Speleothems Reveal 500,000-Year History of Siberian Permafrost

Permafrost Thaw Predictions Permafrost contains twice as much carbon as the atmosphere which could have serious consequences if it were to be released by widespread thawing. Vaks et al. (p. 183, published online 21 February) present a 450,000 year-long record of speleothem growth at selected locations in Siberia, which traces changes in the extent of permafrost over that time period. The authors conclude that conditions only slightly warmer than those of today would cause widespread thawing of continuous permafrost as far north as 60°N. Siberian caves recorded the history of permafrost occurrence during the past 450,000 years. Soils in permafrost regions contain twice as much carbon as the atmosphere, and permafrost has an important influence on the natural and built environment at high northern latitudes. The response of permafrost to warming climate is uncertain and occurs on time scales longer than those assessed by direct observation. We dated periods of speleothem growth in a north-south transect of caves in Siberia to reconstruct the history of permafrost in past climate states. Speleothem growth is restricted to full interglacial conditions in all studied caves. In the northernmost cave (at 60°N), no growth has occurred since Marine Isotopic Stage (MIS) 11. Growth at that time indicates that global climates only slightly warmer than today are sufficient to thaw extensive regions of permafrost.

Soils in permafrost regions contain twice as much carbon as the atmosphere, and permafrost has an important influence on the natural and built environment at high northern latitudes. The response of permafrost to warming climate is uncertain and occurs on time scales longer than those assessed by direct observation. We dated periods of speleothem growth in a north-south transect of caves in Siberia to reconstruct the history of permafrost in past climate states. Speleothem growth is restricted to full interglacial conditions in all studied caves. In the northernmost cave (at 60°N), no growth has occurred since Marine Isotopic Stage (MIS) 11. Growth at that time indicates that global climates only slightly warmer than today are sufficient to thaw extensive regions of permafrost. P ermafrost regions (in which the ground is frozen throughout the year) cover 24% of the Northern Hemisphere land surface and hold~1700 Gt of organic carbon. When it thaws, permafrost releases CO 2 and CH 4 , turning a longterm carbon sink into a source and enhancing the greenhouse effect (1,2). Permafrost degradation also intensifies thermo-karst development, coastline erosion, and liquefaction of ground previously cemented by ice. The latter endangers infrastructure, including major Siberian oil and gas facilities (3). An ability to predict the extent of future permafrost degradation is desirable.
Assessing the response of permafrost to changing climate can be difficult. Instrumental records during the past 20 years indicate substantial warming and thawing of local permafrost conditions (4), but permafrost at regional scales is slow to respond to warming, and instrumental records are insufficient to capture the long-term behavior. To understand the long-term response of permafrost to climate change requires knowledge of past permafrost conditions. Dating of organic material (5) or ground ice (6) can indicate the age of existing permafrost but cannot reveal the longer-term history of permafrost.
Here, we used cave carbonates (speleothems) as a tool to date past permafrost and its relationship to global climate. Vadose speleothems (stalactites, stalagmites, and flowstones) form when meteoric waters (i.e., originating from atmospheric precipitation) seep through the vadose zone into caves. Cave temperatures usually approximate the local mean annual air temperature (MAAT) because of buffering by the surrounding rock (7). When cave temperatures drop below 0°C, waters freeze and speleothem growth ceases. Speleothems found in modern permafrost regions are therefore relicts from warmer periods before permafrost formed (8)(9)(10). Absence of water also prevents speleothem growth in arid settings, so speleothem growth episodes in modern deserts are proxies for past wet periods (11). Because speleothems can be robustly and precisely dated with U-Th techniques, they provide a detailed history of pe-riods when liquid water was available and when both permafrost and desert conditions were absent.
We reconstructed the history of Siberian permafrost (and the aridity of the Gobi Desert) during the past~500,000 years, by means of U-Th dating of speleothems in six caves along a north-south transect in northern Asia from Eastern Siberia at 60.2°N to the Gobi Desert at 42.5°N (Fig. 1). The northernmost cave, Lenskaya Ledyanaya, sits today on the boundary of continuous permafrost with MAAT substantially below 0°C (12). The permafrost type changes toward the southwest to discontinuous, sporadic, and then permafrost-free conditions (13) (Fig. 1). Annual precipitation in this Siberian region is 400 to 600 mm/year, falling mainly during summer. To the south, in the Gobi, MAAT ranges from +2°C to +8°C and little precipitation falls (80 to 200 mm/year) (14).
Speleothem thickness provides an indication of long-term liquid water availability along the transect. Only 8 cm of growth is seen in the northernmost cave, increasing to~70 cm in the www.sciencemag.org SCIENCE VOL 340 12 APRIL 2013 caves of southern Siberia and decreasing again to less than 30 cm in the Gobi. As expected, southern Siberia is more suitable for speleothem growth than the cold north or the dry south. All recovered speleothems show a texture of calcium carbonate layers alternating with growth hiatuses (see supplementary materials).
Thirty-six speleothems were collected from the caves, and 111 U-Th ages were determined ( Fig. 2A). In each speleothem, at least one sample was taken from the outermost layer and from each section of growth (i.e., between hiatuses) inward, until the limits of the U-Th chronology were reached (age~500,000 years) to assess all periods of growth. A full description of the samples and their subsampling and dating is given in the supplementary materials.
