Separation of soil respiration into CO₂ emission sources using ¹³C natural abundance in a deciduous broad-leaved forest in Japan

Abstract

In determining the soil and ecosystem carbon balance, it is necessary to distinguish between autotrophic respiration and heterotrophic respiration. We attempted to measure the contribution of CO₂ emissions from plant roots (R_RHI), from soil organic matter (R_SOM), and from litter (R_L) to CO₂ emissions from the forest floor (soil respiration; R_S) in a deciduous forest of oak (Quercus serrata Thunb.) and hornbeams (Carpinus laxiflora Sieb. et Zucc. Bl., Carpinus tschonoskii Maxim. and Carpinus japonica Bl.) on Andosols in Japan, using a ¹³C natural abundance technique. The ¹³C natural abundances of roots (δ_RHI), litter (δ_L) and SOM (δ_SOM) in the surface soil were −28.9, −30.1 and −24.3‰, respectively. This means that the differences between δ_SOM and δ_RHI are large enough to calculate the contributions of R_RHI, R_SOM and R_L to R_S based on the mass balance of the CO₂ isotope ratios. R_RHI and R_SOM had close relationships with soil temperature, and R_L was influenced by soil temperature and moisture. In summer, under high soil temperatures, R_RHI and R_SOM were the predominant sources of R_S and the proportion of R_RHI to R_SOM to R_L was 51:44:5. In winter, R_L was predominant and the proportion of R_RHI to R_SOM to R_L was 20:11:69. The estimated annual emissions of R_RHI, R_SOM and R_L were 1.45, 2.10 and 1.30 Mg C ha⁻¹, respectively; thus, the proportion of R_RHI to R_SOM to R_L was 30:43:27 on a whole-year basis.

Key words:

• Andosols
• autotrophic respiration
• forest soil
• soil organic matter decomposition
• ¹³C stable isotope

INTRODUCTION

In recent decades, in an attempt to determine the exact capacity of carbon fixation for forests, researchers have examined the carbon balance in forest ecosystems. Net ecosystem production (NEP), calculated as net primary production minus heterotrophic respiration, has been estimated in various forest ecosystems using ecological mass production methods and meteorological flux observations (Baldocchi 2003; Dore et al. 2003; Falge et al. 2002). A challenge in determining NEP is to estimate CO₂ flux from belowground processes; this flux is composed of heterotrophic respiration from decomposing soil organic matter (SOM) and litter, and autotrophic respiration from plant roots in the soil (Raich and Schlesinger 1992). Although separation of these respiration components is crucial for determining NEP, methodological difficulties have not been solved at this stage (Hanson et al. 2000).

Separation of heterotrophic respiration from autotrophic respiration has been carried out using various methods, such as the trenching method, the integrated method, and the isotope method (Hanson et al. 2000). Of these, the isotope method has the advantage of disturbing the field very little (Meharg 1994). When the isotope ratio of the soil differs from that of the plant roots, the isotope method can be used to estimate the relative contributions of heterotrophic and autotrophic respiration to the total respiration by measuring the isotope ratio of soil CO₂ or the isotope ratio of soil-respired CO₂ (Amundson et al. 1998; Hanson et al. 2000). The artificially controlled isotope method has been applied to some forest ecosystems on which free-air CO₂ enrichment experiments have been carried out (Andrews et al. 1999; Matamala and Schlesinger 2000) and to pot experiments (Robinson and Scrimgeour 1995). The ¹³C natural abundance method, which uses the difference in ¹³C ratio between growing plants and soil organic carbon, is simple and has been applied mainly to agricultural fields of C4 plants, such as maize, which have higher isotope ratios than those of their soils (Rochette and Flanagan 1997; Rochette et al. 1999). In forest ecosystems, however, the natural abundance method has not been used because the difference in stable carbon isotope ratios (¹³C/¹²C) between vegetation and SOM is usually less than 1‰ (Boutton 1996) and the isotope ratio of CO₂ emitted from soil respiration is nearly the same as that of CO₂ emitted by vegetation (Cerling et al. 1991).

