Fungal colonization patterns and enzymatic activities of peatland ericaceous plants following long-term nutrient addition

Abstract Northern peatlands are often dominated by ericaceous shrub species which rely on ericoid mycorrhizal fungi (ERM) for access to organic sources of nutrients, such as nitrogen (N) and phosphorus (P), and host abundant dark septate endophytes (DSE). Relationships between hosts and fungal symbionts may change during deposition of anthropogenic N and P. We studied the long-term effects of N and P addition on two ericaceous shrubs, Calluna vulgaris and Erica tetralix, at Whim Bog, Scotland by analyzing fungal colonization of roots, enzymatic activity, and fungal species composition. Unexpectedly, the frequency of typical ERM intracellular colonization did not change while the occurrence of ERM hyphae tended to increase and DSE hyphae to decrease. Our findings indicate that altered nutrient limitations shift root associated fungal colonization patterns as well as affect ericaceous root enzyme activity and thereby decomposition potential. Reduction of recalcitrant fungal biomass in melanized DSE may have implications for peatland C sequestration under nutrient addition.


Introduction
Peatlands in the Northern hemisphere are often nutrient poor ecosystems characterized by acidic, anoxic, water saturated conditions with considerable nitrogen (N) and phosphorus (P) limitations (Aerts et al., 2001). These conditions and the accumulation of recalcitrant vegetation litter containing high concentrations of phenolic compounds and humic acids inhibitory to microorganisms and vegetation are considered to largely suppress decomposition (Leake and Read, 1990;Painter, 1991;Read et al., 2004;van Breemen, 1995). The challenging conditions in peatlands support a unique diversity of vegetation, with ericaceous species comprising one of the most dominant ground cover groups. Ericaceous shrubs are largely dependent on ericoid mycorrhizal fungi (ERM) to provide access to organic N and P which they provide in exchange for photosynthetic carbon (C) from the host plant (Smith and Read, 2008). The ERM fungi are capable of accessing organic N and P via a large variety of degradative enzymes which act primarily on plant cell wall components (Perotto et al., 2018), demonstrating a potential versatility more comparable to saprotrophs than to other types of mycorrhizae. These ericaceous species are also host to abundant dark septate endophytes (DSE) with extracellular enzyme capabilities potentially capable of improving host nutrient uptake Jumpponen, 2005, 2014;Upson et al., 2009).
Over the past 150 years atmospheric deposition of N and P in forms easily accessible to plants has been increasing through combustion of fossil fuels and agricultural fertilization (Galloway et al., 2003(Galloway et al., , 2013Tipping et al., 2014;Wang et al., 2015). As nutrient limitations are alleviated, ericaceous reliance on ERM fungi may be reduced, potentially altering the symbiont community and leading to the loss of mycorrhizal symbionts. As mycorrhizal symbionts likely play a role in protection against pathogens (Vohník et al., 2016;Weiβ et al., 2016), host species may in turn become more vulnerable to infection. Furthermore, the long-term effects of an altered nutrient balance in peatlands may include reduced nutrient acquisition competitiveness for ericaceous species. The ERM fungi are not as efficient decomposers as free-living saprotrophs, which may be naturally kept in check by nutrient limitations and direct competition with ERM fungi (Averill et al., 2014). Reduction of nutrient limitations may free the saprotrophic species' decomposition potential, leading to their dominance and the decline of ERM species and their host plants.
The selective pressure of N deposition on the symbiont community highlights a risk to peatland plant diversity, potentially leading to similar large-scale community shifts as seen in the continental level decline of ectomycorrhizal tree species and the increase in arbuscular tree species described by Averill et al. (2018). Any large-scale changes to peatland microbial and plant communities risk changing the status of peatlands as net C sinks to net sources of greenhouse gas emissions (Andersen et al., 2013), which holds a globally significant potential when considering that peatlands sequester nearly one third of global soil organic carbon C (Gorham, 1991). The risk to C sink potential has been indicated by Larmola et al. (2013), who found that ecosystem C uptake did not increase in a long-term nutrient addition experiment at a nutrient-poor peatland in Canada simulating atmospheric N deposition.
