Wildfire effects on ecosystem nitrogen cycling in a Chinese boreal larch forest , revealed by 15 N natural abundance

Wildfire is reported to exert strong influences on N cycling, particularly during the early succession period immediately after burning (i.e., < 1 year). Previous studies have mainly focused on 10 wildfires influences on inorganic N concentrations and N mineralization rates; but plant and soil N natural abundance (expressed by δN), as a spatial-temporal integrator of ecosystem N cycling, could provide a more comprehensive understanding of wildfire on various N cycling processes at a relatively broader time scale. In this study, we attempted to evaluate legacy effects of wildfire on nitrogen cycling using δN in a boreal forest of northeastern China, which is an important yet understudied ecosystem. 15 We measured inorganic N concentrations (NH4 + and NO3 ) and net N transformation rates (net ammonification, net nitrification, and net mineralization) of organic and mineral soil 4 years after a Biogeosciences Discuss., doi:10.5194/bg-2016-91, 2016 Manuscript under review for journal Biogeosciences Published: 4 April 2016 c © Author(s) 2016. CC-BY 3.0 License.

wildfires influences on inorganic N concentrations and N mineralization rates; but plant and soil 15 N natural abundance (expressed by δ 15 N), as a spatial-temporal integrator of ecosystem N cycling, could provide a more comprehensive understanding of wildfire on various N cycling processes at a relatively broader time scale. In this study, we attempted to evaluate legacy effects of wildfire on nitrogen cycling using δ 15 N in a boreal forest of northeastern China, which is an important yet understudied ecosystem. 15 We measured inorganic N concentrations  and net N transformation rates (net ammonification, net nitrification, and net mineralization) of organic and mineral soil 4 years after a wildfire and compared with unburned area. We also measured δ 15 N of plant and soil samples in 4 and 5 years after the fire. We found that even 4 years after burning, net mineralization and net ammonification in the organic soil were still higher than those in the unburned area. NH 4 + and total inorganic N (TIN) 20 concentrations in the organic soil of the burned area did not significantly differ from those of the unburned area. Organic soil and foliar δ 15 N were significantly higher (by 2. 2‰ and 7.4‰, respectively) in the burned area than those in the unburned area. Five years after fire, δ 15 N of plant tissues such as foliar, branch, fine roots and moss in the burned area were significantly greater (by 1.7‰ to 6.4‰) than that in unburned area. δ 15 N of Oi, Oa+e and 0-10 cm mineral soil were also significantly higher in the 25 burned area than unburned area, but showed no significant difference in deeper layer of mineral soil.
The observed soil 15 N enrichment might be attributed to various mechanisms such as NH 3 volatilization, combustion, litter return, and denitrification. Greater dependence of plant on deeper soil N and less dependence on mycorrhizal fungi in the burned area might also have contributed to the increase of the 15 N in plant and soil. Such 15 N enrichments in soil and plant suggest that N cycling could remain 30 openness years after fire disturbance has occurred, with N supply exceeding demand, leading to a great amount of nitrogen loss from the system for a relatively long time.

