Interpreting patterns of extinction and origination rates from the fossil record can be challenging. Fossil databases have coarse resolution and uneven spacing in time, rendering many conventional time-series analysis tools inapplicable. However, specialized spectral analysis methods developed for unevenly spaced data can provide insight into the mechanisms linking the biological processes of extinction and diversification.

Cross-correlation of extinction and origination statistics shows that originations significantly lag extinctions by roughly10 million years, and this result is robust to the deletion of major extinctions from the record. Yet extinction and origination records are systematically different; extinctions are not autocorrelated over short time scales, while origination statistics are significantly autocorrelated at time scales up to 10 million years. This suggests that the lag between the two results from an intrinsic property of the process generating originations.

My results show that although the long-wavelength variability of diversification rates equals or exceeds that of extinctions, diversification rates are markedly less variable than extinction rates at wavelengths shorter than roughly 25 Myr. This implies that there are intrinsic limits to how much rates of diversification can accelerate to offset bursts of extinction.

Background reading:

Kirchner, J.W., Evolutionary speed limits inferred from the fossil record, Nature, 415, 65-68, 2002.
Kirchner, J.W. and A. Weil, Delayed biological recovery from extinctions throughout the fossil record, Nature, 404, 177-180, 2000.
Kirchner, J.W. and A. Weil, Correlations in fossil extinctions and originations through geologic time, Proceedings of the Royal Society of London, B267, 1301-1309, 2000.
Kirchner, J.W. and A. Weil, No fractals in fossil extinction statistics, Nature, 395, 337-338, 1998.