Abstract

We present a new paleoclimate sensor that reliably estimates mean annual rainfall (MAR) along the arid western slope of the Andes. The sensor exploits the modern relationship between chinchilla rat body size (estimated from the average width of the fecal pellets they produce which accumulate in middens along with plant macrofossils and other remains) and MAR as it varies across 14° of latitude and over 2000 m of altitude. The resulting correlation (R2 = 0.83) generates a transfer function which we then apply to radiocarbon-dated fossil rodent middens that are ubiquitous in northern Chile and southern Peru. Surprisingly, mean annual temperature (the variable typically associated with rodent body size) correlates very poorly with fecal pellet width in our dataset. We present results from two time-series collected in the central (23.5°S) and northern (18.45°S) Atacama Desert. The first series spans the last 14 ka and was chosen because we can compare the results to several other central Atacama paleoclimate records. The second series spans the last 600 yrs but here we first measured the pellets, and then dated the middens to eliminate biases and test reproducibility. For the central Atacama, estimates for MAR indicate a maximum of 110 mm/yr (a fivefold increase over modern) at ~11 ka, in good agreement with other records. A secondary maximum of 80 mm was reached during the Medieval Climate Anomaly. Our youngest midden in this series (0.45 ka) has MAR values similar to modern (~20 mm/yr). In contrast, estimated MAR anomalies from the northern Atacama are either negative or similar to modern between 600-150 cal yrs BP. The record exhibits a major pluvial between 145-80 cal yrs BP with MAR reaching values twice over modern estimates. Again, recent middens exhibit very similar values to modern. We consider our MAR estimates robust because a) the most recent middens estimate modern values to within 20 mm/yr, and b) the high degree of overlap for estimated MAR between middens of similar age. By taking into account the necessary caveats, this method could be applied reliably to estimate past climate change throughout the arid regions of southern South America where chinchilla rats occur.