Biological Extinction in Earth History

David M. Raup

Science, New Series, Vol. 231, No. 4745. (Mar. 28, 1986), pp. 1528-1533.

Stable URL:
http://links.jstor.org/sici?sici=0036-8075%2819860328%293%3A231%3A4745%3C1528%3ABEIEH%3E2.0.CO%3B2-8

Science is currently published by American Association for the Advancement of Science.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/aaas.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org
Fri Aug 17 10:12:14 2007
 Biological Extinction in Earth History

most major extinctions fall at important boundaries in the time scale.

Virtually all plant and animal species that have ever lived The clustering of extinctions may have made it possible for the
on the earth are extinct. For this reason alone, extinction geologic time scale to be established on several continents within
must play an important role in the evolution of life. The just a few decades in the mid-19th century.
five largest mass extinctions of the past 600 million years The Cretaceous-Tertiary (K-T) mass extinction is of interest not
are of greatest interest, but there is also a spectrum of only because of its possible relation to comet or asteroid impact, but
smaller events, many of which indicate biological systems also because it is the best documented of the larger extinction events.
in profound stress. Extinction may be episodic at aU It is celebrated especially because it included the demise of the
scales, with relatively long periods of stability alternating dinosaurs, which, in turn, probably made possible the early Tertiary
with short-lived extinction events. Most extinction epi- evolutionary radiation of manlmals, leading ultimately to human
sodes are biologically selective, and further analysis of the evolution. But the dinosaur extinction was actually a small part of
victims and survivors offers the greatest chance of deduc- the total K-T event: there may have been as few as 20 coexisting
ing the proximal causes of extinction. A drop in sea level dinosaur species in the latest Cretaceous (4) with a total biomass
and climatic change are most frequently invoked to ex- that was trivial in comparison with the rest of the global biota.
plain mass extinctions, but new theories of collisions with Large numbers of species in a variety of habitats died out
extraterrestrial bodies are gaining favor. Extinction may sometime near the end of the Cretaceous. In terrestrial habitats,
be constructive in a Darwinian sense or it may only marsupial mammals were hard hit, but many of them, along with
perturb the system by eliminating those organisms that most other mammalian groups, managed to survive. Other groups,
happen to be susceptible to geologically rare stresses. such as the amphibians and many aquatic reptiles, were relatively
unaffected, and land plants suffered only moderate extinction. This
sort of selectivity characterizes all major extinctions, although the
derails are not consistent from one event to the next.

