(both: internal diameter 0.25 mm, film thickness 0.25 mm; Agilent), a constant He-flux (99.996% pure) of 2 ml min1 and the temperature program presented in Supplementary Table 2. CO2 reference gas (99.995% pure) was injected for 20 s into the detector several times throughout the measurement to identify any drift of the d13C-values. The d13C of the second reference gas peak was defined as 40‰ and all other d13C values were calculated by comparison. The chromatograms were evaluated with ISODAT NT 2.0. Linear regressions were calculated from CO2 reference gas (99.995% pure) peaks to correct for any drift during measurements. To correct for amount-dependent 13C isotopic fractionation during measurements (Schmitt et al., 2003), and for the addition of
C during derivatization, linear and logarithmic regressions of the
external standards d13C-values to their area were calculated. If both
regressions were significant, that with the higher significance was
applied. As the d13C-value for the derivatizing agentswas unknown,
the correction was performed according to Glaser and Amelung
(2002) (Eq. (3)).