Soil microbial community responding to moderately elevated nitrogen deposition in a Japanese cool temperate forest surrounded by fertilized grasslands

ABSTRACT To examine the hypothesis that the soil microbial community in a nitrogen (N) limited forest responds to moderately elevated N deposition (< 10 kg ha−1 yr−1), correlations between N deposition and soil microbial properties were analyzed in a cool temperate forest surrounded by normally fertilized pasture grasslands in northern Japan. Three experimental plots were established in forest edges adjacent to the grasslands, and the other three plots were in forest interiors at least 700 m away from the grasslands. N deposition in each plot was measured from May to November 2018. In August 2018, we collected litter and surface soil samples from all plots to measure net N mineralization and nitrification rates as indicators of microbial activity, and microbial biomass and various gene abundances (i.e., bacterial 16S rRNA gene, fungal ITS region, and bacterial and archaeal amoA genes) as indicators of microbial abundance. N deposition in forest edges was 1.4-fold greater than that in forest interiors, whereas maximum N deposition was 3.7 kg ha−1. N deposition was significantly correlated with net N mineralization and nitrification rates and 16S rRNA and bacterial amoA gene abundances. Microbial community structures analyzed for bacterial 16S rRNA gene and fungal ITS region amplicons were different between litter and soil samples but were similar between the forest edge and interior, although this analysis was made only for single pair of the two plots in the forest edge and interior. N deposition was also correlated with the soil C/N ratio and nitrate and ammonium contents. Thus, it was suggested that some soil microbial activities and abundances in an N limited forest likely responded to moderately elevated N deposition. These findings provide primary information on soil microbial response to moderately elevated N deposition.


Introduction
Nitrogen (N) deposition (deposition of nitrate (NO 3 − ) and ammonium (NH 4 + ) ions) increased by anthropogenic usage of excess N fertilizers and fossil fuels (Galloway et al. 2004) is still a major environmental concern on a global scale (Decina, Hutyra, and Templer 2019;Dentener, Drevet, and Lamarque et al. 2006;Kanakidou et al. 2016;Reay et al. 2008) while reducing trends in N deposition are also observed in some areas of Europe and North America as a result of efforts to mitigate anthropogenic nitrogen oxide emissions (Gilliam et al. 2019;Schwede et al. 2018;Waldner et al. 2014). Increased N deposition variously affects forest ecosystems by altering biodiversity, productivity, biogeochemical cycles, energy dynamics, and so on (Chiwa et al. 2018;Groffman et al. 2018;Janssens et al. 2010;Jia et al. 2020;Reay et al. 2008;Zhang, Chen, and Ruan 2018). These effects of N deposition on forest ecosystems have been investigated mostly by N manipulation experiments with more than 20 kg ha −1 yr −1 of N addition (Janssens et al. 2010;Jia et al. 2020;Zhang, Chen, and Ruan 2018). A substantial extent of forest ecosystems over the world is, however, still receiving less than 10 kg ha −1 yr −1 of N deposition due to limited and heterogeneous distributions of urbanized areas that are receiving extensively elevated N deposition (> 20 kg ha −1 yr −1 ) (Reay et al. 2008;Schwede et al. 2018). Moreover, N limitations in these forests are expected to occur continuously and broadly because of increased N demand of forest vegetations under increasing atmospheric CO 2 concentrations (Groffman et al. 2018;McLauchlan et al. 2017). Therefore, to capture reliable responses of forest ecosystems against changing N status, we need to know more about the effects of moderately elevated N deposition (< 10 kg ha −1 yr −1 ) on N limited forest ecosystems.
Soil microbial community is an ecosystem component sensitive to changes in N deposition (Janssens et al. 2010;Niu et al. 2016;Tian et al. 2017;Zhang, Chen, and Ruan 2018;Waldrop, Zak, and Sinsabaugh 2004). In soils amended with extensively elevated N deposition, CO 2 release resulting from microbial decomposition of soil organic matter has generally been reduced (Janssens et al. 2010;Jia et al. 2020;Zhang, Chen, and Ruan 2018). Microbial transformations of inorganic N compounds, such as nitrification and denitrification, are generally enhanced by N addition (Niu et al. 2016), increasing the risks of N leaching into the surrounding water body and emission of nitrous oxide (a greenhouse gas 300-fold more effective than CO 2 (IPCC 2013)) to the atmosphere (Butterbach-Bahl, Willibald, and Papem 2002;Niu et al. 2016). In contrast to the microbial responses to extensively elevated N deposition, Allison et al. (2009) found that fungal species isolated from boreal forest soils responded parabolically to N addition from 0 to 200 μg. These amounts of added N were equivalent to only 0.1% or less of amounts of organic N substrates in the soils (Allison et al. 2009). Thus, soil microbial community responses to moderately elevated N deposition may differ from the responses against extensively elevated N deposition. However, varied microbial responses depending on the N deposition levels have been less focused in recent metaanalyses examining the effects of N deposition on soil microbial community (Janssens et al. 2010;Jia et al. 2020;Zhang, Chen, and Ruan 2018).
In the present study, we focused on the relationship between spatially varied N deposition and soil microbial properties within a cool temperate forest in the eastern area of Hokkaido, Japan. Eastern Hokkaido, including the investigated forest, receives relatively low N deposition from the atmosphere (2-5 kg ha −1 yr −1 as a typical value) (Chiwa et al. 2015;Network Center for EANET 2021), while N deposition in the boundary area of the forest (i.e., forest edge) is possibly more than that in the interior area of the forest (forest interior) owing to advection of N fertilizer, which farmers supplied to surrounding pasture grasslands (Figure 1) (Reinmann and Hutyra 2017;Remy et al. 2016Remy et al. , 2017Remy et al. , 2018aRemy et al. , 2018b. The grasslands surrounding our forest have received normal agricultural management practices, but not extensive fertilization, since after the land reclamation from forest to grassland in the 1950s. Therefore, investigating the relationship between N deposition and soil microbial properties in the two contrasting areas (i.e., forest interior and edge) of this forest would provide primary information on soil microbial responses to moderately elevated N deposition. In particular, this study was conducted to examine the hypothesis that soil microbial community in an N limited forest responds to moderately elevated N deposition. If the hypothesis is true, soil microbial activities and abundances in an N limited forest will change along the spatially varied amount of N deposition.