I
N 1980, AI,VAREZ t T A L . QROPOSEI> THAT THE MASS EXTINC- In the marine realm, approximately 13 percent of marine families
tion at the end of the Cretaceous was caused by the impact of a and about 50 percent of marine genera died out completely in the
10-kilometer meteorite (asteroid or comet) (I). This proposal Maestrichtian, the final stage of the Cretaceous (5).These figures
has created a storm of controversy, but it has also stimulated underestimate the actual biological cost because they do not include
valuable research on the detailed geology, geochemistry, and paleon- genera and families that lost many species but still managed to
tology of the ternlinal Cretaceous sequence. Some geologists and survive with at least one species. An interpolation based on average
paleobiologists see these developments as heralding major shifts In numbers of species per higher tmon leads to the estimate that
the way we look at earth history and organic evolution. T o others, somewhere between 60 and 75 percent of the marine species existing
the past 5 years have seen a science gone mad. in the latest Cretaceous became extinct (6). Losses included some of
Up to 4 billion species of plants and animals are estimated to have the most abundant species, particularly among marine plankton. Thus,
lived at some tlme m the geologic past (Z), most of these in the last mortality measured in biomass was probably higher.
600 million years (Phaneromic time). Yet there are only a few A key question is the length of time over which the late
million species living today. Thus, extinction of species has been Cretaceous extinctions occurred. This question is important to the
almost as common as origination. ultimate understanding of the causes of mass extinction and also
The data base for extinction analysis is the distribution in space influences the format of any statistical analysis. The percentages
and time of about 250,000 known fossil species (3), an extremely given above are based on global compilations of the time ranges of
small sample of past life because of the negligible probability of fanlilies and genera (5, 7). For the terminal Cretaceous, extinction
preservation and discovery of any given species. Nevertheless, it is a times are grouped into the Maestrichtian stage, the last unit of the
superb data base when compared with that available for many other Cretaceous that can be recognized easily on a worldwide basis.
aspects of Earth history. When the fossil record is viewed at higher Because this interval lasted about 8 million years, the global data do
taxonomic levels (genera, families, and orders), the quality of not distinguish between instantaneous arld protracted events. It is
sampling is substantially better because the probability of finding a often possible t o define the timing of extinctions more precisely.
record of a multispecies group is higher than for a single species.
Extinctions are clustered in time. Not only are there several
profound mass extinctions like that at the end of the Cretaceous, but
relatively sudden and rapid turnovers occur at lesser scales as well.
Variation in Extinction Intensity
Both large and small extinctions were used by 19th-century geolo- Commonly used measures of the intensity of extinction include (i)
gists to define boundaries in the time scale. It is thus no accident that total taxa becoming extinct in a selected unit of time (a geologic
stage), (ii) extinctions in a tilne unit as a percentage of standing
diversity, (iii) number of extinctions per million years, and (iv)
13. M. Raup is in the llepartment ofthc Geopllysical Sciences, U~livcssityof Chicago,
Chicago, IL 60637. percentage of extinctions per million years. All are subject to
SCIENCE, VOL. 231
sampling problems and persistent biases, and the proper selection of Fig. 1. Distribution of extinction
a measure is rarely obvious. Although the percentage of extinction intensities for the 79 generally rec-
ognized geologic stages of Phanero-
per million years should be the most appropriate because it normal-
izes both for standing diversity (taxa available for extinction) and for
zoic time, based on recorded times
of extinction of 2316 marine animal
2 30

2ol
absolute time, it has two serious problems. First, the time dimension families (7).Extinction intensity for
is an estimated duration of a stage or other conventional time unit, the last stage of the cretaceous
and stage durations are uncertain, based as they are on interpola- (Macstrichtian) is indicated for com- Maestrichtian
parison. z 10
tions between widely separated calibration points. Second, the fillly
normalized metric carries the tacit assumption that extinction is a
0
continuous process. If some major extinction events are effectively 0 40 80
Y T - T l
120 160
instantaneous, normalizing for the duration of some arbitrary time Extinctions per stage

unit (such as the Maestrichtian) obscures the real rate. There is no

statistically significant correlation between stage duration and num-
ber of extinctions per stage for the Phanerozoic; this may be because However, foraminifera1 extinction in the Tertiary can be viewed
of errors in stage durations or because the extinction process is differently. Figure 2B shows pseudocohort decay for a fuller data
indeed dominated by widely spaced, point events. set: more cohorts are considered, and each is monitored with higher
In spite of these problems, five extinctions stand out consistently temporal resolution. In Fig. 2B, each line represents a pseudocohort
as the largest and they are conventionally labeled mass extinctions. starting at a different time, and the data points are connected by
The five are terminal Ordovician (-440 million years ago), late straight lines rather than used only as guides for a generalized fit.
Devonian (-365 million years ago), terminal or late Permian The steplike pattern indicates a discontinuous process. The tendency
(-250 million years ago), terminal Triassic (-215 million years for horizontal "treads" and vertical "risers" suggests a series of point
ago), and terminal Cretaceous (65 million years ago). The ranking events separated by intervals of stability (little or no extinction). An
of the five depends somewhat on database and metric, but the especially marked drop in survivorship 13 million years ago corre-
Permian event usually emerges as the largest, with published sponds to a recognized extinction event near the end of the middle
estimates of species kill ranging as high as 96 percent (6). Miocene (11) .
At the other extreme is "background" extinction, seen as the Figure 3 shows pseudocohort survivorship curves at a different
normal or spontaneous rate of replacement of one species by scale. Survivorship is graphed for the entire Phanerozoic, from a
another. Although names may imply that qualitatively different stage-level time scale and records of 2316 extinct families of marine
processes are operating, this may not be the case. A plot of varying animals. The steplike pattern is still seen, with the major extinction
intensity of extinction for the 79 generally recognized stages of the events indicated by sharp drops. The pattern is most evident in the
Phanerozoic (Fig. 1) shows a spectrum of variation and suggests right third of the figure, where taxonomic data and radiometric
that the simple binary classification of extinction is not warranted. dating are most robust. Comparison of the species and family plots
This does not deny the possibility that extinctions are qualitatively (Figs. 2B and 3) suggests that the patterns have a fractal quality of
different from each other, but it does emphasize that variation is self-similarity with changing scale, although this possibility has not
continuous. The problem resembles that faced in the classification of been confirmed.
floods or severe storms, with mass extinction being analogous to the Is the extinction process fundamentally continuous or episodic?
100-year flood or the hurricane. The foregoing discussion implies that it is episodic, but this has not
been demonstrated in the general case. The question remains
important and bears on other questions of the mechanisms of
Episodic Versus Continuous Extinction extinction.

