Estimation of past atmospheric carbon dioxide levels using tree-ring cellulose δ13C

We study the applicability of the Farquhar model for carbon isotopic discrimination (change in carbon isotopic composition from air CO2 to tree-ring cellulose) in C-3 plants to trees growing in the field. Two new carbon isotope datasets from Himalayan conifers with published data from another eight sites across the world show disparate trends in the plot of carbon isotope discrimination versus atmospheric carbon dioxide concentration, in contrast to the model prediction of absence of any trend. This is because the model assumes that the tree adjusts its stomatal conductance for water-use efficiency to maintain a constant ratio of carbon dioxide concentrations inside and outside the leaf and treats the diffusive and biochemical fractionation factors as constants. By introducing a simple linear dependence of these fractionation factors with ambient temperature and humidity, we have enhanced the applicability of the model to naturally growing trees. Further, despite the disparate trends exhibited by the 10 trees, we show using the inverse modelling that it is possible to derive a unique record of past atmospheric CO2 concentrations using tree cellulose delta C-13 data. The reconstructions also replicate the summer pCO(2) gradient from tropics to mid-latitudes. We also discuss the merits and demerits of the model, and compare the model-derived pCO(2) with that of the ice core-based records from Law Dome.