Fossil record of fishes and major evolutionary transitions

Full text

Fossil record of fishes and major evolutionary transitions
Tetsuto Miyashita
a,b,c
,
a
Beaty Centre for Species Discovery and Palaeobiology Section, Canadian Museum of Nature, Ottawa, ON,
Canada;
b
Department of Biology, University of Ottawa, Ottawa, ON, Canada; and
c
Department of Earth Sciences, Carleton University,
Ottawa, ON, Canada
© 2024 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Introduction: Relevance of fossil record to comparative fish physiology 3
An overview of the evolution of “fishes”3
Principles of phylogenetic bracket 9
Physiological adaptations at the origin of vertebrates 11
Contrasting hypotheses for the origin of jaws 11
Model-based approaches for tetrapod origins 12
Comparative biomechanics and ecological interactions 13
Conclusion 15
References 15
Key points
•Over the last half a billion years, fishes have achieved such great disparities that even the breathtaking diversity of modern
fishes represents only a patchy, non-random sample of their evolutionary history, sorted across many extinction events.
•In the absence of support from the fossil record, evolutionary hypotheses generated from a few experimental models may
become rendered tautological, non-falsifiable, or non-discriminable from alternative hypotheses.
•Phylogenetically informed comparison provides a powerful tool to generate and test evolutionary hypotheses, where fossil
forms are more than just calibration points of the tree.
Glossary
Clade A phyletic lineage defined by common ancestry.
Crown group A clade defined by the most recent common ancestry between two living taxa.
Endocast Conventionally a cranial endocast; infilling of the cavity within a neurocranium (braincase).
Ghost lineage A gap in fossil record where a hypothetical ancestral lineage is inferred to exist between the time of divergence
and the nearest occurrence of the taxon.
Histology Microscopic anatomy analyzed in sections at tissue- to cellular-levels.
Lagerstätte A major fossil-rich locality with exceptional state of preservation.
Maximum parsimony Phylogenetic criterion that favors the minimal number of character changes.
Node A point of divergence between clades; hypothetical common ancestry.
Stem group A grade of taxa closer to one crown group than to another.
Total group The most inclusive clade containing a crown and stem group (a crown group and all taxa closer to it than to
another crown group).
Abstract
The fossil record of fishes reveals history of bursts (radiation) and bottlenecks (extinction) that generated the largest
component of the living vertebrate diversity. Although patchy, fragmentary, and often difficult to interpret, fossil fishes
clearly point to cumulative past diversity far richer than can be inferred from the modern assemblage. Thus, fossil fishes
contribute more than just calibration points to the phylogeny. They are critical sources of phenotypic data that are otherwise
inaccessible in modern fishes, and insights generated from the fossils can discriminate competing evolutionary hypotheses.
This chapter briefly surveys the evolution of fishes and discusses how the fossil record informs current debates in or
peripheral to fish physiology.
Reference Collection in Life Sciences https://doi.org/10.1016/B978-0-323-90801-6.00179-8 1

Teaching slide
2Fossil record of fishes and major evolutionary transitions

Introduction: Relevance of fossil record to comparative fish physiology
Once species are viewed in dynamic continuum and not as static datapoints, any comparative work must generate evolutionary
insights, which are ultimately a reference to the common ancestry in geological time. The fossil record provides an independent
test of such inferences. Full implications of this seemingly self-explanatory tenet are difficult to grasp without a phylogenetic
treedsometimes so even with one. In the first cladistic (quantitative phylogenetic) treatment of fish physiology, Sven Løvtrup
blasted that fossils are meaningless and that only extant animals can be properly compared; he then ended up reaching puzzling
conclusion thatdfor exampledcoelacanths and chondrichthyans are each other’s closest relatives (see where these lineages are on
the tree in Fig. 1). Ironically, paleoichthyologists in Løvtrup’s time such as Colin Patterson and Gareth Nelson were leading the
charge in the rise of cladistics. Eventually, the relationships of major vertebrate lineages were to be untangled using the cladistic
methodology Patterson and Nelson helped develop and the set of characters Løvtrup began formulating.
It is now widely appreciated that fossil record provides the depth and shape of a comparative framework (which often comes in
the form of a phylogenetic tree). Despite the general utility, however, fossil record and phylogenetic inferences are sometimes inter-
preted frivolously where only convenient information (such as estimated timings of the most recent common ancestry) is cherry-
picked but counterevidence ignored. This happens typically when insights from the fossil record do not fit a preconceived idea of the
hypothetical ancestor. The vertebrate fossil record often shows much wider disparity and much more complex evolutionary history
than can be derived from modern diversity. So the problem becomes predictably acute when one or more of the extant models are
each used as a surrogate ancestor or a general representative of a larger taxonomic group (e.g., a modern shark treated as a represen-
tative gnathostome). This chapter presents a very brief overview of the fossil record of fishes, and examines a few recent issues in
comparative biology of fishes on which the fossil record has bearing.
An overview of the evolution of “fishes”
The living vertebrate clade (vertebrate crown) is an ancient lineage that extends at least half a billion years. Precise timings of their
origin remain unclear within a loosely constrained window of nearly 200 million years (636e457 Ma). Support has been building
for conodonts (Fig. 2)dan enigmatic lineage of small fish-like animals with mineralized teethdto sit within the vertebrate crown or
just outside it. The earliest paraconodonts occur in the beginning of the Cambrian Period (541 Ma), so accepting their vertebrate
position would push the age of vertebrates at least to that point. There are vertebrate-like forms known from the Early to Middle
Cambrian fossil assemblages: myllokunmingiids, yunnanozoans, and Metaspriggina (Fig. 2). So far, evidence is lacking to place
them within the vertebrate crown. Instead, they are typically nested outside it as stem vertebrates (Fig. 1). But these Cambrian fossils
might appear more primitive than they are because their soft bodies are prone to decay before fossilization, missing precise anatom-
ical details that could help placing them on a phylogenetic tree. These Cambrian “fishes”invariably possess a diminutive head with
an enormous branchial region, implying that the prominent face derived from the vertebrate-specific cell lineage, neural crest, may
have been a later innovation.
The living vertebrate group is split into two different clades: Cyclostomi and Gnathostomata (Fig. 1). Cyclostomes include the
two living lineages of jawless vertebrates, hagfishes and lampreys. Although hagfishes and lampreys probably diverged from each
other by the end of the Ordovician Period (485e444 Ma), they only appear in the fossil record at the end of the Devonian Period
(359 Ma), and only a handful of fossil forms are known from subsequent ages. Two extinct groups of Paleozoic jawless fishes may
also belong to the cyclostome stem: anaspids and conodonts (Fig. 2). Anaspids have mineralized scales, whereas conodonts have
mineralized “teeth.”The implication here is that, despite the primitive appearance of hagfishes and lampreys, they are highly
specialized vertebrates that secondarily lost the three main types of mineralized tissues in a vertebrate skeleton: bone, dentin,
and enamel.
Once past hagfishes and lampreys, all other living vertebrates are gnathostomes. The Gnathostomata are generally understood as
a clade of jawed vertebrates. However, many extinct lineages of jawless vertebrates share more recent common ancestors with living
jawed vertebrates than with hagfishes or lampreys (Fig. 1). These are, in the order of branching from the gnathostome root, pter-
aspidimorphs (arandaspidiforms and heterostracans), thelodonts, galeaspids, pituriaspids, and osteostracans (Fig. 2). Arandaspidi-
forms (arandaspidids, astraspidids, eriptychiids) occur in the Ordovician times (485e444 Ma), and the thelodont record also
extends to the Late Ordovician. Heterostracans were the most abundant jawless fishes from the Silurian (444e419 Ma) to Devonian
(419e359 Ma) periods and also among the longest surviving stem gnathostomes. Galeaspids and osteostracans are important
because their internal anatomy has been extensively studied. They reveal a mosaic of primitive features (the pharyngeal anatomy
is generally similar to lampreys) and derived traits (the endocasts are somewhat intermediate between cyclostomes and jawed verte-
brates). In particular, osteostracans have cellular bone, pectoral fins with endoskeleton, and an epicercal caudal findall of which are
considered as derived gnathostome conditions. Except for arandaspidiforms (exclusively Ordovician) and pituriaspids (Early Devo-
nian), most jawless stem gnathostome lineages appear in the fossil record during the Silurian Period, reached the highest diversity
just before or during the Early Devonian times, and were mostly extinct by the Middle Devonian times. Many stem gnathostomes,
be they jawed or jawless, only appear in the fossil record long after their probable origins, and there is no clear correlation with the
so called Great Ordovician Biodiversification Event. If anything, the end-Ordovician mass extinction preceded the rise of fishes to
the prominence in the Silurian ecosystems, but the causal link between the two phenomena remains to be tested.
Fossil record of fishes and major evolutionary transitions 3

