Wood density and carbon and nitrogen concentrations in deadwood of Chamaecyparis obtusa and Cryptomeria japonica

Abstract

Estimating carbon (C) and nitrogen (N) stocks in deadwood in forests nationwide is required for understanding large-scale C and N cycling. To do so requires estimated values of wood density and C and N concentrations. Additionally, parameters that show variation should be examined. In this study, we clarified the estimated values and the variation in three parameters in each decay class of each of two tree species and examined whether dead log diameter and region contribute to variation in the parameters. Data were collected from 73 Chamaecyparis obtusa (Sieb. et Zucc.) Endl. plantations and 66 Cryptomeria japonica D. Don plantations throughout Japan. Wood densities decreased from 386 to 188 kg m⁻³ for C. obtusa and from 334 to 188 kg m⁻³ for C. japonica in decay classes 1–4. The variation in wood density increased with decay class, and the coefficient of variance increased from 13.9% to 46.4% for C. obtusa and from 15.2% to 48.1% for C. japonica. The N concentrations increased from 1.04 to 4.40 g kg⁻¹ for C. obtusa and from 1.11 to 2.97 g kg⁻¹ for C. japonica in decay classes 1–4. The variation in N concentration increased with decay class, and the coefficient of variance increased from 51.9% to 76.7% for C. obtusa and from 50.3% to 70.4% for C. japonica. Log diameter and region contributed to variations in wood density and N concentration in decay classes 1 and 2 for C. obtusa and C. japonica. However, no relationship was observed between regional climates and the two parameters. In contrast, C concentrations ranged from 507 to 535 g kg⁻¹ and were stable with much lower coefficients of variance throughout the decay classes for both tree species. Thus, we recommend that the same C concentration can be adapted for all decay classes of both tree species.

Keywords:

• carbon stock
• decay class
• diameter
• nitrogen stock
• region.

Introduction

Deadwood plays several important roles, such as storing nutrients and water, contributing to soil development, and providing habitats for living things, in forest ecosystems. The carbon (C) stock in deadwood is one of the large C pools that contribute to C cycling in forest ecosystems (IPCC 2006). Moreover, the nitrogen (N) in deadwood is released at slower rates than fine litter and retained within the ecosystem for a long time, suggesting that deadwood is a pool for slow N circulation (Harmon and Hua 1991; Krankina et al. 1999; Guo et al. 2006; Inagaki et al. 2009). Therefore, estimating C and N stocks in deadwood is key to understanding C and N cycling, as well as maintenance mechanisms, within forest ecosystems.

The United Nations Framework Convention on Climate Change requires the ratifying countries of the Kyoto Protocol to report C stocks in deadwood, as it is one of the most important C pools in forested land. Thus, deadwood measurements are included in the national forest inventories of many countries (Köhl et al. 2000; Ferretti et al. 2009; McRoberts et al. 2009; Fahey et al. 2010). These data are also valid for the forest sector in Japan (Takahashi et al. 2010).

Parameters such as wood density, volume, and C and N concentrations are required to measure C and N stocks in deadwood. In practice, deadwood volume is measured using field surveys. However, wood density and C and N concentrations are estimated values that are determined from other data sets. Thus, the estimated values of wood density and C and N concentrations should be calculated on a national scale to determine nationwide inventories of deadwood C and N stocks. In particular, Chamaecyparis obtusa (Sieb. et Zucc.) Endl. and Cryptomeria japonica D. Don are planted throughout forests in Japan, and these two species cover almost 40% of all private planted land (Forestry Agency 2009). Therefore, the estimated values of wood density and C and N concentrations of these two species were examined.

An estimated value of each parameter is prepared for each decay class (typically five classes are used to categorize deadwood properties) for each tree species (Graham and Cromack 1982; Sollins et al. 1987; Harmon et al. 1995; Skovsgaard et al. 2006). The decay class is a classification system for identifying and categorizing decay states and is correlated with the decrease in wood density as the decomposition process progresses (Harmon et al. 1986; Harmon et al. 2008; Larjavaara and Muller-Landau 2010). The tree species determines the differences in initial wood density, wood tissue structure, and chemical composition (Zabel and Morrell 1992). Indeed, Yatskov et al. (2003) reported that decay class, tree species, and their interaction explain 81.2% (68.0%, 6.1%, and 7.1%, respectively) of the variation in changes in deadwood density in Russian boreal forests. Thus, the value of each parameter should be estimated by decay class for C. obtusa and C. japonica. Additionally, these variations result in estimated C and N stock errors. Thus, variations in these parameters should be clarified for each decay class of the two species.