Introduction
Wildfire-induced nitrogen (N) cycling changes can greatly alter ecosystem structure and functions, such as species composition and biodiversity (Gallant et al., 2003), biogeochemical cycles and productivity 35 (Boerner, 1982;Chorover et al., 1994;Woodmansee and Wallach, 1981), as N is likely to be the most essential element limiting plant growth in terrestrial ecosystems (Elser et al. 2007, Harpole et al. 2011Lebauer and Treseder, 2008;Vitousek and Howarth 1991). In response to recent global climate changes, wildfire frequency and extent in temperate and boreal forests are projected to be enhanced (Flannigan et al., 2009;Lucash et al., 2014;Westerling et al., 2006). Therefore, a better understanding of wildfire 40 effects on N dynamics is of growing importance. Many studies have attempted to examine the effects of fire on N cycles through the analysis of available N concentrations and N mineralization rates (e.g., Turner et al., 2007;Koyama et al., 2010;Deluca et al., 2006). However, two principal limitations have greatly challenged this objective. First, these two indices vary significantly in space and time (Cain et al., 1999;Hu et al., 2013). Second, soil available N concentrations and N mineralization rates only 45 account for a fraction of N cycling processes. Other N-related processes such as denitrification and leaching are also important in forest ecosystems , yet these processes are difficult to measure directly, thus constraining our ability to generalize the response of N cycle to wildfire.
The natural abundance of 15 N/ 14 N of plant and soil is considered as a more time-and space-integrator of N cycling than available N concentration and N mineralization rate, and could reflect the openness of 50 ecosystem N cycling (Robinson, 2001). Soil processes, such as N mineralization, nitrification, denitrification and NH 3 volatilization, discriminate against 15 N and lead to soil N pool with different δ 15 N signatures (Craine et al., 20015), which further express in the values of plant δ 15 N that utilize these N pools for their N demands. Higher values of 15 N in soil and plant generally indicate larger N losses through ammonia volatilization, nitrification or denitrification (Craine et al., 2009;Houle et al., 2014;55 Matsushima et al., 2012). Thus, the ecosystems with a more open N cycle tend to be isotopically enriched in 15 N (Martinelli et al., 1999). In fact, although the responses of available N concentration and N mineralization to wildfire could vary by time and space, we can expect higher values of δ 15 N in plant and soil as a legacy of longer or short term opening in the N cycles when available N supply exceeds demand, resulting in an increase in N loss. Therefore, δ 15 N could provide us a promising and 60 comprehensive tool to detect the legacy effect of wildfire on N cycling openness years after the fire disturbance has occurred.
Boreal forests are typical N limited ecosystems where N cycling is slow and a large proportion of total N capital is tied up in undecomposed organic matter (Hyodo et al., 2013;Metcalfe et al., 2013;Popova et al., 2013;Sah et al., 2006). This is consistent with low δ 15 N values of soil and plant (Hogberg, 65 1997). Wildfire is a primary agent of disturbance in boreal forests and has a profound impact on N cycling (Baird et al., 1999;Bond-Lamberty et al., 2006). Wildfires consume N from vegetation and the upper surface soil layer, resulting in a reduction of N storage in burned forest (Hogberg, 1997;Hyodo et al., 2013). Estimates of post-fire inorganic N concentrations and mineralization rates vary, but most studies show an immediate increase in inorganic nitrogen (Deluca and Sala, 2006;Koyama et al., 2012;70 Turner et al., 2007). Nevertheless, the immediately increased NH 4 + can decline to the pre-fire level within the first year and the elevated NO 3 generally returned to pre-fire level within 5 years (Wan et al. 2001). Studies have also shown a wildfire-induced pulse in foliar δ 15 N (LeDuc et al., 2013). However, this wildfire-induced pulse in foliar δ 15 N is shown short-lived followed by a rapid decline after the first few years (LeDuc et al., 2013;Szpak, 2014). Compared to much attention paid to these changes in N 75 dynamics at the 0-3 years after a fire event in boreal forest (Baird et al., 1999;DeLuca and Zouhar, 2000;White, 1986), little is known about the legacy effects of fires on N cycling over a long time period following this pulse, particularly through the use of natural abundance of N isotopes to detect the responsible processes.
The Great Xing'an Mountains of northeast China are located on the southern extension of the larch 80 forests of the eastern Siberia. It has been reported that the boreal forests in Northeast China stored 1.0 -1.5 Pg C and provided approximately 24 -31% of the total timber production in China (Fang et al., 2001). This region experiences frequent wildfires with historical fire return intervals of 30-120 years (Xu et al., 1997). Despite the knowledge that fires can significantly influence soil N dynamics elsewhere, the understanding on the influence of post-fire N cycling in this region is limited. Previous 6 studies have investigated the responses of soil inorganic N concentration and N mineralization to wildfire in this region (Kong et al., 2015), and showed that a pulse of inorganic nitrogen concentration appeared in one year post-fire. However, there are few studies of response of soil inorganic N and mineralization to fire in this boreal larch forest over a relatively longer post-fire time period, and even less regarding the fire effects on plant and soil δ 15 N. 90 In this study, we compared the N status of burned area (4 and 5 years after a large wildfire) with unburned area through comprehensive analysis of δ 15 N for foliage and soil, inorganic nitrogen concentrations and N mineralization rates. Our overarching goal was to determine the legacy effect of the wildfire on N cycling in boreal forest of Greater Xing'an Mountains and explore the underlying mechanism. Specifically, we expected that: 95 1) The inorganic nitrogen concentration and N mineralization rate in the burned area would be similar with the unburned area since most studies have shown that available N concentrations and mineralization rates declined to the pre-fire level within 5 years after fire; 2) Soil δ 15 N would be higher in the burned area than unburned area, especially in the organic soil, because live biomass in recently burned area are still largely reduced, which would result in a larger 100 inorganic nitrogen loss than its demand; 3) Plants in the burned area would be more enriched in 15 N than unburned forest because plants utilized the N resources that were 15 N-enriched due to N losses.

