Current Biology 23, 1889–1895, October 7, 2013 ª2013 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.cub.2013.07.

055

Report
Rates of Phenotypic and Genomic Evolution
during the Cambrian Explosion

Michael S.Y. Lee,1,2,* Julien Soubrier,1 molecular dating have typically greatly reduced the postulated
and Gregory D. Edgecombe3 cryptic Precambrian branches between phyla [4, 11, 12]. Many
1School of Earth and Environmental Sciences, University of other fundamental aspects of this event remain uncertain,
Adelaide, Adelaide, South Australia 5005, Australia notably the relative importance of environmental, genetic,
2Earth Sciences Section, South Australian Museum, North developmental, and ecological triggers [1–5]. Morphological
Terrace, Adelaide 5000, Australia evolution is also widely postulated to be very rapid during
3Department of Earth Sciences, The Natural History Museum, the Cambrian explosion [1–3]. A relatively short Precambrian
Cromwell Road, London SW7 5BD, UK prelude dictates this pattern, while even a lengthy ‘‘phyloge-
netic fuse’’ can be interpreted to be consistent with it: phyla
could have diverged deep in the Precambrian but still acquired
Summary many novelties simultaneously across the early Cambrian [8,
18] (though it is hard to interpret innovations shared by multi-
The near-simultaneous appearance of most modern animal ple phyla as evolving in the Cambrian under the second sce-
body plans (phyla) w530 million years ago during the nario). Darwin suggested that the sudden appearance of a
Cambrian explosion is strong evidence for a brief interval range of advanced animals in the Cambrian was difficult to
of rapid phenotypic and genetic innovation, yet the exact reconcile with gradual evolution by natural selection [2, 4], as
speed and nature of this grand adaptive radiation remain they would have required a lengthy period of Precambrian evo-
debated [1–12]. Crucially, rates of morphological evolution lution. This view has been echoed by subsequent workers
in the past (i.e., in ancestral lineages) can be inferred from arguing for a lengthy but poorly preserved prelude [6–8]. A
phenotypic differences among living organisms—just as much shorter cryptic period would be required if rates of evo-
molecular evolutionary rates in ancestral lineages can be lution in the late Precambrian and early Cambrian could be
inferred from genetic divergences [13]. We here employed demonstrated to be substantially elevated. However, precise
Bayesian [14] and maximum likelihood [15] phylogenetic estimates of rates of evolution during the Cambrian explosion
clock methods on an extensive anatomical [16] and genomic remain elusive, as the patchy stratigraphic record during this
[17] data set for arthropods, the most diverse phylum in the pivotal interval precludes direct paleontological estimates of
Cambrian and today. Assuming an Ediacaran origin for evolutionary rates.
arthropods, phenotypic evolution was w4 times faster, and Arthropods are the exemplar group for investigating ques-
molecular evolution w5.5 times faster, during the Cambrian tions about such macroevolutionary rates and patterns. They
explosion compared to all subsequent parts of the Phanero- are the most abundant and diverse phylum in the early Paleo-
zoic. These rapid evolutionary rates are robust to assump- zoic, have very complex preserved morphologies, and occu-
tions about the precise age of arthropods. Surprisingly, pied extensive morphospace by the middle Cambrian [1, 19,
these fast early rates do not change substantially even if 20]. However, rapid early expansion of morphospace does
the radiation of arthropods is compressed entirely into the not demonstrate faster early evolution. Steady evolutionary
Cambrian (w542 mega-annum [Ma]) or telescoped into the rates coupled with strong constraints would also produce
Cryogenian (w650 Ma). The fastest inferred rates are still this pattern: morphospace occupation would increase initially
consistent with evolution by natural selection and with before plateauing once constraints are reached, but lineages
data from living organisms, potentially resolving ‘‘Darwin’s could maintain their evolutionary rates even after mor-
dilemma.’’ However, evolution during the Cambrian explo- phospace is fully occupied (by moving around within this
sion was unusual (compared to the subsequent Phanero- permitted morphospace). Thus, although occupation of mor-
zoic) in that fast rates were present across many lineages. phospace and lineage-specific evolutionary rates are often
correlated, they represent independent concepts. Although
Results studies have quantified morphospace occupation in the
Cambrian [19, 20], no analyses have directly quantified the
The abrupt appearance of most modern animal body plans rate of anatomical change during the Cambrian explosion
(often ranked as phyla and classes) over half a billion years and compared it with subsequent evolution. Tracking changes
ago is one of the most important evolutionary events after in individual fossil lineages across stratigraphic sequences is
the origin of life [1–5]. This initial diversification of major meta- problematic, as most continuous sequences spanning sub-
zoan groups—termed the Cambrian explosion—is suggested stantial durations typically preserve only hard external parts,
by the fossil record to have occurred largely in the Terreneu- and the small number of absolute dates for the Cambrian
vian (w542–521 mega-annum [Ma]) [1–3], although molecular means that temporal resolution can be obtained to only within
analyses suggest a cryptic Precambrian interlude of up to a few million years [21]. Burgess Shale-type Lagerstätten pre-
several hundred million years [6–9]. An emerging consensus serving soft tissue are patchily distributed across time and
favors an intermediate interpretation closer to the first sce- space and still retain only a fraction of the anatomy observable
nario: trace fossil, microfossil, and biomolecular evidence from living animals; in arthropods, the most complete fossils
has slightly extended the stratigraphic record of some meta- preserve less than one-third of the anatomical characters
zoan groups [3, 4, 10], while recent studies employing used to infer evolutionary relationships among living forms
[22]. Similarly, there have been few attempts to quantify rates
of genomic evolution in the Cambrian: pioneering efforts
*Correspondence: [email protected] based on short sequences then available suggested elevated
Current Biology Vol 23 No 19
1890

