DOI: 10.1126/science.1173073 
, 733 (2009); 325Science
  et al.Kaustuv Roy,
Bivalves
Phylogenetic Conservatism of Extinctions in Marine
 www.sciencemag.org (this information is current as of August 7, 2009 ):
The following resources related to this article are available online at
 http://www.sciencemag.org/cgi/content/full/325/5941/733
version of this article at: 
 including high-resolution figures, can be found in the onlineUpdated information and services,
 http://www.sciencemag.org/cgi/content/full/325/5941/733/DC1
 can be found at: Supporting Online Material
 http://www.sciencemag.org/cgi/content/full/325/5941/733#otherarticles
, 21 of which can be accessed for free: cites 34 articlesThis article 
 http://www.sciencemag.org/cgi/collection/evolution
Evolution 
: subject collectionsThis article appears in the following 
 http://www.sciencemag.org/about/permissions.dtl
 in whole or in part can be found at: this article
permission to reproduce of this article or about obtaining reprintsInformation about obtaining 
registered trademark of AAAS. 
 is aScience2009 by the American Association for the Advancement of Science; all rights reserved. The title 
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. 
 (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m
 
(10 to 70 Tg/year) and top-down (140 to 910
Tg/year) estimates of global SOA production
(30). Nevertheless, IEPOX is expected to undergo
hundreds of collisions with aerosol surfaces before
reacting with OH, and its detection in the at-
mosphere (fig. S8) suggests that a complex suite
of conditions likely controls its uptake to aero-
sols (e.g., the pH and chemical composition of
aerosol). Furthermore, iSOA formation may
depend on the unquantified differences in the
yields and uptake characteristics of the b- and
d-IEPOX. Quantitative understanding of these
complex interactions is required to assess the ef-
fect of this chemistry on the overall SOA abun-
dance and its associated impacts [e.g., cloud
condensation nuclei (31)].
The efficient formation of dihydroxyepoxides,
a previously unknown class of gas-phase com-
pounds, addresses many of the issues currently
being debated about isoprene chemistry. Because
their formation is accompanied by the reformation
of OH, this chemistry contributes to the remark-
able stability of HOx in remote regions of the
troposphere subjected to high isoprene emissions.
The formation of IEPOX also provides a gas-
phase precursor for the iSOA formation. Further
investigation of the multiphase chemistry of
IEPOX is needed to elucidate the complex
interaction between emissions from the biosphere
and atmospheric composition (32, 33). In partic-
ular, the development of a proper chemical de-
scription of these interactions is essential for
assessing the sensitivity of this chemistry to
changes in isoprene emissions caused by envi-
ronmental changes (e.g., climate change and de-
forestation) and to the further development of
anthropogenic activities and the accompanying
NOx emissions in these regions.
References and Notes
1. A. Guenther et al., Atmos. Chem. Phys. 6, 3181 (2006).
2. P. C. Harley, R. K. Monson, M. T. Lerdau, Oecologia 118,
109 (1999).
3. J. D. Fuentes et al., Bull. Am. Meteorol. Soc. 81, 1537
(2000).
4. T. N. Rosenstiel, M. J. Potosnak, K. L. Griffin, R. Fall,
R. K. Monson, Nature 421, 256 (2003).
5. C. Wiedinmyer, X. Tie, A. Guenther, R. Neilson,
C. Granier, Earth Interact. 10, 1 (2006).
6. R. von Kuhlmann, M. G. Lawrence, U. P?schl, P. J. Crutzen,
Atmos. Chem. Phys. 4, 1 (2004).
7. M. Claeys et al., Science 303, 1173 (2004).
8. P. Crutzen et al., Atmos. Environ. 34, 1161 (2000).
9. C. E. Reeves, S. A. Penkett, Chem. Rev. 103, 5199
(2003).
10. J. A. Thornton et al., J. Geophys. Res. 107, 17 (2002).
11. J. Lelieveld et al., Nature 452, 737 (2008).
12. X. Ren et al., J. Geophys. Res. 113, 5310 (2008).
13. Materials and methods are available as supporting
material on Science Online.
14. J. Peeters, T. L. Nguyen, L. Vereecken, Phys. Chem. Chem.
Phys., 11, 5935 (2009).
15. T. J. Dillon, J. N. Crowley, Atmos. Chem. Phys. 8, 4877
(2008).
16. E. R. Bell, F. F. Rust, W. E. Vaughan, J. Am. Chem. Soc.
72, 337 (1950).
17. W. Wang et al., Rapid Commun. Mass Spectrom. 19,
1343 (2005).
18. J. D. Surratt et al., J. Phys. Chem. A 110, 9665 (2006).
19. E. C. Minerath, M. J. Elrod, Environ. Sci. Technol. 43,
1386 (2009).
