- Edited by Neil H. Shubin, University of Chicago, Chicago, IL, and approved August 19, 2016 (received for review May 6, 2016)
Significance
The
success of teleost fishes, which represent roughly half of all
vertebrate species, has attracted attention since Darwin. Numerous
scenarios invoke elevated diversification in teleosts facilitated by
supposed key innovations, yet claims of teleost exceptionalism are
profoundly biased by the evolutionary “snapshot” of living fishes.
Analysis of 160 million y (Permian–Early Cretaceous) of evolution in
neopterygian fishes reveals that anatomical diversification in Mesozoic
teleosts as a whole differed little from their “living fossil” holostean
sister group. There is some evidence for evolutionary heterogeneity
within teleosts, with early evolving lineages showing the greatest
capacity for evolutionary innovation in body shape among Mesozoic
neopterygians, whereas members of the modern teleost radiation show
higher rates of shape evolution.
Abstract
Since
Darwin, biologists have been struck by the extraordinary diversity of
teleost fishes, particularly in contrast to their closest “living
fossil” holostean relatives. Hypothesized drivers of teleost success
include innovations in jaw mechanics, reproductive biology and,
particularly at present, genomic architecture, yet all scenarios
presuppose enhanced phenotypic diversification in teleosts. We test this
key assumption by quantifying evolutionary rate and capacity for
innovation in size and shape for the first 160 million y (Permian–Early
Cretaceous) of evolution in neopterygian fishes (the more extensive
clade containing teleosts and holosteans). We find that early teleosts
do not show enhanced phenotypic evolution relative to holosteans.
Instead, holostean rates and innovation often match or can even exceed
those of stem-, crown-, and total-group teleosts, belying the living
fossil reputation of their extant representatives. In addition, we find
some evidence for heterogeneity within the teleost lineage. Although
stem teleosts excel at discovering new body shapes, early crown-group
taxa commonly display higher rates of shape evolution. However, the
latter reflects low rates of shape evolution in stem teleosts relative
to all other neopterygian taxa, rather than an exceptional feature of
early crown teleosts. These results complement those emerging from
studies of both extant teleosts as a whole and their sublineages, which
generally fail to detect an association between genome duplication and
significant shifts in rates of lineage diversification.
Numbering
∼29,000 species, teleost fishes account for half of modern vertebrate
richness. In contrast, their holostean sister group, consisting of gars
and the bowfin, represents a mere eight species restricted to the
freshwaters of eastern North America (1). This stark contrast between teleosts and Darwin's original “living fossils” (2) provides the basis for assertions of teleost evolutionary superiority that are central to textbook scenarios (3, 4). Classic explanations for teleost success include key innovations in feeding (3, 5) (e.g., protrusible jaws and pharyngeal jaws) and reproduction (6, 7). More recent work implicates the duplicate genomes of teleosts (8⇓–10) as the driver of their prolific phenotypic diversification (8, 11⇓–13),
concordant with the more general hypothesis that increased
morphological complexity and innovation is an expected consequence of
genome duplication (14, 15).
Most arguments for enhanced phenotypic evolution in teleosts have been asserted rather than demonstrated (8, 11, 12, 15, 16; but see ref. 17),
and draw heavily on the snapshot of taxonomic and phenotypic imbalance
apparent between living holosteans and teleosts. The fossil record
challenges this neontological narrative by revealing the remarkable
taxonomic richness and morphological diversity of extinct holosteans (Fig. 1) (18, 19) and highlights geological intervals when holostean taxonomic richness exceeded that of teleosts (20). This paleontological view has an extensive pedigree. Darwin (2)
invoked a long interval of cryptic teleost evolution preceding the late
Mesozoic diversification of the modern radiation, a view subsequently
supported by the implicit (18) or explicit (19)
association of Triassic–Jurassic species previously recognized as
“holostean ganoids” with the base of teleost phylogeny. This perspective
became enshrined in mid-20th century treatments of actinopterygian
evolution, which recognized an early-mid Mesozoic phase dominated by
holosteans sensu lato and a later interval, extending to the modern day, dominated by teleosts (4, 20, 21). Contemporary paleontological accounts echo the classic interpretation of modest teleost origins (22⇓–24), despite a systematic framework that substantially revises the classifications upon which older scenarios were based (22⇓⇓–25).
Identification of explosive lineage diversification in nested teleost
subclades like otophysans and percomorphs, rather than across the group
as a whole, provides some circumstantial neontological support for this
narrative (26).
Fig. 1.
Phenotypic variation in early crown neopterygians. (A) Total-group holosteans. (B) Stem-group teleosts. (C) Crown-group teleosts. Taxa illustrated to scale.
In contrast to quantified taxonomic patterns (20, 23, 24, 27),
phenotypic evolution in early neopterygians has only been discussed in
qualitative terms. The implicit paleontological model of morphological
conservatism among early teleosts contrasts with the observation that
clades aligned with the teleost stem lineage include some of the most
divergent early neopterygians in terms of both size and shape (Fig. 1) (see, for example, refs. 28 and 29).
