Reply to: Clarifying the marine limit in Northern Spitsbergen and its implications for coastal lake archives

replying to W. R. Farnsworth Communications Earth & Environment https://doi.org/10.1038/s43247-025-02640-3 (2025)

Dr. Farnsworth raises important questions about the elevation of the postglacial marine limit (ML) in northern Spitsbergen and its implications for interpreting lake sediment records. While we agree that raised marine features can suggest a higher ML than previously acknowledged, we find no sedimentological or geochemical evidence for marine conditions within the sediment sequences analysed in our study. We reaffirm our main conclusion, that enhanced snowfall during the Holocene Thermal Maximum likely enabled the persistence of the Åsgardfonna ice cap, as robust regardless of uncertainty surrounding the early Holocene sea-level history.

We thank Dr. Farnsworth for his comment, which highlights additional geomorphological evidence regarding the postglacial marine limit (ML) in northern Spitsbergen. We agree that raised beach ridges and regional sea-level data suggest the local ML likely exceeded the 46 m a.s.l. elevation of Lake Berglivatnet (our primary site) following deglaciation. However, the stratigraphic and geochemical evidence from our lake cores indicates that any marine inundation must have occurred before deposition of the sediments analyzed in our work. In Auer et al.¹ we reported that both basins contain continuous lacustrine sequences beginning ~14.5 cal. ka BP, with no discernible marine phase. Crucially, the lowermost unit of both cores (Unit 4) lacks marine sediments or biota, as reflected by the radiocarbon-dated plant macrofossils and their distinctly terrestrial δ¹³C signature (see Supplementary Table 1 in Auer et al.¹). Moreover, while Calcium (Ca) content - often used as an indicator for marine carbonates²—is high, it co-varies here with Titanium (Ti), a bulk indicator for minerogenic glacigenic input³. Taken together, these observations indicate that sea water did not flow into the Berglivatnet or Lakssjøen lakes during the period captured by our cores, even if the regional high stand exceeded the basin thresholds. In other words, if marine waters inundated the catchments upon deglaciation, this phase is not captured in our sediment archives and might thus have preceded Unit 4.

Dr. Farnsworth proposes that Unit 3 may reflect a marine isolation sequence, deposited during the transition from marine to lacustrine conditions. We disagree: rather than the gradual changes that mark this transition in other nearby (lower-lying) lakes⁴, our sedimentological analysis shows that Unit 3 is an event deposit. The unit is sharply bounded and marked by two abrupt shifts in sediment composition. Notably, the base of Unit 3 shows clear signs of erosion and rapid redeposition. High-resolution X-ray CT imagery reveals soft-sediment deformation structures within Unit 3, including intraclast breccias (rip-up clasts) of less dense mud encased in the denser matrix of underlying Unit 4, as well as steeply tilted laminae indicative of slumping. An erosional unconformity separates these disrupted deposits from the intact sediments below. Such features are illustrated in Supplementary Fig. 8¹, which shows contorted bedding and blocky breccia clasts in Unit 3 of both Berglivatnet and Lakssjøen. Taken together, these distinct depositional features are characteristic for seismically-triggered mass wasting events, as described by Sabatier et al.⁵. We would also like to stress that Unit 3 is dated to around 9.5 ka BP in both investigated basins. Given the ~30 m difference in their elevation, it seems implausible that marine isolation would occur at the same time. Furthermore, the lithological and geochemical fingerprint of Unit 3 supports a two-phase disturbance rather than gradual marine regression. The lower part of Unit 3 (subunit 3b) is relatively organic-rich (loss-on-ignition increases from ~2% to 6%) and shows sharply reduced clastic input (Ti and Ca counts drop) compared to the glacial unit below. In contrast, the overlying subunit 3a is denser and mineral-rich, with a geochemistry resembling the basal Unit 4 (e.g. Ca content approaching that of Unit 4). Principal Component Analysis (PCA) of our data visualizes this pattern: subunit 3b clusters with the younger lacustrine Unit 2, whereas subunit 3a aligns with the older glacial Unit 4. This suggests that Unit 3 contains a mixture of sediments derived from both above and below. We interpret this as further evidence that material from the lake’s already-deposited sediments and underlying glacial deposits was remobilized and redeposited together during a single event. Such reworking is inconsistent with a marine isolation sequence (which typically shows a gradual transition from marine fine-grained sediment, to brackish organic mud, to freshwater peat or gyttja without erosive truncation)⁶. Indeed, a classical isolation sequence would not exhibit an erosional base or intraclast brecciation. Instead, one would expect relatively undisturbed layering as marine conditions slowly gave way to lacustrine deposition. Our cores show no such gradual transition in Unit 3. Therefore, we maintain that Unit 3 represents a synchronous and abrupt mass-wasting event, rather than the gradual isolation of a fjord bay.

It is important to emphasize that Units 3 and 4 lie at the earliest part of the sediment record, outside the interval on which our Holocene paleoclimate interpretations are based. Our study focused on the glacier and hydroclimate history from the Early Holocene (~9.5 ka BP) to present, as recorded in the overlying Units 2 and 1. Thus, the differing interpretations of the basal units do not affect our main conclusions, primarily based on end-member modelling analysis (EMMA) of grain size distributions. In his comment, Dr. Farnsworth acknowledges that a higher marine limit would likely have little impact on our EMMA results. We concur: whether Unit 3 resulted from a seismic event (our interpretation) or from marine isolation, it does not change the post-9.5 ka history of glacier-climate dynamics we reconstruct. And our main conclusion, the survival of the Åsgardfonna ice cap through the Holocene Thermal Maximum, remains supported by our data.

Although the local sea-level history does not impact our findings and was not the focus of our glacier-climate-oriented study, we do agree with Dr. Farnsworth that the local sea-level history is of interest. We also concur with the concluding remark of their comment: that proper (rigorous) dating surveys should be employed. Such efforts should, in our view, incorporate diagnostic provenance indicators, such as stable carbon isotope (δ¹³C) signatures⁷, to better distinguish between marine and terrestrial organic material. In this context, we clarify that in our own work, we chose to report the 39 m a.s.l. sea-level index point in Flatøyrdalen based on a whale cranium found on a clearly defined wave-washed terrace. We considered this a conservative but robust choice, given that alternative material (e.g., shell fragments within a glacial diamict) may be more difficult to interpret due to potential reworking. We acknowledge that this rationale, though stated in internal discussions, was not fully detailed in Auer et al.¹, and we welcome the opportunity to clarify it here.

In summary, we affirm that our original interpretation of the postglacial marine limit around Berglivatnet and Lakssjøen is consistent with the absence of marine signatures in the cores. We conclude that the basal sediments (Units 4 and 3) were deposited in an ice-contact to proximal-proglacial lake setting, punctuated by a rapid mass-wasting (earthquake) event, rather than in a fjord. This clarification does not alter the Holocene interpretations of our records, including our main findings regarding glacier resilience to early Holocene warmth. Also, given the chronological uncertainties of the local sea-level reported in our paper¹, and reiterated above, we believe it is possible that the marine limit exceeded the altitude of (one of) our lakes, but was reached prior to the deposition of the sediments analysed here. Indeed, there is mounting (additional) local evidence for an early onset (~16 ka BP) of deglaciation that supports this notion^8,9,10,11.