The Whim Bog experimental site, located in the Scottish Borders, was established in 2001 to study the effects of different N forms and P addition on an ombrotrophic peat bog (Sheppard et al., 2004). This site allowed us to study nutrient addition effects on the ericaceous species Calluna vulgaris and Erica tetralix and their root associated fungi. The primary goals of this study were to characterize the frequency and morphology of mycorrhizal colonization in these ericaceous shrub species, assess ericoid mycorrhizal root enzyme capability related to organic matter degradation, and identify their fungal symbionts under changing nutrient availability. We hypothesized that: (1) the frequency of microscopically observed fungal colonization in ericaceous shrub roots is reduced across both forms of N and NP nutrient addition treatments, reflecting a reduction in reliance on symbionts for nutrient uptake; (2) nutrient addition treatments alter root associated fungal diversity, as determined by morphotypic analysis and ITS sequencing; and (3) the activities of ericoid mycorrhizal root surface enzymes reflect treatment nutrient limitations.

Study site
The study site, Whim Bog, located in the Scottish Borders, UK (Latitude: 55.76670, Longitude: À 3.26667), has undergone nutrient addition treatment since 2001. The four nutrient addition treatments included in this study received annually 6.4 g N m À 2 either as sodium nitrate (NaNO 3 ) or as ammonium chloride (NH 4 Cl), both with and without P and K (as K 2 HPO 4 ). K 2 HPO 4 was added at a 1:14 P:N ratio to represent the ratio found in amino acids (See Sheppard et al. (2004) and Levy et al. (2019) for details). The ambient deposition in controls was 0.8 g N m À 2 . Precipitation collected at the site was mixed with standard solutions to the required treatment concentrations. When adequate precipitation was collected an automated sprayer-system applied the treatment to the plots, simulating natural rainfall. Natural precipitation was not excluded from plots. Plots received 15 years of nutrient addition resulting in a 96 g N m À 2 cumulative load.

Site measurements and sampling
In August 2016, plant species composition and abundance for each plot were measured via the point-intercept method using a 0.36 m 2 frame on permanent vegetation quadrats established on site. The frame was placed at a height of ca. 1 m relative to the surface of the plot and a graduated pin was used to measure the frame height and vertical location of each vegetation point for 61 intercepts as described in Larmola et al. (2013). Water table (WT) depth for each plot was measured from holes present after extraction of ingrowth cores used in a separate study, relative to moss surface height.
Triplicate plots per treatment and controls were sampled in November 2016 for Calluna vulgaris (L.) Hull and Erica tetralix (L.) plant roots from one individual plant per plot. Fine root sections were collected from several points throughout the root system of each sampled plant and stored at 8 � C prior to transport, followed by storage at À 20 � C. Each root sample was split into two subsamples: one for microscopy and the other for enzymatic and subsequent molecular analyses.
Surface peat from triplicate plots per treatment and controls was sampled to a depth of 20 cm in each plot and stored at 8 � C prior to transport, followed by storage at À 20 � C. After melting, the pH of each sample was measured following homogenization with deionized water at a 1:4 ratio.

Microscopy
The mycorrhizal status of the ericaceous shrub species C. vulgaris and E. tetralix was determined via light microscopy and Trypan Blue staining as described in Kiheri et al. (2017). Using the magnified intersections method described by Mcgonigle et al. (1990), with one slide per species from each of three replicate plots per treatment and controls (15 samples per species), using 300 counts per slide, mycorrhizal colonization was quantified according to different morphological categories, described in Table 1. These categories were used to estimate differences between the frequencies of potential ericoid mycorrhizal (ERM) hyphae and typical dark septate endophyte (DSE) hyphae. In our study, only those fungi which were robust in structure, heavily melanized and displaying septa were assessed to be DSE. This was necessary to prevent misclassification of those ERM species which produce melanized and septate hyphae (Vohník and Albrechtov� a, 2011).
