Shubo Huang · Lixiang Wen · Shuai Yin ·Meng Guo · Fangbing Yu

Abstract The decomposition of coarse woody debris(CWD) af fects the energy f low and nutrient cycling in forest ecosystems. Previous studies on CWD have focused on the input, decomposition, reserve dynamics, and CWD functions, but coarse woody debris decomposition is complex and the results from dif ferent regions vary considerably. It is not clear which factors af fect decay rate ( k), especially at dif ferent decomposition stages. In this study, a singleexponential decay model was used to analyze the characteristics of CWD decomposition in Larix gmelinii forests over the 33 years following a f ire in the Greater Khingan Mountains. The results show that the decay rate of coarse woody debris was positively correlated to decay class. The average decomposition rate was 0.019, and 41 years and 176 years are needed for a 50% and 95% mass loss, respectively. CWD nutrient content, density, and water content could explain the variance in the decay rate (～ 42%) of the decay factors such as amount of leaching, degree of fragmentation, respiration of the debris, and biotransformation, and varied signif icantly between dif ferent decay classes. Using the space–time substitution method, this study arranged the coarse woody debris of dif ferent mortality times to form a 33 year chronosequence which revealed the decomposition process.It was concluded that the decay rate was mainly explained by structural component of the debris and its nitrogen and water contents. This paper quantif ies the indicators af fecting CWD decay to explain the decomposition process.

Keywords Coarse woody debris · Decay rate · Space–time substitution · Boreal forest · Fire disturbance

Introduction

Coarse woody debris (CWD) in the form of snags, downed boles or large branches is an important structural and functional component of forest ecosystems (Harmon et al. 1986).CWD af fects biotic and abiotic processes through physical and biological ef fects and plays an important role in longterm nutrient storage and tree regeneration (Wu et al. 2005).It also provides habitats for various organisms and maintains environmental heterogeneity, biodiversity, and the integrity of ecosystems. In particular, habitat provision and nutrient storage are signif icant for maintaining the continuity of biodiversity and biogeochemical processes (Sturtevant et al. 1997). Research has combined the characteristics of coarse woody debris with forest succession (Carmona et al.2002), community composition (Motta et al. 2006), nutrient cycles (Currie and Nadelhof fer 2002), and forest management (Montes and Cañellas 2006) to study its function and decomposition in forest ecosystems. Coarse woody debris accounts for approximately 5% of carbon (C) storage in terrestrial ecosystems, and in forest ecosystems, for 2–10% of above-ground biomass (Delaney et al. 1998). The proportion of above-ground biomass represented by CWD depends largely on geographical location, forest management practices, disturbances and successional stage. Coarse woody debris biomass in natural forests and plantations usually dif fers. CWD reserves of old forests with little human disturbance can be up to 100 times higher than that in young forests (Carmona et al. 2002). Furthermore, the relative contribution of woody debris to ecosystem C storage increases with increasing stand age (Zhu et al. 2017).

CWD decomposition is a complex process, integrating respiration, biological transformation, leaching, fragmentation, collapse, settling, and weathering under the inf luence of physical, chemical and biological (especially microbial)processes. It is one of the main ways in which C f ixed during photosynthesis is returned to the atmosphere or converted to soil organic C. Decomposition is driven by heterotrophic respiration of decomposers and directly emits CO2into the atmosphere (Mackensen et al. 2003). Dynamic changes in CWD depend on the dif ferences between income and loss(Olson 1963), where income is caused by the natural death of trees or disturbance, and loss is the decomposition of woody debris. The decay class can qualitatively describe the degree of decomposition, expressed by external morphological characteristics, and quantitatively measure decomposition speed via the decay rate, expressed by the decay constantk(Laiho and Prescott 1999; Kruys et al. 2002).The decay rate of CWD is slow and variable and inf luenced by species, temperature, humidity, material quality, size,and decay class. Since water content in wood is the basic environment of fungal growth, it is considered as the most important factor in CWD decomposition (Harmon et al.1986). Moreover, angiosperms have more complex vascular structures than gymnosperms, but the latter have slower rates of decomposition because of lower N contents and higher C/N ratios. The relationship between the structural components and decay rate of gymnosperms is insignif icant, and low N and high lignin contents are closely related to rate of gymnosperm decay (Weedon et al. 2008; Zhang and Wang 2010). The role of coarse woody debris in nutrient cycling of forest ecosystems is generally poorly understood, and the importance of CWD to forest productivity remains controversial (Spies et al. 1988).

