Dynamics of soil nitrogen fractions and their relationship with soil microbial communities in two forest species of northern China

Microbially-mediated soil N mineralization and transformation are crucial to plant growth. However, changes in soil microbial groups and various N components are not clearly understood. To explore the relationship between soil N components and microbial communities, we conducted an in-situ experiment on two typically planted forest species, namely, Sibirica Apricot (SA) and Prunus davidiana Franch (PdF) by using closed-top polyvinyl chloride tubes. Changes in soil inorganic N, organic N (ON) fractions, and levels of microbial phospholipid fatty acids (PLFAs) were measured bimonthly from April 2012 to April 2013. Microbial PLFAs and the concentrations of easily-available microbial biomass N (MBN; ~60 mg kg-1), soluble ON (SON; ~20 mg kg-1), and inorganic N were similar between the two soils whereas the ON (~900 mg kg-1) and its major part total acid-hydrolyzable N (HTN; ~500 mg kg-1), were significantly different (p < 0.05) in most months (5/6 and 4/6; respectively). The canonical correlation analysis of soil N fractions and microbial parameters indicated that the relationship between total PLFAs (total biomass of living cells) and NH4+-N was the most representative. The relative contributions (indicated by the absolute value of canonical coefficient) of NH4+-N were the largest, followed by NO3−-N and MBN. For the HTN component, the relative percentage of hydrolyzable amino acid N and ammonium N decreased markedly in the first half of the year. Canonical variation mainly reflected the relationship between ammonium N and bacterial PLFAs, which were the most sensitive indicators related to soil N changes. The relative contributions of HTN components to the link between soil microbial groups and HTN components were ammonium N > amino acid N > amino sugar N. Observations from our study indicate the sensitivity of soil N mineralization indicators in relation to the temporal variation of soil microbial groups and N fractions.


+
-N was the most representative. The relative contributions (indicated by the absolute value of canonical coefficient) of NH 4 + -N were the largest, followed by NO 3 − -N and MBN. For the HTN component, the relative percentage of hydrolyzable amino acid N and ammonium N decreased markedly in the first half of the year. Canonical variation mainly reflected the relationship between ammonium N and bacterial PLFAs, which were the most sensitive indicators related to soil N changes. The relative contributions of HTN components to the link between soil microbial groups and HTN components were ammonium N > amino acid N > amino sugar N. Observations from our study indicate the sensitivity of soil N mineralization indicators in relation to the temporal variation of soil microbial groups and N fractions. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Introduction nitrification rate, and fungal biomass [41]. The heterotrophic nitrification and immobilization of NO 3 -N may be important N transformation pathways affecting ecosystem productivity. However, the relationships and interplay between changes in the various N fractions and the microbial community remain unclear.
In the present study, we selected a typical revegetated region, the Chinese Loess Plateau in Southern Ningxia. Within the investigated region, the typical cultivated forests, Sibirica Apricot (SA) and Prunus davidiana Franch (PdF), were selected as the representative species to study the relationship between soil N fractions and microbial groups during N mineralization processes. A one-year (from April 2012 to April 2013) in-situ field incubation experiment was conducted in the two typical forested sites. The objectives of this study were to i) quantify the dynamics of soil N fractions, including the net rate of the ammonification, nitrification, and mineralization, and to identify the variation in the percentage of the hydrolyzable N components at the monthly scale ii) identify changes in the microbial communities accompanied by soil N change, and iii) further explore the relationship between soil N and microbial community as affected by plant species and time. We formed the following hypotheses: i) the influence of plant species on soil ON is stronger than that on inorganic N; ii) the effect of temporal variation on soil N and microbial community is stronger than that of plant species; and iii) changes in soil microbes are closely related to easily available N forms.

Site description
The study area is located in the southern mountains of Ningxia, China, at the Shang-Huang Ecological Station (35˚59'-36˚03' N, 106˚26'-106˚30' E; altitude: 1534-1822 m a.s.l.) of Institute of Soil and Water Conservation of the Chinese Academy of Sciences. The local site experiences a monsoon climate with a transition from semi-arid to warm temperate. The mean annual rainfall is approximately 420 mm. The average annual temperature is approximately 6.9˚C. According to the soil classification system of the Food and Agriculture Organization of the United Nations (FAO), the loessial soil [42] represents a silty clay loam texture. The main land-use types are artificial grassland (primarily Medicago sativa L.), artificial forestland (Korshinskii and Pyrus spp. pear), abandoned land (Stipa bungeana Trin., Thymus mongolicus, and Artemisia giraldii Pamp.), and farmland (Triticum aestivum and Zea mays).

