Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Impacts of speciation and extinction measured by an evolutionary decay clock

A Publisher Correction to this article was published on 13 January 2021

This article has been updated

Abstract

The hypothesis that destructive mass extinctions enable creative evolutionary radiations (creative destruction) is central to classic concepts of macroevolution1,2. 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 the 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 events2, seven additional mass extinctions, two combined mass extinction–radiation events and 15 mass radiations. In contrast to narratives that emphasize post-extinction radiations1,3, 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 extinctions4, 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 decay4 (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.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

All prices are NET prices.

Fig. 1: Time structure of the fossil record.
Fig. 2: Macroevolutionary decay.
Fig. 3: Balance between mass radiation and extinction.

Data availability

Raw data are publicly available in the Paleobiology Database at https://paleobiodb.org. Additional source data for Figs. 13 are provided in the Dryad data repository (https://doi.org/10.5061/dryad.b8gtht79t). Additional data are provided as Extended Data Figs. 17.

Code availability

Custom computer code is provided as Supplementary Computer Code Code 16.

Change history

  • 13 January 2021

    A Correction to this paper has been published: https://doi.org/10.1038/s41586-020-03178-4.

References

  1. Simpson, G. G. Tempo and Mode in Evolution (Columbia Univ. Press, 1944).

  2. Raup, D. M. The role of extinction in evolution. Proc. Natl Acad. Sci. USA 91, 6758–6763 (1994).

    Article  ADS  CAS  Google Scholar 

  3. Hull, P. M., Darroch, S. A. F. & Erwin, D. H. Rarity in mass extinctions and the future of ecosystems. Nature 528, 345–351 (2015).

    Article  ADS  CAS  Google Scholar 

  4. Van Valen, L. A new evolutionary law. Evol. Theory 1, 1–30 (1973).

    Google Scholar 

  5. Jablonski, D. Extinctions: a paleontological perspective. Science 253, 754–757 (1991).

    Article  ADS  CAS  Google Scholar 

  6. Jablonski, D. Lessons from the past: evolutionary impacts of mass extinctions. Proc. Natl Acad. Sci. USA 98, 5393–5398 (2001).

    Article  ADS  CAS  Google Scholar 

  7. Budd, G. E. & Mann, R. P. History is written by the victors: the effect of the push of the past on the fossil record. Evolution 72, 2276–2291 (2018).

    Article  Google Scholar 

  8. Lehman, J. & Miikkulainen, R. Extinction events can accelerate evolution. PLoS One 10, e0132886 (2015).

    Article  Google Scholar 

  9. Sepkoski, J. J. A kinetic model of Phanerozoic taxonomic diversity. III. Post-Paleozoic families and mass extinctions. Paleobiology 10, 246–267 (1984).

    Article  Google Scholar 

  10. Stroud, J. T. & Losos, J. B. Ecological opportunity and adaptive radiation. Annu. Rev. Ecol. Evol. Syst. 47, 507–532 (2016).

    Article  Google Scholar 

  11. Field, D. J., Benito, J., Chen, A., Jagt, J. W. M. & Ksepka, D. T. Late Cretaceous neornithine from Europe illuminates the origins of crown birds. Nature 579, 397–401 (2020).

    Article  ADS  CAS  Google Scholar 

  12. Wood, R. et al. Integrated records of environmental change and evolution challenge the Cambrian Explosion. Nat. Ecol. Evol. 3, 528–538 (2019).

    Article  Google Scholar 

  13. Fan, J. X. et al. A high-resolution summary of Cambrian to Early Triassic marine invertebrate biodiversity. Science 367, 272–277 (2020).

    Article  ADS  CAS  Google Scholar 

  14. Muscente, A. D. et al. Quantifying ecological impacts of mass extinctions with network analysis of fossil communities. Proc. Natl Acad. Sci. USA 115, 5217–5222 (2018).

    Article  ADS  CAS  Google Scholar 

  15. Alroy, J. Are Sepkoski’s evolutionary faunas dynamically coherent? Evol. Ecol. Res. 6, 1–32 (2004).

    Google Scholar 

  16. Brett, C. E., Ivany, L. C. & Schopf, K. M. Coordinated stasis: an overview. Palaeogeogr. Palaeoclimatol. Palaeoecol. 127, 1–20 (1996).

