Soil Net Nitrogen Mineralization at Different Ecosystem Development Stages after the Year 2000 Eruption on Miyakejima Island

Nitrogen (N) is one of the vital nutrients for life [1]. It often limits primary production in terrestrial ecosystems [2,3], particularly in ecosystems that have newly established after severe disturbances, such as volcanic eruption [4] and glacial recession [5]. Although it is an abundant element of Earth’s atmosphere and soil, but it is present in the unavailable form to plants. Nitrogen becomes available for plants largely through N mineralization processes in which organic N is converted to inorganic forms, such as NH4 + and NO3 which can be taken up through plant roots. Thus, N mineralization is one of the most important processes for plant growth [6,7]. Because of the importance of N mineralization in the soil, recently soil N mineralization rates and the factors that control it have been measured in situ in various ecosystems: these in situ studies reported huge variations in both time and space [8-10].


Introduction
Nitrogen (N) is one of the vital nutrients for life [1]. It often limits primary production in terrestrial ecosystems [2,3], particularly in ecosystems that have newly established after severe disturbances, such as volcanic eruption [4] and glacial recession [5]. Although it is an abundant element of Earth's atmosphere and soil, but it is present in the unavailable form to plants. Nitrogen becomes available for plants largely through N mineralization processes in which organic N is converted to inorganic forms, such as NH 4 + and NO 3 which can be taken up through plant roots. Thus, N mineralization is one of the most important processes for plant growth [6,7]. Because of the importance of N mineralization in the soil, recently soil N mineralization rates and the factors that control it have been measured in situ in various ecosystems: these in situ studies reported huge variations in both time and space [8][9][10].
Nitrogen mineralization occurs in series which is controlled by many biological processes. Like soil N mineralization rates have been considered to vary with abiotic factors in soil, such as soil temperature; soil moisture content [11,12]; and other soil properties, such as C:N ratio [13,14] and soil organic matter (SOM) [15]. Plants have been also recognized as important in the regulation of soil N mineralization rate because plants provide not only organic matter from both above and below-ground parts, but also habitat for soil microbes. For example, Björk et al. [16] investigated linkages between N turnover including N mineralization and plant community structure in an alpine tundra in Sweden: they showed an indirect effect of plant community on N mineralization. Bengston et al. [17] demonstrated a relationship between plant diversity, productivity, and soil gross N mineralization rate in a mixed beach-oak forest. In a wet grassland in France, Rossignol et al. [18] suggested that plant type altered soil N mineralization via regulation of soil microbial activity.
As for long-term variation in N mineralization rate, succession initiated by severe disturbances may be considered to be important factors, since soil and plant properties show drastic change over long time periods. Plant development and changes in the availability of soil nutrients along with succession have been described by many studies [5,6,[19][20][21][22]. Vitousek et al. [21] reviewed various studies of soil N mineralization rate in primary successions that occur over decades to hundreds of years and demonstrated that soil N mineralization rate showed no constant change along successional gradients. Binkley et al. [23] reported that soil N mineralization rate on riverside terraces did not change directionally along a succession. Hence, long-term patterns of soil N mineralization rate along successional gradients and underlying mechanisms remain unclear due to insufficient in situ observation data.
For present study, site was Miyakejima, a volcanic island near the western rim of the Pacific Ocean; it offers an opportunity to describe how soil N mineralization rate differs among diverse ecosystems from grasslands produced after the latest eruption in 2000 to mature forests. Historical records of eruptions of this island extend back to 1643, over which time there were seven eruptions [24]. These catastrophic disturbances initiated succession; consequently, on Miyakejima there are diverse ecosystems with various types of vegetation and soil properties [25,26]. Also there is a unique shrub, alder (Alnus sieboldiana), which is a N-fixing plant [5]. This shrub species can facilitate plant development [25] by increasing N mineralization and N input to the soil [27][28][29]. So it was hypothesized that soil N mineralization rate at the shrub dominant sites on the island would be higher than those at other sites.
The study was aimed to clarify the relationships between in situ soil N mineralization and environmental factors at various sites on Miyakejima. In this study, ten sites were established where soil N mineralization rate can be measured and so that factors controlling the mineralization can be explored. Using such sites level data, soil N mineralization rates were estimated at typical vegetation types, grassland, shrub land, and forest of Miyakejima and were compared the rates with those in other ecosystems.