Recent reports (Ishizuka et al. 1999; Yoneyama et al. 2001) suggest that the SOM in Andosols (FAO et al. 1998) and in Black soil groups (Forest Soil Division 1976) have higher carbon isotope ratios than those in Cambisols and Brown forest soil groups in Japan. Up to approximately half of the SOM in Andosols originated from Japanese pampas grass (Miscanthus sinensis Anderss.), which was a C4 plant that grew in the past (Hiradate et al. 2004; Ishizuka et al. 1999; Kawamuro and Torii 1986; Shindo et al. 1986). Based on this finding, we expect the difference in carbon isotope ratios between vegetation and SOM in Andosols to be large enough to measure using the ¹³C natural abundance technique. If this method proves successful, it will be possible to distinguish between autotrophic respiration and heterotrophic respiration in areas of Andosols without disturbing the site.

The objectives of the present study were to: (1) distinguish between autotrophic respiration and total soil respiration in forest ecosystems on Andosols by measuring the natural abundance of stable isotopes, (2) examine seasonal changes in the contributions of different CO₂ sources (litter, roots and SOM) to soil respiration. We also discuss limitations in the application of this method.

MATERIALS AND METHODS

Study sites

We established a study site in a deciduous broad-leaved forest in Sakuragawa City, Ibaraki Prefecture, central Japan (36°19′N, 140°9′E). The elevation of the site is 470 m. According to the meteorological station data at Kasama, which is located approximately 10 km northeast from the study site (Japan Meteorological Agency: http://www.data.kishou.go.jp), the mean annual temperature, maximum monthly temperature (August), and minimum monthly temperature (January) in the area are 13.1, 29.3 and −3.5°C, respectively. The average annual precipitation for the previous 20-year period is 1,330 mm, with high precipitation from April to October and low precipitation from November to March. Figure 1 shows seasonal fluctuation in the daily air temperature and the weekly precipitation. In winter, the maximum snow depth is approximately 10 cm and the maximum freezing depth of the soil is approximately 5 cm.

The site is a secondary deciduous broad-leaved forest, mainly consisting of oak (Quercus serrata Thunb.) and hornbeams (Carpinus laxiflora Sieb. et Zucc. Bl., Carpinus tschonoskii Maxim. and Carpinus japonica Bl.), with dwarf bamboo (Pleioblastus chino Franch. et Savat Makino) on the forest floor. In this area, Japanese cypress (Chamaecyparis obtusa Endl.), Japanese red pine (Pinus densiflora S. et Z.) and Japanese cedar (Cryptomeria japonica D. Don) were reforested after clear-cutting in 1969, but did not grow well. After this, broad-leaved trees came to dominate again. The average tree height of the dominant species, stem density and basal area are 8.9 m, 3,500 stems ha⁻¹, and 15.6 m² ha⁻¹, respectively. The site is on a well-drained slope with an inclination of 15°. The soil type is light-colored Black soil (Forest Soil Division 1976) and Fulvic Andosols (Food and Agriculture Organization et al. 1998). The soil has a loamy clay texture with little gravel and the A horizon extends to a depth of 20 cm.

Sampling methods

The CO₂ emitted from the forest floor (soil respiration, R_S) was measured using the static closed chamber method (Sakata et al. 2004). Three chambers (baseless stainless steel cylinders with an internal diameter of 40 cm and a height of 15 cm) were inserted into the forest floor to approximately 5 cm in depth. Three chambers were installed at the site. To measure soil respiration rate, 50 mL of gas in the chamber was sampled at 0, 10, 20 and 40 min after the chamber was sealed. The gas sample was captured in a vacuum vial that had been evacuated to less than 10 Pa. Vials were taken to the laboratory for analysis of CO₂ concentration and stable isotope ratios. The gas was sampled approximately every 2 months from July 2001 to June 2002. After gas sampling, three surface soil (0–5 cm) samples were collected from near the chambers and the soil moister contents were calculated after drying the soil at 105°C for 24 h in the laboratory. The field soil temperature was measured every hour using a thermometer logger (Hioki, Model 3632, Japan) at 5 cm depth (Fig. 1).