Some researchers have emphasized the importance of the decomposition mechanism (fragmentation, respiration,and leaching), and considered the degree of sample fragmentation in terms of mass loss. Various models for calculating the decay constant have been proposed (Sollins 1982; Marra and Edmonds 1994). CWD decay exponential models include single-exponential, double-exponential,multiple-exponential, and lag-time models, where the f irst three consider only respiration and leaching, and the latter consider fragmentation. The dif ferences in decay rates in CWD calculated in current studies ref lect variations in the fore mentioned calculation methods and in ecosystems(Attiwill 1994). Single-exponential models are widely used and assume that the decay rate is constant and CWD is homogeneous. However, decay rates vary throughout the decomposition process and depend on the climate, species,size of material, decay class, slope position, aspect, and site conditions. The decomposition process may be divided into several stages and each stage described by a dif ferent single-exponential model (Yatskov et al. 2003). This process can also be modeled based on physiological assumptions or matrix models (Kruys et al. 2002). Numerous studies have used 50% and 95% of the coarse woody debris mass loss time to represent CWD turnover, and the decay rate of boreal forest species was 0.003–0.071 with a mean value of 0.02.The 95% mass loss, calculated according to the mean value,was about 150 years (Laiho and Prescott 1999). The decay rate of dif ferent tree species is very dif ferent. Xu ( 1988)calculated the average decomposition rate ofLarix gmelinii(Rupr.) Rupr CWD to be 0.013. The decay rate ofTilia amurensis, a common species in the Lesser Khingan and Changbai mountains is 0.028, and forPinus koraiensis Sieb,it is 0.016. The decay rate ofTsuga heterophylla(Raf.) Sarg.,widely distributed in the western Pacif ic Coast of the United States, is 0.016–0.019, and forPseudotsuga menziesii, it is 0.005–0.010 (Chen and Harmon 1992). It indicates that the decay rate ofLarix gmeliniiis much lower than that ofTilia amurensis,Pinus koraiensisandTsuga heterophylla, slightly higher than that ofPseudotsuga menziesii. In the Greater Khingan Mountains, the species with the slowest decay rate isLarix gmeliniiand the fastest isBetula platyphylla Suk.,and the rate forPinus sylvestrisis between the two species(Xu 1988).

Fire changes the succession process in forest ecosystems, af fecting energy f lows, nutrient cycles, and information transfer between various components. Fires also play an important role in maintaining and improving the age distribution as well as maintaining biodiversity and stability. A large amount of CWD is created after a f ire, some is generated immediately and some after a few years. In addition to disturbances such as f ire, a certain amount is generated during succession (Amiro et al. 2006). Forests in northern China account for 14.5% of the world’s land area.From 1950 to 2014, 806,000 forest f ires occurred cumulatively (12,400 annually), and 38.093 million ha of forested area (586,000 ha per year) was af fected (Lierop et al. 2015).Every year, 5–20 million ha of China’s northern forest is burned, releasing about 1.82 × 1011 kg of C and accounting for 9.1% of global C emissions from f ires. Most studies focus on this stage of a forest f ire, i.e., on post-f ire ecological and environmental impacts on vegetation and on habitat restoration, and the adaptation characteristics of organisms to f ire (McLauchlan et al. 2020). Numerous studies have used dynamic vegetation simulation models to re-create the ef fects of f ires on carbon cycling. Forest f ires can reduce the C sink capacity of ecosystems but the extent of such impacts remains uncertain (Hayes et al. 2011; Yue et al. 2015). In addition, accurate prediction of tree mortality is crucial for adaptive forest management and estimation of wood debris input after f ire. Fernandes et al. ( 2008) reviewed forest f ire mortality models of seven European coniferous forests and found that the resistance of pines to low- to moderate-intensity f ire arose from adaptive traits that keep the tree alive;f ire-sensitive species can also tolerate low-intensity f ires.Catry et al. ( 2010, 2013) studied post-f ire mortality rate of mixed forests in central Portugal and found that mortality rates in conifers were as high as 90% four years after a f ire.More than 75% of broad-leaved crowns were mortally damaged whereas the mortality rate of regenerated broad-leaved trees through germination was only 8%. The post-f ire mortality rate of conifers was controlled by f ire intensity, and that of broad-leaved crowns decreased with increasing bark thickness and diameter.