These discrepancies point to considerable ambiguity in initial patterns
of phenotypic diversification that lead to a striking contrast in the
vertebrate tree of life, and underpins one of the most successful
radiations of backboned animals.
Here we tackle this
uncertainty by quantifying rates of phenotypic evolution and capacity
for evolutionary innovation for the first 160 million y of the crown
neopterygian radiation. This late Permian (Wuchiapingian, ca. 260 Ma) to
Cretaceous (Albian, ca. 100 Ma) sampling interval permits incorporation
of diverse fossil holosteans and stem teleosts alongside early
diverging crown teleost taxa (Figs. 1 and 2A and Figs. S1 and S2), resulting in a dataset of 483 nominal species-level lineages roughly divided between the holostean and teleost total groups (Fig. 2B and Fig. S2). Although genera are widely used as the currency in paleobiological studies of fossil fishes (30; but see ref. 31),
we sampled at the species level to circumvent problems associated with
representing geological age and morphology for multiple congeneric
lineages. We gathered size [both log-transformed standard length (SL)
and centroid size (CS); results from both are highly comparable (Figs. S3 and S4);
SL results are reported in the main text] and shape data (the first
three morphospace axes arising from a geometric morphometric analysis) (Fig. 2A and Figs. S1)
from species where possible. To place these data within a phylogenetic
context, we assembled a supertree based on published hypotheses of
relationships. We assigned branch durations to a collection of trees
under two scenarios for the timescale of neopterygian diversification
based on molecular clock and paleontological estimates. Together, these
scenarios bracket a range of plausible evolutionary timelines for this
radiation (Fig. 2B).
We used the samples of trees in conjunction with our morphological
datasets to test for contrasts in rates of, and capacity for, phenotypic
change between different partitions of the neopterygian Tree of Life
(crown-, total-, and stem-group teleosts, total-group holosteans, and
neopterygians minus crown-group teleosts), and the sensitivity of these
conclusions to uncertainty in both relationships and evolutionary
timescale.
Critically, these include comparisons of phenotypic evolution
in early crown-group teleosts—those species that are known with
certainty to possess duplicate genomes—with rates in taxa characterized
largely (neopterygians minus crown teleosts) or exclusively (holosteans)
by unduplicated genomes. By restricting our scope to early diverging
crown teleost lineages, we avoid potentially confounding signals from
highly nested radiations that substantially postdate both genome
duplication and the origin of crown teleosts (26, 32).
This approach provides a test of widely held assumptions about the
nature of morphological evolution in teleosts and their holostean sister
lineage.
Fig. 2.
(A) Morphospace of Permian–Early Cretaceous crown Neopterygii. (B) One supertree subjected to our paleontological (Upper) and molecular (Lower) timescaling procedures to illustrate contrasts in the range of evolutionary timescales considered. Colors of points (A) and branches (B) indicate membership in major partitions of neopterygian phylogeny. Topologies are given in Datasets S4 and S5. See Dataset S6 for source trees.
Fig. S1.
Morphospace
of 398 Permian–Early Cretaceous Neopterygii. Three major axes of shape
variation are presented. Silhouettes and accompanying arrows illustrate
the main anatomical correlates of these principal axes, as described in Table S3.
Actual fossil specimens represented in the dataset are also displayed
to visualize the extremities of morphospace axes. Fossil taxa as
follows: (a) Bavarichthys incognitus; (b) Ellimmichthys longicostatus; (c) Turbomesodon relegans; (d) Allothrissops salmoneus; (e) Pholidophorus germanicus; (f) Macrosemius fourneti.
Fig. S2.
Morphospace of 398 Permian–Early Cretaceous Neopterygii, illustrating the major clades of (A) teleosts and (B) holosteans.
Fig. S3.
Comparisons of size rates between (A) holosteans and teleosts, (B) crown teleosts and all other neopterygians, (C) crown teleosts and stem teleosts, (D) crown teleosts and holosteans, and (E)
stem teleosts and holosteans. Comparisons were made using the full-size
SL dataset, a CS dataset, and a smaller SL dataset pruned to exactly
match the taxon sampling of the CS dataset. Identical taxon sampling
leads the CS and pruned SL datasets to yield near identical results.
Although the larger SL dataset results often differ slightly, the
overall conclusion from each pairwise comparison (i.e., which outcome is
the most likely in an overall majority of trees) is identical in all
but one comparison (E, under molecular timescales).
Fig. S4.
Comparisons of size innovation between (A) holosteans and teleosts, (B) crown teleosts and all other neopterygians, (C) crown teleosts and stem teleosts, (D) crown teleosts and holosteans, and (E)
stem teleosts and holosteans. Comparisons were made using the full-size
SL dataset, a CS dataset, and a smaller SL dataset pruned to exactly
match the taxon sampling of the CS dataset. Comparisons of size
innovation are presented for K value distributions of the three datasets resemble each other closely.
Results and Discussion
Mesozoic Teleosts Do Not Show Enhanced Phenotypic Diversification.
Contrary to expectations ingrained in the neontological literature (3, 4, 8, 9, 12, 13, 15), early teleosts do not possess significantly higher rates of size or shape evolution than holosteans (Fig. 3A). We found no significant difference between rates of body size (measured as SL; other measures deliver comparable results) (Figs. S3 and S4)
evolution between early members of the holostean and teleost total
groups across the majority of sampled topologies under either
paleontological or molecular timescales (Fig. 3A).