Site description
The study was conducted on Miyakejima (34°05′N, 139°31′E) near the western rim of the Pacific ocean, about 180 km south of Tokyo, Japan. The annual mean temperature and precipitation from 2005 to 2014 were: 17.7°C and 2948 mm. The island is an active volcano and the most recent eruption was in 2000; in the summer of that year, the eruption at the summit of Mt. Oyama produced total 9.3 × 10 6 m 3 (dense-rock equivalent) deposits, then formed a new collapsed crater [30]. Harmful gases emission and volcanic ash deposition from the 2000 eruption severely damaged vegetation, especially around the crater [26]. Although the volcanic activity including emission of the harmful gases, such as sulfur dioxide (SO 2 ) has continued, ecosystems damaged by the 2000 eruption have been undergoing restoration succession. Since the type and degree of damage differed among areas, there are various ecosystems on Miyakejima. Ten sites were selected that having various vegetation composition and coverage ( Figure 1). All of the sites were located on the western half of Miyakejima to avoid high SO 2 gas concentrations to the east. Vegetation distribution in each site was visually observed as coverage by each species from understory, mid-height, to canopy level during June to August 2013 (Table 1; Kamijo and others, unpublished data). In the context of this study, the ten sites were converted to three types of vegetation according to vegetation composition of each site; grassland, shrubland, and forest. Sites G1, G2 and G3 were classified as grassland, because Miscanthus condensatus dominated. At site G1, vegetation was only scattered M. condensatus patches with absence of litter and organic accumulation. Sites G2 and G3 were dominated by the grasses M. condensatus and Fallopia japonica. Sites S1, S2 and S3 were classified as shrubland.

Net N mineralization rate measurement
In this study, annual N mineralization rate (ANMR) in the soil (expressed as kg N ha -1 yr -1 ) was defined as the net change in mass of inorganic N forms (NH 4 + -N, NO 3 --N and NO 2 --N) before and after a one-year incubation period from 28 Aug 2013 to 27 Aug 2014. The ANMR in the soil was estimated by the resin core technique [31,32]. The resin core is composed of three polyvinyl chloride (PVC) pipes containing the soil column and resin bags. Briefly, one of the PVC pipes (5.8 cm diameter, 5 cm length) was inserted into the ground to a depth of 5 cm to keep the soil core intact. Both sides of this long PVC pipe were connected to two short PVC pipes (5.8 cm diameter, 1 cm length) with vinyl tape. Resin bags containing 10 g of wet mixed H + and OHion exchange resin were placed tightly into both the top and bottom short PVC pipes. The two short PVC pipes were then covered with polypropylene nets and fastened by vinyl tape to the pipes to minimize disturbance. The upper resin bag traps external inorganic N infiltrating into the core from above, and the lower resin bag traps internal N transported downward from the soil core. For each site, nine replicate resin cores were set up, then reinserted back into the ground and covered with leaf on the surface for in situ field.

Soil properties
Soil temperature (ST) at 5 cm depth was monitored hourly from Aug 2013 to Aug 2014 by sensors (TidbiTv2 Temperature Data Logger, Onset Computer Corporation, Bourne, USA). While setting the resin core, additional three undisturbed soil samples were collected i.e., soil samples per site at 5 cm depth to determine soil organic matter content (SOM), soil total C (TC), total N (TN), and soil porosity (SP). SOM content was determined by loss on ignition with heating of the soil samples to 550°C for 4 h [33]. Total soil C and N contents were determined by a fully automatic CN analyzer (SUMIGRAPH NC-900, SCAS, Ltd, Tokyo, Japan). The air, water and solid volume were determined using a three-phase meter (DIK-1130, Daiki Rika Kogyo Co., Ltd, JPN). The SP is then calculated as the fraction of the air and water volume in total soil volume. Vegetation character in 2013, SO 2 concentration at 1 m height during 2011 to 2012, and volcanic ash depth in 2012 and 2013 were obtained by cooperative study (Kamijo and others, unpublished data).

