A re-assessment of elemental proxies for paleoredox analysis

Paleoredox conditions are commonly evaluated based on elemental proxies but, despite their frequency of use, most of these proxies have received little comparative evaluation or assessment of their range of applicability to paleomarine systems. Here, we evaluate 21 elemental proxies, including six proxies based on the C-S-Fe-P system (TOC, S, TOC/S, DOPT, Fe/Al, C-org/P), nine proxies based on trace-metal enrichment factors (Co-EF, Cr-EF, Cu-EF, Mo-EF, Ni-EF, Pb-EF, U-EF, V-EF, Zn-EF), and six additional proxies from Jones and Manning (1994) (U/Th, U-auth, V/Cr, Ni/Co, Ni/V, (Cu+Mo)/Zn), in 55 Phanerozoic marine formations. We used principal components analysis (PCA) to determine relationships among these 21 proxies in each formation and then sought to identify patterns across the full database. The first principal component (PC1) accounted for 40.1% of total dataset variance on average, with the highest median loadings on trace-metal enrichment factors (Ni-EF 0.82, Mo-EF 0.76, all nine >0.50). The next highest median loadings are on C-S-Fe-P proxies (TOC 0.58, DOPT 0.30, C-org/P 0.28), with bimetal proxies yielding uniformly lower loadings (Ni/Co 0.18, V/Cr 0.13). PCA of the factor loadings for the 55 study formations demonstrated associations among the 21 elemental proxies linked to specific sediment host phases: (1) an organic cluster associated with TOC, Mo, V, and Zn, (2) a uranium cluster associated with all U-based proxies, and (3) a sulfide cluster associated with S and Fe as well as the trace metals Co, Cu, Ni, and Pb (i.e., the major and typical minor constituents of diagenetic pyrite). The findings of the present study have important ramifications for use of elemental proxies for paleoredox analysis. First, all of the proxies examined here are influenced by environmental redox conditions to some degree, although the degree of redox influence on any given proxy can vary considerably from one formation to the next. Second, sedimentary enrichment of most proxies depends on the presence of specific mineral and organic host phases, and evaluation of elemental redox proxy data requires an understanding of how elements are partitioned among those phases. Third, no single proxy is a universally reliable redox indicator, although some are more consistently useful than others-notably, TOC and trace-metal EFs. Fourth, because of this inherent variability in proxy response, adoption of redox proxy thresholds established in earlier published studies is discouraged. Instead, we recommend that future redox studies establish redox thresholds on a formation-specific basis through internal cross-calibration of multiple redox proxies.