A multi-species comparison of delta13C from whole wood, extractive-free wood and holocellulose.

The stable carbon (C) isotope composition (delta13C) of tree rings is a powerful metric for reconstructing past physiological responses to climate variation. However, accurate measurement and interpretation are complicated by diagenesis and the translocation of compounds with distinct isotopic signatures. Isolation and analysis of cellulose minimizes these complications by eliminating variation due to biosynthetic pathways; however, isolation of cellulose is time-consuming and has no clear endpoint. A faster and better-defined analytical method is desirable. Our objectives were to determine if there is a direct relationship between the isotopic compositions of whole wood (WW), whole wood treated with solvents to remove mobile extractives (extractive-free wood; EF) and holocellulose (HC) isolated by extractive removal and subsequent bleaching. We also determined if total C concentration could explain the isotopic composition and variation among these three wood components of each sample. A set of wood samples of diverse phylogeny, anatomy and chemical composition, was examined. The mean offset or difference between HC and EF delta13C was 1.07 +/- 0.09 per thousand and the offset between HC and WW was 1.32 +/- 0.10 per thousand. Equivalence tests (with alpha = 0.05) indicated that the relationship between EF delta13C and HC delta13C had a slope significantly similar to 1 +/- 5.5%, whereas for the WW delta13C: HC delta13C relationship, the slope was significantly similar to 1 +/- 10.08%. A regression model using EF delta13C to predict HC delta13C had a slope of 0.97, which was not significantly different from unity (P = 0.264), whereas the regression for WW had a slope of 0.92 which was significantly different from unity (P = 0.0098). Carbon concentration was correlated with HC:WW offset and cellulose:EF offset (P = 0.0501 and 0.007, respectively), but neither relationship explained much of the variation (r2 = 0.12 and 0.14, respectively). We suggest that HC extraction is unnecessary for most analyses of tree-ring delta13C; a simple solvent extraction is a suitable alternative for many applications.