Statistical analysis
The significance differences of soil ANMR among sites and ANMR, soil SOM, TC, TN and C/N ratio according to vegetation types were test by one-way ANOVA. Different groups were determined by the Duncan's multiple range tests, and similar groups were designated with the same letters and different groups with different letters. The significance of relationships between ANMR and environmental factors was tested by Pearson bivariate correlation analysis. Statistical analyses were performed with SPSS Statistics, version 22 (IBM Corp, Armonk, NY, USA).

Environmental variables
Total vegetation coverage tended to increase from grassland to forest, from G1 to F4 (Table 1). Among sites, SO 2 concentrations were more frequently greater than 1 ppm at G1 and G3 than at other sites; consequently, these two sites were more disturbed by the harmful gas ( Table 2). Annual mean soil temperature at 5 cm depth (ST) varied slightly among sites ( Table 2). The depth of volcanic ash (VAD) varied greatly from 0 cm at F3 and F4, far from the crater, to 43 cm at G2, near the crater. Soil properties, such as SOM, total C, total N, and soil C to N ratio (C/N) also varied greatly among sites (Table 2). SOM was highest at F3 and F4 (18.7% and 17.0%) and lowest at G1 and G2 (2.1, and 2.2%, respectively; Table 2). The patterns of total C and total N were similar to that of SOM. Soil C/N ratios were quite low at G1 and G2 (Table 2). Among the three vegetation types, SOM, TC, TN and C/N ratios gradually increased from grassland to forest ( Figure 2).

Soil net N mineralization
ANMR in the upper 5 cm of soil varied widely among the sites, from approximately 0.9 to 52.5 kg N ha -1 yr -1 (Figure 3). Among all sites, ANMR was significantly higher at S1, S2, and S3 where the N-fixing alder A. sieboldiana was present; ANMR was lowest at F1, where M. thunbergii dominated, but this lowest ANMR did not differ significantly from the other six sites. In addition to the high magnitude of ANMR at the shrubland sites, there was large variation of ANMR within each of these sites, as shown by Figure 3. Mean ANMR among three vegetation types, grassland, shrubland, and forest, ranged from 5 kg N ha -1 yr -1 in grassland to 47 kg N ha -1 yr -1 in shrubland. Mean ANMR was significantly higher in shrubland (Figure 4), but the values in the other two vegetation types did not differ significantly, despite very different environmental factors between the two vegetation stands.   Because of the data logger fault, we missed part of data in G2 and S1 sites; Data are mean with standard deviation (SD) in parentheses (n=3). TN, TC, C/N ratio, or the frequencies of high SO 2 concentrations, neither at the site level nor by vegetation types (Table 3). This lack of correlation suggests that rather than a single factor, complex factors regulated ANMR under field conditions on Miyakejima, or other environmental factors would be important for ANMR.

