Résumé
Accurate climate reconstructions are important to better investigate the Holocene climate changes mainly because this period appears to be a smooth continuity to the current and the future climate warming and also because the human-societies and climate systems interactions begin from this period to be undeniably ravelled.In this context, an important issue address the understanding of the links between proxies (eg pollen and molecular biomarkers) and environmental forcing (climate…) through accurate calibrations on modern samples. Even if this environment-proxy causal relation has to be studied in a global biogeographical context (flowering plants producing pollen and bacteria or archaea producing biomarkers as Glycerol Dialkyl Glyceryl Tetraethers depend on vegetation, geological bedrock and soils as well as climate), some geographical areas are poorly documented. Among them, the Caucasus lowlands of Azerbaijan and the Touran basin to Pamir-Alay piedmonts and mountains of Uzbekistan and Tajikistan have never been investigated.In this study, we have collected and analysed new modern samples for pollen and GDGTs from Azerbaijan, Uzbekistan and Tajikistan. Calibrations are then established. The pollen samples have been compared to vegetation relevés to better understand the vegetation-pollen rain relation, and then, some transfer function methods (MAT and WAPLS) or machine-learning methods (BRT and RF) have been applied to reconstruct climate parameters from pollen assemblages. A particular focus has been carried on local condition for each surface samples: surrounding vegetation, biome extraction, soil description, sample type (top-core, moss polster or soil), in order to unravel the potential local bias in climate reconstruction. Finally, for both proxies, the discussed results and calibrations will be tested on two paleo-records from lakes Fazilman and Tuya (Uzbekistan).This study highlight the difficulties raised by the special conditions of these extreme environments (arid to semi-arid context, impact of soil pH, salinity and gypsum content) on pollen and brGDGTs vs. climate calibrations.