To reveal the mechanism of short- term CWD decomposition, researchers replaced long-term repeated measurements with a space–time substitution method which provided indirect information on decay rate (Harmon et al. 1986; Tyrrell and Crow 1994; Frangi et al. 1997; Lorimer 2011). This study assumed that the time of death of CWD was the year of the burn. Based on a space–time substitution method, it was hypothesized that all the burned patches had similar site conditions, including altitude, slope, aspect and vegetation cover and we want to answer: (1) In the Greater Khingan Mountains, how many years will it take forLarix gmeliniiCWD to decompose following a f ire? (2) What factors af fect the decay rate of CWD; and, (3) How do these factors af fect the decay rate of CWD in dif ferent decomposition phases.

Materials and methods

Study region

Fig. 1 Location of burned patches in study forests in the Greater Khingan Mountains

The forests of the Greater Khingan Mountains are part of the global cold-temperate forests and a southern extension of the eastern Siberian coniferous forests of Northern Eurasia (Fig. 1). The forest area is 11.2% of China’s total forest area and accounts for more than a third of carbon storage (Wang et al. 2001). The main chain of the Greater Khingan Mountains is NNE-SSW and is steeper in the east than in the west. The northern branch is WNW-ESE,235 km long by, 200 km wide, a total of 8.3 × 10 6 km 2 ,with elevations 260–1700 m a.s.l. (Guo et al. 2020).Larix gmeliniioften forms large simple forests, or mixed forests withBetula platyphylla, Pinus sylvestris var.mongolica,Picea asperata, Populus davidiana dode, and other minor species. Soils are brown coniferous forest soil with different soil subclasses under different types of Larch forest(Xu 1988). The climate is a cold temperate continental monsoon with an annual average temperature of approximately − 2.8 °C, maximum and minimum temperatures of 35 °C and − 52 °C, respectively, and an annual average precipitation of 460 mm. The rainy period is from June to September, and a frost-free period lasts 90–110 days(Xu 2018). Winter can extend up to nine months with little precipitation under the control of the Mongolian high pressure system. Summer is brief, preceded by a definite spring and succeeded by a distinct autumn season, all affected by dry winds from Mongolia bringing drastic changes in weather, e.g., high temperatures, low humidities, strong winds, and high forest fire frequency(Hu et al. 2004). The Greater Khingan Mountains area is widely affected by forest fires which are both environmentally beneficial as well as disruptive. The average annual number of fires is over 35 with an average area burned of 76,600 ha (Guo 2007). Fire affects forest soils,hydrology, and biology, which in turn affect the structure,function, and dynamics of individual plants, populations,and communities.

Sample collection and decay classif ication of Larix gmelinii

In July 2019, 60Larix gmeliniisamples were collected from seven burned patches (Fig. 1, Table 1) in the study area (Fig. 2). It was assumed that the coarse woody debris developed during the same year as the f ire disturbance, and samples of dif ferent decay classes were collected according to site conditions. Each 20 cm bole sample was numbered and taken to the laboratory. The sample indices were unevenly distributed from the outside to inside, each divided into several parts, and three randomly selected to measure the indices separately and then averaged to represent the average condition of each sample.

No uniform standard exists for CWD diameter research purposes (Yuan et al. 2011). Harmon et al. ( 1986) proposed that debris with diameters ≥ 2.5 cm qualif ied as CWD. Other studies have used ≥ 2.5 cm (Tang et al. 2003; Zhang et al. 2009; Liu et al. 2011), 7.6 cm (Wu et al. 2011), 8 cm (Yang et al. 2002b)and 10 cm (Yuan et al. 2012). CWD in burned patches was dominated by snags and logs. This study considered woodydebris with diameters ≥ 7.6 cm as CWD and further classif ied it into f ive standards proposed by Sollins ( 1982) and Fogel and Cromack ( 2011), combined with the characteristics ofL. gmelinii(Table 2) to render them suitable for boreal forests.

