Total and Component Carbon Fluxes of a Scots Pine Ecosystem from Chamber Measurements and Eddy Covariance

Abstract

Background and Aims

Distinguishing between, and quantifying, the different components of ecosystem C fluxes is critical in predicting the responses of ecosystem C cycling to climate change. The aims of this study were to quantify the photosynthetic and respiratory fluxes of a 50-year-old Scots pine (Pinus sylvestris) ecosystem, and to distinguish respiration of branches with needles from that of stems, and that of soil.

Methods

The CO₂ flux of the ecosystem was continuously measured using the eddy covariance (EC) method, and its components (respiration and photosynthesis of a branch with needles, stem and soil surface) were measured with an automated chamber system, from 2001 to 2004.

Key Results

All values below are chamber based. The average temperature coefficient (Q₁₀) of respiration was 2·7, 2·2 and 4·0, respectively, for branch (R_bran), stem (R_stem) and the soil surface (R_soil). Respiration at a reference temperature of 15 °C (R₁₅) was 1·27, 0·49 and 4·02 µmol CO₂ m⁻² ground s⁻¹ for the three components, respectively. Over 4 years, the annual R_bran, R_stem and R_soil ranged from 196 to 256, 56 to 83 and 439 to 598 g C m⁻² ground year⁻¹, respectively, with a 4-year average of 227, 72 and 507 g C m⁻² ground year⁻¹. Annual ecosystem respiration (R_eco) was 731, 783, 909 and 751 g C m⁻² ground year⁻¹ in years 2001–2004, respectively, gross primary production (GPP) was 922, 1030, 1138 and 1001 g C m⁻² ground year⁻¹, and net ecosystem production (NEP) was 191, 247, 229 and 251 g C m⁻² ground year⁻¹. The average contribution of R_bran, R_stem and R_soil to R_eco was 29, 9 and 62 %, respectively. Overstorey photosynthesis accounted for 96 % of GPP. The average R_eco/GPP ratio was 0·78. Net primary production (NPP) in the 4 years was 469, 581, 600 and 551 g C m⁻² year⁻¹, respectively, with the NPP/GPP ratio 0·54 averaged over the years.

Conclusions

Respiration from the soil is the dominant component of ecosystem respiration. Differences between years in R_eco were due to differences in temperature during the growing season. R_soil was more sensitive to temperature than R_bran and R_stem, and differences in R_soil were responsible for the differences in R_eco between years.

INTRODUCTION

Forests contain 66–80 % of all C stored in above-ground biomass and about 45 % of that found in below-ground terrestrial pools (Waring et al., 1998). Boreal forests cover a broad circumpolar band across the Eurasian and North American continents, and represent approx. 11 % of the Earth's total land area (Bonan and Shugart, 1989; Gower et al., 2001). They are of global importance because of their large carbon (C) stores and C fluxes, which are substantial components of the Earth's C budget (Landsberg and Gower, 1997; Gower et al., 2001). Thus the magnitude of CO₂ exchange between these forests and the atmosphere is of major importance. Carbon balance of forest ecosystems has become a particular concern as an important environmental issue because of rising atmospheric CO₂ concentration and its influence on climate and vegetation. Forests may be managed in future, not simply for timber and other products, but also for CO₂ sequestration to ameliorate global warming. Ecosystem C management demands an understanding of photosynthesis and respiration, two major processes determining ecosystem C balance.

Net CO₂ exchange of ecosystems, including respiration, has been measured and reliably quantified by eddy covariance (EC) measurements (Black et al., 1996; Aubinet et al., 2000; Baldocchi et al., 2001; Falge et al., 2002; Zha et al., 2004b). However, EC cannot provide information on the C dynamics of components (e.g. photosynthesis and respiration of parts of the vegetation) of an ecosystem. The importance of respiration in the C balance is uncertain (Grace and Rayment, 2000; Valentini et al., 2000), with uncertainties in estimating respiration from meteorological measurements in darkness (Lee, 1998). Component C fluxes (soil, branch, etc.) may be measured continuously with an automated chamber system to estimate their contribution to the total fluxes (Griffis et al., 2004; Zha et al., 2004a, 2005) over long periods. Long-term measurements of components improve the estimates of long-term dynamics of C fluxes compared with measurements made over a limited period. There is still much uncertainty about respiration and photosynthesis of parts of the forest ecosystem and their relative contributions to the total carbon balance. In particular, the processes controlling the sources and dynamics of CO₂ flux from the soil surface remain poorly understood.

