Leaf cellulose δD and δO trends with elevation differ in direction among co-occurring, semiarid plant species

The potential to reconstruct paleoclimate from analyses of stable isotopes in fossil leaf cellulose could be enhanced by adequate calibration. This potential is likely to be particularly great in mid-latitude deserts, where a rich store of fossil leaves is available from rodent middens. Trends in δD and δ¹⁸O of leaf cellulose were examined for three species growing across climatic gradients caused by elevation and slope aspect in southeastern Utah, USA. The species differed in morphology (Pinus edulis vs. Yucca glauca), photosynthetic pathway (C₃Y. glauca vs. CAM Yucca baccata) or both (P. edulis vs. Y. baccata). The δD_LCN (leaf cellulose nitrate) and δ¹⁸O_LC (leaf cellulose) values of P. edulis decreased with elevation. Stem water δD values either increased (in spring) or did not change with elevation (in summer). Needle water δD values usually decreased with elevation and differed greatly with leaf age. These results suggest that δ cellulose values of P. edulis record the effects of climate on the isotopic composition of leaf water but not climate effects on meteoric water. In contrast to P. edulis, δD_LCN values of Y. glauca increased with elevation. The δ¹⁸O_LC values of Y. glauca also increased with elevation but less significantly and only on south-facing slopes. The δ cellulose values in both P. edulis and Y. glauca were most significantly related to changes in temperature, although temperature and precipitation were negatively correlated in the study area. Where all three species co-occurred, their δD_LCN values differed but their δ¹⁸O_LC values were the same. The disparity in δD_LCN between Y. baccata and the other species corresponds to differences in biochemical fractionations associated with photosynthetic pathway. Biochemical fractionations may also contribute to differences between the two C₃ species. Knowledge of factors affecting responses of individual plant species to environment may be required to infer climate from δD_LCN and δ¹⁸O_LC.