Major excursions in sulfur isotopes linked to permafrost change in Eurasia during the last 50,000 years

nature geoscience Brief Communication https://doi.org/10.1038/s41561-025-01760-x Major excursions in sulfur isotopes linked to permafrost change in Eurasia during the last 50,000 years Received: 6 February 2023 Accepted: 1 July 2025 Published online: xx xx xxxx Rhiannon E. Stevens 1 , Hazel Reade 1, Kerry L. Sayle 2, Jennifer A. Tripp 3, Delphine Frémondeau1, Adrian Lister4, Ian Barnes 4, Mietje Germonpré 5, Martin Street6, Julian B. Murton 7, Simon H. Bottrell Daniel H. James 1 & Thomas F. G. Higham 9,10 , 8 Check for updates We identify a major sulfur isotope excursion in Eurasian faunal bone collagen from the last 50,000 years, here termed the Late Pleniglacial Sulfur Excursion. Our analysis suggests this is linked to changing permafrost conditions, presenting the utility of faunal collagen δ34S as a proxy for permafrost dynamics, a critical component of the global carbon cycle. Our findings complicate the use of archaeological faunal sulfur isotopes for mobility and palaeodietary studies. Over the past two decades, sulfur isotope ratios (δ34S) in plant, animal and human tissues have been increasingly used to explore food provenance, present and past diets and human and animal mobility. Most studies use sulfur isotopes as a geolocator, leveraging the spatial variability observed in plant sulfur isotope values, reflecting those of bioavailable sulfur. This variability arises because soil sulfur is primarily derived from mineral weathering of parent bedrock, the δ34S of which varies by rock type1. Additionally, the atmosphere contributes sulfur to soils (via dry deposition, SO42− aerosols or wet deposition of SO42−), although pre-industrial atmospheric inputs contributed <10% of total soil S, excepting narrow zones of strong coastal seawater sulfate spray influence1,2. Recent studies suggest that waterlogged soil conditions may result in distinct bioavailable δ34S values3,4. Minimal fractionation is seen in organic-bound sulfur as it is passed along the food chain (Δ34S tissue-diet ≈ 0 ‰ (refs. 5,6)), so animal δ34S values closely reflect those of the bioavailable δ34S at the base of their food chain7,8. Thus, animal δ34S values have been used to determine origin or mobility/ migratory behaviours. Others use sulfur isotopes as a (palaeo)dietary indicator, as marine resources have high and relatively homogeneous δ34S values (about 20‰), whereas terrestrial δ34S tends to be lower and more variable8. Overall, animal and human δ34S values are commonly interpreted as reflecting one or more stable sources, uninfluenced by environmental change, whereas a few studies argue that archaeological δ34S values reflect locally variable hydrological dynamics9–11. Here we report results of 796 δ34S isotope and 691 accelerator mass spectrometry (AMS) radiocarbon analyses from Late Pleistocene and Holocene fauna from Eurasia (Figs. 1 and 2, Supplementary Discussion 1.1 and Supplementary Data 1). One hundred and five samples come from contexts previously AMS dated. Our results show a high-magnitude excursion in faunal δ34S isotope values between approximately 30 and 15 thousand years (kyr) before present (bp) in some regions of Eurasia (Fig. 2 and Supplementary Fig. 1). This period corresponds to the latter part of the last ice age across much of Marine Isotope Stage 2 (about 29–11.7 kyr bp), including the Last Glacial Maximum (LGM, about 26.5–19 kyr bp). This excursion, which we name the Late Pleniglacial Sulfur Excursion (LPSE), is particularly pronounced in regions where we have good temporal coverage within a discrete geographic area, such as in Britain and Belgium, and is also evident in other regions, such as central Europe north of the Alps (Fig. 3 and Supplementary Fig. 2). However, the temporal and spatial coverage UCL Institute of Archaeology, London, UK. 2Scottish Universities Environmental Research Centre, Rankine Avenue, East Kilbride, UK. 3Department of Chemistry, University of San Francisco, San Francisco, CA, USA. 4Department of Earth Sciences, Natural History Museum, London, UK. 5Royal Belgian Institute of Natural Sciences, Brussels, Belgium. 6MONREPOS Archaeological Research Center and Museum for Human Behavioural Evolution, Römisch-Germanisches Zentralmuseum, Leibniz-Research Institute for Archaeology, Neuwied, Germany. 7Department of Geography, University of Sussex, Brighton, UK. 8School of Earth and Environment, University of Leeds, Leeds, UK. 9Department of Evolutionary Anthropology, Faculty of Life Sciences, University of Vienna, Vienna, Austria. 10Human Evolution and Archaeological Sciences Forschungsverbund, University of Vienna, Vienna, Austria. e-mail: 1 Nature Geoscience Brief Communication https://doi.org/10.1038/s41561-025-01760-x Latitude (° N) 80 70 60 50 40 0 50 100 150 Longitude (° E) Fig. 1 | Geographical distribution of the faunal samples. Pale blue area indicates approximate zone of continuous permafrost at the LGM20. White area indicates LGM extent of ice sheets and glaciers40. Dark blue area indicates zone of present-day continuous and discontinuous permafrost distribution41. Pink squares: samples collected from regions where no permafrost was present during the last 50,000 years. Green circles: samples collected from areas that either had permafrost present or were under ice sheets/alpine glaciers at the LGM but where permafrost/ ice sheets/alpine glaciers are absent today. Yellow triangles: samples collected from regions in which permafrost has been present throughout the past 50,000 years. NGRIP δ18O (‰) prevent us from determining whether the LPSE is time transgressive across this region. The LPSE occurs across multiple species with differing dietary niches and mobility behaviours (Fig. 2). The LPSE magnitude is substantial (up to 35‰), more than double that typically considered to indicate location-based differences12. This suggests that underlying continental-scale processes substantially impacted the terrestrial sulfur cycle during the Late Pleistocene. Our Eurasian samples span the last 50 kyr (Fig. 1). The climate between about 50 and 28 kyr bp (Middle Pleniglacial) featured millennial-scale oscillations between cold stadial and mild interstadial states and fluctuating sea levels, superimposed on a long-term trend towards colder conditions and lower sea levels13,14. The Middle Pleniglacial environment across northwest and central Europe was wet and densely vegetated, with long periods of seasonal frost, and some discontinuous permafrost15,16. The onset of the Late Pleniglacial (about 28–14.7 kyr bp) saw the major expansion of European ice sheets17,18. Maximum ice-sheet extent occurred during the LGM, when sea levels were about 130 m lower than today19. Continuous permafrost (ground that remains ≤0 °C for at least two consecutive years) was widespread across northern Eurasia at this time20. Between about 25 and 22 kyr bp, increasing aridity in northern Europe induced widespread fluvio–aeolian deposition in river vall (...truncated)