Figure 8: Frequency distributions of the seasonal origin index (SOI) for the summer half of the year (May through October) for beech(a) , spruce (b) and young spruce xylem (c)(upper row), mobile soil waters in 10 cm (d) , 20 cm(e) and 40cm (f) (middle row) and bulk soil waters in 10 cm (g) , 20 cm (h) , 40cm (i) (bottom row). The colored and gray bars indicate the distributions for campaigns with low and high antecedent precipitation, respectively, in the 30-day period prior sampling. The asterisk indicates that there are significant differences (p<0.05, t-test) between SOI in dry and wet antecedent conditions.
Our results suggest that although mean annual precipitation has impact on the lag between precipitation seasonality and xylem seasonality (Goldsmith et al. , 2022; Guo et al. , 2018; Williams and Ehleringer, 2000), antecedent wetness (precipitation in 30 days prior to the sampling dates) appears to have no significant effect (Figure 8). Thus, we hypothesize that overall root structures of the trees at a specific site, which are probably adapted to the average precipitation that a location receives, are dominantly affecting the lag between precipitation seasonality and xylem seasonality. Short-term (i.e., last 30 days) moisture availability had little effect on the seasonal signals in the xylem, thus we infer no shifts in the accessed water pools and uptake depths at our site for these two non-drought years. However, this might not hold for exceptionally dry years, as previous studies showed that trees might shift their water uptake depths during water limitations (e.g., Brinkmann et al. , 2019; Gartner et al. , 2009; Meißner et al. , 2012).
Methodological limitations of xylem and soil water signal interpretation
Although we took regular bulk soil samples with an auger, we did not see a change in the bulk soil signal (see Figure 2) resulting from the artificial creation of preferential flow paths through drilling (von Freyberg et al. , 2020). Recent studies have pointed to potential extraction bias when using cryogenic vacuum extraction (e.g., Chenet al. , 2020). Whereas for bulk soil water extractions, the bias is potentially small and thus negligible (Newberry et al. , 2017), Chen et al. (2020) clearly documented δ2H offsets in xylem waters. These offsets were attributed to the exchange of 2H in the wood tissue with waters in the xylem. However, in a more recent study, Diao et al. (2022) showed that these offsets are potentially small when large amounts (i.e., > 600 µl) were extracted. This was true for most of the xylem samples (>95%) we presented in the study, where we typically extracted > 1 ml of xylem water. Another effect on the xylem signal can originate from water stored in xylem not actively contributing to transpiration. In a recent study, Barbetaet al. (2022) used a cavitron centrifuge at specific spinning rates to specifically extract sap water from xylem and intra-cellular water stored in the xylem tissue. Barbeta et al. (2022) found that the sap xylem water matched the irrigation water (with no effects of isotopic fractionation during root water uptake), but the water extracted from xylem tissue storage was always depleted in δ2H; this depletion could potentially also influence our results. However, we calculated the effect of the potential offset by δ2H depletion, i.e., a bias of -6.1 ‰; Allen and Kirchner, 2022), and found that our major conclusion derived from Figure 6 does not change and forest trees (at our site) were indeed containing a mixture of water dominated by winter precipitation throughout the entire year (see Supplementary Material – Figure S3). Therefore, our main conclusions are robust against the potential biases introduced by water extraction that have been described elsewhere.