Introduction

Efflux of CO2 from soils can account for over 70% of the respiration in some ecosystems (Raich and Schlesinger 1992). Law et al. (1999) estimated that 76% of the CO2 lost from a ponderosa pine (Pinus ponderosa) ecosystem was derived from soil respiration (SR). Respiration from soils is comprised of both plant-derived “autotrophic” respiration from roots and mycorrhizae, and “heterotrophic” respiration from free-living soil bacteria, fungi, and fauna (Nguyen 2003; Hanson et al. 2000; Whipps 1990).

To understand the role of environmental factors on ecosystem carbon flux, it is important to clarify the seasonal course of SR, and how it is partitioned among autotrophic and heterotrophic components. Widen and Majdi (2001) found that the contribution of roots to total SR in a coniferous forest varies seasonally and was greater in the spring than the fall. Elevated temperature, limiting or supra-optimal soil water availability, nutrient limitations, and air pollution have all been shown to alter carbon allocation in trees, often resulting in a change in autotrophic, and hence total SR (Andersen 2003; Vose and Ryan 2002; Scagel and Andersen 1997; Edwards 1991). Understanding the contribution of roots/mycorrhizae (autotrophs) and decomposers (heterotrophs), i.e., the two major biotic compartments in belowground respiration, to total SR will promote the reliability of models on forest productivity and atmospheric impact on ecosystem carbon flux (Vose and Ryan 2002; Epron et al. 2001; Ryan et al. 1996).

Given the importance of quantifying carbon flux from soils, relatively little is known about the source components of this flux. Unfortunately, it is extremely difficult to separate the individual components of SR since they are intimately linked, and manipulation of one component can lead to a change in respiration from another component (Hanson et al. 2000). One non-invasive approach involves removing the carbon source for root growth and maintenance, either by cutting the shoot (Nakane et al. 1996) or by girdling its phloem (Högberg et al. 2001), and following the change in respiration over time. Roots of trees store significant carbon reserves, so girdling may not always lead to immediate root mortality and reductions in respiration (Högberg et al. 2002). However, previous studies have shown a rapid decrease in SR upon girdling, suggesting that roots and associated mycorrhizae are highly dependent on current photosynthate for their maintenance (Högberg et al. 2001). A rapid decline in respiration following girdling also may be related to decreased nutrient uptake, reduced root growth, and loss of mycorrhizae; over time, however, root mortality will eliminate autotrophic respiration from total SR, thus providing an estimate of heterotrophic respiration. Here we report on an assessment of autotrophic respiration around mature beech and spruce trees in a mixed stand, based on reductions in SR following girdling.

Materials and methods

The study was conducted in a mixed forest (Kranzberger Forst) near Freising, Germany (485 m a.s.l., 48°N 11°E), where a set of long-term research plots is located (Pretzsch et al. 1998). The soil was a luvisol derived from loess over tertiary sediments with good water and nutrient supply. Five mature beech (Fagus sylvatica, dbh 11.7 cm, SD = 3.7 cm) and five Norway spruce trees (Picea abies, dbh 20.4 cm, SD = 5.5 cm) aged between 40 and 50 years were girdled at breast height on August 19, 2002 (plot no. 6, Pretzsch et al. 1998, 48°24′45′′ N 11°39′50′′ E). SR was measured prior to and after girdling along transects at 0.5, 1.0, and 1.5 m from the base of beech, and 0.5, 1.0, and 2.0 m from the base of spruce (Fig. 1a, b). Transects were established in a direction to minimize the possibility of influence from roots of non-girdled trees, and coring at the end of the experiment confirmed that roots of non-girdled trees were not occupying the girdled tree sample locations (data not shown). Distances from the bole were selected to be within the canopy drip line for each species. An additional five trees of each species were selected as controls (beech dbh 13.4 cm, SD = 3.8; spruce dbh 27.8, SD = 8.8) and similar transects established to follow CO2 efflux.