    Article  Google Scholar 

  17. Blanchet, F. G., Cazelles, K. & Gravel, D. Co-occurrence is not evidence of ecological interactions. Ecol. Lett. 23, 1050–1063 (2020).

    Article  Google Scholar 

  18. Sepkoski, J. J., Jr. Rates of speciation in the fossil record. Phil. Trans. R. Soc. Lond. B 353, 315–326 (1998).

    Article  Google Scholar 

  19. Sadler, P. M. Quantitative biostratigraphy—achieving finer resolution in global correlation. Annu. Rev. Earth Planet. Sci. 32, 187–213 (2004).

    Article  ADS  CAS  Google Scholar 

  20. Alroy, J. et al. Phanerozoic trends in the global diversity of marine invertebrates. Science 321, 97–100 (2008).

    Article  ADS  CAS  Google Scholar 

  21. Na, L. & Kiessling, W. Diversity partitioning during the Cambrian radiation. Proc. Natl Acad. Sci. USA 112, 4702–4706 (2015).

    Article  ADS  CAS  Google Scholar 

  22. Kearsey, T. I. et al. The terrestrial landscapes of tetrapod evolution in earliest Carboniferous seasonal wetlands of SE Scotland. Palaeogeogr. Palaeoclimatol. Palaeoecol. 457, 52–69 (2016).

    Article  Google Scholar 

  23. Van Valen, L. Adaptive zones and the orders of mammals. Evolution 25, 420–428 (1971).

    Article  Google Scholar 

  24. Benton, M. J. The Red Queen and the Court Jester: species diversity and the role of biotic and abiotic factors through time. Science 323, 728–732 (2009).

    Article  ADS  CAS  Google Scholar 

  25. Newman, M. E. J. & Eble, G. J. Decline in extinction rates and scale invariance in the fossil record. Paleobiology 25, 434–439 (1999).

    Article  Google Scholar 

  26. Fischer, A. G. & Arthur, M. A.  in Deep Water Carbonate Environments (eds Cook, H. E. & Enos, P. E.) 10–50 (Society of Economic Paleontologists and Mineralogists, 1977).

  27. Raup, D. M. & Sepkoski, J. J. Jr. Periodicity of extinctions in the geologic past. Proc. Natl Acad. Sci. USA 81, 801–805 (1984).

    Article  ADS  CAS  Google Scholar 

  28. Gilinsky, N. L. Volatility and the Phanerozoic decline of background extinction intensity. Paleobiology 20, 445–458 (1994).

    Article  Google Scholar 

  29. Pimiento, C. et al. The Pliocene marine megafauna extinction and its impact on functional diversity. Nat. Ecol. Evol. 1, 1100–1106 (2017).

    Article  Google Scholar 

  30. Gradstein, F. M., Ogg, J. G. & Smith, A. G. A Geologic Time Scale 2004 (Cambridge Univ. Press, 2004).

  31. Ross, R. J., Adler, F. J., Amsden, T. W., Bergstrom, D. & Bergström, S. M. The Ordovician System in the United States: Correlation Chart and Explanatory Note (International Union of Geological Scientists, 1982).

  32. Walker, J. D., Geissman, J. W., Bowring, S. A. & Babcock, L. E. The Geological Society of America geologic time scale. Geol. Soc. Am. Bull. 125, 259–272 (2013).

    Article  ADS  CAS  Google Scholar 

  33. Gilinsky, N. L. & Bambach, R. K. Asymmetrical patterns of origination and extinction in higher taxa. Paleobiology 13, 427–445 (1987).

    Article  Google Scholar 

  34. Peters, S. E. & McClennen, M. The Paleobiology Database application programming interface. Paleobiology 42, 1–7 (2016).

    Article  Google Scholar 

  35. Caswell, B. A. & Frid, C. L. J. Learning from the past: functional ecology of marine benthos during eight million years of aperiodic hypoxia, lessons from the Late Jurassic. Oikos 122, 1687–1699 (2013).

    Article  Google Scholar 

  36. van der Maaten, L. & Hinton, G. Visualizing data using t-SNE. J. Mach. Learn. Res. 9, 2579–2605 (2008).

    MATH  Google Scholar 

  37. Mikolov, T., Chen, K., Corrado, G. S. & Dean, J. A. Efficient estimation of word representations in vector space.. Preprint at https://arxiv.org/abs/1301.3781 (2013).