20. J. D. Crounse, K. A. McKinney, A. J. Kwan, P. O.
Wennberg, Anal. Chem. 78, 6726 (2006).
21. F. Paulot et al., Atmos. Chem. Phys. 9, 1479 (2009).
22. F. W. McLafferty, Science 214, 280 (1981).
23. L. Ruppert, K. H. Becker, Atmos. Environ. 34, 1529
(2000).
24. J. D. Surratt et al., J. Phys. Chem. A 112, 8345
(2008).
25. S. Gao et al., Environ. Sci. Technol. 38, 6582 (2004).
26. I. Bey et al., J. Geophys. Res. 106, 73 (2001).
27. J. D. Surratt et al., Environ. Sci. Technol. 41, 5363
(2007).
28. J. A. de Gouw et al., J. Geophys. Res. 110, D16305 (2005).
29. P. Tunved et al., Science 312, 261 (2006).
30. A. H. Goldstein, I. E. Galbally, Environ. Sci. Technol. 41,
1515 (2007).
31. V.-M. Kerminen, H. Lihavainen, M. Komppula, Y. Viisanen,
M. Kulmala, Geophys. Res. Lett. 32, L14803 (2005).
32. F. W. Went, Nature 187, 641 (1960).
33. M. O. Andreae, P. J. Crutzen, Science 276, 1052 (1997).
34. We thank X. Levine, H. O. T. Pye, and the Harvard GEOS
CHEM team (Daniel J. Jacob, principal investigator)
for their help in setting up the GEOS-CHEM model;
A. J. Kwan, A. W. Chan, P. S. Chhabra, and N. Eddingsaas
for experimental assistance; J. D. Surratt for providing
the speciation of the SOA resulting from BEPOX
reactive uptake; and J. Lane, I. Maxwell-Cameron, and
S. J?rgensen for helpful discussions regarding the
quantum calculations. F.P. was partially supported by the
William and Sonya Davidow fellowship. J.D.C. thanks
the EPA Science to Achieve Results (STAR) Fellowship
Program (FP916334012) for providing partial support.
The mass spectrometer used in this study was purchased
as part of a major research instrumentation grant from
the National Science Foundation (ATM-0619783).
Assembly and testing of the CIMS instrument was
supported by the Davidow Discovery Fund. The numerical
simulations for this research were performed on Caltech?s
Division of Geological and Planetary Sciences Dell
Cluster. This work was supported by the Office of
Science (Biological and Environmental Research), U.S.
Department of Energy grant DE-FG02-05ER63983,
U.S. Environmental Protection Agency STAR agreement
RD-833749, and the Marsden Fund administrated by
the Royal Society of New Zealand. The TC4 and ARCTAS
campaigns were supported by NASA grants NNX07AL33G
and NNX08AD29G. This work has not been formally
reviewed by the EPA. The views expressed in this document
are solely those of the authors, and the EPA does not
endorse any products or commercial services mentioned in
this publication.
Supporting Online Material
www.sciencemag.org/cgi/content/full/325/5941/730/DC1
Materials and Methods
Figs. S1 to S9
Tables S1 to S9
References
2 March 2009; accepted 24 June 2009
10.1126/science.1172910
Phylogenetic Conservatism of
Extinctions in Marine Bivalves
Kaustuv Roy,1* Gene Hunt,2 David Jablonski3
Evolutionary histories of species and lineages can influence their vulnerabilities to extinction,
but the importance of this effect remains poorly explored for extinctions in the geologic past. When
analyzed using a standardized taxonomy within a phylogenetic framework, extinction rates of
marine bivalves estimated from the fossil record for the last ~200 million years show conservatism
at multiple levels of evolutionary divergence, both within individual families and among related
families. The strength of such phylogenetic clustering varies over time and is influenced by
earlier extinction history, especially by the demise of volatile taxa in the end-Cretaceous mass
extinction. Analyses of the evolutionary roles of ancient extinctions and predictive models of
vulnerability of taxa to future natural and anthropogenic stressors should take phylogenetic
relationships and extinction history into account.
Groups of organisms differ in their vulner-ability to extinction (1?5), but the natureand magnitude of that variation is still
poorly quantified. Extinction risk of species and
lineages is determined by a variety of ecological
and life history traits (2), as well as emergent
properties such as geographic range (5?8). Many
of these extinction-related traits are phylogeneti-
cally conserved, suggesting that extinctions should
be phylogenetically clustered: Taxa in some clades
should be consistently more extinction-prone than
others (3, 9, 10). Consistent with this idea, current
extinction risk and documented anthropogenic
extinctions are nonrandomly distributed among
vertebrate lineages (9, 11?15), but whether such
phylogenetic selectivity holds in general, includ-
ing for extinctions in the geologic past, remains
poorly known. In this study, we used theMesozoic-
Cenozoic fossil record of marine bivalves, in con-
junction with molecular phylogenies, to test for
phylogenetic clustering of extinctions within and
among bivalve families and how this clustering
varies over time.