The objectives of the present study are (a) to quantify seasonal differences, and those between years, in respiration and photosynthesis for a forest ecosystem and its components; (b) to quantify the relative contribution of respiration of different components to the ecosystem carbon budget; and (c) to compare the estimate of ecosystem from automated chambers with those from the EC.

MATERIALS AND METHODS

Site description

The research was conducted in a 50-year-old stand of Scots pine (Pinus sylvestris L.) at Huhus (62°52′N, 30°49′E, 145 m a.s.l.), eastern Finland. The climate is characterized by a long, cold winter. The mean monthly temperature is lowest in January, −10·4 °C, and highest in July, 15·8 °C. The average annual precipitation at the site (1961–2000) is 724 mm, of which 38 % falls as snow. The site is flat, and has a homogeneous overstorey and understoreys. Soil is a sandy podzol with the top 50 cm containing an average volumetric mineral fraction of 47 % and organic matter fraction of 21 %, and has a mean bulk density of 1·34 g cm⁻³. The depth of the soil organic layer is about 15–20 cm. There are 1176 trees ha⁻¹, with a mean height of 13·5 m above the ground and a mean diameter at breast height of 11·2 cm. The soil surface is covered by small patches of litter or lichen (30 and 65 % of area, respectively). The understorey is principally mosses (Dicranum spp, Pleurozium schreberi) and dwarf shrubs (Vaccinium vitis-idaea, Calluna vulgaris), so the site is a Calluna type ecosystem on a sandy soil with a low nitrogen supply.

Measurements of CO₂ exchange

Net ecosystem CO₂ exchange was measured by EC, and, concurrently, vertical temperature, humidity and radiation profiles as described by Zha et al. (2004b). The component CO₂ fluxes of branches with needles, tree stems and soil surface were measured for 60 s every 30 min with automated chamber systems based on a CO₂ and H₂O analyser (Hartmann and Braun URAS 14, Software Version 1·3, Hartmann and Braun GmbH and Co. KG, Germany), from April to November each year from 2001 to 2004. The chamber system had an open, flow-through design in which CO₂ efflux was measured as the difference between the CO₂ entering and leaving the chamber. The flow rate of the reference air was 1·0 L min⁻¹ and that of sample air from the chamber was 1·0 L min⁻¹ for stem, and 1·5 L min⁻¹ for both branch and soil, and the rates were regulated by separate pumps and mass flow controllers (5850E, Brooks Instrument, Veenendaal, The Netherlands). The branch and soil chambers were equipped with pneumatically operating lids which closed during the measurement periods for 60 s once every 30 min, and opened between the measurements. Sample air flowed to an analyser through a 20–30 m long Teflon tube with inner diameter 4 mm. Condensation on the gas line was avoided by warming the air tube with heating tapes, which slightly affected the temperature of the sample air. The effect of the dilution and mixing in sampling tube on measurements was minimized by removing the first measurement in the calculation of fluxes.

It was assumed that the variation in C fluxes along a bole is small. Stem respiration was measured on three trees, 10·4, 11·7 and 23·8 cm in diameter 1·3 m height above ground, using a cylindrical respiration chamber of high density foam [expanded polystyrene (EPS), UK-Muovi Oy, Iisalmi, Finland], 16 cm long with two halves, completely encircling a segment of the tree stem at a height of 1·3 m above the ground. The two halves of the chamber were attached to the bark with 20 mm thick neoprene gaskets glued to the inside on both ends. The chambers were covered and lined with aluminium foil to prevent overheating. Details of measurements are given in Zha et al. (2004a, 2005).

