A continental-scale chironomid training set for reconstructing Arctic temperatures

Abstract

We present chironomid species assemblage data from 402 lakes across northern North America, Greenland, Iceland, and Svalbard to inform interpretations of Holocene subfossil chironomid assemblages used in paleolimnological reconstruction. This calibration-set was developed by re-identifying and taxonomically harmonizing chironomids in previously described surface sediment samples, with identifications made at finer taxonomic resolution than in original publications. The large geographic coverage of this dataset is intended to provide climatic analogues for a wide range of Holocene climates in the northwest North Atlantic region and North American Arctic, including Greenland. For many of these regions, modern calibration data are sparse despite keen interest in paleoclimate reconstructions from high latitudes. A suite of chironomid-based temperature models based upon this training set are evaluated here and the best statistical model is used to reconstruct late glacial (Allerød and Younger Dryas) and Holocene paleotemperatures at five non-glacial lakes representing a wide range of climate zones across Greenland. The new continent-scale training set offers more analogues for the majority of Greenland subfossil assemblages than existing smaller training sets, with many in Iceland and northern Canada. We find strong agreement between chironomid-based reconstructions derived from the new model and independent glacier-based evidence for multi-millennial Holocene temperature trends. Some of the new Holocene reconstructions are very similar to published data, but at a subset of sites and time periods we find improved paleotemperature reconstructions attributable both to the new model's finer taxonomic resolution and to its expanded geographic/climatic coverage, which resulted in improved characterization of species optima. In the late glacial, the new model's finer taxonomic resolution yields a unique ability to resolve temperatures of the Allerød from colder temperatures of the Younger Dryas, although the magnitude of that temperature difference may be underestimated. This study demonstrates the value of geographically and climatically broad paleoecological training sets. The large, taxonomically harmonized dataset presented here should be useful for a wide range of future investigations, including but not limited to paleotemperature reconstructions across the Arctic.