The CO₂ emission from litter (R_L) was measured in the field at the same time as the soil respiration was measured. Litter was cut from three circles, each 30 cm in diameter, and transferred to a plastic cylindrical container (30 cm in diameter, 15 cm in height) with as little disturbance as possible. After the container was sealed, 40 mL of gas in the container was sampled after 0, 15 and 30 min. The gas samples were captured in a vacuum vial and taken to the laboratory for analysis.

Litter was cut from three squares, each 20 cm × 20 cm, and collected in July 2001 to measure its deposition mass. Litterfall was collected using five litter traps (42 cm × 30 cm) from September to November 2001, which is the defoliation period. Fine roots (diameter ≤ 2 mm) were collected from the A horizon (0–20 cm depth) in May (three replications) and November 2001 (ten replications) and washed to remove adhering soil. There were few fine roots in the litter layer. We did not separate live roots from dead roots. Dry weights of litter and fine roots were measured after oven-drying at 70°C for 48 h. Soil samples and soil cores (100 mL) were collected from the A horizon in May 2001, with three replications for each. The gas phase volume of the soil core was measured using a gas pycnometer (Daiki, DIK1121, Saitama, Japan). Then, the bulk density of the soil was calculated after drying the soil at 105°C for 24 h. The dried litter, fine root, and soil samples were ground in an agate mortar before analysis.

Gas and isotope analysis

The CO₂ concentration was measured using a gas chromatograph (GC14B-TCD, Shimadzu, Kyoto, Japan) or an infrared gas analyzer (ZFP9, Fuji Electric, Tokyo, Japan). The stable carbon isotope ratio of the CO₂ in the gas samples was analyzed using combustion gas chromatograph coupled isotope ratio mass spectrometry (HP6890-MAT252, Thermo Finnigan, Bremen, Germany). The isotope ratios of the litter, soil and fine roots were determined using automated nitrogen–carbon analysis mass spectrometry (NC2500-MAT252, Thermo Finnigan). Isotopic compositions are expressed using delta notation:

(1)

where δ¹³C is the stable isotope ratio in parts per thousand (‰), ¹³C/¹²C_sample is ¹³C/¹²C for the sample, and ¹³C/¹²C_standard is ¹³C/¹²C for the Pee Dee Belemnite (PDB).

Calculating respiration, emission and isotope ratios

The CO₂ concentrations after 0, 10 and 20 min in the chamber were used to calculate the rates of soil respiration and the CO₂ emission from litter by substituting those concentrations into the following equation (Hutchinson and Mosier 1981):

(2)

where R_S (kg CO₂ m⁻² s⁻¹) is the soil respiration rate, R_L (kg CO₂ m⁻² s⁻¹) is the CO₂ emission rate from the litter, ρ (kg m⁻³) is the gas density (1.96 for CO₂), V (m³) is the volume of the chamber, A (m²) is the area of the chamber's base, t (s) is the gas sampling interval (10 min in this case), C_i (kg m⁻³) is the CO₂ concentration in the chamber at time i (min), and T (°C) is the air temperature.

The δ¹³C value of the CO₂ emitted from the forest floor was calculated by substituting the CO₂ concentrations and the δ¹³C values into the following mass balance expression:

(3)

where δR_S (‰) is the δ¹³C value of the CO₂ emitted from the forest floor and δC _i (‰) is the δ¹³C value of the gas in the chamber at time i (min).

Soil respiration rate is the sum of the CO₂ emission rates of roots, SOM and litter:

where R_SOM and R_RHI are the emission rates from SOM and roots, respectively.

Under normal conditions, such as when the soil is well aerated and in a steady state, the δ¹³C value of the soil-respired CO₂ is approximately equal to that of the CO₂ emitted from the source (Amundson et al. 1998). Therefore, we can make the following mass balance equation for the CO₂ isotope ratio:

where δR_S, δR_RHI, δR_SOM and δR_L (‰) are the δ¹³C values of CO₂ emitted from the forest floor, roots, SOM and litter, respectively, and f_RHI, f_SOM and f_L are the proportions of R_RHI, R_SOM and R_L in R_S, respectively.