Four branch chambers, 14 cm in diameter and 23 cm in height, were installed at four different crown positions (four different whorls along the vertical transect) located in four individual trees. Each chamber enclosed a small branch with needles of classes similar to those on the tree. Needle area was estimated once every month in the growing season from nearby branches with similar size and properties to the enclosed branch. The needle area from the sampling branch was measured using a scanner and the winNEEDLE program (v. 4·0, Régent Instruments Inc., Canada). The photosynthetically active radiation (PAR) inside each of the chambers was measured by an LI-190SB quantum sensor.

Four soil chambers 19 cm in diameter and 20 cm in height were placed on areas representative of the average characteristics of the site. Two centimetres of the lower part of the chamber was inserted into the soil and the chamber was fixed permanently on the ground surface. The PAR inside the chamber was measured by an LI-190SB quantum sensor, mounted 2 cm inside to the upper edge of the chamber.

Calculation of photosynthesis and respiration

The model used to relate daytime (PAR>3 µmol m⁻² s⁻¹) CO₂ exchange to PAR was (Hollinger et al., 1994):

1

where A_max is the maximum photosynthetic rate (μmol m⁻² s⁻¹), R_day is respiration during daytime, A_n is daytime net CO₂ exchange, and α is the initial slope of the light response curve (μmol CO₂ μmol⁻¹ of photon), i.e. the quantum yield. For this analysis, it is assumed that carbon assimilation is driven by light only, and does not consider any other limitations (temperature, vapour pressure deficit, drought stress, etc.).

The model used to describe the effects of temperature on night-time (PAR≤3 µmol m⁻² s⁻¹) CO₂ exchange was

2

where R is respiration, R₁₅ is respiration at a reference temperature of 15°C, T is the temperature, and Q₁₀ is the temperature coefficient of respiration.

The night-time net CO₂ exchange was assumed to be caused by respiration, and the annual total for each component was separately estimated by eqn (2) using parameters derived from regression of mean daily respiration at night against the corresponding temperature. Temperatures used in eqn (2) were mean daily air temperature at canopy height (12 m above the ground), stem temperature 5 mm below the bark, and soil temperature at 10 cm below the ground surface. Annual total photosynthetic rates for branches with needles (GPP1) were estimated by eqn (1) in which the parameters derived from the regression of mean daily branch CO₂ fluxes against the corresponding PAR. For soil surface CO₂ fluxes, the daytime values fit eqn (2) well (R²>0·70). Therefore, the photosynthetic rates of the understorey (GPP2) were estimated as the difference between estimated daytime net CO₂ exchange and estimated daytime soil respiration. The daytime net CO₂ exchange was estimated with eqn (2) using parameters derived from the daytime soil surface CO₂ fluxes and the corresponding soil temperature 10 cm below the ground. Daytime soil respiration was estimated using eqn (2) with parameters from the regression of night soil surface CO₂ fluxes against corresponding soil temperatures 10 cm below ground. Analyses of EC data including data screening, gap filling and calculation of total fluxes have been described (Zha et al., 2004b).

Scaling chamber measurements

Soil temperature at 2·5, 3·5 and 10 cm depth below the ground surface in the organic layer were measured using thermocouple probes (HFP01, Seattle, WA, USA). The water content at 0–5 cm depth below the mineral soil surface was monitored using a water content reflectometer (CS615, Campbell Scientific, Shepshed, Leicestershire, UK). Detailed information is given by Kellomäki and Wang (1999), Wang et al. (2004) and Zha et al. (2004b).

The stem temperature was measured at hourly intervals with copper–constantan thermocouples 5 mm under the bark at 15, 285 and 305 cm above the ground throughout the 4 years (2001–2004). The average stem temperature at the three heights was used in the analysis. Diameter growth at breast height was monitored at hourly intervals with a dendrograph (ELPA 93, University of Oulu, Finland) consisting of a stainless-steel band and a displacement potentiometer. The precision was 0·5 mm. The stem respiration per unit ground area was calculated using annual stem volume per unit ground area. Annual stem volume was calculated by measuring both diameters at 1·3 m above ground and tree heights for 32 trees around the tower in October each year. Trees were assumed to have cylindrical dimensions.

