolytic potentials has been highlighted as a mechanistic link between fertilization and hampered decomposition after fertilization in North American temperate forests (Entwistle et al., 2017). However, in our forests, saprotrophic agaricomycetes were not affected by fertilization or thinning (Figure 1c), nor did they correlate with C stocks (Figure 6d). Instead, we found that the relative abundance of ectomycorrhizal fungi responded negatively to fertilization, with Cortinarius species being particularly sensitive (Figure 3a,d), in agreement with hypothesis (2a) and previous studies (Hasselquist & Högberg, 2014; Högberg et al., 2017; Lilleskov et al., 2019; van der Linde et al., 2018; Wallander et al., 2011). The genus Piloderma, which lack ligninolytic capacity (Kohler et al., 2015), was also negatively affected by fertilization, but less so than Cortinarius, leading to increased relative dominance of Piloderma in the ectomycorrhizal community after fertilization. The abundance of ectomycorrhizal Cortinarius species correlated significantly and negatively with C stocks (Figure 6f). One species, Cortinarius acutus, responded particularly strongly (Figure 4), and was specifically found to correlate negatively with soil C storage in a field survey of Swedish forests (Lindahl et al., 2021). Suillus variegatus, which is likely to use the Fenton reaction to oxidize organic matter (Shah et al., 2016), also responded negatively to fertilization. Supposedly, these ectomycorrhizal decomposers take over the role of saprotrophic agaricomycetes under the strongly N-limited conditions that prevail in boreal forests (Lindahl et al., 2021), but their N-mining activity becomes obsolete when inorganic N is provided through fertilization (Bödeker et al., 2014). The importance of N-mining by ectomycorrhizal decomposers was recently supported by a global meta-analysis, which found that stimulated tree growth at elevated CO2 occurred at the expense of below-ground organic stocks, and primarily in forests with ectomycorrhizal hosts (Terrer et al., 2021). Similar to our study, the effect was nullified by fertilization.

We did, however, not find a significant decline in manganese peroxidase activity after fertilization, and therefore little direct support for hypothesis (2b), although there was a trend towards decreased activity when thinning and fertilization were combined (Figure 5a). Previous studies have found correlations between fungal community composition and manganese peroxidase activity (Bödeker et al., 2014; Entwistle et al., 2017; Kyaschenko et al., 2017a), and linked this to changes in soil C stocks (Kyaschenko et al., 2017b; Stendahl et al.,