COMPARING EXTINCTION CONFIDENCE INTERVALS FROM INDEPENDENT K-PG FOSSIL DATA SETS FROM SEYMOUR ISLAND, ANTARCTICA

Seymour Island (Antarctic Peninsula) is an excellent location for testing the causes and effects of the K-Pg mass extinction due to continuous deposition and likely unchanging depositional environment across the K-Pg boundary and the exceptional quantity and quality of fossil preservation. In 2012, Tobin et al. reported evidence that there were two separate extinction events recorded on Seymour Island, one simultaneous with K-Pg boundary, and another roughly 150 ka earlier, associated an increase in temperature and contemporaneous with Deccan Traps eruptions. Their interpretation was that the Deccan Traps may have been a significant contributing factor to the overall K-Pg extinction, at least at high latitudes. Since then, Witts et al. (2016) interpreted a composite of several fossil datasets from Seymour Island as indicative of a single extinction at the K-Pg boundary, with no evidence for an earlier extinction. While Witts et al. evaluated confidence intervals on individual species ranges, they did not include a statistical evaluation of the multi-species extinctions.

Here I modify the approach of Wang & Marshall (2004) to evaluate multi-species extinctions in data sets of fossils from Seymour Island. In this method, a stratigraphic multi-species extinction confidence interval is assessed by including progressively more species until all species present are included in the analysis. Intervals of multi-species extinction are indicated by significantly narrow confidence intervals, in comparison with simulated extinction-free fossil datasets of identical size. This method was also tested using random data sets that were generated including zero, one, and two extinctions and randomized fossil recovery potential. It was successful in identifying single and double extinctions without prior knowledge of their location, while very rarely generating false positives when no extinctions were simulated. I analyze several independent fossil data sets from Seymour Island using the same techniques, including those of Witts et al. (2016), Stilwell et al. (2004), Zinsmeister et al. (1989), Huber (1988), Harwood (1988), and Macellari (1986). I find equivocal results: all show a clear extinction that is synchronous with the K-Pg boundary, while several, but not all, show evidence for separate, earlier extinction.