Accounting for atmospheric carbon dioxide variations in pollen-based reconstruction of past hydroclimates

Abstract

Changes in atmospheric carbon dioxide (CO 2 ) concentration directly influence the ratio of stomatal water loss to carbon uptake. This ratio ( e ) is a fundamental quantity for terrestrial ecosystems, as it defines the water requirement for plant growth. Statistical and analogue-based methods used to reconstruct past hydroclimate variables from fossil pollen assemblages do not take account of the effect of CO 2 variations on e . Here we present a general, globally applicable method to correct for this effect. The method involves solving an equation that relates e to a climatic moisture index (MI, the ratio of mean annual precipitation to mean annual potential evapotranspiration), mean growing-season temperature, and ambient CO 2 . The equation is based on the least-cost optimality hypothesis, which predicts how the ratio (χ) of leaf-internal to ambient CO 2 varies with vapour pressure deficit (vpd), growing-season temperature and atmospheric pressure, combined with experimental evidence on the response of χ to the CO 2 level at which plants have been grown. An empirical relationship based on global climate data is used to relate vpd to MI and growing-season temperature. The solution to the equation allows past MI to be estimated from pollen-reconstructed MI, given past CO 2 and temperature. This MI value can be used to estimate mean annual precipitation, accounting for the effects of orbital variations, temperature and cloud cover (inferred from MI) on potential evapotranspiration. A pollen record from semi-arid Spain that spans the last glacial interval is used to illustrate the method. Low CO 2 leads to estimated MI being larger than reconstructed MI during glacial times. The CO 2 effect on inferred precipitation was partly offset by increased cloud cover; nonetheless, inferred precipitation was greater than present almost throughout the glacial period. This method allows a more robust reconstruction of past hydroclimatic variations than currently available tools. • A model using the least-cost optimality hypothesis to account for physiological effects of CO 2 • The model provides a correction for statistical reconstructions of the moisture index • The model provides a new way to reconstruct mean annual precipitation • Application to glacial pollen data shows climate wetter than inferred statistically