Temperature Controls on No-Analog Community Establishment in the Great Lakes Region

Abstract

Temperature reconstructions in eastern North America from pollen are rife in the literature, but these reconstructions cannot be used to study the sensitivity and response time of vegetation to temperature changes. Branched glycerol dialkyl glycerol tetraethers (brGDGT) have recently been applied to reconstruct temperature at Silver Lake, OH, where the resulting temperature estimates closely tracked a regional pollen temperature reconstruction for the southern Great Lakes region. Here, we present the second brGDGT temperature record in this region, at Bonnet Lake, OH, using a new method for brGDGT detection and four alternative calibration functions, and reanalyze sediments from Silver Lake. We compare results across brGDGT detection methods, between sites and among calibration functions, and assess calibration uncertainty using a Bayesian linear regression. Of the calibration functions, MAT MBT′5Me reproduced the regional pollen stack and existing temperature record from Silver Lake most closely but is ~2°C warmer on average than the existing brGDGT temperature record. In the Bayesian regression analyses, the 95% credible interval when using the calibration from Weijers et al. (2007) was +/- 12.3°C. The MAT MBT′5Me calibration from De Jonge et al. (2014) had a smaller uncertainty (+/- 9.8°C 95% credible interval), likely resulting from a reduced sensitivity of brGDGT methylation to pH. These uncertainties are based only on analyses of the calibration data and likely overestimate reconstruction uncertainty, because calibration soil samples are heterogenous at the sampling scale and are at locations distant from the corresponding temperature measurement. Uncertainty can be reduced by creating improved calibration datasets of modern lake sediments. Despite uncertainty in absolute temperature and range, climatic trends are closely similar among brGDGT reconstructions and provide insight into the temperature drivers of past vegetation dynamics. Picea decline begins shortly after the start of Bølling-Allerød warming. The establishment and disappearance of no-analog communities lag temperature change by 200-500 years at Bonnet Lake and Silver Lake. The lack of synchrony in the timing of warming and no-analog community establishment at Bonnet Lake and Silver Lake, but agreement in the sequence of events, indicates that local climate is an important control on vegetation assemblages. The extension of brGDGTs to the Great Lakes Region is further supported in this study, but uncertainty in temperature estimates emphasizes the need for improved calibration datasets.