rates [23, 24] but employed problematic methods that could [4, 11, 12] (i.e., 542–650 Ma). The first occurrence of many
have generated such patterns even on rate-constant data advanced arthropod clades in the Cambrian, as underscored
[25, 26]. by recent discoveries of several crustacean lineages [34]
We here simultaneously infer rates of phenotypic and (Table S1), implies that the major arthropod clades all diverged
genomic change in arthropods during the Cambrian explosion and evolved within a brief time window (w40 million years
and subsequent Phanerozoic, using a novel approach that [Myr]; Figure 1A). This compresses into a very narrow interval
exploits (1) the extensive phenotypic and genomic data avail- the extensive molecular (Figure 1B) and phenotypic (Figure 1C)
able for living arthropods, (2) the calibration information avail- changes that occurred on the branches leading to euchelicer-
able from the rich arthropod fossil record, and (3) adapting ates, myriapods, mandibulates, pancrustaceans, oligostra-
molecular clock methods for use on both genetic and morpho- cans, vericrustaceans, and miracrustaceans (crown ages all
logical data (see the Supplemental Experimental Procedures >500 Ma). Accordingly, rates of evolution in most early
online for detailed methods, and see Data Set S1 online for arthropod lineages are substantially faster than rates for the
data sets). Given DNA sequences from living taxa, and (ideally subsequent Phanerozoic (Figures 2A, 2D, 3A, and 3D). The
multiple) calibration points, there are numerous phylogenetic fastest molecular rates occur in the branches leading to
dating methods that can estimate evolutionary relationships Arthropoda and Pancrustacea (>103 average subsequent
and divergence dates. For each branch, such methods simul- rates); the fastest phenotypic rate occurs in Mandibulata
taneously infer both duration (length in time units) and change (>163). The average rate of phenotypic evolution in early
(length in molecular substitutions), thus directly computing Cambrian lineages (0.561% per million years [pMyr]) is w43
evolutionary rates. Notably, ‘‘relaxed clock’’ methods [13] the average rate in subsequent lineages (0.136% pMyr; Tables
can reveal how rates have changed with time by permitting S2 and S3; Figure S3). Similarly, the average Cambrian rate of
rates of molecular evolution to vary across branches, both in- molecular evolution (0.117% pMyr) is w5.53 the average
ternal (between two nodes, always extinct) and external (lead- subsequent rate (0.022% pMyr); the latter rate is highly con-
ing to a tip, which is typically extant). Such approaches have sistent with rates inferred for conservative nuclear genetic
been recently applied to morphological data [27, 28] but data in modern invertebrates [23, 24, 35].
have not been used to explicitly address the issue of past rates These rate estimates are robust to a wide range of root age
of evolutionary change. We here analyze an extensive data set assumptions (Table S3). Surprisingly, early rates increase only
of arthropods, consisting of 395 phenotypic characters [16] slightly under an extreme Cambrian explosion scenario, where
and 62 protein-coding genes [17, 29], with 20 calibration points maximum root age is tightened to 542 Ma and basal arthropod
taken from the fossil record (Table S1). The combined data divergences are compressed into <10 Myr within the lower-
were analyzed using Bayesian relaxed-clock methods in most Cambrian. Early rates also remain highly elevated if
BEASTMC3 [14] to simultaneously estimate tree topology, root age is relaxed to 650 Ma, resulting in a ‘‘long fuse,’’ where
divergence dates, and morphological and molecular evolu- basal arthropod divergences are telescoped across >100 Myr.