Fig. 1
figure 1

Map of basal stem areas of the assessed control and girdled beech (a) and spruce trees (b) in relation to neighboring tree individuals on the site. Symbols size is proportional to stem width at breast height, and arrows show direction of respiration measurement points

Full size image

SR was measured biweekly from mid-August 2002 to November 2002, and then monthly from March to June 2003, using an infrared CO2 analyzer system (EGM-3 Environmental Gas Monitor, PP Systems, UK). Precipitation and soil water content (SWC) during the study period are given in Figs. 2 and 3. During each measurement period, the cuvette of the SR assessment system was placed at positions marked by a PVC ring on the soil surface, and the rate of CO2 efflux was followed until a steady state was achieved (a maximum of 80–120 s). Readings were taken at cuvette CO2 concentrations no more than 50–70 ppm above the starting level of 350–400 ppm. Soil temperature was measured at 3 cm depth with a stainless steel probe (Soil Temperature Probe, PP Systems, UK) and varied less than ±1.0°C across the site on any sample date. Measurements of the volumetric SWC by time domain reflectrometry technique (TDR) were provided by a monitoring station on an adjacent site within “Kranzberger Forst” [Bayerische Landesanstalt für Wald und Forstwirtschaft (LWF), 48°24′24′′N 11°39′22′′E] to show general patterns in soil moisture during the dry summer of 2003 (Fig. 3). After the study was completed, control plots were cored to a depth of 10 cm and roots were divided into fine (roots <2 mm) and small (2–5 mm) root classes to identify patterns of root density in relation to distance from the bole (Table 1).

Table 1 Fine (<2 mm) and small (2–5 mm) root density (g l−1 ± SE) in the upper 10 cm of soil and average SR at three distances from the bole of control beech and spruce
Full size table

The relationship between soil temperature and SR was analyzed using linear regression and by comparing means for each measurement location. To estimate the contribution of autotrophic (root + mycorrhizal) respiration to total SR, the mean decrease in SR following girdling was calculated for each sample date ([SRGird/SRCont−1] × 100), where SRGird = SR in girdled tree plots and SRCont = SR in control plots, which eliminated the confounding effects of moisture and temperature over the course of the season. To examine spatial changes in respiration in relation to distance from the bole, SR in control plots was adjusted to a temperature of 15°C using a Q10 of two and compared to small root biomass (Table 1).

Results

SR in control plots, corrected to a constant temperature of 15°C, did not show a strong pattern relating to root density or distance from the bole (Table 1). There was a slight decrease in spruce SR from 0.5 to 2.0 m from the bole, which was weakly related to fine + small root density (upto 5 mm). Fine and small root density was patchy and showed no significant pattern in relation to distance from the bole in either species.

Soil temperature explained roughly two-third of the seasonal variation in SR observed from August 2002 to May 2003 (Tables 2, 3). The response to soil temperature over this time period was linear, with rates of SR increasing with increasing soil temperature. The slopes were similar in both beech and spruce, and did not differ between controls and girdled trees. During the extraordinary dry period from May to September 2003, temperature and SR appeared uncoupled and SR was strongly related to SWC (Tables 2, 3). SWC explained approximately 90% of the seasonal variation in beech and 46% and 84% of the variation in SR in control and girdled spruce, respectively. In both spruce and beech, SR was lower in August 2003 than August 2002, which corresponded to lower rainfall amounts during the exceptionally warm and dry summer of 2003 (Fig. 2; Gietl 2004). The strong influence of soil temperature on SR during 2002 and the absence of such an effect in 2003 can also be seen in Figs. 4 and 5.

Table 2 Linear relationship between SR and soil temperature and volumetric soil water content for different time periods throughout the study for girdled and control beech
Full size table
Table 3 Linear relationship between SR and soil temperature and volumetric soil water content for different time periods throughout the study for girdled and control Norway spruce
Full size table
Fig. 2
figure 2

Monthly rainfall amounts (mm) for 2002 and 2003 (Gietl 2004)

Full size image

An immediate decrease in SR was observed in beech upon girdling compared to control plots (Fig. 4). Differences were greatest at distances of 0.5 and 1.0 m from the base of girdled beech trees, both during 2002 and 2003. SR was also decreased at 1.5 m from girdled trees relative to controls, but differences were smaller in magnitude.

