1 Introduction

Forest ecosystems are predominantly sinks for atmospheric CO2 (Gower et al. 2001; Trumbore et al. 2015). However, disturbances caused by climatic extremes and human activities (e.g., wildfire, storm, drought, prescribed burning, nitrogen addition, and land use management) can significantly and rapidly affect the C cycle of forest ecosystems and its feedback to the atmospheric system (Fang et al. 2018; Frank et al. 2015; Hu et al. 2017b; Plaza-Álvarez et al. 2017; Sheng et al. 2009). In recent years, extreme climate events, such as intensified El Niño and drought, which are associated with climate warming, have meant that the frequency, severity, and burn area of wildfires are expected to increase in the near future (Abatzoglou and Williams 2016; Santin et al. 2016; Seidl et al. 2017). Quantifying the effect of forest fire disturbance on the C dynamics of forest ecosystems is a key component for lowering the uncertainties associated with C sink estimates (Kasischke and Stocks 2000; Liu et al. 2014; Schimel and Baker 2002).

Soil respiration (Rs) is the sum of soil autotrophic respiration (Ra) (from root systems and root-associated microorganisms) and soil heterotrophic respiration (Rh) (from the decomposition of organic material by free-living microbes) (Chen et al. 2016b; Davidson and Janssens 2006; Luo and Zhou 2006). Soil respiration is the second largest C efflux (80–98 Pg C·yr−1) in terrestrial ecosystems and the global Rh from soils has been estimated at 53–57 Pg C·yr−1 (Bond-Lamberty and Thomson 2010). A recent study observed that the soil surface Rh:Rs ratio significantly increased from 0.54 to 0.63 between 1990 and 2014 due to environmental change (Bond-Lamberty et al. 2018). Functionally, disturbance (e.g., wildfires) can have significant effects on soil respiration components such as Rh and while the control of soil respiration components by some environmental factors has been identified, how each environmental factor interacts with disturbance remains an open question (Harmon et al. 2011).

Many studies have focused on forest soil respiration during the growing season (Chen et al. 2016a; Decina et al. 2016; Pries et al. 2016; Zhou et al. 2016) and have estimated the annual soil respiration by assuming that the respiration efflux is near zero during the non-growing season (Fahnestock et al. 1998). However, other studies have shown that the non-growing season soil respiration could account for 2–37% of annual soil respiration and that it significantly affects the C balance of forest ecosystems (Brooks et al. 2005; Wang et al. 2006; Wang et al. 2013; Wang et al. 2014b). Forest fire disturbance can alter both the input and output of organic C stored in the soil. Thus, a remaining challenge is to understand the variation and dominating environmental factors of the non-growing season Rs and its components (Rh and Ra) after fire disturbance (Song et al. 2018). In particular, the response of Rh to forest fire might be the key factor influencing the amount of net C stored in forest ecosystems.

The degree of influence of fire on the soil C pool depends on fire severity and duration (Marañón-Jiménez et al. 2011). Soil respiration and its components are potentially controlled by the variation of soil environmental factors after fire disturbance (Pereira et al. 2016). For instance, forest fires can affect soil respiration by decreasing vegetation cover and increasing albedo, which can increase soil temperatures and litter decomposition rates (Jiang et al. 2015; Throop et al. 2017). Additionally, fire can increase soil hydrophobicity, which may indirectly control the components of soil respiration by reducing soil moisture infiltration and increasing surface runoff (O’Donnell et al. 2009). Previous studies have indicated that soil temperature and moisture are the dominant environmental factors of soil respiration variation during the growing season (Raich and Schlesinger 1992; Yi et al. 2020).

The northern hemisphere has a greater area of winter snow compared to the southern hemisphere, and snow in the north is much more vulnerable to climate change (Cohen and Entekhabi 1999; Danco et al. 2016). Snow creates an insulating layer that might increase soil temperature, and soil temperature and moisture changes after disturbance have a strong effect on the snow-cover depth (Groisman et al. 1994; Uchida et al. 2005). The interaction effect between environmental factors can change biological and chemical processes such as microbial decomposition, enzyme and rhizosphere organism activity (Monson et al. 2006; Tucker et al. 2014). However, few studies have focused on the effect of environmental factors on the components of the soil respiration during the non-growing season, leading to great uncertainty about the variation in soil respiration during the non-growing season (Barba et al. 2018; Hibbard et al. 2005; Rustad et al. 2001).

