[Palaeontology, 2017, pp.

1–9]

RAPID COMMUNICATION

MASS EXTINCTIONS OVER THE LAST 500 MYR: AN

ASTRONOMICAL CAUSE?
by ANATOLY D. ERLYKIN 1 , DAVID A. T. HARPER 2 , 3 * , TERRY SLOAN 4 and
ARNOLD W. WOLFENDALE 5
1
Lebedev Physical Institute, Russian Academy of Sciences, Moscow, Russia; [email protected]
2
Palaeoecosystems Group, Department of Earth Sciences, Durham University, Durham, DH1 3LE, UK; [email protected]
3
Department of Geology, Lund University, S€olvegatan 12, Lund, SE-223 62, Sweden
4
Department of Physics, Lancaster University, Lancaster, LA1 4YB, UK; [email protected]
5
Department of Physics, Durham University, Durham, DH1 3LE, UK; [email protected]
*Corresponding author

Typescript received 9 September 2016; accepted in revised form 19 January 2017

Abstract: A Fourier analysis of the magnitudes and timing Fourier periodograms. This leads us to conclude that, apart
of the Phanerozoic mass extinctions (MEs) demonstrates that from possibly a small number of major events, astronomical
many of the periodicities claimed in other analyses are not causes for MEs can largely be ruled out.
statistically significant. Moreover we show that the periodici-
ties associated with oscillations of the Solar System about the Key words: mass extinctions, periodicity, astronomical pro-
galactic plane are too irregular to give narrow peaks in the cesses, Phanerozoic.

T H E cause (or causes) of mass extinctions (MEs) of mar- showers (Alvarez & Muller 1984). To these can be added
ine and terrestrial biological genera has been debated for some explanations published prior to the Raup & Sep-
many decades (see Hallam 2004) and there is, not sur- koski analysis, including periodic doses of cosmic rays
prisingly, an extensive and impressive portfolio of controlled by reversals in the Earth’s magnetic field (Hat-
research in this area (see e.g. MacLeod 2014, together field & Camp 1970) and climate change based on fluctu-
with Bond & Grasby in press for more recent reviews). ating solar energy and rhythms in mantle convection and
Insofar as the problem is germane to understanding the associated processes (Fischer 1977). The concept of peri-
evolution of life on Earth, its solution is important. odicity, however, has not received universal acceptance.
The cause can either be astrophysical, such as the In a critique of the flurry of astronomical papers, Hallam
impact of asteroids, or terrestrial, due to changes in habi- (1984) noted the many terrestrial causes of mass extinc-
tat together with drama induced by climate change and tion including climate and sea-level changes together with
plate tectonic movements, or both. Our aim here is, volcanicity while emphasising the shortcomings of the
specifically, to determine the existence, or lack, of an fossil record at that time in providing an accurate time
astrophysical influence on MEs and the Earth’s eco- frame. Benton’s (1993, 1995) updated analysis of the fos-
systems through deep time. sil record (Harland et al. 1967) indicated that only three
Periodicity in fossil range data, in a loose sense, has of the ten peaks cited by Raup & Sepkoski (1984) were
been recognized for some time (Newell 1952). The initial real mass extinctions and his data did not validate the
quantification, however, of periodicity in marine mass other peaks. Many of these claims were thus dismissed
extinctions (Raup & Sepkoski 1982) prompted a range of due to inadequate data and poorly calibrated time scales
astronomical explanations: the Sun’s oscillation about a (e.g. Patterson & Smith 1987). In a series of key studies
solar plane (Schwartz & James 1984); oscillation of the Bambach (2006) and his colleagues (Bambach et al. 2004)
Solar System vertically about a galactic plane (Rampino & re-evaluated the data, stating firstly that there were only
Stothers 1984); the presence of a distant solar companion, three major (or big) MEs in the fossil record (end
Nemesis (Davis et al. 1984; Whitmire & Jackson 1984); Ordovician, end Permian and end Cretaceous) and sec-
the existence of a tenth planet (Whitmire & Matese 1985) ondly that ME events were not homogeneous, suggesting
that is beyond the orbit of Pluto; and periodic comet the lack of a common effect and causation. In addition,

© The Palaeontological Association doi: 10.1111/pala.12283 1

2 PALAEONTOLOGY

palaeontological textbooks on both sides of the Atlantic 0.7

(e.g. Benton & Harper 2009; Foote & Miller 2007) have
paid scant attention to periodicity as a key pattern in the 0.6
history of life. Thus the growing body of evidence sug-
gested that each major ME was different and there was