The fossil record of marine bivalves preserves
a rich history of past extinctions, and although
this record is not free of taphonomic biases, such
biases are increasingly well understood (16, 17).
We used a taxonomically standardized database
1Section of Ecology, Behavior and Evolution, University of
California San Diego, La Jolla, CA 92093?0116, USA.
2Department of Paleobiology, National Museum of Natural
History, Smithsonian Institution, Washington, DC 20013?
7012, USA. 3Department of Geophysical Sciences, Uni-
versity of Chicago, 5734 South Ellis Avenue, Chicago, IL
60637, USA.
*To whom correspondence should be addressed. E-mail:
kroy@ucsd.edu
www.sciencemag.org SCIENCE VOL 325 7 AUGUST 2009 733
REPORTS
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m
 
of stratigraphic ranges of marine bivalves (18)
to calculate background extinction rates (i.e., for
times other than the massive end-Cretaceous event)
of 1678 genera and subgenera (hereafter termed
genera) over the last ~200 million years (Jurassic
to Pleistocene). Genera are the preferred units
of large-scale paleontological analyses because,
relative to species, their taxonomy is better stan-
dardized and more stable, and their fossil record
is far more complete and more robust to tapho-
nomic biases (19, 20). Furthermore, comparative
analyses indicate that morphologically defined
molluscan genera generally reflect the topologies
of molecular phylogenies (21). Taxonomic stan-
dardization is clearly a prerequisite for any quan-
titative analysis of extinction rates, and the data
we used were subjected to extensive revisions
and standardization (18, 19, 22). We used the
family level for analyses of phylogenetic cluster-
ing of extinction rates because families provide
the necessary balance between adequate sample
size and phylogenetic resolution. In general, fam-
ilies of marine bivalves have proven to be robust
taxonomic units, and recent molecular phyloge-
nies suggest that none of the major families of
marine bivalves are blatantly polyphyletic (23).
Although some bivalve families may prove to
be paraphyletic when a more complete molecular
phylogeny of the group is available, for our analy-
ses, paraphyly is more likely to add noise than to
produce artifactual trends.
If extinctions of bivalve genera were random
with respect to family membership, then an index
of taxonomic clustering for individual extinction
events [RCL (18)] should not differ systematically
from zero across a time series of such events.
Positive values of RCL indicate more (and nega-
tive values less) clustering than random extinc-
tion (18). Of the 26 standard time intervals in our
data (18) for which RCL could be reliably calcu-
lated, 21 (81%) have positive values (Fig. 1), a
result that is highly unlikely under a model of
phylogenetically random extinctions (P = 0.002,
exact binomial test). When RCL for individual time
intervals is compared with the null distribution for
that interval (18), 8 out of the 26 intervals show
significantly positive clustering (i.e., the observed
RCL is at least as great as the upper 95% confi-
dence limit), and no interval is significantly less
clustered than random (Fig. 1). These eight in-
tervals include the end-Cretaceous event, the only
major mass extinction in our data. A ninth interval,
the Campanian, is marginally significant. Thus
extinctions of marine bivalves over the last 200
million years show a general pattern of phyloge-
netic conservatism within families, both during
background and mass extinctions, but the strength
of such clustering varies over time, and not all
extinctions show significant clustering.
Our data also reveal that extinctionmagnitude
is not correlated with phylogenetic clustering.
The highest RCL value is associated with the end-
Cretaceous mass extinction (Fig. 2), but many
high-extinction intervals, such as the Late Eocene,
lack strong clustering (table S1), and the overall
relation between phylogenetic clustering and ex-
tinction intensity is not significant (Spearman rank
correlation, rs = 0.20, P = 0.34). Extinction rates
declined significantly over time, but this decline
was caused by culling of volatile clades rather
than by a decrease in extinction intensity within
individual clades. Overall, rates before the end-
Cretaceous extinction (excluding the Maastrichtian
stage, which ends with the mass extinction) are
higher than in the Cenozoic (median for Mesozoic
stages = 0.087, median for Cenozoic stages =
0.029; Wilcoxon rank sum test, W = 157, P =
0.054). For families well-represented in both the
Mesozoic and the Cenozoic, extinction rates do
not differ significantly before and after the end-
Cretaceous event (Wilcoxon paired signed rank
test, V = 61, P = 0.74, n = 16 families), and
families show high rank-order agreement between
Mesozoic and Cenozoic background rates of ex-
tinction (Fig. 3) (Spearman rank correlation, rs =
0.75,P= 0.0008). The lower Cenozoic extinction
rates instead result from preferential losses during
the end-Cretaceous extinction in families with
inherently high extinction rates (Fig. 4), so that
Cenozoic bivalve diversity was dominated by the
more extinction-resistant families. The three fam-
ilies with the highest Mesozoic background ex-
tinction rates went globally extinct at the end of
the Cretaceous, and other high-rate Mesozoic
clades (e.g., trigoniids and arcticids, both with
background rates >0.15, at least twice theMesozoic
median) (Fig. 4) were severely hit and have since
remained minor components of the bivalve fauna.