Fig. 1 A simplified phylogeny for an overview of fish diversity. A majority of the order-level clades of fishesdextinct or extantdhas been
assembled according to the conventional understanding of their interrelationships. Polytomies indicate the lack of consensus among analyses (see
text for the explanation of major phylogenetic conflicts and sources of the phylogenetic hypotheses). The terminal branches containing living taxa are
indicated in red; stem lineages and extinct branches are in dark. Node colors indicate estimated timing of divergence. Multiple independent lineages
have been abbreviated to simplify the scheme. The names of these abbreviated para- and polyphyletic groups are shown in 50% opacity. Many
clades have been entirely omitted because they are minor, obscure, uncertain, poorly understood, and/or non-critical to the overall understanding of
the current fish phylogeny. In particular, several such clades are missing in the holocephalan stem, neopterygian stem, teleost stem, and percomorph
crown. The labels for acanthodians and placoderms are set in a box to indicate that these are paraphyletic assemblages (grades). A zebrafish is
shown in original color to illustrate the sense of proportion for a single experimental model to represent a larger, highly diverse clade such as
ostariophysans, teleosts, actinopterygians, osteichthyans, or even gnathostomes.
4Fossil record of fishes and major evolutionary transitions

Confusingly, the first jawed vertebrate is neither the first gnathostome nor the most recent common ancestor of all living gna-
thostomes (Fig. 1). The total gnathostome node contains all vertebrates closer to Ichiro Suzuki (Homo sapiens) than to a sea lamprey
(Petromyzon marinus), inclusive of jawless stem gnathostomes. The crown gnathostome node defines the last common ancestor
between Suzuki and a dogfish (Squalus acanthias) and excludes a major assemblage of jawed stem gnathostomes. This major group
of jawed stem gnathostomes is known as the Placodermi, so the first jawed vertebrate wasdso far as we knowda placoderm. For the
purpose of this overview, placoderms provide two key insights. First, the range of anatomical variation among placoderms far
exceeds the general depiction of this group represented by the hypercarnivorous Dunkleosteus (Fig. 3). Once controlling for
Fig. 2 Diversity of fossil jawless vertebrates. Not to scale. See Fig. 1 for their interrelationships.
Fig. 3 Jawed stem gnathsotomes, conventionally referred to as “placoderms.”Interrelationships of placoderms remain contentious (see many early
branches collapse into a polytomy, which is a point of phylogenetic uncertainty). Representative forms were illustrated by Philippe Janvier. Not to
scale.
Fossil record of fishes and major evolutionary transitions 5

differences in species richness and chronological existence, the disparity among placoderms would qualitatively rival that for the
total group of chondrichthyans. Second, relationships among placoderm lineages remain unresolved. There is good evidence for
each distinct placoderm lineage as a natural group (acanthothoracids, antiarchs, arthrodires, brindabellaspidids, maxillates, petal-
ichthyids, pseudopetalichthyids, ptyctodonts, rhenanids, stensioellids), but there is little consensus from one analysis to the next
about their interrelationships. Antiarchs can be among the earliest branching placoderms or the proximate lineage to the gnathos-
tome crown. Acanthothoracids, brindabellaspidids, pseudopetalichthyids, rhenanids, and stensioellids are also considered primi-
tive, whereas maxillates form a probable sister group to the gnathostome crown (Fig. 3). The hypothesis that all or most placoderm
lineages form a clade has not been entirely ruled out yet. Anatomically, some exceptionally preserved placoderm fossils have gener-
ated important and surprising discoveries about the physiology and ecology of early-branching jawed vertebrates. These include
a sigmoidal heart, livers, muscles, and even umbilical cords attached to the embryos in utero (Long, 2010;Trinajstic et al.,
2022). As for reproduction, fertilization is internal for placoderms and live birth occurs at least in ptyctodonts. Substantial variation
exists in the intromittent organs: the claspers are modified parts of the dermal skeleton in antiarchs, whereas they are internally
ossified in arthrodires and ptyctodonts (unlike chondrichthyans, the placoderm claspers are independent of the pelvic girdle)
(Long et al., 2015). A recent discovery in China put the earliest occurrence of placoderms in the Early Silurian Period at 436 Ma
(Zhu et al., 2022).
All stem cyclostome and stem gnathostome lineages mentioned thus far (except for conodonts) were extinct by or at the
Devonian-Carboniferous boundary (359 Ma). The end-Devonian mass extinction consists of a series of distinct events, one of which
impacted vertebrates disproportionately and is known as the Hangenberg event (Sallan and Coates, 2010). This was the most conse-
quential of all mass extinctions in vertebrate evolution because the major crown gnathostome lineages (actinopterygians, chon-
drichthyans, and sarcopterygians) replaced the pre-Hangenberg vertebrate fauna and went on to shape the spectacular diversity
of vertebrates we recognize today (Fig. 4). These lineages existed long before the end-Devonian mass extinction, and chon-
drichthyans and sarcopterygians each had high disparity and species richness across the Devonian Period. However, the post-
Devonian vertebrate fauna has little continuity from the Devonian counterpart at the taxonomic levels of orders and families (Sallan
and Coates, 2010).
On one side of the gnathostome crown (the living group of jawed vertebrates), chondrichthyans (cartilaginous fishes) have the
longest fossil record among any living vertebrate lineages. Chondrichthyan-like scales occur as early as arandaspids and astraspids in
the Middle Ordovician strata, but the earliest teeth and body fossils are from the Early Silurian Period at 436e439 Ma (Andreev
et al., 2022). The Ordovician chondrichthyans would force a minimum 30e40 My-long window of missing record for other jawed
vertebrate lineages. Filling this gap remains as a great challenge of vertebrate paleontology. One great advance in this field over the
last decade and a half has been the growing consensus that acanthodiansda Paleozoic group of fishes characterized by fin spines
(Figs. 4 and 5A)dpopulate the chondrichthyan stem (Fig. 1)(Brazeau and Friedman, 2015). Many anatomical differences between
’typical’acanthodians and modern chondrichthyans indicate that the latter is highly derived, despite the widespread assumption
that sharks and skates are a paragon of primitive vertebrate conditions. Indeed, the Devonian taxa such as Janusiscus and Ramirosuar-
ezia continue to shift their positions from just outside the gnathostome crown to either the chondrichthyan or osteichthyan stem,
thereby shaking up and shuffling branches in this part of the tree (Fig. 1). This instability illustrates how anatomically similar the
earliest-branching chondrichthyans are to osteichthyans and stem gnathostomes.
Fig. 4 Consequences of the Hangenberg event (a major pulse of the end-Devonian mass extinction). A representative shallow marine fauna in the
Late Devonian times (Bad Wildungen) is dominated by morphologically diverse and speciose placoderms. In a counterpart fauna from the mid-
Carboniferous (late Mississippian) times (Bear Gulch), these ecomorphs are occupied by the diversifying chondrichthyans and actinopterygians. Color
codes: blue ¼placoderms; green ¼chondrichthyans; red ¼actinopterygians; yellow ¼sarcopterygians. Modified from Friedman and Sallan (2012).
6Fossil record of fishes and major evolutionary transitions