Soil N mineralization at site level on Miyakejima varies widely
Various vegetation with different soil properties are fundamental features of Miyakejima (Tables 1 and 2). In the latest eruption, a large amount of volcanic ash and emission harmful gas SO 2 disturbed the    ecosystems on the upper elevations of the island [34]. Hence, the landscape was reduced almost to bareland (such as at G1) near the crater, but forests far from the crater (such as at F3 and F4) were only minimally damaged; all the sites have been changed continuously. The study reported that the ANMR changed with vegetation and soil conditions at site level ( Figure 3 and Table 2). However, ANMR was significantly higher at only three sites (G1, G2, and G3), and the differences among the low values at the other seven sites were not significant ( Figure 3). These results indicate that although vegetation coverage and its composition are important controlling factors for ANMR, one-directional ecosystem development from grassland to forest could not sufficiently explain the spatial variation in ANMR on Miyakejima.
Considering the vegetation and soil properties of each site, it was hypothesised that the three sites i.e., S1, S2, and S3 are "hot-spots" for ANMR is because of the presence of N-fixing A. sieboldiana. The results of this study are consistent with those of previous studies which also reported that soil N mineralization beneath stands of such vegetation was high compared with that beneath stands of non-Nfixing vegetation [27,28,35]. Although the mechanisms of higher soil ANMR beneath N-fixing vegetation remain obscure, there are two possible mechanisms. First, N-fixing vegetation, such as alder, can increase input into the soil of easily available organic N obtained from atmospheric inactive N 2 . Second, the N-fixing vegetation considerably contribute to the supply of organic substrates that are easily decomposed and thereby decrease C/N ratios [36,37]. Thus, the N-fixing vegetation provides both accessible and plenty of resources for soil microbes, and facilitate N mineralization in the soil. Additionally, it was reported that there were quite low ANMR points even under the A. sieboldiana in the shrubland type. The actual range of ANMR from incubated samples from sites S1, S2 and S3 was 7.9 to 107, 14.3 to 67.8, 10.1 to 114.8 kg N ha -1 yr -1 , respectively. Such high spatial variation over small distances (cm level) suggests the importance of existence of litter or root nodules of A. sieboldiana due to uneven distribution of the litter fall and root nodules in both above-and under-ground parts. Further study will be needed on the impact of litter and root nodules from N-fixing vegetation, which directly affects ANMR through changes in available N and organic matter at the soil surface, in addition to the impact on ANMR of underground parts of the N-fixing vegetation and related microbes.
Many factors may contribute to the lower range of ANMR at the other seven sites, since soil and vegetation properties differed considerably among the sites. Among these seven sites, two sites (G1 and G2) would be bareland rather than grassland due to low vegetation coverage (Table 1) with low SOM, total C and total N ( Table 2). Thus, these two sites can be assumed as the early stage of succession. A number of studies have indicated that net N mineralization rates in early successional stages are low, probably due to scarcity of organic matter and lower substrate quantity [21,38]. Yoshitake et al. [39] demonstrated that soil microbial biomass in the Mt. Fuji volcanic desert in Japan was very low in the early successional stage and strongly correlated with soil total C, total N, and SOM contents. Thus, possible reasons for the very low ANMR at sites G1 and G2 are insufficient resources for N mineralization and lower soil microbial biomass than those at the other sites. But, the reason why ANMR was low in the G3 site with relatively high SOM was different from that in G1 and G2 sites. Although this study could not clarify the mechanisms of lower ANMR in the G3 site, high frequency of high SO 2 concentrations might account for the low ANMR at G3 (Table 2). Previous studies have demonstrated that exposure to high SO 2 concentrations could inhibit both ammonification [40] and nitrification in acid soil [41]. However, the direct effects of high SO 2 concentrations on N mineralization processes were not assessed in the present study; further studies of this are needed.
The other four sites, F1, F2, F3, and F4, including forest type can be regarded as mature ecosystems because of relatively high vegetation coverage, SOM, total C, and total N ( Table 2). Another common feature of the four sites was high C/N ratio (16.2, 16.0, 15.2, and 14.0 in F1, F2, F3 and F4, respectively), which may be the reason for the low measured ANMR. Several authors reported that substrate C/N ratio strongly influences N mineralization [14,42]. Booth et al. [43] also suggested that increase in C/N ratio decreases N mineralization. Among other controlling factors leading to the lower ANMR at F3 and F4, the large proportion of mature trees in the total vegetation would be important: the proportion of mature C. sieboldini, one of the climax species on the island, was more than half at F3 and F4 (Table 1). In general, large, mature trees contain a large fraction of recalcitrant substances, such as lignin, which decompose slowly [44]. Hence, in addition to the quantity of substrates for N mineralization, the quality and composition of each substrate may affect ANMR.