Hereon, putative ericoid mycorrhizal morphotypes are referred to with categories ERM1-4 and dark septate endophytes with DSE1-4 (Table 1). Of the morphotypic categories, only ERM category 1 (intracellular coiling) was interpreted as mycorrhizal colonization frequency. Categories 2-4 were interpreted as potential changes in fungal diversity and function.

Enzyme assays
The enzymatic activities of root samples from C. vulgaris and E. tetralix were determined using a multi-enzyme assay described by Pritsch et al. (2011), originally performed for assessing ectomycorrhizal exo-enzyme potential. From triplicate plots of each of the five treatments, rhizomes of both ericaceous species were each sampled for 9 individual ca. 1 cm root pieces, for a total of 270 root samples (n ¼ 5 treatments x 3 plots x 2 species x 9 root samples). To measure eight different hydrolytic and oxidative root surface enzyme potentials we used the method described by Velmala et al. (2014)  Washington, NY, USA) in buffers containing enzyme specific 7-amino-4-methylcoumarine (AMC) or 4-methylumbelliferone (MU) substrates. Incubation times for each enzyme followed the protocol of Pritsch et al. (2011). The respective substrates used were Leucine-AMC, MU-xylopyranoside, MU-β-D-glucuronide, MU-cellobiohydrofuran, MU-N-acetyl-β-D-glucosaminide, MU-β-D-glucopyranoside, and MU-phosphate. After incubation, substrate solutions were collected by centrifugation with a 96-well plate adapter at 3200 rpm onto Optiplate-96F reading plates (PerkinElmer, Waltham, MA, USA) containing stop buffer (pH 10). Each substrate's fluorescence was measured using a Victor 3 1420 multilabel plate counter (PerkinElmer, Waltham, MA, USA) at an excitation wavelength of 355 nm and an emission wavelength of 460 nm. Standard solutions were prepared using aminomethylcoumarin (AMC) and 4-methylumbelliferone (MUF) and used to calculate enzymatic activities from concentrations of released AMC or MUF according to their respective substrates. All standard and enzyme substrate solutions were purchased from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany).
Following enzymatic measurements, the individual root pieces were scanned at high resolution (650 dpi) and their surface area measured using winRHIZO Pro (ver. 2017, Regent Instruments Inc.) software. These surface area values were used to convert enzymatic activity to pmol mm À 2 min À 1 .

Molecular methods and sequencing
From triplicate plots of each of the five treatments, rhizomes of both ericaceous species were each sampled for 9 individual ca. 1 cm root pieces, for a total of 270 root samples (n ¼ 5 treatments x 3 plots x 2 species x 9 root samples). Following enzyme assays and winRHIZO analysis on these samples, a randomized subset of 1 of 3 treatment replicates was taken from the 270 samples for sequencing. These 90 samples, plus four repeated samples, were then directly amplified using a Phire Plant Direct PCR kit's plant leaf protocol (Thermo Fisher Scientific, Waltham, MA, USA) and PCR using the ITS1F (Gardes and Bruns, 1993) and ITS4 (White et al., 1990) primer pair. Individual root pieces were manually crushed using sterile pestles in provided dilution buffer, in order to release fungal cells both on the root surface and within the root structure, and 1 μl of each mixture was then used as a template in a 20 μl PCR reaction. Cycling conditions for the Direct PCR were: initial denaturation 98 � C 5 min, 40 cycles (98 � C for 5 s, 57 � C for 5 s, 72 � C for 20 s), and final extension 72 � C for 1 min. The Direct PCR products were separated in a 2% agarose gel in 1X TAE buffer at 120V for 2 h and each ITS band excised and purified using a Nucelospin® Gel and PCR Clean-Up kit (Macherey-Nagel GmbH & Co. KG, Düren, Germany). The purified products were then further amplified using a DreamTaq PCR mastermix (Thermo Fisher Scientific, Waltham, MA, USA) by using 1 μl of Direct PCR product as the template in a 20 μl DreamTaq PCR reaction with the same ITS1F-ITS4 primer pair. The DreamTaq PCR was performed with the following program: initial denaturation 95 � C for 3 min, 35 cycles (95 � C