Abstract

Based on modern pollen samples from different vegetation units in southern Patagonia, showing a close relation to yearly amounts of precipitation and mean annual temperatures, different pollen-climate transfer functions are developed and tested. Comparing the performance of MAT (Modern Analogue Techniques), WA (Weighted Average), as well as WAPLS (Weighted Average Partial Least Square) statistical techniques, it is possible to determine the statistically most robust model (WAPLS for precipitation). This transfer function is then used to estimate palaeoprecipitation amounts based on Laguna Potrok Aike pollen results for the last 16,000 years. Generally, the results of the precipitation model indicate less precipitation during the Lateglacial and alternating wet and dry periods during the Holocene. The Holocene started with higher amounts of precipitation until about 8 ka cal. BP, followed by a period with lower amounts between 8 and 2.5 ka cal. BP, while the Late Holocene shows a general increase in precipitation. Comparisons with former shoreline reconstructions and carbonate concentrations in the sediments of Laguna Potrok Aike not always show similarities due to the complex environmental factors recorded by these proxies. Moreover, changes in the moisture availability due to the interplay of precipitation and temperature, cannot be reconstructed directly. Nevertheless, the general long-term trend of palaeoprecipitation is in accordance with the absolute moisture content in the air, which is determined mainly by temperature: during cold periods with less absolute moisture, the model shows less precipitation. Moreover, the model also points to a relation with the position and strength of the Southern Hemisphere Westerlies. (c) 2013 Elsevier Ltd. All rights reserved.