In the present study, we used a Quercus mongolica forest immediately after fire disturbance. The objectives of the study were to determine how wildfires affect the non-growing season components of soil respiration and to determine the dominating environmental factors that drive variation in soil respiration after fire disturbance. We postulated that the components of soil respiration are potentially controlled by interactions among soil environmental factors such as soil temperature, soil moisture and snow depth that are altered by a fire event. In a previous study, a forest fire was found to significantly decrease Ra (Hu et al. 2017b); thus, we hypothesized that the non-growing season total Rs was dominated by soil Rh after fire disturbance. As snowpack is an important environmental factor during the non-growing season and is vulnerable to temperature change, we further postulated that snowpack depth may be related to fire disturbance and that it will influence the components of the non-growing season soil respiration (Rh and Ra).

2 Materials and methods

2.1 Study area

The present study was conducted at the Maoershan Forest Ecosystem Research Station, northeast China (45°20′–45°25′ N, 27°30′–127°34′ E, 400 m above sea level). The parent material is granite bedrock, and the soil is classified as a Haplumbrepts in the United States Soil Taxonomy (Soil Survey Staff 2014). The climate is continental monsoon with a dry and cold winter. The annual total precipitation varies from 600 mm to 800 mm, of which ~50% falls between June and August (summer dominated). The mean annual, maximum, and minimum air temperatures are 2.7 °C, 18.0 °C, and −12.1 °C, respectively. During the sampling years (2017–2018), the maximum and minimum air temperatures were 32.5 °C and −31.2 °C, respectively. Snowpack lasted for 154 days, with the snowpack depth varying from 0 to 31.2 cm, with a mean depth of 14.1 cm. The dominate tree species of our research stand is Q. mongolica (> 80% total basal area) and mixed with naturally regenerated tree species that include Betula platyphylla and Populus davidiana. The dominant herb species during the study period were Anisodus acutangulus, Adenocaulon himalaicum, Dryopteris crassirhizoma, and Aegopodium alpestre.

2.2 Site description

In the present study, the non-growing season experimental period was from mid-November 2017 to mid-April 2018 and was approximately 150 days. The definition of the non-growing season follows that of previous phenological studies (Piao et al. 2007; Xu et al. 2017), meta-analysis of the winter ecosystem (Wang et al. 2011) and C flux research of temperate Korean Pine (Pinus koraiensis Sieb. et Zucc.) in the Maoershan area (Wang et al. 2013). The first span of at least 5 days with daily mean air temperatures below 5 °C was defined as the start of the non-growing season. Similarly, the first span of 5 days with daily mean air temperatures above 5 °C was defined as the end of the non-growing season. The freeze-thaw cycle (FTC) period in spring was defined as 5 cm of soil above 0 °C (i.e., the start of the snowmelt), to the end of the non-growing season (i.e., the snow completely melted). The non-growing season included the snow-cover winter period and the FTC period (Table 1).

Table 1 Timing of the non-growing season, winter period, and spring freeze-thaw cycle (FTC) period
Full size table

In April 2016, forest fires caused by lightning occurred at Maoershan Forest Ecosystem Research Station, northeast China. The total area burned was approximately 500 ha and provided an opportunity for us to study the effects of fire disturbance on soil respiration and its components. The burn severity was moderate in the burned area; severity was determined by the depth of the burned organic soil, the consumption of the aboveground biomass, tree mortality, and the bark char height (Keeley 2009). In the burned area, approximately 50% of the understory shrubs were burned, the bark char height was 1.8–2.4 m and tree mortality was approximately 36%. We selected three replicate stands in the burned area to conduct our investigation and selected nearby unburned areas as the control stands. The size of each stand was 400 m2 (20 m × 20 m) and all stands were established in April 2017. The specific information of the stands and soil characteristic is shown in Table 2. The leaves of Q. mongolica were persistent; a large number of leaves did not fall off the trees and instead existed in the canopy even in the non-growing season, which led to the difference in canopy coverage between the control and the burned stands in the non-growing season.

