On uncertainties in the plant source water isotopic composition extracted with cryogenic vacuum distillation

Recent studies challenge the use of plant water from cryogenic vacuum distillation (CVD) extraction in accurately representing the hydrogen and oxygen isotopic composition (δ²H and δ¹⁸O) of plant source water. This is hypothesized to be because the δ²H in extracted water depends on tissue relative water content (RWC), which might be explained by the exchange of H-atoms between water and organic material. Secondary hypotheses focus on extraction artefacts related to evaporation and sublimation, but clear evidence is lacking. Here, we hypothesized that the observed δ²H and δ¹⁸O offsets (Δ²H and Δ¹⁸O) are influenced by (i) an H-exchange effect, (ii) tissue water amount or RWC and (iii) evaporation and sublimation enrichments.
The hypotheses were systematically tested by three corresponding experiments. Firstly, we added a range of strongly depleted reference water (δ²H: ca. -460‰; δ¹⁸O: ca. -170‰; 50–1200 μl) to organic materials (with and without exchangeable H) of constant weight (200 mg), followed by a 24 h incubation. In addition, the same range of pure reference water and tap water without any material were used as controls. Secondly, we incubated dry stem segments (Larix decidua) of different sizes in excess of reference water for 24 hours, then they were took out for extracting known water contents from samples with known RWC. Accordingly, fresh twig segments from the same species were prepared for extracting water with natural abundance. Thirdly, a range of different amounts of reference water (50–1200 μl) was added directly into the water collection tubes of the CVD extraction system. In addition, 2 ml glass vials containing the same range of reference water amounts were incubated in a climate chamber at 25 °C and 50 % relative humidity with lids open for 2 hours. All the samples, except the water in the glass vials, were extracted using a standard CVD extraction method for 2 hours. 
We found that both Δ²H and Δ¹⁸O values were not related to changes in RWC. In contrast, we observed an inversely proportional relationships with water amount, i.e., the lower the water amount, the higher the Δ²H and Δ¹⁸O. For Δ²H, the pattern was more pronounced for materials with exchangeable H, which reached 150‰ at the lowest water amount and decreased to -20‰ with increasing water amounts when the depleted reference water was used. However, the pattern was much less pronounced for the samples with natural isotopic abundance, indicating that the magnitude of the pattern is probably dependent on isotope ratios of plant water and water vapour in the laboratory. The evaporation and sublimation tests both showed that the pattern was partly caused by an increasing isotopic enrichment with decreasing water amount.
In conclusion, we identified a significant artefact of CVD when water is present in small amounts, particularly when δ²H and δ¹⁸O of the water was below natural isotope abundances. We therefore recommend extracting > 600 μl of water. Moreover, we provide first evidence of a significant H-exchange effect, suggesting that using hydrogen isotopes for estimating plant source water will remain challenging in future.