The chondrichthyan crown (the living group of chondrichthyans) is split into the holocephalan (including chimaeras) and the
elasmobranch (including sharks, rays, and skates) total groups (Fig. 1). In the fossil record, stem holocehaphans were much more
diverse during the Carboniferous (359e299 Ma) and Permian (299e252 Ma) periods than the modern representatives suggest,
including strange forms like chondrenchelyiforms, eugeneneodontids, iniopterygians, menaspiforms, and petalodontiforms.
Each of these lineages acquired a unique combinations of unparalleled morphology (e.g., tall attachment for wing-like pectoral
fins in iniopterygians; symphysial tooth whorls in eugeneodontids) and traits convergent with elasmobranchs and actinopterygians
(e.g., anguilliform body profile of chondrenchelyiforms; crushing dentition in petalodontiforms) (Fig. 4). There is also an emerging
consensus for a stem holocephalan assemblage referred to as symoriiforms, which also includes the classically “shark-like”Devo-
nian taxon Cladoselache and stethacanthids (another strange group with the hypertrophied first dorsal fin) (Fig. 4). On the elasmo-
branch stem, three major branches thrived across the late Paleozoic into the Mesozoic eras: xenacanthiforms, ctenacanthiforms, and
hybodontiforms in the order of divergence (Fig. 5B). There was a significant loss of diversity across the end-Triassic mass extinction
(201 Ma) for stem elasmobranchs, but hybodontiforms survived as a species-rich group until the end of the Cretaceous Period
(66 Ma). The loss was effectively filled in by the Mesozoic diversification of the elasmobranch crown group (the living group of
sharks and skates). Major living lineages of elasmobranchs appeared in the fossil record during the Jurassic (201e145 Ma) and
Cretaceous (145e66 Ma) periods. Despite the bottleneck across the end-Cretaceous extinction and a slow recovery that followed,
their overall ecological diversity was restored in the Cenozoic times and largely maintained to the modern day (Fig. 5C). Interre-
lationships of crown elasmobranchs both within and among the orders remain in flux, with the most fundamental controversy
being the placement of batoids (rays and skates) as the sister group to sharks or nested among squalomorphs. Recent molecular
phylogenetics favor the former hypothesis. Among extinct elasmobranch lineages, synechodontiforms form a clade outside
Fig. 5 Representative fossils of crown gnathostomes (the clade originated in the last common ancestor of all living jawed vertebrates) preserved in
different modes and in various states. AeC, chondrichthyans; D, sarcopterygian; EeH, actinopterygians. (A) Diplacanthus acui (Albany Museum
specimen number 5739), an acanthodian (stem chondrichthyan) from the latest Devonian times, preserved in carbon-rich black metashale of the
Waterloo Farm Laggerstätte. Hard and soft tissues are preserved in kaolinite (clay). Photograph by coutersy of Robert Gess, (B) A hatchling of
Bandringa rayi (Field Museum of Natural History specimen PF 15382), a stem elasmobranch (crown chondrichthyan) from the Late Carboniferous
times, preserved in an iron-rich nodule of the Mazon Creek Laggerstätte. Soft tissues are preserved as stains and impressions, often infilled by
precipitates of secondary minerals. (C) Galeorhinus cuvieri (Museo Geologico Giovanni Capellini specimen 1976), a carcharhiniform shark from the
Eocene times, preserved in the micritic limestone of Pesciara di Bolca Lagerstätte. Soft visceral and nervous tissues, such as the intestine and brain,
are preserved in this specimen, along with a stomach content of a barracuda. (D) Eusthenopteron foordi (Cleveland Museum of Natural History
specimen 8158), a tristichopterid stem tetrapod from the Late Devonian times, preserved three-dimensionally in the sandstone bed of the Miguasha
Lagerstätte. (E) Cheirolepis canadensis (Royal Ontario Museum specimen 1936), a cheirolepiform (stem actinopterygian) from the Late Devonian
times, preserved in two-dimensional, laterally compressed state in the mudstone of the Miguasha Lagerstätte. (F) Canobius modulus (Royal Ontario
Museum specimen 1227), a potential stem actinopterygian of uncertain affinity from the Early Carboniferous times, preserved in two-dimensional,
laterally compressed state in the shale of the Albert Mine Lagerstätte. Some soft tissues (e.g., eyes) are preserved as carbon-rich stains. (G)
Boreosomus gillioti (Smithsonian Institute specimen 21,468), a ptycholepiform (potential stem neoptegyan) from the Early Triassic times, preserved
three-dimensionally as a mold (negative print) within an iron-rich sandstone nodule from Ambaracasaca. The dermal skeleton is visible here after all
the internal skeleton and organs had dissolved. (H) A braincase of Cladocyclus pankowskii (Smithsonian Institute specimen 521,360), an
ichthyodectiform (stem teleost) from the Early Cretaceous times, preserved three-dimensionally as an isolated element from the red sandstone of the
Kem Kem beds. Specimens not to scale.
Fossil record of fishes and major evolutionary transitions 7

