Although belemnites form a major clade of extinct cephalopods, the early stage of their diversification remains poorly known in time and space. Here we investigate the first diversification episodes of belemnites (order Belemnitida) using a new species-level database encompassing the Hettangian-Aalenian interval (Early Jurassic-earliest Middle Jurassic) and covering the Western Tethys. Rarefied richness shows a four-fold increase from the Planorbis chronozone to the Ibex chronozone, a strong decrease between the Margaritatus and Spinatum chronozones, followed by a drop in the Spinatum-Serpentinum interval that is coeval with the second-order Toarcian biological crisis. The Bifrons chronozone records a high richness that departs significantly from those of the surrounding chronozones. A last richness peak in the Dispansum chronozone precedes a decrease towards the Aalenian. Biogeographic analyses do not reveal any clear large-scale provincialism for belemnites, in sharp contrast with ammonoids. Such a long-term homogeneous spatial distribution of belemnites is probably due to: (1) the relatively poorly-documented fossil record of belemnites, especially in Mediterranean localities and (2) contrasted dispersal abilities of belemnites compared to ammonoids over the studied time interval.

The database of the present work is constructed upon a near-exhaustive, species-level taxonomic revision of the available literature, covering 150 years of palaeontological investigations (1867-2018): 64 publications (Appendix S1) were selected according to the criteria described hereafter and analysed. Some publications are monographs dealing with a large palaeogeographic area, whereas others are more geographically restricted. Only rarely were data from the authors’ own samples or from museum collections added (Appendix S2). Because published data are far from being homogeneous, we constructed a dataset respecting several strict criteria. Data from the literature have been evaluated only if they contained specimen illustrations (drawings and/or photographs) in order to check the original species attribution. Occurrences that were simply cited were ignored. Once this first criterion was validated, geographical and stratigraphic information were checked. In cases where one of these two types of data was not adequately reported for a species, the latter was rejected from the dataset. In terms of temporal resolution, we retained only occurrence data available at the resolution of the ammonite chronozone. Finally, a consistent taxonomic evaluation has been made by the same experts (RW & PN) over the entire dataset in order to check for synonymy and for a qualitative appreciation of intra-specific variability. The result is the taxonomic appendix given in full (Appendix S3).

Species richness was evaluated using raw data but also sample-based and coverage-based rarefaction and extrapolation (Appendix S4).

Biogeographic contrasts among belemnite assemblages were tested statistically using ANOSIM. This non-parametric technique tests for significant differences in similarity between two or more assemblages based on a distance measure, here the Raup-Crick coefficient. All these biogeographical analyses were conducted at the scale of the chronozone, sub-stage and stage with the software PAST 3.15. The results are given in Appendix S5.