LAUSR

Paleo-dune deposits have been widely used as a proxy indicator of past dune activity, which is further used to reconstruct paleoclimates. However, recent studies have critically examined the reproducibility of dune chronologies and the complexity of paleo-dune deposits as paleoenvironmental records. This paper addresses questions on the paleoenvironmental implications of dune chronostratigraphies that have been raised by those reviews, in the specific case of crescentic dunes, using a case study from the Mu Us dune field, north-central China. The processes of turn-over and stabilization of relatively small crescentic dunes are first investigated by observational evidence. In combination with the analysis of a simplified sand preservation model and stratigraphic records, the effect of dune morphodynamics on sand preservation is demonstrated. It is especially evident that thick, nearly instantaneously deposited sand units record dune stabilization near the very end of a dune activity episode, while thin sand units may signal the preservation of sand deposited earlier in episodes of activity. Interpreting the distribution of luminescence ages that indicate sand deposition over time is not as simple as assumed in some previous work. Low frequency of sand ages could indicate an interval of either dune field stabilization or extensive dune activity but poor sand preservation. A peak of sand age frequency likely represents a shift in dune field activity towards stabilization, not a peak of active dune extent, especially if it partially overlaps with an independently identified interval of stabilization (e.g. one recorded by paleosols). The nature and magnitude of these biases in the distribution of sand ages over time are strongly affected by the magnitude of net sand accumulation, which is in turn related to sand supply, transport capacity and sand availability, and ultimately climate change. Relatively short dune stabilization and turn-over time (101 to 102 yrs) indicate that dune geomorphic processes can quickly respond to short-term disturbance, but the chronology of that response must be interpreted in light of how those processes influence age distributions.