The only family for which background extinction
rates were much lower in the Cenozoic than the
Mesozoic is the Veneridae, which also suffered
major losses at the end of the Cretaceous. Thus,
the end-Cretaceous extinction had a filtering ef-
fect on lineage-specific extinction rates, removing
the most volatile families but not systematically
altering within-family extinction rates.
Extinction rates analyzed in conjunction with
recently published molecular phylogenies of living
bivalve families (18) also indicate phylogenetic
conservatism at deeper levels in the bivalve tree.
Extinction rates of closely related families are
significantly more similar to each other than is
expected by chance (Fig. 5) [P = 0.014 using a
permutation test (18); the maximum-likelihood
estimate of l, an index of phylogenetic depen-
dence (24), for within-family extinction rate is
0.84, a value within the range typically found for
ecological and morphological traits (24) and sig-
nificantly different from zero, P < 0.0004; see
(18)]. The phylogenetic signal remains significant
(P= 0.049) under an alternativemodel of character
change (18). Thus, the taxonomic and phylogenetic
analyses together suggest that extinction rates in
bivalves are conserved at multiple levels of evo-
lutionary divergence, within individual families
as well as among related families.
Stratigraphic ranges of taxa can be distorted
by preservational and sampling biases (17). Our
analyses hinged on differences between clades
rather than differences between temporal bins, so
the primary concern is not temporal variation in
the quality of the fossil record (25, 26), but sys-
tematic differences in preservation potential across
bivalve lineages. Extinction-rate estimates are ro-
bust to differences in shell composition (16), but
other variables known to influence preservation,
such as shell size, thickness, and preferred hab-
itats, may be important (17). To evaluate the
robustness of our results to preservational biases,
we repeated all analyses after omitting families
identified by Valentine et al. (17) as having low
-200 -150 -100 -50 0
-
0.
02
0.
00
0.
02
0.
04
0.
06
0.
08
0.
10
Age (Millions of years ago)
Jurassic Cretaceous Paleogene Neogene
To
ar
ci
an
B
ar
re
m
ia
n
A
pt
ia
n
A
lb
ia
n
Ce
no
m
an
ia
n
Ca
m
pa
ni
an
M
aa
st
ric
ht
ia
n
m
id
-M
io
ce
ne
Pl
io
ce
ne
Ph
yl
og
en
et
ic
 C
lu
st
er
in
g 
of
 E
xt
in
ct
io
ns
 (R
CL
)
Fig. 1. Temporal trend in phylogenetic clustering of extinctions (RCL). Shaded bars represent 95%
confidence intervals around the expected value of RCL (18). The intervals showing statistically significant
phylogenetic clustering of extinctions are labeled in bold; an additional interval, the Campanian, is
marginally significant. Only intervals with enough extinctions to analyze (eight or more) are plotted.
The dotted line indicates the value expected if extinctions were not phylogenetically clustered.
7 AUGUST 2009 VOL 325 SCIENCE www.sciencemag.org734
REPORTS
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m
 
preservation potential. The results were qualita-
tively unchanged (fig. S2). Another potential con-
cern is that the observed differences in extinction
rates between the Mesozoic and Cenozoic reflect
taxonomic oversplitting of some Cretaceous groups
relative to others. However, the families with dis-
proportionate extinction represent a large and ec-
ologically diverse assemblage (table S2), and
multiple lines of evidence (18) suggest that the
Mesozoic-Cenozoic difference is unlikely to sim-
ply reflect taxonomic practices. Phylogenetic clus-
tering also does not appear to be driven only by
extinctions associated with major environmental
changes. For example, the overall signal remains
highly significant, even when the Toarcian, Aptian,
and Cenomanian stages, all of which include
oceanic anoxic events (27), are excluded from
the analysis, along with the Maastrichtian mass
extinction (17 out of 22 extinctions with RCL >
0, P = 0.017, exact binomial test).
Taken together, our results show that extinc-
tion rates of marine bivalve genera tend to be
phylogenetically conserved, but the strength of
this effect varies over time and can be substan-
tially and permanently changed by a mass extinc-
tion. Lineage-level clustering of extinction rates,
as seen here, is expected to follow from phyloge-
netic conservatism of traits that correlate with ex-
tinction vulnerability (28), making some lineages
more prone to extinction than others (2, 9, 10).