Extinction proportion P
no common cause (e.g. Bambach et al. 2004; Bambach 0.5

2006; Brenchley & Harper 1998). Extinctions, moreover,

were clearly episodic, a series of separate events, rather 0.4
than periodic, occurring at regular intervals.
Within the last decade there has been a renewed interest
0.3
in periodicity with better calibrated time-series data, larger
databases of taxon-range information at the genus level
and more sophisticated analytical techniques. Periodicities 0.2
in fossil-range data have been re-established by a number
of author groups predicting causality from coincident peri- 0.1
odic processes, some astronomical. For example, Rohde &
Muller (2005) demonstrated a 62
3 myr cycle, which is
0
particularly evident in shorter-lived genera. More recently, 0 100 200 300 400 500
Melott et al. (2010) similarly described a 62
3 myr Age of extinction (Ma)
cycle, associated with cosmic rays; Melott & Bambach
FIG. 1. The depth of the extinction or extinction proportion
(2011a) noted a 62 myr cycle with the signal strength
(P) of the genus extinctions as a function of time for the extinc-
decreasing in time due to the accumulation of long-lived tion events. The solid line shows the linear fit up to age 460 Ma
genera; Melott & Bambach (2011b) favoured periodic and the dashed line that for all the data referred to in the text.
sea-level change or astronomical causes to explain that
cycle; Melott et al. (2012) linked the biotic data to a
59.3
3 myr cycle in the strontium isotope record that by examining the time series of MEs from the work of
may be associated with mantle or plate tectonic events; Bambach (2006), Melott & Bambach (2011a, b, 2013,
Melott & Bambach (2010) calculated a 27 myr cycle that 2014) which gives the proportion, P, and age of each genus
ruled out the influence of the distant Nemesis; finally in a extinction as shown in Figure 1. There are two widely used
recalibrated dataset with reference to the most recent geo- databases (see MacLeod 2014): the much-updated range
logical timescale (Gradstein et al. 2012), 27 and 62 myr distribution of families and genera initiated and established
cycles have been detected shifting in and out of phase by the late Jack Sepkoski (http://strata.geology.wisc.edu/jac
(Melott & Bambach 2013, 2014). The causes are unknown. k/) and the occurrence database that forms the basis for
In addition a 56-myr rhythm has been identified in sedi- the Paleobiology Database (https://paleobiodb.org/). We
mentary cycles during the Phanerozoic in North America chose to analyse the former, firstly since both databases
(Meyers & Peters 2011) and developed in terms of marine appeared to perform similarly in time-series analysis
biodiversity change and its relationship to ocean redox (Melott & Bambach 2014) and secondly because that data-
conditions and long-term sea-level fluctuations driven by base, updated where relevant particularly taking account of
plate tectonics (Hannisdal & Peters 2011). Two areas, how- new absolute age constraints, was made available to us
ever, have particularly enlivened the debate. Firstly, through the kindness of Dr Richard Bambach. In all there
Rampino (2015) and Rampino & Caldeira (2015) have re- were 163 genus extinction events; 147, if the large extinc-
introduced the coincidence of asteroid craters with mass tion peaks around 250 Ma and >470 Ma are excluded (see
extinction events, noting a 26–30 myr cycle for extinctions below). The distribution of P values was examined and
and 31
5 myr for cratering. Secondly, this apparently forms the basis for discussion. This was followed by a
matches the Sun’s vertical oscillations through the galactic search for periodicities in the P-record and also in 37
disc (32–42 myr) between crossings, invoking the influence meteorite craters by Fourier analysis. The significance of
of the mid-plane Oort Cloud and a dark matter disc, the the peaks in the Fourier periodograms was examined in
latter providing a topical connection between the evolution some detail and conclusions drawn.
of life, extinctions and events in space (Randall 2015). Additional, complementary and confirmatory analyses
These studies suggest that both biological and geological of more historic datasets provided by Shanan Peters
evolution on Earth may be controlled by a periodicity in together with Rohde & Muller (2005, supporting infor-
galactic dynamics. mation) (see Bambach 2006, supporting information) are
In order to investigate further the reality of periodicity noted below and the details provided in Erlykin et al.
and its relevance for the history of life on Earth, we start (2017).
 ERLYKIN ET AL.: MASS EXTINCTIONS 3