for 30 s, 57 � C for 30 s, 72 � C for 1 min), and final extension 72 � C for 10 min. This was done as a quick method to identify common fungal species in each root sample which could then be linked to their associated root surface enzyme activities. These ITS products were Sanger sequenced using the ITS4 primer (Macrogen Europe, Amsterdam, NL). This produced ITS fragments ranging from 98 to 884 bp in length. This method resulted in the successful sequencing of 60% of (57 out of 94) ITS products (Supplementary File 1), of which 38 sequences were more than 200 bp in length and of high enough quality required for NCBI Gen Bank submission. The sequences are disposed to Gen Bank under the accession numbers MN059889-MN059927. See Supplementary File 1 for all 57 FASTA sequences. The high proportion of sequencing failure was likely due to mixed ITS products from multiple fungal species and inhibitory chemistry. Sequence identification was performed using the Unite massBLASTer analysis (Nilsson et al., 2019) on manually trimmed sequences. Analyzed sequences had 86-100% sequence similarity to existing reference or representative sequences within the INSD or Environmental databases and the most likely species hypotheses (SH) were selected as their fungal identities. Functional roles were then assigned to these fungal identities according to the web based FUNGuild bioinformatic tool (Nguyen et al., 2016).

Statistical analyses
Hereon, α level for statistical significance is defined as p � 0.05 and indicative as 0.05 < p � 0.1 in all cases, and the term significant specifically indicates statistical significance as p � 0.05. All analyses were calculated using values from treatment means (n ¼ 3). Differences between treatments, for each host plant separately, were determined by analysis of variance (ANOVA) on logarithmically transformed data followed by pairwise comparisons using the parametric Tukey's HSD and nonparametric Games-Howell post hoc tests, using IBM SPSS Statistics 25. Fungal-Enzyme activity profiles in Fig. 8 were prepared using Ori-ginPro 2018 by assigning fungal identifications to enzyme activities on an individual root piece basis.
The combined data for both host plants' mycorrhizal morphotype categories ERM1-4 and DSE1-4 (Table 1), root enzyme activities, plot level vegetation abundance, surface peat pH, and water table depth were analyzed using correlation analyses. Correlation analyses were performed using rcorr-function with Spearman rank based correlation from package Hmisc v4.1-1 (Harrell et al., 2014) and plotted using corrplot-function from package corrplot v0.84 (Wei and Simko, 2016) in the R programming environment (R Core Team, 2017).

Vegetation, peat pH, and water table
In all four nutrient addition treatments, the dominant ericaceous shrub, Calluna vulgaris, tended to decrease in abundance while Erica tetralix tended to increase, especially in NaNO 3 treatments (Table 2). Nutrient addition treatments showed decreasing trends in the abundance of Sphagnum, when compared with controls. The sedge Eriophorum vaginatum showed increasing trends in abundance in all treatments except NaNO 3 , when compared with controls. The high abundance of reported E. vaginatum in the NaNO 3 þPK treatment was largely an effect of one plot where the point-intercept measurements were performed within a large E. vaginatum tussock. Furthermore, the different forms of N addition were found to cause opposite changes in peat pH, with NaNO 3 increasing pH by ca. 0.2-0.3 units and NH 4 Cl decreasing pH by ca. 0.2-0.4 units ( Table 2). Effects of treatments on peat pH were found to be statistically significant (F (4,10) ¼ 6.410), with pairwise comparisons finding statistically significant differences between NaNO 3 and NH 4 ClþPK treatments and between NaNO 3 þPK and NH 4 ClþPK treatments. Mean water table (WT) depth, measured relative to moss surface, was eight cm closer to the moss surface in NaNO 3 þPK treatments than in control plots, while NH 4 Cl treatments affected WT the least. This was likely affected by the loss of moss abundance and subsequent subsidence in NaNO 3 þPK treatments.