Table 1 Basic information of burned patches

Fig. 2 Burned patches in the Greater Khingan Mountains in 2003 a and 2010 b

Space–time substitution

Space–time substitution is a commonly used method for studying vegetation succession and restoration. In forest ecosystems, complete restoration may often take several decades or centuries. To determine the short-term decay rate of CWD,researchers have developed a space–time substitution sampling method (Harmon et al. 1986; Tyrrell and Crow 1994; Frangi et al. 1997; Lorimer 2011). Typically sites at dif ferent succession stages are chosen in a region or in dif ferent regions with similar environments. The decay rate of CWD at dif ferent times varies with changes in its density and represents the actual decomposition dynamics.In order to calculate the decay rate of CWD, the burn year was determined from remote sensing images, and was considered as the time of tree mortality. CWD with conf irmed times of mortality was arranged to a time series. In this way,the change of decomposition rate with time can be calculated by comparing dif ferent sample densities.

Experimental methods

The fresh weight (M, g) was obtained by weighing a part of the volume of each sample, the dry weight (m, g) by drying a part of the volume of each sample at 80 °C for 48 h to a constant weight. The sample volume (V, cm 3 ) was measured using the displacement method due to the irregular shapes of the samples. CWD water content (WCWD , %) was calculated by subtracting the dry weight from the fresh weight (WCWD;Eq. 1), and the sample density (ρ; Eq. 2) by dividing the dry weight by the volume.

The decay rate was estimated based on the changes in density expressed as the decay constantk. The year of tree death was assumed to be the same as the year the f ires occurred, and a single-exponential decay model (k;Eq. 3)was used to calculatek(Olson 1963). T0.5and T0.95, representing the decomposition of 50% and 95% of the samples,were calculated using Eqs. 4 and 5, respectively (Mackensen and Bauhus 2003):

After measuring the WCWDand ρ of each sample, a grinding machine was used to crush the samples, and an EA3000 Elemental Analyzer (EuroVector SRL, Pavia, Italy) used to determine carbon (Cc, %) and nitrogen (Nc, %) contents to analyze their changes with decomposition and the ef fect on decay rates. Carbon (Cd, g cm −3 ) and nitrogen (Nd, g.cm −3 )densities were obtained by multiplying Ccand Ncby the sample density. Using a lignin analyzer for neutral and acid washing, and concentrated sulfuric acid for soaking, lignin (Lic,%), cellulose (Cec, %), and hemicellulose (Hcc, %) contents were measured (Eqs. 6– 8). Lignin (Lid, g cm −3 ), cellulose(Ced, g cm −3 ), and hemicellulose (Hcd, g cm −3 ) densities were obtained by multiplying the content of these substances byρ.

Table 2 Decay classes of coarse woody debris in this study

To determine the changes ink,ρ,WCWD, C, and N turnover, structural components of the debris decomposition process were determined. Ordinary least squares regression and analysis of variance (ANOVA) were used to study the changes ink,ρ,WCWD, Cc, Nc, Cd, Nd, C/N, Lic, Cec, Hcc,Lid, Ced, Hcd, and lignin/N between dif ferent decay classes and to calculate the regression equation ofk. To explain the ef fect ofWCWD, Nc, Cc, C/N, Lic, Cec, Hcc, lignin/N,andρonk, multi-regression equations were established.The log transformation of variables before ordinary least squares regression was used to pass assumptions of normality, linearity, and homoscedasticity of the model. All statistical analyses and charting were carried out in R3.5.1 (2018)software.

Results

Decay rate ( k) based on density change

Thek, T0.5, and T0.95values were calculated from Eqs. 3– 5(Table 3). The results show that CWD decomposition was slow in early stages and more rapid in the late stages, andkincreases faster in the late decomposition stage. In the entire decomposition process, it took 41 years for 50% decomposition and 176 years for 95%. The decay rate in each burned patch was positively correlated to decay class, and the rate of increase dif fered between classes: decay class I was the lowest and decay class IV the highest, with the rate almost doubling. This means that with an increasing number of years after the f ire, both the CWD decay class and rate increased.

