Abstract

Over the last decade we have mapped and radiocarbon dated in-stream wetland and ground-water discharge deposits along a north-south transect in the Atacama Desert to reconstruct changes in late Quaternary hydrology. Additionally, our research group has collected and analyzed over 250 rodent middens from the same region to identify vegetation migrations associated with fluctuations in precipitation. This combined record of changes in hillslope vegetation and hydrology away from fluvial systems (groundwater-discharge deposits) allows us to assess the influence of climate change on streams with in-stream wetlands. Specifically, this combined record allows us to evaluate the relative role of stream power versus the resistance of the streambed to erosion during climate change, determine the sensitivity of different size stream catchments to climatic perturbations, and constrain the response times of in-stream wetlands to climatic events. Modern in-stream wetlands are widespread in the driest sector of the Atacama Desert, between ~20� - 26�S. They occur in diverse settings including deeply incised volcanic canyons and meandering streams that flow through basin centers. Although some in-stream wetland systems experience large discharge events (e.g. 310m3s-1) compared to normal stream flow of ~1m3s-1, the number of annual discharge events is limited (~5 to 10) and confined to the austral summer. Seven different in-stream wetlands possess time-stratigraphic units of similar age, indicating that geomorphic change in these systems is largely controlled by climate. Comparison of these time-stratigraphic units with groundwater-discharge deposits and rodent middens indicate that all in-stream wetlands aggrade during wet intervals and incise during dry periods, suggesting that the resistance of the sreambed to erosion is the main control on alluvial cut-and-fill cycles in these systems. Most cut-and-fill cycles span between 5000 to 3000 years, although during the late Holocene the cut-and-fill cyclicity increased dramatically with events lasting 500-1000 years, potentially due to enhanced ENSO variation. In general there does not appear to be a difference in the response times between large and small stream catchments to climate change, with all system responding in <500 years to climatic events.