As extinction-prone taxa are winnowed out, both
the rate of extinction and the associated phylo-
genetic clustering are expected to decrease. The
stronger phylogenetic clustering seen for the Cre-
taceous extinctions is likely to reflect the promi-
nence during this time of clades with volatile
dynamics (table S2). The demise of these high-
rate taxa at the end of the Cretaceous shifted the
overall distribution to lower values and also re-
duced the range of variation of within-family
extinction rates (Fig. 4). Such hardening of the
biota over evolutionary time has been hypothe-
sized before (29?31). Though we cannot reject
the alternative hypothesis that the decline in ex-
tinction rates fromMesozoic to Cenozoic is due to
a systematic decrease in extinction forcing mech-
anisms, we see no reason to assume that forcing
mechanisms became less intense, especially given
that the Cenozoic is characterized by dramatic
shifts in climate, occurring on multiple temporal
scales (32). The Mesozoic-Cenozoic differences
also do not reflect differences in statistical power,
because each of these eras has similar average
numbers of extinctions per time interval (table S1).
Other factors such as the nature of the ex-
tinction mechanism can also contribute to the
observed variations in phylogenetic clustering.
Different kinds of environmental stresses are
likely to cause extinctions that are selective with
respect to different traits, and we might expect
phylogenetic clustering of extinctions to track, in
part, the degree to which the relevant traits are
conserved over phylogeny. Extinction triggers that
disproportionately affect specific regions or envi-
ronments might also contribute to clustered ex-
tinctions in families with restricted distributions.
However, virtually all bivalve families are geo-
graphically and environmentally widespread (33),
and such spatial effects are weak in the end-
Cretaceous extinction (5, 6). Information on en-
vironmental drivers of past extinctions and their
spatial heterogeneity is currently insufficient to
permit more detailed exploration of these factors.
Irrespective of the underlying causes, the influence
of previous extinctions on both the magnitude of
extinction rates and the pattern of phylogenetic
conservatism suggests that attempts to understand
the biological basis for differential extinction vul-
nerabilities of clades should take into account their
past history of extinctions. These results also cor-
roborate a peculiarity of the end-Cretaceous mass
extinction (and perhaps of major extinctions in
general), where the phylogenetic pattern of ex-
tinction is consistent with preceding intervals, as
shown here, but the functional or ecological se-
lectivity is not (5, 29, 34).
Phylogenetic nonrandomness and the tempo-
ral decline in extinction rates documented here
are potentially problematic for calculating spe-
ciation and diversification rates from molecular
phylogenies because they violate the assumption
that extinction rates are stochastically constant over
time (35, 36), although recent work has started to
Fig. 3. Mesozoic versus Cenozoic back-
ground extinction rates for families well
represented in both intervals. The dashed
line represents equality in rates between
the two eras (y = x).
0.00 0.05 0.10 0.15
0.
00
0.
05
0.
10
0.
15
Cenozoic extinction rate
M
es
oz
oi
c 
ex
tin
ct
io
n 
ra
te
Arcidae
Astartidae
Cardiidae
Corbulidae
Gryphaeidae
Limidae
Lucinidae
Mytilidae
Nuculanidae
Nuculidae
Ostreidae
Pectinidae
Pinnidae
Pteriidae
Tellinidae
Veneridae
Fig. 2. Phylogenetic clustering of extinc-
tions (RCL) as a function of extinction rate,
for each time interval in Fig. 1. The relation
is not significant. The dotted line indicates
the value expected if extinctions were not
phylogenetically clustered.
0.02 0.05 0.10 0.20 0.50 1.00
0.
00
0.
02
0.
04
0.
06
0.
08
Extinction Rate by Interval
Ph
yl
og
en
et
ic
 C
lu
st
er
in
g 
of
 E
xt
in
ct
io
n 
by
 In
te
rv
al
Maastrichtian
0.0 0.1 0.2 0.3 0.4 0.5
0
2
4
6
8
10
12
0.0 0.1 0.2 0.3 0.4 0.5
0
5
10
15
20
25
30
Background Extinction Rate
N
o.
 o
f F
am
ili
es
Mesozoic
Cenozoic
N
o.
 o
f F
am
ili
es
Fig. 4. Distribution of background extinction rates
by family during the Mesozoic and Cenozoic eras.
Black bars indicate families that went extinct dur-
ing the end-Cretaceous mass extinction; gray bars
indicate surviving families.
www.sciencemag.org SCIENCE VOL 325 7 AUGUST 2009 735
REPORTS
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m
 
consider less restrictive models (37, 38). On the
other hand, if lineages or biotas tend to harden
over time, perhaps through the filter of mass ex-
tinctions, then the assumption of stochastically
constant extinction rates may be more reasonable
for later histories of lineages once the more vola-
tile taxa have been winnowed. Caution is needed,
however, because such hardening might operate at
multiple levels, ranging from the well-documented
decline in background extinction rates through the
Phanerozoic attributed to the culling of the more
volatile Paleozoic fauna in favor of the more re-
sistant Modern fauna (39, 40) to the species-level
selectivities seen during the marine extinction pulse
near the onset of Pleistocene glacial cycles (41).
