The understanding of the dynamics who controlled, since the last glacial optimum, the arid and hyper-arid ecosystems mosaic in the Arid Central Asia (from the Caucasus to Mongolia) is determining to improve our response to the up-coming climate change. Especially because these desert-steppe areas will quickly be spread due to the the climate change. Thus, the paleoclimate and paleoenvironment issues in these ecosystems have to be studied using proxies conserved in archives mainly from lake and peat. These proxies (pollen, fungal spores, bacterial lipid biomarkers…) must be calibrated globally and locally to determine a signal much reliable as possible. This proxy signal is then convert in climate or vegetation reconstruction (using transfer functions, machine learning, multi-regression calibrations or indices…) and compared with surrounding archives. Finally, the ecological approach and archeological studies coupling permit us to draw more precisely the history of the nomads and pastoral societies in Central Asia.