Greenland Ice Sheet mass loss during the Last Interglacial and middle Holocene despite paleolimnological evidence for increased atmospheric moisture

Melt and snowfall over the Greenland Ice Sheet are both expected to increase in the coming century as air temperatures rise. How these competing levers impact the long-term surface mass balance of the ice sheet is an important question, relevant to predicting future rates of global sea level rise. Here we present reconstructions of the isotopes of lake water and isotopes of terrestrial leaf water at a lake in northwest Greenland, which preserves sediments from both the Holocene and the Last Interglacial. We use the oxygen isotopic composition (δ¹⁸O) of chironomid larvae chitin, hydrogen isotopic composition (δ²H) of sedimentary leaf wax (nC₂₉ alkanes), and chironomid species assemblages to reconstruct precipitation isotopes, leaf water isotopes and summer air temperatures respectively. By comparing parallel reconstructions of precipitation and leaf water, we estimate the time-dependent transpirative enrichment of leaf water ²H/H ratios. We use this result to show that atmospheric moisture decreased substantially during cool late Holocene summers, relative to the warm early Holocene (+4-7°C warmer than modern) and peak temperatures of the Last Interglacial (+5.5-8.5°C warmer than modern)(McFarlin et al., 2018 PNAS). This suggests that future warming will be accompanied by an increase in relative humidity and precipitable moisture over northwest Greenland, which is consistent with current model projections. However, by comparing our data to published records of Greenland Ice Sheet fluctuations, we demonstrate that ice expansion occurred during the coldest and regionally driest conditions of the Holocene. In contrast, the GrIS strongly retracted during the warm and regionally wet conditions of the middle Holocene and Last Interglacial. This indicates that temperature strongly controlled regional ice mass balance, outpacing any effects of changing precipitable moisture. Our results therefore support models indicating that warming-driven ice loss in this region will be at most partially mitigated but not substantially slowed by increased accumulation.