The hypothesis that destructive mass extinctions enable creative evolutionary radiations (creative destruction) is central to classic concepts of macroevolution[1,2]. However, the relative impacts of extinction and radiation on the co-occurrence of species have not been directly quantitatively compared across the Phanerozoic eon. Here we apply machine learning to generate a spatial embedding (multidimensional ordination) of the temporal co-occurrence structure of the Phanerozoic fossil record, covering 1,273,254 occurrences in the Paleobiology Database for 171,231 embedded species. This facilitates simultaneous comparison of macroevolutionary disruptions, using measures independent of secular diversity trends. Among the 5% most significant periods of disruption, we identify the ‘big five’ mass extinction events[2], seven additional mass extinctions, two combined mass extinction–radiation events and 15 mass radiations. In contrast to narratives that emphasize post-extinction radiations[1,3], we find that the proportionally most comparable mass radiations and extinctions (such as the Cambrian explosion and the end-Permian mass extinction) are typically decoupled in time, refuting any direct causal relationship between them. Moreover, in addition to extinctions[4], evolutionary radiations themselves cause evolutionary decay (modelled co-occurrence probability and shared fraction of species between times approaching zero), a concept that we describe as destructive creation. A direct test of the time to over-threshold macroevolutionary decay[4] (shared fraction of species between two times ≤ 0.1), counted by the decay clock, reveals saw-toothed fluctuations around a Phanerozoic mean of 18.6 million years. As the Quaternary period began at a below-average decay-clock time of 11 million years, modern extinctions further increase life’s decay-clock debt.

Data accompanying Hoyal Cuthill et al 2020, Nature, DOI 10.1038/s41586-020-3003-4, including the occurrence numbers for the source datasets used in the study (downloaded from the Paleobiology Database at https://paleobiodb.org) and custom data generated in the study.

In use of provided custom data please cite the associated paper (Hoyal Cuthill et al 2020, Impacts of speciation and extinction measured by an evolutionary decay clock. Nature, DOI: 10.1038/s41586-020-3003-4).

ReadMe
For Dryad dataset https://doi.org/10.5061/dryad.b8gtht79t
Hoyal Cuthill et al Nature DOI 10.1038/s41586-020-3003-4

Files and usage notes:

fossil_embeddings.txt
Spatial embedding of fossil species produced in the associated publication (Hoyal Cuthill et al, Nature).
This file can be opened with a text editor or read in by the Supplementary Computer code of the associated publication (Hoyal Cuthill et al, Nature).
Rows correspond to fossil species, columns correspond to axes of the multidimensional spatial embedding, values give locations of species on these axes.

fossil_time_embeddings.th
This file is a PyTorch state dictionary object, recording machine learning parameter values from the analysis of the associated publication (Hoyal Cuthill et al, Nature).
This file can be opened with a CUDA enabled PyTorch instance. For example see instructions for handling PyTorch state dictionary files at:
https://pytorch.org/tutorials/beginner/saving_loading_models.html

OccNs_pbdb_data_1lineheader.csv
This is a comma separated values file containing the occurrence numbers of the Paleobiology Database download of the complete dataset used in the associated publication (Hoyal Cuthill et al, Nature). Corresponding full raw data can be downloaded from https://paleobiodb.org/ using the following data download url:
http://paleobiodb.org/data1.2/occs/list.csv?datainfo&rowcount&taxon_reso=genus&max_ma=1000&min_ma=0&show=class,coords,env

OccNs_pbdb_data19_03_20_species_lineheader.csv
This is a comma separated values file containing the occurrence numbers of the Paleobiology Database download of the taxonomically screened dataset used in the associated publication (Hoyal Cuthill et al, Nature).
Corresponding full raw data can be downloaded from https://paleobiodb.org/ using the following data download url:
http://paleobiodb.org/data1.2/occs/list.csv?datainfo&rowcount&taxon_reso=species&max_ma=1000&min_ma=0&show=class,acconly,coords,paleoloc,env,acconly

stimes.txt
This is a text file containing time range mid points for analysed fossil species as described further in the Supplementary Computer code of the associated publication (Hoyal Cuthill et al, Nature).