The estimated errors of C and N stocks can be predicted if the structure of variation in the three parameters in each decay class of the two species is known. Wood density and C and N concentrations vary within the same class due to the heterogeneous progression of deadwood decomposition (Pyle and Brown 1999). As factors contributing to heterogeneous deadwood density, Yatskov et al. (2003) indicated that deadwood size and region explain 1.1% and 0.9%, respectively, of the total variation in deadwood density in boreal forests. Log diameter is related to initial wood conditions, such as fresh wood density (Fearnside 1997; Fujiwara et al. 2004) and N concentration (Meerts 2002). Thus, wood density and N concentrations will likely vary with deadwood size. Regional climate influences the deadwood decay process, and the wood density in a given decay class may differ among regions even for the same plantation species (Yatskov et al. 2003; Harmon et al. 2008). Therefore, we examined the effects of both region and log diameter on variations in wood density and the C and N concentrations within each decay class of each of our study tree species.

The objectives of our study were to clarify estimated values of wood density and C and N concentrations, as well as variations in each decay class, for C. obtusa and C. japonica and to examine the contribution of region and dead log diameter to variation in the parameters of each decay class of the two species. To achieve these objectives, we sampled fallen logs produced by noncommercial thinning in C. obtusa and C. japonica plantations throughout Japan, measured the wood density and C and N concentrations, and analyzed their estimated values and variation, as well as the relationship with region and dead log diameter.

Materials and Methods

Study sites

We randomly selected planted forests that had been thinned and that contained fallen logs for which the thinning year had been recorded, resulting in the inclusion of 66 C. obtusa plantations in six prefectures and 73 C. japonica plantations in another six prefectures (hereafter called “regions”). The fallen logs produced by thinning had been left whole on the forest floor in some cases or had been cross cut into short lengths in other cases. The regions were distributed from northern to southern areas of Japan and covered most climate conditions (Fig. 1). The respective annual temperatures in the studied forests of C. obtusa and C. japonica were 8.5–16.0°C and 7.4–15.5°C, annual precipitation amounts were 1150–2682 mm and 1284–2837 mm, and maximum snow depths were 1–186 cm and 2–205 cm (Table 1). We established a 10 × 10-m quadrat as a representative forest floor in each plantation for fallen log sampling. Detailed information about the study sites was presented in Sakai et al. (2008).

Figure 1. Locations of the sampling prefectures for Chamaecyparis obtusa (Sieb. et Zucc.) Endl. (squares) and Cryptomeria japonica D. Don (circles).

Table 1. Location, management history, and climate conditions of the study stands in 12 prefectures

Tree species	Region	Number of sites	Sampling area		Elevation (m a.s.l.)	Stand age (years)	Years since death	Mean annual temperature (°C)	Mean annual precipitation (mm)	Maximum snow depth (cm)
			Latitude, N.	Longitude, E.	Min∼Max		Min∼Max	Min∼Max	Min∼Max	Min∼Max
Chamaecyparis obtusa (Sieb. et Zucc.) Endl.
	Niigata	15	37°37′ ∼ 38°32′	138°49′ ∼ 139°38′	60 ∼ 155	12 ∼ 25	1 ∼ 6	11.1 ∼ 12.8	1786 ∼ 2654	5 ∼ 186
	Saitama	8	36°00′ ∼ 36°05′	138°52′ ∼ 139°09′	430 ∼ 930	20 ∼ 46	1 ∼ 15	8.5 ∼ 11.5	1150 ∼ 1495	9 ∼ 11
	Kyoto	8	35°14′ ∼ 35°20′	135°05′ ∼ 135°30′	80 ∼ 450	27 ∼ 38	1 ∼ 20	11.1 ∼ 13.4	1450 ∼ 1596	21 ∼ 26
	Hiroshima	8	34°31′ ∼ 35°03′	132°51′ ∼ 133°10′	295 ∼ 748	17 ∼ 41	1 ∼ 13	9.3 ∼ 12.7	1284 ∼ 2079	8 ∼ 75
	Mie	8	34°20′ ∼ 34°24′	136°20′ ∼ 136°24′	135 ∼ 693	29 ∼ 53	0 ∼ 13	10.8 ∼ 14.0	2132 ∼ 2682	2 ∼ 3
	Nagasaki	19	32°49′ ∼ 32°58′	130°01′ ∼ 130°11′	50 ∼ 820	24 ∼ 44	1 ∼ 18	10.9 ∼ 16.0	2156 ∼ 2837	1 ∼ 9
Cryptomeria japonica D. Don
	Akita	10	39°42′ ∼ 40°16′	140°24′ ∼ 140°41′	110 ∼ 450	28 ∼ 76	1 ∼ 20	7.4 ∼ 9.5	1738 ∼ 2279	67 ∼ 205
	Ibaraki	15	36°35′ ∼ 36°36′	140°30′ ∼ 140°34′	130 ∼ 390	32 ∼ 46	1 ∼ 17	10.9 ∼ 12.2	1321 ∼ 1487	16 ∼ 17
	Toyama	8	36°33′ ∼ 36°33′	137°10′ ∼ 137°11′	165 ∼ 210	23 ∼ 26	0 ∼ 6	8.2 ∼ 12.3	1367 ∼ 2486	30 ∼ 106
	Chiba	24	35°09′ ∼ 35°15′	139°57′ ∼ 140°03′	110 ∼ 300	19 ∼ 50	1 ∼ 10	13.3 ∼ 14.0	1709 ∼ 1895	11 ∼ 12
	Yamaguchi	8	34°05′ ∼ 34°14′	131°22′ ∼ 131°32′	150 ∼ 390	21 ∼ 46	1 ∼ 18	12.4 ∼ 14.1	1707 ∼ 2122	7 ∼ 20
	Oita	8	33°08′ ∼ 32°54′	131°33′ ∼ 131°54′	130 ∼ 470	24 ∼ 34	1 ∼ 13	12.6 ∼ 15.5	1673 ∼ 2257	2 ∼ 4
a.s.l., above sea level.

