Abstract Geochemical patterns in the environment are always the result of certain processes. Therefore, it is essential to decipher a process to properly evaluate the environmental role and potential of chemical elements/compounds. This allows the distinction between natural and anthropogenic influence on elemental concentrations. However, if the compositional nature of geochemical data is neglected, erroneous or misleading conclusions regarding the processes involved are probable. In this study the reconstruction of depositional environments and processes through the Holocene in two sediment cores obtained from submerged sinkholes located on the island of Mljet, Croatia, was performed by taking into account the compositional nature of geochemical, mineralogical and grainsize data. Problems involving compositional data are always multivariate; for example, the concentration of a single element does not carry any interpretative information, as only the ratios between elements do. This has led to the discovery of a large number of geochemical proxies based on elemental ratios, which describe certain environmental conditions and processes involved. Nevertheless, some proxies have been found to be restricted to only some specific environments, thus preventing them from being used in general; therefore, some kind of relation between different proxies is necessary to obtain final conclusions. However, when using simple elemental ratios, those correlations cannot be obtained due to the nature of compositional data. With a sequential binary partition of a compositional vector, orthonormal log ratio (olr) coordinates (proxies) can be constructed. When based on expert knowledge, those proxies fully acknowledge the geochemical properties of the chosen elements with one major difference - that the correlation between newly obtained variables is mathematically well grounded. As a result, the final conclusion is more accurate. In this research, geochemical proxies obtained as a representation in olr coordinates of the elements that are enriched compared to the local soil were used to perform principal component analyses. It helped to unravel the evolution of sedimentary environments. Mineralogical (XRD and heavy mineral data) and grain size analyses supported the conclusions obtained based solely on geochemical data. Furthermore, data analysis suggests that the proxies for redox conditions described in the literature should be used with caution, as their use is somewhat limited.

中文翻译：

---

重建涉及全新世海相淹没沿海岩溶盆地演化的地球化学过程的组合方法（克罗地亚姆列特岛）

摘要 环境中的地球化学格局总是某些过程的结果。因此，必须破译一个过程以正确评估化学元素/化合物的环境作用和潜力。这允许区分自然和人为对元素浓度的影响。但是，如果忽略地球化学数据的组成性质，则可能会得出关于所涉及过程的错误或误导性结论。在这项研究中，通过考虑地球化学、矿物学和粒度数据的组成性质，从位于克罗地亚姆列特岛的水下沉洞中获得的两个沉积岩芯中，通过全新世重建沉积环境和过程。涉及组成数据的问题总是多变量的；例如，单个元素的浓度不带有任何解释性信息，只有元素之间的比率才带有任何解释性信息。这导致发现了大量基于元素比率的地球化学代理，这些代理描述了某些环境条件和所涉及的过程。然而，一些代理被发现仅限于某些特定环境，从而阻止它们被普遍使用；因此，不同代理之间的某种关系是获得最终结论所必需的。然而，当使用简单的元素比率时，由于成分数据的性质，无法获得这些相关性。通过组合向量的连续二进制分区，可以构建正交对数比 (olr) 坐标（代理）。当基于专业知识时，这些代理完全承认所选元素的地球化学特性，但有一个主要区别——新获得的变量之间的相关性在数学上是有充分根据的。因此，最终的结论更为准确。在这项研究中，作为与当地土壤相比富集的元素的 olr 坐标表示形式获得的地球化学代理被用于执行主成分分析。它有助于解开沉积环境的演变。矿物学（XRD 和重矿物数据）和粒度分析支持仅基于地球化学数据得出的结论。此外，数据分析表明，应谨慎使用文献中描述的氧化还原条件代理，因为它们的使用有些有限。