(Table 1) Age determination of sediment core SO136-003GC

We compile and compare data for the last 150,000 years from four deep-sea cores in the midlatitude zone of the Southern Hemisphere. We recalculate sea surface temperature estimates derived from foraminifera and compare these with estimates derived from alkenones and magnesium/calcium ratios in foraminiferal carbonate and with accompanying sedimentological and pollen records on a common absolute timescale. Using a stack of the highest-resolution records, we find that first-order climate change occurs in concert with changes in insolation in the Northern Hemisphere. Glacier extent and inferred vegetation changes in Australia and New Zealand vary in tandem with sea surface temperatures, signifying close links between oceanic and terrestrial temperature. In the Southern Ocean, rapid temperature change of the order of 6°C occurs within a few centuries and appears to have played an important role in midlatitude climate change. Sea surface temperature changes over longer periods closely match proxy temperature records from Antarctic ice cores. Warm events correlate with Antarctic events A1-A4 and appear to occur just before Dansgaard-Oeschger events 8, 12, 14, and 17 in Greenland.

(Table 1) Sea surface temperature reconstruction for eastern equatorial Pacific surface sediment samples

Significant uncertainties persist in the reconstruction of past sea surface temperatures in the eastern equatorial Pacific, especially regarding the amplitude of the glacial cooling and the details of the post-glacial warming. Here we present the first regional calibration of alkenone unsaturation in surface sediments versus mean annual sea surface temperatures (maSST). Based on 81 new and 48 previously published data points, it is shown that open ocean samples conform to established global regressions of Uk'37 versus maSST and that there is no systematic bias from seasonality in the production or export of alkenones, or from surface ocean nutrient concentrations or salinity. The flattening of the regression at the highest maSSTs is found to be statistically insignificant. For the near-coastal Peru upwelling zone between 11-15°S and 76-79°W, however, we corroborate earlier observations that Uk'37 SST estimates significantly over-estimate maSSTs at many sites. We posit that this is caused either by uncertainties in the determination of maSSTs in this highly dynamic environment, or by biasing of the alkenone paleothermometer toward El Niño events as postulated by Rein et al. (2005).

Water temperature reconstructions based on paired Uk'37 and Tex86 records, and paired surface and subsurface Mg/Ca records

This dataset contains the collection of available published paired Uk'37 and Tex86 records spanning multi-millennial to multi-million year time scales, as well as a collection of Mg/Ca-derived temperatures measured in parallel on surface and subsurface dwelling foraminifera, both used in the analyses of Ho and Laepple, Nature Geoscience 2016. As the signal-to-noise ratios of proxy-derived Holocene temperatures are relatively low, we selected records that contain at least the last deglaciation (oldest sample >18kyr BP).