tionary rates across branches (and thus across time). Pheno- Panarthropods are unlikely to be younger than 542 Ma (when
typic data were allowed to influence inferred tree topology there are already putative arthropod trace fossils; Table S1)
and divergence dates, as such data, even in combination or older than 650 Ma (which exceeds many recent molecular
with genomic data, can be crucial for reconstructing topology estimates for the age of panarthropods [4, 11, 12] and the
and thus relative and absolute divergence times [30]. Unlike age of the oldest fossil evidence not only for arthropods but
other recent studies [4, 9, 11, 12], which estimated divergence for any animals [3, 4, 10]). Placing a soft bound on root age re-
dates, our study explicitly aimed to estimate how rates of evo- veals that, in order to reduce early Cambrian evolutionary rates
lution varied through time. We also tested the sensitivity of to subsequent levels, one has to assume that panarthropods
these results to analytical methods (by using maximum likeli- originated w940 Ma (Table S3).
hood to infer tree topology [31] as well as divergence dates These rate patterns also persist but are slightly dampened
and evolutionary rates [15]), to data (by analyzing the molecu- (Table S3) if the phylogeny is constrained to conform to an
lar data alone), and to topology (by enforcing novel arthropod alternative recent phylogeny of arthropods in which pancrus-
clades recently proposed [32]). We also test robustness to taceans form two primary clades [32]. The pattern of elevated
calibration assumptions by randomly deleting internal calibra- early rates also holds in analyses in which tree topology and
tions, as well as employing a range of root age constraints for branch lengths are based solely on molecular data (Figure 4),
panarthropods (hard bounds of <542 to <700 Ma, as well as if the internal calibrations are greatly relaxed (e.g., by deleting
soft bounds) that encompass the majority of recent age esti- 50% of the calibrations) and if rates are compared across a
mates [4, 9, 11, 12]. consistent interval (50 Myr sliding window; Table S3B).
The topology of the dated arthropod trees (Figures 1 and 2) Maximum likelihood analysis (Supplemental Experimental
is broadly consistent with the trees obtained from the same Procedures SI_12), using solely molecular data to first infer
nuclear genes alone across a broader sample of exemplars only topology (using RAxML) and then dating this topology
[9, 12, 17, 29, 32] as well as from other extensive molecular (using r8s), retrieves a very similar chronogram (Figure S2C)
and combined data sets [11, 16, 33]. The relative amounts of and very similar patterns and rates of evolution (Figures 2B,
molecular and phenotypic evolutionary change across the 2E, 3B, and 3E; Table S3); average rates of phenotypic and
tree (Figures 1B and 1C) are also consistent with other ana- molecular evolution after the early Cambrian are within 5% of
lyses (e.g., [16, 17, 29]): most basal branches of arthropods Bayesian estimates. Phenotypic traits diverge at 1.17% pMyr
exhibit substantial molecular and phenotypic change (many in the early Cambrian, w83 the average rate for the remainder
synapomorphies). However, a striking pattern involves the of the Phanerozoic (0.139% pMyr); molecular divergence is
time frame for these basal arthropod divergences, and atten- 0.204% pMyr, w93 the subsequent rate (0.021%).
dant rates of molecular and phenotypic evolution, under all All of these sensitivity analyses are discussed in Supple-
root age constraints broadly consistent with the fossil record mental Experimental Procedures SI_1–SI_12, along with cor-
[1–5, 10] and the majority of recent molecular studies rections for other potential biases: oversampling of basally
Evolutionary Rates during the Cambrian Explosion
1891

