Xyl(t=2) = f1 * bulksoil(t=1) + f2 * P(t1->t2)
Where bulksoil(t=1) is the isotopic composition of water in the bulk soil at the previous timestep andP(t1->ts) is the volume weighed isotopic composition of precipitation that fell between the previous and most recent xylem sampling dates (typically the last three weeks).
Results and discussion
Isotopic variation of precipitation, soil waters and xylem water
From April 2020 to March 2022, we observed seasonal cycles in isotope ratios (shown for δ2H in Figure 2 and for δ18O in the supplementary material Figure S1) in incoming precipitation that corresponded with transitions between summer and winter seasons. Isotope ratios were lighter in winter and heavier in summer, with volume-weighted annual precipitation δ2H averaging ‑63.9 (and δ18O averaging -9.5 ‰), as shown by the dotted lines in Figure 2 b-d. Mixtures of precipitation in soil or plants that lie above that line (Figure 2) contain more summer precipitation than annual precipitation does (i.e., summer precipitation is over-represented in those soil and xylem samples), and mixtures below that line contain more winter precipitation than annual precipitation does (i.e., winter precipitation is over-represented in those soil and xylem samples).
Mobile and bulk soil waters showed a seasonal cycle in δ18O, similar to that of precipitation but with a dampened amplitude (Figures 2c&d); however, the beech, spruce and young spruce samples did not show a clear seasonal cycle (Figure 2b; Figure S1b). Readers should note that several of the collection dates occurred during the dormant season for beech, during which some of the highest and most variable xylem δ2H values were observed. Overall, most of the bulk soil samples plotted below the mean precipitation line, indicating that they over-represent winter precipitation; in contrast, the mobile soil water samples were quite evenly distributed around the mean precipitation signatures, suggesting that they, on average, reflect a less directionally biased mixture of precipitation. The mean mobile soil water δ2H (‑64.8 ± 0.5, mean ± standard error) was similar to the mean annual precipitation δ2H (‑63.9 ± 2.5), whereas the mean bulk soil water δ2H was substantially lighter (‑79.2 ± 0.7). Given the consistency among soil water samplers and close tracking of precipitation fluctuations, mobile soil waters seemed to reflect recent precipitation more than bulk soil waters did.