Statistical analyses of grain-size, geochemical and mineralogical data in core CM92-43, Central Adriatic basin

A statistical strategy feasible for paleoenviromental studies is described. We present some examples of multivariate methods (principal component, factor and cluster analyses) on chemical, sedimentological and mineralogical data, from a core (CM92-43) collected in the Mid Adriatic Deep (MAD). This core, raised in 250 m water depth, provides a continuous expanded record of the late-Quaternary sea level rise and highstand. Problems related to peculiar characteristiscs of the data set (the presence of unusual sporadic events and the constant sum) are discussed, and some solutions are presented. Specific patterns resulting from the multivariate statistics are compared to independent results, indicating good correspondance to specific climatic events recorded by other proxies (age boundaries, chemostratigraphy, foraminifera ecozones, palynology).