SCIENCE sciencemag.orgGRAPHIC: N. DESAI/
SC
IENCE
By Peter WagnerD
iscussions about the quality of the
fossil record frequently focus on com-
pleteness—that is, the proportion of
species that existed for which scien-
tists have fossil samples. An equally
important aspect of quality is how
finely the ages of fossils and durations of
major evolutionary events can be resolved.
Paleontologists and other biologists typi-
cally date fossils using the general ages of
the chronostratigraphic units assigned to the
rock strata yielding the fossils. For example,
if a fossiliferous bed comes
from the Rhuddanian Gas-
works Sandstone, then the
ages assigned to the fossils
are usually 443.8 to 440.8
million years. On page 272
of this issue, Fan et al. ( 1 ) re-
port on their use of constrained optimization
(CONOP) ( 2 ) in biochronological analyses
of fossil-bearing layers containing 11,000+
marine invertebrate species from 3000+ sec-
tions of sedimentary rocks, which constrain
ages to a fraction of that range.
Their data, taken from the Geobiodiver-
sity Database (GBDB) ( 3 ), span the Paleo-
zoic through the Early Triassic. Although
GBDB data are limited (largely) to China,
the different continental plates that com-
pose modern China fortuitously span multi-
ple biogeographic realms throughout most
of the Paleozoic. CONOP aligns fossil-bear-
ing beds from different rock sections (e.g.,
spans of rocks from quarries, road cuts, and
natural erosion on hills and mountains) by
maximizing sets of species with first and
last appearances in the same order in as
many sections as possible. By integrating
these analyses with radiometric dates and
other geological data ( 4 ), the authors re-
solved the age of an average bed (and thus
the average first and last appearance times
of species) to within 26,000 years. This new
level of dating specificity is similar to mov-
ing from a system in which all people who
lived in the same century are considered
to be contemporaries to one in which only
people who lived during the same 6-monthperiod are deemed to be contemporaries.
Armed with exact first and last appear-
ance dates for thousands of species, the
authors then developed what are, as they
themselves stress, a set of fairly preliminary
analyses of Paleozoic diversity. The authors
deliberately eschew sophisticated analyses
of origination and extinction dynamics or
sampling dynamics used in other works
with much coarser temporal resolution
than that yielded by CONOP ( 5 , 6 ). This ap-
proach permits contrasting the basic pat-
terns with those that would be implied had
the authors chosen to “bin” these data intopredefined chronostratigraphic analyses, as
nearly all prior diversity studies have done.
In doing so, some noteworthy patterns
emerge. For example, despite failing to cor-
rect for the Signor-Lipps effect (imperfect
sampling that causes sudden extinctions to
look gradual or rapid radiations to look pro-
tracted) ( 7 ), CONOP results indicate rapid
declines in richness of marine invertebrate
species prior to the end-Ordovician and
end-Permian extinctions. Similarly, Fan et al.
found the rebound after the end-Ordovician
extinction to be much more rapid than prior
studies suggest. Furthermore, CONOP recon-
structions show patterns over hundreds of
thousands of years that prior studies, which
lumped together occurrences over millions of
years, cannot hope to reveal.
Analysis of the Great Ordovician Biodi-
versification Event (GOBE) ( 8 ) offered a dif-
ferent type of unexpected result. Fan et al.’s
study suggests that the GOBE began 20+
million years earlier than expected, during
the late Cambrian. Given that late Cambrian
sampling for many marine invertebrates is
relatively poor compared to their fossil re-
cords earlier in the Cambrian and in the Or-
dovician ( 9 ), one would predict emergence of
an opposite pattern: Any late Cambrian di-
versification should not have appeared until
the Ordovician.
The CONOP analyses of GBDB data have
implications for how organismal biologists
and molecular phylogeneticists perform
analyses of species, including fossils, in thefuture. Recent years have seen a prolifera-
tion of methods such as tip-data ( 10 ) and
fossilized birth-death models ( 11 ), which in-
clude fossilized anatomical data and modern
anatomical and molecular data. Accounting
for uncertainty in the true ages of first and
last appearances can strongly affect the
results of such analyses ( 12 ). Current ap-
proaches for dealing with this uncertainty
often assume that uncertainty is large; for
example, if a species’s first appearance is in
the Rhuddanian age, then all dates within
the Rhuddanian are equally probable first
appearances ( 13 ). However, as the new
study makes clear, biochro-
nological analyses will make
some dates within the Rhud-
danian much more probable
than others.
Of course, many extant
organisms do not fossilize as
well as shell-bearing marine invertebrates,
and their oldest fossil representatives often
are known from exceptional preservation
in the lagerstätte sedimentary deposits ( 14 ).
CONOP and related biochronology methods
( 15 ) require species from multiple beds in
multiple sections to infer ages for fossil occur-
rences. However, fossiliferous beds preserv-
ing typical assemblages of shell-containing
invertebrates usually occur above and below
the lagerstätte. Constraining the ages of those
beds will simultaneously constrain the ages
of the lagerstätte. With respect to important
evolutionary questions, the development of
new datasets and methods of the sort used
by Fan et al. will further evolutionary biology
as a whole, not just paleontology. jREFERENCES AND NOTES
1. J. -x. Fa n et al., Science 367 , 272 (2020).- P. M. Sadler, R. A. Cooper, in High-Resolution Stratigraphic
Approaches in Paleontology, P. Harries, Ed. (Plenum,
2003), vol. 21, pp. 49–94. - GBDB, http://www.geobiodiversity.com.
- P. M. Sadler, Bull. Geosci. 87 , 681 (2012).
- M. Fo o te et al., Proc. Biol. Sci. 285 , 20180122 (2018).
- M. Fo o te et al., J. Geol. Soc. London 176 , 1038 (2019).
7. P. W. Signor, J. H. Lipps, Spec. Pap. Geol. Soc. Am. 190 , 291
(1982). - T. Servais et al., Palaeogeogr. Palaeoclimatol. Palaeoecol.
294 , 99 (2010). - Y.-D. Zhang et al., Palaeoworld 28 , 1 (2019).
- F. Ronquist et al., Syst. Biol. 61 , 973 (2012).
- T. A. Heath, J. P. Huelsenbeck, T. Stadler, Proc. Natl. Acad.
Sci. U.S.A. 111 , E2957 (2014). - J. Barido-Sottani et al., Proc. Biol. Sci. 286 , 20190685
(2019). - D. Silvestro et al., Paleobiology 45 , 546 (2019).
- D. T. Ksepka et al., Syst. Biol. 64 , 853 (2015).
- J. Alroy, Paleobiology 20 , 191 (1994).
10.1126/science.aba4348
Refined ages of marine fossils clarify the timing of diversification and extinction events
Department of Earth and Atmospheric Sciences, and
School of Biological Sciences, University of Nebraska
Lincoln, Lincoln, NE 68588-0340, USA.
Email: [email protected]EVOLUTIONARY BIOLOGYHigh-resolution dating of Paleozoic fossils
“...constrained optimization results indicate rapid
declines in richness of marine invertebrate species prior
to the end-Ordovician and end-Permian extinctions.”
17 JANUARY 2020 • VOL 367 ISSUE 6475 249
Published by AAAS