HIGHER TAXA AS PROXIES FOR SPECIES DIVERSITY: A GLOBAL SPECIES DATABASE OF TRILOBITA

A comprehensive global species database of Trilobita includes 20, 262 records of which 18, 576 are valid species. The database was compiled directly from the primary literature, and includes classification to traditional genus, subfamily, family, suborder, and order level. Stratigraphic information includes primary data such as regional zone, stage, etc., and also a series of 37 global sampling bins from Lower Cambrian to Upper Permian. Geographic information was recorded as both modern geopolitical entities and tectonic elements.

The compilation permits the first direct assessment of the performance of higher taxa as proxies of sampled species diversity. Trilobites have been proffered as an example of potential bias in this proxy relationship due to different amounts of taxonomic "splitting" in different time intervals, particularly in Cambrian versus post-Cambrian taxonomy. If the average number of species per genus changes non-randomly through time, higher taxic patterns might depart from species patterns.

Global diversity was calculated in each of the 37 bins at family, genus, and species levels, including Lazarus taxa in each case. The diversity profiles were then compared by calculating the predicted percent diversity change and its direction from bin to bin, and comparing the result between taxonomic levels. Comparisons were carried out using nonparametric Wilcoxon signed-rank tests and Spearman's rank correlation.

A bias in trilobite taxonomic practice does exist: there are 3.77 valid species per Cambrian genus versus 4.73 in the Ordovician, 4.78 in the Silurian, and 4.82 in the Devonian. However, this has no evident effect on the proxy relationship. No significant differences exist between any of the profiles using the Wilcoxon signed-rank test. All paired profiles are also significantly positively correlated, although correlation is strong between genera and species (r=.961, p<0.0001) but rather weak between families and species (r=.491, p=0.0037). Hence, the use of higher taxa, particularly genera, as proxies for species diversity is validated. However, the results belie the frequently advanced claim that higher taxa have advantages over raw species data in muting the effects of rock record bias. Rather, higher taxa appear to be surprisingly direct proxies for sampled species diversity.