Experimental design and field incubation
Sibirica Apricot (SA) and Prunus davidiana Franch (PdF) are typical plants cultivated on the Chinese Loess Plateau to reduce the rate of soil erosion with an average survival rate of~50% [43]. The experiment was conducted on the planted field of SA and PdF forestland. For controlling soil spatial variability, we selected two sites with a close distance (<1 km) and similar topography (hillside field) and land-use history (abandoned cropland with previous cultivation of wheat (Triticum aestivuml). Soils of the study area were all developed from the same loessial parent soil material. The detailed geographical characteristics are shown in the Table 1. Meanwhile, for minimizing plot effects and obtaining representative soil samples, three 10 m × 10 m replicate subplots were established at each plant species site in April 2012.
Annual in-situ net N mineralization was measured by the buried soil core method [44]. Specifically, after removal of surface litter, six polyvinyl chloride (PVC) cylinders (7 cm in diameter and 11 cm in length) were installed in each subplot, at a depth of 10 cm into the soil and an adjacent distance of 50 cm. One soil core was taken for immediate analysis, the remaining intact soil cores were placed into a PVC collar, and then the soil cores were sealed with plastic wraps on the top to minimize evaporation. Absorbent cotton was placed on the bottom to maintain enough ventilation. All of the sealed PVC collars were placed back into soils at a depth of 10 cm. Aboveground litter was replaced on the top of the PVC cylinders for in-situ incubation.

Soil sampling and analysis
At each subplot, three replicate soil samples were collected from PVC cylinders from April 2012 to April 2013 by removing one cylinder at each time, with an interval of 60, 120, 180, 240, and 360 days, separately. Soil horizons included in the cores were simply A horizon. When soil cores were collected, stones and coarse roots were removed from the soil. The samples were stored in cooling boxes and transported to the laboratory, where soil samples were homogenized with 5 mm-sized mesh sieves. One-third of the homogenized samples were frozen at -20˚C for microbial phospholipid fatty acid (PLFA) analysis while the remaining portions were air-dried, sieved through 2 mm mesh, and stored at +4˚C (<48 h) for other chemical analyses.

Basic physical and chemical characteristic analyses
Soil temperature at 5 cm depth was measured with a mercury thermometer (1/20˚C). Soil moisture was measured by oven drying the soil at 105˚C for 24 h and measuring the weight loss. Soil pH was measured using a soil suspension extracted at a 1:2.5 (w/w) soil:water ratio. Soil bulk density was measured using the core method [45]. Soil organic carbon was determined via wet oxidation using dichromate in an acid medium, followed by the FeSO 4 titration method [46]. Total N was measured by Kjeldahl digestion and distillation azotometry [47]. -N) was determined by measuring extracts with an automated Continuous-Flow Auto Analyzer (Bran Luebbe AA3, German). Net N ammonification and nitrification rates were calculated from the differences of soil NH 4 + and NO 3 − concentrations between days 0, 60, 120, 180, 240, and 360.

Inorganic N and mineralization rate analyses
Mineral N (N min ) was calculated by summing the concentration of NO 3--N and NH 4 + -N. Net mineralization was calculated from the difference of soil inorganic N (NH 4 + -N and NO 3--N) before and after incubation in the PVC core.

MBN and SON analyses
Soil microbial biomass nitrogen (MBN) was measured using the chloroform fumigationextraction method [48]. Soil samples subjected to fumigation and non-fumigation treatments were extracted in 0.5 M of K 2 SO 4 at a ratio of 1:4. The concentration of K 2 SO 4 -extracted total N was analyzed using a modified method of alkaline persulfate oxidation [49] and nitrate was determined by ultraviolet spectrophotometry analysis in a spectrophotometer (Hitachi, UV2300) at 220 and 275 nm. Microbial biomass nitrogen was calculated using a K EN factor of 0.45 [48]. Total soluble N was determined by the extracts of non-fumigation soil samples. Soluble organic nitrogen (SON) was calculated by subtracting the concentration of inorganic N from total soluble N.

