CLUMPED-ISOTOPE TEMPERATURES FROM GAR SCALES, WITH AN APPLICATION TO A TERRESTRIAL K-PG SECTION

Gar fish have scales made of biogenic apatite, with a high concentration of carbonate in the anion sites. Biogenic apatite is fairly resistant to diagenetic alteration, and its well-defined initial crystalline texture provides a useful reference for evaluating alteration. We have found that outer enamel layer of the gar scale, along with the following measurement protocol, give excellent results in terms of reproducibility and a credible calibration trend. First, we acid digest a sample to produce CO₂, which is then treated to remove water and sulfur. The gas is further cleaned through a gas chromatograph column using long transport times to ensure good separation of pure CO₂. The columns are then well baked between measurements.

Our calibration curve is based on modern biogenic apatite samples from sites with known mean annual temperatures, including gar scales from fish collected by us, and mammal, alligator, and shark data measured by Eagle et al. (2010) and Wacker et al. (2016). The calibration samples show a well-defined linear trend on a Δ₄₇ vs. T^-2 plot.

We applied this method to stratigraphically well-constrained gar scales collected from a K-Pg section (Hell Creek and Fort Union formations) in North Dakota, USA. Previous work at this site has shown a short-lived (300 ka) warm event that preceded the K-Pg. The temperature proxies used in these previous studies are based on leaf fossils, paleosols, and stable isotopes from mollusks. Marine benthic forams show a 4 to 5 °C warming during this event. Our gar scale Δ₄₇ temperatures show the same warming event, and indicate a terrestrial temperature excursion of 10 °C. A major advantage of this method is that slow and fairly steady growth of gar scales mean that the measured temperature is similar to long-term mean annual water temperature. In turn, shallow-water temperature is within ~1 °C of surface air temperature, which means that gar-scale Δ₄₇ temperature should be directly comparable to surface air temperatures, as predicted by GCM models.