In terms of evolution in overall body shape (as indicated by scores on
the first three shape axes of our morphospace), we found no consistent
signal, whether rates were higher in total-group teleosts or holosteans (Fig. 3A).
However, total-group holosteans possess significantly higher rates of
overall shape change than total-group teleosts in a majority of
topologies when these were timescaled to match published divergence-time
estimates made using the molecular clock (Fig. 3A).
Fig. 3.
Comparisons of phenotypic rate and innovation between (A) holosteans and teleosts, (B) crown teleosts and all other neopterygians, and (C)
crown teleosts and stem teleosts. Rate results are conveyed with pie
charts, where the proportion of sampled supertrees in support of
significantly higher rates (as determined from two-way P values
with α = 0.05) in a given neopterygian partition are color coded to
match that partition; white sections correspond to supertrees indicating
no significant rate differences. Violin plots capturing the
distribution of K values obtained from 100 sampled supertrees allow for investigation of innovation. Higher K
values suggest greater phylogenetic signal, and correspondingly,
greater innovation, in contrast to the iterative “rediscovery” of
similar phenotypes (Methods). K interpretations are contextualised with rate in Table S1.
We calculated Blomberg’s K for specific clades to summarize how efficiently they explore phenotypic space (Methods). Because interpretation of K in isolation can be misleading (33), it is useful to consider K alongside information on evolutionary rate (34) (Table S1). We report K values directly here and in subsequent sections, drawing on our comparisons of evolutionary rate to provide necessary context. Table S1 provides more detailed interpretations of K using rate information. Concerning size innovation, K distributions for the teleost and holostean total-groups overlap considerably on paleontological timescales (Fig. 3A), whereas holostean K values are distinctly higher than teleost values on molecular timescales (Fig. 3A). Taken together with the suggestion of broadly comparable rates of evolution in the two clades (Fig. 3A and Table S1), these results imply that total-group holosteans either match or exceed total-group teleosts in their size innovation. (Fig. 3A and Table S1). K distributions for shape in holosteans and teleosts are comparable regardless of timescale (Fig. 3A and Table S1), suggesting they were similarly innovative.
Table S1.
Innovation interpretations, based upon K value distributions, contextualized with rate
The Broader Holostean Radiation Does Not Fulfill Numerous Living Fossil Expectations.
Darwin articulated several concepts of what it means to be a living fossil (2),
such as taxa that are “remnants of a once preponderant order,” or are
“slowly formed” (i.e., showing low rates of lineage diversification or
trait evolution). Extant holosteans, which are among Darwin’s archetypal
living fossils (2),
embody these features clearly: they represent the last survivors of a
once diverse radiation and demonstrate low rates of lineage
diversification (26, 32, 35). The modern genera Amia, Lepisosteus, and Atractosteus show negligible anatomical change since their first appearance in the Late Cretaceous and Paleocene (1, 36), and living gars show low rates of body size evolution (17).
In contrast to this pattern from living species, Mesozoic holosteans show comparable rates of size change to total- (Fig. 3A), crown- (Fig. S5A), and stem-group teleosts (Fig. S5B).
These patterns are seen across a majority of topologies regardless of
timescale. Rates of shape evolution in holosteans are broadly comparable
to those of teleosts, but often exceed those of total-group teleosts on
molecular timescales (Fig. 3A) and stem teleosts on both timescales (Fig. S5B).
Fig. S5.
Comparisons of phenotypic rate and innovation between (A) holosteans and crown teleosts; (B)
holosteans and stem teleosts. Rate results are conveyed with pie
charts, where the proportion of sampled supertrees in support of
significantly higher rates (as determined from two-way P values
with α = 0.05) in a given neopterygian partition are color-coded to
match that partition; white sections correspond to supertrees indicating
no significant rate differences. Violin plots capturing the
distribution of K values obtained from 100 sampled supertrees allow for investigation of innovation. Higher K
values suggest greater phylogenetic signal, and correspondingly,
greater innovation, in contrast to the iterative “rediscovery” of
similar phenotypes (Methods). K interpretations are contextualized with the rate in Table S1.
There is no clear difference in size innovation between holosteans and total- (Fig. 3A), crown- (Fig. S5A), and stem- (Fig. S5B) group teleosts on paleontological timescales (Table S1). On molecular timescales, holostean-size K values are marginally larger than those of total- (Fig. 3A) and crown-group teleosts (Fig. S5A), and clearly larger than stem teleosts (Fig. S5B), suggesting holosteans are more innovative in these instances (Table S1).
Unlike evolutionary rates and size innovation, there are some scenarios
in which holosteans are marginally poorer shape innovators than
teleosts. For example, on paleontological timescales, holosteans show
marginally less shape innovation than crown teleosts (Fig. S5A and Table S1) and appear less innovative than stem teleosts (Fig. S5B and Table S1) regardless of timescale.
Crown Teleosts Display Comparable Patterns of Phenotypic Evolution to other Mesozoic Neopterygians.