Soil hydrolyzable N fractions analyses
Total soil hydrolyzable N (HTN) and its components were analyzed according to the method from Bremner (1965) [50]. In brief, total soil acid-hydrolyzable N was fractionated by mixing 5 mL of 6 M of HCl hydrolysis and 2 mL of 5 M H 2 SO 4 in Kjeldahl bottles. In addition, the following three acid hydrolysis solutions of each soil were prepared and analyzed: acid-hydrolyzable ammonium-N (HAN) by adding 2.5 mL of 3.5% MgO to 10 mL of acid hydrolysis solution; acid-hydrolyzable amino sugar N (HASN) by mixing phosphate-borate buffer (pH 11.2) and acid hydrolysis solution in 1:1 ratio (v/v); and acid-hydrolyzable amino acid-N (HAAN) by mixing acid hydrolysis solution and 0.5 M NaOH at a ratio of 5:1 (v/v). Then, the nitrogen concentrations in all of these treated solutions were determined by an automatic azotometer with a blank test conducted synchronously. Acid-unhydrolyzable N (UHN) was calculated by subtracting HTN from total N.

Statistical analysis
The differences in soil N and microbial parameters over the period of six months were compared by ANOVA followed by Duncan's post hoc test at a significance level of 5%. An independent two-sample t-test was further used to examine the differences within a single To determine how microbial PLFAs were related to different soil N forms, soil N groups were split into two the sub-groups, namely, the total soil hydrolyzable N components (N1, N2, N3, N4, N5 and N6) and the rest N fractions (N7, N8, N9, N10, N11 and N12). Then, the two groups of soil N indicators were separately subjected to CCA with the PLFA data.

Physical and chemical soil characteristics
In the SA and PdF soils, temperature at 5 cm depth in the soil varied between 1.24˚C in December and 25.12˚C in August (Table 2). Soil moisture ranged from 7% to 20%. Averaged soil moisture was higher in SA soil than that in PdF soil. By contrast, the bulk density of PdF was higher than that of the SA. For each soil sample, pH did not differ significantly within those months (Table 2). By comparing the two-time points of April 2012 and April 2013, contents of SOC significantly increased by 4.7% and 6.0% in SA and PdF; soil C/N ratio showed a significant increase of 12.0% and 30.4% in SA and PdF, respectively.

Ammonification, mineralization and nitrification
Average annual rates of ammonification, nitrification and mineralization were similar between the two soils, whereas the three rates varied strongly with time (Table 3). During the first incubation period (0-60 days, from April to June 2012), the ammonification rate was 0.016 mg kg −1 day −1 in the SA soil and −0.024 mg kg −1 day −1 in the PdF soil ( Table 3). The nitrification rate was three times higher in SA (0.046 mg kg −1 day −1 ) than in PdF (0.015 mg kg −1 day −1 ) (P < 0.05). The mineralization rates were 0.062 mg kg −1 day −1 in SA soil and −0.009 mg kg −1 day −1 in PdF soil. During the days of 60-120 (from June to August), negative rates were observed for the ammonification (−0.13 mg kg −1 day −1 ), nitrification (−0.114 mg kg −1 day −1 ) and mineralization (−0.244 mg kg −1 day −1 ), with a relatively higher incidence of these rates in SA soil compared with PdF soil (Table 3). No significant changes in inorganic N transformation rates were noted from days 120−180 (from August to October). However, the rates of nitrification and mineralization were significantly higher in SA (0.012 and 0.021 mg kg −1 day −1 ) than in PdF (0.004 and 0.013 mg kg −1 d −1 ) during days 180−240 (from October to December). During days 240−360 of experiment (from December 2012 to April 2013), PdF presented higher ammonification, nitrification and mineralization rates (0.066, 0.060, and 0.043 mg kg −1 day −1 , respectively) compared with SA (0.033, 0.010, and 0.043 mg kg −1 day −1 , respectively; Table 3).

ON, SON, and MBN
Organic nitrogen (ON) concentration remained stable over the year for each soil (Fig 1D). Annual average ON concentration was higher in SA (960 mg kg −1 ) soil than that in PdF (770 mg kg -1 ) soil. The concentrations of SON and MBN varied similarly (Fig 1E and 1F). After 2 months of incubation from April to June, the SON concentrations increased by approximately fivefold and onefold in SA and PdF, respectively ( Fig 1E). Microbial biomass nitrogen concentration increased by 130.5% in SA and decreased by 40.7% in PdF (Fig 1F).
The lowest values were observed after 4 months of incubation in August (Fig 1E and 1F); SON and MBN concentrations decreased by 25.6% and 79.1% in SA and by 54.4% and 75.8% in PdF, respectively. After this time point, the concentrations all increased continuously with a strong MBN trend. At the end of the incubation trial in April 2013, the content of MBN and SON increased by~7 and >3 times, respectively (Fig 1E and 1F). The averaged concentrations of MBN (calculated on an annual base) and SON were approximately 55 and 15 mg kg −1 , and no significant difference can be detected between the two soils.

