Palaeo-productivity record from Norwegian Sea enables North Atlantic Oscillation (NAO) reconstruction for the last 8000 years

The North Atlantic Oscillation (NAO) is a dominant atmospheric driver of North Atlantic climate variability, influencing the strength and position of the westerlies and subsequently impacting surface ocean circulation, current speeds, and mixing in the region. A positive NAO phase results in warmer and wetter conditions in northwestern Europe, while a negative phase leads to colder and drier conditions. The NAO's influence extends across various timescales, from days to centuries, and its impact on primary productivity through air-sea fluxes and circulation patterns has been previously suggested. While several NAO reconstructions exist, utilizing various paleodata, there's a scarcity of reconstructions based on open-ocean sediment records, largely due to challenges in achieving sufficient temporal resolution and precise age control. This study addresses these challenges to reconstruct NAO variability using an open-ocean proxy record. Previous research has utilized variability in biogenic calcium carbonate (CaCO3) content in marine sediments to infer paleoceanographic changes, but the complexities of the relationships between sediment carbonate and sea-surface conditions necessitate further investigation. The sources of CaCO3 in deep-sea sediments are varied and its concentration can be impacted by factors such as the productivity of CaCO3-secreting organisms (coccolithophores, foraminifera, pteropods, calcareous dinoflagellates), dilution by terrigenous material, and chemical dissolution. This study aims to determine if ITRAX-derived Ca/Fe variability can serve as a proxy for changes in sea-surface primary productivity and how this proxy reflects oceanic and atmospheric variability in the southeastern Norwegian Sea. Two well-dated, high-resolution sediment cores from this region, spanning the last 8000 years, were analyzed.

Numerous studies have explored the North Atlantic Oscillation (NAO) and its impact on climate variability, using various proxy records and modeling approaches. Existing NAO reconstructions predominantly utilize terrestrial datasets such as tree rings or ice cores. However, these reconstructions often have limitations in terms of temporal resolution and spatial representativeness. Studies have highlighted the influence of NAO on the dynamics of phytoplankton productivity, suggesting a correlation between NAO phases and changes in primary production. Previous research also utilized biogenic calcium carbonate (CaCO3) in marine sediments for understanding paleoceanographic changes, although the complexity of the relationship between sediment carbonate and sea-surface conditions made generalization challenging. The use of CaCO3 as a proxy for productivity was acknowledged as being dependent on other factors such as the productivity of CaCO3 secreting organisms, dilution by terrigenous material, and chemical dissolution. This study aimed to determine if Ca/Fe ratios, derived from ITRAX X-ray fluorescence core scanning, could overcome the challenges of the previously used proxy-based reconstructions.

Two sediment cores, GS13-182-01CC (GS13) and P1-003, were collected from the southeastern Norwegian margin at water depths of 960 m and 850 m, respectively. The cores penetrate into the Storegga Slide debris and contain a record of the last 8000 years. Age models for both cores were established using Icelandic tephra, radiocarbon, and 210Pb dates. A Bayesian approach generated an ensemble of 19,000 equally likely age models, which accounts for age model uncertainties between 2 and 13 years for the period 1992-1850 AD, increasing to 17–190 years between 1850 AD and 8000 years BP. ITRAX X-ray fluorescence (XRF) core scanning provided high-resolution data on bulk elemental composition. Ca/Fe ratios were used as a proxy for CaCO3 variability, accounting for closed-sum constraints of compositional data. The study assessed the influence of grain size and terrigenous input on Ca/Fe variability. The primary sources of biogenic CaCO3 were investigated, with a focus on coccolithophores and foraminifera. Coccolithophore species abundance and their contribution to CaCO3 flux were reviewed. To validate the Ca/Fe proxy, cross-correlation analysis was performed against available instrumental observations of primary productivity from the Continuous Plankton Recorder (CPR) survey (since 1958) and dissolved O2 data from Ocean Weather Ship Mike (OWSM, integrated over 1-50 m water depth) for the period of overlapping proxy and instrumental data. The CPR data was used as the most direct measure of primary productivity, with total dinoflagellate and diatom counts serving as an approximation of primary productivity changes. The study noted that valid coccolithophore counts were only available between 1999 and 2016, but these showed a strong correlation with total phytoplankton count during the period of overlap. Cross-correlation analysis was conducted across various lead/lag times for different seasons. The Ca/Fe record was further compared to instrumentally measured seasonal NAO indices (1824-1992 AD), the Atlantic Multidecadal Oscillation (AMO), and the subpolar gyre index (SPG). For the long-term NAO reconstruction, a bandpass filter (10-500 a) was applied to the Ca/Fe record to remove lower-frequency components potentially related to changes in sediment supply. High-pass filtered (periods <100 a) data were then used to compare the reconstruction to existing NAO reconstructions.

