Deriving xylem water isotopic compositions from in situ transpiration measurements: opportunity for plant source water identification?

The isotopic signature of xylem water (δ_X) is of great interest for plant source water studies. δ_X is usually derived by destructive sampling and subsequent cryogenic vacuum extraction (CVE). However, numerous studies have criticized this approach due its methodological constraints and analytical artifacts. New in situ methods to derive δ_X have been proposed in recent years. Yet, they are still in the development- and test phase, and their application highly intrusive. Gas exchange chamber techniques, on the other hand, have been well established for decades and allow the isotopic signature of transpired water (δ_T) to be monitored in high temporal resolution. As δ_T values approach δ_x values when transpiration is at isotopic steady state, measurements of δ_T may provide a relatively non-intrusive method to derive δ_X values.  

While conducting a large-scale long-term drought experiment of 90 days in a model rainforest ecosystem (Biosphere 2, WALD project), we monitored dynamics in δ_T values in two tropical plant species, one understory and one canopy species. Severe drought was ended with a deep water pulse strongly enriched in ²H. By connecting flow-through leaf chambers to a water isotope analyzer, we measured δ_T in a 2-h resolution and observed its response to increasing drought, deep labelling and subsequent recovery. Parallel to continuous measurements of δ_T, branch samples were collected at 5 time points throughout the experiment to determine δ_X values from CVE.

We found that δ_T values provide a good proxy for δ_X values when using the daily averages of δ_T values, weighted by the transpiration flux; derived δ_X values matched well with isotopic compositions of soil water. In situ-δ¹⁸O_X agreed well with values from CVE. CVE-δ²H_X values, however, were strongly enriched in ²H in comparison to in situ derived values, which is probably linked to the isotopic effects of CVE on δ²H as recently found in several studies. Moreover, by adding a ²H deep water pulse, δ_T allowed us to distinguish clearly between deep and shallow soil water use as well as show uptake velocities of newly added water. Monitoring  δ_T using gas exchange chambers provides a good proxy for δ_X values to address research questions concerning plant-available water sources and their usage, and at the same time give additional information on the plant water status.