galeomorphs and squalomorphs, although their relationship with batoids remains uncertain. Another key taxon for estimating ages
of the living elasmobranch groups, Protospinax is a Late Jurassic crown elasmobranch that likely falls among squalomorphs.
The osteichthyan (bony fishes) clade has a sparsely populated stem and a spectacularly diverse crown (Fig. 1). The Silurian forms
such as Andreolepis and Lophosteus occupy the osteichthyan stem, and there are other Silurian to Devonian forms that may fall within
or outside the osteichthyan crown: Achoania,Dialipina,Guiyu,Ligulalepis,Meemania, and Psarolepis. Depending on their placement,
the hard minimum age for the osteichthyan crown can be as old as 425 Ma (Late Silurian Period).
Sarcopterygians (lobe-finned fishes) achieved high levels of diversity and disparity during the Devonian times mostly as large,
carnivorous fish. Beside the three living lineages (coelacanths, lungfishes, and tetrapods), onychodontids (potential sister group to
actinistians), porolepiforms (sister group to dipnoans), rhizodontiforms, canowindrids, osteolepiforms (megalichthyiforms), tris-
tichopterids, and elpistostegalians (sister group to tetrapods) each went through a rapid rise and fall during the Devonian Period
(Figs. 1 and 5D). All branches after the dipnoan total group sit on the tetrapod stem and thus have received extensive scrutiny over
the last half a century to understand the fin-to-limb transitiondparticularly elpistostegalians. The earliest body fossils of tetrapods
come from the Late Devonian times. After the Hangenberg event (259 Ma), however, there is a notable gap in the fossil record of
tetrapods that could otherwise link the tetrapod crown with the Devonian forms such as Acanthostega and Ichthyostega. This is known
as Romer’s Gap, but the tetrapod fossil record is beyond the scope of this review. It suffices to mention that both actinistians (coela-
canths) and dipnoans (lungfishes) continued as a relatively speciose group of predatory fish into the Mesozoic Era, but now only
a few species each survive at the tip of such a long, skinny branch that some refer to them as the “living fossils.”
Actinopterygians (ray-finned fishes) are by far the most successful major lineage of vertebrates, taking up half the living verte-
brate diversity: one out of every two living vertebrate species is an actinopterygian! Despite their evolutionary and ecological impor-
tance, early actinopterygian relationships remain in contention. There is little dispute about the stem actinopterygian status for
some Devonian taxa such as Cheirolepis (Fig. 5E) but others such as Moythomasia (Fig. 4) may fall inside or outside the crown. There
is an enormous assemblage of extinct actinopterygians that has been, for the lack of better taxonomic assignment, referred to as
palaeonisciforms or palaeoniscoids (Fig. 5F and G). Some chronologically younger forms in this assemblage are seemingly better
compared with modern actinopterygian lineages. “Palaeoniscoids”are certainly a wastebasket category for indeterminate actino-
pterygians, and the standing challenge is to find their branches in the actinopterygian tree. Some, many, or most of them may
form sister groups to the living actinopterygian lineages from cladistians (bichirs) to neopterygians, sit among them, slip more
stem-ward than them, or cluster more crown-ward than them (Fig. 1)(Sallan, 2014). One consequence of these different topologies
is the lack of consensus on estimated ages for many actinopterygian groups. Nevertheless, recent studies have begun to cultivate
previously inaccessible data from the internal anatomy of these fossils, and the growing consensus points to a large stem assemblage
that represents radiations before and after the end-Devonian mass extinction. This is a rapidly moving front of vertebrate paleon-
tology today.
Interspersed among these Paleozoic forms, the extant non-teleost lineages also diverged (Fig. 1). These are cladistians, chondros-
teans (sturgeons and paddlefishes), and holosteans (gars and bowfins). Among them, holosteans and teleosts form a clade Neo-
pterygii. Each of these four living actinopterygian lineages (cladistians, chondrosteans, holosteans, and teleosts) currently suffers
the lack of early fossil record and has a nearly 100 My-long ghost lineage (Friedman, 2015). Characteristically, the stem assemblage
for each crown group is mainly known from the Mesozoic times (scanilepiforms for Polypterus; semionotiforms for gars; parasemio-
notiforms and ionoscopiforms for bowfins; pycnodontiforms, pachychormiforms, aspidorhynchiforms, phoridophoriforms, lepto-
lepiforms, ichthyodectiforms, and others for teleosts) (Fig. 5H). This problem may be resolved with better understanding of the
phylogenetic relationships of the Carboniferous and Permian actinopterygians. In that sense, a recent breakthrough was the rein-
terpretation of scanilepiforms as stem cladistians, thereby greatly reducing that ghost lineage for this deepest branch within the acti-
nopterygian crown (Giles et al., 2017).
After these early actinopterygian lineages comes the colossal clade Teleostei. It requires an encyclopedia of its own to review all
major teleost clades, and this is certainly beyond the scope of this review. Teleosts are so speciose and so diverse that no reliable
morphological traits are known to unite all the ingroups. Despite this explosive diversification, their early fossil record is poor;
most stem teleosts are known from the Mesozoic Era and their relationships remain in flux (Friedman, 2015). One defining feature
of teleostsdat least that of the crowndmay be the whole genome duplication, as all living teleost lineages have inherited signatures
of a duplicated genome where homologous genes (paralogues) occur in the same order (synteny) on different chromosomes. This is
the third round (3R) of whole genome duplication in vertebrate evolution, where the first (1R) occurred before the vertebrate crown
and the second (2R) sometime before the gnathostome crown. Mesozoic diversification of neopterygians (inclusive of teleosts) saw
the development of ecological guilds and morphotypes similar to modern counterparts, only that these roles were occupied by
different lineages (e.g., the laterally compressed, deep-bodied, durophagous pycnodontiforms reminiscent of modern reef duroph-
ages). One ecomorph of the Mesozoic neopterygians missing in the modern teleost diversity is giant pelagic suspension feeders,
which in Mesozoic times were represented by pachycormiforms like Leedsichthys. After the Cretaceous Period, giant pelagic suspen-
sion feeders evolved among chondrichthyans and cetaceans, potentially excluding neopterygians from re-entering this ecological
space.
In the aftermath of the K-Pg event (end-Cretaceous mass extinction; 66 Ma), two major radiations became apparent in the
recovery fauna: Acanthomorpha (Acanthopterygii in particular) and Ostariophysi, each a teleost mega-clade that dwarfs all other
contemporaneous vertebrate lineages in species richness (>10,000 spp.) and morphological disparity. The former is nested within
the Euteleostei, whereas the latter sits in the Otocephala. The Euteleostei and Otocephala form an ingroup clade of teleosts (Clu-
peocephala) to the exclusion of elopomorphs and osteoglossomorphs (Fig. 1). The fossil record extends well into the Cretaceous
8Fossil record of fishes and major evolutionary transitions

Period for both acanthomorphs and ostariophysans. The earliest crown ostariophysan (Rubiesichthys, 126 Ma) is older than the
earliest crown acanthomorph (Stichocentrus, 98 Ma). In ostariophysans, two out of the five orders (gonorhynchiforms and characi-
forms) are known from the Cretaceous times. With several exceptions (e.g., stem lampridiforms), most order-level acanthomorph
clades appear in the fossil record in the Cenozoic Era. The explosive diversification of acanthomorphs during the Paleogene and
Eocene outpaced other vertebrate lineages (Friedman, 2010), and (although resolved to a lesser degree at this point) ostariophysans
also achieved similar levels of species richness across the Cenozoic Era, mostly in freshwater systems. This habitat constraint makes
ostariophysans a great model system in biogeography where the fossil record is a critical source of calibration points with which to
scale the phylogeny over time. In the marine realm, modern biodiversity hotspots occur in the coral reef systems, which are domi-
nated by acanthomorphs. The rise of reef communities is correlated with the Paleogene-Eocene diversification within that clade;
therefore, it is possible to compare distinct ecomorphs for similarities, differences, and phylogenetic structures between the modern
and fossil faunas.
As can be seen in these two mega-clades, the Cenozoic fossil fishes in general facilitate modern analyses of evolutionary
dynamics both as a data source and as calibration points. This general importance rings particularly true in the two intensively
studied models of adaptive radiation (sticklebacks of the Northern Hemisphere and cichlids of the Great Rift Valley). In the case
of sticklebacks, the fossil beds formed in seasonal, annual, or decadal cycles allow tracking populations over tens to hundreds of
thousands of years. Combined with modern genetics, such reciprocally calibrated datasets can reveal tempo and mode of speciation,
strength and direction of selection, and rise and fall of a population with unprecedented precision.
Many significant lineages, events, and localities are omitted in this cursory overview of the fossil record of fishes. These topics are
treated in more depth and greater precision by excellent reviews elsewhere. To name just several, for the Paleozoic fishes Early Verte-
brates (Janvier, 1996) has served the community for decades as a timeless masterpiece, and The Rise of Fishes (Long, 2010) provides
a great introduction to the topic and offers dazzling specimen photography. For the living lineages, Fishes of the World (Nelson et al.,
2016) stands as a taxonomic compendium, although the phylogeny is a rapidly moving front. A synthetic review by Friedman and
Sallan (2012) merits a special mention for tracking the fish evolution by geological events and patterns of diversity through time.
Forward and reverse literature search starting from these titles will provide a vantage point of fish evolution over the last half a billion
years.
Principles of phylogenetic bracket
In modern biology, comparison assumes evolutionary relationships where characters vary in their states and have polarity from
primitive to derived. The simplest cladistic scheme of such comparison is a three-taxon tree, and maximum parsimony (the algo-
rithm that favors the smallest possible number of changes) is used as a criterion to discriminate potential scenarios of character
transformation. In the example (Fig. 6A), Taxon A is a sister group to the clade Taxon B þTaxon C, and the smallest number of
changes required to produce the distribution of the character states is that the state 1 arose in the last common ancestor of Taxon
B and C. However, placing this in a larger tree or scaling the branches to time illustrates the problem underlying this theoretical
construct. In reality, homoplasies (independent evolution of character states) are pervasive across phylogeny; the likelihood of char-
acter change increases over time; and a single outgroup taxon is (even under parsimony) never enough to polarize a character tran-
sition because it requires more outgroups to test the assumption that Taxon A retained an ancestral condition. So any three-taxon
comparison in practice requires four or more taxa, and multiple outgroups contribute outside the frame of comparison to anchor
a primitive condition.
This cladistic thinking demonstrates the value of fossil record. It is needed in modern comparative biology to: (a) calibrate the
depths of common ancestries; and (b) document patterns of character distribution through time. On one hand, the fish fossil record
(Fig. 1) illustrates that the use of fossil data involves more than just searching for the oldest occurrence of the living group of interest
(because there may be extinct relatives that push origins of the lineages or characters under investigation) or adapting a single fossil
outgroup to establish character polarity (because that fossil itself may be derived already). On the other hand, modeling a hypothet-
ical ancestor after a living taxon uncritically is to downplay the fact that evolution is complex. After a good long stare at the
phylogeny (a dramatically abbreviated summary tree is presented in Fig. 1)dafter numerous divergences and adaptive radiations,
the rises and falls of lineages over 100 s My, and cataclysmic mass extinctions and many more species-sorting eventsdwhat are the
chances that a living taxon can serve as a surrogate for a common ancestor when its extinct relatives look nothing like it?
Paleontologists have faced this dilemma in a reverse direction. The fossil record is inherently biased toward hard tissues. To
reconstruct soft tissues in a fossil taxon, it usually requires: (a) skeletal correlates (e.g., muscle attachment on a particular bony
process); and (b) extant models that possess a homologous tissue. The Extant Phylogenetic Bracket (Witmer, 1997) was generated
to frame this comparison in a phylogenetic context (Fig. 6A). It compares distribution of the skeletal correlates between extant taxa
that “bracket”the fossil taxon in question. The skeletal correlates may be present in all three taxa (Level III inference), in one of the
extant taxa and the fossil taxon (Level II inference), or only in the fossil taxon (Level I inference). The levels of inference essentially
reflect degrees of phylogenetic congruence for the skeletal correlates to be used as evidence for the non-preserved soft tissue. The
theoretical framework for the Extant Phylogenetic Bracket is broadly applicable to other comparative questions and has been
used implicitly or explicitly in the literature.
The value of this bracketing approach is its ability to compare relative fit of competing hypotheses to the existing phylogeny
(Fig. 6B). An evolutionary narrative often emphasizes whether one hypothesis is compatible with phylogeny, but alternative
Fossil record of fishes and major evolutionary transitions 9