Site description
This study was conducted in a natural, deciduous, broad-leaved forest in the Shibecha branch of the Hokkaido Forest Research Station, Field Science Education and Research Center, Kyoto University (N43° 24.2ʹ, E144° 38.5ʹ, 115 m above sea level) in eastern Hokkaido, northern Japan. The station is registered as an associate site of JaLTER (Shibecha/Shiranuka forest, http:// www.jalter.org/en/researchsites/) and is used in a diverse range of ecological research (Christopher et al. 2008;Hosokawa et al. 2017;Isobe et al. 2018;Nakayama et al. 2019;Nakayama and Tateno 2018;Tateno et al. 2019;Urakawa et al. 2014Urakawa et al. , 2016. Briefly, mean annual air temperature and precipitation for 1981-2010 were 6.2°C and 1169.7 mm, respectively. The growing season is usually from June to October. Season with a persistent snowpack is generally from December to April. The annual maximum snow depth was 64 cm, as an average for 1981-2010. More detailed features of this forest are given in Christopher et al. (2008).
Pasture grassland surrounding the forest is fertilized with ammonium-rich materials derived from livestock manure and slurry. The fertilization was usually conducted in May. Several farmers also conducted the fertilization in September. The major direction of wind flow in both months was from north to south clockwise ( Figure 1). The normal amount of fertilization for pasture grassland in this region was 6 and 4.5 × 10 3 kg ha −1 yr −1 as livestock manure and slurry, respectively (Hokkaido Government, http://www.kushiro.pref.hokkaido.lg.jp/ss/nkc/ gijyutu/H29/JA04hon.htm, access on May 2021). These amounts of normal fertilization were equivalent to 60 and 90 kg N ha −1 yr −1 for livestock manure and slurry, respectively.