GENUS EXTINCTION PROPORTIONS the addition of the large ME values beyond 470 Ma.
THROUGH TIME (P) Hence we conclude that the data are well represented by
a Gaussian distribution and an exponential tail.
As is well known, the mean P value increases with age in The implication of such a Gaussian form at small val-
an approximately linear fashion (see Fig. 1, solid (dashed) ues of DP is that each P value is the resultant of smaller
line excludes (includes) the large extinctions around 250 scale, that is less catastrophic, events. For DP > 0.1 the
and >470 Ma). Linear fits give a reasonable representation Gaussian component is negligible and the exponential tail
of the data and these are adopted rather than more com- dominates; this strongly suggests contributions from
plicated ones. There are, however, large deviations from mechanisms which caused more catastrophic damage.
the median line. Figure 2 shows the frequency distribu-
tion of DP, the displacement of the P value from the two
linear fits shown in Figure 1. The solid smooth curves in THE SEARCH FOR PERIODICITIES
Figure 2 show a maximum likelihood fit to the data of a
Gaussian distribution plus an exponential tail; a Gaussian Fourier analysis
being a natural curve to fit, not least because it fits so
well for negative DP values. Good fits were obtained with Much has been written about Fourier analysis and the
the value of the Pearson test statistic v2 = 10.9 for 13 statistical methods used to judge the significance of any
degrees of freedom in Figure 2A and v2 = 18.4 for 13 result. Omerbashich (2006) showed that, if a Gauss–
degrees of freedom in Figure 2B. The data at ages beyond Vanıcek spectral analysis of the same data used by Melott
470 Ma have very large positive and negative fluctuations et al. (2010) to deduce the presence of their 62 myr peak,
from the linear fit and therefore seem somewhat anoma- the peak disappears. This shows that manipulation of data
lous, perhaps reflecting the instability and lack of resili- can introduce bias. In this paper we adopt a simple
ence of the Cambrian ecosystem, its different composition approach which does not need binning, manipulation of
and structure (Bambach 1983, 1985; Bush & Bambach the data to fill in gaps or interpolation to fixed time
2011). However, Figure 2B shows that if the whole age intervals. The avoidance of such data manipulation
range is fitted, similar results are obtained to those up to should lead to fewer biases in the analysis. However, to
age 470 Ma in Figure 2A with the exponential tail avoid generating spurious peaks in the Fourier analysis
approximately doubled in amplitude mainly because of some detrending of the data is necessary. Here we adopt

25 30
A B

25
20

20
Number of events

Number of events

15

15

10
10

5
5

0 0
–0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 –0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4
ΔP=deviation from linear fit ΔP=deviation from linear fit

FIG. 2. Frequency distribution of the ‘amplitude’ of the probability of genus extinctions, DP. By amplitude is meant the excursion
from the linear fits in Figure 1. A, for the data from 0 to 470 Ma. B, for data from 0 to 530 Ma. The smooth solid curves shows the
maximum likelihood fit of a Gaussian distribution plus an exponential tail described in the text. The dashed curves show the individ-
ual contributions of the Gaussian and the exponential tail.
4 PALAEONTOLOGY

the simplest method of subtracting the appropriate trend

A
line shown in Figure 1. Detrending by more complicated
curves such as polynomials would only reduce the signifi- 0.04

cance of any Fourier peaks and thereby may lead to valu-

able information being discarded. 0.02

The Fourier integrals for a particular angular frequency

x are deduced by simply averaging the readings. Thus: 0
B
2DT X
N

Genus proportion amplitude

0.04
RðxÞ ¼ DPi cos xti
N i¼1
0.02
and

2DT X
N 0
IðxÞ ¼ DPi sin xti C
N i¼1
0.04

where DPi is the deviation of the P value from the trend

0.02
line of the event at age ti, N is the total number of events
considered and DT is the total time range over which the
0
sample of data is taken. The absolute amplitude of the 0.8 D
Fourier component with frequency x is then given by
0.6
p
AðxÞ ¼ ðRðxÞ2 þ IðxÞ2 Þ 0.4

In order to judge the significance of any observed peak, 0.2

random values of Pi and ti were generated and passed 0
0 10 20 30 40 50 60 70 80 90 100
through the analysis program. The process was repeated
many times and the significance of a peak in the data Period (myr)
judged by the number of occurrences of peaks from the FIG. 3. Fourier amplitude of the genus extinction proportion
random distribution with greater amplitude, and there- as a function of period for the extinction and crater data. A, all
fore significance, than the one observed in the data. 163 extinctions over 530 Ma (detrended by the linear fit to all
data in Figure 1). B, extinctions younger than 250 Ma. C,
extinctions between 270 and 470 Ma. D, including the 37 craters
Periodicity in the time series of P values younger than 250 Ma, each with unit weight. Note that the large
groups of extinctions at around 260 Ma and more than 470 Ma
have been excluded from B and C. The data in B and C were
There is a wealth of literature on claims for periodicities
detrended using the linear fit from 1 to 460 Ma in Figure 1.
in the extinction records (see above), with periodicities
ranging from 13 to 64 myr (Bambach 2006). Currently,
27 myr is marginally favoured, but this is largely because
the perceived frequency of the Solar System oscillating and 4B, C but are included in Figures 3A and 4A. Com-
around the galactic plane is of a similar magnitude (Bah- parison shows that the effects of these peaks on the Four-
call & Bahcall 1985; Shaviv 2002a, b). ier analysis are insignificant.
Figure 3A–C shows periodograms from Fourier analy- To check the statistical significance of the peaks in Fig-
ses of the genus extinction proportions with age from ures 3 and 4, random genus proportions and dates were
Bambach (2006), shown in Figure 1. Since the craters are passed through the analysis chain. The random events
only assigned unit weight, Figure 4 shows for comparison were generated with a distribution of genus proportion of
similar periodograms of the extinctions with each given a similar shape to the data (Figure 2) about the trend
unit weight rather than weighted by the genus propor- line. The process was repeated 1000 times. It was
tion, DP, as in Figure 3. The data are shown separately observed that 10% of the random extinction data had
for the periods 1–250 myr and 270–470 myr which each peaks which were larger, that is, more significant than
correspond roughly to one orbit of the Solar System those seen in the vicinity of 27 myr in Figures 3A–C and
around the Galaxy. Various peaks occur including peaks 60% of those in Figures 4A–C. These fractions show that
around a period near to 27 myr. To see if the large the observed peaks in the data have limited statistical sig-
groups of extinctions around 260 and 500 Ma affect the nificance. It is therefore plausible that the peaks are the
Fourier analyses they are excluded from Figures 3B, C results of statistical fluctuations rather than a repetitive
 ERLYKIN ET AL.: MASS EXTINCTIONS 5