Figure 2A shows a survivorship curve for species of marine

planktonic foraminifera that is based on census data at intervals of
approximately 5 million years over the past 30 million years. The Dating Extinction Events
point on the upper left represents the group of species found in the Although global data locate extinctions only to the nearest
fossil record 30 million years ago. The fate of this group, or geologic stage, the situation is not quite as bad as this would imply.
"pse~~docohort,"is monitored through geologic time to show the Stage boundaries have traditionally been defined at points of
decay of the group by extinction of its constituent species (8). evolutionary turnover, and, in a number of cases, there isindepen-
Origination of new species plays no role. dent evidence that this is true. It is often possible to date extinctions
A logarithmic ordinate is used in Fig. 2A so that the decay will be on a local or regional scale with considerable precision.
linear if extinction is a continuous, stochastic process with constant Figure 4 shows the occurrences of 50 brachiopod species in the
probability of extinction. The array of seven points appears to be few meters above and below the K-T boundary at one locality in
linear and can be described by the equation Denmark. In this section, the K-T boundary is marked by a
distinctive clay layer ("marl"), within 1meter of which 20 brachio-
pod species disappear. This fits the episodic model and suggests a
where So and S,are the numbers of survivors at some initial time and sudden environmental or other pem~rbation.However, species 21
at time t, respectively, and g is the probability of extinction of a to 26 in Fig. 4 disappear from the record near the K-T boundary
species per million years. A best-fit line in this case yields g = 0.13, only to reappear four or more meters higher. Clearly, these six
the reciprocal of which is an estimate of mean species duration (7.7 species did not actually die out. This represents a common situation
million years) for the pseudocohort. which has been called the Lazarus effect by Jablonski (12).All 26
Figure 2A and its linear interpretation are completely compatible species may have lived during all or part of the barren interval but
with the model of continuous background extinction known as Van were not preserved because of changes in the depositional regime,
Valen's law (9). This model is attractive because it makes the postdepositional chemical solution, or greatly reduced population
extinction process mathematically tractable and predictable (10). sizes. It is also possible that these species simply migrated away from
28 MARCH 1986 AKTICLES 1529
 environmental changes (sudden or gradual) that cause hiatuses in
sedimentation may also cause biological extinction. This is true for
the anoxic events associated with cxtinctions at the base of the
Silurian, in the late Devonian, at the top of the Triassic, and at the
tops of the Pliensbachian (Jurassic), Cenomanian (Cretaceous), and
Maestrichtian (Cretaceous) stages (13).
In the examples just described, extinction events may appear to be
sharper than they actually are, but there is another set of conditions
that has the opposite effect. For most organisms, it is unlikely that
the true last occurrence of an extinct species or family will be
recorded. Therefore, almost all observed time ranges are truncated.