A Branches proportional to Time (million yrs) B

500 400 300 200 100 0
Tardigrada (Milnesium)
Tardigrada (Hypsibiinae)
Onychophora (Euperipatoides)
Onychophora (Peripadae)
Pycnogonida (Colossendeis) Branches proportional to
Pycnogonida (Ammotheidae)
Pycnogonida (Endeis) Pycnogonida Molecular Change
Xiphosura (Carcinoscorpius) Xiphosura (thick = Early Cambrian & latest Ediacaran)
Xiphosura (Limulus)
Acari (Amblyomma/Opilioacarus)
Opiliones (Leiobunum/Siro)
Solifugae (Eremocosta)
00.655 Ricinulei (Cryptocellus) Arachnida
Scorpiones (Scorpionidae)
Araneae (Mygalomorphae)
Uropygi (Masgoproctus)
Chilopoda (Scugera)
Chilopoda (Lithobius)
0.67
Chilopoda (Scolopendra)
Chilopoda (Craterosgmus)
Diplopoda (Polyxenus)
Diplopoda (Narceus)
Pauropoda (Pauropodinae) Myriapoda 3%
Symphyla (Hanseniella)
Symphyla (Scugerella)
0.81
Ostracoda (Cypridopsis)
Mystacocarida (Derocheilocaris)
0.81 Branchiura (Argulus)
Pentastomida (Armillifer/Railieella)Oligostraca
C Branches proportional to
Branchiopoda (Artemia)
Branchiopoda (Streptocephalidae) Morphological Change
0.88
0. Branchiopoda (Triops) (thick = Early Cambrian & latest Ediacaran)
Branchiopoda (Lynceus)
Branchiopoda (Daphnia)
0.88 Branchiopoda (Limnadia)
Cirripedia (Semibalanus)
Copepoda (Calanoida) Vericrustacea
Malacostraca (Nebalia)
Malacostraca (Stomatopoda)
Malacostraca (Decapoda)
0.76 Malacostraca (Oniscidea)
Cephalocarida (Hutchinsoniella)
Remipedia (Speleonectes/Remipedia) Xenocarida
Collembola (Tomocerus)
Diplura (Campodeidae)
Diplura (Japygidae)
Archaeognatha (Meinertellidae)
Archaeognatha (Machilidae) 8%
Zygentoma (Lepismadae)
Ephemeroptera (Ephemerella/Callibaes)
Diptera (Panopoida)
Posterior Probability = 1.0 Orthoptera (Acheta/Locusta) Hexapoda
unless otherwise indicated Blaodea (Periplaneta)

Figure 1. Arthropod Lineages during the Cambrian Explosion Are Short in Duration yet Undergo Large Amounts of Phenotypic and Molecular Change
Bayesian (BEASTMC3) analysis of 395 phenotypic characters (Supplemental Experimental Procedures SI_2) and 62 nuclear genes (Supplemental Experi-
mental Procedures SI_3). Branches and taxa in these three trees are in identical order and have the same color coding. Note the large amount of change
indicated by the long bold branches in (B) and (C), and the corresponding short durations of these branches in (A). For details of specimen illustrations,
see Supplemental Experimental Procedures SI_14.
(A) Tree with branch lengths and divergence dates in terms of time; pink shading highlights the period before the late Cambrian (>500 Ma), during which most
high-level (‘‘phylum’’ and ‘‘class’’) diversity appears in the animal fossil record. Blue bars denote 95% highest posterior densities (HPDs) for divergences,
which are often large due to extreme rate heterogeneity.
(B) Tree with branch lengths proportional to molecular change; bold denotes branches with ages > 500 Ma (branch age = midpoint of upper and lower node).
(C) Tree with branch lengths proportional to phenotypic change; bold denotes branches with ages > 500 Ma.

changing traits (and undersampling of unique derived traits) in [35]. However, the results offer less support for scenarios
the morphological data, saturation and/or underparameteriza- that do not necessarily entail elevated rates across both
tion of molecular sequence data, and parameter uncertainty morphology and housekeeping genes: for example, that the
associated with chronologically short Cambrian branches. explosion was driven largely by changes in gene regulation
(e.g., [18, 36]) or by evolutionarily labile developmental sys-
Discussion tems that congealed after the Cambrian (e.g., [37]). Such sce-
narios do not predict that genes involved in basic metabolic
This study provides the first explicit estimates of rates of both and cellular processes common to all panarthropods should
phenotypic and molecular evolution during the Cambrian have faster evolutionary rates in earlier panarthropod lineages.
explosion, permitting illuminating comparisons with later The elevated Cambrian rates of phenotypic evolution are
Phanerozoic rates. The 4- and 5.5-fold increases in phenotypic unlikely to be an artifact of oversampling of characters chang-
and molecular evolutionary rates respectively provide quanti- ing on early arthropod branches, for two reasons. First, the
tative support for the widespread view that evolutionary rates data set explicitly surveyed phenotypic characters changing
were elevated during the Cambrian explosion. Notably, both at all levels in the phylogeny, not only those occurring on early,
the patterns and magnitude of the rate elevations are strikingly basal branches (see Supplemental Experimental Procedures
similar for two very different suites of characters (one set SI_1). Second, the molecular data exhibit an almost identical
dominated by anatomical traits, the other consisting of pro- Cambrian rate burst yet cannot suffer such ascertainment
tein-coding nuclear genes). This is most consistent with pro- bias (because gene sequencing samples every single nucleo-
posed drivers that could directly affect these two disparate tide in a gene region, not just sites changing on particular
sets of features (and by implication most other systems), branches). The congruence of the morphological pattern with
such as ecological opportunism coupled with a more complex the molecular pattern suggests that the morphological pattern
fitness landscape, which might have unleashed latent evolu- is biologically meaningful rather than artifactual.
tionary capabilities [4, 8], or smaller body size and shorter gen- These findings also directly answer a powerful argument for
eration time in basal metazoan (including arthropod) lineages a lengthy Precambrian phylogenetic fuse. One of the few
Current Biology Vol 23 No 19
1892

