RESUMEN
The forest-steppe ecotone in southern Siberia is highly sensitive to climate change; global warming is expected to push the ecotone northwards, at the same time resulting in degradation of the underlying permafrost. To gain a deeper understanding of long-term forest-steppe carbon dynamics, we use a highly resolved, multiproxy, palaeolimnological approach, based on sediment records from Lake Baikal. We reconstruct proxies that are relevant to understanding carbon dynamics including carbon mass accumulation rates (CMAR; g C m-2 yr-1 ) and isotope composition of organic matter (δ13 CTOC ). Forest-steppe dynamics were reconstructed using pollen, and diatom records provided measures of primary production from near- and off-shore communities. We used a generalized additive model (GAM) to identify significant change points in temporal series, and by applying generalized linear least-squares regression modelling to components of the multiproxy data, we address (1) What factors influence carbon dynamics during early Holocene warming and late Holocene cooling? (2) How did carbon dynamics respond to abrupt sub-Milankovitch scale events? and (3) What is the Holocene carbon storage budget for Lake Baikal. CMAR values range between 2.8 and 12.5 g C m-2 yr-1 . Peak burial rates (and greatest variability) occurred during the early Holocene, associated with melting permafrost and retreating glaciers, while lowest burial rates occurred during the neoglacial. Significant shifts in carbon dynamics at 10.3, 4.1 and 2.8 kyr bp provide compelling evidence for the sensitivity of the region to sub-Milankovitch drivers of climate change. We estimate that 1.03 Pg C was buried in Lake Baikal sediments during the Holocene, almost one-quarter of which was buried during the early Holocene alone. Combined, our results highlight the importance of understanding the close linkages between carbon cycling and hydrological processes, not just temperatures, in southern Siberian environments.