Factors af fecting decay rate (k)

Univariate regression between log (k) andWCWD, Nc, Cc,C/N, Li c , Ce c , Hc c , lignin/N, andρare shown in Table 4.Most explanatory variables had a signif icant linear relationship with log (k) (P< 0.001), and theR2 values of the linear regression model varied according to the decay class(Table 4). Furthermore,ρwas the most signif icant factor af fectingk(R2 = 0.294–0.454).R2 values increased with increasing decay class and signif icantly af fectedkin each class. Log (k) was negatively correlated withρ, which indicated that density decreased with ongoing decomposition andkincreased accordingly. Decay rate in the early stages (decay class I) was mainly af fected by CWD structural components, Lic, Cec, and Hcc(lignin, celluloseand hemicellulose) contents. In the middle stages (decay classes II–IV), C/N and lignin/N signif icantly af fected the rate of decay. The water content (WCWD) inf luenced the rate of decay in later stages (decay classes IV and V). Carbon levels had no signif icant ef fect onk. With increasing decay class, factors af fecting the decay ratekgradually changed from intrinsic factors (ρ, Lic, Cec, Hcc, Nc) to an environmental factor (WCWD).

Table 3 Decay rate ( k) and time required for 50% (T 0.5 ) and 95%(T 0.95 ) mass loss of coarse woody debris

To maximize the explanatory ability of the model with few variables, all subsets regression was used to test possible multi-regression models and selected the “best”model (Table 5). Multi-regression analysis could explain the factors that af fectedkin dif ferent decay classes, and all explanatory variables could explain the variance of log (k) at an average of about 42% (R2 = 0.422). Intrinsic factors (Nc, Cc,ρ, Cec, and Hcc) had a greater inf luence onkthan the environmental factor (WCWD). The multiregression analysis showed that the inf luencing factors inthis study were among many in decomposition that are jointly af fected by dif ferent explanatory variables at different decomposition stages.

Table 4 Univariate regression results

Table 5 All-subsets regression results

Table 6 Average value of each variable in f ive decay classes

Relationship between decay class and variables

Fig. 3 Average value of variables in f ive decay classes in dif ferent burned patches; W CWD , coarse wood debris water content; ρ, coarse wood debris density; N c , nitrogen content; C d , carbon density; Ce d , cellulose density; Hc d , hemicellulose density

Average value of each variable in dif ferent decay classesWith increasing decay class (Ι-V),k,WCWD, Nc, Ndincreased,ρ, Cd, Hcc, Cec, Hcd, Ced, Liddecreased (Table 6).However, carbon and lignin contents and nitrogen density(Cc, Licand Nd) increase with an increase of decay class(Ι-IV), decrease at the f ifth decay class but are still greater than in the f irst four decay classes. Decay rate tripled from decay class I to IV, nitrogen levels increased from 0.09 to 0.20%, average carbon content was about 50%, and carbon density decreased by 34%. The average proportion of CWD structural components was 74%, with cellulose content (Ce c )the highest value, followed by Li c , and Hc c with the lowest.With increasing decay class, the density of lignin, hemicellulose and cellulose (Lid, Hcd, and Ced) decreased by 26.7%,48.3%, and 58.7%, respectively. The most readily decomposed variable was cellulose and lignin the least.

Figure 3 shows the variables with the same trend with increasing decay class in dif ferent burned patches (WCWD,ρ,Nc, Cd, Ced, and Hcd). Water content (WCWD) increased signif icantly in the later stages of decomposition. With increasing decay, nitrogen gradually accumulated and Cd, Ced,Hcd,andρgradually decreased. Compared with cellulose and hemicellulose contents, their densities changed uniformly between dif ferent decay classes, but nitrogen did not, which may be related to low nitrogen levels in coarse woody debris.The inconsistent trend in each variable in the same decay class between the dif ferent burned patches ref lect the main source of error in the space–time substitution. Conversely,the consistent trend between dif ferent decay classes in the same burned patch ref lect the inf luencing factors related to decomposition. The six variables in Fig. 3 are consistent with the variables af fectingkin Table 4, which are related to WCWD, ρ, N, C, Ce, and Hc, indicating that the decay classes provided a standard for the qualitative description of CWD decomposition dynamics. A one-way ANOVA was used to further study the signif icance of changes in each variable in dif ferent decay classes.

