Vapor pressure deficit governs the relative contribution of leaf and source water to intra-annual δ18O variations of tree rings

Oxygen isotopes (δ¹⁸O) in tree rings carry a strong potential to retrospectively evaluate tree water uptake and physiological response to climate. Their interpretation can, however, be challenging due to the complexity of the isotopic fractionations along the soil-tree-atmosphere continuum. Indeed, several processes play a role in defining the final tree-ring isotopic signal: source water variations, evaporative processes at the soil surface and leaf level, and mixing of xylem water that might exchange with new assimilates associated with phloem transport and synthesis of wood constituents. Disentangling these influences along the growing season and how climate conditions modify them are remaining challenges to exploit the full potential of δ¹⁸O tree-ring records as a climate proxy.

In this study, we aim at identifying the contribution of leaf water enrichment and source water on the tree-ring δ¹⁸O signature by assessing intra-annual variations of δ¹⁸O along the soil-leaf-tree ring pathway of larch (Larix decidua Mill.). We focus on two sites with contrasting water availability in the Lötschental valley (Swiss Alps) and three consecutive growing seasons (2011-2013). Our approach takes into consideration specific timing of the involved processes with a high spatio-temporal resolution: environmental conditions, diurnal sapflow-derived transpiration rates, δ¹⁸O analysis of xylem and leaf water, and intra-annual tree-ring δ¹⁸O measurements coupled with wood formation kinetics. Structural equation models (SEM) were applied to statistically assess the relations among δ¹⁸O values of the different pathway components. Furthermore, we calibrated mechanistic models of leaf-water and tree-ring cellulose δ¹⁸O to explore site-specific contributions of the fractionation processes (e.g., Péclet effect and the proportion of xylem-cellulose synthesis exchange [P_ex]) and investigated their climatic drivers.

Our results showed that intra-annual xylem water δ¹⁸O and transpiration rates differed between sites and years whereas needle water δ¹⁸O did not differ significantly between sites (but between years). However, tree-ring cellulose δ¹⁸O values were higher at the dry site resembling those differences observed in xylem water δ¹⁸O. SEMs reinforced these results since xylem water δ¹⁸O contributed more to cellulose δ¹⁸O in comparison to needle water δ¹⁸O, and this effect was more prominent at the dry site. Vapor pressure deficit (VPD) had strong control on the overall leaf water-related ¹⁸O-fractionations. However, mechanistic leaf-water δ¹⁸O models did not indicate a relevant role of the Péclet effect in our study. Most importantly, mechanistic models of cellulose δ¹⁸O revealed that P_ex was variable along the growing season and its variability was significantly associated with variations in VPD.

Our study suggests that the imprint of the source water signal on the δ¹⁸O signature in tree rings is highly dominant, particularly during episodes of high VPD, potentially overwriting signals coming from leaf fractionation processes.