TESTING THE COMMON CAUSE HYPOTHESIS BY COMPARING SAMPLING BIAS IN THE DEEP SEA AND LAND BASED ROCK RECORDS

The Common Cause hypothesis predicts that rock record bias and sampled diversity are correlated because both are independently being driven by a third factor, namely sea-level. We test the this hypothesis by comparing patterns of coccolithophorid diversity over the last 150 Ma that are captured by two very different geological records: that of the land and deep sea. Based on a new taxonomically standardized database comprising 19,232 land-based and 28,222 deep-sea species occurrences from the North Atlantic and bordering land areas we show that sampling curves of land based and deep-sea deposits yielding coccolithophorids follow very different trajectories. We find that sampled coccolithophorid species diversity curves from these two environments diverge markedly after the Lower Cretaceous and in both cases are strongly correlated to their own rock record quality. Furthermore, subsampling and modelling applied to each dataset independently suggest that the true deep sea and land-based diversity curves are very similar, remaining relatively uniform over time. We conclude that rock sampling bias is a significant confounding factor for biodiversity analyses rather than a harmless correlate.