physical process for either the genus proportion (Fig-

A
ure 3) or single events (Figure 4). Other factors which
show that the 27 myr peak is unlikely to be related to an 0.4
astrophysical mechanism of any known kind are as fol-
lows:
1. The peak in the region of 27 myr is present only for 0.2
the interval 1–250 Ma in Figure 3B. It has a different
character in time range 270–470 Ma in Figure 3C.
(Note that the time for the Solar System to orbit the

Relative Fourier amplitude

0
Galaxy is in the order 250 myr). If the signal were real, B
the peak value should be similar in each time range. 0.4
2. The wide range of other peaks at periods with no
astrophysical significance means that the cluster
around 27 myr could be accidental and not related to
0.2
a repetitive astrophysical source.
3. As shown elsewhere (e.g. Wolfendale & Wilkinson
1988) there is no evidence, nor theoretical justifica-
0
tion, for precise ‘bursts’ of asteroids or comets when C
the Solar System crosses the galactic plane.
4. Similarly, cosmic ray effects are negligible insofar as 0.4
changes in cosmic ray intensity variation over the
500 myr interval should be too small to produce MEs
(Bailey et al. 1987; Shaviv, 2002a, b; Sloan & Wolfen- 0.2
dale 2008, 2013 and references therein).
As noted above, two additional, albeit historical, data-
sets were also analysed (see Erlykin et al. (2017) for 0
0 10 20 30 40 50 60 70 80 90 100
details). Fourier analyses of the Bambach dataset gener-
Period (myr)
ated in detail herein, and those for the Rohde & Muller
and Peters datasets, show large peaks at the following fre- FIG. 4. Fourier analysis of all extinctions with each extinction
quencies: (24, 27, 38, 47 and 60 myr), (24.5, 27, 38, 48, given unit weight, for comparison with the crater data in Fig-
ure 3D, rather than weighted by the genus proportion as in Fig-
61 myr) and (25, 27, 38, 47 and 62 myr), respectively. All
ure 3. A, 1–530 Ma. B, 1–250 Ma. C, 270–470 Ma.
three datasets display their major peaks with probabilities
> 10% that they occurred by chance, and thus are not
significant. Understandably, the heights of the peaks differ 1992). (A simple model consisting of a uniform slab
across the analyses, but the shapes of the distributions of matter shows that the oscillation period varies as
(N > P vs P) are the same. the square root of the density in the galactic plane.)
2. Dark matter has two effects. Firstly the effect on the
total mass density and, secondly, the effect of discrete
The variability of the oscillation period of the Solar System ‘clumps’ in deflecting the orbit. Insofar as the total
mass of dark matter in clumps is probably about 10%
The Solar System in its orbital journey round the Galaxy of the total mass, the effect on the period is not neg-
oscillates above and below the galactic plane. It encoun- ligible. Using data from Charbonnier et al. (2012) we
ters different concentrations of mass in this journey, for estimate a 10% variation in the oscillation period of
example it moves into and out of the spiral arms of the the Solar System about the galactic plane from such
Galaxy. In consequence it is continually accelerating and clumps. In fact, the data referred to indicate a ‘signifi-
decelerating. Hence its period and phase are rather vari- cant collision’ every 50 myr. Furthermore, reference
able. Phenomena which cause variations in period and needs to be made to the thin disk of dark matter
phase from one oscillation to the next and, thereby ‘jitter’ model of Randall & Reece (2014). Such a thin disk
in the periodicity, are listed as follows (the references in could cause further changes in the oscillation period
brackets refer to the source of data used to calculate the of the Solar System about the galactic plane as well as
standard deviation in the period). leading to several problems such as the effect on stel-
1. Stellar mass density varies from place to place by lar dynamics.
about 40% during the orbit of the Solar System lead- Taking these factors together it is estimated that the
ing to a 20% variation in period (Scheffler & Els€asser period of successive oscillations varies by at least
20%.
6 PALAEONTOLOGY

