Estimating intraseasonal intrinsic water‐use efficiency from high‐resolution tree‐ring δ13C data in boreal Scots pine forests

Summary Intrinsic water‐use efficiency (iWUE), a key index for carbon and water balance, has been widely estimated from tree‐ring δ13C at annual resolution, but rarely at high‐resolution intraseasonal scale. We estimated high‐resolution iWUE from laser‐ablation δ13C analysis of tree‐rings (iWUEiso) and compared it with iWUE derived from gas exchange (iWUEgas) and eddy covariance (iWUEEC) data for two Pinus sylvestris forests from 2002 to 2019. By carefully timing iWUEiso via modeled tree‐ring growth, iWUEiso aligned well with iWUEgas and iWUEEC at intraseasonal scale. However, year‐to‐year patterns of iWUEgas, iWUEiso, and iWUEEC were different, possibly due to distinct environmental drivers on iWUE across leaf, tree, and ecosystem scales. We quantified the modification of iWUEiso by postphotosynthetic δ13C enrichment from leaf sucrose to tree rings and by nonexplicit inclusion of mesophyll and photorespiration terms in photosynthetic discrimination model, which resulted in overestimation of iWUEiso by up to 11% and 14%, respectively. We thus extended the application of tree‐ring δ13C for iWUE estimates to high‐resolution intraseasonal scale. The comparison of iWUEgas, iWUEiso, and iWUEEC provides important insights into physiological acclimation of trees across leaf, tree, and ecosystem scales under climate change and improves the upscaling of ecological models.


Fig. S1
Locations and photographs of the study sites in Finland.            photorespiratory assumptions. iWUE was derived from gas exchange (iWUEgas), tree-ring δ 13 C (iWUEiso) and eddy covariance data (iWUEEC). gm is mesophyll conductance, and f is the fractionation during photorespiration. Horizontal line represents the median, box represents the interquartile range, the tails extend to 1.5 times of the interquartile range, and dots represent outliers that are outside 1.5 times of the interquartile range. Letters indicate different correlation coefficient across different assumptions (t-test).

Fig. S9
Across-border correlations in tree-ring δ 13 C of Scots pine, which denotes the degree of use of previous-year reserves (Fonti et al., 2018). y axis is the first δ 13 C observation in the current-year tree ring and x axis is the last δ 13 C observation in previous-year tree ring. Pearson correlation coefficient and P value between x and y values are given for individual trees (blue) and site-representative mean values. In (a) and (b), the tree-ring δ 13 C data were corrected by the trend in δ 13 C of atmospheric CO2. In (c) and (d), the low-frequency trends in tree-ring δ 13 C data were removed by first differencing, as suggested by McCarroll et al. (2017). (a) Kolari et al. (2022).

Methods S1 LA-IRMS system
The LA-IRMS system is comprised of a laser unit (213 nm UV laser, LSX-213 G2+, by Teledyne Photon Machines), a combustion unit, a CO2 collection unit and an IRMS. The laser ablation system is equipped with a custom-made laser cell with an inner chamber volume of circa 6 cm 3 (TerraAnalytic). The laser is operated using separate software that allows the user to select appropriate lasing settings (laser energy, fire mode, spot size, scan speed etc.) and set up a sampling sequence for the analytical run. For the analysis of resin-extracted wood, we found the following settings to be appropriate: laser energy of 40% ( to determine the number of current-year tracheids in the enlargement, wall-thickening and lignification, and mature phases (Jyske et al., 2014). The growth curves for tracheid production and tracheid maturation were obtained via Gompertz fitting (Zeide, 1993) on the number of total and mature current-year tracheids, respectively, using the 'nlsLM' function of R package 'minpack.lm' (Elzhov et al., 2010).
Meanwhile, the dimensional growth curve of tracheid production and the growth curve of mature tracheid number were modeled in CASSIA with input parameters validated for Hyytiä lä and Vä rriö (Schiestl-Aalto et al., 2015). Based on tracheid cell dimensional measurements in radial direction (Jyske et al., 2014), we constructed a non-linear fitting curve between cell number and tree-ring dimension in CurveExpert 1.6.0 and transferred the number-based growth curves to dimensional growth curves. We evaluated the quality of CASSIA model results via comparing them with the observational results (Fig. S4).
However, there are several sources of uncertainties in tracheid growth prediction via the CASSIA model (Fig. S4), which uses air T as the main driver. First, even though temperature response function has proven to be good in most conditions (Schiestl-Aalto et al., 2015), it may underestimate or overestimate growth or development rate during long lasting unusually cold or warm periods. Second, although temperature is the most important factor determining tracheid growth (Jyske et al., 2014), other factors, such as photoperiod (Rossi et al., 2006) or water availability (Gruber et al., 2010), may also affect growth rhythm. Third, the model applies a simplistic description of different growth phases (Schiestl-Aalto et al., 2015), and thus the timings of, for example, enlargement and lignification period in relation to each other, may not be fully considered.
Considering the uncertainties in tracheid growth modeling, the modeled growth curves were shifted by at most 10 d per year and site. The results which gave best intraseasonal alignment between iWUEiso, iWUEgas and iWUEEC were reported.