A B

D E

Morph Rates

Morph Rates

Figure 2. Fastest Rates of Phenotypic and Molecular Change in Arthropods Are Concentrated in Lineages Spanning the Cambrian Explosion
Warmest colors (red) denote fastest-evolving branches; numbers beside branches in (A) and (D) denote evolutionary rates in % change per million years
(pMyr) (95% HPDs in Table S2). These results are robust to dating methods and calibration assumptions (Table S3). Taxa are color coded according to major
clade as in Figure 1. Branch rates for (B), (C), (E), and (F) are detailed in Figure S2.
(A–C) Molecular evolutionary rates inferred using BEASTMC3 assuming panarthropods are no older than 558 Ma (A), maximum likelihood and penalized
likelihood rate smoothing (B), and BEASTMC3 assuming panarthropods are no older than 650 Ma (C).
(D–F) Phenotypic (morphological) evolutionary rates inferred in the same three analyses as (A)–(C).
Evolutionary Rates during the Cambrian Explosion
1893

A B
0.50%
Mol. Rates
0.40% (Maximum Likelihood)
0.25% 11.61
Molecular Rates (Bayesian, root < 558 Ma) 0.30%

0.20% 9.29 0.20%

Ingroup branch
0.10%
Outgroup branch
6.97 0

ingroup branches (molecular) or ingroup internal branches (morphological)

1 = base rate, ie overall rate since upper Cambrian across all
0.15% Trendline 600 400 200 0
(50 Myr moving average of )
C

4.65 0.10%
0.10%

Relative Rate of Change

Mol. Rates
(percentage change per million years)

(Bayesian, root < 650 Ma)

Absolute Rate of Change

0.05% 2.33 0.05%

0 0 0
600 400 200 0
E
8%

6% Morph. Rates
2% 14.75 (Maximum Likelihood)
Morphological Rates (Bayesian, root < 558 Ma) 4%

1.5% 11.07 2%
Ingroup internal branch (internode)
Ingroup external branch (tip) 0
D 1% Outgroup branch 600 400 200 0
7.38
Trendline
(50 Myr moving average of )
F
1%

Morph. Rates
0.5% 3.69
(Bayesian, root < 650 Ma)
0.5%

0 0
600 500 400 300 200 100 0 0
600 400 200 0
Branch Age (age of midpoint, in millions of years)

Figure 3. Rates of Phenotypic and Molecular Evolution in Arthropods Increased w4 and w5.5-Fold during the Cambrian Explosion
Rates (in % change pMyr) plotted against branch age (= midpoint of branch). Fastest branches (warm colors) are concentrated in an interval from 558 to 500
Ma (pink shading). Brown line represents the moving average inferred using a sliding window of 50 Myr. Because of potential bias, outgroup branches (for
molecular change) and outgroup and terminal/tip branches (for morphological change) are plotted using different symbols and excluded from sliding-win-
dow calculations. See Supplemental Experimental Procedures SI_1–SI_12 for full methods and Figures 1 and S2 for rates on all individual branches.
(A–C) Molecular evolutionary rates inferred using BEASTMC3 assuming panarthropods are no older than 558 Ma (A), maximum likelihood and penalized
likelihood rate smoothing (B), and BEASTMC3 assuming panarthropods are no older than 650 Ma (C).
(D–F) Phenotypic (morphological) evolutionary rates inferred in the same three analyses as (A)–(C).