Such variability in the period will influence any Fourier crossings of the galactic plane because of the variation in
amplitude peak which is caused by a repetitive process phase and period expected in the Galaxy. Figure 5 shows
such as repetitive crossings of the galactic plane. that astronomical processes with the expected variable
The sensitivity of the Fourier analysis to the variability periodicity cannot leave a discernible spectral peak; in
of the sinusoidal period was investigated by passing which case the significance of peaks in extinctions is irrel-
through the analysis program, samples of events generated evant to the search for astronomical causes.
at random times with a pure sine wave distribution of From this we conclude that there is little evidence that
genus proportions. The starting period of the sine wave MEs have an extra-terrestrial origin (apart from the
was chosen to be 27 myr, and then varied by a fraction Chicxulub asteroid noted below; Alvarez et al. 1980).
generated randomly between events. Figure 5A shows the
results for a pure sine wave and Figure 5B–D as the per-
iod is varied. The variations in period were chosen to be ANALYSIS OF THE CRATER AGES
Gaussian distributed with standard deviations of 2%, 4%
and 6% of 27 myr. It can be seen that the peak broadens The 37 (‘meteroritic’) craters from Rampino (2015) and
and disappears to be less than the noise level if the varia- Rampino & Caldeira (2015) were Fourier analysed. These
tion of the period was generated with more than 5% of craters have relatively well-defined ages. The analysis shows
27 myr. As explained above, any astronomical cause that a peak in the Fourier amplitudes occurs at a period
would be expected to have a larger variation in period near 27 myr (see Figure 3D). Again to test the statistical
and phase than this. The observed Fourier peak at significance of the peaks, 1000 groups of 37 random crater
27 myr is therefore too distinct to be caused by repetitive ages were passed through the Fourier analysis program.
These showed that 39% of the random spectra had larger
(i.e. more significant) peaks, than the one observed in the
A data. This shows that there is a high probability that the
0.04 peak is a statistical fluctuation and hence is not statistically
significant. The evidence that the peak has a repetitive
0.02 astrophysical cause is therefore statistically weak.
The quality of the data is degraded by many effects,
0 such as the rather strange groupings over very short (few
B myr) intervals, the loss of craters which have disappeared
0.04 under the oceans, those prior to the Jurassic largely lost
due to subduction processes, and the degradation of the
0.02 craters due to long term weathering. The latter effect
probably causes the very large differences in frequency of
Amplitude

0 detected craters from place to place over the land.

C A search was made for a correlation between the P value
0.04 for an extinction and the diameter, D, of the nearest crater
in time. No correlation could be found. Hence there seems
0.02
to be no general connection between craters and MEs
(apart from Chicxulub 65 Ma). Neither is there a connec-
tion between the distribution of the integral P values
0
D (N (> P) vs P) and that of bolide energies (represented by
E = RD4; where R is a constant) and the integral energy
0.04
distribution (N (> E) vs E). One would have expected that
P and E would be related if MEs and asteroid impacts
0.02
were strongly correlated. Other candidates such as the
giant Wilkes Land Crater have been associated with the
0
0 10 20 30 40 50 60 70 80 90 100 end Permian extinction; but neither the age of that crater
Period (myr) nor its association with the Permian–Triassic events are
FIG. 5. Typical Fourier analyses of samples of 147 events gen-
proven. Its location under the Antarctic ice (Weihaupt
erated as a pure sine wave distributed as P(t) = 0.04 sin xt. A, 2010) is a formidable barrier to any further investigation
pure sine wave; the value of x is fixed to correspond to a period at present.
of 27 myr. B–D, periods between events were varied by a ran- From this we conclude that there is little evidence from
dom amount with Gaussian distributions of standard deviation: craters that there is a connection between MEs and astro-
B, 0.02; C, 0.04; D, 0.06 times 27 myr. nomical events.
 ERLYKIN ET AL.: MASS EXTINCTIONS 7