Multiple comparison of each variable in dif ferent decay classes

The one-way ANOVA results are shown in Fig. 4. Decay classes are signif icantly related to values ofρ, Nc, C/N,WCWD, Cd, Lid, Cec, Hcc, Ced, and Hcdindices but were not signif icant with Nd, Cc, Lic, and lignin/N (P> 0.1) and therefore are not described here. The multiple comparison results of Ce d , Hc d , C d , ρ, and W CWD were more signif icant. Ce d ,Hcd, Cd, and ρ all decreased signif icantly between the f irst two decay classes. Compared with decay class I, Ced, Hcd,Cdandρdecreased signif icantly in decay class II but not signif icantly among other classes (II–V). WCWDincreased from 18 to 113%, almost unchanged in the f irst three decay classes but was signif icantly greater in class V than in the other classes. Nitrogen levels changed smoothly between decay classes and were signif icantly higher in decay classes IV and V than in classes I and II, and no signif icant dif ferences existed between the two adjacent decay classes (I–V).C/N ratios decreased with an increase of decay but the f luctuation was not obvious. Adjusted nutrient densities of carbon, lignin, cellulose and hemicellulose (Cd, Lid, Ced, and Hcd), in terms of mass loss, showed more signif icant changes between decay classes than unadjusted nutrient concentrations did (C c , Li c , Ce c , and Hc c ).

Discussion

Previous research on decay rate ( k)

Numerous researchers have used single-exponential decay models to calculate the decay rate of coarse woody debris in dif ferent regions for dif ferent species and have obtained varying results (Table 7). Most studies used space–time substitution, with the time series spanning 13–140 years.Only a few studies have used repeated measurements.Because of the dif ferent research methods and areas,khas varied greatly between the research results, and dif ferences between species in the same area ref lect the impact of CWD factors onk. Mattson et al. ( 1987) found thatkchanged 10 times among tree species, and broad-leaved species generally decayed faster than coniferous species.The dif ference between the same species in dif ferent areas ref lected the inf luence of environmental factors on decay rates. In these study, because of the local environment and regional climate,kdid not increase with increasing latitude. Repeated measurements could describekmore accurately, and the space–time substitution method could elucidate changes in decay rate over the long term. Nevertheless, some errors did occur.

Compared with the results of repeated measurements,the space–time substitution method underestimated the actualk, and the length of the established chronosequence impactedk. This study shows that, with increasing decomposition time,kincreased with increasing decay class, and the average decay rate in dif ferent decomposition periods did not vary substantially. In general, the averagek-value ofL. gmeliniicoarse woody debris over 33 years was 0.019, and 50% and 95% mass loss occurred over 41 and 176 years, respectively (Table 3). Over 140 years of decomposition, the average decay rate ofL.gmeliniiwas 0.013 (0.010–0.014), 50% mass loss took 53 years (50–67 years), and 95% mass loss took 231 years(219–289 years); the decay rate of snags and logs was not significant (Xu 1988). Using the same method of space–time substitution, other researchers obtainedkvalues higher than in this study (Chambers et al. 2001; Mackensen et al. 2003; Lv et al. 2006) and some lower (Chen and Harmon 1992; Harmon et al. 2000; Yang et al. 2002)because of dif ferent species and locations. The decomposition ofL. gmeliniiwas rapid initially and slower in later stages, which may be related to factors such as the high organic matter of CWD, strong respiration of the material,snags in later decomposition stages in contact with the ground, and changes in temperature and moisture content.Compared with previous studies, thek-value ofL. gmeliniiin this study was higher which may be related to the short chronosequence (33 years). Various methods are currently being used to determine the decay rate. In future research,the most appropriatek-values need to be used and research methods and contexts need to be considered. Even without information on species and site-specif ic decay rates,decomposition can be modeled and used to indicate expected decomposition times of individual samples.