Hydrolyzable total N (HTN) and its components
Hydrolysable total N (HTN) content significantly increased (p < 0.05) from August to October, and significant difference was observed from August to December (p < 0.05; Fig 2A) in both soils. In comparing monthly change of HTN between the two soils, PdF showed a stable change with no significant difference (p > 0.05) in April, October, December 2012, and April Table 3. Ammonification, mineralization, and nitrification rates (mg kg −1 d −1 ) in SA and PdF soils along temporal patterns. 2013. Hydrolysable total N content of SA was significantly higher (p < 0.05) in October and December than that in other months (Fig 2A). For hydrolysable total N components, HAAN was dominant except for the months of October 2012 and April 2013. In October, the two soils showed highly significant difference (p < 0.01) for HTN content (Fig 2A) whereas the relative percentage of HTN components was similar: the highest percentage of HUN (~50%), followed by HAN, HAAN, and HASN ( Fig  2B). By comparison, from April to August, HTN was dominated by the fractions of the HAAN and HAN in both soils. However, the fractions of HUN and HASN gradually increased in the subsequent months (Fig 2B). In the course of a year (from April of 2002 to 2003), within the HTN fraction, soil HSAN exhibited a marked increase of 10 and 5 times in SA and PdF, respectively ( Fig 2B).

Change in microbial community structure as determined by PLFA
Both soils of SA and PdF were bacteria dominated, with a relative abundance exceeding 60%. No significant plant species effect on microbial PLFA was detected at the start and end of the experiment (Fig 3). However, in warm months (June and August), the contents of Gram-positive PLFA and actinomycetes PLFA were significantly higher in PdF soil than those in SA soil (Fig 3B and  3F), whereas in October, the total PLFA and the PLFAs belonging to all bacteria, Gram-negative and fungi, were all significantly higher in SA soil compared with PdF soil (Fig 3).

Correlation between soil N and microbial indicators
Canonical correlation analysis (CCA) was performed using soil N and microbial PLFA data. In total, six pairs of canonical variates (CVs) were extracted individually (Tables 4 and 5). For the HTN components, the canonical correlation between the first soil N canonical variate (N-CV1) and the first microbial canonical variate (P-CV1) was significant (R = 0.989) and showed a good fit (p = 0.0001). The first CV mainly reflected the relationship between the acid hydrolyzable ammonium (N2) and bacterial PLFA (P2). Approximately 43% of the variance in P-CV1 was explained by the N-CV1 (as indicated by the proportion that explained from between-cluster; Table 4). The contributions (evaluated by the absolute value of canonical coefficient) of the different forms of acid hydrolyzable N as evaluated by the canonical coefficient of CV were in the order of HAN > HAAN > HASN > unknown N > UHN > HTN. By contrast, microbial PLFA contributions were in the order of bacteria > Gram (−) > fungi > Gram (+) > total PLFA > Actinomycetes.
As for the inorganic N and other ON fractions, the canonical correlation between the first soil N CV (N-CV1) and the first PLFA canonical variate (P-CV1) was significant (R = 0.994) with a favorable fit (p = 0.0001). The first CV mainly reflected the relationship between the NH 4 + -N (N2) and the total PLFA (P2). Approximately 50% of the variance in the P-CV1 was explained by the N-CV1 (Table 5;