studies comparing Cambrian and subsequent evolutionary arthropod divergences are compressed into w40 Myr, with a
rates investigated phylogenetic signal in simulated DNA data 5.4-fold increase in the early Cambrian (Figures 1 and 3).
[7]. That study suggested that reconstructing relationships The present study reveals that in order to equalize morpho-
between Cambrian lineages would be essentially impossible logical and molecular evolutionary rates before and after the
assuming constant evolutionary rates, unless they diverged Cambrian explosion, the root of panarthropods needs to be
gradually over a time period of >100 Myr. However, the same w940 Myr old (Table S3: soft-bounded root). Thus, the
study [7] showed that phylogenetic relationships in a assumption of rate constancy through time means that exten-
‘‘Cambrian bush’’ are resolvable if elevated evolutionary rates sive morphological and molecular divergences between the
during the critical time window generated large suites of primary arthropod groups must be accommodated on chrono-
changes on chronologically short branches. Our results sup- logically long branches, leading to very early basal diver-
port this possibility, and the numbers are highly consistent. gences inconsistent with the fossil record. Notably, other
The simulations found that the optimal DNA substitution rate recent molecular studies, without hard bounds on root ages,
for recovering Cambrian evolutionary relationships was obtained similar dates for panarthropods [9, 38]; such ap-
0.01% pMyr; the average substitution rate inferred here is proaches allow the analysis to freely ‘‘telescope’’ the age of
0.02% pMyr, confirming the appropriateness of this filtered the root to reduce rate heterogeneity across time. However,
genomic data for inferring deep arthropod divergences [29] those studies did not investigate how rates vary if the
(Figure S1). Simulations suggested that ability to recover a maximum root age is successively reduced to be compatible
series of divergences compressed into w35 Myr improved with the fossil record (and the majority of recent molecular
significantly if evolutionary rates in this interval were increased studies). The present study reveals that constraining root
5-fold; in the focal analysis (root < 558 Ma), most basal age to any figure even loosely consistent with the fossil record
Current Biology Vol 23 No 19
1894

A B

Figure 4. Evolutionary Rates in Arthropods Based on Molecular Data Only

Arthropod phylogeny, divergence dates, and evolutionary rates based on a BEASTMC3 analysis of 62 protein-coding genes (Supplemental Experimental
Procedures SI_9); the patterns retrieved here are very similar to those found in the context of a combined morphological and molecular analysis (Figures
1 and 2).
(A) Consensus tree with branch lengths and divergence dates in terms of time; all posterior probabilities = 1.0 unless indicated. Shaded bars denote 95%
HPDs for divergences, which are often large due to extreme rate heterogeneity.
(B) Tree with branch lengths proportional to amount of molecular change; taxon order is identical to (A). Large amounts of change occur on chronologically
short basal branches.
(C) Tree with rates of evolutionary change on each branch (% pMyr); warmest colors (red) denote fastest-evolving branches.

(e.g., <650 Ma) implies elevated rates of morphological and to explain the Cambrian explosion of arthropods, even under
molecular evolution in the late Precambrian. an extreme scenario in which all divergences are compressed
Darwin famously considered that the sudden appearance into the Cambrian. Rather, the pattern is consistent with
of complex morphologies in the lower Cambrian was at many Cambrian lineages exhibiting accelerated—yet plau-
odds with normal evolutionary processes [2, 4]. Many sub- sible—rates of morphological and molecular evolution.
sequent workers have reasonably contended that this pulse Typical directional selection can increase phenotypic evolu-
of diverse fossils is not explicable without either positing a tionary rates by orders of magnitude over short timescales
lengthy cryptic Precambrian prelude or invoking ‘‘unknown [39], and even conserved genomic regions can exhibit 10-
evolutionary mechanisms’’ [6–8]. Similarly, initial molecular fold differences in evolutionary rates in living sister lineages
analyses concluded that the extensive molecular divergences [40]. More specifically, in arthropods, data sets of first- and
between arthropods, echinoderms, and chordates could not second-position codons alone often exhibit 2-fold, and occa-
be fully reconciled with the abrupt Cambrian fossil record, sionally 5-fold, differences between closely related taxa
assuming even the fastest modern rates of molecular evolu- (Figure S2D).
tion [35]. These legitimate reservations have predictably While this study examined only arthropods, the patterns
been exploited by opponents of evolution. However, ‘‘Dar- found here may be general across other metazoan groups,
win’s dilemma’’ [2, 4] might be resolvable. The initial mo- though broader taxonomic studies are required. Arthropods
lecular analyses [35] demonstrated that observed rates of are by far the most abundant and well-known Cambrian
molecular evolution could be reconciled with divergences phylum, typically representing nearly 40% of species and
between metazoan phyla as recent as w586 Ma, which more than half of the specimens in Burgess Shale-type biotas
(although still pre-dating the Cambrian) is now broadly [41], a dominance that has persisted until the present [33].
congruent with recent discoveries of the earliest metazoans The patterns found in arthropods have accordingly been
(Table S1). Our results, using updated methods on morpho- routinely extrapolated as representative for all Cambrian taxa
logical and genomic-scale data, show potentially even (e.g., [6]). Clades characterized by major phenotypic innova-
greater congruence. Inexplicably fast rates are not required tions often exhibit higher rates of evolution initially (e.g.,
Evolutionary Rates during the Cambrian Explosion
1895