Fig. 4 Multiple comparisons of single factor analysis of variance;same letter between groups show mean dif ference was insignif icant( P > 0.05) and letters between groups show that mean dif ference was signif icant ( P < 0.05); C d , carbon density; Li d , lignin density;Ce c , cellulose content; Ce d , cellulose density; Hc c , hemicellulose content; Hc d , hemicellulose density; N c , nitrogen content; ρ, coarse wood debris density; W CWD , coarse wood debris water content; ***,P < 0.001; **, P < 0.01; *, P < 0.05

Table 7 Decay rate ( k) of coarse wood debris

Decomposition process

The decay rate is af fected by environmental and intrinsic factors. By establishing regression equations fork(Table 5), the interpretability of each explanatory variable was about 42% (R2 = 0.375–0.484). Moreover,ρstrongly inf luencedk(R2 = 0.294–0.454). In lower decay classes(I–III), structural components (lignin, cellulose and hemicellulose) af fectedk. C/N and lignin/N ratios could also be used as indicators of the rate of decomposition. In highly decomposed classes IV and V, a certain degree of fragmentation occurred, and WCWDstarted to af fectk(Table 4).As decomposition progressed andkincreased, the water,lignin and nitrogen contents were positively correlated but levels of cellulose, hemicellulose and density were negatively correlated (Table 6). Structural components as well as Ccand Ncare related to species, and there are great dif ferences between species. Nitrogen content in CWD varied between species and decay classes (0.15%–0.82%),and C/N ratios showed a decreasing trend with increasing decay class (Noh et al. 2017). For six common species of deciduous temperate forests over 40 months, mass loss was negatively correlated with initial lignin concentration and positively correlated with initial cellulose concentration and density (Cha et al. 2017). Furthermore, decreased lignin concentration increased thek.

The water content of coarse woody debris (WCWD) is af fected by environmental factors, which in turn af fectkby controlling microbial activity. Water makes up 30–160% of fresh weight, levels which are the most suitable for facilitating microbial growth (Harmon et al. 1986). Research on northern Canadian forest ecosystems found that WCWDaf fectedkwhen the humidity was below 43%, but higher humidities had little ef fect on decomposition (Chambers et al. 2001). Zell et al. ( 2009) carried out a meta-analysis of nine European and 30 North American studies from 1973 to 2005 and proposed an equation for an invariant decay constant, indicating that k was closely related to species,diameter, density, average July temperatures and annual precipitation. The k-value of conifers was 0.63 times higher than that for broad-leaved species and 0.99 times higher than the original value when diameters increased by 10 cm.Mass loss was 1.87 times higher than density loss (impact of fragmentation), and decomposition accelerated with increasing temperature. Rainfall was 1100–1300 mm, and the maximum k occurred at 1226 mm. Mackensen et al. ( 2003)believed thatkreaches a maximum when rainfall range is 1200–1300 mm. Other studies have indicated that large-scale decomposition is related to forest type, days with temperatures exceeding 5 °C, initial decay class, length of coarse debris but not diameter (Russell et al. 2014). Thek-value is not a constant, and dif ferences in research methods and study areas must be considered before using the estimatedk-value to compare dif ferent forest ecosystems. Among the factors that af fectk, the existing observation indices are meaningful for understanding the complex decomposition process and for quantifying the inf luencing factors. Decomposition was def ined as the process of carbon release, structural composition loss, and nitrogen accumulation. These observation indexes provide an indirect means of predictingk.

The decay classes provided a means of better understanding decomposition. Densities of components (Ced, Hcd, Cd),andρrapidly decreased in the early stages of decomposition,WCWDincreased signif icantly in the later stages and N accumulated gradually. With increasing level of decomposition, thek-value tripled,ρdecreased from 0.51 to 0.35 g/cm 3 ,WCWDincreased from 18 to 113%, and N increased from 0.093 to 0.197% (Fig. 4.). In terms of fragmentation,densities (Cd, Lid, Ced, and Hcd) changed more signif icantly between dif ferent decomposition stages than the unadjusted concentrations (Cc, Lic, Cec, and Hcc) did. Most studies showed that CWD decomposition consists of N accumulation and C release. Thus C/N ratios also decreased with ongoing decomposition (Mackensen and Bauhus 2003;Bütler et al. 2007). Some studies have indicated that the nitrogen in woody debris decomposition is related to the decomposition period and tree species. After a 14 year study,Laiho and Prescott ( 1999) found that the nitrogen content(Nc) ofPinus contortaincreased with ongoing decomposition. The NcinPicea engelmanniiwas relatively constant,and inAbies lasiocarpa, it decreased.