38 30 20 10
- - I-
038
(Samples)

30
---
20
L

10
Y L " ' ,

0
This causes a "smearing" of the record of an extinction event
backward in time and is called the Signor-Lipps effect (14).The fact
lo6 Years ;go that the highest known dinosaur bone in the K-T section in eastern
Fig. 2. Sunivorship of species of planktonic foraminifera. (A) Survivorship Montana is 3 meters below the iridium anomaly is attributed by
of a cohort ~nonitoredat approxitnately 5-n~illion-yearintervals, suggesting some to the Signor-Lipps effect. On a larger scale, extinction of
constant probability of extinction in a purely stochastic systern (data from families in the late Perniia~rappears to cxtend ovcr the last sevcral
(33)].(B) More complete survivorship pattern for cohorts of planktonic stagcs of the Pcrmian, but this may just bc a rcsult of range
foraminifera, suggesting a11 irregular and distinctly episodic sunrivorship truncation during a long interval of lowercd sea levcl and incomplete
pattern; sample points are shown inside the horizontal scale [redrawn from
figure 9C in (.33)1. sedimentary sections. The Signor-Lipps cffcct is difficult to prove
unless Lazarus taxa arc available to evaluatc the probability of-
nonprcscrvation.
the site. In any event, the significancc of thc "extinctions" noted at The dating of cxtinction events requires much morc work.
the beginning of the intenla1 is jeopardized. Because of the problenis and ambiguities found throughout the
The Lazarus effcct is known to occur on a large scale aftcr the record, capable geologists and palcontologists diffcr sharply on
Permian cxtinctions: several million years of the succceding Triassic interyrctations of obscnlations.
rccord show exccllcnt fossil prcscrvation but lack all traces of several
major groups of rnarinc organisms (Lazan~staxa) known to have
survived the Pcrnmian cxtinction.
Figure 5 shows detailcd records of planktonic foraminifera1 Selectivity of Extinction
spccics through about 15 million years of-the Eocene and Oligoccnc Major cxtinction events are selective-that is, the victims and
epochs of the Tertiary. Again, thcrc is evidcncc of steplike extinc- sunlivors arc not random san~plesof the pre-extinction biota. This
tion, but there are altcrnativc cxplanations. In this case, thcrc is opcrls the way to carcful analysis of t l ~ cphysiological, ecological,
indcpendcnt cvidence for hianises in sedimentary dcposition (wavy and biogeographic common denominators of the surviving or
lincs at three horizons). Such hiatuses triulcatc rangcs of species and nonsurviving spccics. With this inforniation, a fuller utldcrsta~rding
thus cnhancc the imprcssion of simultaneous extinction. The termi- of the causes of extinction should be straightforward.
nation of the first four specics in Fig. 5 falls just below a gap in Sonie aspccts of organismal biology appear to bc relatcd to
sampling. resistance to cxtinction. Largc population size, broad gcographic
Even if a hiatus is known to occur at an apparcnt extinction point, distribution, and high dispersal potential should help protect spccies
simultaneous extinction can occasionally be dcmonstratcd. Hiatuses and higher taxa from extinction, and this appears to be the case for
in the geologic rccord are commonly associated with true (indcpcn- periods of background extinction. Jablonski has shown that for rrlost
dently verified) cxtinction events. Especially in decp-sea situations, of the last third of the Cretaceous, rrlollusks with high dispersal and

Fig. 3. Sutvivorship of 2316 families

of marine animals over the past 600
million years ( 7 ) . Each line is a
"pseeudocohort" which starts (upper
left) with thc families present in the
fossil rccord at a point in time. Dc-
cay of the cohorts is monitored for
all rccognizcd geologic stages (indi-
cated by thc small boxes along the
base of the graph). Mass cxtinctions
appear as sharp drops in sutvivor-
ship. Because the sample lacks fam-
ilies still living today, the intensity o f
extinction evctlts is exaggerated to-
ward thc right.

1q30 SCIENCE, VOI,. 2 3 1

 1 ("

:I
I0

4,
(~arn~ies)F
- -- --

Lazarus
taxa

I.
- -

Dantan t a x a
-- -- -- - .
Fig 4. Ranges of brachlopod spe-
cles m about 15 meters of the sedl-
mentary sequence near the Creta-
ceous-Tert~ary boundary at Nye
Kl$v, Ileiimark (34) Sample p a n t s
are shown 111scdethe left scale.
; B-;

