Deciphering ∼45.000 years of Arctic Ocean lithostratigraphic variability through multivariate statistical analysis

Our understanding of past climate conditions in the Arctic Ocean has been hampered by poor age control caused in part by low sedimentation rates (< 1 cm/kyr), hiatuses during glacial intervals as well as the scarcity and poor preservation of calcareous nanno-and microfossils in the sediments. Although recent advances using variations in single element (e.g. Mn) content or physical sediment properties (e.g. bulk density, grain size, colour) of the recovered sediments have aided Arctic core-to-core correlations, unique depositional events and post-depositional changes can complicate stratigraphic interpretations based on individual or even multiple, physical or chemical parameters. Furthermore, clear correlations between cores using physical and chemical parameters are not always possible to establish. To tackle this issue, we developed an algorithm that combines clustering and multivariate ordination to test the interrelation of multiple input parameters (e.g. an array of individual XRF elemental contents), and subsequently identifies statistically significant stratigraphic units on centimetre to decimetre scales. Our preliminary results show that a distinct sedimentological pattern during the past 45,000 years characterizes cores from the region of the Morris Jesup Rise and the Greenland side of the Lomonosov Ridge. Stratigraphic patterns of the Siberian Side of the Lomonosov Ridge yield distinct differences, thus allowing for novel insights into sedimentary processes shaping the different regions within the Arctic Ocean. We also argue that our approach can compensate for some of the weakness of single element or proxy applications, and hence aid the construction of a robust stratigraphic framework for a wide geographical range of Arctic Ocean sediments.