Despite
possession of duplicate genomes, we find only ambiguous evidence for
elevated shape evolution in early crown teleosts relative to rates in
other neopterygian lineages. Rates are significantly higher for only a
small majority of topologies on paleontological timescales (Fig. 3B), and fewer than half on molecular timescales (Fig. 3B).
Evidence for higher rates in crown teleosts is even less compelling for
size evolution, where a majority of trees display no significant
difference in rate between crown teleosts and other neopterygians
regardless of timescale (Fig. 3B).
Matching
our inferences concerning evolutionary rate, we find no clear evidence
to support the notion that early crown teleosts are better size or shape
innovators than other neopterygian fishes as a whole. Regarding size,
crown teleost K values are comparable to those of other neopterygians regardless of timescale (Fig. 3B), suggesting they are similarly innovative (Table S1). Regarding shape, crown teleost K values are either comparable to (on paleontological timescales) (Fig. 3B) or marginally lower than (on molecular timescales) (Fig. 3B) those of other neopterygians, suggesting they are similarly or less innovative, respectively (Table S1).
Capacity for Innovation and Rates of Phenotypic Change Vary Within the Teleost Total Group.
Unremarkable
evolutionary patterns across teleosts mask heterogeneities within the
teleost total group. For example, crown-group teleosts show
significantly elevated rates of shape evolution relative to stem
teleosts in a small majority of topologies under both molecular and
paleontological timescales (Fig. 3C).
However, evidence for elevated rates of shape change in crown-group
teleosts is by no means unambiguous; we also find no rate difference
between crown teleosts and stem teleosts, or significantly higher rates
in stem teleosts, in a nontrivial fraction of topologies (Fig. 3C).
Furthermore, finding higher shape rates in crown teleosts relative to
stem taxa does not demonstrate uniquely enhanced shape diversification
in crown teleosts, because holosteans also demonstrate higher shape
rates than stem teleosts in a similar fraction of topologies (Fig. S5B).
In contrast to these patterns for shape evolution, we find little
evidence for elevated rates of size evolution in crown teleosts relative
to members of the stem (Fig. 3C).
Possible
contrasts in rates of shape evolution between crown teleosts and stem
teleosts do not align with patterns of evolutionary innovation. K
distributions point to moderately (paleontological timescale) or
substantially (molecular timescale) lower capacity for evolutionary
innovation in members of the crown compared with those on the stem (Fig. 3C).
Concerning differences in body-size evolution, there is little support
for major differences between crown- and stem-group teleosts (Fig. 3C).
Tenuous Links Between Genome Duplication and Enhanced Evolutionary Rate and Innovation in Fishes.
The
staggering ecological and anatomical diversity of extant teleosts,
especially in comparison with living nonteleost actinopterygian
lineages, has long been taken as prima facie evidence of enhanced
capacity for phenotypic evolution in this enormously successful
vertebrate radiation (3, 4, 8, 9, 12).
Based on a more balanced taxon sample incorporating roughly equal
numbers of early teleost and holostean species, we find that evidence
for this widely held assumption is at best equivocal. Teleosts as a
whole cannot be reliably distinguished from holosteans in terms of
either rate of phenotypic change or capacity for evolutionary
innovation. The most consistent contrasts we find concern patterns of
shape change between crown- and stem-group teleosts, but these do not
align: stem teleosts potentially show a higher capacity for evolutionary
innovation, whereas crown teleosts are characterized by higher rates of
phenotypic change. However, both must be viewed as ambiguous in light
of the multiple pairwise comparisons made between partitions of
neopterygian phylogeny.
With the obvious caveat that our
paleontological sample excludes some of the most divergent modern
teleost body plans, these results call into question the search for key
innovations fueling the success of modern teleosts in sum. Many such
features of teleost biology have been proposed (3⇓⇓⇓–7),
but the duplicate genomes that characterize all living members of this
group represent the most popular candidate in recent literature (8, 11⇓–13).
The connection between genome duplication and shifts in evolutionary
patterns in teleosts has, to date, been addressed in terms of rates of
lineage diversification in modern species alone. As with our own
examination of phenotypic evolution, these studies yield ambiguous
results. Elevated rates of lineage diversification are not uniformly
detected for crown teleosts as a whole, but there is a consistent and
strong signal for exceptional shifts in rate associated with the
hyperdiverse and phylogenetically nested otophysan and percomorph
radiations (26, 32, 35).
The origin of these clades substantially postdates the teleost-specific
whole-genome duplication, which molecular-clock dating of paralogue
pairs (10) localizes to the middle of the teleost stem lineage rather than near the origin of the crown radiation (Table S2).
Table S2.
Relative time of the genome duplication event upon the teleost stem lineage
Further
polyploid events within actinopterygians provide additional natural
experiments for examining the consequences of genome duplications for
subsequent patterns of evolutionary diversification. Here, too, results
provide little direct support for increased rates of lineage
diversification in polyploid groups relative to close relatives that
have not undergone duplication events (37).
In the case of groups like salmonids, geologically recent ecological
shifts—rather than more ancient changes in genomic architecture—appear
more closely linked with increased rates of lineage diversification (38, 39).