Establishment of experimental plots
In May 2018, we established six experimental plots in the forest (Figure 1(a)), expecting those plots to have different levels of N deposition. Three of those plots were located at the forest edge (i.e., Edge 1 to 3), a boundary between the forest and adjacent pasture grasslands, while the other three were located in the forest interior (i.e., Interior 1 to 3) at least 700 m away from the grasslands (Figure 1(a)). All experimental plots were 10 m × 40 m in size. Soils in this forested area have been classified as Andosols, using the classification of the Food and Agriculture Organization (IUSS Working Group WRB 2015). Dominant vegetations of the plots were natural, deciduous broadleaved trees (mainly Quercus crispula know as Japanese oak) with dense understory vegetations of Sasa nipponica. There was no remarkable difference in vegetation status (i.e., species composition, standing tree density, and canopy structure) between the forest edge and interior. Briefly, the mean diameter at breast height of standing trees was 17.3 cm throughout the plots. The mean stand density was 829 trees ha −1 . The maximum height of Sasa vegetation during the growing season was 80-100 cm.

N deposition observation
The amounts of N deposition in the six experimental plots were measured from May 9th to November 20th, 2018, by continuously collecting throughfall water from the atmosphere to the ground through the canopy vegetation. Seven shaded plastic buckets equipped with collecting tubes and funnels (21 cm in diameter) were randomly put on each experimental plot. Throughfall water was collected by the buckets at an almost biweekly interval and then filtered using a 0.45 μm pore-sized membrane filter (ADVANTEC 25CS045AN, Toyo Roshi Kaisya LTD., Tokyo, Japan). Then, concentrations of NO 3 − and NH 4 + were measured using ion chromatography (Dionex-Integrion, Thermo Fisher Scientific, MA, USA). The amount of N deposition for the individual collection interval was quantified by multiplying the ion concentration in the collected water sample to the amount of throughfall. Then, the total amount of N deposition during the six-month observation period was quantified by summing up the N deposition for all collection intervals.

Litter and soil sampling
Litter and surface mineral soil (0-5 cm depth) samples were collected on August 10th, 2018. Three sets of litter and soil samples were collected from each of the experimental plots to obtain the representative mean and the interspatial variation of soil microbial properties within a plot. Here, we determined the litter and soil sampling locations, avoiding areas directly below trees to reduce the possibility of specific effects from roots and rhizospheres on collected samples. Litter samples were collected by gloved hands from an area of 30 cm × 30 cm, which were randomly selected within the plot. Soil samples were collected using a shovel. Collected litter and soil samples were cooled and transferred to the laboratory within a day. Soil samples were gently passed through a 4-mm sieve to remove gravel and plant tissues. Litter samples were pieced into a smaller size (ca. less than 2 mm × 2 mm) to obtain a homogenized sample. The prepared litter and soil samples were immediately applied to soil microbial analysis. Portions of these samples were air-dried and analyzed for total carbon (C) and N contents (Koarashi et al. 2018) and pH (H 2 O), as presented in Table 1. Contents of NH 4 + and NO 3 − in fresh litter and soil samples (Table 1) were measured by the colorimetric method coupling with the 2 M KCl extraction of NH 4 + and NO 3 − (Urakawa et al. 2014(Urakawa et al. , 2016. All data of soil properties in this study are presented with the unit per area after the conversion with measured bulk density in Table S1.

