Detecting the influence of secondary environmental gradients on chironomid-inferred paleotemperature reconstructions in northern North America

We examine the influence of multiple environmental factors on quantitative reconstructions of past climate that are based on conventional transfer function approaches using subfossil midge remains preserved in lake sediments. Chironomid assemblages from the uppermost sediments of 366 lakes spanning northern North America were compared to environmental parameters using direct gradient analysis. While temperature expectedly explained the largest amount of variation in the chironomid assemblages, dissolved organic carbon (DOC) and total Kjeldahl nitrogen (TKN) accounted for significant amounts of the remaining variance. By constraining chironomid assemblages to mean-July air temperature (JulyT), we found a cluster of 70 lakes that were orthogonal to the primary temperature gradient in ordination. These lakes tended to be cold with higher DOC and TKN concentrations, yet had chironomid assemblages that were similar to the assemblages found in less productive, warmer lakes. In order to examine how secondary gradients affected paleotemperature reconstructions, three chironomid-based transfer function models were generated: a full dataset model (all 366 lakes), a reduced temperature-constrained model (296 lakes), and a reduced secondary-constrained model (70 lakes). Application of these models to four previously published Holocene paleotemperature reconstructions were used to distinguish the influence of temperature versus secondary-gradient effects. While unproductive Arctic lake paleotemperature reconstructions were found to be robust, reconstructions for two lakes in the boreal-tundra ecotone suggest secondary gradients influenced parts of their records. We recommend that passive core trajectory analysis upon our expansive surface-sediment calibration set provides a means to assess the veracity of paleotemperature reconstructions and potential influence of secondary gradients.