High-Dimensional Chemostratigraphy of Pelagic Clay in the Western North Pacific Ocean Revealed via an Unsupervised Clustering Approach

Pelagic clay constitutes massive and apparently uniform lithology that limits the stratigraphic correlation between neighboring sediment core samples. Recent studies on the pelagic clay in the western North Pacific Ocean demonstrated that the bulk chemical composition of sediments constitutes multielemental chemostratigraphy, deducing the correlation between visibly featureless pelagic clay layers across several cores. However, this heuristic approach utilized only a few elements. Therefore, this study employed multivariate statistical techniques, including k-means cluster analysis, to analyze the chemical composition data set of 1,646 samples x 41 elements of the western North Pacific pelagic clay. The pelagic clay was classified into 10 clusters systematically aligned from the seafloor to the depth in a specific order, constituting stratigraphic units that reflected the high-dimensional geochemical features of these 41 elements. This finding strongly supports the statistical robustness of the latent chemostratigraphy in the western North Pacific pelagic clay. Additionally, we performed Sr-Nd-Pb isotope analyses of the detrital silicate fractions of the centroid samples representing each cluster. The multi-isotopic features of the detrital fraction varied from a mixture of North American and Asian dusts to a predominance of Asian dust superimposed by volcanic inputs. This secular variation in the matrix components is attributable to the northwestward motion of the Pacific Plate and the time-varying influence of arc volcanism throughout sedimentary history. The proposed integrated approach of multivariate statistical and isotopic analyses effectively extracted the essential information hidden in the pelagic clay, which may postulate a new protocol for paleoceanographic reconstructions targeting the pelagic realm. Plain Language Summary The examination of deep-sea sediments presents beneficial insights regarding past environmental changes. Although pelagic clay cover the vast deep-sea floor, geological information cannot be easily extracted from them owing to their uniform appearance. This study applied a multivariate statistical technique called k-means cluster analysis to a large data set of the chemical composition of pelagic clay in the western North Pacific Ocean. The results revealed that visibly featureless clay can be classified into 10 clusters corresponding to a series of sediment layers. Accordingly, we selected representative samples from each cluster (layer) and performed Sr-Nd-Pb isotopic analyses of the aluminosilicate fractions. The isotopic data indicated that the dust source progressed from a mixture of North America and Asia to a predominantly Asian interior because of the northwestward motion of the Pacific Plate. Furthermore, arc volcanic materials were supplied over temporal variations. Thus, pelagic clay can be utilized as a past environmental archive using an integrated approach of statistical and isotopic analyses.