Harmon et al. ( 1994) found that N in CWD was released continuously during the f irst seven years of decomposition;the increased nutrient concentration may have resulted from fungal activity and N f ixation by microorganisms. N f ixation enhances growth and possibly microbial growth in most temperate forest ecosystems (Date 1973). Studies in North America have shown an average nitrogen content in CWD of 0.2–2.1 kg ha −1 a −1 , and could be as high as 5.9 kg ha −1 a −1 (Harmon et al. 1986). Coarse woody debris can also af fect circulation of soil elements by dissolving organic C,resulting in N f ixation (Hafner and Grof fman 2005). In USA black pine forests, Busse ( 1994) found that the proportion of downed logs occupied less than 3% of the soil nutrient pool and 68% of the total C pool, but the estimation of nutrient release from CWD was still limited and the formation mechanism remains unclear.The present study showed that, with decomposition, nitrogen levels doubled, carbon contents changed by approximately 50%, and structural components changed by about 74%. CWD was not an important source of available N in the study forest ecosystems. In contrast, the coarse woody debris contributed more to the C pool and relatively little to nutrient cycling. To date, no consistent conclusion can be drawn about C and N turnover in CWD decomposition. Lignin, cellulose and hemicellulose are the main components of plant cell walls, and cellulose is the most easily decomposed, followed by hemicellulose, whereas lignin is the least readily decomposed because of its complex structure (Harmon et al. 1986). In this study, the highest percentage of cellulose was followed by lignin, and hemicellulose was the lowest in CWD, and Lid, Ced, Hcddecreased by 26.7%, 58.7%, and 48.3%, respectively, with increasing decay class, which is consistent with previous research results.

Limitations

By using a space–time substitution method, the decomposition rate of post-f ireLarix gmelinicoarse woody debris was estimated and factors af fecting it were analyzed. There are some limitations:

1. This study only estimated the changes of decomposition of coarse woody debris and measured the density, moisture content, structural composition and C, N contents,without considering human activities, f ire conditions,stand structure, and soil conditions which can af fect the decomposition process.

2. A space–time substitution was used to establish a 33 year chronosequence from seven burned patches. The coarse woody debris ofLarix gmeliniwas still in the initial stages of decomposition 33 years after f ire, and the number of burned patches needs to be increased to establish a longer chronosequence.

3. An average decomposition rate for debris samples was used to represent an average level in the study forests without considering difference between snags and downed boles. In the future, the decomposition process should be described according to dif ferent spatial states of coarse woody debris.

Conclusions

Wildf ires occur frequently in boreal forests and large volumes of coarse woody debris are produced. Quantifying the conversion and decomposition of this material is necessary to clarify the role of boreal forests in carbon cycling. To understand the decomposition process and inf luencing factors, a single-exponential decay model was used to calculate the k-value ofLarix gmelinii.

The average decomposition rate was 0.019 and the average 50% and 95% mass loss will take approximately 41 years and 176 years, respectively. The decomposition rate was higher than that of tree species in high northern latitudes.Because of the short chronosequence, the decomposition rate of coarse woody debris ofLarix gmeliniiwas overestimated.

The rate of decomposition increased with decay class.In the lower decay classes, structural components af fected decomposition rate, and in the higher decay classes, Water content of the debris had a strong inf luence on decomposition. Although nitrogen was a small proportion of the debris(0.093%–0.197%), it had a signif icant inf luence on rate of decomposition. With regards to factors that influenced decomposition, each variable could explain about 42%.

ANOVA results showed that the debris rapidly released nutrients in the early stages of decomposition. Nitrogen gradually accumulated and water contents increased rapidly in the later stages of decomposition with the loss of structural components. C/N and lignin/N ratios were indicators of decomposition rate but they were not suf ficient to distinguish decay classes.

This study provides a new perspective to quantify the interaction between energy f low in coarse woody debris and nitrogen cycling, and provides a deeper understanding of the decomposition process. However, more research is needed to calculate decomposition rates accurately and to analyze environmental factors that af fect the decomposition process.

Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source,provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.