Leaf area index (LAI) in 2001 and 2002 that was used to calculate the branch photosynthesis and respiration per unit ground area was reported previously (Wang et al., 2004); that in 2003 and 2004 was estimated by the equation derived from regressing the seasonal LAI in 2001 and 2002 against corresponding tree diameters (y=0·85x−10·71, R²=0·87, where x represents the diameter 1·3 m from the ground).

Statistical analysis

Statistical analysis was conducted using SPSS Version 11·5 for Windows software (SPSS, Chicago, IL, USA). Differences between years were analysed using a general linear model (GLM) repeated-measures analysis of variance (ANOVA). If the difference was significant, Tukey's h.s.d of the post hoc multiple comparison test was performed to compare the differences between years. Student's t-test was used to analyse the difference between EC- and chamber-based estimates. All tests were based on a 0·05 significance level.

RESULTS

Photosynthesis and respiration

Net CO₂ exchange during darkness (PAR≤3 µmol ;m⁻² s⁻¹) was assumed to be respiration. The respiration of branch (R_bran), stem (R_stem) and soil surface (R_soil) was exponentially related to the air temperature, stem temperature and soil temperature, respectively (Table 1). The data fitted eqn (2) with R²>0·56. Respiration at the reference temperature of 15°C (R₁₅) for branch, stem and soil surface ranged over 4 years (2001–2004) from 1·04 to 1·74, 0·44 to 0·58 and 3·15 to 4·68 µmol CO₂ m⁻² ground s⁻¹, respectively. The 4-year averages (±s.e.) were 1·27±0·16, 0·49±0·03 and 4·02±0·32 µmol CO₂ m⁻² ground s⁻¹. Temperature coefficients of respiration (Q₁₀) for branches ranged, over 4 years, from 2·18 to 3·77 with a 4-year average of 2·7±0·36; the corresponding values for stems were 1·99–2·69 (average 2·2±0·16), and for soil 2·39–5·71 (average 4·0±0·83).

Table 1.

Temperature coefficients (Q₁₀) of respiration of branches with foliage, stems and soil surface (μmol CO₂ m⁻² ground s⁻¹) at a reference temperature of 15 °C(R₁₅) and determination coefficient R²

Year		Q10	s.e.	R15	s.e.	R2
2001	Branch	2·34a	0·14	1·18a	0·03	0·56
	Stem	2·09b	0·04	0·45b	0·01	0·92
	Soil	5·71c	0·37	4·68f	0·12	0·84
2002	Branch	2·56a	0·11	1·04a	0·01	0·81
	Stem	1·99b	0·04	0·44b	0·01	0·90
	Soil	2·39c	0·11	3·15c	0·07	0·71
2003	Branch	2·18a	0·08	1·12a	0·02	0·73
	Stem	2·03b	0·06	0·49b	0·02	0·82
	Soil	2·8c	0·13	4·01c	0·09	0·74
2004	Branch	3·77d	0·16	1·74d	0·02	0·85
	Stem	2·69e	0·11	0·58e	0·03	0·79
	Soil	4·11f	0·36	4·27f	0·15	0·72

Values of parameters were from exponential regression of branch with foliage respiration and corresponding air temperature at crown height, stem surface respiration and stem temperature at 5 mm depth under the bark, and soil surface respiration and soil temperature at 10 cm depth. Values not followed by the same letter are significantly different from each other at P=0·05.

When daytime net CO₂ exchange for branches was regressed against PAR, there was a good fit to eqn (1) (R²>0·59, Table 2). Light-saturated rates of gross photosynthetic assimilation (A_max) ranged over 4 years (2001–2004) from 8·06 to 9·43 (4-year average 9·03±0·33) μmol CO₂ m⁻² ground s⁻¹. Quantum yield of branches ranged over 4 years from 0·014 to 0·026 (average 0·022±0·003) μmol µmol⁻¹.