Finally, our results combined with previous studies
(9, 12, 14, 28) imply that evolutionary histories of
individual lineages are important determinants of
extinction vulnerabilities of their constituent taxa,
under both natural and anthropogenic forcing. This
commonality suggests that more detailed studies
of phylogenetic selectivity of extinctions in the
geological past, and the traits involved, should
provide useful insights about the consequences
of extinctions unfolding today. For example, if
phylogenetic clustering is a general rule, then an-
thropogenic extinctions are likely to eliminate sub-
stantially more evolutionary history in the near
future than models based on random extinctions
of the same intensity would predict (15).
References and Notes
1. D. Jablonski, Philos. Trans. R. Soc. London Ser. B 344, 11
(1994).
2. M. L. McKinney, Annu. Rev. Ecol. Syst. 28, 495 (1997).
3. D. M. Raup, Extinction. Bad Genes or Bad Luck? (Norton,
New York, 1991).
4. S. C. Wang, A. M. Bush, Paleobiology 34, 434 (2008).
5. D. Jablonski, Paleobiology 31 (suppl.), 192 (2005).
6. D. Jablonski, Proc. Natl. Acad. Sci. U.S.A. 105, 11528
(2008).
7. J. L. Payne, S. Finnegan, Proc. Natl. Acad. Sci. U.S.A.
104, 10506 (2007).
8. M. G. Powell, Paleobiology 33, 530 (2007).
9. P. M. Bennett, I. P. F. Owens, D. Nussey, S. T. Garnett,
G. M. Crowley, in Phylogeny and Conservation, A. Purvis,
J. L. Gittleman, T. Brooks, Eds. (Cambridge Univ. Press,
Cambridge, 2005), pp. 317?336.
10. A. Purvis, M. Cardillo, R. Grenyer, B. Collen, in Phylogeny
and Conservation, A. Purvis, J. L. Gittleman, T. Brooks,
Eds. (Cambridge Univ. Press, Cambridge, 2005),
pp. 295?316.
11. G. J. Russell, T. M. Brooks, M. M. McKinney, C. G. Anderson,
Conserv. Biol. 12, 1365 (1998).
12. J. L. Lockwood et al., Conserv. Biol. 16, 1137 (2002).
13. A. Purvis, P.-M. Agapow, J. L. Gittleman, G. M. Mace,
Science 288, 328 (2000).
14. J. Bielby, A. A. Cunningham, A. Purvis, Anim. Conserv. 9,
135 (2006).
15. A. Purvis, Annu. Rev. Ecol. Evol. Syst. 39, 301
(2008).
16. S. M. Kidwell, Science 307, 914 (2005).
Solemyidae
Nuculidae
Neilonellidae
Yoldiidae
Nuculanidae
Glycymerididae
Arcidae
Noetiidae
Mytilidae
Isognomonidae
Pteriidae
Gryphaeidae
Ostreidae
Pinnidae
Limidae
Plicatulidae
Anomiidae
Pectinidae
Carditidae
Crassatellidae
Astartidae
Trigoniidae
Thyasiridae
Laternulidae
Thraciidae
Lucinidae
Pharidae
Hiatellidae
Gastrochaenidae
Lasaeidae
Sportellidae
Donacidae
Psammobiidae
Tellinidae
Semelidae
Cardiidae
Veneridae
Arcticidae
Chamidae
Mactridae
Ungulinidae
Teredinidae
Pholadidae
Corbulidae
Myidae
500 400 300 200 100 0
Extinction Rate
0.01
0.10
0.25
Paleozoic Mesozoic Cenozoic
Fig. 5. Mean extinction rates for living families of bivalves mapped
on the composite phylogeny used in this study (18). Only families
for which background extinction rates could be reliably estimated (18)
are shown. The dark bar along each branch denotes the known strat-
igraphic range of that taxon, omitting extinct stem groups. The size of the
symbol for each family is proportional to its extinction rate. Extinction
rates of closely related families are significantly more similar to each other
than is expected by chance.
7 AUGUST 2009 VOL 325 SCIENCE www.sciencemag.org736
REPORTS
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m
 
17. J. W. Valentine, D. Jablonski, S. M. Kidwell, K. Roy,
Proc. Natl. Acad. Sci. U.S.A. 103, 6599 (2006).
18. Materials and Methods are available as supporting
material on Science Online.
19. D. Jablonski, K. Roy, J. W. Valentine, Science 314, 102
(2006).