Influence of plant species
Plant species influence soil nutrient availability through their effects on litter decomposition, nutrient uptake, inputs, and losses [55,56]. Sibirica Apricot (SA) and Prunus davidiana Franch    Relationship between soil N fractions and microbial communities (PdF) are two typical revegatation types planted on the Loess Plateau in China to reduce the rate of soil erosion. Our results showed the stronger effects of variations in plant species on soil ON than that in soil inorganic N stocks, which is further in line with our first hypothesis. The average concentrations of soil NH 4 + -N did not differ significantly between the SA and PdF. This finding agrees with those of Ren et al. (2011), who reported that NH 4 + -N concentration was unvaried even under different vegetation types among coniferous, mixed, and broadleaf forests. However, this result contradicts previous findings [57,58] where site specific characters, such as soil and bed rock type, temperature, moisture, and vegetation performed a key role in modifying N stocks in forest soils. Sibirica Apricot soil showed higher ON content than PdF. This finding may be attribute to inherent soil/site difference and plant species. However, given the similar land-use history and soil type of the two sites, we mainly ascribe the reason of increased accumulation of N in soils to aboveground leaf litter traits and its decomposition rates. Lignin is considered as a major recalcitrant part of leaf litter and is found to be relatively higher in SA (14%) than that in PdF (12%). This may further lead to increased accumulation of N in SA soil since high lignin content in leaf litter inhibits the mineralization of organic N fractions in most cases [59][60][61]. For hydrolysable total N components, a stable variation in amino acid concentration was detected, as is consistent with a study performed by Jones et al. (2009), at a global scale that covered 40 sites [62]. The steady change of amino acid was probably due to the fact that it is the major component of slowly-decomposing organic N. Moreover, the decomposing process that converts high weight organic matter (such as humus and lignin fractions) into low weight matters (such as amino acid and amino sugar), is the rate-limiting step during soil organic matter decomposition. Differences in aboveground vegetation reportedly affect soil microbial communities [63,64]. Our results showed that high relative abundance of microbial groups exist in the soils with high relative percentage of HASN (Fig 2) and N min . A possible connection between the N fractions and the microbes is the high relative abundance of microbial groups, which can result in high microbial activities, microbial metabolism, and ultimately high microbial residues (which can be marked by HASN; [10]). This increment can further accelerate soil N mineralization (resulting in increased content of N min ); therefore, UHN fractions, such as plant lignin fractions and/or dominant structural components of humic compounds, are correspondingly high [65].
In summary, the comparison between two sites planted with different plant species showed minor changes in soil microbial community and soil N (Table 3 and Fig 3). Therefore, we expected that the real differences were induced by temporal effects.

Temporal effects on soil N and microbial community
This analysis suggests that in comparison with variations in plant species, temporal variations result in additional changes in soil microbial communities and N contents, which support the second hypothesis of our study. Therefore, we further focused on whether soil measurement varies synchronously with time. Given that monthly-based changes are mainly reflected by the changes in abiotic factors, such as temperature, soil moisture, and rainfall, we compared the measured soil parameters and abiotic factors. The lowest soil temperature (~1.5˚C) was observed in December 2012 (winter) with SA and PdF treatment, whereas, in spring, the temperature increased up to 3.5˚C in April 2013 (Table 2). Meanwhile, the changes of microbial biomass and abundances of all microbial groups were synchronous with time, lowest values in December and increased in the following year (Fig 3). Temperature-related microbial shifts during the winter-spring transition are common features across various systems, and reveals a widespread biogeochemical pattern in seasonally frozen ecosystems [66]. In our study, temporal variations from December 2012 to April 2013 (accompanied by seasonal freeze-thaw transitions) were followed by non-synchronous changes in microbial biomass and soil N contents. We observed a general decrease in microbial biomass and an increase in large parts of the contents of inorganic N and SON, which are related to the acceleration of the release of soil N components at higher temperature; by contrast, soil microorganisms tend to lag behind when available nutrients are incorporated into their biomass to maintain their self-metabolism [67,68]. In general, temporal variations of individual microbial groups (determined by PLFAs analysis) were synchronous regardless of the differences in plant species (Fig 3). This result agrees with those of Liu et al. (2016), who observed that microbial communities are strongly affected by environmental constraints, such as sampling time, soil moisture, and air temperature [69]. In a similar study conducted on the soils of tall birch (Betula glandulosa) and surrounding dwarf birch hummock vegetation, the fungal dominance, principal fungal, and bacterial types also exhibited synchronous variation with time [27].
In comparison with microbial groups, soil N fractions varied non-synchronously. The concentrations of NO  (Table 3). These results were comparable with those of a study considering different vegetation types in northern China [70]. From June to August, the high soil C:N ratios (>9; Table 2) coupled with high soil rainfall (averaged 95 mm; Table 2) and moderate temperature (15-25˚C; Table 2) possibly promoted net N immobilization in the form of amino acid N because the highest relative percentage of the HAAN during the same period was detected (Fig 2). This finding is consistent and supported by results from other studies showing that HAAN is closely associated with microbial metabolism and functions as an important storage pool for immobilized N [10,71,72].

