PP43F-02
Pollen reconstructions, tree-rings and early climate data from Minnesota, USA: a cautionary tale of bias and signal attentuation

Thursday, 17 December 2015: 13:55
2003 (Moscone West)
Jeannine-Marie St-Jacques1,2, Brian F Cumming3, John P. Smol3 and David Sauchyn2, (1)University of Regina, Regina, SK, Canada, (2)Prairie Adaptation Research Collaborative (PARC), Regina, SK, Canada, (3)Queen's University, Kingston, ON, Canada
Abstract:
High-resolution proxy reconstructions are essential to assess the rate and magnitude of anthropogenic global warming. High-resolution pollen records are being critically examined for the production of accurate climate reconstructions of the last millennium, often as extensions of tree-ring records. Past climate inference from a sedimentary pollen record depends upon the stationarity of the pollen-climate relationship. However, humans have directly altered vegetation, and hence modern pollen deposition is a product of landscape disturbance and climate, unlike in the past with its dominance of climate-derived processes. This could cause serious bias in pollen reconstructions.

In the US Midwest, direct human impacts have greatly altered the vegetation and pollen rain since Euro-American settlement in the mid-19th century. Using instrumental climate data from the early 1800s from Fort Snelling (Minnesota), we assessed the bias from the conventional method of inferring climate from pollen assemblages in comparison to a calibration set from pre-settlement pollen assemblages and the earliest instrumental climate data. The pre-settlement calibration set provides more accurate reconstructions of 19th century temperature than the modern set does. When both calibration sets are used to reconstruct temperatures since AD 1116 from a varve-dated pollen record from Lake Mina, Minnesota, the conventional method produces significant low-frequency (centennial-scale) signal attenuation and positive bias of 0.8-1.7 oC, resulting in an overestimation of Little Ice Age temperature and an underestimation of anthropogenic warming. We also compared the pollen-inferred moisture reconstruction to a four-century tree-ring-inferred moisture record from Minnesota and Dakotas, which shows that the tree-ring reconstruction is biased towards dry conditions and records wet periods relatively poorly, giving a false impression of regional aridity. The tree-ring chronology also suggests varve chronology problems.

It remains to be explored how widespread this landscape disturbance problem is when conventional pollen-based inference methods are used, and consequently how seriously regional manifestations of global warming might have been underestimated with traditional pollen-based techniques.