Fig. 6 Phylogenetically informed comparison can test differential fit of hypotheses. Character code: 0 ¼absent; 1 ¼present. Extinct taxa are shown
in gray scale. (A) The concept of Extant Phylogenetic Bracket. This is a classical three-taxon problem. (B) Outgroup comparison can distinguish
conflicting hypotheses. Within the phylogenetic bracket, H1 (independent gains) and H2 (independent losses) are equivalent for both Trait 1 and 2.
With the information on the outgroup taxon, one hypothesis can be shown to better fit the phylogeny than the other (H1 for Trait 1; H2 for Trait 2).
(C) Comparison of two hypotheses for the origin(s) of a filter-feeding larval phase with just living taxa. It is not clear whether the larval phase of
a lamprey is derived independently from invertebrate chordates (H1) or a retention of the primitive feature (H2). (D) Comparison of two hypotheses
for the origin(s) of a filter-feeding larval phase including the fossil data from Paleozoic stem lampreys. Four out of five Paleozoic stem lampreys have
been conclusively shown to lack a filter-feeding larval phase. Inclusion of these taxa favors the interpretation that the filter-feeding larval phase
evolved in lampreys independently. (E) Separating cause and effects and testing the evolutionary narrative for gill functions. Comparison of the
deuterostome taxa (phylogenetic bracket 1) suggests that the gills acquired respiratory function only at the origin of vertebrates. Because this
comparison hinges on the larval lamprey being a functional analog to the vertebrate ancestor, however, it is not clear whether the gills in a larval
lamprey perform gas exchange because of the ancestry or because of the combination of physiological variables. In theory, this can be tested in
future using other living vertebrates that can separate some of these variables (phylogenetic bracket 2). Simultaneously, each bracket serves as an
extant phylogenetic bracket for stem taxa. (F) Conflicting interpretations and phylogenetic fit of jaw-origin hypotheses. The Gill Arch and Assimilation
hypotheses both accept serial nature of the pharyngeal arches and are equivalent with one another, but make different interpretations on the anatomy
of Arch I (mandibular arch) in fossil taxa. These differences percolate to hypothesis fitting on the tree.
10 Fossil record of fishes and major evolutionary transitions

hypotheses may also be phylogenetically consistent. Therefore, the main question is what fits the phylogeny better. In its simplest
form, the Phylogenetic Bracket is a three-taxon comparison; so a choice of terminal taxa will require some arbitrary decisions when
the character evolves in a complex manner (e.g., when reasonable evidence suggests that the closest outgroup lost the ancestral trait
and evolved a specialized condition). For the purpose of this review, it suffices to state that: (a) as many outgroups as relevant
should be considered outside the bracket of three taxa; and (b) whether the outgroup taxon is primitive or derived requires an inde-
pendent test.
Physiological adaptations at the origin of vertebrates
The filter-feeding larval phase of modern lampreys has served as a classic model of the vertebrate ancestry. With comparison made
mainly to invertebrate cephalochordates (amphioxus), their similarities in ecology and anatomy were interpreted as the retention
from their shared ancestry, rather than as convergence due to similar lifestyles as sand-burrowing filter feeders. These two scenarios
were previously equivalent on phylogeny under the parsimony principle (Fig. 6C) because it was unclear at what point in vertebrate
evolution the filter-feeding life phase would become lost. However, recent fossil discoveries support the latter scenario of conver-
gence. Stem lampreys from the Devonian and Carboniferous periods did not undergo a filter-feeding larval phase. As these taxa
represent at least three independent branches off the lamprey stem, the filter-feeding larval phase of modern lampreys is now an
isolated condition of the crown group, likely a secondarily derived state (Fig. 6D). This does not mean that modern lamprey larvae
are useless in comparison; only that they make a poor surrogate ancestor. The larvae could, at best, serve as an analog, within the
bounds of phylogenetically informed comparison.
This unequivocal fossil evidence now requires comparative studies to justify incorporating lamprey development in the discus-
sion of the origin and early evolution of vertebrates. For example, a recent comparison of gill functions among a hemichordate,
a cephalochordate, and a lamprey larva (Sackville et al., 2022) offers fair rationales that their choice of models is based on “being
small, worm-shaped burrowers with filter-feeding gills”and that, convergent or not, these traits must “constrain the function of gills
and skin [.] in similar ways.”Having said that, one side effect of treating the lamprey larva as an analog is the difficulty to distin-
guish phylogeny (similarities due to common ancestry) from convergence (similarities due to function). In this case, Sackville et al.
(2022) found that the gills are the dominant site of gas exchange only in larval lampreys (with increasing body size) among the
three models. They conclude that this function likely evolved in vertebrates, but it is not clear whether as an adaptive change or
as a product of allometric constraints such as flow rates over the gills. Indeed, teleost larvae are typically smaller than lamprey larvae,
and the skin is the dominant site of gas exchange because of high surface:volume ratio and because the flow through the branchial
chambers are limited under low Reynold’s number. These observations raise the possibility that the gill functions, particularly gas
exchange, may have evolved passively with physiological constraintsde.g., more as a consequence of increasing flow volume in the
gill chambers than as an adaptation to facilitate high metabolic rates of vertebrates. The cause and effects are not clearly separated
here.
To test these alternative hypotheses, a comparative framework will need to incorporate the taxa that allow discriminating these
possible correlates (Fig. 6E). The comparative analysis should then shift from early-branching chordates to crown-group vertebrates,
survey a variety of teleost larvae, or incorporate an outgroup that evolved gas exchange at the gills independently. From the fossil
record, the presence of dermal skeleton is strongly inferred for the last common ancestor of all living vertebrates. The forms
belonging to this “ostracoderm”assemblage, such as anaspids and arandaspids, are covered with mineralized scales and typically
no smaller than 5 cm in body length. Given that they also have a prominent branchial apparatus, the dominant site of gas exchange
was likely in the gills and not the skin. However, such an active lifestyle precedes the vertebrate crown group, and the Cambrian stem
vertebrates are smaller than most ostracoderms, naked, and associated with a prominent branchial apparatus. If scaling relation-
ships for gas exchange can be identified in extant taxa against body size, body and gill surface areas, and dimensions of the branchial
apparatus, this may provide a range of estimates for these Cambrian forms.
Contrasting hypotheses for the origin of jaws
The origin of the vertebrate jaws represents an acquisition of novel morphology rather than of specific physiological adaptations,
but it is relevant to comparative physiology and biomechanics because of the magnitude of functional changes. Jawed vertebrates
feature a long list of sensory, ventilatory, locomotory, and metabolic correlates that do not occur in jawless vertebrates, including:
three semicircular canals, the dual pumping mechanism for ventilation, two sets of paired appendages, endochondral ossification,
and synovial joints. The list grows even longer with such physiological correlates that are observed in just a small fraction of jawless
vertebrates, including: paired nostrils (present in arandaspidids), separation of nasal and hypophyseal systems (present in galeas-
pids), cellular bone, pectoral fins, and epicercal tail (present in osteostracans) (Janvier, 1996). The existing hypotheses about the
origin of the jaws can be categorized by whether they postulate the jaws to have evolved from a classic gill arch pattern (Hirschberger
et al., 2021) or from a distinct pattern by assimilating the gill arches (Miyashita, 2016). Fossil data are critical to this debate because
the origin of jawed vertebrates is deeply embedded in the gnathostome stem (Fig. 1)dno comparison between living vertebrates
can constrain this node.
Fossil record of fishes and major evolutionary transitions 11