In Norway spruce, there was no apparent effect of girdling on SR until October, approximately 6 weeks after girdling (Fig. 5). From October to November 2002, SR was significantly reduced (51.2%±2.3 SE) at 0.5 m from the bole of the tree. The general pattern of decline in girdled spruce persisted until August 2003, although the magnitude of response was less than at the end of 2002. At 2.0 m from the base of girdled spruce, there was no apparent change in SR following girdling, except during two sample periods in the summer of 2003.

The contribution of autotrophic respiration to total SR was estimated for two time periods: August to November 2002; and May to September 2003 (Fig. 6). In beech, SR decreased by 49.8 (SE = 2.6) and 49.0% (SE = 3.5) during the 3 months following girdling at 0.5 and 1.0 m, respectively, from the base of the tree (Fig. 6a). A smaller decrease was observed in beech at 1.5 m from the tree. In spruce, SR decreased by 37.7% (SE = 5.8) at 0.5 m and less than 15% at 1.0 m during the 3 months following girdling (Fig. 6b). There was no indication that girdling had any effect at 2.0 m from the base of spruce trees. During 2003, there was a slight decrease in SR at 0.5 m and no decrease at 1.0 and 2.0 m.

Discussion

During most of the study period, temperature appeared to be the dominant factor influencing seasonal changes in SR (Tables 2 and 3). Others have shown the dominant role of temperature in controlling SR, particularly in the absence of soil water stress (Janssens and Pilegaard 2003; Yuste et al. 2003; Raich and Schlesinger 1992; Hanson et al. 2000). Temperature and SWC explained 95% of the variation found in a Pinus sylvestris L. stand in Belgium (Yuste et al. 2003). In the current study, soil temperature explained roughly two-third of the seasonal variation in SR observed from August 2002 to May 2003. But during the dry period from June to September 2003, there was no apparent correlation between soil temperature and SR, while SWC collected from an adjacent site showed a strong correlation with SR (Tables 2 and 3). In 2003, volumetric SWC dropped below 25% in June at 5 cm depth, corresponding to a soil water potential of less than −0.5 MPa (Hammel and Kennel 2001), and remained below 25% until end of December (Fig. 3). Water reserves were depleted until mid-August. In these soils water is not available for plant uptake below 14% SWC due to the clay content of the soil. In an adjacent stand at “Kranzberger Forst” (Nunn et al. 2002; “plot no. 1” in Pretzsch et al. 1998), pre-dawn water potentials of beech and spruce trees in August 2003 were found to be as low as −1.4 and −1.6 MPa, respectively (M. Löw, personal communication), which represents exceptionally severe water tension for forest site conditions of Central Europe. Yuste et al. (2003) and Rey et al. (2002) both found that temperature was a dominant factor controlling SR when soil water was not limiting, but when SWC dropped below 15–20%, temperature was not correlated with SR.

Fig. 3
figure 3

Soil water content (vol%) measured on an adjacent site during 2002 and 2003 at a depth of 5 cm using TDR probes (Raspe et al. 2004)

Full size image

The immediate decrease in SR around beech trees following girdling was similar to what was found after girdling Scots pine trees, and illustrates the importance of current photosynthate in maintaining autotrophic, and hence total, SR (Ekblad and Högberg 2001; Högberg et al. 2001). In beech, a drop of nearly 50% was observed at both 0.5 and 1.0 m distances from the bole shortly after girdling (Fig. 4). Although the quantity of starch reserves in the course roots can supply respiratory substrate for at least several days, past studies have shown rapid decline in respiration after girdling. Decreased carbohydrate transport to roots following girdling could rapidly lead to down-regulation of growth and maintenance processes via sugar signaling (Koch 1996), even when sufficient reserves are available. Högberg et al. (2001) reported rates of autotrophic respiration approaching 56% in a boreal Scots pine forest during the first summer after girdling, increasing to 65% during the second year after girdling (Bhupinderpal-Singh et al. 2003). In our study, estimates of autotrophic respiration were lower (18–30%) during the second year, possibly due to the severe drought during the summer of 2003. Drought may not affect autotrophic and heterotrophic respiration to the same extent, a possibility supported by our preliminary results employing different methods on an adjacent site (P. Nikolova, unpublished data).