How rates of phenotypic evolution might relate to these polyploid
events has not been explored specifically, although at least two
lineages characterized largely (Acipenseridae) or exclusively
(Salmonidae) by polyploid species show elevated rates of body-size
evolution relative to background actinopterygian rates (17).
Thus, although genome duplication represents a seductive and widely
enlisted hypothesis for explaining the taxonomic and especially
morphological proliferation of clades, a synoptic view of the
consequences of polyploid events on rates of lineage diversification and
patterns of phenotypic change remains elusive.
Although
our results do not strongly suggest any immediate consequences of
genome duplication for morphological evolution in early crown-group
teleosts, we anticipate that future developments could help to better
constrain these patterns. Inferences about rate heterogeneity can vary
substantially within our pool of sampled topologies. A more robustly
constrained hypothesis of relationships and times of evolutionary
divergence among early neopterygians, therefore, represents a first step
to more decisive detection of shifts in the nature of morphological
evolution across this major radiation, the early history of which has
received substantially less systematic attention than more species-poor
groups like birds (40⇓–42) and mammals (43⇓–45). Divergence estimates for paralogues (10) provide a loose constraint for the timing of the teleost-specific genome duplication (Table S2),
but cannot identify which members of the teleost stem lineage were
polyploid. There is the possibility that a paleogenomic approach using
the size of osteocyte lacunae to estimate genome sizes in extinct
lineages could more precisely pinpoint the phylogenetic position of the
genome duplication. In addition to permitting more finely defined
contrasts than those applied here, estimation of genome size in fossil
teleosts would allow direct investigation of rates of genome reduction
following duplication (46).
Methods
Phenotypic Datasets.
Phenotypic
data were collected from photographs of museum specimens of
neopterygians ranging in age from Wuchiapingian (late Permian, ∼260 Ma)
to Albian (Early Cretaceous, ∼100 Ma), supplemented by high-quality
images in the primary literature. The phenotypic datasets represent a
combined total of 1,170 unique specimen images assigned to 483 species.
Our
phenotypic datasets are divided into those describing variation in size
and those capturing differences in shape. Because of varying degrees of
completeness between fossil specimens, these datasets do not contain
identical sets of taxa, although the degree of overlap between any two
datasets is high. We obtained SL for 949 specimens assigned to 468
species, and CS (based on our constellation of landmarks for geometric
morphometic analyses; see below) for 626 individuals assigned to 382
species. Size values within species were averaged, and all resultant
species sizes were log-transformed before analysis (SL in Dataset S1; CS in Dataset S2).
We used a 2D geometric morphometric approach using a constellation of 25 landmarks to quantify shape variation (Fig. S6) using the software package tpsDig2 (47). The shape dataset consisted of 774 specimen images assigned to 398 species (Fig. 2A and Figs. S1 and S2).
Both fixed landmarks and semilandmarks were used to capture overall
body shape and fin position, based on schemes applied previously to
living (48) and fossil (31)
fishes. Landmarked specimen data were aligned using orthogonal
generalized Procrustes superimposition analysis (GPA), permitting shape
values to be averaged within species. The averaged species data were
then aligned with GPA and subject to a relative warp (RW) analysis in
tpsRelw v1.54 (49).
Of the four axes that described >5% of overall variation, the first
three (RW1 to 3) captured clear biological features (rather than
differences potentially related to preservation) and formed the basis of
all shape analyses in Dataset S3. RW1 to 3 explained 42.53%, 21.43%, and 13.52% of the variation respectively. The Supporting Information details anatomical correlates of these axes (Table S3).
Fig. S6.
Landmark
scheme used to quantify shape. Fixed landmarks are marked by red
circles; semilandmarks are marked by black circles with white fill. Red
circles containing a white asterisk act as anchor points for the
semilandmarks in-between. Fixed landmarks document: 1) anterior tip of
the upper jaw (premaxilla); 2) the central, ventral surface of the
orbit; 3) the center of the orbit; 4) the central, dorsal surface of the
orbit; 5) the dorsal surface of the skull immediately above the eye; 6)
postero-dorsal tip of braincase; 7) anterior insertion of dorsal fin;
8) posterior insertion of dorsal fin; 9) dorsal surface representation
of the last vertebral centra; 10) ventral surface representation of the
last vertebral centra; 11) posterior insertion of anal fin; 12, anterior
insertion of anal fin; 13) anterior insertion of the pelvic fin; 14)
anterior insertion of the pectoral fin; 15) lower jaw joint. Landmarked
specimen is Amiopsis lepidota (SMNS 80251) from the Staatliches Museum für Naturkunde, Stuttgart.
Table S3.
The first four axes derived from a relative warp analysis and their anatomical correlates
Tree Construction.
Summarizing existing topologies.
Because
there is no densely sampled phylogenetic hypothesis available for early
fossil neopterygians, we adopted a “supertree” approach to produce a
sample of trees for comparative analyses. Topologies were constructed
using matrix representation with parsimony (MRP), drawing upon 120
source topologies (Dataset S6)
to summarize relationships and capture phylogenetic uncertainty among
671 (mostly Mesozoic, but some living) neopterygian species. We adopted
MRP because many trees lacked the data matrix used to create them (e.g.,
ref. 50) or were hand-constructed (e.g., ref. 51).
