The fine-grained sediments of the Cariaco Basin, Venezuela, of the last 130 ky, whose deposition history is well characterized, were analyzed geochemically in order to test the validity of sediment bulk geochemistry as an indicator of detrital provenance. Several binary and ternary diagrams as well as the chemical index of alteration (CIA) were tested for their capacity to discriminate the poorly contrasted detrital sources to the Cariaco Basin, and to describe the temporal evolution of the contributions of these different sources. Most of the diagrams tested did not allow a good discrimination of sources or, when sources were well discriminated, did not allow an interpretation of the temporal variations consistent with the known history. A relatively good discrimination of sources and a consistent interpretation of temporal variations were however obtained using Hf vs. Th and La/Yb vs. Gd/Yb binary diagrams, as well as Ti-Zr-Th, Ti-Zr-La, and Lu-Hf-Th ternary diagrams. Compared to the previous studies of the detrital content of the Cariaco Basin sediments, the geochemical approach permitted the recognition of a sediment contribution eroded from the Unare platform and Gulf of Cariaco during rapid sea level oscillations, and the contribution of Saharan eolian particles during the Younger Dryas-Preboreal and MIS6-5 transition. The choice of plotted elements was determined after considering carrier minerals, so that different elements may be informative in different sedimentary contexts. Overall, mineral sorting during transport appears as a major limit to quantitative estimation of the different contributions. In particular mineral sorting leads to the selective enrichment of elements associated with clays (Al, Rb, Th and LREE) in sediments deposited in the basin. Unless the geochemical effect of mineral sorting can be measured, it appears that quantitative provenance analysis should be performed on fractions of similar grain size instead of bulk sediment.
