New cosmogenic nuclide constraints on Late Glacial and Holocene glacier fluctuations in the sub-Antarctic Indian Ocean (Kerguelen Islands, 49°S)

Cosmogenic nuclide dating of glacial landforms on the Kerguelen Archipelago (49°S, 69°E) gives the opportunity to study multi-millennial glacier fluctuations within the sub-Antarctic sector of the Indian Ocean. We here dated such geomorphic features to provide time constraints over the last 17,000 years using in situ-produced 36Cl in three glacial valleys: Val Travers valley, Ampere Glacier valley and Arago Glacier valley. For the first time, a combination of in situ-produced 36Cl and 10Be dating and 26Al/10Be ratios analysis was performed in the quartz-bearing syenite boulders of the Arago Glacier site. In addition, a Bayesian approach was computed to obtain a better constraint on moraine dating. Glacial advances occurred during the Late Glacial at 16.0 ± 1.9 ka and at 12.9 ± 1.7 ka in Val Travers, and at 13.6 ± 1.8 ka in Arago Glacier valley, probably linked to the Heinrich Stadial 1 and/or Antarctic Cold Reversal events, respectively. This suggests that all glaciers at this latitude were broadly sensitive to the large-scale climatic signal of the Antarctic Cold Reversal. So far, no Early nor Mid-Holocene moraines have been found in the glacial valleys on Kerguelen, indicating that the glaciers had probably receded significantly during these periods. This is in agreement with previously determined 14C ages from peat bogs, which suggest extensive deglaciation during several millennia of the Holocene period. Samples from glacially-polished bedrock surfaces (ranging from ∼4.4 ka to ∼14 ka) at Ampere Glacier site also suggest that this valley was ice free for several millennia during the Holocene. Finally, glaciers seem to have re-advanced only during the Late Holocene, especially within the last millennium, at ∼1 ka, ∼ 430 yr and ∼300 yr. A comparison of this new dataset with the available 10Be ages from other southern mid-latitude regions during the Holocene allows the identification of three different glacier evolution patterns. We suspect that variations of Kergue