All junior synonyms present in the source trees were replaced by their
correct senior synonyms to ensure all taxa were correctly represented in
the source topologies. We included a “seed” (i.e., backbone) tree built
with reliable taxonomic information that contained every species (52).
This process permitted inclusion of large numbers of taxonomically
assigned species that have not otherwise been included in a formal
phylogenetic analysis. Use of taxonomic information is further
vindicated given that paleontological trees derived from taxonomies can
deliver comparable results using comparative methods to those derived
from cladistic phylogenies (53).
The taxonomy seed tree was also treated as a constraint on the
supertree analysis to ensure that strongly corroborated placements could
not be overruled by the source trees (e.g., holostean monophyly, which
is well-supported by modern molecular and morphological analyses, but
not recovered by older studies). We purposely left the seed tree poorly
resolved to allow source trees to dictate relationships where there is
genuine uncertainty. A second constraint was applied to ensure that the
relationships between major living teleost clades matched those arising
from recent molecular phylogenetic studies (54).
To implement both constraints, the nodes (expressed as characters in
the MRP matrix) defining the relationships of the taxonomic and
molecular trees were upweighted to 1,000 (the maximum) in our MRP data
matrix.
Safe taxonomic reduction (55) was performed upon the MRP matrix using Claddis v0.1 (56) before phylogenetic analysis in TNT v1.1 (57).
Twenty replicates of new technology searches were performed, saving
1,000 trees each time, with each replicate starting from a random tree.
Ten-thousand MPTs were then obtained from these saved replicates,
followed by a final search for remaining MPTs with tree bisection and
reconnection, delivering a total of 10,500 MPTs. Taxa removed by safe
taxonomic reduction were reinserted into every MPT, either into their
sole possible position or, if multiple positions were equally likely,
one was chosen at random. One-hundred trees were then selected at random
from this pool for downstream comparative analyses, with any remaining
polytomies randomly resolved using the “multi2di” function in APE (58).
Timescaling topologies.
Living
species were pruned from our 100 supertrees before timescaling using
the timePaleoPhy function of the paleotree package in R (59). As illustrated in Fig. 2B, we adopted two end-member timescaling procedures: (i) a paleontological timescale to reflect divergence times based solely upon fossils, and (ii) a molecular timescale to reflect some of the oldest neopterygian divergence estimates in recent clock studies.
The
tip age of every species was randomized (with a uniform distribution)
between its oldest potential age (i.e., the oldest lower boundary age of
all of the deposits where the species is found) and its oldest reliable
minimum age (i.e., the oldest upper boundary age of all of the deposits
where the species is found). This randomization procedure was carried
out for each tree individually. We used the node-dating procedure of
Hedman (60)
to provide an estimate for the neopterygian crown node (i.e., the root
of the supertree) as the first step in timescaling topologies under our
paleontological approach. This approach delivered a mean estimate of 280
Ma for the neopterygian crown, which we set as the root age. For our
molecular timescaling procedure, we constrained the age of three nodes
based upon the clock estimates of Near et al. (54)
(crown Neopterygii: 361.2 Ma; crown Holostei: 271.9 Ma; crown
Teleostei: 307.1 Ma). For both paleontolgical and molecular timescaling,
we used the “equal” method implemented in timePaleoPhy.
Quantifying Phenotypic Rates.
Size rates were quantified using the Bayesian approach of Eastman et al. (61)
implemented over 1,000,000 generations, discarding the first 250,000
generations as burn-in. Randomization tests (implemented via the
“compare.rates” function of the auteur package in R v2.15.3) provided a
two-way P value (α = 0.05) to test for differences between neopterygian partitions. The Adams (62)
method permits estimation of evolutionary rate on multivariate data,
and was applied to our shape dataset. Simulation of each supertree
topology under a null model of equal rates was used to generate a null
distribution of rate ratios for each of our five comparisons. The
observed rate ratio for a given comparison can then be compared with the
simulated distribution of rate ratios to derive a two-way P value to test for differences between two sets of taxa (α = 0.05).
Quantifying Phenotypic Innovation.
Blomberg’s K quantifies whether closely related taxa in a clade of interest are either more (K > 1) or less similar (K < 1) with respect to a trait value than expected under a Brownian motion model of evolution (K
= 1). Therefore, a more innovative clade (i.e., one that is efficient
at exploring new regions of trait space) should have a larger K
value than a less innovative clade. This is because the lineages of an
innovative clade should spread apart from one another, occupying
different regions of trait space so that more closely related taxa
appear more similar in trait value than more distantly related taxa. A
less-innovative clade should express lower K values, as
multiple lineages overlap and re-explore similar phenotypes, eroding
phylogenetic signal. Variation in rate of phenotypic change between
focal groups can, however, distort this simple relationship. For
example, in a clade that shows high rates of phenotypic change relative
to its boundaries in phenotypic space, K can be degraded (33, 34). K
values are interpreted to reflect innovation in the main text with
appropriate caveats given potential differences in rate, with all
comparisons further contextualized in Table S1.