Soil N Mineralization at three vegetation types on Miyakejima
The sampling sites on Miyakejima were categorized by vegetation coverage and composition, and soil properties (Tables 1 and 2) can be regarded as three vegetation types. Mean ANMR among the three vegetation types on Miyakejima ranged from 3.0 to 47.0 kg N ha -1 yr -1 . With variation in ANMR at site level, the highest ANMR in the shrubland was found in which the N-fixing A. sieboldiana (alder) was the dominant species (Table 1). Since statistical tests show that there were no significant controlling factors (Table 3), the highest ANMR in shrubland could be due to the existence of A. sieboldiana. Many previous studies also reported differences in net N mineralization rate at various vegetation types, such as cropland, grassland, shrubland, and forest [27,28,45,46]. By chronosequences approach using unique sites, such as riverside terrace [47] and volcanic mountain side [48] where can speculate successional change of ecosystem structure, some of the previous studies mentioned about successional change of the net N mineralization rate [23,49]. In present study, if different vegetation type were assumed along with sequential vegetation types in succession then there will be more chance of predicting the long-term change of ANMR with succession, often initiated by volcanic eruption in Miyakejima. Since limited data and information are available at present, further studies will be needed to understand not only spatial but also temporal variation in ANMR related to succession, a common phenomenon, in Miyakejima.

Characteristics of soil N mineralization on Miyakejima
ANMR was reviewed in previous studies and this study (Table  4). Not only the magnitude of ANMR on Miyakejima, but also the range of ANMR are comparable to those from other studies in various ecosystems; also similar were the large spatial variation in ANMR at both plot and vegetation-scale. Similar to the Miyakejima results, ANMR in most stands of N-fixing vegetation are higher than those in stands of non-N-fixing vegetation (Table 4). In plots of ANMR versus environmental factors with data from both previous studies and this study, no significant controlling factors for ANMR were apparent ( Figure 5). ANMR tended to increase with annual mean temperature from sub-zero to 30°C, but decrease with annual mean precipitation from 0 to ca. 6200 mm yr -1 ( Figure 5). These relationships between N mineralization and temperature and moisture conditions are considered  to be common in various processes related to decomposition of organic matter and respiration [50][51][52]. As for soil N mineralization rate, some studies reported that N mineralization rate tends to decrease under conditions of excess soil moisture [53,54]. Thus, the relatively lower ANMR values on Miyakejima except for that from shrubland were probably due to the high precipitation, ca. more than 3000 mm yr -1 .
Soil C/N ratio is commonly considered to be a good indicator of N mineralization [55]. Higher soil C/N ratios are generally thought to hinder mineralization [56]. When the C/N ratio of substrate is low (12-20:1) decomposers are not N limited, and a net release of inorganic N to the soil solution occurs [57]. The majority of soil C/N ratios that were measured on Miyakejima, are within this range and often N was mineralized in these soils. The relationship between ANMR and soil C/N ratios would show an appealing single-peak curve, which is ANMR was also low under low soil C/N ratios less than ca. 10. This is different from the general understanding that low soil C/N ratios promote N mineralization [58]. The mechanisms that keep low N mineralization under low soil C/N ratios remain unclear, even though several factors seem to be involved, such as microbial properties [59,60], vegetation properties [61][62][63][64][65][66][67][68][69], and extraordinary disturbance like high SO 2 gas exposure in volcanic area of current study.
In-situ data demonstrated that ANMR in the Miyakejima, reported the huge spatial variation and suggest that the spatial variation could not be explained by single environmental factor though, could be partly explained by some unique factors in the volcanic island, such as existence of N-fixing and high SO 2 gas exposure.

Conclusions
In this study soil N mineralization rate were in-situ at a series of sites that were disturbed to various extents by the latest 2000 eruption on the volcanic island Miyakejima. ANMR varied greatly among sites, from about 0.9 to 52.5 kg N ha -1 yr -1 . The highest rates were at sites where the N-fixing vegetation A. sieboldiana was present as the dominant species. Low ANMR was observed in grassland and forest-vegetation type; the former would be due to lower SOM, high SO 2 concentration, or both and the latter would be due to relatively high C/N ratio. Compared to previously published studies of other locales, the relatively lower ANMR were observed on Miyakejima (except for those at shrubland)