Table 2.

Comparison of photosynthetic parameters for branches derived from fitting eqn (1) to net branch CO₂ exchange during the day (PAR>3 μmol m⁻² s⁻¹)

	Amax	s.e.	q	s.e.	R2
2001	9·29a	0·74	0·014b	0·001	0·72
2002	9·35a	0·35	0·026a	0·002	0·80
2003	9·43a	0·39	0·024a	0·001	0·86
2004	8·06b	0·55	0·024a	0·003	0·59

Values not followed by the same letter are significantly different from each other at P=0·05.

q (μmol µmol⁻¹), quantum yield; A_max (μmol m⁻² ground area s⁻¹), light-saturated rate of gross photosynthetic assimilation; R², coefficient of determination.

Seasonal and yearly differences in CO₂ fluxes

The modelled mean daily respiration for branch, stem and soil surface matched the corresponding observed respiration well (Fig. 2). There were similar seasonal trends in the respiration of branch (R_bran) and stem (R_stem), and from the soil surface (R_soil), in relation to temperature and radiation (Figs 1 and 2). The largest values were in late July or early August when the mean daily temperature and radiation were highest. The smallest values of respiration occurred in winter. Estimated mean daily R_bran maximized at 1·97, 1·35, 1·52 and 2·02 µmol CO₂ m⁻² ground s⁻¹ in the 4 years (2001–2004), respectively, R_stem maximized at 0·61, 0·63, 0·69 and 0·60 µmol CO₂ m⁻² ground s⁻¹, and R_soil maximized at 5·92, 4·87, 6·71 and 4·71 µmol CO₂ m⁻² ground s⁻¹ (Fig. 2).

Fig. 2.

Observed and modelled mean daily respiration per unit ground area, for a branch with needles (R_bran), stem (R_stem) and soil surface (R_soil) in 4 years (2001–2004).

Fig. 1.

Daily total precipitation (P), mean daily photosynthetically active radiation above the canopy (PAR), mean daily air temperature 12 m above ground, mean daily soil temperature 10 cm below the soil surface and volumetric soil water content (θ_soil) 5 cm below the mineral soil surface from 2001 to 2004.

The monthly total respiration of components was greatest in July and smallest in January or December (Fig. 3). Annual R_bran, R_stem and R_soil ranges in 2001–2004 and 4-year averages (in parentheses) were 196–256 (225±12) g C m⁻², 56–83 (72±6) g C m⁻² and 438–598 (497±38) g C m⁻², respectively (Table 3).

Fig. 3.

Total monthly respiration per unit ground area for branches with needles (R_bran), stem (R_stem) and soil surface (R_soil), as well as chamber- (R_eco) and EC-based ecosystem respiration (R_EC) over 4 years (2001–2004).

Table 3.

Annual estimates of ecosystem and component C fluxes (g C m⁻² ground year⁻¹) for the period of 2001–2004

	Estimates from chamber measurements				Estimates from eddy covariance measurements
	2001	2002	2003	2004	2001	2002	2003	2004
Ecosystem respiration Reco	731 (75)	782 (88)	909 (96)	751 (92)	752	879	817	691
Branch respiration Rbran	217 (31)	196 (16)	228 (25)	256 (36)
Stem respiration Rstem	76 (11)	74 (8)	83 (6)	56 (9)
Soil surface respiration Rsoil	438 (71)	512 (77)	598 (65)	439 (68)
Root respiration Rr	159 (21)	178 (26)	227 (23)	138 (19)
Heterotrophic respiration Rh	279 (35)	334 (41)	372 (32)	301 (43)
Autotrophic respiration Ra	452 (41)	449 (58)	537 (46)	450 (64)
Ra/Reco	0·62	0·57	0·59	0·60
Ecosystem GPP	922 (97)	1030 (112)	1138 (121)	1001 (135)	924	1084	985	869
Overstory GPP1	870 (95)	996 (106)	1101 (118)	960 (130)
Understory GPP2	52 (11)	34 (6)	37 (5)	41 (8)
Net primary production NPP	470 (52)	582 (63)	601 (57)	552 (75)
Net ecosystem production NEP	191 (25)	247 (33)	229 (28)	251 (37)	172	205	168	179
Reco/GPP	0·79	0·76	0·80	0·75	0·81	0·81	0·83	0·79
NPP/GPP	0·51	0·56	0·53	0·55
Rbran/Reco	0·30	0·25	0·25	0·34
Rstem/Reco	0·10	0·09	0·09	0·07
Rsoil/Reco	0·60	0·65	0·66	0·58
Rr/Reco	0·20	0·23	0·25	0·18
Rh/Reco	0·38	0·43	0·41	0·40