  38. Schroff, F., Kalenichenko, D. & Philbin, J. in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 815–823 (Institute of Electrical and Electronics, 2015).

  39. Hoyal Cuthill, J. F., Guttenberg, N., Ledger, S., Crowther, R. & Huertas, B. Deep learning on butterfly phenotypes tests evolution’s oldest mathematical model. Sci. Adv. 5, eaaw4967 (2019).

    Article  ADS  Google Scholar 

  40. Ruder, S. An overview of gradient descent optimization algorithms. Preprint at https://arxiv.org/abs/1609.04747 (2017).

  41. Dietterich, T. Overfitting and undercomputing in machine learning. ACM Comput. Surv. 27, 326–327 (1995).

    Article  Google Scholar 

  42. Goldberg, Y. & Levy, O. word2vec explained: deriving Mikolov et al.’s negative-sampling word-embedding method. Preprint at https://arxiv.org/abs/1402.3722 (2014).

  43. Heim, N. A. & Peters, S. E. Covariation in macrostratigraphic and macroevolutionary patterns in the marine record of North America. Geol. Soc. Am. Bull. 123, 620–630 (2011).

    Article  ADS  Google Scholar 

  44. Bacaro, G. & Ricotta, C. A spatially explicit measure of beta diversity. Community Ecol. 8, 41–46 (2007).

    Article  Google Scholar 

Download references

Acknowledgements

This research was supported by funding from an IADS Research Fellowship (J.F.H.C.), EON Research Fellowship (J.F.H.C.) at the Tokyo Institute of Technology (supported by a grant from the John Templeton Foundation), Earth–Life Science Institute Research Interactions Committee Visitor Fund (N.G. and J.F.H.C.) and Swedish Research Council (VR grant no. 2015-04726, G.E.B.). We thank S. Newman and L. Schalkwyk for computing time and support, S. Conway Morris and E. Mitchell for highly constructive comments on the manuscript.

Author information

Authors and Affiliations

Authors

Contributions

J.F.H.C., N.G and G.E.B. designed research, N.G. and J.F.H.C. wrote computer code and performed analyses and J.F.H.C. wrote the paper with input from all authors.

Corresponding author

Correspondence to Jennifer F. Hoyal Cuthill.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Emily Mitchell and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Methods graphical summary and effects of computer simulated diversity increases.

a, Graphical summary of the ML method. bg, Computer simulations of secular variation in diversity, testing effects on measures of co-occurrence structure used in this study. bg, Linear (bd) and exponential (eg) diversity increases (Supplementary Computer Code 3). b, e, Heatmaps visualizing the machine learnt spatial embedding distance between mean species locations at different times: yellow, closest; purple, farthest. c, f, Plot of embedding distances between successive times. d, g, Plot of first two principal component axes from the 16-dimensional spatial embedding. ML training times were 3,000 training epochs.

Extended Data Fig. 2 Bootstrap data-resampling results and shared fraction of species between successive times versus mean embedding distance.

a, Bootstrap data-resampling results: differences in embedding distances for 60 reference fossils, compared within 20 A, B, C triplets over 18 technical replicates of bootstrap data re-sampling and ML embedding training. Error bars, s.d. of the distance absolute(A–B) – absolute(A–C): mean 0.77. We expect the embedding distances to be comparatively stable within the time range over which co-occurrence probability is within the evolutionary decay range (observed to be mean 30 Myr for co-occurrence probability to reach 0.1 in the complete data set). b, d, Fraction of species shared between successive times, calculated exhaustively from raw species time ranges (histogram, Extended Data Fig. 6a). c, e, Distance in the ML spatial embedding between mean species locations at successive times. Compared times are at increments of 1 Myr. b, c, Complete fossil occurrence data set. d, e, Taxonomically screened data set. Vertical lines indicate the 5% most significant times of fractional species turnover (Fig. 3, Extended Data Fig. 4): mass extinctions (red), mass radiations (blue) and mixed mass extinction–radiations (magenta).

Extended Data Fig. 3 Proportions of species originating versus going extinct.

1-Myr increments from 600 to 0 Ma with a threshold of 30% species entry/exit threshold, grey square. This threshold highlights the top 66 times of turnover from 222 total turnover times identified among 600 times included in the analysis. Red, mass extinctions; blue, mass radiations; magenta, mixed mass extinction–radiations.