Methods

All fallen logs with diameters >5 cm were classified into four diameter classes (5–10, 10–15, 15–20, 20–40 cm) in each quadrat. In each diameter class, three to five fallen logs were collected randomly when three or more logs were present, and all logs were collected when fewer than three logs were present. A disk was cut from a random location on each fallen log within the range of the diameter class using a chain saw or a hand saw. Disks of 10–15 cm in thickness were cut from logs 5–10 cm in diameter, and disks of 3–8 cm in thickness were cut from logs >10 cm in diameter. For C. obtusa, we sampled 60–218 disks of fallen logs in each region and collected 635 disks in total. For C. japonica, we sampled 76–297 fallen logs in each region and collected 893 disks (Appendix 1). All samples were collected between July 2005 and February 2006.

We categorized all fallen logs into five decay classes based on their appearance using the criteria outlined in Table 2 (Graham and Cromack 1982). Samples were cut transversely into disks and diameter and thickness were measured to calculate wood density. Each sample disk was oven-dried at 70°C to a constant weight and then weighed. The sample volume (m³) was calculated using Equation 1 (Smalian's formula; Oh-uchi 1989):

where l is the thickness (in m), and d_b and d_u are the diameter of the lower and upper cutting planes of the disk (in m), respectively. Wood density was determined by dividing sample dry weight by volume.

Table 2. Criteria determining the decay classes of fallen logs

Decay class	Appearance of decaying fallen logs
1	No decay
	Fine twigs remaining
	Bark completely remaining
	Wood is clear or blue, slightly stained
	No moss
2	Slightly decayed
	No fine twigs
	Most bark present
	Sapwood rotted but heartwood mostly sound
3	Moderately decayed, but fallen log could support itself
	Branch stubs remaining
	Some bark present
	Sapwood and heartwood show strong signs of decay
4	Fallen log could no longer support itself
	All bark often absent
	Sapwood often absent and deep red-brown heartwood could be crumbled manually
5	Wood resembles red powder with a disassembled structure
	The bark is completely absent
	The cross section of the bole is no longer round, but elliptical

For measurements of C and N concentrations, we selected 437 samples that were representative of the average wood density of each diameter class of a stand: 5–10, 10–15, 15–20, and 20–40 cm (Appendix 1). Each disk was ground into fine particles (<0.25 mm in diameter) using a cutting mill (Pulverisette 14; Fritsch GmbH, Idar-Oberstein, Germany). In cases of larger diameter disks, before the grinding, a wedge considering proportion of sapwood to heartwood were taken from each disk. Concentrations of C and N were measured using the dry combustion method with an NC analyzer (Vario MAX CN; Elementar Analysensysteme GmbH, Hanau, Germany).

Statistical analysis

Differences in parameters (wood density and C and N concentrations) among decay classes and between tree species were analyzed using a generalized linear model (GLM) with Gaussian errors (i.e., two-way analysis of variance; Equation 2). Decay class and tree species were used as categorical variables. The parameters were also compared between pairs of decay classes and tree species using Tukey's honestly significant difference test. However, decay class 5 and decay classes 4 and 5 were excluded from the analyses of wood density and C and N concentrations, respectively, due to insufficient sample sizes (Appendix 1).