II
>. C

-5
L
-
0 1 0 1
g 7!
6
3
1 1 4 3 4 4 4 5 4 6 47

3 ~ 3 G 3 7 339e 4 0 4 1
+

2 5~
A
4 -1
I ,
2 8 2'3 30 31 32

I-
27
2
m

1, Maestrichtlan extinctions

broad geographic range survived longer, but Jablonski has also The two most commonly cited causes of mass extinction are
shown that this can break down at mass extinctions (15). Gastro- change in sea level and climatic deterioriation. Strong evidence exists
pods and bivalves with long-lived larvae and wide geographic for an association of extinction with times of lobering sea level, with
distributions have no higher survival rates at the K-T event than loss of habitable area on the marine continental shelves being
other groups. Also, ironically, highly species-rich groups do not suggested as the primary cause of extinction (17). Climatic deterio-
appear to have an advantage at this mass extinction. ration has been claimed as the'prbxiinal cause in a number of cases,
A few other tentative generalizations apply to one or more of the most notably in recent studies by Stanley of a regional extinction
big mass extinctions. 1,arge body size appears to be a disadvantage, event in the Pliocene (18).
at least among terrestrial animals. Tropical biotas seem to be prone The main problem with,these explanatidns is that both kinds of
to extinction, and biological groups that consistently show high change are common in earth history. There are many documented
background rates of speciation and extinction are most likely to be times of sea level lowering, some apparently caused by glaciation
eliminated at Inass extinction events. and others not, and many indications of long-term climatic change
In general, however, biological selectivity in extinction is poorly independent of sea level. There are also many extinction events.
studied and little understood. This is perhaps the most cn~cialarea Thus, to establish cause and effect is difficult and requires rigorous
for hture research, with rigorous comparisons among extinction assessment of probabilities in a complex time series. The K-T
events having the highest priority. A full understanding of selectivity extinction event was preceded by several million years of global
is the most promising route to discovering the environmental cooling and substantial sea-level lowering. But the association of
stresses that cause extinction. physical and biological events by itself does not prove cause and
effect.
Comet or asteroid scenarios have a long but not very distin-
guished history. As long as comets have been known, claims have
Causes of Extinction been made that their collisions with the earth have caused devasta-
Paleontology and geology have adhered to a strong conventional tion and destruction. Rut no empirical evidence was oEered until
wisdom, perhaps a dogma, since the days of Lyell and Darwin. The recently. In 1973, Urey presented a reasonable statistical argument,
basic tenets are as follows: (i) extinction is a gradual process driven based on ages of tektites and extinctions, for comets as a cause of
by the intricacies of interactions among species and between species several of the lesser extinction events in the Tertiary (19).
and their physical environment; (ii) because of the complexities of The first hard evidence canle with the report of Alvarez et al. in
multiple, independent causes, major extinction events are randomly 1980 of the iridium anomaly at the K-T boundary (1). The main
distributed in time; and (iii) all major extinctions are different, and a arguments in favor of the Alvarez scenario are embodied in interpre-
search for a single, unif)ing cause is futile. It is implicit in the tations of iridium enrichment (1,20), osmium isotope ratios (21),
foregoing that the causes of extinction are not to be found in the shocked quartz (22), and spherules interpreted to be microtektites
alien world of cosmic collisions. 1,yell put it this way in 1833: (23). The case fbr impact is strong but has its detractors (24).
The more important question in the present context is whether
111 our attempt to unravel these difficult clucstions, we shall [restrict]
ourselves to the lu~ownor possible operations of existing causes; feeling the impact caused the terminal Cretaceous extinctions. This, again,
assured that we have not yet exhausted the resources which the study of the comes down to questions of the probability of co-occurrence of
present course of nature (nay provide, slid therefore, that we are not events. Mass extinctions of the magnitude of the K-T event are
authorized in the infalicy of our science, to recur to extraordinary agents. We relatively rare in Phanero7mic time (perhaps five in 600 million
shall adhere to this plau . . . because . . . history informs us that this method
has always put geologists on the road that leads to truth (16, pp. 4 and 5).
years). Similarly, collisions with 10-kilometer objects are also rare:
perhaps ten such events in the last 600 million years (25). I t should
It is not surprising that the hypothesis of mass extinction by be simple to compute the appropriate probabilities if the timing of
meteorite impact is anathema to many paleontologists and geolo- both could be tied down precisely, but the problem is complex
gists. because of uncertainties in the actual timing of the terminal Creta-
28 MARCH 1986 ARTICLES 1531
Fig. 5. Rmges of planktonic forami-
niferal species from Deep Sea Drill-
ing I'roject site 253 in the southcrn
Indian Ocean (3.5). Wavy lincs at
three levels indicate hiatuses in thc
sedimentary record. Sample points
arc shown inside thc lcft scale.