Fig. 4
figure 4

Soil respiration (SR, g CO2 m−2 h−1) at three distances from the bole of control and girdled beech trees from August 16, 2002 to September 15, 2003. Mean soil temperature did not vary significantly across the site and is shown in the bottom panel. Error bars represent ±1 standard error. Treatment trees were girdled after the first SR measurement was made

Full size image

Estimates of the contribution of autotrophic to total SR were less certain for spruce than beech. Spruce girdling resulted in nearly a 40% drop in SR at 0.5 m from the bole, averaged over 3 months following girdling (Fig. 6b). However, SR did not change in response to girdling at any distance from the bole for the first 6 weeks after girdling (Fig. 5). The 6-week delay in response to girdling may be due to greater stored carbohydrate reserves in spruce than in beech. If the first 6 weeks after girdling are removed from the analysis, estimates of autotrophic respiration were greater than 50% of total SR at 0.5 m during the fall of 2002. Similar to beech, estimates of autotrophic respiration were less during the drought of 2003, with autotrophic respiration comprising approximately 18% of total respiration at 0.5 m.

Fig. 5
figure 5

Soil respiration (SR, g CO2 m−2 h−1) at three distances from the bole of control and girdled Norway spruce trees from August 16, 2002 to September 15, 2003. Mean soil temperature did not vary significantly across the site and is shown in the bottom panel. Error bars represent ±1 standard error. Treatment trees were girdled after the first SR measurement was made

Full size image
Fig. 6
figure 6

Percent change in soil respiration following girdling, at three distances from the bole of beech (a) and spruce trees (b). Columns represent data averaged from August to November 2002 and from May to September 2003. Percent change calculated as (SRgirdled/SRcontrol−1) × 100 for each sample date, averaged over the time interval. Error bars represent ±1 standard error

Full size image

Fine and small root density (0–5 mm diameter) was patchy within the experimental transects and there was no clear pattern of respiration relating to SR in either beech or spruce (Table 1). Mean SR of beech controls, adjusted to a constant temperature of 15°C, was 0.61, 0.72, and 0.52 g CO2 m−2 h−1 at 0.5, 1.0, and 1.5 m from the bole, respectively. A slight decrease in SR was found with increasing distance from spruce boles, with mean SR decreasing from 0.75 g CO2 m−2 h−1 at 0.5 m to 0.51 and 0.48 g CO2 m−2 h−1 at 1.0 and 2.0 m from the bole, respectively. It is generally assumed that total root density decreases with distance from tree bole (Ammer and Wagner 2002); however, cores taken in beech and spruce control plots after the experiment completely revealed high spatial variability of roots less than 5 mm in diameter, with no apparent pattern relating to distance from the bole in beech and only a weak (ns) pattern in spruce (Table 1). Decreased SR with increasing distance from the bole in spruce may reflect decreased heterotrophic or mycorrhizal activity as well as fewer fine and small woody roots. The decrease in estimated autotrophic respiration with distance from the bole (Fig. 6) was not due to interference by roots of adjacent trees since cores in girdled plots revealed essentially no recolonization by new roots at the end of the study (data not shown).

Estimates of the contribution of autotrophic respiration to SR found in this study are similar to reports in other forests using different methods (Hanson et al. 2000). Nakane et al. (1996) compared forested and clear-cut areas and found that autotrophic respiration contributed 51% to total SR in a deciduous forest in Japan. Brumme (1995) estimated autotrophic respiration by comparing SR in a 146-year-old beech forest to SR in an adjacent gap (without roots), and concluded that the autotrophic fraction comprised approximately 40% of total SR. More recently, Epron et al. (2001) estimated root-rhizosphere respiration in a 30-year-old beech forest in France and found rates ranging from 30% to 60%, with a seasonal mean of 52%. Our results approaching 50%, based on younger trees, are consistent with earlier reports and suggest that autotrophic respiration may comprise a larger portion of total SR in younger than older forests. In addition, our estimates suggest that the autotrophic contribution to total SR may be greater during wet than dry years, a possibility that warrants additional investigation. Collectively, the results show that autotrophic respiration contributes significantly to total SR, and illustrate the need to understand factors that control carbon allocation belowground and hence root respiration in forested ecosystems.