Figure 4. Comparison of the Holocene brGDGTs-based MAAT record with temperature reconstructions and model simulations. (a ) BrGDGTs-based MAAT from Huguangyan Maar Lake (this study); (b ) Global mean surface temperature (Temp12k; Kaufman et al., 2020); (c ) Proxy-based mean temperature of the Northern Hemisphere (90º–30º N; Marcott et al., 2013); (d ) Pollen-based mean annual temperature anomaly (MATA) for North America and Europe (Marsicek et al., 2018); (e ) Model-based simulation of the temperature of the Northern Hemisphere (90º–30 ºN; Liu et al., 2014); (f ) Seasonally-unadjusted SST record (SSTSN, 23.5º–23.5ºN; Bova et al., 2021); (g ) Tropical mean annual SST record (MASST, 23.5ºS–23.5ºN; Bova et al., 2021).
Spectral analysis of the brGDGTs-based temperature series since 8.0 ka BP revealed several statistically significant quasi-periodicities, from the multi-decadal to the centennial scale: 201-202, 111-112, 95-96, 92-93, 78-82, and 64-65 years. Several of these periodicities are close to those of significant solar cycles, such as the Gleissberg cycle (~70–100 yr) and the de Vries cycle (~200-210 yr), although the latter is less significant in the time series (see Supplementary Figure S2).
The record from Maar Lake Huguangyan is indicative of regional-scale temperature change. A comparison of the record with subcontinental and hemispheric temperature time series is shown in Figure 4. In terms of general trends, the brGDGTs-based MAAT record shows a decreasing trend similar to that of global mean surface temperature (Temp12k; Kaufman et al., 2020), and proxy-based mean temperature in the Northern Hemisphere (90º-30ºN; Marcott et al., 2013), but it is in conflict with model simulations of the temperature of the Northern Hemisphere (90º-30 ºN; Liu et al., 2014) and of tropical mean annual sea surface temperature (MASST, 23.5ºS-23.5ºN; Bova et al., 2021). Thus, the Holocene temperature evolution at Huguangyan Maar Lake contrasts with the results of model simulations. One of the reasons for this may be the existence of a model simulation bias in SST reconstructions and uncertainty in the weighting of different forcings.