Major Axes of Shape Variation and Their Anatomical Correlates
All relative warp axes that individually account for >5% of the variation across the species sampled are displayed in Table S3.
Morphospaces derived from the first three axes, containing all 398
Mesozoic neopterygian species in our shape dataset, are presented in Fig. S1. Images of sampled fossil specimens are also included in Fig. S1
to illustrate the anatomical correlates of shape axes. The positions of
major neopterygian clades in morphospace are indicated by different
colors in Fig. S2. Major teleost clades are presented in Fig. S2A and major holostean clades in Fig. S2B.
RW1 captures 42.53% of the variance, reflecting changes from slender-bodied taxa to deep-bodied taxa (Fig. S1 and Table S3). Highly positive scores on RW1 are dominated by Pycnodontiformes (Fig. S2A). The most negative scores on RW1 (highly slender bodies) belong to Aspidorhynchiformes (Fig. S2A).
RW2, representing 21.43% of variance, captures the position of the dorsal fin relative to the anal fin (Fig. S1 and Table S3).
Those taxa with highly positive scores have dorsal fins that insert far
anterior to the anal fin. Highly positive scores are dominated by the
Macrosemiiformes (a clade of stem gars; Fig. S2B), although numerous crown teleosts, stem teleosts, and halecomorphs also share this region (Fig. S2).
Highly negative scores characterize taxa whose dorsal fin inserts more
posteriorly than the anal fin, the most extreme example of which is the
pachycormid Euthynotus incognitus. Other Pachycormiformes, as
well as Aspidorhynchiformes, Ichthyodectiformes, “pholidophoriforms,”
and Osteoglossomorpha possess highly negative scores (Fig. S2).
RW3 explains 13.52% of variance and captures variation in the length of the dorsal fin base (Fig. S1 and Table S3).
Highly positive scores characterize taxa with small dorsal fin bases
relative to the length of the dorsal body surface, and include
Ellimichthyiforms (Clupeomorpha); some pholidophoriforms and a few
Ginglymodi and Halecomorphi (Fig. S2). The most negative scores (less than −0.2) are restricted to Macrosemiiformes (Fig. S2B).
RW4,
representing only 6.10% of the variation, appears to summarize
ventral-dorsal flexion, commonly exhibited by some fish after death.
Although care was taken to remove highly distorted taxa from the
dataset, some expressing mild degrees of bending clearly remain, and
appear to inform this axis. Therefore, we do not analyze variation in
RW4, and instead focus upon RW1 to 3.
SL vs. CS Comparisons
CS offers a measure of size independent of shape (63),
whereas SL does not. Therefore, we may expect to observe some
differences in the analytical results derived from these two types of
dataset. When the SL dataset is pruned to match the CS dataset for taxon
sampling, it yields near identical results regarding evolutionary rate
regardless of timescale choice (Fig. S3). The same is also true of innovation, where K distributions in taxonomic comparisons are very similar for both the pruned SL and CS datasets (Fig. S4). The larger SL dataset produces highly similar rate and innovation results compared with the CS and pruned SL dataset (Figs. S3 and S4).
The only minor differences observed occur exclusively in rate, where
the larger SL dataset recovers significantly higher rates in crown and
total group teleosts in a higher proportion of trees than the CS and
pruned SL datasets, regardless of timescale choice (Fig. S3).
However, these small differences have almost no influence on our
interpretations, because there is only a single instance where the
increased frequency of high rate results changes the overall conclusion
for a comparison: the crown teleost vs. stem teleost comparison on
molecular timescales. Here, the larger SL dataset delivers a majority of
trees where crown teleosts possess significantly higher rates than stem
teleosts, whereas the CS and pruned SL datasets find this result in a
large minority (Fig. S3).
Overall,
these results suggest that choice of size metric is relatively
unimportant for our dataset, and that the overall size and taxonomic
samplings of the dataset are more likely to influence subsequent
results, despite those factors having a relatively small influence here.
Nevertheless, choice of metric may be important for other datasets
(e.g., different groups of organisms or datasets of other
biological/nonbiological structures), because it is possible to envisage
scenarios where the choice of size metric would matter. For example, it
is possible that two fishes, identical in SL, could differ greatly in
CS if their shapes differed greatly, such as between an extremely
deep-bodied taxon (e.g., Opah) and a highly shallow-bodied taxon (e.g.,
needlefish). This discrepancy could present downstream effects on
analyses if a hypothetical “clade A” contained deep-bodied taxa that
regularly evolved into shallow-bodied taxa of similar SL (and vice
versa), whereas “clade B” consisted solely of similar sized deep-bodied
taxa. As a result, these two clades may show similar rates of evolution
using SL but different rates using CS (because clade A would be expected
to possess much higher rates of CS change). In our own data, the fact
that SL and CS datasets deliver near identical results suggests such
transitions are rare enough not to create such issues. This theory
matches what we expect given the topology of our tree, where organisms
that contrast the most in shape, such as highly deep-bodied (e.g.,
Pycnodontiformes) and shallow-bodied taxa (e.g., Aspidorhychiformes),
form distinct, distantly related clades, meaning rates will rarely be
calculated upon transitions between these distinct shape clusters.