Root associated fungal morphology and frequency
The mean proportion of root intersects which were microscopically quantified as presenting fungal colonization increased from 78% to 83% in control plots for C. vulgaris and E. tetralix roots, respectively, to 89% and 91% in nutrient addition plots, respectively. Of these root intersections which contained fungal structures, both ericaceous host species showed trends of increasing ERM hyphal frequency and decreasing DSE hyphal frequency when under nutrient addition (Fig. 1). Nutrient addition increased mean ERM hyphal frequency in C. vulgaris and E. tetralix roots by 31% and 16%, respectively. However, mean DSE hyphal frequency decreased by 30% in C. vulgaris roots and 22% in E. tetralix roots. Though these trends are considerable, they were not found to be statistically significant due to high natural variation between samples.

Root enzymatic activity
The different forms of N addition, NH 4 þ and NO 3 À , induced variable and opposing effects on C and N acquiring enzymes on the two ericaceous shrubs' root surfaces, when compared to controls (Figs. 6 and 7). Treatment with NaNO 3 tended to reduce all C. vulgaris root enzymatic activities except acid phosphatase, suggesting that C. vulgaris or its root associated fungi are more sensitive to NaNO 3 than E. tetralix, which did not show this effect. In C. vulgaris roots under NH 4 Cl addition, the activities of the C acquiring enzymes β-xylosidase and β-glucosidase did not change compared to controls, while activity of β-glucuronidase tended to increase two-fold and N-acetylglucosaminidase tended to decrease by nearly half. In contrast, the effects of both forms of N on E. tetralix roots were similar for all C and N acquiring enzymes, while leucine aminopeptidase activity tended to be suppressed under NH 4 Cl addition to less than 25% of controls and N-acetylglucosaminidase decreased by less than half, compared to controls. Laccase activity was not detected in any samples. Compared to control plots, addition of both forms of N alone tended to induce an approximately 25% increase in acid phosphatase activity in both plant species (Figs. 6 and 7). In contrast, treatments with additional PK reduced acid phosphatase activity to 1/3rd of controls in both ericaceous species (Fig. 7). In C. vulgaris roots, the effect of nutrient additions on acid phosphatase activity was statistically significant (F (4,10) ¼ 8.163), with significant differences between NaNO 3 and NaNO 3 þPK and NH 4 ClþPK treatments, as well as between NH 4 Cl and NaNO 3 þPK and NH 4 ClþPK treatments. In E. tetralix roots the effect of nutrient addition on acid phosphatase activity was significant (F (4,10) ¼ 11.400), with significant differences between control and NH 4 ClþPK treatments, between NaNO 3 and NaNO 3 þPK and NH 4 ClþPK treatments, and also between NH 4 Cl and NaNO 3 þPK and NH 4 ClþPK treatments.
Interestingly, in both ericaceous species the NPK treatments induced higher, although statistically non-significant, activities in many C and N acquiring enzymes compared to N alone (Figs. 6 and 7). The exceptions to this were the suppression of β-xylosidase and β-glucuronidase activities in C. vulgaris roots under NH 4 ClþPK addition. Additionally, there was a significant positive correlation between C. vulgaris ERM intracellular hyphal frequency (ERM2) and β-glucuronidase activity (r ¼ 0.66) and a statistically indicative positive correlation with β-glucosidase activity (r ¼ 0.50) (Fig. 4).

Ericaceous root enzyme activity profiles
When fungal sequence identifications were linked to the enzymatic activities of their respective root samples, potential species related enzyme activity patterns emerged (Table 3, Fig. 8). C. vulgaris root samples from NaNO 3 , NaNO 3 þPK, and NH 4 ClþPK treatments and hosting Serendipitaceae (Clade-B Sebacinales) or Serendipita sp. produced the highest detected leucine aminopeptidase activities. A C. vulgaris root sample in the NH 4 ClþPK treatment and hosting Hyaloscypha sp. was highly active in β-xylosidase, N-acetylglucosaminidase, and β-glucosidase, while samples hosting Hyaloscypha sp. in control and NH 4 Cl treatments were ca. 50-75% less active for the same enzymes. Additionally, a C. vulgaris root sample in the NH 4 Cl treatment simultaneously hosting Hyaloscypha hepaticicola and Ceratobasidium sp. was one of the most active samples, across all enzymes.