Values are the mean of chambers (±s.e.).

The monthly total of photosynthesis was highest in July, except for the understorey in 2002 (Fig. 4). The overstorey photosynthesis (GPP1) ranged over 4 years from 870 to 1101 (average 982±48) g C m⁻² year⁻¹, and that for the understorey ranged from 35 to 52 g C (average 41±4) m⁻² year⁻¹ (Table 3).

Fig. 4.

Monthly total of photosynthesis per unit ground area for overstorey (GPP1) and understorey (GPP2), as well as chamber- (GPP) and EC-based ecosystem photosynthesis (GPP_EC) over 4 years (2001–2004).

Component fluxes and contribution

Among the respiration of components per unit ground area, respiration from the soil surface (R_soil) was largest and stem respiration (R_stem) was smallest throughout the growing season (Figs 2 and 3). Branch respiration (R_bran) accounted for 25–34 % over 4 years, R_stem for 7–10 % and R_soil for 58–66 %. Overall, R_bran, R_stem and R_soil accounted for 29, 9 and 62 % of R_eco, respectively. Overstorey photosynthesis GPP1 accounted for 94, 97, 97 and 96 % of ecosystem photosynthesis (GPP, sum of overstorey and understorey photosynthesis) in 4 years (2001–2004), respectively, and understorey photosynthesis (GPP2) for 6, 3, 3 and 4 % of GPP. Overall, GPP1 and GPP2 contributed 96 and 4 % to GPP, respectively.

A general equation (based on 54 sets of published data encompassing 54 forest sites) of the relationship between root respiration (R_r) and soil respiration R_soil (R_r^0·5=−7·97+0·93R_soil^0·5, R²=0·87, P<0·001; Bond-Lamberty et al., 2004) was used to calculate root respiration (R_r), thereby estimating ecosystem autotrophic respiration (R_a=R_bran+R_stem+R_r) and net primary production (NPP=GPP−R_a). Estimated NPP ranged over 4 years from 470 to 601 g C m⁻² year⁻¹ with a 4-year average of 551±29 g C m⁻² year⁻¹. Overall, NPP accounted for 54 % of GPP.

Comparison of total ecosystem CO₂ exchange from chamber measurements with eddy covariance

Total ecosystem respiration (R_eco) from chamber measurements for 2001 and 2002 was 3 and 10 % smaller than corresponding values from EC, respectively (Table 3), while those values for 2003 and 2004 were 11 and 8 % larger than those from EC. Chamber-based estimates of annual GPP for 2001 and 2002 were close to the corresponding values from EC, while those values in 2003 and 2004 were 16 and 15 % higher than corresponding values from EC, respectively. In contrast, the chamber estimates of annual net ecosystem production (NEP) in 4 years were 11, 20, 36 and 40 % greater than corresponding values from EC, respectively.