105
Our study area was in Huzhong National Natural Reserve (51°17′42″ N to 51°56′31″ N, 122°42′ 14″ E to 123°18′05″ E), which is located in the Great Xing'an Mountains of northeastern China ( Fig. 1). This reserve encompassed 167,213 ha and experienced a terrestrial monsoon climate, characterized by a long and severe winder. The average annual temperature was -4.7 ℃ and mean annual precipitation was ca 500 mm. More than 60% of the annual precipitation fallen in the summer 110 season from June to August (Liu et al. 2012;Zhou, 1991 The historical fire regime in this region is described as frequent, surface fires mixed with infrequent, 120 stand-replacing crown fires, with fire-free interval ranged from 30 to 120 years (Xu et al., 1997;Liu et al., 2012). However, climate change, forest management and human activities have altered fire regimes in this region (Jackson et al., 1997;Wang et al., 2007). Although the dominant tree species Dahurian larch is regarded as a fire-tolerant species with thick bark near the stem bottom, its post-fire mortality rate is still high, mainly due to a horizontal shallow-distributed root system  125 Vijayakumar et al., 2016). A stand-replacing wildfire，which was ignited by lighting, burned 600 ha of Huzhong National Natural Reserve on June 26th, 2010. This fire provided an ideal opportunity to study the effects of fire on soil N dynamics in this ecosystem.

Experimental design and field sampling
In early June 2014, we randomly selected 12 plots (each 10 m ×10 m) in the burned area with six plots 130 at northern and southern slopes, respectively. In the unburned area, we also randomly set six plots (each 20 m × 20 m, Fig. 1) with three plots at each slope (northern and southern slope). To mitigate edge effects, we located these plots at least 100 m away from the roads. In addition, each plot was 200 m away from each other in order to minimize samples' spatial autocorrelation.
Soil samples were taken from two layers (organic layer and 0-20 cm mineral layer) at five random 135 locations within each plot and were composited. We also recorded the temperature of organic layer by soil thermometer at the soil depth of 5 cm (whenever applicable). The soil temperature was measured between 10am and 4pm. To account for the inherent hourly and daily temperature variations, we also measured soil temperatures at two fixed places at the hourly basis and used them as the baseline temperature data to correct such sources of uncertainty. The corrected values would be used to compare 140 the difference in mean soil temperature between burned and unburned areas. On the same day the soil samples were collected, 7.5 g fresh organic soil passed through 5 mm sieve and 30 g fresh mineral soil passed through 2 mm sieve were extracted by 75 ml 2 M KCl solution, shaking for 1 h at 160 r/m and then filtered. The extracts were frozen and maintained at -20 ℃ until later laboratory analysis. For each plant species, the foliage was sampled from at least 5 separate individuals within the same plot.

Laboratory treatment and chemical analysis
Subsamples of organic and mineral soils were air-dried, crushed and sieved through 5 mm and 2 mm mesh, respectively, for chemical analysis. Soil pH was measured in H 2 O employing a soil: solution ratio 160 of 1:10. Soil samples were dried at 105 ℃ for 48 h to measure soil water content. Ammonium in the extract was determined by the indophenol blue method followed by colorimetry, and NO 3 was determined colorimely using the same autoanalyser in the form of NO 2 after reduction of NO 3 in a Cd-Cu column followed by the reaction of NO 2 with N-1-napthylethylenediamine to produce a chromophore (Rivas et al., 2012). where R sample represents the isotope ratio ( 15 N/ 14 N) of sample and R standard is the 15 N/ 14 N for atmospheric N 2 . The analytical precision for δ 15 N was in general better than 0.2‰. 175 In order to examine the net N mineralization rate, we collected soil from the same plots in early August of 2014 with the same sampling method. We used the soil samples collected in the late growing season for this purpose because we didn't have low-temperature sample transportation facilities during the first-round soil collection in early June. Net N mineralization rates were estimated using laboratory soil incubations. 7.5 g fresh organic soil passed through 5 mm sieve and 30 g fresh mineral soil passed 180 through 2 mm sieve were put into a plastic cups with polyethylene film to minimize moisture evaporation and incubated at 20 ℃ for 1 week without light. Incubated soil mineral N (NH 4 + and NO 3 -) was extracted and measured as above mentioned. On the same day, the soil samples were extracted by 75 ml 2 M KCl solution by the same method as used in the June. The extracts were frozen and maintained at -20 ℃ until later laboratory analysis. Net N mineralization potentials were calculated as 185 the difference between final and initial inorganic N (NH 4 + + NO 3 -) concentrations divided by the number of incubation days. The expression "N mineralization potential" is used to designate soil samples that produced net amounts of inorganic N.

