Plant- and micromammal-based paleoprecipitation proxies : Comparing results of the Coexistence and Climate-Diversity Approach

Abstract Both plant and vertebrate communities have been used to reconstruct paleoprecipitation. However, direct stratigraphic comparisons between the two types of proxies have hardly been performed, which is due to the fact that plant and vertebrate fossils usually do not occur together in single geological beds. Here, we focus on a series of 18 sites from the Neogene of Europe and Anatolia that contain both types of fossils, and compare paleoprecipitation predictions produced by the Coexistence Approach (plants) and the Climate-Diversity approach (micromammals). Most of the sites have overlapping uncertainty intervals for mean annual precipitation as generated by the two methods, pointing to a common precipitation-related signal in flora and fauna. Nevertheless, a systematic difference appears to characterize drier sites, for which micromammal richness-based estimates for annual precipitation tend to be lower than plant-based coexistence-based estimates. Driest-month precipitations are in agreement for drier sites, but strongly differ for wetter sites, with micromammal richness-based estimates being higher. Overlap is poor for estimates of wettest-month precipitation, which are higher when based on plant-based coexistence. Possible sources for a bias towards higher annual values in plants are related to the inclusion of local, wetland elements in the analysis and/or to a lower taxonomic resolution for pollen and spores compared to macroflora. Micromammal-based annual estimates could be biased towards drier values, when smaller samples lack rare species that could be preferentially wet-adapted. Several of the higher driest- and lower wettest-month micromammal-based estimates could result from over-extrapolation beyond the calibration domain for arboreality as currently existing in the modern Western Old World. On the other hand, the use of several SE Asian reference taxa could add a high-precipitation (monsoonal) component to the plant-based predictions. In addition, wettest-month estimates for both plants and micromammals probably suffer from the poor performance of water excess as a limiting factor. An attempt to further increase the predictive power of the micromammal-based prediction equations was performed by incorporating non-linear terms, and by using an extended calibration set including selected plant-based precipitation estimates for micromammal assemblages without modern analogue in the Western Old World (“proxy inter-calibration”).