DISCUSSION

Respiration and photosynthesis parameters

There was little change in annual temperature coefficients of respiration (Q₁₀) and respiration at the reference temperature of 15 °C (R₁₅) for branches and stem over the 3 years 2001–2003 (Table 1). However, both respiration parameters for soil surface ranged widely from year to year, exhibiting higher Q₁₀ and R₁₅ in 2001 and 2004. This indicates that soil respiration is more sensitive than branch and stem respiration to temperature. Large changes in soil surface respiration between years may be associated with the differences in soil temperature between years. Mean monthly temperatures in the main growing seasons in 2001 and 2004 were cooler than in the other 2 years (Fig. 5C). Higher Q₁₀ and R₁₅ for the soil surface in both 2001 and 2004 coincided with lower soil temperatures in both years.

Fig. 5.

Monthly mean of phtosynthetically active radiation (PAR), air temperature 12 m above ground (T_a), soil temperature 10 cm below ground (T_soil) and soil volumetric water content (θ_soil) 5 cm below the mineral soil surface over 4 years (2001–2004).

The quantum yield of branches (α) varied from 0·014 to 0·026 over 4 years (average 0·022, Table 2), similar to that (0·029) for a mid-latitude Scots pine forest ecosystem (Dolman et al., 2002) and that (0·026) for a Scots pine ecosystem at the same site (Zha et al., 2004b). The average was close to the value of 0·024 for coniferous forests obtained by Ruimy et al. (1996). There is no significant difference in A_max values between years (P=0·129).

Variation between years

There were changes in annual C fluxes of both ecosystem and its components (Table 3). Ecosystem respiration (R_eco) from chamber measurements was largest in 2003 and smallest in 2001, corresponding to differences in soil surface respiration between years, greatest in 2003 and smallest in 2001 (Fig. 5C). These differences are associated with soil temperature in the main growing season (July–September; Fig. 5C). Warmer temperature in 2003 led to faster soil respiration, the main contribution to R_eco, which thereby was large. Conversely, cooler soil temperatures in 2001 and 2004 led to smaller R_soil, and thus R_eco (Table 3, Fig. 5C). There was no significant relationship between photosynthesis (P=0·225) or respiration (P=0·186) and soil water content. In addition, as the soil was well drained, water logging was not apparent. Therefore, soil water content was probably not a factor limiting photosynthesis and respiration. Similar differences between years in C fluxes of forest ecosystems have been reported (Barr et al., 2004; Epron et al., 2004).

Component fluxes

The contribution of respiration from branch (R_bran), stem (R_stem) and soil surface (R_soil) to total ecosystem respiration (R_eco) was 29, 9 and 62 % averaged over 4 years, respectively. The results are comparable with the 17·6, 6 and 76·4 % for R_bran, R_stem and R_soil, respectively, of a ponderosa pine forest in Oregon (Law et al., 1999), and close to the proportions of 25·4, 9·5 and 64·8 % in another ponderosa pine forest (Xu et al., 2001). The proportion of soil surface respiration (62 %) is within the estimated range of 48–71 % for soil surface C efflux by Raich and Schlesinger (1992), and similar to the 67 % of a mixed Scots pine and oak forest (Yuste et al., 2005). The contribution of R_soil to R_eco was relatively stable in winter and summer, with slight fluctuations (Fig. 6). However, there was a decline in the contribution of R_soil to R_eco and an increase in the contribution of R_bran and R_stem to R_eco in spring (April or May; Fig. 6). An explanation is that wetter soil and cooler temperature caused by snow melting in April and May slowed decomposition of soil organic matter and decreased root respiration (Fig. 5). Needles and stems become active when snow starts melting as the air temperature gradually increases. Therefore, both R_bran and R_stem increased much more than that R_soil, leading to the decline in contribution of R_soil to R_eco in April and May. A similar trend was observed at another site of Scots pine (Yuste et al., 2005).

Fig. 6.

Monthly contributions of branch, stem and soil respiration to corresponding ecosystem respiration over 4 years (2001–2004).