CONCLUSIONS DATA ARCHIVING STATEMENT

There is strong evidence that the frequency distribution Data for this study are available in the Dryad Digital Repository:
of the probability of genus extinctions has two compo- https://doi.org/10.5061/dryad.dk385
nents: a near-Gaussian distribution and a small exponen-
Editor. Andrew Smith
tial tail. The mean probability has fallen with time. This
is a consequence of the planet’s increasing biodiversity,
possibly populated too by evolutionarily more stable,
longer-ranging species. REFERENCES
Based on the one event and the energetics of asteroid
impacts (in the order of 1023 Joules for a major impact; A L V A R E Z , W. and M U L L E R , R. A. 1984. Evidence for cra-
see R. Marcus et al., http://impact.ese.ic.ac.uk/ImpactEffec ter ages for periodic impacts on the Earth. Nature, 308, 718–
720.
ts; Shulte et al. 2010), a case can be made for the few
A L V A R E Z , L. W., A L V A R E Z , W., A S A R O , F. and
events in the exponential tail being due to such impacts,
M I C H E L , H. V. 1980. Extraterrestrial cause for the Creta-
although high-energy terrestrial causes, such as those ceous-Tertiary extinction. Science, 208, 1095–1108.
associated with volcanicity (in the order of 2 9 1021 B A H C A L L , J. N. and B A H C A L L , S. 1985. The Sun’s motion
Joules for a major eruption; Blong 1984) or intense cli- perpendicular to the galactic plane. Nature, 316, 706–708.
mate change (e.g. Benton & Twitchett 2003; Harper et al. B A I L E Y , M. E., W I L K I N S O N , D. A. and W O L F E N -
2014; Finnegan et al. 2016) are equally as likely in the D A L E , A. W. 1987. Can episodic comet showers explain the
absence of any geological evidence of impact. The extinc- 30-Myr cyclicity in the terrestrial record? Monthly Notices of
tions in the main Gaussian region are likely to be due to the Royal Astronomical Society, 227, 863–885.
many different causes, for example thermal effects of ter- BAMBACH, R. K. 1983. Ecospace utilization and guilds in marine
restrial origin (e.g. those associated with climate fluctua- communities through the Phanerozoic. 719–746. In T E V E S Z ,
M. and M c C A L L , P. (eds). Biotic interactions in recent and fos-
tions (e.g. Mayhew et al. 2008, 2012) and plate tectonic
sil benthic communities. Topics in Geobiology, 3, Springer.
processes, particularly the effects of large igneous pro-
-1985. Classes and adaptive variety: the ecology of diversifi-
vinces (e.g. Bond & Grasby in press)). cation in marine faunas through the Phanerozoic. 191–253. In
Using Fourier analysis, we show that the evidence for V A L E N T I N E , J. W. (ed.) Phanerozoic diversity patterns: pro-
periodicities in the extinction record is statistically weak. files in macroevolution. Princeton University Press.
Furthermore, we show that periodicity of the oscillation - 2006. Phanerozoic biodiversity mass extinctions. Annual
of the Solar System about the galactic plane is too vari- Review of Earth & Planetary Sciences, 34, 127–155.
able to produce a narrow peak in such a Fourier analysis. - K N O L L , A. H. and W A N G , S. C. 2004. Origination,
Hence the claim of such regular astronomical phenomena extinction and mass depletions of marine diversity. Paleobiol-
contributing to mass extinctions is not well founded. ogy, 30, 522–542.
Instead terrestrial causes are favoured for the vast major- B E N T O N , M. J. (ed.) 1993 The fossil record 2. Chapman &
Hall, 845 pp.
ity of MEs (see also MacLeod 1998, 2005, 2014; Bond &
-1995. Diversity and extinction in the history of life. Science,
Grasby in press).
268, 52–58.
-and H A R P E R , D. A. T. 2009. Introduction to paleobiology
Acknowledgements. We thank Dr Alistair McGowan and an and the fossil record. John Wiley & Sons, 342 pp.
anonymous reviewer for their careful and detailed comments -and T W I T C H E T T , R. 2003. How to kill (almost) all life:
that improved the manuscript. The latter suggested we should the end-Permian extinction event. Trends in Ecology & Evolu-
analyse a couple of the more historic datasets, which we did. Dr tion, 18, 358–365.
Richard Bambach generously permitted use of his updated data- B L O N G , R. J. 1984. Volcanic hazards: a sourcebook on the
base and engaged in robust discussions of our findings. TS, ADE effects of eruptions. Academic Press, 424 pp.
and AWW are grateful to the Kohn Foundation for financial B O N D , D. P. G. and G R A S B Y , S. E. in press. On the causes
support. DATH acknowledges the receipt of a research fellow- of mass extinctions. Palaeogeography, Palaeoclimatology,
ship from the Leverhulme Trust and support from the Wenner- Palaeoecology. doi: 10.1016/j.palaeo.2016.11.005
Gren Foundation (Sweden). B R E N C H L E Y , P. J. and H A R P E R , D. A. T. 1998. Palaeoe-
cology: ecosystems, environments and evolution. Chapman &
Note added in proof. While our paper was in press, DATH was Hall, 402 pp.
made aware of a more detailed, independent study by Matthias B U S H , A. M. and B A M B A C H , R. K. 2011. Paleoecologic
Meier and Sanna Holm-Alwmark (Meier & Holm-Alwmark in megatrends in marine Metazoa. Annual Review of Earth &
press) on the record of impact craters. This is a more focused Planetary Sciences, 39, 241–269.
challenge of the cratering record, emphasising a lack of demon- C H A R B O N N I E R , A., C O M B E T , C. and M A U R I N , D.
strable periodicity, and confirming extravagantly our own 2012. CLUMPY: a code for gamma ray signals from dark matter
conclusions in that area. structures. Computer Physics Communications, 183, 656–668.
 8 PALAEONTOLOGY