Statistical analysis
We used one-way analyses of variance (ANOVA) to test whether wildfire significantly affected soil N 190 availability and examine the differences of δ 15 N (‰) of foliage, organic soil and mineral soil in burned and unburned area. Significance level was set at a P value of 0.05 unless otherwise stated. Significant differences among treatment means of soil properties were analyzed using One-way ANOVA. Data were statistically analyzed in R (R Core Team, 2014).

Basic soil properties
The alteration of soil basic properties in the burned area 4 years after the wildfire was mainly found in the organic soil, not in the mineral soil (Table 1). The soil water content (SWC), total nitrogen (TN), total carbon (TC), C:N were lower in the burned soil, but only the reduction of SWC and TC reached the significant level (p<= 0.05). Mean soil water content at the organic layer was significantly lower in 200 the burned area when compared to the unburned area (41.2% vs. 117.6%). Mean TC at the organic layer in the burned area was 9.2%, which was significantly lower than that in the unburned area (29.2%). In contrast of those properties that were reduced after fire, pH and temperature were increased. The mean organic soil temperature in the burned area was 10.0 o C and was significantly higher than that in the unburned area (2.9 o C). 205

Soil inorganic nitrogen concentrations
At the beginning of growing season (early June), total inorganic N pools were greater in the organic soil than the mineral soil both in burned and unburned area ( Fig. 2A). However, the significant increases in soil inorganic N concentrations in response to wildfire were only observed in the mineral soil. Mean total inorganic N concentration in the mineral soil in the burned area (5.55 mg N· kg -1 ) was significantly 210 higher than that (2.22 mg N· kg -1 ) in the unburned area. Compared to unburned area (1.6 mg N· kg -1 ), the amount of NH 4 + was significantly higher (5.0 mg N· kg -1 ) in mineral soil of burned area (Fig. 2B). NO 3 concentrations were consistently low in both organic and mineral soil, and had no difference between burned and unburned area (Fig. 2C).
At the end of growing season (early August), there were no significant differences in total inorganic 215 N, ammonium and nitrate concentrations between burned and unburned area (Figs. 2D and 2F).
However, the significant decreases in soil inorganic N concentrations in response to wildfire were only observed in the mineral soil. Mean total inorganic N and ammonium concentrations in the organic soil were 5.86 and 4.27 mg N· kg -1 in the burned area, which were significantly lower than those (12.07 and 10.35 mg N· kg -1 , respectively) in the unburned area (Figs. 2D and 2E). 220

Nitrogen transformation rates
The response pattern of N transformation after the fire was similar to that of soil inorganic N concentrations. Both mean net mineralization and ammonification rates (0.056 mg N kg -1 d -1 and 0.029 mg N kg -1 d -1 ) were significantly higher in the organic soil of the burned area compared to the unburned area of -0.653 mg N kg -1 d -1 and -0.579 mg N kg -1 d -1 , respectively (Figs. 3A and 3B). In contrast, 225 ANOVA test revealed no significant differences on mineralization and ammonification rates between burned and unburned mineral soil. There was no significant difference in net nitrification either in the organic or in the mineral soil between burned and unburned soil (Fig. 3C).