To evaluate the contribution of region and deadwood size to variations in the parameters of each decay class and tree species, we analyzed the differences in parameters among regions and log diameters using a generalized linear model (GLM) with Gaussian errors (i.e., analysis of covariance; Equation 3). The regions were considered categorical variables. A few regions with no or one measurement in each decay class were excluded from the GLM analysis; these included wood density for one region in decay class 3 and three regions in decay class 4, and C and N concentrations for three regions in decay class 3 (see Table S1). We did not use log diameter classes as variables in the analysis, but included the measured values of all samples in a continuous log diameter variable.

We determined the significance of explanatory variables using randomization tests in all GLMs, as some samples were collected from the same stand and were not statistically independent of each other. The F-value distribution was determined by calculating the F-values for fixed effects 10,000 times using randomizations of the response variable (i.e., each parameter). The significance level was determined by comparing the actual F-value to the prepared distribution of F-values. All statistical analyses were conducted using R 2.13.2 (R Development Core Team 2011). A P-value <0.05 was considered significant.

Results

Wood density

The wood density values (mean ± standard deviation) of C. obtusa were 386 ± 54, 302 ± 58, 212 ± 87, and 188 ± 88 kg m⁻³ in decay classes 1–4, respectively (Fig. 2). For C. japonica, the respective wood density values were 334 ± 51, 270 ± 61, 198 ± 62, and 188 ± 91 kg m^–3. The coefficient of variance increased from decay class 1 to 4: 13.9%, 19.4%, 40.9%, and 46.4% for C. obtusa and 15.2%, 22.4%, 31.4%, and 48.1% for C. japonica, respectively. Wood density varied significantly among decay classes (P < 0.001) and between tree species (P < 0.001; Table 3). A significant interaction between decay class and tree species was also detected (P = 0.001; Table 3). Wood density decreased significantly from decay class 1 to 3 for C. obtusa (P < 0.001) and C. japonica (P < 0.001; Fig. 2). However, the standard deviation became larger from decay class 3 to 4 for both tree species, resulting in no significant difference between decay classes 3 and 4 (C. obtusa, P = 0.520; C. japonica, P = 0.906). C. obtusa wood density was significantly higher than that of C. japonica in decay classes 1 (P < 0.001) and 2 (P < 0.001).

Figure 2. Wood density of fallen logs of Chamaecyparis obtusa (Sieb. et Zucc.) Endl. (squares) and Cryptomeria japonica D. Don (circles) in decay classes 1–4. Bars are standard deviations. Different letters indicate significant differences (P < 0.05).

Table 3. Results of a two-way analysis of variance for the wood density of fallen logs by decay class and tree species, and their interaction

Dependent variable	Fixed factor	Df	Mean Sq	F value	P value
Wood density	Decay class	3	1304965	403.95	<0.001
	Tree species	1	635571	196.74	<0.001
	Decay class × Tree species	3	22449	6.95	0.001
	Residuals	1515	3231
Df, degrees of freedom.

Region significantly affected wood density in all C. obtusa decay classes (P ≤ 0.018; Table 4). A significant effect for C. japonica was detected in decay classes 1–3 (P ≤ 0.032). Log diameter significantly affected wood density in C. obtusa decay classes 1 and 2 (P ≤ 0.001) and C. japonica decay classes 1 and 2 (P ≤ 0.046; Table 4). Additionally, a significant interaction between region and log diameter was detected in decay class 4 (P = 0.011) for C. obtusa. In decay classes 1–3 for C. obtusa and C. japonica, the mean wood density showed little tendency to correlate with annual temperature and annual precipitation of the regions (see Fig. S1). Wood density decreased with log diameter, especially in decay class 1 (Fig. 3).

Figure 3. Wood density of fallen logs changed with log diameter in decay class 1 (filled circles), decay class 2 (open circles) and decay class 3 (triangles) of Chamaecyparis obtusa (Sieb. et Zucc.) Endl. and Cryptomeria japonica D. Don.

Table 4. Results of an analysis of covariance for the wood density of fallen logs by region and log diameter, and their interaction, in each decay class and tree species