D A V I S , M., H U T , P. and M U L L E R , R. A. 1984. Extinction impact cratering record. Monthly Notices of the Royal Astro-
of species by comet showers. Nature, 308, 715–717. nomical Society.
E R L Y K I N , A. D., H A R P E R , D. A. T., S L O A N , T. and M E L O T T , A. L. and B A M B A C H , R. K. 2010. Nemesis
W O L F E N D A L E , A. W. 2017. Data from: Mass extinctions reconsidered. Monthly Notices of the Royal Astronomical Soci-
over the last 500 myr: an astronomical cause? Dryad Data ety, 407, 99–102.
Repository. doi: 10.5061/dryad.dk385 -- 2011a. A ubiquitous ~62-myr periodic fluctuation
F I N N E G A N , S., R A S M U S S E N , C. M. Ø. and H A R P E R , superimposed on general trends in fossil biodiversity. I. Docu-
D. A. T. 2016. Biogeographic and bathymetric determinants of mentation. Paleobiology, 37, 92–112.
brachiopod extinction and survival during the Late Ordovician -- 2011b. A ubiquitous ~62-myr periodic fluctuation
mass extinction. Proceedings of the Royal Society B, 283 (1829), superimposed on general trends in fossil biodiversity. II. Evo-
20160007. lutionary dynamics associated with periodic fluctuation in
F I S C H E R , A. G. 1977. Secular variations in the pelagic realm. marine diversity. Paleobiology, 37, 383–408.
Special Publication, Society of Economic Paleontologists & Min- -- 2013. Do periodicities in extinction—with possible
eralogists, 25, 19–50. astronomical connections—survive a revision of the geological
F O O T E , M. and M I L L E R , A. I. 2007. Principles of paleontol- timescale? The Astrophysical Journal, 773, 6.
ogy, 3rd edn. W.H. Freeman, 354 pp. - -2014. Analysis of periodicity of extinction using the
G R A D S T E I N , F., O G G , J., S C H M I T Z , M. and O G G , G. 2012 geological timescale. Paleobiology, 40, 177–196.
2012. The geologic time scale 2012. Elsevier, 1144 pp. -K R E J C I , A. J., T H O M A S , B. C., M E D V E D D E V , M.
H A L L A M , A. 1984. The causes of mass extinctions. Nature, V., W I L S O N , G. W. and M U R R A Y , M. J. 2010. Atmo-
308, 686–687. spheric consequences of cosmic ray variability in the extra-
-2004. Catastrophes and lesser calamities: the causes of mass galactic shock model: 2. Revised ionization levels and their
extinctions. Oxford University Press, 274 pp. consequences. Journal of Geophysical Research: Planets, 115,
H A N N I S D A L , B. and P E T E R S , S. E. 2011. Phanerozoic E08002.
Earth system evolution and marine biodiversity. Science, 334, - BAMBACH, R. K., P E T E R S E N , K. D. and
1121–1124. M c A R T H U R , J. M. 2012. An ~60-million-year periodicity is
H A R L A N D , W. B., H O L L A N D , C. H., H O U S E , M. R., common to marine 87Sr/86Sr, fossil biodiversity, and large-
H U G H E S , N. F., R E Y N O L D S , A. B., R U D W I C K , M. scale sedimentation: what does the periodicity reflect? Journal
J. S., S A T T E R T H W A I T E , G. E., T A R L O , L. B. H. of Geology, 120, 217–226.
and W I L L E Y , E. C. (eds) 1967. The fossil record: a sympo- M E Y E R S , S. R. and P E T E R S , S. E. 2011. A 56 million
sium with documentation. Geological Society of London, 827 year rhythm in North American sedimentation during the
pp. Phanerozoic. Earth & Planetary Science Letters, 303, 174–
H A R P E R , D. A. T., H A M M A R L U N D , E. U. and R A S - 180.
M U S S E N , C. M. Ø. 2014. End Ordovician extinctions: a N E W E L L , N. D. 1952. Periodicity in invertebrate evolution.
coincidence of causes. Gondwana Research, 25, 1294–1307. Journal of Paleontology, 26, 371–385.
H A T F I E L D , C. B. and C A M P , M. J. 1970. Mass extinctions O M E R B A S H I C H , M. 2006. Gauss-Vanıcek spectral analysis
correlated with periodic galactic events. Bulletin of the Geologi- of the Sepkoski compendium: no new life cycles. Computing
cal Society of America, 81, 911–914. in Science & Engineering, 8, 26–30.
M A C L E O D , N. 1998. Impacts and marine invertebrate extinc- P A T T E R S O N , C. and S M I T H , A. B. 1987. Is the period-
tions. 217–246. In G R A D Y , M. M., H U T C H I S O N , R., icity of extinctions a taxonomic artefact? Nature, 330,
M c C A L L , G. J. H. and R O T H E R Y , D. A. (eds). Meteorites: 248–251.
flux with time and impact effects. Geological Society, London, R A M P I N O , M. R. 2015. Disc dark matter in the Galaxy and
Special Publication, 140. potential cycles of extraterrestrial impacts, mass extinctions
- 2005. Mass extinction causality: statistical assessment of and geological events. Monthly Notices of the Royal Astronomi-
multiple-cause scenarios. Russian Journal of Geology & Geo- cal Society, 448, 1816–1820.
physics, 9, 979–987. -and C A L D E I R A , K. 2015. Periodic impact cratering and
-2014. The geological extinction record: history, data, biases, extinction events over the last 260 million years. Monthly
and testing. 1–28. In K E L L E R , G. and K E R R , A. C. (eds). Notices of the Royal Astronomical Society, 454, 3480–3484.
Volcanism, impacts, and mass extinctions: causes and effects. -and S T O T H E R S , R. B. 1984. Terrestrial mass extinctions,
Geological Society of America Special Paper, 505. cometary impacts and the Sun’s motion perpendicular to the
M A Y H E W , P. J., J E N K I N S , G. B. and B E N T O N , T. G. galactic plane. Nature, 308, 709–712.
2008. A long-term association between global temperature and R A N D A L L , L. 2015. Dark matter and the dinosaurs. Harper
biodiversity, origination and extinction in the fossil record. Collins, 432 pp.
Proceedings of the Royal Society of London B, 275, 47–53. -and R E E C E , M. 2014. Dark matter as a trigger for peri-
-B E L L , M. A., B E N T O N , T. G. and M c G O W A N , A. J. odic comet impacts. Physical Review Letters, 112, 161301.
2012. Biodiversity tracks temperature over time. Proceedings of R A U P , D. M. and S E P K O S K I , J. J. Jr 1982. Mass extinctions
the National Academy of Sciences, 109, 15141–15145. in the marine fossil record. Science, 215, 1501–1503.
M E I E R , M.M.M. and H O L M - A L W M A R K , S. in press. A - -1984. Periodicity of extinctions in the geologic past.
tale of clusters: no resolvable periodicity in the terrestrial Proceedings of the National Academy of Sciences, 81, 801–805.
 ERLYKIN ET AL.: MASS EXTINCTIONS 9