ceous extinctions on a global basis. If all extinctions are assumed to to make the extinctio~~-by-impact proposals plausible, with or
coincide with those in the micro plank to^^ at Gubbio, Italy, the without periodicity. By the same token, periodicity need not require
probability of chance co-occurrence is negligible (26), but this impact by a large body. The recent success in relating Milankovich
assumption is difficult to prove. This uncertainty, more than cycles to the history of the younger terrestrial glaciations demo11-
anything else, has led competent researchers to conflicting conclu- strates that solar system processes call have recognizable effects on
sions on the likelihood that the impact of a large body caused a mass the earth (32). Thus, periodicity and impact should be viewed as
extinction. separate, altllough possibly linked, phenomena.
Broadly based studies of impact craters and of geophysical and
geochemical indicators of impact are needed before it is possible to
construct a hller chronology of the earth's impact history for
comparison with the biological record. Only then can the uncertair- Evolutionary Significance of Extinction
ties inherent in the interpretation of a single event be accommodat- Because the half-life of a biological species in geologic time is very
ed. T o date, there are reports of five impact-extinction pairs other short (generally less than 10 million years), turnover rates are high.
than the K-T event, but each raises substantial questions. All five are Extinction must therefore be important to the total evolutionary
based on evidence of iridium enrichment, and some have other process; to ignore it would be as inappropriate as for a population
features suggesting extraterrestrial input (27). It is too early to say biologist to ignore mortality or a sedimentologist to ignore erosion.
definitely whether impacts of large bodies caused these extinctions. It has been conventional t o view extinction as a constructive force
It may even turn out that impacts cause mass extinctions but only by which less well-adapted organisms are eliminated, leading to
when biological systems are already strained by other kinds of improvement in the mean adaptive level of the total biota. For
environmental stress, such as a change in sea level or climate. example, it has been assumed that Cretaceous mammals were better
adapted than the large reptiles. In general, however, it has been
impossible to verify the constructive aspect of extinction.
It may be that extinction, although selective, is not constructive. If
Periodicity of Extinction mass extinctions are the result of environmental stresses so rare as to
According to a number of analyses, major extinction events are bc beyond the "experience" of the organisms, extinction may be just
regularly spaced in geologic time ( I 1, 28, 29)-a further departme a matter of the chance susceptibility of the organisms to these rare
from the conventional wisdom of Lycll. The periods claimed range stresses. C~nsider,as a I~ypotheticalexlunple, the effects that large
from 26 to 32 million years, and most studies have been limited to doses of ionizing radiation would have 011 present-day biota. In
the last 250 million years of the l'hanerozoic, which permits terrestrial habitats, one can imagine radiation levels that would kill
relatively high resolution. If periodicity can be established, a com- all exposed mamnlals but have negligible effects on most insects and
mon cause is virtually required. plants. The result would be a highly selective extinction, but one
After Sepkoski and I proposed a 26-million-pear stationary extinc- having no constructive effect in terms of the general success of
tion periodicity ( I I ) , several astrophysical explanations were pro- organisms in normal times. Only if such crises were common in
posed (30). Most suggested that perturbations of comets in the Oort geologic time would the evolutionary system be able to cope
Cloud raise the probability of comet impact on the earth. Thus, a through natural selection.
link is suggested benveen the claimed periodicity of extinction and This model provides for profound effects on evolving systems, but
the impact theory of Alvarez et al. (1). This link has been supported the effects are not constructive in the usual Darwinian sense. It is
by two studies suggesting that impact craters on the earth show a conceivable that something like this model was operating among
periodicity compatible with that for extinction (29, 31). land vertebrates in the late Cretaceous. Mammals and dinosaurs had
Periodicity and extinction by impact are not necessarily linked, coexisted for more than 100 million years. Aker the large reptiles
however. The flux of large comets and asteroids has been sufficient became extinct, mammals ut-tdenvent an explosive evolutionary