Based on reports that the δ¹³C values of CO₂ emitted from roots are nearly the same as those emitted from other plant organs (Amundson et al. 1998; Cerling et al. 1991; Lin and Ehleringer 1997) and that isotope fractionation by soil microbial respiration is negligible (Balesdent and Mariotti 1996; Blair et al. 1985; Lin et al. 1999), we assumed that the δ¹³C values of CO₂ emitted from the litter, roots and SOM were the same as the δ¹³C values of organic matter in their sources Under this assumption, Eq. 5 can be converted to Eq. 5′, and Eq. 6 and Eq. 5′ can then be transformed into Eq. 7 and Eq. 8:

(7)

(8)

Figure 1  Seasonal fluctuations in (a) air temperature, (b) precipitation, (c) soil temperature and soil moisture content at 5 cm depth and (d) CO2 emission rates from forest floor (RS) and litter (RL). Error bars represent the standard error.

It is noted that R_RHI in the present study is actually composed of root respiration and heterotrophic respiration in the rhizosphere to some extent, because rhizosphere microbes metabolize root exudates and root deposits with quick turnover rates; metabolism that is difficult to distinguish from root respiration. Consequently, R_SOM originates from SOM in the non-rhizosphere soil.

Calculated errors of contribution ratios of R_RHI and R_SOM

If we make the reasonable assumption that δR_S, δ_RHI, δ_SOM and δ_L are independently measured, then a first-order Taylor series approximation of the variance of f_RHI evaluated at δR_S, δ_RHI, δ_SOM and δ_L can be calculated using partial derivatives (Taylor, 1997):

(9)

which reduces to:

(10)

where σ²δR_S, σ²δ_RHI, σ²δ_SOM and σ²δ_L (‰) represent variances of δR_S, δ_RHI, δ_SOM and δ_L, respectively (i.e. the square of the standard errors [SE]: Phillips and Gregg 2001). In the same way, the variance of f_SOM can be calculated:

(11)

RESULTS

Masses and stable carbon isotope ratios of soil, litter and roots

In the A horizon at this site, the bulk density was 357 ± 19.3 kg m⁻³ and the gas phase was 0.48 ± 0.02 m³ m⁻³. The surface soil had relatively high permeability. The litter in July, 0.440 kg m⁻² (Table 1), was almost the same as the litterfall that occurred in the defoliation period from September to November (0.422 ± 0.812 kg m⁻²). In the A horizon (0–20 cm) the total mass of fine roots (0–2 mm in diameter) was 0.306 kg m⁻² in May and 0.247 kg m⁻² in November (Table 1), which is not significantly different (t-test, P = 0.25). The δ¹³C values of plant roots (δ_RHI), litter (δ_L) and SOM (δ_SOM) were −30.1, −28.9 and −24.3‰, respectively. Thus, δ_SOM was higher than δ_RHI and δ_L, and the δ¹³C values of the litter were nearly equal to those of the plant roots.

Fluctuations in soil respiration and annual CO₂ emission

The soil respiration rate (R_S) showed seasonal fluctuations, large in summer and autumn (July and September) and small in winter (December and February) (Fig. 1). R_S correlated exponentially with soil temperature:

where T_S (°C) is the soil temperature at 5 cm depth. Seasonal changes in soil moisture and temperature had the opposite pattern, but R_S was not significantly correlated

Table 1 Masses and stable carbon isotope ratios of soil, litter and roots

	Dry weight (kg m^−2)	Carbon content (kg kg^−3)	δ^13C value (‰)
Litter	0.440 ± 0.083	0.485 ± 0.009	30.1 ± 0.1
Fine roots†	0.306 ± 0.096‡ 0.247 ± 0.049‡	0.435 ± 0.004	28.9 ± 0.1
Soil†	71.4 ± 3.86	0.104 ± 0.002	24.3 ± 0.5
†Samples were collected from a depth of 0–20 cm.
^‡The upper and lower data were sampled in July 2001 and December 2001, respectively.

with the soil moisture (r ² = 0.416, P = 0.118). The emission rates from the litter (R_L) peaked in two seasons: autumn (September and November) and spring (April). They were low in summer (July) and winter. Thus, soil temperature alone had no correlation with R_L (r ² = 0.069, P = 0.572). Soil moisture also influenced R_L, and a multiple correlation of soil temperature and soil moisture with R_L was found, although it was not significant (r ² = 0.679, P = 0.097).