Activity and abundance
Litter and soil samples were also applied to measurements of net N mineralization and nitrification rates, microbial biomass C and N, and various gene abundances, such as bacterial 16S rRNA gene, fungal ITS region, and bacterial and archaeal amoA genes. Mineralization and nitrification rates were measured as indicators of microbial activity, and microbial biomass and gene abundance were indicators of microbial abundance. The net N mineralization and nitrification rates were determined, respectively, as the changes in the concentrations of total inorganic N (NO 3 − + NH 4 + ) and NO 3 − alone between initial and after aerobic incubation of soils at 25°C for 4 weeks (Urakawa et al. 2014(Urakawa et al. , 2016. In the investigated forest, these net mineralization and nitrification rates can indicate gross mineralization and nitrification rates, respectively (Urakawa et al. 2016). Moreover, the well-known correlation between N mineralization rate and microbial CO 2 production rate (Haney, Brinton, and Evans 2008;Rustad et al. 2001;Zak et al. 1999) enables us to infer behaviors of organic matter decomposition and consequent CO 2 release. Microbial biomass C and N were measured using the chloroform fumigation extraction method (Vance, Brookes, and Jenkinson 1987). Total DNA was extracted from 0.1-0.5 g of fresh litter and soil samples using a DNeasy Power Soil Kit (Qiagen, Hilden, Germany). Abundances of microbial genes were then quantified with an Illumina's Eco Real-Time PCR System (Illumina, CA, USA) and commercial reagent kits or primer sets targeting specific gene regions. Femto bacterial and fungal DNA quantification kits (Zymo Research, CA, USA) were used for bacterial 16S rRNA gene and fungal ITS region. The primer sets for bacterial 16S rRNA gene and fungal ITS region were 8 F/357 R (Turner et al. 1999) and ITS1-F/ITS-2-R (White et al. 1990), respectively. For bacterial and archaeal amoA genes, the primer sets of amoA1f/ amoA2r (Rotthauwe, Witzel, and Liesack 1997) and CrenamoA23f/Cremamo616r (Tourna et al. 2008) were used, respectively, with FastStart Essential DNA Green Master (Roche, Basel, Switzerland) as a PCR reaction mixture. The PCR conditions are shown in Table S2.

Community structure
Microbial community structures in litter and soil samples were also analyzed for bacterial 16S rRNA gene and fungal ITS region amplicons using a high-throughput DNA sequencer (MiSeq, Illumina). Due to the limitation of funding ability, DNA samples extracted from Edge 2 and Interior 1 plots were used for this evaluation to briefly capture the difference in the microbial community structure between these two contrasting plots. Interior 1 was located farthest from the surrounding grasslands, while N deposition at Edge 2 was middle of the three edge plots (Figure 1). The amplicon libraries of bacterial 16S rRNA gene and fungal ITS region were prepared using the 16S (V3-V4) Metagenomic Library Construction Kit for NGS (TaKaRa Bio) with the primer set of 341 f /806 R (Caporaso et al. 2011;Muzer, Waal, Uitterlinden, 1993) and the 2× Kapa HiFi HotStart ReadyMix (Roche, Swiss) with the primer set of ITS3-F/ITS4-R (Waud et al. 2014), respectively, in the coupling with the Nextera XT Index Kit (Illumina). A total of 0.16 million of 2 × 250 bp paired-end reads and a total of 1.26 million of 2 × 150 bp paired-end reads were obtained for bacterial 16S rRNA gene and fungal ITS region, respectively, from 12 total genomic DNA samples (2 depths × 2 sites × 3 replications). These sequences were then binned into operational taxonomic units (OTUs) of 1,592 for bacterial 16S rRNA gene and 2,626 for fungal ITS region using CD-HIT-OTU (Li et al. 2012), which were configured with a clustering threshold value of 0.97 and a perbase PCR error value of 0.01. Taxonomies of the OTUs were determined with Quantitative Insights Into Microbial Ecology (QIIME, an open-source software pipeline for analysis of microbial community sequence data) (Caporaso et al. 2010). In the QIIME analysis, OTU sequences for 16S rRNA gene were classified using the RDP classifier (Wang et al. 2007) at 80% confidence threshold with the GreenGenes database (gg_13_18 ver.; DeSantis et al. 2006). The OTU sequences for fungal ITS region were classified using the UCLUST program at 90% confidence threshold with the UNITE database (7.2 2017-12-01 ver.; Abarenkov et al. 2010). Finally, 164,714 reads in 1,591 5.0 ± 0.1 4.9 ± 0.3 5.3 ± 0.1 4.9 ± 0.1 4.9 ± 0.2 4.9 ± 0.1 0.10 a) Mean ± Standard deviation for 3 replicates. b) p < 0.05 is defined as the probability level suggesting statistically significant. Upward and downward arrows indicate significantly positive and negative correlations to N deposition, respectively.
OTUs of bacterial 16S rRNA gene (10,870-16,620 reads/sample) and 1,188,300 reads in 1,946 OTUs of fungal ITS region (66,942-115,420 reads/sample) were applied to the analysis comparing the microbial community structures among samples, after excluding non-bacterial and non-fungal OTU reads that were 3 reads in 1 OTU for bacterial 16S rRNA gene and 79,093 reads in 680 OTUs for fungal ITS region.