Interestingly, a root sample from C. vulgaris in the NaNO 3 þPK treatment was linked to a fungal identification of Pseudogymnoascus sp., which produced some of the highest activities for all enzymes in this treatment, except leucine aminopeptidase. A sample of C. vulgaris root from the NH4ClþPK treatment which hosted the species Acrodontium crateriforme indicated activities of cellobiohydrolase, N-acetylglucosaminidase, and β-glucosidase several times higher than any other samples measured. In the control treatment, C. vulgaris root samples hosting a member of the Rhytismatales showed the highest activity for leucine aminopeptidase and acid phosphatase while E. tetralix samples hosting Rhytismatales showed the highest activities for all enzymes in that treatment, except leucine aminopeptidase.

Discussion
The increases in overall fungal colonization for both Calluna vulgaris and Erica tetralix under nutrient addition were unexpected and the unchanged frequency of ERM intracellular coiling showed that long-term N and NPK addition did not reduce mycorrhizal colonization. This suggests that the ericaceous host plants are unable to restrict fungal colonization of their roots, despite access to excess inorganic N and P. Alternatively, unchanged mycorrhizal colonization rates may indicate that the fungi provide benefits to the plant beyond N and P.
The observed reduction of Sphagnum abundance and the decreased frequency of DSE associated with ericaceous roots may be linked. DSE fungi have been shown to be a common occurrence in submerged aquatic plants (Kohout et al., 2012) and are capable of propagating and existing as saprobes of moss gametophytes (Day and Currah, 2011). This capability and their prevalence in aquatic plants suggest that DSE fungi are well adapted to periodic waterlogged conditions in peatlands, perhaps providing a niche during the annual senescence of their ericaceous hosts. A similar relationship between ERM fungi and liverworts, common species in peatlands, may also provide a niche during host senescence (Kowal et al., 2018(Kowal et al., , 2015. As the loss of Sphagnum species is one of the most obvious effects of nutrient addition treatments (Bubier et al., 2007;Levy et al., 2019), this may remove an important ecosystem niche for DSE fungi. Our finding of positive correlation between E. tetralix root surface DSE colonization frequency and Sphagnum abundance supports this possibility.
The significantly increased frequencies of ERM hyphal morphotypes and decreased DSE hyphal morphotypes in ericaceous roots strongly suggests that long-term nutrient addition resulted in an altered fungal community. Furthermore, the different forms of N addition may have had selective effects on the fungal community as we identified different root associated Helotiales members only in controls, NH 4 Cl or NH 4 ClþPK treatments and members of the Pezizales only in NaNO 3 þPK treatments. The Helotiales may prefer NH 4 þ as a substrate compared to NO 3 À , as experimental evidence has shown for Hyaloscypha hepaticicola (Cairney et al., 2000). The presence of the Pezizales in only NaNO 3 þPK treatments may imply they prefer this N source, though their Fig. 8. Ericaceous root enzyme activities (Leu: leucine aminopeptidase, Glr: β-glucuronidase, Xyl: β-xylosidase, Cell: cellobiohydrolase, Gls: β-glucosidase, Nag: N-acetylglucosaminidase, Pho: acid phosphatase) for individual root pieces with identified fungal sequences, as well as the mean activities for root samples without sequencing (Black X Symbol, 20<n<25, per treatment). Error bars indicate �1 standard deviation. Not shown; Acrodontium crateriforme: Xyl ¼ 0.61, Nag ¼ 7.5, (C. vulgaris, NH 4 ClþPK). Pseudogymnoascus sp.: Cell ¼ 0.87, (C. vulgaris, NaNO 3 þPK). mycorrhizal status is unclear it has been suggested for some families by Hobbie et al. (2001). Sequence identification of ITS amplicons from root samples of both ericaceous species revealed several ERM and DSE species commonly found to associate with ericaceous roots, as well as common peatland saprotrophs and pathogens (Sieti€ o et al., 2018;Thormann, 2006;Thormann and Rice, 2007). Both ericaceous species shared several family and genus level groups, indicating a common symbiont community among the ericaceous hosts. This is in line with the findings of Kjøller et al. (2010) who showed that several ericaceous species in a subarctic mire shared fungal communities when in close proximity. While ERM symbionts are generally ascomycetes, recent studies have found that the basidiomycete fungi Serendipitaceae (Clade-B Sebacinales) are common in ericaceous roots and capable of forming mycorrhizal structures (Brundrett and Tedersoo, 2018;Vohník et al., 2016;Weiβ et al., 2016), and potentially capable of utilizing photosynthetic C (Sieti€ o et al., 2018). Our identifications of members of the Serendipitaceae and Serendipita sp. in both ericaceous hosts support their likely role as root symbionts. Interestingly, in C. vulgaris roots from the NaNO 3 þPK treatment we found a potential ERM symbiont, Pseudogymnoascus sp., which is a genus that may form ERM associations, as shown between Pseudogymnoascus roseus and Vaccinium angustifolium (Dalp� e, 1989).