The annual R_S was 1.78 kg CO₂ m⁻² (4.85 Mg C ha⁻¹), which was estimated by summing up the hourly soil respiration calculated by inserting hourly soil temperature into Eq. 12 from July 2001 to June 2002. Although our observations were conducted over the course of only 1 year, this annual CO₂ emission was comparable to previous reports from temperate forests in Japan (3.1–10.6 Mg C ha⁻¹; Ishizuka et al. 2006) and to a worldwide review of temperate deciduous forests (3.0–14.1 Mg C ha⁻¹; Hibbard et al. 2005).

Fluctuations in δ¹³C values and error analysis

The δ¹³C values of soil respiration (δR_S) also showed seasonal fluctuations (Fig. 2). The δR_S values ranged from −34.7 to −26.2‰, but were lower than those of SOM (δ_SOM) and higher than those of plant roots (δ_RHI) and litter (δ_L) except in February (Fig. 2). This may be attributable to surface-soil freezing in December and snow coverage of the forest floor in February; it has been reported that during freezing and thawing periods the δ¹³C values of soil air change as a result of alterations to soil physical properties (Dudziak and Halas 1996).

The greater the difference between δ_RHI and δ_SOM, the smaller the errors of f_RHI and f_SOM (Eq. 10). We used a significance level of 5%; thus, the errors of δR_S need to be 0.21‰ or smaller because the isotopic difference between δ_RHI and δ_SOM is 4.2‰ at this site. As the standard errors of δR_S were 0.23–3.98‰ (Fig. 2), the relative errors of estimated ratios were 5–93%. The error of the δR_S was large when soil respiration was small, for example, in winter.

Figure 2  Seasonal fluctuations in the carbon isotope ratio of CO2 emitted by soil respiration. Error bars represent the standard error.

Contribution of R_RHI, R_SOM and R_L to R_S

The ratios of R_SOM to R_S calculated using Eq. 7 and Eq. 8 tended to increase in summer and decrease in winter and ranged from 8 to 67% of R_S (Fig. 3). Those of R_RHI to R_S showed a similar seasonal trend and ranged from 10 to 51%. R_RHI and R_SOM were exponentially correlated with soil temperature:

and the maximum values of R_RHI and R_SOM occurred in July; 59.0 and 52.6 µg CO₂ m⁻² s⁻¹, respectively. Accordingly, in summer R_RHI and R_SOM were the predominant sources of R_S, that is, the ratio of R_RHI to R_SOM to R_L was 51:44:5 in July. The contribution ratios of R_L to R_S ranged from 5 to 69%; that ratio was at a maximum in December and a minimum in July (Fig. 3). Therefore, R_L was predominant in winter and the ratio was 20:11:69 in December. The annual R_RHI and R_SOM were 0.53 and 0.77 kg CO₂ m⁻² (1.45 and 2.10 Mg C ha⁻¹), respectively, which were estimated by summing up the hourly R_RHI and R_RHI calculated by inserting hourly soil temperature into Eq. 13 and Eq. 14 from July 2001 to June 2002. The annual R_L was 0.48 kg CO₂ m⁻² (1.30 Mg C ha⁻¹), which was estimated by the annual R_S minus the annual R_RHI minus the annual R_SOM; thus, the contributions of R_RHI, R_SOM and R_L were 30:43:27 on a whole-year basis.

DISCUSSION

As in many previous reports, the R_RHI and the contribution of R_RHI to R_S were found to increase in summer (Fig. 3). For example, the contribution of R_RHI increased from spring to summer in Japanese cedar forests (Lee et al. 2003; Ohashi et al. 2000), an oak forest (Takahashi et al. 2002) and a boreal pine forest (Hogberg et al. 2001). This seasonal pattern may result from root phenology, such as a large mass of fine roots and high root respiration in summer. The mass of fine roots did tend to be larger in summer (July) than in winter (December) at our study site. Furthermore, the amount of roots produced in summer (June to September) was greater than that produced in other seasons as observed using the ingrowth coring method at this site (Mizoguchi T, 2002, pers. comm.).