Statistical analysis
Statistical analysis in the present study was performed using R software ver. 3.6 (R Core Team 2017). The probability level suggesting statistical significance was defined as less than 0.05 in the present study (i.e., p < 0.05). The two-way ANCOVA using glm function in base package (R Core Team 2017) was applied to examine the significant correlation between N deposition and soil microbial properties and also to examine the significant interactive effect of different soil layers (i.e., litter vs. 0-5 cm soil) on these correlations. The ANCOVA was applied to a total of 36 data points (6 plots × 2 soil layers × 3 replications). The correlation between N deposition and possibly confounding environmental factors, i.e., mean soil water content and temperature for the observation period, was also examined. Soil water content and temperature can vary between forest interior and edge locations due to differences in ecological and meteorological features between forest interiors and edges (Reinmann and Hutyra 2017;Remy et al. 2016Remy et al. , 2017Remy et al. , 2018aRemy et al. , 2018b. The correlations between N deposition and soil chemical properties, i.e., C and N contents, pH (H 2 O), and NO 3 − and NH 4 + contents, were also examined to capture their possible changes associated with moderately elevated N deposition and microbial properties. The correlation between N deposition and soil chemical property was examined individually for litter and surface soil layers using simple linear regression with glm function in R.
The bacterial and fungal community structures were compared among different soil layers and sites with the permutational multivariate analysis of variance (perMANOVA, 9999 random permutations) by adonis function in vegan package (Oksanen, Guillaume Blanchet, and Friendly et al. 2018). Here, the number of sequences per sample was rarefied to 10,870 for bacterial 16S rRNA gene and 66,942 for fungal ITS region by random sampling of rrarefy function in vegan package (Okansen et al., 2018).

Summary of N deposition
Cumulative N deposition via throughfall for the 6 months (from May to November) ranged from 2.2 to 3.7 kg ha −1 in the six experimental plots (Figure 1). The maximum and minimum amounts of N deposition were observed in the most northern plot of the forest edge (Edge 3) and the most northern plot of the forest interior (Interior 3), respectively. In summary, the mean N deposition for three plots of the forest edge was 3.5 ± 0.9 kg ha −1 , which was 1.4-fold higher than that of the forest interior (2.5 ± 0.7 kg ha −1 ). This difference in N deposition between the forest interiors and edges was considered statistically significant by the t-test. A large proportion (> 76%) of the N deposition was in the form of NH 4 + .

Soil microbial properties vs. N deposition
Net N mineralization and nitrification rates showed significantly positive correlations to N deposition (Figure 2). The interactive effect between soil layer and N deposition was not significant (p > 0.05). Abundances of 16S rRNA gene showed significantly positive correlations to N deposition (Figure 3), however the interactive effect between soil layer and N deposition was not significant (p > 0.05). The positive correlation between bacterial amoA gene abundance and N deposition was also statistically significant, showing a significant interactive effect from the soil layer and N deposition combinations. The slope value for the relationship between N deposition and bacterial amoA gene abundance in the surface mineral soils was 3.5-fold greater than that in the litter layer. There was no significant correlation between other microbial properties and N deposition (p > 0.05). There was no significant difference in microbial community structure between Edge 2 and Interior 1, while microbial community structure was significantly different between the litter and soil layers (Figure 4).

Environmental factors vs. N deposition
There was no apparent correlation between N deposition and environmental factors (i.e., temperature and soil water content) ( Figure 5). The differences in mean values of these environmental factors between the forest edge and interior were only 0.2°C in temperature and 1% in soil water content, without any significant differences by t-test (p > 0.05).

Soil chemical properties vs. N deposition
In contrast to environmental factors, some soil chemical properties were found to be significantly correlated with N deposition (Table 1). C/N ratios of litter and soil samples were negatively correlated with N deposition. In soil samples, NO 3 − content showed a positive correlation to N deposition, while NH 4 + content showed a negative correlation to N deposition.