Extended Data Fig. 4 Times of greatest fractional species turnover in the Phanerozoic fossil record.

Top 5% most significant proportionate extinction or origination times (corresponding to the 30 labelled and coloured times in Fig. 3, > 42% species entry/exit threshold). Drill plots for focal times (key, top left) comparing stratigraphic ranges of all species occurring within 1 Myr of the focal time, vertically sorted into originations, extinctions and crossing ranges. Colours indicate over threshold mass extinctions (red), mass radiations (blue) and mixed mass extinction–radiations (magenta). Relevant stratigraphic unit names, dates and corresponding references are those used in the PBDB (*ref. 30; **refs. 31,32).

Extended Data Fig. 5 Breakdown by phylum of species extinctions and originations at the top 5% of evolutionary disruption times.

Associated with Fig. 3, Extended Data Fig. 4. Proportions of species entering (dark blue) or exiting (dark red) the fossil record are shown for the 19 most prevalent phyla in the data set (taxonomically screened data set).

Extended Data Fig. 6 Raw species time ranges and diversity counts and examples of the decay in probability of temporal co-occurrence.

a, Raw species time ranges: time ranges (maximum occurrence – minimum occurrence) for 137,779 fossil species (taxonomically screened data set). Taxonomically screened Phanerozoic data set (535–0 Ma): median = 6.5 Myr, mean = 9.95 Myr, s.d. = 12.86. Complete data set: median 7 Myr, mean 14.4 s.d. = 28.1 Myr. b, Raw diversity counts: sampled-in-bin taxonomic diversity of genera (grey dashed line) and families (black line) for the complete data set, output by the PBDB within the default time bin of geological ages (at maximum Ma). cf, Examples of decays in co-occurrence probability (c, e) or in shared fraction of species (d, f), from base times 1 Myr before versus after major evolutionary disturbance events. Grey dashed lines indicate a value of 0.1. c, d, End-Permian mass extinction at 252 Ma. e, f, Carboniferous mass radiation at 358 Ma. Following a disturbance event, co-occurrence probabilities and shared fractions of species fall more rapidly to low levels because comparatively few living species co-occur with any species that were present in the past.

Extended Data Fig. 7 Conceptual diagram comparing measures of macroevolutionary decay, decay-clock detail focusing on the last 40 Myr and statistical relationships between measures of macroevolutionary disturbance and time.

a, Set representation of the shared fraction of species between compared times (for example, times t1 and t2). This measure is used in this study and is closely conceptually related to the co-occurrence probability calculated using the ML spatial embedding (see Methods for further details). b, Fraction of surviving species, a core concept of standard methods of survivor analysis for example4. These measures (a, b) will be equal if no new species have originated by time t2 (scenario in c). Where new species have instead originated by time t2, their effect will be picked up by the measures used in this study (a), whereas the impact of new species would not be considered by measures only of the fraction of survivors from t1 (b). d, Vertical lines indicate times of evolutionary disturbance (blue, mass radiations; red, mass extinctions, corresponding to Fig. 3; grey, turnover events below the mass-event threshold). e, (1), measures of disturbance to co-occurrence structure calculated between consecutive time windows are largely independent of Phanerozoic time (over which there have been secular trends in raw diversity20). The shared fraction of species shows no significant relationship with time (taxonomically screened data set). The embedding distance (complete data set) shows a weak relationship across the whole Phanerozoic that is removed when Cenozoic data are excluded (data excluded in order to isolate hypothesized effect after initial data analysis), consistent with a weak effect on Cenozoic embedding distance from fossils with ranges extending to 0 Ma (which are particularly abundant in the data set). (2), proportions of species exiting or entering the fossil record within 1 Myr of a given time show no significant relationship with time (taxonomically screened data set). All statistical tests are two-tailed.

Supplementary information

Reporting Summary

Supplementary Data

This zip folder contains Supplementary Computer Code 1-6, with Python program files supplied in py or ipynb file format and in pdf format. Descriptions of each code are included in an accompanying text file.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Hoyal Cuthill, J.F., Guttenberg, N. & Budd, G.E. Impacts of speciation and extinction measured by an evolutionary decay clock. Nature 588, 636–641 (2020). https://doi.org/10.1038/s41586-020-3003-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41586-020-3003-4

This article is cited by

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing