Uncertainties in static closed chamber measurements of the carbon isotopic ratio of soil-respired CO2

The δ13C of soil-respired CO2 (δr) is frequently determined using static closed chamber methods. δr is obtained as the intercept of the least squares linear regression of δ vs 1/C*, where measured δ13C–CO2 (δ) and volume fraction of CO2 (C*) values of chamber headspace samples are used. Theoretically, we show that the variance of the estimate of δr can be reduced by extending the 1/C* interval of the regression towards (i) higher or (ii) lower values, or (iii) distributing the 1/C* values optimally within the pre-selected headspace CO2 sampling time period. Experimental applications of these approaches indicated that: (1) lowering the initial CO2 level, thereby increasing 1/C*, yielded a positive bias to the δr result. (2) It was feasible to obtain lower variance in the δr estimate by lowering 1/C* values through extended CO2 sampling time. We also recommend that each chamber is sampled only once, mainly because this allows freedom to select the sampling times, in order to optimize the distribution of 1/C* values.