Methodological basis for quantitative reconstruction of air temperature and sunshine from pollen assemblages in Arctic Canada and Greenland
Résumé
This study presents a modern database including 831 pollen assemblages from lakes of the Boreal, Subarctic and Arctic biomes of North America and Greenland, and corresponding temperature, sunshine and precipitation. Pollen data include the 39 most common vascular taxa (18 woody plants and 21 herbs). Multivariate ordinations using correspondence analysis (CA) and canonical correspondence analysis (CCA) reveal that temperature and sunshine account, respectively, for 45.5% and 44.4% of the variance within pollen assemblages, whereas precipitation only accounts for 10.2% of total variance. CCA further demonstrates that the climatic information encapsulated in pollen assemblages is not the same for the three biomes. In the Boreal biome, precipitation and temperature account for most of the variance, whereas sunshine and precipitation are more determinant in the Subarctic biome, and the temperature and sunshine seem to exert the main control on the pollen distribution in the Arctic biome. The modern analogue technique (MAT) and CCA regressions were tested for reconstructing climate parameters. MAT yields better results than CCA regressions and validation tests indicate a root mean square error (RMSE) of 1.35 degrees C for July air temperature, 2.3% for July sunshine, 3.10 degrees C for January air temperature and 143 mm for annual precipitation. Two examples of reconstructions are presented from recent (<450 years) lake-sediment cores of eastern Baffin Island, Arctic Canada and southwest Greenland. They demonstrate that July air temperature and July sunshine can be reconstructed independently from Arctic pollen assemblages, but January air temperature and annual precipitation must be interpreted with caution. They also suggest that reconstruction of past sunshine variations can serve to document climate changes at synoptic scale including the North Atlantic Oscillation (NAO). (C) 2008 Elsevier Ltd. All rights reserved.