Tree species	Decay class	Factor	Df	Mean Sq	F value	P value
Chamaecyparis obtusa (Sieb. et Zucc.) Endl.	1	Region	5	7115	2.82	0.017
		Diameter	1	127373	50.54	<0.001
		Region × Diameter	5	2473	0.98	0.433
		Residuals	376	2520
	2	Region	5	10688	3.46	0.004
		Diameter	1	34548	11.18	0.001
		Region × Diameter	5	2079	0.67	0.639
		Residuals	181	3091
	3	Region	4	46305	14.41	<0.001
		Diameter	1	6407	1.99	0.160
		Region × Diameter	4	7271	2.26	0.104
		Residuals	35	3214
	4	Region	3	11701	15.74	0.018
		Diameter	1	3937	5.30	0.101
		Region × Diameter	1	20191	27.16	0.011
		Residuals	3	744
Cryptomeria japonica D. Don	1	Region	5	5892	2.44	0.032
		Diameter	1	68447	28.29	<0.001
		Region × Diameter	5	3834	1.59	0.162
		Residuals	577	2420
	2	Region	5	10523	3.11	0.010
		Diameter	1	14047	4.16	0.046
		Region × Diameter	5	5561	1.65	0.160
		Residuals	186	3381
	3	Region	5	15668	5.14	0.001
		Diameter	1	562	0.19	0.678
		Region × Diameter	5	3308	1.09	0.365
		Residuals	65	3048
	4	Region	4	11286	1.42	0.277
		Diameter	1	2783	0.35	0.553
		Region × Diameter	3	7249	0.91	0.449
		Residuals	15	7950
Df, degrees of freedom.

Carbon concentration

The C. obtusa C concentrations (mean ± sample standard deviation) were 518 ± 11, 527 ± 7, 535 ± 17, and 527 ± 9 g kg⁻¹ in decay classes 1–4, respectively (Fig. 4). The respective values of C concentration for C. japonica were 513 ± 16, 507 ± 14, 511 ± 14, and 527 ± 13 g kg⁻¹. The coefficients of variance in C concentration were small compared to those of wood density: 2.1%, 1.4%, 3.2%, and 1.6% for C. obtusa, and 3.1%, 2.8%, 2.8%, and 2.5% for C. japonica in order of decay classes. Carbon concentration varied significantly among decay classes (P = 0.005; Table 5) and between tree species (P < 0.001). A significant interaction between decay class and tree species was also observed (P < 0.001). Carbon concentrations in C. obtusa increased significantly with progression from decay class 1 to 3 (decay classes 1 and 2, P < 0.001; decay classes 2 and 3, P = 0.043; Fig. 4). Carbon concentrations differed significantly for C. japonica between decay classes 1 and 2 (P = 0.004) and between decay classes 3 and 4 (P = 0.040). C. obtusa exhibited significantly higher C concentrations than C. japonica for decay classes 1–3 (decay class 1, P = 0.003; decay class 2, P < 0.001; decay class 3, P < 0.001).

Figure 4. Fallen log carbon concentrations of Chamaecyparis obtusa (Sieb. et Zucc.) Endl. (squares) and Cryptomeria japonica D. Don (circles) in decay classes 1–4. Bars are sample standard deviations. Different letters indicate significant differences (P < 0.05).

Table 5. Results of a two-way analysis of variance for the carbon concentrations of fallen logs by decay class and tree species, and their interaction

Dependent variable	Fixed factor	Df	Mean Sq	F value	P value
Carbon concentration	Decay class	3	0.85	4.53	0.005
	Tree species	1	12.15	64.98	<0.001
	Decay class × Tree species	3	2.11	11.28	<0.001
	Residuals	424	0.19
Df, degrees of freedom.

Region significantly affected C concentration in C. obtusa decay class 1 (P < 0.001) and in all decay classes of C. japonica (P < 0.001; Table 6). Log diameter significantly affected C concentration in C. obtusa decay class 1 (P < 0.001) and in C. japonica decay classes 1 (P < 0.001) and 2 (P < 0.001; Table 6). Mean C concentrations in all decay classes of both tree species showed little tendency to correlate with annual regional temperature and precipitation (see Fig. S2). Carbon concentrations increased with log diameter in all decay classes of both species, but this tendency was not clear (Fig. 5).

Figure 5. Concentrations of (a) carbon and (b) nitrogen changed with diameter in decay class 1 (filled circles), decay class 2 (open circles) and decay class 3 (triangles) of Chamaecyparis obtusa (Sieb. et Zucc.) Endl. and Cryptomeria japonica D. Don.

