Biochemical composition is not the main factor influencing variability in carbon isotope composition of tree rings
Résumé
From June to December, we determined the effects of variations in biochemical composition on δ13C of tree rings of 2-year-old oaks (Quercus petraea (Matt.) Liebl.) growing under semi-natural conditions, and the dependence of these effects of water stress during the growth season. Percent abundance, carbon concentration and δ13C were measured in holocellulose, lignin, extractive-free wood, starch and a water-soluble fraction. Relative concentrations of lignin and holocellulose in the extractive-free wood varied little during the season or in response to water stress, indicating that these compounds could not quantitatively explain the variations in whole-wood δ13C. Among all sampled tree rings, the relative concentration of each structural compound (holocellulose and lignin) accounted for less than 5% of the δ13C variability in whole wood. Variations in holocellulose and extractive-free wood δ13C between tree rings were almost identical (r > 0.95), whereas variations in lignin δ13C were less well correlated to these compounds. Whole-wood δ13C had a slightly altered isotopic signal compared with that of the structural compounds because of the presence of the extractive component. These results showed that variations in lignin δ13C and lignin concentration have little influence on extractive-free wood δ13C and whole-wood δ13C. Rather, holocellulose influences δ13C the most. Thus, we confirmed that, for climatic reconstruction from tree rings, removal of extractives by soxhlet is generally sufficient and sometimes unnecessary. Our findings also indicate that, in the case of rapid and severe water stress, the structural component did not accurately record the associated increase in δ13C because of dilution with previously formed organic matter and cessation of trunk growth. The effect of drought on carbon isotope ratios was more pronounced in the extractive compounds, making them good water stress indicators but only on a scale of days to months.