Table 2 Basic information of research stands. Values are the means (±standard deviation) of three replicates with repeat measurements (n = 3)
Full size table

3 Soil CO2 efflux and its environmental factors

Soil CO2 effluxes (including Rh and Ra) were measured using an Li-8100 portable automatic measuring system for soil C flux (Li-Cor, Inc.; Lincoln, NE, USA). Five polyvinylchloride (PVC) soil rings (internal diameter 19 cm and height 7 cm) for measuring Rs were randomly placed in each stand. A trenching approach was used to separate Rh and Ra. Three trenched plots in each stand were dug down to either bedrock or to a maximum depth of 80 cm, each encompassing an area of 1.5 m × 1.5 m. All roots within the trenches were severed and plastic lining was installed to inhibit root and mycorrhizal in-growth, and ground vegetation was absent (Liu et al. 2016; Zeng et al. 2016). All PVC rings remained in the same position throughout the study period. Soil CO2 effluxes measured from trenched plots were assumed to be Rh. Ra was calculated as the difference between the mean values of Rs and Rh in each stand. All trenched plots were established in May 2016, approximately 12 months before the measurement of soil respiration and its components, to ensure that the disturbance caused by trenching on soil respiration and its components had subsided. The soil CO2 efflux measurement method in the corresponding unburned control stands was the same as that for the burned stands. Soil CO2 effluxes were measured monthly from November to April in 2017–2018. The measurement time lasted approximately two minutes for each soil respiration ring. Each measurement was conducted from 9:00 am to 11:00 am for a total of 48 (30 non-trenched soil rings for Rs + 18 trenched soil rings for Rh) measurements over 2 days.

Soil temperature and moisture were measured using a temperature probe (Licor p/n8100–201) and soil volumetric water content probe (ECH20 EC-5; p/n 8,100,202), respectively, at a depth of 5 cm; snowpack depth was measured with a ruler. The measurement of soil temperature, moisture, and snowpack depth was synchronized with the measurement of soil CO2 effluxes (Hu and Sun 2021).

3.1 Models of soil respiration and its components

An exponential model and the temperature sensitivity of soil respiration (Q10) were used to describe the relationship between soil respiration and soil temperature, which was determined by fitting the exponential function (see Eq. (1) and Eq. (2) below) for burned and unburned control stands (Lloyd and Taylor 1994):

$$ SR={\alpha \mathrm{e}}^{\beta \mathrm{T}} $$
(1)
$$ {Q}_{10}={\mathrm{e}}^{10\beta } $$
(2)

where SR is the measured total soil respiration and its components (Rh and Ra), T is the soil temperature (°C) at 5 cm, α and β are regression coefficients, e is the nature constant, and Q10 is the factor by which soil respiration and its components increase during a temperature increase of 10 °C.

To describe the relationship among soil respiration and soil moisture and the depth of the snowpack, linear, exponential, and quadratic functions were tested for using the data from burned and unburned control stands (Eqs. 3–8); we selected the best fitted model based on higher R2 results (Davidson et al. 1998; Lai et al. 2012):

$$ \mathrm{SR}=\alpha +\beta \mathrm{W} $$
(3)
$$ \mathrm{SR}={\alpha \mathrm{e}}^{\beta \mathrm{W}} $$
(4)
$$ \mathrm{SR}=\alpha +\beta \mathrm{W}+{\omega \mathrm{W}}^2 $$
(5)
$$ \mathrm{S}\mathrm{R}=\alpha +\beta \mathrm{S} $$
(6)
$$ \mathrm{S}\mathrm{R}={\alpha \mathrm{e}}^{\beta \mathrm{S}} $$
(7)
$$ \mathrm{S}\mathrm{R}=\alpha +\beta \mathrm{S}+{\omega \mathrm{S}}^2 $$
(8)

where SR is the measured total soil respiration and its components (Rh and Ra); W is the soil moisture (%) at 5 cm; S is snowpack depth (cm); and α, β, and ω are the constant values of the regression model coefficients.