Table 6. Results of an analysis of covariance for the carbon concentrations of fallen logs by region and log diameter, and their interaction, in each decay class and tree species

Tree species	Decay class	Factor	Df	Mean Sq	F value	P value
Chamaecyparis obtusa (Sieb. et Zucc.) Endl.	1	Region	5	1.76	61.19	<0.001
		Diameter	1	0.66	23.04	<0.001
		Region × Diameter	5	0.04	1.33	0.260
		Residuals	94	0.03
	2	Region	5	0.12	2.43	0.054
		Diameter	1	0.01	0.11	0.743
		Region × Diameter	4	0.03	0.58	0.674
		Residuals	43	0.05
	3	Region	3	0.34	0.87	0.502
		Diameter	1	0.04	0.10	0.724
		Region × Diameter	3	0.09	0.23	0.756
		Residuals	8	0.40
Cryptomeria japonica D. Don	1	Region	5	6.82	135.81	<0.001
		Diameter	1	1.02	20.33	<0.001
		Region × Diameter	5	0.04	0.88	0.511
		Residuals	157	0.05
	2	Region	5	1.84	54.35	< 0.001
		Diameter	1	0.47	13.83	<0.001
		Region × Diameter	4	0.08	2.32	0.068
		Residuals	47	0.03
	3	Region	4	0.80	12.87	<0.001
		Diameter	1	0.00	0.01	0.946
		Region × Diameter	3	0.01	0.19	0.920
		Residuals	12	0.06
Df, degrees of freedom.

Nitrogen concentration

The N concentrations (mean ± standard deviation) of C. obtusa were 1.04 ± 0.54, 1.17 ± 0.64, 1.84 ± 1.16, and 4.40 ± 2.32 g kg⁻¹ in decay classes 1–4, respectively (Fig. 6). The respective N concentration values for C. japonica were 1.11 ± 0.56, 1.43 ± 0.80, 1.52 ± 1.04, and 2.97 ± 2.10 g kg⁻¹ (Fig. 6). The N concentration coefficients of variance were high compared to those of wood density and were highest in decay class 4: 51.9%, 54.3%, 62.9%, and 76.7% for C. obtusa and 50.3%, 56.1%, 68.2%, and 70.4% for C. japonica in order of decay classes. Nitrogen concentrations differed significantly among decay classes (P < 0.001; Table 7), and the interaction between decay class and tree species was also significant (P = 0.007). In both species, N concentrations increased dramatically from decay class 3 to 4 (C. obtusa, P < 0.001; C. japonica, P = 0.003), although N concentrations in decay class 4 had a large standard deviation. In contrast, the increase in N concentration from decay class 1 to 3 was smaller than that from decay class 3 to 4.

Figure 6. Fallen log nitrogen concentrations of Chamaecyparis obtusa (Sieb. et Zucc.) Endl. (squares) and Cryptomeria japonica D. Don (circles) in decay classes 1–4. Bars are sample standard deviations. Different letters indicate significant differences (P < 0.05).

Table 7. Results of a two-way analysis of variance for the nitrogen concentrations of fallen logs by decay class and tree species, and their interaction

Dependent variable	Fixed factor	Df	Mean Sq	F value	P value
Nitrogen concentration	Decay class	3	2.05 E−02	40.86	<0.001
	Tree species	1	3.62 E−04	0.72	0.399
	Decay class × Tree species	3	2.30 E−03	4.59	0.007
	Residuals	424	5.02 E−04
Df, degrees of freedom.

Region significantly affected N concentration in decay classes 1 and 2 for both C. obtusa and C. japonica (decay class 1, P < 0.001; decay class 2, P < 0.001; Table 8). Log diameter also significantly affected N concentration in decay classes 1 (P < 0.001) and 2 (P < 0.001; Table 8). Additionally, a significant interaction between region and log diameter was observed in C. obtusa decay class 2 (P = 0.033). Nitrogen concentrations were not related to annual regional temperature and precipitation (see Fig. S2). The N concentrations showed large variations but decreased with log diameter (Fig. 5).

Table 8. Results of an analysis of covariance for the nitrogen concentrations of fallen logs by region and log diameter, and their interaction, in each decay class and tree species

Tree species	Decay class	Factor	Df	Mean Sq	F value	P value
Chamaecyparis obtusa (Sieb. et Zucc.) Endl.	1	Region	5	2.71 E−03	21.09	<0.001
		Diameter	1	4.30 E−03	33.52	<0.001
		Region × Diameter	5	1.52 E−04	1.18	0.319
		Residuals	94	1.28 E−04
	2	Region	5	2.95 E−03	34.31	<0.001
		Diameter	1	2.05 E−03	23.81	<0.001
		Region × Diameter	4	2.60 E−04	3.02	0.033
		Residuals	43	8.60 E−05
	3	Region	3	2.33 E−03	2.70	0.119
		Diameter	1	3.57 E−03	4.14	0.078
		Region × Diameter	3	9.25 E−04	1.07	0.404
		Residuals	8	8.62 E−04
Cryptomeria japonica D. Don	1	Region	5	3.79 E−03	20.79	<0.001
		Diameter	1	4.23 E−03	23.20	<0.001
		Region × Diameter	5	2.01 E−04	1.10	0.369
		Residuals	157	1.82 E−04
	2	Region	5	5.51 E−03	47.59	<0.001
		Diameter	1	2.91 E−03	25.17	<0.001
		Region × Diameter	4	2.37 E−04	2.05	0.111
		Residuals	47	1.16 E−04
	3	Region	4	6.02 E−04	0.49	0.721
		Diameter	1	1.46 E−03	1.18	0.301
		Region × Diameter	3	9.20 E−04	0.74	0.478
		Residuals	12	1.24 E−03
Df, degrees of freedom.

