Global plate model choice impacts reconstructions of the latitudinal biodiversity gradient

Key results

1. 01

   ShownThe four Global Plate Models (GPMs) often fail to agree on which palaeolatitudinal bin holds peak genus richness. Agreement is lowest in the Palaeozoic and slightly higher in the Northern than the Southern Hemisphere.

   All four models agreed on the peak bin in ~19% of Palaeozoic, ~47% of Mesozoic and ~26% of Cenozoic stages. Across the Phanerozoic, agreement was ~49% in the Northern vs ~43% in the Southern Hemisphere (coverage-based estimates, quorum 0.4). · Quantifying the LBG; Palaeogeographic reconstructions

2. 02

   ShownAll four models recover the same long-term shift of peak diversity, from the Southern Hemisphere before the Permian to the Northern Hemisphere after it. Even so, model pairs often rank latitudinal bins by richness differently, and they differ least in the Mesozoic.

   Normalised mean pairwise rank-order difference was ~0.25 in the Mesozoic (no pair above ~0.78), ~0.45 in the Palaeozoic and ~0.27 in the Cenozoic; in the latter two, some pairs disagreed completely (score 1). Raw-richness gradients showed this less often than coverage-based estimates. · Quantifying the LBG; Palaeogeographic reconstructions

3. 03

   ShownIn a subset of stage-hemisphere combinations, the choice of model changes the type of gradient: one GPM gives a tropical peak and another a polar or temperate peak. These conflicts are more frequent in the Southern Hemisphere.

   Tropical vs polar peak conflicts occurred in 25 stage-hemisphere combinations (7 Northern, 18 Southern; e.g. Southern Hemisphere Givetian). Tropical vs temperate peak conflicts occurred in 68 combinations (33 Northern, 35 Southern). · Discussion

4. 04

   ShownDifferences in palaeolatitude between GPMs generally grow with the age of the reconstruction, but not steadily: the Permian has low differences, and the spread is wider before the Late Jurassic.

   Average median pairwise palaeolatitudinal difference per stage was 2.51º in the Cenozoic, 4.49º in the Mesozoic and 10.34º in the Palaeozoic. Some collections differ by more than 25º latitude, e.g. Cambrian (n = 847), Ordovician (n = 680) and Jurassic (n = 488). · Quantifying the LBG; Palaeogeographic reconstructions; Conclusion

5. 05

   ShownEach GPM can place a different set of localities, because of the plate ages it assumes, and this changes bin sample sizes. The authors attribute gradient differences in young intervals mainly to this, not to shifts in reconstructed position.

   Lutetian: of 601 localities, 597 (PALEOMAP), 592 (GOLONKA), 546 (MERDITH2021) and 496 (TorsvikCocks2017) were reconstructed. Anisian: 553 of 879 (~63%) with TorsvikCocks2017. None of 52 Tortonian Panama localities could be placed with GOLONKA or MERDITH2021. Usable bins differed by one bin in 47 and two in 17 of 97 stages. · Palaeogeographic reconstructions; Discussion