The Importance of Timescale and Topology
Examination of Fig. 3 and Figs. S4 and S5
clearly demonstrates the ability for evolutionary timescale and
topology to alter specific rate and innovation analyses. For example,
timescale can dictate whether crown teleosts show higher size rates than
stem teleosts in a minority or a majority of trees (Fig. 3C), whereas topology can dictate whether holosteans or teleosts are more likely to possess higher rates of shape evolution (Fig. 3A).
Furthermore, the observation that molecular and paleontological
timescales can deliver differing results has considerable implications
for paleontology because the vast majority of analyses in the field are,
often by necessity, conducted solely upon paleontological timescales.
Therefore, where possible (when crown nodes are present in
paleontological trees to which various molecular estimates could be
applied) it should prove enlightening to incorporate the full range of
known timescale uncertainty, especially in cases where large mismatches
still persist between clock and rock estimates, such as for arthropods
(e.g., ref. 64) or land plants (e.g., ref. 65).
Similarly, paleontological analyses that do not contain crown nodes may
benefit from the introduction of additional timescale uncertainty, to
discover the temporal limits of their findings.
Our
analyses also provide insight into the relative importance of timescale
versus topology for understanding Mesozoic neopterygian evolution.
Although neopterygian timescales have estimated the origin of crown
teleosts to range anywhere from the Carboniferous to the Lower Jurassic (54, 66),
we demonstrate that even the most dramatic timescale differences under
our current methodology have little bearing upon our main findings.
Instead, the pool of sampled topologies typically set the range of
outcomes for a given analysis (Fig. 3 and Fig. S5),
while changes to the timescale subtly adjust the proportions of each
potential outcome, rather than bringing about conclusive outcomes in
favor of one or another hypothesis (Fig. 3 and Fig. S5).
These observations suggest refinement of the fossil neopterygian
phylogentic framework could be more important to Mesozoic-specific
questions than additional revisions to the neontological neopterygian
timescale. However, this does not undermine the importance of timescale
(indeed, changes to the topology themselves alter the timescale by
altering branch durations), but simply highlights that reductions in
uncertainty will be primarily delivered by improvements to
paleontological topology and timescale, obtainable via more systematic
studies of fishes and the creation of paleontological databases that
will permit use of probabilistic timescaling approaches (67).
Estimating the Position of the Genome Duplication upon the Teleost Stem
Hurley et al. (10)
performed a molecular clock analysis on the basis of four paralogous
genes. This approach allowed them to date not only the major divergences
between taxa, but the divergence between the paralogs themselves,
providing an estimate for the timing of the genome duplication event.
Hurley et al. (10)
performed clock analyses upon both a halecostome and a holostean
topology, and although the holostean topology is preferable in the light
of recent analyses (e.g., refs. 1 and 54); see also figure 2 in ref. 10),
both topologies still provide an estimate for the origin of the teleost
stem lineage. Different paralogue groups (a and b) also provide two
independent estimates for the age of the teleost crown (an estimate from
“a” paralogues and a separate estimate derived only from “b”
paralogues); we averaged these to provide a single crown estimate.
Finally, a clock estimate between the pairs of paralogues (a vs b)
delivered an absolute age estimate for the genome duplication event.
Altogether, these analyses delivered both the total duration of the
teleost stem, and an absolute estimate for the genome duplication event.
For three analyses (tables 3–5 in the supplement of ref. 10)
it was therefore possible to calculate the relative timing of the
genome duplication on the teleost stem, providing estimates of 62%, 54%,
and 64%, respectively, yielding an average estimate of 60% (Table S2). Therefore, current estimates suggest that the duplication occurred just over half way up the teleost stem lineage.
Acknowledgments
We
thank L. Sallan, R. Benson, R. Close, and L. Soul for useful
discussion; and the collections managers, curators, and research
scientists at numerous institutions for their assistance and access to
fossil specimens. This work was supported by a Palaeontological
Association Whittington Award and a Natural Environment Research Council
Cohort grant (to J.T.C.); Australian Research Council Grant DE140101879
(to G.T.L.); Philip Leverhulme Prize PLP 2012-130 (to M.F.); Natural
Environment Research Council Award NE ∕ I005536 ∕1 (to M.F.); and the
John Fell Fund (M.F.).
Footnotes
- ↵1To whom correspondence should be addressed. Email: j.clarke.paleo@gmail.com.
- ↵2Present address: Museum of Paleontology and Department of Earth and Environmental Science, University of Michigan, Ann Arbor, MI 48108-1079.
- Author contributions: J.T.C. and M.F. designed research; J.T.C. performed research; G.T.L. contributed new reagents/analytic tools; J.T.C. and G.T.L. analyzed data; and J.T.C., G.T.L., and M.F. wrote the paper.
- The authors declare no conflict of interest.
- This article is a PNAS Direct Submission.
- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1607237113/-/DCSupplemental.
Freely available online through the PNAS open access option.
Nenhum comentário:
Postar um comentário
Observação: somente um membro deste blog pode postar um comentário.