Our findings on the varying effects of N addition on root surface enzymatic activities in the two ericaceous species, C. vulgaris and E. tetralix, indicate that they may have functionally different root symbionts and decomposition potentials. In NPK treatments both ericaceous plants displayed highly suppressed acid phosphatase activity as the roots and root associated fungi did not need to access organic P sources. Conversely, N treatments increased acid phosphatase activity, reflecting the colimitation of N and P found in peatlands (Pinsonneault et al., 2016;Wang et al., 2015) and confirming that ericoid mycorrhizal root enzymatic activities reflect nutrient limitations. Almost every enzyme activity increased with additional P compared to either form of N alone, while NaNO 3 addition was found to generally decrease enzyme activities for C. vulgaris. Additionally, both forms of N addition led to reductions in chitinase activity, which is similar to the findings of Bragazza et al. (2006), who suggested that this indicates an alleviation of N limitation.
In C. vulgaris roots β-xylosidase, β-glucuronidase, and β-glucosidase activities under NH 4 Cl addition were comparable to or higher than their activities in controls and NaNO 3 þPK treatments, suggesting that NH 4 Cl promotes overall decomposition activity, as these enzymes primarily degrade plant cell wall components (Dunn et al., 2014). This may reflect a reduction in N limitation for the C. vulgaris mycorrhizal symbiont identified as Hyaloscypha hepaticicola, as experimental in vitro data has shown that this species may preferentially utilize NH 4 þ as a source of N compared to organic sources (Cairney et al., 2000). Additionally, this is supported by the colonization morphotype data showing that NH 4 Cl addition induced the largest significant increase in intracellular ERM hyphae in C. vulgaris, implying increases in fungal biomass and therefore higher decomposition potential. Furthermore, the sequence identification of H. hepaticicola in a C. vulgaris root sample from the NH 4 Cl treatment was linked to relatively higher enzymatic activities compared to other root samples in the same treatment. Calluna vulgaris roots hosting Hyaloscypha sp. displayed higher enzymatic activity under nutrient addition than in control conditions, indicating a response to increased N or NPK availability. Our results demonstrate that it is necessary for studies of mycorrhizal fungi to include measurements of enzyme activities in natural conditions in order to more precisely estimate their roles in nutrient cycles. Though there is extensive research on the enzymatic activity of mycorrhizal fungi in sterile systems, few studies have measured the activity of mycorrhizal roots in their natural environment. While these in vitro enzyme activities of mycorrhizal fungi are often interpreted as their natural activities, work by Timonen and Sen (1998) showed that enzyme expression levels in Pinus sylvestris mycorrhizal fungi are locally regulated in the mycorrhizosphere, highlighting the variability in fungal enzyme expression which is not apparent from in vitro studies.
