Climatic information from δ13C in plants by combining statistical and mechanistic approaches.

The approach commonly used to assess the potential for climate reconstruction is to use linear regressions to compare the isotopic signal stored in archives to instrumental climatic data sets. A new method is proposed that combines statistical and mechanistic approaches to extract climatic information from δ13C records in organic matter. Both a spatial and a temporal gradient of 13C discrimination in a moss species commonly found in temperate and tropical peat bogs are compared to meteorological records. The relevance of fossil and modern analogues to elucidate palaeoenvironment records are tested. It was found that the magnitude and, in some cases, the direction of the impact of temperature, humidity and CO2 atmospheric concentration on 13C discrimination depend on the calibration set considered. The use of a mechanistic model is shown to help greatly in specifying the joint influence of the climatic variables. [ABSTRACT FROM AUTHOR]