Discussion

Wood density

As decay progressed, the mean wood densities of fallen logs decreased while the variation in wood density increased (Fig. 2), indicating that heterogeneity in the decay process leads to larger variation in wood density later in decomposition.

The large variation in wood density late in the decay process resulted in no difference between the two study species in the late decay classes (Fig. 2). In contrast, wood density differed between C. obtusa and C. japonica in the early decay classes (Fig. 2). These findings indicate that tree species identity is essential in the early but not in the late decay classes. The difference in wood density between the two species is consistent with that observed for fresh logs (Fujiwara et al. 2007). Thus, we suggest that initial wood density affects interspecific differences in wood density in the early decay classes.

In decay classes 1 and 2, variation in wood density was influenced by log diameter (Fig. 3, Table 4). The fresh wood densities of C. obtusa and C. japonica increased with immature wood ratios, which are related to fresh log diameter (Kano 1962; Watanabe 1978; Minato et al. 1989; Fujiwara et al. 2004). Therefore, differences in wood maturity may be related to the relationship between wood density and dead log diameter in the early decay classes.

Fujiwara et al. (2007) observed no significant regional effect on C. obtusa and C. japonica fresh wood density. However, wood densities were affected by region in decay classes 1–4 of both species, with the exception of decay class 4 of C. japonica (Table 4). Regional variation in environmental conditions such as temperature and moisture may influence the wood structure, species composition, and microbe activity of decomposing deadwood. However, the mean values by region were not clearly related to regional climate parameters, such as annual temperature and precipitation. Thus, the effects of regional climate on initial wood density and the decomposition process may not be important. Further research is needed to clarify the cause(s) of regional differences in wood density, which may include differences in site environments (Yatskov et al. 2003; Harmon et al. 2008) and local cultivars (Yahata et al. 1987; Miyajima 1989).

Carbon concentration

Carbon concentrations varied among decay classes, but no consistent pattern was observed (Table 5, Fig. 4). Moreover, the variation in C concentrations among decay classes appeared to be less than that between the two species (see F-values in Table 5).

Carbon concentrations in deadwood were higher for C. obtusa than for C japonica. Strictly speaking, this result suggests that different C concentration parameters should be used for C. obtusa and C. japonica. However, the coefficient of variance was smaller for C concentration than for wood density, and C stock was estimated using not only C concentration, but also wood density. Therefore, the same C concentration can be used to estimate the C stock of both C. obtusa and C. japonica. Given the smaller variation among decay classes than between tree species, we recommend the use of a C concentration of 519 g kg⁻¹, which is the mean value across all decay classes for both tree species.

Variations in C concentration were affected by dead log diameter in decay classes 1 and 2 and by region in decay classes 1–3 (Table 6). The mean C concentrations were not related to regional climate, indicating that climate conditions had no effect. Moreover, C concentrations increased with log diameter in all decay classes. However, the variation in C concentrations was small, even within each decay class of each species (Table 5). Thus, accounting for dead log diameter and region may not be particularly effective for explaining variation in C concentrations.

Nitrogen concentration

A significant increase in N concentration was observed as logs decayed (Fig. 6), highlighting a general trend in decaying logs (Lambert et al. 1980; Palviainen et al. 2010). Nitrogen concentrations increased dramatically from decay class 3 to 4, although wood density did not change between these two decay classes (Fig. 2). Thus, we hypothesize that the increase in N concentration was partially caused by fungal N translocation (Hart 1999) and bacterial N fixation (Hicks et al. 2003), particularly late in the decay process. On the other hand, amounts of N per initial volume (N density) were almost constant from decay class 1 to 4, except decay class 4 of C. obtusa (see Fig. S3). Therefore, the increase in N concentration caused by fungal N translocation (Hart 1999) and bacterial N fixation (Hicks et al. 2003) may be restricted in the deadwood decay process (Inagaki et al. 2009). The high N density of decay class 4 of C. obtusa had a large variation, and further study is needed to confirm the increase.

N concentrations did not differ between the two study species in any decay class (Table 7). Therefore, the same N concentration parameter can be used to estimate N stocks in both C. obtusa and C. japonica plantations.