The youngest speleothem growth in the region of modern continuous permafrost (i.e., at 60°N) occurred during interglacial Marine Isotopic Stage 11 (MIS-11), contrasting with the center of the transect where speleothems grew during all interglacials (Fig. 2, A and B). Age ranges in southern Siberia also demonstrate that the duration of speleothem deposition in MIS-11 was longer than during subsequent interglacials. These observations indicate that permafrost thawing during MIS-11 was more extensive than at any other point during the last 450,000 years and extended northward of 60°N, significantly further north than the present limit of continuous permafrost. Some similar thawing may also have occurred at MIS-13 in this most northerly cave. The absence of any observed speleothem growth since MIS-11 in the northerly Lenskaya Ledyanaya cave, despite dating the outer edges of eight speleothems, suggests the permanent presence of permafrost at this latitude since the end of MIS-11. Speleothem growth in this cave occurred during early MIS-11, ruling out the possibility that the unusual length of MIS-11 caused the permafrost thawing.
MIS-11 was also characterized by wetter conditions in the Mongolian Gobi Desert, as shown by two ages from Shar-Khana Cave speleothems ( Fig. 2A), which contrast with the absence of growth during subsequent interglacials. The existence of a humid event in the Gobi during early MIS-11 is supported by mollusk assemblages from the Chinese Loess Plateau (15) and by the dominance of input into Lake Baikal via the Selenga River during MIS-11 (16).
The degradation of permafrost at 60°N during MIS-11 allows an assessment of the warming required globally to cause such extensive change in the permafrost boundary. There is compelling evidence that MIS-11 was the warmest of recent interglacials, including the presence of boreal forest on South Greenland at that time (17), the absence of ice-rafted debris in the North Atlantic (18), increased sea levels (19), and higher sea surface temperatures (SSTs) in the tropical Pacific (20)(21)(22). Mg/Ca reconstructions (21,22) indicate that the SST of the Pacific Warm Pool (PWP) reached >30°C in early MIS-11, compared to 29.5°C in MIS-5.5 and~28.5°C during the pre-industrial Late Holocene (Fig. 2D). This tropical heat was transported poleward (23), and there is evidence of unusual warmth in Siberia during MIS-11, evidenced by the high fraction of biogenic silica in the sediments of Lake Baikal (24) (Fig. 2C) and high spruce pollen content in Lake El'gygytgyn, suggesting local temperatures 4°to 5°C above the present (25). When PWP www.sciencemag.org SCIENCE VOL 340 12 APRIL 2013 temperatures reached 30°C, this appears to have caused more pronounced warming of northern continents and to have led to substantial northward migration of the permafrost boundary. Periods of Siberian speleothem growth since MIS-11 suggest a close link between greenhouse warming (hence global temperatures) and permafrost extent. After a brief hiatus in growth after MIS-11 (from 370,000 to 355,000 years ago), coinciding with a minimum in atmospheric CO 2 and in PWP SST during MIS-10 (Fig. 2, D and F), large thicknesses of speleothem grew in Southern Siberia during MIS-9 as greenhouse gases returned to higher values. Speleothems also grew actively during MIS-5.5 and the Holocene (>5 cm) when CO 2 levels were high. In contrast, growth during MIS-7, a period of lower CO 2 and cooler global conditions, is minimal (maximum 1.5 cm in any studied cave), and no growth is observed during MIS-5.4 to 5.1. Conditions during MIS-7 were at the very limit for growth in southern Siberia: Speleothems grew during MIS-7.3 and 7.1 in Okhotnichya Cave (52°N) but only during MIS-7.1 just to the north at Botovskaya Cave (55°N). No growth occurred during MIS-7.5 at either cave despite higher concentrations of CO 2 and CH 4 than later in MIS-7 (26,27) and high PWP SST (Fig. 2, D to F) (21,22). Lake Baikal biogenic silica (24) and the percentage of arboreal pollen in Lake El'gygytgyn sediments (28) are also lower during MIS-7.5 than during MIS-7.3 and 7.1. Lower local summer insolation during MIS-7.5 (Fig. 2G) (29) suggests a role for local insolation in overprinting a Siberian climate dominantly controlled by global greenhouse gas levels.
U-Th dating of Siberian speleothem growth during recent interglacials allows detailed comparison of permafrost history with other aspects of the global climate system (Fig. 3). During MIS-5.5, speleothems started growing between 128,700 and 127,300 years ago, and this growth ended between 119,200 and 118,100 years ago (as determined from Bayesian analysis of U-Th data using OxCal-4.1; see supplementary materials). The permafrost thawing initiated when insolation was close to its maximum and greenhouse gases had just reached maximum values. Holocene permafrost degradation at our sites lags maximum insolation and greenhouse gas concentrations slightly, and starts between 10,000 and 9800 years ago. This lag may be due to the time required for permafrost to thaw at the slightly lower insolation and CO 2 levels of the Holocene (relative to MIS-5.5).
Overall, dated periods of speleothem growth allow an assessment of the relationship between global temperature and permafrost extent. PWP SST was 0.5°to 1.0°C higher during MIS-5.5 and~1.5°C higher during early MIS-11 relative to the pre-industrial Late Holocene (Fig. 2D) (21,22). Using PWP SST as a surrogate for global temperature (21) suggests that an increase in global temperatures by 0.5°to 1.0°C will degrade only noncontinuous permafrost in southern Siberia, with the Gobi Desert remaining arid. Warming of~1 .5°C (i.e., as in MIS-11) may cause a substantial thaw of continuous permafrost as far north as 60°N and may create wetter conditions in the Gobi Desert. Such warming is therefore expected to markedly change the environment of continental Asia and can potentially lead to substantial release of carbon trapped in the permafrost into the atmosphere.