[42]); if this pattern characterized most metazoan phyla, rates revealed by phylogenomic analysis of nuclear protein-coding
in the Cambrian would be uniformly elevated [1–4]. sequences. Nature 463, 1079–1083.
18. Davidson, E.H. (2006). The Regulatory Genome: Gene Regulatory
Networks in Development and Evolution (New York: Academic Press).
Supplemental Information
19. Briggs, D.E.G., Fortey, R.A., and Wills, M.A. (1992). Morphological
disparity in the Cambrian. Science 256, 1670–1673.
Supplemental Information includes three figures, three tables, Supple-
20. Thomas, R.D.K., Shearman, R.M., and Stewart, G.W. (2000).
mental Experimental Procedures SI_1–SI_14, and two supplemental data
Evolutionary exploitation of design options by the first animals with
files and can be found with this article online at http://dx.doi.org/10.1016/
hard skeletons. Science 288, 1239–1242.
j.cub.2013.07.055.
21. Zhu, M.-Y., Babcock, L.E., and Peng, S.-C. (2006). Advances in
Cambrian stratigraphy and paleontology: Integrating correlation tech-
Acknowledgments
niques, paleobiology, taphonomy and paleoenvironmental reconstruc-
tion. Palaeoworld 15, 217–222.
We thank John Paterson, Jim Gehling, Erik Sperling, and the journal’s ref-
22. Edgecombe, G.D. (2010). Palaeomorphology: fossils and the inference
erees for comments; Sean Reilly and eResearch South Australia for grid
of cladistic relationships. Acta Zool. 91, 72–80.
computing resources and technical assistance; and the Australian Research
23. Aris-Brosou, S., and Yang, Z. (2002). Effects of models of rate evolution
Council and The Environment Institute (University of Adelaide) for funding.
on estimation of divergence dates with special reference to the meta-
zoan 18S ribosomal RNA phylogeny. Syst. Biol. 51, 703–714.
Received: May 14, 2013 24. Aris-Brosou, S., and Yang, Z. (2003). Bayesian models of episodic evo-
Revised: June 19, 2013 lution support a late Precambrian explosive diversification of the
Accepted: July 16, 2013 Metazoa. Mol. Biol. Evol. 20, 1947–1954.
Published: September 12, 2013 25. Welch, J.J., Fontanillas, E., and Bromham, L. (2005). Molecular dates for
the ‘‘Cambrian explosion’’: the influence of prior assumptions. Syst.
References Biol. 54, 672–678.
26. Ho, S.Y.W., Phillips, M.J., Drummond, A.J., and Cooper, A. (2005).
1. Gould, S.J. (1989). Wonderful Life: The Burgess Shale and the Nature of Accuracy of rate estimation using relaxed-clock models with a critical
History (New York: W.W. Norton). focus on the early metazoan radiation. Mol. Biol. Evol. 22, 1355–1363.
2. Budd, G.E. (2008). The earliest fossil record of the animals and its signif- 27. Ronquist, F., Klopfstein, S., Vilhelmsen, L., Schulmeister, S., Murray,
icance. Philos. Trans. R. Soc. Lond. B Biol. Sci. 363, 1425–1434. D.L., and Rasnitsyn, A.P. (2012). A total-evidence approach to dating
3. Conway Morris, S. (2006). Darwin’s dilemma: the realities of the with fossils, applied to the early radiation of the Hymenoptera. Syst.
Cambrian ‘explosion’. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361, Biol. 61, 973–999.
1069–1083. 28. Pyron, R.A. (2011). Divergence time estimation using fossils as terminal
4. Erwin, D.H., Laflamme, M., Tweedt, S.M., Sperling, E.A., Pisani, D., and taxa and the origins of Lissamphibia. Syst. Biol. 60, 466–481.
Peterson, K.J. (2011). The Cambrian conundrum: early divergence and 29. Regier, J.C., and Zwick, A. (2011). Sources of signal in 62 protein-coding
later ecological success in the early history of animals. Science 334, nuclear genes for higher-level phylogenetics of arthropods. PLoS ONE
1091–1097. 6, e23408.
5. Peters, S.E., and Gaines, R.R. (2012). Formation of the ‘Great 30. Lee, M.S.Y., and Camens, A.B. (2009). Strong morphological support for
Unconformity’ as a trigger for the Cambrian explosion. Nature 484, the molecular evolutionary tree of placental mammals. J. Evol. Biol. 22,
363–366. 2243–2257.
6. Fortey, R.A., Briggs, D.E.G., and Wills, M.A. (1996). The Cambrian evolu- 31. Stamatakis, A. (2006). RAxML-VI-HPC: maximum likelihood-based
