TESTING THE QUALITY OF THE FOSSIL RECORD THROUGH GEOLOGICAL TIME

Tracking large-scale changes in biodiversity through geological time is an important aspect of paleobiological research and it is currently debated whether patterns observed in the fossil record can be trusted at face value. Many studies have suggested that the fossil record is severely biased by the amount of sedimentary rock preserved, and thus available for sampling, per geological time bin. However, most studies are carried out at global or continental scales using untested, vague and imprecise sampling proxies. This raises questions about the validity of interpretations on the correlations between various supposed sampling proxies and diversity estimates. Using novel GIS and remote sensing techniques, I have shown that commonly used sampling proxies (i.e. outcrop area, number of formations etc.) do not correlate well with one another, or with other metrics representing rock accessibility and worker effort. Studies carried out at local, regional and national scales in the British Triassic and Jurassic show that correlations between sampling proxies and paleodiversity are not consistent and are heavily influenced by the facies and lithologies preserved at certain time periods. The precision of sampling proxies improves at finer temporal and geographical scales, questioning the accuracy of global Phanerozoic paleodiversity studies that undoubtedly use vague and imprecise sampling proxies. Analysis also shows that formation counts obtained directly from the Paleobiology Database (i.e. the most common sampling proxy used in global Phanerozoic diversity studies) are flawed by the common occurrence of missing data, synonymous entries, and spelling mistakes. These results suggests that, although some of the diversity signal in the fossil record can be attributed to sedimentary rock bias, the complete picture is much more complicated with the influence of changing paleoenviroments, both ecological and preservational, appearing to be stronger. It is therefore premature to assume we can detect, or correct for, bias in the fossil record using simple proxies for rock availability.