Although we could not analyze factors contributing to N concentration in decay class 4 due to insufficient sample sizes, the variation in N concentration was influenced by dead log diameter and region in decay classes 1 and 2, but not in decay class 3 (Table 8). In fresh logs, N concentration varies with the relative proportions of bark and sapwood with higher N concentrations, and heartwood with lower N concentrations (Harada and Sato 1966; Meerts 2002), which may have led to the association of higher N concentration with smaller dead log diameter in the early decay classes. We also observed a significant effect of region on the variation in N concentrations. However, we found no trend associated with climate conditions such as annual temperature and precipitation. Additionally, the variation in N concentration increased dramatically in the late decay classes (Fig. 6). Thus, we suggest that the variation in N dynamics may be larger among logs than among regions in the late decay process.

Conclusions

Our results show the estimated values and the variation of wood density and C and N concentrations by decay class and tree species. The estimated parameter values can be applied to estimate C and N stocks of deadwood by decay class in C. obtusa and C. japonica plantation forests in Japan. We recommend the consideration of interspecific differences in wood density, but not N concentration, in early decay classes. However, we should recognize that variations in wood density and N concentration increased with decay class and were influenced by region and log diameter. In contrast, we recommend the use of a C concentration of 519 g kg⁻¹ for C. obtusa and C. japonica in Japan, which is the mean value across all decay classes of both tree species, because the variation in C concentrations was small.

In this study, we were not able to collect a sufficient number of fallen logs in decay class 5. In Japan, few deadwood logs decompose into decay class 5 after thinning. Further research should seek to determine the estimated parameters for decay class 5. On the other hand, similar wood density values have been found for decay classes 4 and 5 of softwoods (131–201 and 108–220 kg m⁻³, respectively) (Laiho and Prescott 1999; Yatskov et al. 2003; Tobin et al. 2007; Harmon et al. 2008). Thus, we may be able to apply the wood density value for decay class 4 (188 kg m⁻³) to deadwood in decay class 5. Moreover, the C concentration of 519 g kg⁻¹ may be used for deadwood in decay class 5 because C concentrations were almost constant during decomposition. Few studies have reported N concentrations in decay classes 4 and 5, prohibiting the use of the N concentration for decay class 4 as that for decay class 5.

The parameter values obtained in the present study enable the estimation of current C and N stocks in deadwood by decay class in C. obtusa and C. japonica plantation forests in Japan. Further studies that estimate decay rates and deadwood inputs are needed to estimate changes in C and N stocks.

Acknowledgments

This research was funded by the Forestry Agency, Ministry of Agriculture, Forestry and Fisheries, Japan (the Organization Development Project for Measurement and Use of Forest Sinks and the National Forest Soil Carbon Inventory Project). We thank Satoshi Sawata, Yoshitaka Houjyou, Yasuhiko Tamaki, Tamami Shirai, Hiroshi Takeda, Hideharu Aiura, Hiromasa Shimada, Teppei Iwatsuki, Atsushi Yamaba, Takanobu Yamada, Hajime Maeda, and Masamichi Muro for their support during field research. We also thank Dr. Shinji Kaneko and Dr. Mitsuo Matsumoto for their support.

Supplementary Material

Table S1 and Figs. S1-S3 are available online at http://dx.doi.org/10.1080/00380768.2012.710526

Appendix 1

Number of disks collected in each decay class and region for the measurement of wood density. Numbers of samples submitted to carbon (C) and nitrogen (N) analysis are presented in parentheses

		Decay class
Tree	Region	1	2	3	4	5
Chamaecyparis obtuse (Sieb. et Zucc.) Endl.	Niigata	115(30)	8(1)	0(0)	0(0)	0(0)
	Saitama	16(5)	37(11)	2(0)	5(3)	0(0)
	Kyoto	45(13)	21(6)	15(4)	1(0)	0(0)
	Hiroshima	42(13)	15(5)	6(2)	0(0)	0(0)
	Mie	52(16)	31(9)	5(3)	1(1)	0(0)
	Nagasaki	118(30)	81(22)	17(7)	2(1)	0(0)
Cryptomeria japonica D. Don	Akita	56(18)	35(10)	13(4)	4(1)	0(0)
	Ibaraki	175(48)	10(3)	7(1)	0(0)	0(0)
	Toyama	70(19)	39(9)	3(0)	4(1)	0(0)
	Chiba	163(54)	99(31)	34(9)	1(0)	0(0)
	Yamaguchi	57(17)	5(1)	10(4)	4(0)	0(0)
	Oita	68(16)	10(4)	10(3)	11(2)	5(0)