tionary ‘explosion’: Decoupling cladogenesis from morphological phylogenetic analyses with thousands of taxa and mixed models.
disparity. Biol. J. Linn. Soc. Lond. 57, 13–33. Bioinformatics 22, 2688–2690.
7. Levinton, J., Dubb, L., and Wray, G.A. (2004). Simulations of evolu- 32. Rota-Stabelli, O., Lartillot, N., Philippe, H., and Pisani, D. (2013). Serine
tionary radiations and their application to understanding the probability codon-usage bias in deep phylogenomics: pancrustacean relationships
of a Cambrian explosion. J. Paleontol. 78, 31–38. as a case study. Syst. Biol. 62, 121–133.
8. Valentine, J.W. (2004). On the Origin of Phyla (Chicago: University of 33. Giribet, G., and Edgecombe, G.D. (2012). Reevaluating the arthropod
Chicago Press). tree of life. Annu. Rev. Entomol. 57, 167–186.
9. Wheat, C.W., and Wahlberg, N. (2013). Phylogenomic insights into the 34. Harvey, T.H.P., Vélez, M.I., and Butterfield, N.J. (2012). Exceptionally
Cambrian explosion, the colonization of land and the evolution of flight preserved crustaceans from western Canada reveal a cryptic
in arthropoda. Syst. Biol. 62, 93–109. Cambrian radiation. Proc. Natl. Acad. Sci. USA 109, 1589–1594.
10. Jensen, S., Droser, M.L., and Gehling, J.G. (2005). Trace fossil preserva- 35. Bromham, L.D., and Hendy, M.D. (2000). Can fast early rates reconcile
tion and the early evolution of animals. Palaeogeogr. Palaeoclimatol. molecular dates with the Cambrian explosion? Proc. Biol. Sci. 267,
Palaeoecol. 220, 19–29. 1041–1047.
11. Rehm, P., Borner, J., Meusemann, K., von Reumont, B.M., Simon, S., 36. Vavouri, T., and Lehner, B. (2009). Conserved noncoding elements and
Hadrys, H., Misof, B., and Burmester, T. (2011). Dating the arthropod the evolution of animal body plans. Bioessays 31, 727–735.
tree based on large-scale transcriptome data. Mol. Phylogenet. Evol. 37. Jacobs, D.K. (1990). Selector genes and the Cambrian radiation of
61, 880–887. Bilateria. Proc. Natl. Acad. Sci. USA 87, 4406–4410.
12. Rota-Stabelli, O., Daley, A.C., and Pisani, D. (2013). Molecular timetrees 38. Sanders, K.L., and Lee, M.S.Y. (2010). Arthropod molecular diver-
reveal a Cambrian colonization of land and a new scenario for ecdyso- gence times and the Cambrian origin of pentastomids. Syst.
zoan evolution. Curr. Biol. 23, 392–398. Biodivers. 8, 63–74.
13. Drummond, A.J., Ho, S.Y.W., Phillips, M.J., and Rambaut, A. (2006). 39. Reznick, D.N., Shaw, F.H., Rodd, F.H., and Shaw, R.G. (1997).
Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88. Evaluation of the rate of evolution in natural populations of guppies
14. Drummond, A.J., Suchard, M.A., Xie, D., and Rambaut, A. (2012). (Poecilia reticulata). Science 275, 1934–1937.
Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. 40. Crawford, N.G., Faircloth, B.C., McCormack, J.E., Brumfield, R.T.,
Evol. 29, 1969–1973. Winker, K., and Glenn, T.C. (2012). More than 1000 ultraconserved ele-
15. Sanderson, M.J. (2003). r8s: inferring absolute rates of molecular evolu- ments provide evidence that turtles are the sister group of archosaurs.
tion and divergence times in the absence of a molecular clock. Biol. Lett. 8, 783–786.
Bioinformatics 19, 301–302. 41. Zhao, F., Zhu, M., and Hu, S. (2010). Community structure and compo-
16. Rota-Stabelli, O., Campbell, L., Brinkmann, H., Edgecombe, G.D., sition of the Cambrian Chengjiang biota. Sci. China. Earth Sci. 53, 1784–
Longhorn, S.J., Peterson, K.J., Pisani, D., Philippe, H., and Telford, 1799.
M.J. (2011). A congruent solution to arthropod phylogeny: phylogenom- 42. Lloyd, G.T., Wang, S.C., and Brusatte, S.L. (2012). Identifying hetero-
ics, microRNAs and morphology support monophyletic Mandibulata. geneity in rates of morphological evolution: discrete character
Proc. Biol. Sci. 278, 298–306. change in the evolution of lungfish (Sarcopterygii; Dipnoi). Evolution
17. Regier, J.C., Shultz, J.W., Zwick, A., Hussey, A., Ball, B., Wetzer, R., 66, 330–348.
Martin, J.W., and Cunningham, C.W. (2010). Arthropod relationships