Statistically coherent approach involving log-ratio transformation of geochemical data enabled tephra correlations of two late Pleistocene tephra from the eastern Adriatic shelf

Tephrostratigraphic correlations commonly rely on geochemical composition supported by additional constraints (e.g., multiple stratigraphically ordered tephra, geochronological-stratigraphical constraints, and isotopic determinations), which provide key clues to restrict the number of possible candidates and disambiguate the correlation of a specific tephra among compositionally similar volcanic sources/tephra. However, such additional data may not be available or acquirable, leaving the geochemical data as sole, but challenging viable approach. In this study, two geochronologically poorly constrained late Pleistocene tephra from the eastern Adriatic -from a sand profile on Mljet Island (M-53/2) and from a marine sediment core from Pirovac Bay (PROS 721)- were correlated to known eruptions using only geochemical data (major and trace elements of glass shards), which were treated using both log ratio transformed and raw data. After the statistical treatment of the geochemical data using bivariate plots, linear discrimination analyses and selbal algorithm, the tephra M-53/2 and PROS 721 were suitably correlated with the widespread tephra generated during the Campi Flegrei eruptions of Massereia del Monte (Y-3 marine tephra, 29.0 +/- 0.8 ka) and Neapolitan Yellow Tuff (14.5 +/- 0.4 ka), respectively. This study showed that the correlation was hardly tenable when using the raw data, as opposed to compositional approach, which yielded satisfactory results. As a consequence, the distribution of Massereia del Monte/Y-3 tephra extended far toward the northeast, while a better chronological model, for reconstructing the paleoenvironmental changes at the Pirovac Bay location and the Holocene sea-level dynamics, could be obtained.

Automated summary

Tephrostratigraphic correlations rely on geochemical composition and additional constraints, but in cases where additional data is not available, the statistical treatment of geochemical data provides a viable approach. This study successfully correlated two poorly constrained late Pleistocene tephra using only geochemical data, which resulted in new insights into the distribution and chronology of volcanic eruptions in the eastern Adriatic region. The use of compositional analysis proved to be more reliable than raw data.