Relationship between microbial populations and N
Considering that soil N mineralization is a biological process driven by microbial activity, we tracked the changes in soil microbial parameters accompanied by N mineralization processes via in-situ buried soil core method. This method has been used as a common method for estimating N mineralization rates in soils [73,74]. From a microbial perspective, the in-situ core method exerts certain influences soil microbes due to the absence of C input above-and belowground. For instance, as a result of top-sealing the cores (with the purpose of preventing N deposition and leaching loss), external C-substrate availability becomes a limiting factor for soil microbial communities [75] because the quantity and chemical composition of aboveground leaf litter play an important role in shaping microbial community structure in forest soils [69,76]. At the early stages of litter decomposition, increased input of leaf litter-dissolved organic matter favors the growth of bacteria over that of fungi [77,78]. Therefore, in comparison to the soil inside the core, surrounding soil tends to harbor high amounts of bacterial biomass. Given that this method is equivalent to root exclusion treatments (which eliminate C inputs except via capillary flow through the bottom cotton), the soil core approach decreases fungal biomass and alters the bacterial community structure in forest soils [79][80][81].
Although absence of roots and leaf litter induced by the soil core method may lead to a decrease in fungal and bacterial biomass, this method remains reliable in maintaining similar soil microenvironment for soil microbes during in-situ incubation [73,74]. Moreover, the possible differences from the method itself can be treated as systematic errors and are largely ignored given the following considerations: i) the two forested soils were subjected to the same (same PVC cylinder material, buried at the same depth) soil core incubation; and ii) the main aims of our experiment were not to compare the changes of soil microbial communities inside and outside of the core but to determine their temporal variations and relationships with soil N fractions. In view of the temporal patterns of soil N and microbial community, we aimed to associate measured soil N and microbial variables by using CCA. The identified relationship was in line with our third hypothesis. Canonical correlation analysis showed the relationship between the NH 4 + -N and the total PLFAs, which were the most sensitive indicators related to microbially-mediated soil N variation ( Table 5). Given that the total PLFA generally comprises the total biomass of living cells, the identified relationship highlights the link among all living microbes and easily available concentration of NH 4 + -N. Owing to the non-synchronous change in soil N fractions (as discussed before), the mechanisms by which soil organic N components interact with individual microbial groups should be clarified. The relationship between ON components and microbial PLFAs were indicated by the link between bacterial PLFAs and the easily available form of HAN (Table 4). For the N source, a large fraction of soil HAN (contributes about 20-35%) was derived from acid-labile organic constituents, such as exchangeable and clay-fixed NH 4 + [2,3]. Therefore, soil bacterial groups are possibly prone to utilizing N on labile substrates. Bell et al. (2008) found that seasonal and annual variability of soil bacterial activity was the most closely associated with extractable NH 4 + -N, pH and SOM. However, the mechanism of how bacterial groups interact with various NH 4 + forms cannot be clearly uncovered by the present study because NH 4 + -N production rates were correlated positively with large pools and production rates of dissolved soil C and N, high quality litter inputs, and low soil C concentration [27]. Tahovská et al. (2013) also proposed that the structure of the bacterial community was related to the dissolved organic C and the concentrations of C and N in microbial biomass [82]. We expect that future research, combined with not only soil N but also C and easily available fractions, will provide further understanding of this relationship. In summary, we used the multivariate analysis of CCA to clarify the main link during a microbially-mediated N change. However, we need to carefully explain the two relationships. The obtained links help us to focus on the main features of soil N-microbes variation, but these findings does not simply mean that bacteria were closely correlated with organic HAN and that the easily available concentration of NH 4 + -N is preferred by all living microbes.

Conclusions
To improve our understanding of soil N and microbial changes on a broad forested region in the Chinese Loess Plateau, we selected two commonly cultivated forests sites as representative ecological systems and explored the relationship between soil N fractions and microbial groups during in-situ N mineralization. A comprehensive investigation on soil N fractions revealed that the total ON (accounting for the highest percentage within soil N) exhibited a minor temporal change, whereas the other forms of N exhibited a non-synchronous variation with time. As a microbially-mediated process, the dynamics of different soil microbial groups were more affected by time than by plant species. Our data highlighted the importance of total PLFAs, microbial PLFAs belonged bacteria, and easily-accessible inorganic NH 4 + -N and organic HAN in understating the main links between soil N fractions and microbial groups.
Supporting information S1