Plant and soil δ 15 N
Mean foliar δ 15 N in the burned sites was 3.7‰ and was significantly higher compared to the mean foliar and -2.4‰, respectively (Table 2). A significant difference of δ 15 N between the burned (3.6‰) and unburned (1.3‰) area was also found in the organic soil (Appendix 1). However, there was no significant difference in the mineral soil between burned (4.9‰) and unburned (4.8‰) area (Appendix 1).
The effects of wildfire on δ 15 N were also detected in plants' aboveground parts in the burned area 5 240 years after the fire (Fig. 4). Mean foliar and branch δ 15 N were 2.3‰ and 1.5‰, respectively, in the burned area, and were significantly (p<0.001) greater than those in the unburned area (both -4.1‰). The moss δ 15 N ranged from 0.9‰ to 1.7‰, and the mean was 0.7‰, which was significantly (p<0.001) higher than the moss collected in the unburned area. Fine root δ 15 N was 0.7‰ in burned area and was significantly (p<0.001) higher than that in unburned area (-0.9‰). As the various sub soil organic layers, 245 the Oi was more depleted in 15 N (-3.6‰) in unburned area than in burned area (-2.4‰). The wildfire also significantly increased the δ 15 N of Oa+e and 0-10 cm mineral soil, but had no significant effects on the deeper mineral soil layers. 250 We initially expected that the inorganic N concentrations and N mineralization rates in the burned area would have recovered to the pre-fire level 4 years after fire. However, our data didn't fully support this hypothesis. In contrast, our study showed that wildfire still had a strong effect on inorganic nitrogen concentrations and N mineralization rates. Specifically, we observed higher than the control level NH 4 + and TIN concentrations in the mineral soil in June, and higher net mineralization and ammonification 255 rates in the organic soil in August. Although the N mineralization rates were high in the organic soil of the burned area, we found that TIN and NH 4 + concentrations in the organic soil of the burned area were significantly lower than those in the unburned area. Such lower amount of TIN and NH 4 + may be due to plant uptake or N loss through gases and leaching. Large decreases in TIN and NH 4 + concentrations in the mineral soil of burned area after a growing season (from June to August) might be contributed to 260 plant uptake as the fine roots were mainly distributed in the mineral soil after fire (Appendix 2).

The effects of fire on soil inorganic nitrogen concentrations and transformation rates
The significantly increased rates of net mineralization and net ammonification in organic soil after fire might be attributed to following three reasons: (1) increased available organic matter to microbes (Dannenmann et al., 2011), as shown in our results that C:N ratio decreased from 27.5 to 20.8 (Table 1), may enhance microbial activities to decompose litter.
(2) post-fire abiotic environments such as 265 increased soil pH resulted from increased base cation availability and temperature (Table 1) tend to be more suitable for microbial activities (Smithwick et al., 2005). Increased temperature might have played a key role in N transformation (Klopatek et al.,1990) because decomposition rates may increase by 50% -100% when soil temperatures increase 5 o C -10 o C (Richter et al., 2000). In this study, organic soil temperature was increased (by 7.1 o C) significantly after fire, mainly due to combustion of thick organic 270 layer (the thickness was decreased from 22.2 cm to 5.3 cm) and the removal of overstory tree, leading to more solar radiation reaching ground surface (Christensen and Muller, 1975). On the contrary, net nitrification rate remained unchanged after fire, despite increased soil net ammonification (Fig. 3). This could be because nitrifier population size may be too low after fire and it may take some time to increase (Turner et al., 2007). In addition, fire might have an adverse effect on nitrifying microbes, as 275 some previous studies have suggested nitrifiers are more susceptible to fire than other soil microbial groups (Hart et al., 2005).
Other studies have also observed increase in inorganic N after fire (Certini, 2005;Gomez-Rey and Gonzalez-Prieto, 2013;Koyama et al., 2012). Turner (2007) studied inorganic N pools and mineralization rates in the first 3 years after a stand-replacing wildfire in the Greater Yellowstone 280 ecosystems and found that soil NH 4 + concentration increased and followed by increases in soil NO 3 -, but fire had a net negative influence on N mineralization due to microbe immobilization. Koyama et al. (2010) found soil NO 3 concentrations elevated in the 2 years after wildfire in the coniferous forests of central Idaho resulted from reduced microbial NO 3 uptake capacity, but NH 4 + concentrations between the treatments were not significantly different. They also suggested that reduced available C was the key 285 factor regulating soil N cycling after fire. On the contrary, Deluca and Sala (2006) showed recurrent, low-severity fire had a different effect on N in ponderosa pine forests. In their study, post-fire soil total N concentrations and potential mineral N (PMN) rates using the 14-day anaerobic incubation procedure decreased, and the concentrations of NH 4 + and NO 3 were not in line with the changes of total N pool and PMN rate. These studies collectively showed fire severity, time after fire, vegetation type and soil 290 sampling depth may be responsible for the inconsistency of the reported findings (Wan et al., 2001;Wang et al., 2014).