20. J. J. Sepkoski, Philos. Trans. R. Soc. London Ser. B 353,
315 (1998).
21. D. Jablonski, J. A. Finarelli, Proc. Natl. Acad. Sci. U.S.A.
106, 8262 (2009).
22. D. Jablonski, K. Roy, J. W. Valentine, R. M. Price,
P. S. Anderson, Science 300, 1133 (2003).
23. J. D. Taylor, S. T. Williams, E. A. Glover, P. Dyal, Zool. Scr.
36, 587 (2007).
24. R. P. Freckleton, P. H. Harvey, M. Pagel, Am. Nat. 160,
712 (2002).
25. S. E. Peters, Proc. Natl. Acad. Sci. U.S.A. 102, 12326
(2005).
26. A. B. Smith, A. J. McGowan, Palaeontology 50, 765 (2007).
27. R. D. Pancost et al., J. Geol. Soc. London 161, 353 (2004).
28. M. L. McKinney, Paleobiology 21, 300 (1995).
29. D. Jablonski, Proc. Natl. Acad. Sci. U.S.A. 98, 5393
(2001).
30. J. B. C. Jackson, in Extinction Rates, J. H. Lawton, R. M. May,
Eds. (Oxford Univ. Press, Oxford, 1995), pp. 45?54.
31. S. M. Stanley, Paleobiology 16, 401 (1990).
32. J. Zachos, M. Pagani, L. Sloan, E. Thomas, K. Billups,
Science 292, 686 (2001).
33. J. A. Crame, Paleobiology 26, 188 (2000).
34. S. M. Stanley, Paleobiology 33 (suppl.), 1 (2007).
35. S. Nee, Annu. Rev. Ecol. Syst. 37, 1 (2006).
36. S. Nee, E. C. Holmes, R. M. May, P. H. Harvey, Philos. Trans.
R. Soc. London Ser. B 344, 77 (1994).
37. D. L. Rabosky, I. J. Lovette, Proc. R. Soc. London Ser. B.
Biol. Sci. 275, 2363 (2008).
38. D. L. Rabosky, I. J. Lovette, Evolution 62, 1866 (2008).
39. J. J. Sepkoski Jr., in Global Events and Event Stratigraphy,
O. H. Walliser, Ed. (Springer, Berlin, 1996), pp. 35?51.
40. J. W. Valentine, in Causes of Evolution: A Paleontological
Perspective, R. M. Ross, W. D. Allmon, Eds. (Univ. of
Chicago Press, Chicago, 1990), pp. 128?150.
41. J. T. Smith, K. Roy, Paleobiology 32, 408 (2006).
42. We thank S. M. Kidwell and J. W. Valentine for discussions,
I. T?mkin and T. R. Waller for taxonomic advice, and two
anonymous reviewers for insightful comments. This work
was supported by a grant from NASA.
Supporting Online Material
www.sciencemag.org/cgi/content/full/325/5941/733/DC1
Materials and Methods
SOM Text
Figs. S1 to S5
Tables S1 and S2
References
4 March 2009; accepted 19 June 2009
10.1126/science.1173073
Genetic Properties of the Maize
Nested Association Mapping Population
Michael D. McMullen,1,2|| Stephen Kresovich,3|| Hector Sanchez Villeda,2* Peter Bradbury,1,3
Huihui Li,4,5,3 Qi Sun,6 Sherry Flint-Garcia,1,2 Jeffry Thornsberry,7 Charlotte Acharya,3
Christopher Bottoms,2 Patrick Brown,3 Chris Browne,1 Magen Eller,1 Kate Guill,1
Carlos Harjes,3? Dallas Kroon,3 Nick Lepak,1 Sharon E. Mitchell,3 Brooke Peterson,1
Gael Pressoir,3? Susan Romero,1 Marco Oropeza Rosas,8? Stella Salvo,1 Heather Yates,3
Mark Hanson,9 Elizabeth Jones,10 Stephen Smith,10 Jeffrey C. Glaubitz,11 Major Goodman,8
Doreen Ware,1,12 James B. Holland,1,8|| Edward S. Buckler1,3,13||
Maize genetic diversity has been used to understand the molecular basis of phenotypic variation
and to improve agricultural efficiency and sustainability. We crossed 25 diverse inbred maize lines
to the B73 reference line, capturing a total of 136,000 recombination events. Variation for
recombination frequencies was observed among families, influenced by local (cis) genetic
variation. We identified evidence for numerous minor single-locus effects but little two-locus
linkage disequilibrium or segregation distortion, which indicated a limited role for genes with large
effects and epistatic interactions on fitness. We observed excess residual heterozygosity in
pericentromeric regions, which suggested that selection in inbred lines has been less efficient in
these regions because of reduced recombination frequency. This implies that pericentromeric
regions may contribute disproportionally to heterosis.
