400-million years of 'fishes': a survey of sampling biases based on the UK record

Fishes comprise a majority of living vertebrate species, with an extensive fossil record spanning most of the Phanerozoic. Surprisingly few analyses have explored patterns of taxonomic diversity in fossil fishes since these were first investigated by Agassiz more than a century and a half ago. Paleobiodiversity trajectories in fossil fishes are thought to be disproportionately driven by exceptional deposits (“Lagerstätten effects”), but the degree to which these patterns are influenced by the rock record more generally remains unquantified. Today, as in Agassiz’s time, perspectives on fish evolution are strongly influenced by the well-studied UK record, which yields a series of important faunas ranging from Silurian (Llandovery) to latest Eocene (Priabonian) in age. We assembled a database spanning this interval, comprising 1678 species occurrences from 100 localities spanning 70 geologic formations and representing 443 genus-level lineages encompassing a diverse taxonomic assemblage spanning ‘agnathans’ to fin-bearing members of the tetrapod stem lineage. Some patterns revealed by our UK survey match features in fish data derived from Sepkoski’s genus-level and Benton’s (Fossil Record 2) family-level global compendia, most strikingly an increase in taxonomic diversity approaching the Recent. Unlike those analyses, we sampled our data in-bin and excluded extant taxa, suggesting the observed rise is not attributable to a “Pull of the Recent”. Furthermore, we show that when accounting for potential sampling bias - through rarefaction, shareholder quorum subsampling or a modelling approach using either the number of fish-bearing formations or localities as a rock record proxy - this signal remains, with a prominent Paleogene peak. Breaking out the two largest extant clades - Chondrichthyes and Osteichthyes - shows that these groups have similar post-Devonian patterns, both to each other and to the cumulative ‘fish’ curve. However, we do show the bony fishes make more statistically significant excursions from a model that assumes true diversity is constant and observed diversity is driven purely by sampling. Overall, although correlations between taxonomic diversity and our rock record proxies are significant, fish diversity appears to be less biased by sampling than many terrestrial vertebrate groups, such as dinosaurs.