We observed that long-term nutrient addition resulted in a reduction in C. vulgaris abundance, potentially due to a reduction in competitive fitness, leaving an ecosystem gap that was rapidly occupied by other fast-growing species such as the non-mycorrhizal sedge Eriophorum vaginatum, which is not reliant on symbionts for organic N uptake (Chapin et al., 1993). The suppressive nature of NaNO 3 on C. vulgaris root enzymatic activities, compared to E. tetralix, suggests that C. vulgaris and its symbionts are more sensitive to NaNO 3 and its effects on peat properties, such as pH. This is also reflected in the vegetation abundance data for NaNO 3 where C. vulgaris abundance is reduced while E. tetralix abundance increases. This sensitivity to NaNO 3 may put C. vulgaris at a competitive disadvantage to other ericaceous species during NO 3 À deposition.
Our findings of the cumulative effects of nutrient addition treatments at Whim Bog on abundances of peatland vegetation are similar to those detailed in Levy et al. (2019) which describes the decline of several plant species and the increase of E. vaginatum as the major effects over the entire timespan of the experimental site. The loss of Sphagnum may also directly reduce the ability of ericaceous species to uptake nutrients, as the upper moss layer is heavily inhabited by ericaceous roots, forming a thick layer which receives nutrients from litter and the atmosphere before it reaches the lower layers (Read et al., 2004). As the living Sphagnum layer is lost and forms bare, decaying peat, it collapses and becomes more submerged and anoxic, becoming an environment that ericaceous roots are less likely to inhabit. This loss of aerobic substrate for ericaceous species to inhabit and uptake nutrients from may be an underlying cause of the observed reduction in C. vulgaris abundance. The loss of the moss layer may also lead to the subsidence of peat, as has been observed by Juutinen et al. (2018) to be a result of nutrient addition at another long-term nutrient addition experiment, Mer Bleue Bog, located in Ontario, Canada. This subsidence was indicated by our water table depth measurements, as they were made relative to the moss surface and the treatments with the highest water table values, NaNO 3 and NaNO 3 þPK, also showed the largest reductions in moss abundance. Furthermore, the effects of the treatments on peat pH should not be overlooked as a significant source of variation. Long-term alteration of pH is likely directly linked to the observed differences in fungal colonization patterns as well as root enzymatic activity.
Loss of ericaceous vegetation and Sphagnum are key examples of the risks posed by anthropogenic N and P deposition. Current research has lacked a clear picture of how ericaceous root associated fungi, ERM and DSE, are involved in these processes. Our findings indicate that altered nutrient limitations shift root associated fungal diversity and morphology, with direct effects on enzyme activity and thereby decomposition potential. The losses of C. vulgaris and Sphagnum due to nutrient addition and the subsequent reduction in DSE colonization rates may have additional consequences. Dark septate endophytes are by nature heavily melanized and may contribute a significant source of recalcitrant C (Fernandez and Koide, 2014). Subsequently, the potential loss of recalcitrant fungal biomass may lead to lower peatland C sequestration. As suggested by Averill et al. (2014) and Orwin et al. (2011), ERM symbionts may be responsible for limiting the decomposition potential of free-living saprotrophs and the microbial community by increasing N and P limitation in soil. Addition of inorganic N and P may upset this limitation, leading to increased decomposition which releases C limitation for the more efficient saprotrophs, which in turn outcompete the mycorrhizal fungi, thereby limiting ericaceous nutrient access. Further research is necessary to determine the potential long-term risks of N and P deposition and the role of peatland mycorrhizal fungi in C sequestration.

Conclusions
The total frequency of fungal colonization at Whim Bog, Scotland, did not decrease under nutrient addition treatments but instead tended to increase by ca. 10% in both Calluna vulgaris and Erica tetralix, refuting our hypothesis (1) which expected a reduction in fungal colonization rates. The considerable increase in ERM hyphal frequency (ca. 20-30%) in both host species was accompanied by a significant suppression of DSE hyphal frequency (ca. 20-30%) under nutrient addition, indicating a strong treatment effect on the root symbiont community. The altered fungal morphotype frequencies and identified fungal species agree with our hypothesis (2) of mycorrhizal diversity indicating nutrient addition effects and suggests that altered nutrient availability is a selective pressure upon the root associated fungal community. The enzymatic activities of both ericaceous shrub roots and their associated fungi strongly support our hypothesis (3) of mycorrhizal root enzyme activities reflecting nutrient limitations.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.