R O H D E , R. A. and M U L L E R , R. A. 2005. Cycles in fossil V I E Y R A , M., R E I M O L D , W. U., R O B I N , E., S A L G E ,

diversity. Nature, 434, 208–210. T., S P E I J E R , R. P., S W E E T , A. R., U R R U T I A - F U C U -
S C H E F F L E R , H. and E L S A € S S E R , H. 1992. Bau und Physik G A U C H I , J., V A J D A , V., W H A L E N , M. T. and W I L -
der Galaxis. Wissenschaftsverlag, Mannheim, 642 pp. L U M S E N , P. S. 2010. The Chicxulub asteroid impact and
S C H W A R T Z , R. D. and J A M E S , P. B. 1984. Periodic mass mass extinction at the Cretaceous-Paleogene boundary.
extinctions and the Sun’s oscillation about the galactic plane. Science, 327, 1214–1218.
Nature, 308, 712–713. S L O A N , T. and W O L F E N D A L E , A. W. 2008. Testing the
S H A V I V , N. J. 2002a. Cosmic ray diffusion from the galactic proposed causal link between cosmic rays and cloud cover.
spiral arms, iron meteorites, and a possible climatic connec- Environmental Research Letters, 3, 024001.
tion. Physical Review Letters, 89, 051102. - -2013. Cosmic rays and climate change over the past
-2002b. The spiral structure of the Milky Way, cosmic rays 1000 million years. New Astronomy, 25, 44–49.
and ice age epochs on Earth. New Astronomy, 8, 39–77. W E I H A U P T , J. G. 2010. Gravity anomalies of the Antarctic
S H U L T E , P., A L E G R E T , L., A R E N I L L A S , I., A R Z , J. A., lithosphere. Lithosphere, 2, 454–461.
B A R T O N , P. J., B O W N , P. B. , B R A L O W E R , T. J., W H I T M I R E , D. P. and M A T E S E , J. J. 1985. Periodic comet
C H R I S T E N S O N , G. L., C L A E Y S , P., C O C K E L L , C. S., showers and planet X. Nature, 313, 36–38.
C O L L I N S , G. S., D E U T S C H , A., G O L D I N , T. J., -and J A C K S O N , A. A. IV 1984. Are periodic mass extinc-
G O T O , K., G R A J A L E S - N I S H I M U R A , J. M., G R I E V E , tions driven by a distant solar companion? Nature, 308, 713–
R. F., G U L I C K , S. P. S., J O H N S O N , K. R., K I E S S L I N G , 715.
W., K O E B E R L , C., K R I N G , D. A., M a c L E O D , K. G., W O L F E N D A L E , A. W. and W I L K I N S O N , D. A. 1988.
M A T S U I , T., M E L O S H , J., M O N T A N A R I , A., M O R - Periodic mass extinctions: some astronomical difficulties. 231–
G A N , J. V., N E A L , C. R., N I C H O L S , D. J., N O R R I S , 239. In C L U B E , S. V. M. (ed.) Catastrophes and evolution.
R.D., P I E R A Z Z O , E., R A V I Z Z A , G., R E B O L L E D O - Cambridge University Press.