The majority of phenotypic variation in nat-ural populations and agricultural plants andanimals is determined by quantitative ge-
netic traits (1). Maize (Zea mays L.) exhibits ex-
tensive molecular and phenotypic variation (2?4).
Understanding the genetic basis of quantitative
traits in maize is essential to predictive crop im-
provement. However, only slow progress has been
made in identifying the genes controlling quan-
titative agronomic traits because of limitations
in the scope of allelic diversity and resolution
in available genetic mapping resources. Linkage
mapping generally focuses on the construction
and analysis of large families from two inbred
lines to detect quantitative trait loci (QTLs) (5).
However, resolution of these QTLs can be poor
because of the limited number of recombination
events that occur during population development.
Association analysis takes advantage of historic
recombination from deep coalescent history as
linkage disequilibrium (LD) generally decays with-
in 2 kb (1, 6). However, because of the number
of single-nucleotide polymorphisms (SNPs) re-
quired and the confounding effects of population
structure, whole-genome association analysis can
be difficult in maize (4).
To provide a genetic resource for quantita-
tive trait analysis in maize, we have created the
nested association mapping (NAM) population.
NAM was constructed to enable high power and
high resolution through joint linkage-association
analysis, by capturing the best features of pre-
vious approaches (7, 8). The genetic structure
of the NAM population is a reference design
of 25 families of 200 recombinant inbred lines
(RILs) per family (fig. S1). The inbred B73 was
chosen as the reference inbred line because of its
use for the public physical map (9) and for the
Maize Sequencing Project (www.maizesequence.
org). The other 25 parents [named the 25 diverse
lines (25DL)] maximize the genetic diversity of
the RIL families (8, 10), independent of any spe-
cific phenotype. The lines were chosen to repre-
sent the diversity of maize?more than half are
tropical in origin, nine are temperate lines, two
are sweet corn lines (representing Northern Flint),
and one is a popcorn inbred line (fig. S2).
The NAM genetic map is a composite map
created with 4699 RILs combined across the 25
families, representing 1106 loci, with an average
marker density of one marker every 1.3 centi-
morgans (cM) (fig. S3 and table S1). The pro-
portion of SNP loci from the composite map
polymorphic in an individual family ranged from
63 to 74%. Among RILs, 48.7% of all marker geno-
types were inherited from B73, 47.6% were in-
herited from the 25DL parent, and 3.6% were
heterozygous, which suggests that they were
broadly representative of the parents and fall with-
in the expected range for S5 generation RILs. The
NAM population captured ~136,000 crossover
events, corresponding, on average, to three cross-
over events per gene. This allows genetic fac-
tors to be mapped to very specific regions of the
1United States Department of Agriculture?Agriculture Research
Service (USDA?ARS). 2Division of Plant Sciences, University of
Missouri, Columbia, MO 65211, USA. 3Institute for Genomic
Diversity, Cornell University, Ithaca, NY 14853, USA. 4School of
Mathematical Science, Beijing Normal University, Beijing
100875, China. 5Institute of Crop Science, Chinese Academy
of Agricultural Sciences, Beijing 100081, China. 6Computational
Biology Service Unit, Cornell University, Ithaca, NY 14853, USA.
7Northwest Missouri State University, Maryville, MO 64468,
USA. 8Department of Crop Science, North Carolina State
University, Raleigh, NC 27695, USA. 9Illumina Inc., San Diego,
CA 92121, USA. 10Pioneer Hi-Bred, Johnston, IA 50131, USA.
11Laboratory of Genetics, University of Wisconsin, Madison, WI
53706, USA. 12Cold Spring Harbor Laboratory, Cold Spring
Harbor, NY 11724, USA. 13Department of Plant Breeding and
Genetics, Cornell University, Ithaca, NY 14853, USA.
*Present address: International Maize and Wheat Improvement
Center (CIMMYT), kilometer 45, Carretera Mex-Veracruz, El Batan,
Texcoco, Mexico.
?Present address: Monsanto, Leesburg, GA 31763, USA.
?Present address: Fondation CHIBAS, 30 Rue Pacot, Port-au-
Prince, Haiti.
?Present address: Delta Pine/Monsanto, Post Office Box 194,
Scott, MS 38772, USA.
||To whom correspondence should be addressed. E-mail:
mcmullenm@missouri.edu (M.D.M.); sk20@cornell.edu (S.K.);
james_holland@ncsu.edu (J.B.H.); esb33@cornell.edu (E.S.B.)
www.sciencemag.org SCIENCE VOL 325 7 AUGUST 2009 737
REPORTS
 
o
n
 A
ug
us
t 7
, 2
00
9 
w
w
w
.s
ci
en
ce
m
ag
.o
rg
D
ow
nl
oa
de
d 
fro
m