PhanDA stage temperatures give a weak link (rho 0.27, 77 stages) between temperature change and PBDB marine genus extinction, but their uncertainty does not exclude a strong link

Question: does an independent temperature history support the strong link between temperature change and marine genus extinction that Song et al. (2021) reported (Spearman rho 0.63 on 45 bins)? Approach: we used 841,812 Paleobiology Database occurrences of 11 marine invertebrate groups (downloaded 2026-10-24) in 80 PhanDA stage bins (Tremadocian to Quaternary). We calculated gap-filler and three-timer genus extinction rates, raw and with SQS. We compared them with stage-to-stage global mean surface temperature changes from the PhanDA posterior ensemble (Judd et al. 2024). Results: extinction had a weak positive rank correlation with the change across the top boundary of each stage (rho = 0.27, n = 77, AR(1)-surrogate p = 0.024). Without the Big Five stages, rho was 0.20 (p = 0.12). The rate of change per Myr had no association (rho = 0.02), and GMST level had a weak association (rho 0.11 to 0.22). On Song et al.'s own bins, their within-bin temperature excursions did not agree with PhanDA stage changes (rho = 0.09, n = 43). The PhanDA posterior SD of a stage-to-stage change (3.2 to 4.1 C) is as large as the spread of changes between stages. Simulations show that a true correlation of 0.63 would give an observed rho of only 0.23 to 0.31. Thus the observed value agrees with true correlations from about 0.15 to above 0.9. Conclusion and limitation: with PhanDA stage means, the link is positive but weak and not clear without mass extinctions. PhanDA uncertainty and stage averaging do not let this test confirm or reject the strong link. Stage means also can hide short events. The posterior error structure between adjacent stages is assumed.

mass extinction · Phanerozoic · paleoclimate · Paleobiology Database · PhanDA · gap-filler rates · temperature change · data assimilation uncertainty · marine invertebrates · replication