SCIENCE, VOL. 233
radiation made possible by the absence of the dinosaurs. Rut the C. Lycll, Principles of Geology (JohnMurray, London, 1833),vol. 3.
N. D. Newell, Geol. Soc.Am. Spec. Pap. 89 (1967),p. 63; T . J . M. Schopf,J. Geol.
dinosaurs may not have done anything "wrong" in a Darwinian 82, 129 (1974);D. S. SimbcrloE, ibid., p. 267.
sense. S. M. Stanley, Geolvgy 12, 205 (1984); -and L. D. Campbell, Nature
(Londun) 293, 457 (1981).
The nonconstnlctive role of extinction is only hypothesis, just as H. C. Urcy, Nature (London) 242, 32 (1973).
the conventional Darwinian role is hypothesis. With the recent W . Alvarez, L. W . Alvarez, F . Asaro, H. V . Michel, Science 223, 1183 (1984).
J .M. Luck and K. K. Turekian, ibid. 222, 613 (1983).
renewal of interest in extinction, thanks to new theories of mass B. F. Bohor, E. E. Foord, P. J .Modrcski, D. M . Triplehorn,ibid. 224,867 (1984).
extinction, we may expect that this and other significant questions J . Smit and G. Klaver, Nature (Londun) 292, 47 (1981);A. Montanari et al.,
Geology 11,668 (1983).
can be investigated fruitfully. C. B. Officerand C. L. Drake, Science 227, 1161 (1985).
E. M. Shoemaker,Annu. Rev. Earth Planet. Sci. 11, 461 (1983).
L. W . Alvarez, Experimental Evidence that an Asteroid Impact Led to the Extinction of
REFERENCES AND NOTES
Many Species 65 Millwn Yea%& (Publ. LBL-14809, Lawrcncc Bcrkcley Labora-
~ , 1982). -
tory: ~ i r k e l eCA,
1. L. W . Alvarcz, W . Alvarez, F . Asaro, H. V . Michcl, Science 208, 1095 (1980). Terminal Eocene: W . Alvarez, F . Asaro, H. V . Michel, L. W .Alvarcz, Science 216,
2. G. G. Simpson, Evolution 6 , 342 (1952). 886 (1982); R Ganapathy, ibid., p. 885 Middle-upper Jurassic boundary: W .
3. D. M. Raup, Paleobwlop~y2, 279 (1976). Rrochwicz-Lewinski et al.. Int. Geol. Cwelatwn Pro~~r.
- - - -- - - - 199 (abstract,Mav 1985).
4 T . J M. Schopf, in GeologicalImplicatwns of Impacts of Lave Asteroidr and C;omets on Permian-Triassic boundaj: Y.-Y. Sun et al., in ~&elopmen&in Geoscienie (Acad.
the Earth, L. T . Silver and P. H. Schuln, Eds. (Geological Society o f America, Sinica, Science Prcss, Beijing, 1984), p 235-245 Late Devonian: P. E. Playford,
Bouldcr, CO, 1982), pp. 415-422. D. J. McLaren, C. J . Orth, J . S. Gihore, W . D: Goodfcllow, Science 226, 437
5. J .J . Sepkoski, Jr., in Pattern and Process in the H k q ofLife, D. Jablonski and D. (1984).Early Cambrian: K. J . Hsu, in Pattm and Prom in the H k q $Life, D.
M. Raup, Eds. (Springer-Verlag,Bcrlin, in prcss). Jablonski and D. M. Raup, Eds. (Springer-Verlag,Berlin, in press).
6. Bascd on methods dcvelopcd by D. M. Ran [Science206,217 (1979)l. A. G Fischcr and M. A. Arthur, in Deep-Water Carbonate Environments, H . E.
7. J . J . Sepkoski, Jr., Milwaukee Contvib. Bid! Geol. 51, 1 (19821, with updates Cook and P. Enos (Societyo f Economic Paleontologist$ and Mineralogists,Tulsa,
through October 1984. O K , 1977), pp. 19-50.
. .
8. For morc on andvsis o f taxonomic cohorts. see D. M. Raup IPde~biohflY
(1978);ibid. 11, 42 (1985)l.
-. 4, 1 M. R. Rampino and R. B. Stothers, Nature ( L o d n ) 308, 709 (1984)
Companionstar: M. Davis, P. Hut, R. A. Mullet, ibid., p. 715;D. P. Whitmire and
9. L. Van Valcn, Evol. Themy 1, 1 (19'73);D. M. Raup, l'aleobiolony 1, 82 (1975). A. A. JacksonN, ibid.,p. 713. Galactic plane oscillation:R. D. Schwartz and P. B.
10. D. M. Ran Acta Geol. Hispanics 16 (No. l ) , 25 (1981). James,ibid., p. 712; (32). Planet X : D. P. Whitmirc and J .J. Matese, ibid. 313, 36
11. and!i. J . Sepkoski, Jr.,Proc.Natl.Acad.Sd. U.S.A. 81,801 (1984);Science (1985).
231. 833 11986). W . Alvarcz and R. A. Muller, ibid. 308, 718 (1984).
12. 6: ~ablonski,Pahobwlogy 9, 315 (1983);inllynamiw ofExtinction, D. Elliott, Ed. J .Imbrie and J . 2. Imbrie, Science 207, 943 (1980).
(Wiley, Ncw York, in press). A. Hoffmanand J . A. Kitchell, Paleobiology 10, 9 (1984).
13. K. J . Hsii, in Pattern and Process in the Histmy ofL@, D. Jablonskiand D. M. Raup, F . Surlyk and M. B. Johansen,Science 223, 1177 (1984).
Eds. (Springer-Verlag,Berlin, in press); A. Hallam, Nature (Londun) 308, 686 G. Keller, Mar. Miwapaleontol. 7 , 463 (1982).
I 1 QRL;\
\-'--I'
Supported by National Aeronautics and Spacc Adnlinistration grant NAG 2-237. I
14. Based on P. W . Signor 111 and J . H. Lipps [in Geolofld Implications oflmpacts of thank the membcrs o f the 1985 Dahlcm Conferenceon Pattcrn and Process in thc
Lave Asteroidr and Comets on the Earth, L. T . Silver and P. H. Schultz, Eds. History o f Lifc for many stimulating discussions during the develo ment o f this
(Geological Society o f America, Bouldcr, CO, 1982), pp. 291-2961. review. Also, I thank D. Jablonski,W . S. McKerrow, and J . J. Sep!.oski, Jr., for
15. D. Jablonski,Science 231, 129 (1986). helpful comments on the manuscript.

