Holocene climatic change and the development of the lake-effect snowbelt in Michigan, USA

Lake-effect snow is an important constraint on ecological and socio-economic systems near the North American Great Lakes. Little is known about the Holocene history of lake-effect snowbelts, and it is difficult to decipher how lake-effect snowfall abundance affected ecosystem development. We conducted oxygen-isotope analysis of calcite in lake-sediment cores from northern Lower Michigan to infer Holocene climatic variation and assess snowbelt development. The two lakes experience the same synoptic-scale climatic systems, but only one of them (Huffman Lake) receives a significant amount of lake-effect snow. A 177-cm difference in annual snowfall causes groundwater inflow at Huffman Lake to be 18O-depleted by 2.3‰ relative to O'Brien Lake. To assess when the lake-effect snowbelt became established, we compared calcite-δ18O profiles of the last 11,500 years from these two sites. The chronologies are based on accelerator-mass-spectrometry 14C ages of 11 and 17 terrestrial-plant samples from Huffman and O'Brien lakes, respectively. The values of δ18O are low at both sites from 11,500 to 9500 cal yr BP when the Laurentide Ice Sheet (LIS) exerted a dominant control over the regional climate and provided periodic pulses of meltwater to the Great Lakes basin. Carbonate δ18O increases by 2.6‰ at O'Brien Lake and by 1.4‰ at Huffman Lake between 9500 and 7000 cal yr BP, suggesting a regional decline in the proportion of runoff derived from winter precipitation. The Great Lakes snowbelt probably developed between 9500 and 5500 cal yr BP as inferred from the progressive 18O-depletion at Huffman Lake relative to O'Brien Lake, with the largest increase of lake-effect snow around 7000 cal yr BP. Lake-effect snow became possible at this time because of increasing contact between the Great Lakes and frigid arctic air. These changes resulted from enhanced westerly flow over the Great Lakes as the LIS collapsed, and from rapidly rising Great Lakes levels during the Nipissing Transgression. The δ18O difference between Huffman and O'Brien lakes declines after 5500 cal yr BP, probably because of a northward shift of the polar vortex that brought increasing winter precipitation to the entire region. However, δ18O remains depleted at Huffman Lake relative to O'Brien Lake because of the continued production of lake-effect snow