A tidal wave of gene expression data has swept across this century-old debate, with the most significant input coming from chon-
drichthyans. They invariably confirm either that the pharyngeal arches (which give rise to the jaw and gill arches in jawed verte-
brates) share gene expression profiles, or that the jaw cartilages and gill arches are patterned dorsoventrally using the same
genetic codes. Additionally, the pseudobranchda gill-like epithelial folding attached to the jaw archdhas shared expression
profiles and cellular origins consistent with the serial homology with the gill arches (Hirschberger and Gillis, 2022). At a glance,
these data seem to support the idea that the jaws evolved by modifying a gill arch.
However, there is no difference between the hypotheses in fitness to these data. Phylogenetically, the data from chondrichthyans
cannot differentiate these hypotheses because they both predict that the jaw and gill arches are serially patterned in jawed verte-
brates (Fig. 6F). So the shared patterning mechanism could still result from assimilation of the gill-arch pattern at the origin of
jawed vertebrates. More fundamentally, both hypotheses accept that the pharyngeal arches are serial homologs. They differ in
how the structures derived in these arches are patterned in the stem of jawed vertebrates, where one predicts a series of skeletal
gill arches (Hirschberger et al., 2021) and the other posits a differentiated oral skeleton followed by the skeletal gill arches (Miya-
shita, 2016). The former renders the condition found in cyclostomes and jawless stem gnathostomes as independently derived,
whereas the latter sees it primitive.
At its core, the gill-arch hypothesis predicts the ancestor of jawed vertebrates to have a series of the gill arches without specialized
oral structures in the presumptive jaw region (mandibular arch). There is no strong case made yet for such morphology in the out-
groups of jawed vertebrates. Compelling evidence suggests distinct patterning of the mandibular domain in cyclostomes (hagfishes
and lampreys) and osteostracans, whereas other jawless vertebrates also seem to conform to the cyclostome-like pattern. Two poten-
tial exceptions are the Cambrian stem vertebrates and galeaspids. The Cambrian yunnanozoans have a series of gill arches beginning
just behind the mouth. However, yunnanozoans and other stem vertebrates are so far removed from the origin of jawed vertebrates
(Fig. 1) that a direct evolutionary link is tenuous. Furthermore, it takes circular reasoning to interpret the yunnanozoan condition as
ancestral to jawed vertebrates: there is no anatomical correlate to positively identify the yunnanozoan arch I as a homolog of the
jaw, unless assuming (tautologically) the most anterior gill arch as the mandibular. Classical hypotheses posited a premandibular
arch in this position, whereas it is also possible that the mandibular skeleton was either not preserved or even diminutive/absent. In
the second example, galeaspids have been recently reconstructed with a gill at the mandibular position (Gai et al., 2022). This new
reconstruction contradicts itself for having spatially incompatible structures in the velum (ventilatory pump) and “mandibular gill.”
These are two separate products that derive from the pharyngeal pouch I and functionally conflict each other; therefore they cannot
coexist in the same fish. As such, the jaw origin debate remains far from settled. Differential testing of the predictions by each
hypothesis will be crucial to shift the balance of the debate.
Model-based approaches for tetrapod origins
The rise of tetrapods caps the list of major evolutionary transitions in the Devonian Period (419e359 Ma), also known as the Age of
Fishes. The origin and early evolution of tetrapods has been framed in two interrelated lines of investigation: morphological (¼fin-
to-limb transition) and ecological (¼tetrapod invasion of terrestrial habitats). Although often confused with each other, they have
different phylogenetic scopes and are chronologically distinct processes. The fin-to-limb transition occurred across the tetrapod stem
across the Devonian Period, but the terrestrial invasion was a much longer, and overall later process. Early tetrapods vary consid-
erably in their degrees of “terrestrial”adaptations, and their radiations mostly occurred in the aftermath of the end-Devonian mass
extinction. So the whole transition (the origin and terrestrial invasion of tetrapods) cannot be characterized with a linear progres-
sion or captured in a single evolutionary event. Accepting this complexity, the emphasis here is on the latter ecological perspective
on tetrapod origins. This area of investigation offers abundant opportunities to integrate physiology of living models with the
fossils: an explicit focus on functional questions allows a model-based approach where analogs can circumvent strict evolutionary
conservation of the characters.
To illustrate this, modern amphibious fishes provide accessible analogs to study the ecological transition of early tetrapods from
aquatic to marginal (and eventually to terrestrial) habitats. These fish (perhaps except for lungfishes) are distantly related to tetra-
pods. However, they live under similar physiological constraints and therefore serve as viable models. The insights generated from
these fish are often complementary to paleontological findings because it is difficult in the fossil record to measure physiological
parameters or distinguish phenotypic response to environmental input (phenotypic plasticity) from the genetic background.
In this sense, Polypterus (bichir) represents an emerging model system. When reared under terrestrial conditions, individuals of
Polypterus accommodate the challenge by changing their behavior, accompanied by a morphological shift in the pectoral fins and
girdles that may be interpreted as paralleling the character transitions in stem tetrapods (Standen et al., 2014). This remarkable
system confirms that phenotypic plasticity is generally adaptive and requires concerted changes across multiple tissues. How this
might fit the fossil data to explain the tetrapod origins remains open to further inquiries, but the terrestrial Polypterus is a powerful
model becausedparadoxicallydit is demonstrably analogous so as to allow distinguishing functional correlates from phylogenetic
inertia, and because the dataset it offers is otherwise inaccessible in the fossil record.
These results set the morphological comparison of early tetrapods in perspective and can help untangle some tautological argu-
ments. For example, the functional “adaptive landscape”has been recently generated for the tetrapod transition based on the shapes
of humeri (Dickson et al., 2021). In that study, the cause and effects are not separated between the morphospace and the perfor-
mance surface, and the latter is confused with fitness. When the morphospace occupation by early tetrapods does not follow the
12 Fossil record of fishes and major evolutionary transitions