Comparison with EC values

A number of studies have reported larger estimates of respiration from chamber measurements than EC (Goulden et al., 1996; Lavigne et al., 1997; Law et al., 2000; Griffis et al., 2004). In the present case, the difference between EC- and chamber-based estimates in both respiration (R_eco) and photosynthesis (GPP) differed from year to year. Overall, chamber-based estimates of R_eco, GPP and NEP were, respectively, 2, 6 and 27 % greater than from the EC estimate. The estimates of component C fluxes might be underestimated due to dilution in the sampling tube, to measurements of stem respiration in the lower part of the stem (1·3 m above ground) or to neglecting the contribution of small branches in woody tissue respiration.

The R_eco/GPP ratio was relatively constant over 4 years. Moreover, the 4-year average for chamber-based R_eco/GPP ratios was 0·78 compared with 0·81 from EC, slightly greater than the average of 0·76 of four boreal forest sites (Lindroth et al., 1998; Vesala et al., 1998; Law et al., 2001; Markkanen et al., 2001), but very similar to the ratio of 0·82 for temperate coniferous forests and 0·77 for temperate broad-leaved deciduous forests (FluxNet sites; Falge et al., 2002).

Net primary production

In order to estimate NPP, R_soil was partitioned into autotrophic (R_r) and heterotrophic respiration (R_h) by using the general equation (R_r^0·5=−7·97+0·93R_soil^0·5, r²=0·87, P<0·001; Bond-Lamberty et al., 2004). From this, root respiration accounted for 31–38 % (average 35 %) of total soil surface respiration over 4 years (Table 3). Our estimate is smaller than a mean of 45·8 % for forest species from 37 studies in the literature (Hanson et al., 2000), but within the range of 30–90 % for the contribution of root respiration to total soil surface CO₂ efflux (Bowden et al., 1993; Thieron and Laudelout, 1996; Epron et al., 1999). Nakane et al. (1996) concluded that the proportion of root respiration to soil surface CO₂ efflux may approach 50 % irrespective of forest type. Given this uncertainty, further measurements of root respiration are required.

The estimated NPP/GPP ratio ranged from 0·51 to 0·56, over 4 years (average 0·54), and is greater than the 0·32 for a 100-year-old Scots pine forest in eastern Finland (Helmisaari et al., 2002) and smaller than the values of 0·66–0·85 reported for Scots pine forests (Gower et al., 1994). The NPP/GPP ratio did not differ significantly between years (P=0·481).

Estimation of component fluxes

The goodness of fit of the regression between temperature and respiration may depend on which temperature profile is chosen. Therefore night-time net CO₂ exchanges, from both chamber and EC measurements, were regressed against air (12 m above canopy) and soil (2·5, 3·5 and 10 cm depth below the ground surface) temperatures. There were no significant differences in the regression slopes for these temperatures during the growing season (P>0·163). Therefore, canopy air (12 m above ground), stem and soil (10 cm depth) temperatures were used in data analysis. Night-time net CO₂ flux of all components fitted eqn (2) well (R²>0·6, Table 1), while daytime net CO₂ exchange for branch and EC fit eqn (1) (R²>0·6, Table 2). We are therefore confident of the estimation of annual respiration of each component.

CONCLUSIONS

Branch (R_bran), stem (R_stem) and soil (R_soil) respiration changed with the seasons, corresponding to changes of temperature, differences in which between years affected the respiration of components and the total ecosystem respiration (R_eco). The main source of differences in R_eco between years was R_soil, which contributed 62 %, compared with 29 % from R_bran and 9 % from R_stem. Overstorey photosynthesis contributed 96 % to GPP, and understorey photosynthesis contributed 4%. Soil respiration was sensitive to temperature, with a 4-year average Q₁₀ of 4·0, indicating that temperature affected R_soil more than it affected R_bran and R_stem. Estimates of R_eco, GPP and NEP from chamber measurements were comparable with those from EC. The R_eco/GPP ratios were, on average, 0·78 and 0·81 for chamber- and EC-based estimates, respectively. The NPP/GPP ratio was 0·54.

In conclusion, seasonal and interannual variations in R_bran, R_stem, R_soil and R_eco were mostly controlled by temperature. Respiration from the soil surface dominated the ecosystem respiration.