Carbon fluxes to the soil in a mature temperate forest assessed by ¹³C isotope tracing

Abstract

Photosynthetic carbon uptake and respiratory C release from soil are major components of the global carbon balance. The use of ¹³C depleted CO₂ (δ¹³C = −30‰) in a free air CO₂ enrichment experiment in a mature deciduous forest permitted us to trace the carbon transfer from tree crowns to the rhizosphere of 100–120 years old trees. During the first season of CO₂ enrichment the CO₂ released from soil originated substantially from concurrent assimilation. The small contribution of recent carbon in fine roots suggests a much slower fine root turnover than is often assumed.¹³C abundance in soil air correlated best with temperature data taken from 4 to 10 days before air sampling time and is thus rapidly available for root and rhizosphere respiration. The spatial variability of δ¹³C in soil air showed relationships to above ground tree types such as conifers versus broad-leaved trees. Considering the complexity and strong overlap of roots from different individuals in a forest, this finding opens an exciting new possibility of associating respiration with different species. What might be seen as signal noise does in fact contain valuable information on the spatial heterogeneity of tree-soil interaction.

Appendix

The regression line of a Keeling plot can be derived from a simple mass balance equation, where the isotopic ¹³C signature of atmospheric CO₂ mixes with the ¹³C signature of soil respiratory CO₂ (see Eqs. 1, 2):

$$ \delta ^{{13}} {\text{C}}_{{{\text{mixed}}}} [{\text{CO}}_{2} ]_{{{\text{mixed}}}} = \delta ^{{13}} {\text{C}}_{{{\text{atmosphere}}}} [{\text{CO}}_{2} ]_{{{\text{atmosphere}}}} + \delta ^{{13}} {\text{C}}_{{{\text{respired}}}} \;[{\text{CO}}_{2} ]_{{{\text{respired}}}} $$

(1)

$$[{\text{CO}}_{2} ]_{{{\text{mixed}}}} = [{\text{CO}}_{2} ]_{{{\text{atmosphere}}}} + [{\text{CO}}_{2} ]_{{{\text{respired}}}} $$

(2)

where δ¹³C_mixed is the δ¹³C of the analyzed soil gas sample, [CO₂]_mixed is the CO₂-concentration of the analyzed soil gas sample, δ¹³C_atmosphere is the atmospheric δ¹³C, [CO₂]_atmosphere is above the canopy the atmospheric CO₂-concentration, δ¹³C_respired is the δ¹³C of respiratory CO₂, and [CO₂]_respired is the CO₂-concentration of respiratory CO₂ in the soil.

Except for δ¹³C_respired and [CO₂] _respired all parameters can be measured. Thus the equations can be solved for δ¹³C_respired. The equation solved for δ¹³ C_respired corresponds to an individual Keeling plot based on two points (one point defined by atmospheric CO₂ concentration and its δ¹³C, the other by the CO₂ concentration of the soil air sample and its measured δ¹³C). The difference to the classic Keeling approach is that δ¹³C_atmosphere and [CO₂]_atmosphere are assumed to be known.

Air samples collected 5–10 m above the canopy, showed highly variable values in both atmospheric CO₂ concentrations and δ¹³C. It is likely that this variation was influenced by the vegetation and the CO₂ enrichment experiment. Therefore we extrapolated the CO₂ concentrations and δ¹³C values from different monitoring stations (Ireland, Mace Head, Count Galway 53°20′N 9°54′E 25 m a.s.l.; Azores, Terceira Island 38°46′N 27°23′E, 30 m a.s.l; Malta, Gozo 36°03′N 14°11′E 475 m a.s.l; Tenerife, Izana Observatory 28°18′N 2°00′E 7 m a.s.l) to the year 2001. Since both atmospheric CO₂ and its appropriate δ¹³C vary seasonally due to a seasonal change in respiratory and photosynthetic activity of the vegetation, we used monthly data of the respective parameters. We assumed that the average of these data taken from different northern latitudes would give an approximate description of the free atmospheric CO₂ around our experimental site and thus could be used for both, calculation of soil respiratory δ¹³C in the treated and control area.

Sensitivity analysis of the Keeling approach

Since our new approach of using the Keeling plot is not yet established, we tested its robustness by calculating the C isotopic signature for pure respiratory soil CO₂ for every combination of atmospheric conditions in CO₂ concentration (between 350 and 400 ppm in 1 ppm steps) and δ¹³C (between −9 and −7‰ in 0.1‰ steps). In general the sensitivity of the calculated soil respiratory δ¹³C towards changing atmospheric conditions decreases with increasing soil air CO₂ concentrations. To be on the safe side, we performed a sensitivity analysis for a soil air sample with a CO₂ concentration of 2,300 ppm (δ¹³ C=−20.8‰), as was found in October and November, the 2  months with lowest soil air CO₂ concentrations. As shown in Fig. 8 the calculated isotopic value for pure respiratory CO₂ changes less than 0.015 ‰ when atmospheric CO₂ concentration shifts from 360 to 370 ppm (a shift which is related to ca. 0.5‰ in atmospheric δ¹³C). Even when the atmospheric δ¹³C is not adjusted to a change in atmospheric CO₂ concentration, a shift of 10 ppm in atmospheric CO₂ is followed by a potential error of less than 0.08‰ in the calculated respiratory δ¹³C. Therefore in the range of 360 ppm a change in atmospheric CO₂ concentration of 10 ppm does not significantly affect the calculated δ¹³C value of respired CO₂. The same modeling applied to a soil air sample with a CO₂ concentration of 1,000 ppm (δ¹³C=−18.4‰) revealed a shift of less than 0.1 ‰ when the atmospheric CO₂ concentration changed from 360 to 370 ppm. The maximal potential error was 0.2 ‰ when the related shift of 0.5‰ in atmospheric δ¹³C was not considered. Less than 3% of all soil air samples had a CO₂ concentration lower than 1,000 ppm. In general the soil samples showed a high enough CO₂ concentration so that the potential error of the calculated respiratory δ¹³C was below the measuring precision. Thus we conclude that our method of calculating the δ¹³C values of pure respiratory CO₂ of every single soil air sample is sufficiently accurate and can be used to visualize the spatial variability of the biological δ¹³C input into the ground.

Fig. 8

δ¹³C of pure respiratory CO₂ for varying atmospheric CO₂ concentration and atmospheric δ¹³C. For the simulation typical soil air CO₂ concentration (2,300 ppm) and δ¹³C (−20.8‰) were used. A 10 ppm increase in atmospheric CO₂, (roughly related to a decrease in atmospheric δ¹³C of 0.5‰, indicated by the arrow), results in a potential error of less than 0.1‰ (open triangle) in calculated soil respiratory CO₂. See Appendix for details