Soil microbial activity vs. N deposition
N disposition was significantly greater in the forest edge than in the forest interior (Figure 1). Given that NH 4 + consisted of more than 76% of the deposited N and the fertilization in surrounding pasture grasslands were conducted with ammonium-rich materials derived from cattle manure and slurry, N fertilizer likely had significant contributions to the forest N deposition through the ammonia volatilization (Hayashi and Yan 2010). However, these amounts of N deposition in this forest were less than half of the threshold N amount, which causes adverse effects on temperate and boreal forest ecosystems (i.e., 10-15 kg ha −1 yr −1 ) (Bobbink et al. 2010;Nordin et al. 2005). Thus, the bioavailability of N in the investigated soils was likely limited even in the edge plots.
Soil microbial activity at the forest edge was likely enhanced by moderately elevated N deposition because significant correlations were found between N deposition and net mineralization and nitrification rates (Figure 2). This microbial response is different from observations of the reduction in soil microbial CO 2 release in forests under extensively elevated N deposition (Janssens et al. 2010;Jia et al. 2020;Zhang, Chen, and Ruan 2018). The enhancements of microbial activity and presumably soil organic matter decomposition at our forest edges are partially supported by the observations of the negative correlations between N deposition and soil C/N ratio (Table 1). The relative abundance of C to N in soil organic matter generally decreases with the progress of microbial decomposition, where organic C is mineralized to and released as CO 2 , whereas N is retained and reutilized by the soil microbial community (Koarashi et al. 2014;Kramer, Lajtha, and Aufdenkampe 2017). Correlations between N deposition and individual content of soil inorganic N species (Table 1) probably resulted from the enhanced NH 4 + consumption and NO 3 − production through the nitrification under moderately elevated N deposition (Figure 2). This postulation is supported by a consistent result that correlations between soil inorganic N availability and microbial property (such as net N mineralization and nitrification rate, and 16S rRNA and bacterial amoA gene contents) in surface mineral soils were sufficiently strong (Pearson's r values were −0.46 to −0.86 for soil NH 4 + pool and 0.68 to 0.90 for soil NO 3 − pool). Such an enhancement of soil microbial activity under moderately elevated N deposition (Figure 2) can contribute to increased CO 2 production through the decomposition of soil organic matter, thus increasing atmospheric CO 2 concentration. Moreover, considering the previously known sensitive responses of microbial processes to N addition (Allison et al. 2009;Butterbach-Bahl, Willibald, and Papen 2002;Jassal et al. 2011;Niu et al. 2016;Smith et al. 2000), not only the extensive elevation but also the moderate elevation of N deposition may significantly alter the soil nutrition dynamics, particularly at forest boundary edge areas.

Soil microbial abundances vs. N deposition
There was a remarkable difference between bacterial (16S rRNA) and fungal (ITS) abundances in terms of their correlations to N deposition (Figure 3). This difference between bacteria and fungi is considered to be the reflection of different N demands between these two different microbial groups (Strickland and Rousk 2010). In general, bacterial biomass is relatively enriched with N rather than fungal biomass does, suggesting a higher N demand for the bacterial body (Strickland and Rousk 2010). Therefore, the observed linkage between bacterial abundance and N deposition is reliable under the assumption of the high sensitivity of bacteria to N availability. These different responses to N deposition between bacteria and fungi were also inferred from a lower ratio of microbial biomass C to N and a lower ratio of fungal to bacterial gene abundance at forest edges than at forest interiors (Figure 3), whereas differences in these ratios between the forest edges and interiors were not statistically significant (p > 0.05).
In forest soils with higher levels of N deposition (e.g., >20 kg ha −1 yr −1 ), both community structures of bacteria and fungi would respond to increased N deposition (Isobe et al. 2019;Kjoller et al. 2012). Isobe et al. (2019) showed phylogenetically conserved responses of soil bacterial taxa to N addition, performing a meta-analysis of 13 field experiments for soil bacterial communities across 5 continents. Additionally, in a Norway spruce forest in Denmark, Kjoller et al. (2012) found that fungal community structure and its function significantly responded against N deposition gradient from 27 to 43 kg ha −1 yr −1 . Thus, relatively lower levels of N deposition in our forests (Figure 1(b)) likely enabled us to detect microbial responses differing between bacteria and fungi (Figure 3). Note that our results for the microbial DNA amplicon analysis using MiSeq (Figure 4) may still be insufficient to achieve a consistent insight into the soil microbial community under moderately elevated N deposition because there was only a result from single pair of two plots in forest interior and edge.
The bacterial amoA gene abundance appeared to respond to N deposition, but the archaeal amoA gene abundance was not (Figure 3). In the investigated forest, Isobe et al. (2018) also found gross nitrification rate synchronously changing with bacterial amoA gene abundances during winter. These microbial features at our forest site are somewhat different from a previous study finding that archaeal ammonia oxidizers play an important role in the soil nitrification process in temperate forests and agricultural upland soils in Europe (Leininger et al. 2006). One of the possible interpretations for this discrepancy between European and Japanese forest soils is that soil conditions of our forest are preferable for bacterial ammonia oxidizers, which have larger cell bodies and higher cell-specific-unit activity compared with archaeal ones (Jia and Conrad 2009). Then, the specific dependence of bacterial ammonia oxidizers on autotrophic growth and contrasting dependence of archaeal oxidizers on heterotrophic growth (Jia and Conrad 2009) might result in the dominant contribution of bacterial community to nitrification in our forest soils.
Moreover, this specific sensitivity of bacterial ammonia oxidizers might be associated with changes in the species composition of these bacteria (Isobe et al. 2020). While the overall compositions of bacterial and fungal communities in our forest soils were less sensitive to moderately elevated N deposition (Figure 4), Isobe et al. (2020) found significant changes in the species composition of bacterial ammonia oxidizers along with forest slope gradients. Accordingly, Isobe et al. (2020) pointed out the importance of specific microbial community compositions in elucidating soil N dynamics under changing environmental conditions.

