Fossil-based analyses of clades’ diversification patterns require taxonomic expertise and appropriate methodology

Summary Estimating deep-time diversification patterns and the establishment of extant biodiversity represent major challenges in macroevolution. Fossil record data provide essential information to address these topics, but their heterogeneous temporal and geographical distributions require using analytical approaches to process these data. Gardiner et al. 1 (hereafter GEA) used a deep-learning model 2 and a fossil-occurrences dataset 3 to estimate neoselachian richness over the last 145 myr. Results and Discussion GEA 1 found that neoselachian diversity increased throughout the Cretaceous, was little impacted by the Cretaceous-Paleogene (K/Pg) mass extinction (~10% species loss), and peaked in the mid-Eocene but declined until the Present. While the Cretaceous increase in neoselachian richness is well known 4 , the other findings of GEA 1 are at odds with current knowledge. With the exception of lamniform sharks, the perceived decrease in species richness in the recent past is most likely due to a drop in available fossil record data combined with difficulties in identifying extant species in the fossil record 5 . Similarly, all previous analyses of the impact of the K/Pg mass extinction on elasmobranch diversification have reported high extinction rates, a marked diversity drop, and delayed recovery 6–7 , despite heterogeneity across clades, ecology, and geographical distribution 7 . Taking the K/Pg as an example, we demonstrate that the discrepancies between GEA 1 ’s results and current consensus is most likely due to a combination of incomplete, unverified, and incorrect fossil-occurrence data with inappropriate methodology.