Carbon isotope ratios of plants are highly informative for the reconstruction of ancient environments and for the interpretation of plant physiological processes to climate, but their responses to changing atmospheric CO2 concentration are currently debated. Moreover, plants in the geological past have experienced long-term low CO2 concentration (LC). However, the effects of LC on the plant C isotope ratios are still elusive. To investigate effects of atmospheric CO2 concentration ([CO2]) and drought on isotope ratios of plant metabolites we grew winter wheat (Triticum aestivum) in climate-controlled chambers under different [CO2] covering glacial, pre-industrial, and present concentrations (170, 280, and 400 ppm) and water regimes (well-watered and drought). First, we quantified total C isotope discrimination between plant and atmosphere (Delta) using C-13 on-line measurements of plant gas exchange and C-13 values of plant metabolites, i.e., cellulose, n-alkane, and phospholipid fatty acids (PLFA). We found that LC yielded a higher Delta regardless of water regime, i.e. more C-13-depleted values were found under LC; the effect was stronger for n-C-29 alkane (1.5 parts per thousand/100 ppm) and C16:0 PLFA (1.1 parts per thousand/100 ppm) than that for cellulose (0.6 parts per thousand/100 ppm). We then calculated post-photosynthetic C isotope shift (epsilon) between specific metabolites and plant bulk isotope values. delta C-13(n-C29) (alkane )and delta C-13(P)LFA were 8.3 parts per thousand and 7.3 parts per thousand lighter than the delta C-13(bulk) under 400 ppm; these depletions became higher (9.8 parts per thousand and 8.2 parts per thousand lighter than the delta C-13(bulk) for n-C-29 alkane and PLFA, respectively) under 170 ppm. In contrast, delta C-13(cellulose) was 1.2 parts per thousand heavier than the delta(13)C(bulk )under 400 ppm while this enrichment became higher (1.6 parts per thousand) under 170 ppm. Changes in atmospheric [CO2] affected C fractionation not only via photosynthetic but also post-photosynthetic processes and thus must be taken into account when interpreting C isotopes for paleoclimate reconstruction and future global C cycle prediction. (C) 2018 Elsevier Ltd. All rights reserved.