Extinction periodicity or episodicity

Mention has already been made (in Chapter 3) of Jack Sep-koski's huge compendium of Phanerozoic marine families, which was the basis for the recognition with his colleague Dave Raup of the 'big five' mass-extinction events. Later on, using a combination of methods involving Fourier analysis and Monte Carlo simulation, they discovered to their great excitement a statistically highly significant 26 million-year periodicity in extinction events during the past 250 million years. This was a most surprising result, because rare events such as floods and hurricanes are apparently randomly distributed in time. In their paper in the Proceedings of the National Academy of Sciences published early in 1984 they concluded that 'it seems inescapable that the post-late Permian extinction record contains a 26 Myr periodicity.' Since they could not conceive of any terrestrial process that would be so periodic they were inclined to favour an extraterrestrial cause.

Well before Raup and Sepkoski's paper was published many people knew about their astonishing findings. Sepkoski had presented the results at a sympsium on extinctions held at Flagstaff, Arizona, the previous autumn, preprints had been circulated, and the media had been aroused, with articles appearing in both the scientific and popular press. Thus it was that only two months after the Raup and Sepkoski paper appeared in print, the issue of Nature for 19 April 1984 contained five papers that were written in direct response to it. This phenomenon led the distinguished editor of Nature, John Maddox, to complain in that issue about the circulation of preprints to a select inner circle, cutting out others who were not in the network. (Raup subsequently endeavoured to exculpate himself from this venal sin by pointing out that the scientists in question had requested preprints, having heard of the periodicity work through articles in the press.)

The five papers fell into three categories. Walter Alvarez and his astrophysicist colleague at Berkeley, Richard Muller, had teamed up to subject the record of impact craters on the Earth to time-series analysis. Only 13 craters, of a much larger number of known craters, were dated adequately enough but these were claimed to be sufficient to enable a 28-million-year periodicity to be detected. Then there were two astronomical hypotheses relating extinction events to comet impact, the galactic plane, and the companion star, each put forward independently by two groups of researchers.

The galactic-plane hypothesis was proposed independently by two groups of American astronomers. It takes account of the fact that the Sun crosses the plane of our galaxy twice in a cycle lasting between about 62 and 67 million years. Close to the galactic plane there is an increased likelihood of encountering giant clouds of gas and dust, known as the molecular clouds. These could perturb the cloud of numerous comets close to the Solar System known as the Oort cloud - widely accepted by astronomers though never observed - causing a few of those comets to collide with the Earth. Several of these comets could perhaps hit the Earth in a period of time up to about a million years, with disastrous consequences for many organisms. Periodic extinctions might thus be explained, although the period would be nearer to 30 than 26 million years.

The alternative hypothesis, by two other groups of Ameri can astronomers including Richard Muller, postulated that the Sun had an unseen companion star, subsequently dubbed 'Nemesis' after the ancient Greek goddess who ensured that no mere mortal ever challenged the dominance of the gods. This star had a highly eccentric orbit. When near the perihelion (the point at which it was nearest to the Sun) it was supposed to perturb the orbits of comets in the Oort cloud, thereby initiating an intense comet shower upon the Earth. The end result from the Earth's point of view was thus much the same, whichever hypothesis might be preferred.

Until this time I had been an interested bystander concerning the extinctions controversy, but I began to be drawn into it by accepting an editorial invitation to contribute a 'News and Views' piece for Nature, commenting on the five papers and the article that had started it all. My general reaction was one of scepticism. I questioned the reliability of the geological timescale used by Raup and Sepkoski. Because of the poor quality of many of the data on which it was based, and the large errors involved, the use of other timescales might have found no periodicity. I also noted that some of their extinction events were extremely dubious; others were minor, and deviated from the purported periodicity by several million years. The argument for periodic cratering through the Phanerozoic, based on so few data points, was not readily believable. What I reacted to most strongly was that a group of astronomers seemed to be blithely entering the extinctions debate provoked by the original Alvarez paper without making any attempt to learn more about what geology had to say on the subject; for instance about global changes of sea level and climate. 'Before astronomers indulge in further speculations about the cause of mass extinctions they would do well to learn something about the rich stratigraphic record of their own planet.' I was left with the impression of a science in which too many theoreticians were chasing too few facts, a situation very different from geology.

Two different controversies were generated as a result of Raup and Sepkoski's article and its immediate aftermath. The first of these concerned the rival merits of the astronomical hypotheses. The Nemesis hypothesis was criticized on two counts: first, because the companion star has never been observed (and still has not, despite years of intensive searching by Muller); and secondly, because the orbit of the supposed companion star would be unstable because of the gravitational deflection induced by passing stars; thus there could be no periodicity. The galactic-plane hypothesis is less easy to dismiss, and the well-established periodicity is indeed tantalizingly close to that claimed by Raup and Sepkoski. However, it faces the serious problem that the Sun is currently very close to the galactic plane but the last purported extinction event was 11 million years ago. According to the hypothesis the Sun should now be at the maximum distance from the galactic plane. A group of British astronomers have pointed out another serious if not insuperable difficulty with the hypothesis: the molecular clouds required for cometary perturbation are too sparsely distributed to make encounters with the Oort cloud plausible.

The second controversy concerned the extinction periodicity. An exciting result based on statistics is bound to attract statisticians into the fray, and two University of Chicago colleagues of Raup and Sepkoski duly obliged. Stigler and Wagner observed that a major component of their extinction periodicity analysis was a significance test that decisively rejected the alternative hypothesis that extinctions occurred randomly. Stigler and Wagner confirmed this result, but discussed two things that led them to conclude that the apparent periodicity could be a statistical artefact. Certain types of error in measurement can enhance a periodic signal or cause a pseudoperiodic signal to emerge from data that are aperiodic. The 'hypothesis of a periodic dynamic structure is so powerful in its implications, and so selective in the case with which it imposes itself on us with limited data sets such as this one, that it must be required to pass a stringent test.'

Raul and Sepkoski countered my criticism of their use of a particular timescale by repeating their analysis with other timescales, and showing that the periodicity persisted, although the error margins were greater. Later use of improved time-scales, such as the Gradstein timescale for the Mesozoic and Harland's timescale for the Cenozoic, has however considerably weakened Raup and Sepkoski's case. The best fit using these newer and better timescales occurs with a periodicity of 20.5 million years centred on the end-Triassic and early Toarcian extinction events. This produces a 50 per cent success rate; only six of the predicted twelve mass extinctions occur at the predicted times. A fit to a 28.5-million-year periodicity centred on the end-Cretaceous and end-Cenomanian events most closely approximates to the periodicity originally proposed by Raup and Sepkoski, although only four of a predicted nine mass extinctions occur near the correct time. Furthermore, as indicated in my 1997 book with Paul Wignall, there are strong reasons for doubting the reality of a number of Raup and Sepkoski's claimed mass-extinction events in the mid- and end-Jurassic, early Cretaceous, late Eocene, and mid-Miocene. An independent compilation of all Phanerozoic families, including terrestrial ones, under the editorship of Mike Benton of the University of Bristol enabled Benton to calculate his own extinction metrics. His cogent conclusion is that 'The present data do not lend strong support to this idea [of periodic mass extinctions], because only six, or perhaps seven, of the events are evident.'

Raup and Sepkoski themselves admitted that they could find no evidence in their analysis of Sepkoski's database for any periodicity in the Palaeozoic. Nowadays there is little talk in the palaeobiological community of any periodicity in mass-extinction events. Their occurrence can be considered with confidence as being random, but I prefer the term 'episodic'.

0 0

Post a comment