3.2 Statistical analysis

Data were processed and analyzed using R statistical software version 3.5.2 (R Core Team 2018), using R packages “car” (Fox 2012), “agricolae” (Mendiburu 2017), and “lavann” (Rosseel 2012). Differences in variables between the burned and control stands were tested by analysis of variance (ANOVA) and comparisons between means were performed using the least-significant differences test. Repeated-measures ANOVA was used to determine the direct and interactive effect of fire disturbance and measurement date on soil respiration components (Rs, Rh, and Ra), soil temperature, soil moisture, and snowpack depth. Linear, exponential, and quadratic function models were used to evaluate the relationship among the soil respiration components (Rs, Rh, and Ra) and environmental factors (soil temperature, soil moisture, and snowpack depth). Structural equation modeling (SEM) was used to determine how environmental factors affected soil respiration. A conceptual meta-model was developed, including direct and indirect pathways between theoretical drivers of the components of soil respiration. Only the environmental factors that had a significant correlation with the components of soil respiration were included in the meta-model. Parameters were linked to the model either directly or as a composite variable. Non-significant P-values (P > 0.05) of the chi-square test in SEM suggest a good fit between the model and data. Differences were considered statistically significant at P-values < 0.05.

4 Results

4.1 The effect of fire disturbance on soil environmental factors

The average soil temperature of the non-trenched control and burned stands was −2.73 ± 1.68 °C and −0.91 ± 1.38 °C, respectively. The soil temperature of the trenched control and burned stands was −1.98 ± 2.56 °C and −0.22 ± 1.04 °C, respectively (Fig. 1a and b). The soil temperature of the non-trenched control and burned stands showed a similar variation over time, decreasing at the beginning of the non-growing season and remaining at its minimum value from December 2017 to January 2018, after which there was an increasing trend to the end of the non-growing season, reaching the maximum value from March to April 2018.

Fig. 1
figure 1

Non-growing season soil temperature, soil moisture at a depth of 5 cm, and depth of snowpack measured at non-trenched (a, c, and e) and trenched (b, d, and f) plots in control and burned stands. Error bars represent standard deviations, and the shadowed period indicates the spring freeze-thaw cycle (FTC). Values represent the average of three technical replications and repeated measurements

Full size image

The average soil moisture of the non-trenched control and burned stands was 50.14 ± 15% and 34.98 ± 4.81%, respectively. The average soil moisture of the trenched control and burned stands was 56.10 ± 6.64% and 32.74 ± 9.87%, respectively (Fig. 1c and d). The average snowpack depth at the non-trenched control and burned stands was 14.07 ± 3.05 cm and 10.45 ± 3.30 cm, respectively. The average depth of snowpack at the trenched control and burned stands was 13.58 ± 6.64 cm and 10.74 ± 2.66 cm, respectively (Fig. 1e and f). No significant differences in soil temperature, soil moisture, or snowpack depth were detected between the trenched and non-trenched plots at the different areas (Table 3). Soil temperature was significantly higher at the fire disturbed stands than at the control stands, whereas the soil moisture and average snowpack depth were significantly lower in the fire disturbed area than in the control stands (Table 3). The measurement date had a significant effect on the soil temperature and snowpack depth; however, it did not have a significant effect on soil moisture. Therefore, soil moisture did not show a significant dynamic variation trend similar to that of soil temperature and snowpack depth during the non-growing season (Table 3).

Table 3 Results (F-values) of repeated-measures analysis of variance on the effects of fire disturbance (F), measurement date (D), trench effect (TE), and their interaction on soil temperature (T, °C), soil moisture (W, %), and snowpack depth (S, cm)
Full size table

4.2 Effect of fire disturbance on soil respiration and its components

Rs, Rh, and Ra all showed significant variation during the sampling period (Fig. 2). The Rs trend followed that of the soil temperature in that there was an increasing trend during the non-growing season (Fig. 2a). The mean values of Rs in the control and burned stands was 0.59 ± 0.19 μmol CO2 m−2 s−1 and 0.72 ± 0.15 μmol CO2 m−2 s−1, respectively. The average Rs was not observably significantly different between the two treatments (P > 0.05; Table 4).

Fig. 2
figure 2

Dynamic variations of the non-growing season (a) total soil respiration rate (Rs), (b) soil heterotrophic respiration (Rh), and (c) soil autotrophic respiration (Ra) in control and burned stands. The shadowed period indicates the spring freeze-thaw cycle (FTC). Values represent the average of three technical replications and repeated measurements with standard deviations

Full size image
Table 4 Results (F-values) of repeated-measures analysis of variance of the effects of fire disturbance (F), measurement date (D), and their interaction on soil total respiration (Rs) (μmol CO2 m−2 s−1), soil heterotrophic respiration (Rh) (μmol CO2 m−2 s−1), and soil autotrophic respiration (Ra) (μmol CO2 m−2 s−1)
Full size table

The dynamic patterns of Rh were similar to those of Rs and showed a similar trend to that of soil temperature (Fig. 2a and b), whereas the dynamic pattern of Ra followed that of snowpack depth, which showed a single peak and maximum values in February (Fig. 2c). The mean non-growing season Rh in the control and burned stands was 0.32 ± 0.14 μmol CO2 m−2 s−1 and 0.47 ± 0.15 μmol CO2 m−2 s−1, respectively. The average non-growing season Rh in the burned stands was significantly higher than the control stands by approximately 47% (P < 0.05) (Table 4). The mean non-growing season Ra in the control and burned stands was 0.27 ± 0.09 μmol CO2 m−2 s−1 and 0.25 ± 0.07 μmol CO2 m−2 s−1, respectively. No significant differences were found between the Ra in the control stands and that in the burned stands (P > 0.05) (Table 4). Compared with the control stand, the average Rh:Rs increased from 0.57 ± 0.18 to 0.66 ± 0.19 in the burned stands (Fig. 6 in Appendix). Compared with the entire non-growing season, the FTC period increased Rs, Rh, and Ra 1.8, 1.9, and 1.6 times, respectively, in the control stands. In the burned stands, Rs and Rh increased by 1.6 and 1.9 times, respectively. There was no change between the non-growing season Ra and the FTC Ra in the burned stands (Table 6 in Appendix).

4.3 Relationships among soil respiration components and environmental factors

The exponential regressions with soil temperature as a single controlling factor for Rs, Rh, and Ra were significant (P < 0.01) for both control and burned stands. However, only Rh showed a significant quadratic relationship with soil moisture in the control stand (Fig. 3, Tables 5, and 7 in Appendix).

Fig. 3
figure 3

Relationships between soil total respiration (Rs) and its components (Rh and Ra) and soil temperature (Fig.3a–c), soil moisture (Fig.3d–f), and snowpack depth (Fig.3g–i) in control and burned stands. Equations and statistical parameters are shown in Table 5

Full size image
Table 5 Parameters of the equations for the relationship of soil respiration and its components (Rs, Rh, Ra) (μmol CO2 m−2 s−1) with soil temperature (T), soil moisture (W), and snowpack depth (S) (cm) for the control and burned stands.
Full size table

The quadratic function regressions with the snowpack depth as the single explanatory variable of Rs, Rh, and Ra were the best fitted models to describe the relationship among snowpack depth and soil respiration and its components (Fig. 3, Tables 5, and 7 in Appendix). In the control stands, snowpack depth pack showed a significant relationship with Rh and Ra; snowpack depth showed a significant relationship with soil Rs, Rh, and Ra in the burned stands (Fig. 3 and Table 5).

Based on the model fitting results, soil temperature and snowpack depth were introduced in the structural equation model to describe the relationship between Rs and its components (Rh and Ra) and soil environmental factors. All structural equation models fitted well (P > 0.05), explaining approximately 86.1%, 84.9%, and 66.6% of the variation for Rs, Rh, and Ra, respectively (Fig. 4).

Fig. 4
figure 4

Structural equation model describing the influence of soil temperature and snowpack depth as drivers of soil total respiration (Rs) and its components (Rh and Ra) after fire disturbance. Solid boxes represent observed variables, while the hexagonal box depicts a composite variable (to account for a polynomial model structure). Single and double headed arrows represent relationships and correlations between variables, respectively; the strength and sign of relationships and correlations are depicted by standardized path coefficient. *, **, and *** represent significance at P < 0.05, 0.01, and 0.001, respectively

Full size image

The structural equation models revealed direct effects of environmental factors on Rs, Rh, and Ra, and also helped explain the interaction effect between soil temperature and the snowpack depth composite. The structural equation models revealed that soil temperature directly affected Rs, Rh, and Ra (Fig. 4). Although Rs and its components (Rh and Ra) all showed significant quadratic relationships with snowpack depth, this composite only had a significant direct effect on Rs and Ra (Fig. 4).

4.4 Changes in Q 10 induced by fire disturbance

Compared with the control stands, the Q10 of Rs and Ra was lower by approximately 6.8% and 15% than that in the burned stands, respectively; however, these results were not significant (P > 0.05) (Fig. 5 and Table 5). The Q10 of Rh in the burned stands was significantly greater by approximately 27.3% than that of the control stands (P < 0.05) (Fig. 5 and Table 5).

Fig. 5
figure 5

Temperature sensitivity (Q10) of non-growing season soil total respiration (Rs), soil heterotrophic respiration (Rh), and soil autotrophic respiration (Ra) in control and burned stands. Different letters represent significance at P < 0.05. Values presented represent the average of three technical replications and repeated measurements with standard deviations

Full size image

5 Discussion

5.1 Seasonal variation of the non-growing season soil respiration and its components

In the present study, the mean Rs during the non-growing season in the temperate Q. mongolica forest was 0.59 ± 0.19 μmol CO2 m−2 s−1. This result was higher than that of a previous study that found that the non-growing season Rs in a boreal forest ecosystem of China was 0.29 ± 0.06 μmol CO2 m−2 s−1 (Hu et al. 2017a). This difference might be due to a lower temperature occurring in the higher latitudes of China. We found synchronous responses to Rs and Rh with a peak in mid-April, whereas Ra tended to follow the dynamics of snowpack depth, which showed a single peak curve with maximum values occurring in February (Fig. 2). In agreement with our hypothesis, Rh stemming from the microbial decomposition of soil organic matter was the dominant component during the non-growing season Rs. Several studies indicated that Rh dominated Rs during the non-growing season (Gaumont-Guay et al., 2008; Hanson et al., 2000; Jiao and Wang, 2019; Savage et al., 2013; Tucker et al. 2014). At the same time, the non-growing season Rh accounted for a larger proportion of Rs than in the growing season (Ruehr and Buchmann, 2010; Shi et al., 2012; Tang et al., 2005). This may be caused by a difference in how Rh and Ra respond to soil environmental changes (Li et al., 2013; Zou et al., 2018). Our results indicated that Rh was more sensitive to environmental changes than Ra in the non-growing season. Despite the low temperature in the non-growing season, soil microbes were still active and were the dominant biotic controller of the non-growing season Rh due to the regulatory effect of snowpack depth (Yi et al., 2020). Several studies have shown that the duration and depth of snow cover can significantly influence soil temperature and that correspondingly, soil temperature can manipulate snow depth, which could therefore significantly influence Rs and its components (Gavazov et al. 2017; Nobrega and Grogan 2007; Reinmann and Templer 2018; Uchida et al. 2005). Our finding highlights that the interaction coupling effect between soil temperature and snowpack depth must be considered when studying the components of non-growing season Rs (Gao et al. 2018).

5.2 Effects of fire disturbance on non-growing season soil respiration and its components

Our findings show that fire has different effects on non-growing season Rs. Non-growing season Rh was significantly increased by approximately 47% after fire disturbance and Rh:Rs increased from 0.57 to 0.66, whereas the non-growing season Ra showed no change after fire. This result is in agreement with our hypothesis that the non-growing season Rs was dominated by Rh after fire disturbance. The non-growing season Rs increased after fire disturbance; however, this increasing trend was not significant for the non-growing season Ra after fire disturbance. There are several reasons to explain our findings, the first being that fire changes the quality and quantity of detritus, which may promote higher decomposition rates by microbes. Our results agree with the finding from a previous study of increased Rs in a boreal forest of interior Alaska after prescribed burning, which was mainly attributed to higher Rh after fire disturbance (Kim 2013). Fire burned the vegetation and soil organic layer, which increased the availability of nutrients, thus promoting microbial activity and changing the decomposition rate (Song et al. 2017; Wüthrich et al. 2002). Second, post-fire environmental factors control the variation of components of non-growing season Rs after fire disturbance. Forest fires decrease the forest canopy, which will directly increase the soil temperature after fire disturbance (Munozrojas et al. 2016). Owing to the persistent leaves of Q. mongolica, canopy cover of the burned stands was lower than that of the control stands in the non-growing season, which directly led to the higher surface solar radiation of non-growing season in the burned stands. In the present study, soil temperature significantly increased by approximately 2 °C after fire disturbance and snowpack depth and organic material layer depth were significantly decreased in the burned stands. Recent studies have shown that non-growing season Rs is almost entirely driven by microbial decomposition, which is a temperature-dependent biological process, and that soil temperatures between −2 °C and 0 °C strongly affect substrate supply and soil microbial activity (Monson et al. 2006; Tucker 2014). We therefore suggest that higher soil temperature and changed substrate supply from burned debris may be driving the increase of non-growing season Rh:Rs after fire.

Fire severity has a strong effect on the components of non-growing season Rs with the effect of fire depending on severity and duration, which may account for the divergence in our Rs response to fire (Czimczik et al. 2006; Meigs et al. 2009; Nave et al. 2011; Richards et al. 2012; Song et al. 2018; Uribe et al. 2013). High severity fires have greater negative effects on ecosystem processes than that of low severity fires (Dooley and Treseder 2012; Martínez-García et al. 2017; Plaza-Álvarez et al. 2017). Previous studies have shown that the non-growing season Rs of Dahurian Larch in the high latitudes of China decreased by approximately 55% in burned stands, which may be attributed to the decrease of Ra after a high severity fire (Hu et al. 2017a). This is because high severity fires result in the understory shrubs, litter, and duff layers being completely burned, causing damage to plant roots (Hu et al. 2017a; O’Donnell et al. 2009). This result is inconsistent with the findings from our study, possibly because the fire in our study was only of medium severity. The rapid recovery of pioneer vegetation after fire promoted the recovery of plant roots, which may be the main reason why there was no significant difference in non-growing season Ra between the control and burned stands (Hart et al. 2005; Johnson and Curtis 2001).

Recent meta-analyses and long-term experiments have shown that global Rh is increasing, probably in response to environmental changes; therefore, climate-driven losses of soil C are currently occurring across many ecosystems, with a detectable and sustained trend emerging at the global scale (Crowther et al. 2016; Melillo et al. 2017; Wang et al. 2014a; Zhou et al. 2016). Our results indicate that an increase in forest fire frequency might accelerate the process of C loss in said ecosystems within the context of global warming and the intensification of the El Niño effect (Jolly et al. 2015; Yin et al. 2016). Fire disturbance will convert live vegetation into dead material that decomposes, changes ambient soil conditions and temporally decreases the ability of the ecosystem to gain C via plant photosynthesis, which in turn will change the relationship between net primary production (NPP) and net ecosystem production (NEP) (Keeley 2009; Smithwick et al. 2007). Fire could then drive NEP (i.e., NEP = NPP – Rh) to be negative, and the ecosystem to become a source of C to the atmosphere (Harmon et al. 2011). Thus, fire as a potent factor should not be ignored in forest ecosystems, especially during the non-growing season as it is vulnerable to micro-environmental variation.

5.3 Effect of environmental factors on non-growing season soil respiration and its components

Previous studies have reported that the components of soil respiration exponentially increased with temperature increase during both the growing and non-growing seasons (Bondlamberty et al. 2004; Mo et al. 2005; Monson et al. 2006; Yi et al. 2020). Our study showed that wildfire increased the non-growing season Rs. Consistent with our hypothesis and results from previous studies, snow depth as an insulating layer influences the non-growing season (i.e., winter) respiration (Aanderud et al. 2013; Brooks et al. 2011; Wang et al. 2010). Although snowpack depth decreased in the burned stands, it still had a significant quadratic function relationship with Rs and components after fire. Thus, the interaction between soil temperature and snowpack depth was the driving environmental factor controlling the non-growing season soil respiration and its components after fire disturbance.

Previous studies have found that higher soil moisture stimulated soil respiration when the soil water content was below optimum (Rey et al. 2011; Yohannes et al. 2011). Although we did find lower soil moisture in the burned stands, probably due to the higher solar radiation and thinner snow depth, soil moisture did not show a significant correlation with the non-growing season Rs and its components after fire disturbance. This result may be due to the non-growing season soil moisture being abundant (> 30%) in the burned stands. Therefore, the effects of soil moisture on the non-growing season Rs and its components were minor or were counterbalanced by other environmental factors after fire.

In our study, the mean Q10 of the non-growing season Rs were 2.53 and 2.36 in the unburned control and burned stands, respectively. The results were higher than the global scale estimate (1.69) (Zhou et al. 2009). In addition, compared with the unburned control stands, the Q10 of Rh was significantly greater in the burned stands (2.39 vs. 3.12), whereas there was no significant difference in Q10 of Ra between the unburned control stands and burned stands (2.69 vs. 2.29). These results were inconsistent with a previous study in which a high severity fire decreased the growing season Q10 of Rs and Rh in a forest of boreal China (Hu et al. 2017b). These contrasting results might be attributed to the high severity fire destroying the plant root structure and a loss of the labile fraction of soil organic C to the atmosphere, which restrained root and rhizosphere respiration and limited soil microorganisms activity (Conant et al. 2011; Thornley and Cannell 2001). Rh was the dominant component of the non-growing season soil respiration efflux. The mean Rh of the FTC period was 1.9 times greater than the non-growing season; Rh accounted for 71% to 84% during this period. The higher temperature and soil nutrients could provide more activation energy based on Arrhenius kinetic theory. Activation energy is one of the dominant abiotic factors that is directly related to the substrate supply (Schipper et al. 2014). More recalcitrant substrates in burned areas, which are complex molecules and have higher activation energy, should have higher temperature sensitivity than those in unburned areas (Davidson and Janssens 2006; O’Neill et al. 2006). Therefore, the higher solar radiation and temperature combined with soil nutrient content in the burned areas could promote the microbial decomposition leading to the higher non-growing season Q10 of Rh after fire (Mikan et al. 2002; Pan et al. 2013).

In the present study, we found that fire led to the non-growing season Rh significantly increasing after a fire disturbance. Considering that temperate and boreal forests have been experiencing a significant increasing trend of fire occurrence caused by global warming and cold months have been facing even faster warming than the growing season during the past decades (Hantson et al. 2015; Jolly et al. 2015; Piao et al. 2007; Zhang et al. 2013), our study suggests that forest fires create an increase of non-growing season Rh:Rs, which will potentially decrease the amount of net C stored in forest ecosystems.

6 Conclusion

In summary, the present study explored the effects of recent fire disturbance on the components of non-growing season Rs (Rh and Ra) as well as their Q10 in a cold temperate forest in northeast China. Our results revealed that forest fires significantly increased the non-growing season Rh and also drove the Rh:Rs increase that was found in burned stands. The Q10 of Rh significantly increased in the burned stands. The interaction between soil temperature and snowpack depth was the driving environmental factor controlling the non-growing season soil respiration and its components after fire. Our study highlights that fire is a potent factor on the components of the soil respiration and should not be ignored in forest ecosystem C cycling, especially during the non-growing season as it is vulnerable to micro-environmental variation. Considering that temperate and boreal forests have been experiencing a significant increasing trend of fire occurrence caused by global warming and that cold months have been facing even faster warming than the growing season during the past decades, long-term studies involving diverse ecosystems are required to better elucidate mechanisms that have been found during the non-growing season Rs under an increasing trend of fire occurrence.