Contributions of evaporation, isotopic non-steady state transpiration and atmospheric mixing on the delta18O of water vapour in Pacific Northwest coniferous forests.

Changes in the 2H and 18O of atmospheric water vapour provide information for integrating aspects of gas exchange within forest canopies. In this study, we show that diurnal fluctuations in the oxygen isotope ratio (delta 18O) as high as 4% per hundred were observed for water vapour (delta (18)Ovp) above and within an old-growth coniferous forest in the Pacific Northwest region of the United States. Values of delta 18Ovp decreased in the morning, reached a minimum at midday, and recovered to early-morning values in the late afternoon, creating a nearly symmetrical diurnal pattern for two consecutive summer days. A mass balance budget was derived and assessed for the 18O of canopy water vapour over a 2-d period by considering the 18O-isoflux of canopy transpiration, soil evaporation and the air entering the canopy column. The budget was used to address two questions: (1) do delta 18O values of canopy water vapour reflect the biospheric influence, or are such signals swamped by atmospheric mixing? and (2) what mechanisms drive temporal variations of delta 18Ovp? Model calculations show that the entry of air into the canopy column resulted in an isotopically depleted 18O-isoflux in the morning of day 1, causing values of delta 18Ovp, to decrease. An isotopically enriched 18O-isoflux resulting from transpiration then offset this decreased delta 18Ovp later during the day. Contributions of 18O-isoflux from soil evaporation were relatively small on day 1 but were more significant on day 2, despite the small H2(16)O fluxes. From measurements of leaf water volume and sapflux, we determined the turnover time of leaf water in the needles of Douglas-fir trees as approximately 11 h at midday. Such an extended turnover time suggests that transpiration may not have occurred at the commonly assumed isotopic steady state. We tested a non-steady state model for predicting delta 18O of leaf water. Our model calculations show that assuming isotopic steady state increased isoflux of transpiration. The impact of this increase on the modelled delta 18Ovp was clearly detectable, suggesting the importance of considering isotopic non-steady state of transpiration in studies of forest 18O water balance.