The effect of fire on soil δ 15 N
Our results showed that 15 N natural abundance in organic soil was significantly higher in the burned area than unburned area (Fig. 4). These results are consistent with our expectation that soil δ 15 N would 295 be higher in the burned area than unburned area, especially in the organic soil. Similar results were reported in other forest ecosystems (LeDuc et al., 2013;Schafer and Mack, 2010). Combustion of the upper δ 15 N-depleted surface soil layer and enhanced nitrification are the two widely-recognized mechanisms to explain 15 N enrichment in organic soil (Hogberg, 1997;LeDuc et al., 2013;Schafer and Mack, 2010;Szpak, 2014). However, other mechanism such as NH 3 volatilization, combustion, litter 300 return, nitrate leaching, denitrification can also contribute to the observed 15 N enrichment in our study.
Our results showed higher net mineralization and net ammonification didn't lead to higher ammonium concentrations in burned organic soils (Figs. 2-3). On the contrary, the ammonium and total inorganic nitrogen in the organic soil of the burned area were significantly lower than those in the unburned area (Fig. 2). Such lower NH 4 + and TIN concentrations in the burned soil were likely due to 305 NH 3 volatilization --although several other mechanisms such as surface run-off and filtration to mineral soil might also contribute to this observed pattern. Higher soil temperature and pH values in the burned area as observed in the present study could enhance NH 3 volatilization (Nelson and Conrad, 1982;Raison, 1979). NH 3 volatilization is associated with strong fractionation against 15 N and higher gaseous losses of 15 N-depleted NH 3 , and leads to the remaining soil NH 4 + to be enriched in 15 N (Hobbie and 310 Ouimette, 2009). Therefore, fire-stimulated NH 3 volatilization, associated with strong isotopic fractionation and subsequent export of 15 N-deplted NO 3 -, is considered as one likely being responsible for 15 N enrichment of organic soil.
Combustion of surface soil layer could also cause the upper soil to be enriched in 15 N since high, sustained fire temperatures cause a greater loss of 14 N compared to 15 N (Huber et al., 2013;Schafer and 315 Mack, 2010). In our study, wildfire, characterized with high temperature, combusted the thick organic layer, leading to a significant higher δ 15 N in the burned organic soil. For the mineral soil, the significant higher δ 15 N was only observed in 0-10cm mineral soil but not in deeper mineral soil. This pattern may have resulted from the insulation of underlying mineral soil from heating and limited downward conduction of heat from soil surface to deep soil (Smithwick, et al., 2005). litter with a higher δ 15 N value, which contribute to a relatively lower δ 15 N in Oi than that in leaf in the burned area.  reported that denitrification was an important N loss pathway and could account for 48% to 86% total NO 3 loss in forest ecosystems. Although we didn't measure this N process directly, we also considered denitrification as a potential mechanism for the higher δ 15 N in the organic 330 soil. On one hand, lower plant and microbial biomass in the burned area would result in lower N need and more NO 3 loss through denitrification. On the other hand, the lack of increase in net nitrification of the burned soil resulted from 7-day laboratory incubation might be due to an enhanced denitrification, which is associated with strong fractionation against 15 N and higher gaseous losses of 15 N -depleted N 2 or N 2 O, remaining soil NO 3 to be enriched in 15 N (Hobbie and Ouimette, 2009;Robinson, 2001).  (Hogberg, 1997;Sah et al., 2006). This assumption is supported by our field experiment in 2015. We found the root was significantly lower in the organic layer and fine roots were mainly distributed in the 0-20 cm mineral soil in the burned area; while in the unburned area, fine roots were mainly distributed in the organic soil layer (Appendix 2). Secondly, part of 15 N-enriched NH 4 + and NO 3 infiltrated into the 345 deeper mineral soil with rainfall from the organic layer, which leads to the remaining soil N pool to be enriched in 15 N and further expressed in the values of δ 15 N in plant that utilized these N pools for their N demand. Thirdly, increased N availability could lead to a lower dependence of plant N nutrition upon mycorrhizal fungi, which provide their host plants with 15 N-depleted N relative to the soil N sources (Craine et al., 2009;Hobbie et al., 2008). Boreal forest is a typical N-limited ecosystem and plants 350 usually associated with mycorrhizal fungi to meet their N demand (Craine et al., 2009;Hobbie et al., 2008;Nasholm et al., 2013). Larch is the dominant species in the unburned area and is often associated with ECM. Ledum spp. and Vaccinium spp. are the main understory species and are often associated with ERM (Michelsen et al., 1998). Numerous studies have suggested that mycorrhizal fungi preferentially transfer isotopically depleted nitrogen to their host plants (Hobbie and Agerer, 2010;355 Högberg 1997; Whiteside et al., 2012   Different letters indicate significant difference between burned and unburned plots. circles represent burned plots, solid black squares represent unburned plots, respectively. Data show means ± standard error. One asterisk indicate significant difference among forests at p ≤ 0.05, two asterisks indicate significant difference among forests at p ≤ 0.01, three asterisks indicate significant difference among forests at p ≤ 0.001.