Long-term cycles in the history of life: Periodic biodiversity in the Paleobiology Database

Time series analysis of fossil biodiversity of marine invertebrates in the Paleobiology Database (PBDB) shows a significant periodicity at approximately 63 My, in agreement with previous analyses based on the Sepkoski database. I discuss how this result did not appear in a previous analysis of the PBDB. The existence of the 63 My periodicity, despite very different treatment of systematic error in both PBDB and Sepkoski databases strongly argues for consideration of its reality in the fossil record. Cross-spectral analysis of the two datasets finds that a 62 My periodicity coincides in phase by 1.6 My, equivalent to better than the errors in either measurement. Consequently, the two data sets not only contain the same strong periodicity, but its peaks and valleys closely correspond in time. Two other spectral peaks appear in the PBDB analysis, but appear to be artifacts associated with detrending and with the increased interval length. Sampling-standardization procedures implemented by the PBDB collaboration suggest that the signal is not an artifact of sampling bias. Further work should focus on finding the cause of the 62 My periodicity.

💡 Research Summary

In this paper Adrian L. Melott revisits the long‑standing claim of a ~62 Myr periodicity in marine invertebrate biodiversity, originally identified in the Sepkoski compendium, by applying modern statistical techniques to the Paleobiology Database (PBDB). The PBDB was specifically constructed to reduce sampling bias through rigorous standardization and subsampling, addressing criticisms that earlier detections might be artifacts of uneven fossil record quality. Melott first detrended the PBDB time series using a cubic polynomial, which best captured the overall increase in genera and a mid‑Phanerozoic plateau, outperforming linear, quadratic, exponential, logarithmic, power‑law, and hyperbolic alternatives.

Two parallel approaches were taken to examine periodic signals. In the first, the detrended data were linearly interpolated to a uniform 1 Myr grid and analyzed with a conventional Fast Fourier Transform (FFT). In the second, the original unevenly spaced data were subjected to a Lomb‑Scargle periodogram, which avoids interpolation and is robust to irregular sampling. Both methods yielded virtually identical spectra, confirming that the interpolation step did not introduce spurious peaks.

The power spectra display three prominent peaks: a low‑frequency component near 157 Myr, a middle‑frequency component at 63 ± 0.7 Myr, and a higher‑frequency component around 46 Myr. Significance was assessed against a red‑noise (AR(1)) background, with the 63 Myr peak reaching p < 0.001, the 46 Myr peak also exceeding the 0.001 threshold, and the 157 Myr peak significant at p ≈ 0.01. Autocorrelation analysis of the detrended series reveals a damped oscillatory pattern with a ~150 Myr envelope, consistent with the low‑frequency peak, and a Hurst exponent of 0.92, indicating long‑range memory rather than white noise.

Melott notes that the 46 Myr and 157 Myr signals are sensitive to methodological choices. The interval lengths in the PBDB vary, showing a strong 39 Myr periodicity in sampling resolution. A beat between this 39 Myr sampling rhythm and the genuine 63 Myr biodiversity cycle could generate side‑bands near 46 Myr, explaining the ambiguous nature of that peak. Moreover, the 157 Myr component diminishes markedly when a simple linear detrending is used instead of the cubic, suggesting it may be an artifact of the chosen trend model.

To test whether the 63 Myr signal appears independently in both databases, Melott computed the cross‑spectrum of the PBDB and Sepkoski series. The real part of the cross‑spectrum peaks sharply at 61.7 ± 0.4 Myr, with a phase offset of only 0.16 rad (≈1.6 Myr), indicating that peaks and troughs line up almost perfectly between the two independent compilations. By contrast, the 157 Myr and 47 Myr cross‑spectral peaks show larger phase mismatches (1.34 rad and 0.68 rad respectively), weakening the case for their biological reality.

Overall, the analysis demonstrates that the ~62 Myr biodiversity cycle is robust to data set, sampling standardization, detrending method, and spectral technique. It accounts for roughly 20 % of the variance in the PBDB after trend removal, whereas the other identified periodicities appear to be methodological artifacts. The persistence of the 62 Myr signal across two independently curated fossil databases strongly argues that it reflects a genuine, recurring perturbation in Earth’s biosphere, possibly linked to astronomical or geophysical processes (e.g., galactic plane crossings, mantle plume cycles, or climate oscillations). Melott concludes that future work should focus on refining temporal resolution in the PBDB, exploring independent proxies, and developing mechanistic models capable of producing a ~62 Myr rhythm in marine biodiversity.