Setting the Record Straight: The Importance of Sample Preservation in Proxy Applications

Good proxies are useless without good samples. More than anything, chemical preservation of samples is the single factor most responsible for inaccurate results and incorrect interpretations. This is especially true for studies of Mesozoic and Paleozoic sediments, but can apply to Cenozoic studies as well. It is impossible to prove a specimen preserves 100% of its original chemical and isotopic composition, but a case can be made for the reliability of proxy data based on (1) preservation of original texture, microtexture, color (e.g., conodonts) and mineralogy (e.g., aragonite), (2) crystal clarity (absence of inclusions), (3) trace element compositions similar to those of modern specimens (e.g., enriched B, Sr, and SO4}2- and little or no Mn and Fe in carbonates), and (4) preservation of chemical or isotopic differences between coeval taxa. For Paleozoic fossils, thin-sectioning specimens and targeting nonluminescent (NL) shell under the cathodoluminescent stage has proved more effective in eliminating diagenetically-altered carbonate than relying on just grain clarity and trace element analyses. Rarely, modern specimens show cathodoluminescence, but these examples can be distinguished from the cathodoluminescence of altered ancient shells. Specimens may show differing tendencies to luminescence with alteration depending on the availability of Mn in diagenetic fluid. Thus, diagenetic components should always be examined for cathodoluminence to ensure that Mn-bearing fluids were present. When fossils are not available for isotopic studies, researchers rely on fine-grained carbonate sediments. Diagenesis largely resets δ18O because of high water:rock conditions, but the low water:rock ratios for carbon generally permit the features of the carbon isotope record to remain. Such δ13C records become suspect when carbonate contents are low or when exposure surfaces are present. For example, a negative δ13C spike at the Mid-Carboniferous boundary is not seen in the δ13C of NL brachiopods. Redundant proxies provide additional tests for fidelity, especially when diagenesis would alter proxies in a sense opposite of that of the environment. For example, Sr/Ca increases and δ18O decreases with temperature in marine mollusk shells, whereas diagenesis in a meteoric environment will cause both measures to decrease. Similar trace-element/δ18O validation has proved successful for Carboniferous brachiopod shells and Cenozoic benthic foraminifera. No sample screening method is fool-proof. As we continue to develop exciting new proxies for Earth system history, we must in parallel develop better proxies for screening sample preservation.