The analysis of the Ca/Fe ratios in the sediment cores revealed distinct changes in Ca concentration, centered around 5000, 3000, 2000, and 0 a BP. The grain size variations were found to have a negligible effect on the Ca/Fe ratio, suggesting that the fluctuations in the Ca/Fe ratio mainly reflected changes in CaCO3 concentrations. The study found that foraminifera contributed insignificantly to the total CaCO3 in the cores. The authors propose that the variability in the Ca/Fe ratio primarily reflects the variability in coccolith deposition, specifically with Coccolithus pelagicus as the main contributor. The cross-correlation analysis demonstrated a statistically significant correlation (P < 0.05) between the Ca/Fe record and summer total phytoplankton concentrations from the CPR survey (standard area B4), with about 84% of all 19,000 age models showing a significant correlation, indicating that the Ca/Fe ratio serves as a robust proxy for changes in primary productivity. The highest cross-correlation was found with summer data, suggesting a lag of 0 or 1 year. This relationship was supported by the close resemblance between Ca/Fe trends and previously published chlorophyll concentration trends in the eastern North Atlantic. The Ca/Fe record also demonstrated a robust anti-correlation with the winter NAO index (1824-1992 AD), indicating a negative relationship between primary productivity and the NAO. Specifically, a significant anti-correlation was found with the winter NAO between 1950-1992 AD (median R = -0.48, best fit R = -0.72) and the correlation persisted but weakened, when the correlation was extended to 1824 AD. This aligns with the established mechanistic link between NAO and primary productivity, whereby increased winter NAO activity increases wind-driven mixing, deepens the mixed layer, and delays or reduces the spring bloom. The 8000-year Ca/Fe record reveals both higher-frequency (periods <500 a) and lower-frequency variability. The higher-frequency components were used to reconstruct NAO variability. This NAO index (NAO_Ca/Fe) showed a significant correlation with existing NAO reconstructions spanning the last 1000 years, with particularly strong agreement with the NAO_growth reconstruction over the last 300 years. The NAO_Ca/Fe record also agreed well with a previously published likelihood record for the NAO, indicating an agreement of approximately 77% after introduction of a 25-year lag. However, the study found that the variance in the Ca/Fe record could not be entirely explained by NAO variability, suggesting the influence of other climate regimes like the AMO and SPG. The long-term analysis of the NAO_Ca/Fe index did not show distinct changes in NAO amplitude during the Little Ice Age (LIA), the Medieval Warm Period (MWP), or the Holocene Thermal Maximum (HTM), suggesting that the NAO's variability was not directly coupled with the amplitudes of these climatic periods.

The findings of this study support the hypothesis that primary productivity in the southeastern Norwegian Sea serves as a robust proxy for NAO variability. The strong correlation between the Ca/Fe proxy and the instrumental NAO data validates the use of this proxy for reconstructing past NAO variations. The study's methodology successfully overcame limitations of previous reconstructions by utilizing high-resolution data and advanced age modelling techniques to account for uncertainties. The anti-correlation between NAO and primary productivity is consistent with existing mechanistic understanding of their relationship and has implications for understanding the coupled ocean-atmosphere system. The 8000-year NAO reconstruction provides a unique long-term perspective, allowing for analysis of NAO behavior across different climatic periods. The agreement between the high-frequency components of the NAO_Ca/Fe index and existing reconstructions enhances confidence in the reconstruction's reliability. However, it's important to acknowledge the potential influence of other climate modes on the Ca/Fe variability, necessitating further research to isolate the unique contribution of NAO. The long-term record enables analysis of NAO behavior across various climate periods and offers valuable insights into the dynamics of ocean-atmosphere interaction.

This study presents a novel sub-decadally resolved reconstruction of the North Atlantic Oscillation (NAO) for the last 8000 years using Ca/Fe ratios in sediment cores from the southeastern Norwegian Sea as a proxy for primary productivity. The robust correlation between the proxy and instrumental NAO data validates the reconstruction's accuracy. The findings show that the NAO’s amplitude remained relatively consistent across major climatic events such as the LIA, MWP, and HTM. Future research can further explore the relationship between the lower-frequency variations in the Ca/Fe record and other climate modes, potentially leading to a more comprehensive understanding of long-term climate dynamics.

The study acknowledges several limitations. Chronological uncertainties, particularly for periods before 1850 AD, could influence the accuracy of the reconstruction. The influence of other climate modes (AMO and SPG) on the Ca/Fe variability was noted, indicating that a portion of the signal may not be solely attributable to the NAO. The proxy mainly reflects summer productivity, limiting the comprehensive representation of NAO's seasonal variability. Finally, the spatial representativeness of the sediment core record might limit the extrapolation of findings to the entire North Atlantic region.