We confirmed the annual contribution ratios of R_RHI, R_SOM and R_L from the balance of carbon in the soil. The annual carbon emission from litter (R_L) was 1.30 Mg C ha⁻¹, which corresponded to 61% of the carbon in the litter in July. This value also corresponded to 64% for the leaf litterfall (2.05 Mg C ha⁻¹) at the site. According to a review article (Paul et al. 2002), the soil carbon content in various forests did not significantly change in the 30 years after trees were planted. If we assume that the carbon balance of the soil at this site is almost at equilibrium, then the turnover rate of roots can be calculated. When the annual carbon emission from the SOM is balanced with carbon supplied from roots and the litter layer to the soil, the annual deposition from roots can be explained by the annual R_SOM (2.10 Mg C ha⁻¹) plus R_L minus leaf litterfall. Root deposition was found to be 1.35 Mg C ha⁻¹ in this study. This deposition was almost equivalent to the fine root mass 1.07–1.33 Mg C ha⁻¹ (0.247–0.306 kg m⁻²) in the surface soil. In other words, fine root turnover was estimated to be approximately 1 year, which is close to the values reported in previous studies (Chen et al. 2004; Gill and Jackson 2000; Majdi et al. 2005; Meharg 1994).

It is essential that there be a significant difference in ¹³C values between roots (δ_RHI) and SOM (δ_SOM) when applying the ¹³C natural abundance technique to

Figure 3  Seasonal fluctuations of CO2 emissions and the contribution ratios from roots (RRHI), soil organic matter (RSOM) and litter (RL). Error bars represent the standard error. N.E., values that were not able to be estimated.

separate root respiration from heterotrophic respiration in the field. Because we found that R_RHI and R_SOM had large seasonal variation, the large difference between δ_RHI and δ_SOM is advantageous for estimating the contribution of CO₂ sources accurately. Usually, in Japan there are greater differences between δ_RHI and δ_SOM than we found at our study site (Ishizuka et al. 1999; Ishizuka S et al., 2004, unpubl. data). Because Andosols are relatively widely distributed, accounting for 13% of the forested area in Japan (Morisada et al. 2004), the ¹³C natural abundance technique can be applied to various forest ecosystems in Japan. However, for Andosols in other countries, the difference between δ_RHI and δ_SOM needs to be confirmed before applying this method. In addition, the errors in the measurements should be considered. As we showed, the errors of δR_S were large when the soil respiration was small. This may be attributable to the smallness of the changes in CO₂ concentration and δ¹³C in the chamber for the measurement period. We need to evaluate δR_S more accurately, for example, by leaving a longer interval between the first and second sampling of the gas in the chamber after sealing in winter. More precise measurement in winter is necessary not only for accurate estimation of the contribution ratio of R_RHI and R_SOM to the total respiration, but also for clarification of the effect of soil freezing and snow cover on the δ¹³C of soil air and soil respiration (Dudziak and Halas 1996).

We should consider the isotope fractionations of the CO₂ emissions from SOM, roots and litter. If the isotope fractionations occurred, this isotope method would have larger estimated errors. Although there are some reports examining isotope fractionation (Amundson et al. 1998; Balesdent and Mariotti 1996; Blair et al. 1985; Cerling et al. 1991; Lin and Ehleringer 1997; Lin et al. 1999), in general, the mechanisms of the isotopic fractionations during SOM decomposition and root respiration are not well understood. That isotopic fractionations during decomposition and root respiration even occur is currently under debate (Crow et al. 2006; Fernandez et al. 2003; Klumpp et al. 2005). In the future, to estimate more accurately the ratio of the CO₂ emission from each source using this technique, we should not only measure the δR_S more accurately, but also clarify the isotope fractionations.

ACKNOWLEDGMENTS

The authors wish to thank Drs N. Okada of Kyoto University and Y. Tsuboyama of the Forestry and Forest Products Research Institute for their help in the use and maintenance of the isotope analysis equipment. We appreciate being allowed to use the research site by the Ibaraki District Forest Office. This work was supported by research grant numbers 199903 and 200303 from the Forestry and Forest Products Research Institute, Japan.