Environmental factors vs. N deposition
In our forest, there was little evidence indicating that environmental factors other than N deposition had caused the pseudocorrelation between N deposition and soil microbial properties. This was based on our findings that environmental factors such as soil water content and temperature were similar between the forest edge and interior ( Figure 5). Additionally, soil pH was less correlated with the amount of N deposition (Table 1), while Urakawa et al. (2016) suggested that soil acidity would be a significant factor affecting N mineralization activity in Japanese Andosols. Thus, the moderately elevated N deposition in our forest might be insufficient to create Figure 4. Comparison of fungal and bacterial community structures between Edge 2 and Interior 1 plots and between litter and soil layers. The probability level (p value) for statistical significance examined by perMANOVA was presented above panels. The phyla which individually had less than 1% of proportional composition was binned into other phyla. a significant gradient of those environmental and soil physicochemical factors that can directly affect microbial properties. The situation indicating similar environmental and soil physicochemical factors between the forest interiors and edges was strictly different from the situations in previous studies in forests in Europe (Remy et al. 2016(Remy et al. , 2017(Remy et al. , 2018a(Remy et al. , 2018b and USA (Reinmann and Hutyra 2017), where not only N deposition but also other environmental and soil physicochemical factors have changed gradually from edge to interior.

Conclusions
In a Japanese cool-temperate forest surrounded by pasture grasslands, we found that soil microbial activities and their abundances increased along with spatial gradients of N deposition from forest interior to boundary edge area. Still, elevated levels of N deposition in the forest edges were moderate (<10 kg ha −1 yr −1 ) rather than extreme (>20 kg ha −1 yr −1 ). Our finding was different from most previous studies, which mainly focused on the effects of >20 kg ha −1 of annual N deposition and showed a reduction in soil organic matter decomposition and microbial CO 2 release (Janssens et al. 2010;Jia et al. 2020;Zhang, Chen, and Ruan 2018). Because of the significant coverage of N limited forest ecosystems (Groffman et al. 2018;McLauchlan et al. 2017;Reay et al. 2008) and the significant increase in fragmented forest ecosystems over the world (Haddad et al. 2015;Smith et al. 2018), understanding the responses of soil microbial communities to moderately elevated N deposition is essential, to capture reliable changes in C and nutrient cycles under changing environments. Through this study, we provided primary information on soil microbial response to moderately elevated N deposition. Figure 5. Correlations of soil water content and temperature to N deposition. The presented soil water content and temperature are seasonal means for the observation period from May to November 2018. No significant correlation was observed between N deposition and the environmental factors without any significant difference by t-test (p > 0.05).