least resistance line toward a presumed functional optimum (“adaptive peak”in Dickson et al., 2021), possible interpretations are
either that such seemingly underperforming forms represent transitional phenotypes from one optimum to another, or simply that
predicted performance of one bone (humerus) by just a handful of locomotory correlates does not translate into an adaptive land-
scape. However, available data cannot test either explanation, thereby rendering the evolutionary narrative unfalsifiable. A wide
range of experimental models are now available to provide an independent test on these differential predictions, including killifish,
lungfish, mudskippers, and salamanders. Coupling the insights generated by these living models with the quantitative analysis of
early tetrapods remains a promising venue of research.
Meanwhile, emerging to terrestrial environment requires not just a shift in locomotion, but a long list of physiological adapta-
tions including aerial respiration and osmoregulation. These physiological traits are the subject of active investigation in various
modern amphibious fish. For example, a new scoring system for their land usage and terrestrial emersion tolerance brings a breath
of fresh air (Fig. 7A) (Turko et al., 2021). Despite the expectation that these two scores should be correlated with each other (high
emersion tolerance should promote greater land usage, and vice versa) only a few taxa exhibit high scores in both. This distribution
clearly refutes the idea that the water-to-land transition constitutes a smooth continuum in these amphibious fishes. This view is
overall incompatible with an easy characterization of tetrapod origins as a trek across the morphospace from one functional
optimum to another.
Among the promising approaches to integrate paleontological and neontological data, quantitative analysis of bone histology
provides a potential interface between early tetrapods and modern functional analogs. There has been a rapid accumulation of
information and insights from 3D bone histology in early vertebrates, and particularly in stem tetrapods, where the microstructures
are meticulously reconstructed to complement the data on ossification and growth patterns gained through the traditional
histology. These recent works offer profound implications on bone physiology, such as the role of bone marrow in long bone elon-
gation and the formation of hematopoietic stem cell niche (Estefa et al., 2021) and reveal surprising insight into the life history and
ecology of stem tetrapods. For example, the largest individuals known for the iconic early tetrapod Acanthostega are juveniles (San-
chez et al., 2016).
Beyond tetrapod origins, quantitative bone histology and physiology is a moving front of vertebrate paleontology. Recent tech-
nological advances include focused ion beam scanning electron microscopy (Haridy et al., 2021). This approach has only generated
a limited dataset so far, but with a broader and precise comparative framework it promises to be a powerful method to reconstruct
subcellular correlates of bone metabolism. Possible correlation with genome size highlights the significance of osteocyte dimen-
sions as a marker to constrain the timing of whole genome duplication in the teleost stem (Davesne et al., 2021). In a final example
of notable recent findings, stable isotope records referenced by osteohistology of paddlefishes have led to constraining the season at
the formation of a mass death assemblage linked to an end-Cretaceous bolide impact (During et al., 2022).
Comparative biomechanics and ecological interactions
Biomechanics and ecological dynamics each have a long history in paleontological research and are relevant to modern fish phys-
iology, but neither leaves abundant direct evidence in the fossil record (e.g., trace fossils, stomach contents). As reviewed in the
previous sections, the relationship between paleontological and neontological data are reciprocal here as well. Through compar-
ison, the fossil record reveals evolutionary history on certain structures, whereas the modern experiments establish morphological
correlates of functions that can be interpreted in the fossils. The challenge for comparative biomechanics of extinct forms remains
ground-truthing of inferences, and that for ecological dynamics is testing strength of the hypothetical link between form and func-
tion. Over the last two decades, there have been notable advances in biomechanics of fossil fishes. These include computational
fluid dynamics of jawless stem gnathostomes (Ferrón et al., 2020), ecomorphology of early jawed vertebrates (Anderson et al.,
2011), intracranial kinesis of stem tetrapods (Lemberg et al., 2021), and comparison of body profiles with reference to modern
teleost diversity (Friedman, 2010). In particular, the fluid dynamics of stem gnathostomes are noteworthy for analyzing the forms
that do not have apparent modern analogs. The promises of the method will be fully realized when combined with modern fish
kinematics and fluid dynamics and when tested extensively for sensitivity of the results.
For integrating the two moving frontiers, an exemplar to highlight is suction feeding reconstructed for an early hybodontiform
(stem elasmobranch) Tristychius (Coates et al., 2019). The advent of X-ROMM (X-Ray Reconstruction of Moving Morphology) has
opened a floodgate of data to reveal the disparity and diversity of vertebrate motions. One burgeoning venue of X-ROMM inves-
tigation is skeletal kinematics of suction feeding in various modern fishes. Coates et al. (2019) combined rich insights from this
emerging field with the 3D reconstruction of the skull of Tristychius to generate kinematic simulations of its feeding apparatus
(Fig. 7B). This simulation predicts substantial oral expansion (by 60%) required for high-power suction feeding. Even though
high-performance suction feeding is generally considered a hallmark of actinopterygian evolution, the earliest actinopterygian
capable of equivalent performance only occurs in the Late Permian Period (Acentrophorus), some 50 million years after Tristychius.
This highly derived feeding function was previously beyond detection in the disparity of 2D morphology.
At the interface of comparative biomechanics and trophic relationships, the historical development of reef fish communities
across the Cenozoic Era has become an emerging paradigm. Most modern families of reef fish had appeared by the end of the
Eocene Epoch (34 mya) with matching functional groups identified in the modern coral reefs, but diversification within each func-
tional group only began more recently in the last 5.3 My across the Pliocene and Pleistocene epochs (Bellwood et al., 2017). This
progress has been propelled by the advanced application of phylogenetics with the time calibrated trees and biogeographic
Fossil record of fishes and major evolutionary transitions 13

reconstruction, both of which require significant input from the fossil record. With the coral reefs being biodiversity hotspots, the
focus on trophic or functional specialists have generated these insights, including acanthurids (surgeonfishes), chaetodontids
(butterflyfishes), and labrids (wrasses). Generality of these trends remain to be tested in other lineages and habitats, but this is
an exciting frontier driven by interdisciplinary approaches.
Fig. 7 Two examples from the recent literature that integrate data and insights from fossil and living fishes. (A) A scoring system devised by Turko
et al. (2021) for amphibious fishes reveals the surprising lack of strong correlation between land use and emersion tolerance. Ongoing investigation
in these living taxa will generate valuable resources for paleontologists studying stem and early tetrapods, and vice versa. (B) A marriage of 3D
methodsdreconstruction of a skull of an early stem elasmobranch Tristychius and kinematics of modern suction-feeding fishdgave rise to a set of
quantitative estimates during a cycle of mouth opening and closure (Coates et al., 2019). The estimated values indicate high-power suction feeding in
this “shark,”preceding actinopterygians by 50 My. Both reproduced with the permission from the authors.
14 Fossil record of fishes and major evolutionary transitions