.
AAAS-Newcomb Cleveland Prize
To Be Awarded for an Article or a Report Published in Science

The AAAS-Newcomb Cleveland Prize is awarded to the nominate papers appearing in the Reports or Articles sections.
author of an outstanding paper published in Science. The value of Nominations must be typed, and the following information
the prize is $5000; the winner also receives a bronze medal. The provided: the title of the paper, issue in which it was published,
current competition period begins with the 3 January 1986 issue author's name, and a brief statement of justification for nomina-
and ends with the issue of 29 May 1987. tion. Nominations should be submitted to the AAAS-Newcomb
Reports and Articles that include original research data, theo- Cleveland Prize, AAAS, 1333 H Street, NW, Washington, DC
ries, or syntheses and are fimdamental contributions to basic 20005, and must be received on or before 30 June 1987. Final
knowledge or technical achievements of far-reaching conse- selection will rest with a panel of distinguished scientists appoint-
quence are eligible for consideration for the prize. The paper ed by the editor of Science.
must be a first-time publication of the author's own work. The award will be presented at a ceremony preceding the
Reference to pertinent earlier work by the author may be President's Public Lecture at the 1988 AAAS annual meeting to
included to give perspective. be held in Boston. In cases of multiple authorship, the prize will
Throughout the competition period, readers are invited to be divided equally between or among the authors.

28 MARCH 1986 ARTICLES 1533