Conclusion
The fossil record of fishes provides a wealth of comparative information to modern zoologists. Reciprocally, the study of modern
fishes generates datasets that are not readily preserved in fossils, from the molecules (e.g., genomics) to organs (e.g., various soft
tissues) and from development to behavior. As reviewed in this chapter, the barrier for an integrative approach has been gradually
torn down. Hypothetical scenarios for the origin of jawed vertebrates were generated based on the fossil record and are evaluated
with transcriptomics of modern fishes, whereas insights from X-ROMM of modern actinopterygians aids in reconstructing 3D kine-
matics of suction feeding in a Paleozoic shark. In years down the road, it may become difficult to distinguish whether a particular
study is research on fossils informed by modern species or the opposite. The standard of evidence going forward is accordingly
higher, and neither neontological nor paleontological approach should abuse comparative data by oversimplification. Modern
phylogenetics provide a framework in which to contrast alternative hypotheses, partly through improved understanding of inter-
relationships and partly through theoretical constructs such as phylogenetic bracketing. These progresses paint bright future with
new technologies and fossil discoveries, as this is a fundamentally additive process where information only accrues. However,
increasing information does not promisedand may in fact negatively correlate withdclarity on the many issues, as the vertebrate
paleontologist Alfred Romer famously quipped in his 1962 colloquium. That still may be the only path to deepen our under-
standing of this greatest continual radiation in the history of vertebrate animals.
References
Anderson, P.S.L., Friedman, M., Brazeau, M.D., Rayfield, E.J., 2011. Initial radiation of jaws demonstrated stability despite faunal and environmental change. Nature 476 (7359),
206e209. https://doi.org/10.1038/nature10207.
Andreev, P.S., Sansom, I.J., Li, Q., Zhao, W., Wang, J., Wang, C.-C., Peng, L., Jia, L., Qiao, T., Zhu, M., 2022. Spiny chondrichthyan from the lower Silurian of South China. Nature
609 (7929), 969e974. https://doi.org/10.1038/s41586-022-05233-8.
Bellwood, D.R., Goatley, C.H.R., Bellwood, O., 2017. The evolution of fishes and corals on reefs: form, function and interdependence. Biol. Rev. 92 (2), 878e901. https://doi.org/
10.1111/brv.12259.
Brazeau, M.D., Friedman, M., 2015. The origin and early phylogenetic history of jawed vertebrates. Nature 520 (7548), 490e497. https://doi.org/10.1038/nature14438.
Coates, M.I., Tietjen, K., Olsen, A.M., Finarelli, J.A., 2019. High-performance suction feeding in an early elasmobranch. Sci. Adv. 5 (9), eaax2742. https://doi.org/10.1126/
sciadv.aax2742.
Davesne, D., Friedman, M., Schmitt, A.D., Fernandez, V., Carnevale, G., Ahlberg, P.E., Sanchez, S., Benson, R.B.J., 2021. Fossilized cell structures identify an ancient origin for the
teleost whole-genome duplication. Proc. Natl. Acad. Sci. U. S. A. 118 (30), e2101780118. https://doi.org/10.1073/pnas.2101780118.
Dickson, B.V., Clack, J.A., Smithson, T.R., Pierce, S.E., 2021. Functional adaptive landscapes predict terrestrial capacity at the origin of limbs. Nature 589 (7841), 242e245.
https://doi.org/10.1038/s41586-020-2974-5.
During, M.A.D., Smit, J., Voeten, D.F.A.E., Berruyer, C., Tafforeau, P., Sanchez, S., Stein, K.H.W., Verdegaal-Warmerdam, S.J.A., van der Lubbe, J.H.J.L., 2022. The Mesozoic
terminated in boreal spring. Nature 603 (7899), 91e94. Nature Publishing Group. https://doi.org/10.1038/s41586-022-04446-1.
Estefa, J., Tafforeau, P., Clement, A.M., Klembara, J., Nied
zwiedzki, G., Berruyer, C., Sanchez, S., 2021. New light shed on the early evolution of limb-bone growth plate and bone
marrow. Elife 10, e51581. eLife Sciences Publications, Ltd. https://doi.org/10.7554/eLife.51581.
Ferrón, H.G., Martínez-Pérez, C., Rahman, I.A., Selles de Lucas, V., Botella, H., Donoghue, P.C.J., 2020. Computational fluid dynamics suggests ecological diversification among
stem-gnathostomes. Curr. Biol. 30 (23), 4808e4813.e3. https://doi.org/10.1016/j.cub.2020.09.031.
Friedman, M., 2010. Explosive morphological diversification of spiny-finned teleost fishes in the aftermath of the end-cretaceous extinction. Proc. R. Soc. Lond. B Biol. Sci.,
rspb20092177 https://doi.org/10.1098/rspb.2009.2177.
Friedman, M., 2015. The early evolution of ray-finned fishes. Palaeontology 58 (2), 213e228. https://doi.org/10.1111/pala.12150.
Friedman, M., Sallan, L.C., 2012. Five hundred million years of extinction and recovery: a phanerozoic survey of large-scale diversity patterns in fishes. Palaeontology 55 (4),
707e742. https://doi.org/10.1111/j.1475-4983.2012.01165.x.
Gai, Z., Zhu, M., Ahlberg, P.E., Donoghue, P.C.J., 2022. The evolution of the spiracular region from jawless fishes to tetrapods. Front. Ecol. Evol. 10. https://doi.org/10.3389/
fevo.2022.887172.
Giles, S., Xu, G.-H., Near, T.J., Friedman, M., 2017. Early members of “living fossil”lineage imply later origin of modern ray-finned fishes. Nature 549 (7671), 265e268. https://
doi.org/10.1038/nature23654.
Haridy, Y., Osenberg, M., Hilger, A., Manke, I., Davesne, D., Witzmann, F., 2021. Bone metabolism and evolutionary origin of osteocytes: novel application of FIB-SEM tomography.
Sci. Adv. 7 (14), eabb9113. American Association for the Advancement of Science. https://doi.org/10.1126/sciadv.abb9113.
Hirschberger, C., Gillis, J.A., 2022. The pseudobranch of jawed vertebrates is a mandibular arch-derived gill. Development 149 (13), dev200184. https://doi.org/10.1242/
dev.200184.
Hirschberger, C., Sleight, V.A., Criswell, K.E., Clark, S.J., Gillis, J.A., 2021. Conserved and unique transcriptional features of pharyngeal arches in the skate (Leucoraja erinacea) and
evolution of the jaw. Mol. Biol. Evol. 38 (10), 4187e4204. https://doi.org/10.1093/molbev/msab123.
Janvier, P., 1996. Early Vertebrates. Clarendon Press, Oxford.
Lemberg, J.B., Daeschler, E.B., Shubin, N.H., 2021. The feeding system of Tiktaalik roseae: an intermediate between suction feeding and biting. Proc. Natl. Acad. Sci. USA 118 (7),
e2016421118. https://doi.org/10.1073/pnas.2016421118.
Long, J.A., 2010. The Rise of Fishes: 500 Million Years of Evolution, second ed. Johns Hopkins University Press, Baltimore.
Long, J.A., Mark-Kurik, E., Johanson, Z., Lee, M.S.Y., Young, G.C., Min, Z., Ahlberg, P.E., Newman, M., Jones, R., den Blaauwen, J., Choo, B., Trinajstic, K., 2015. Copulation in
antiarch placoderms and the origin of gnathostome internal fertilization. Nature 517 (7533), 196e199. https://doi.org/10.1038/nature13825.
Miyashita, T., 2016. Fishing for jaws in early vertebrate evolution: a novel hypothesis of mandibular confinement. Biol. Rev. 91 (3), 611e657. https://doi.org/10.1111/brv.12187.
Nelson, J.S., Grande, T.C., Wilson, M.V.H., 2016. Fishes of the World, fifth ed. Wiley, New York.
Sackville, M.A., Cameron, C.B., Gillis, J.A., Brauner, C.J., 2022. Ion regulation at gills precedes gas exchange and the origin of vertebrates. Nature 610 (7933), 699e703. https://
doi.org/10.1038/s41586-022-05331-7.
Sallan, L.C., 2014. Major issues in the origins of ray-finned fish (Actinopterygii) biodiversity. Biol. Rev. 89 (4), 950e971. https://doi.org/10.1111/brv.12086.
Sallan, L.C., Coates, M.I., 2010. End-Devonian extinction and a bottleneck in the early evolution of modern jawed vertebrates. Proc. Natl. Acad. Sci. USA 107 (22), 10131e10135.
https://doi.org/10.1073/pnas.0914000107.
Fossil record of fishes and major evolutionary transitions 15

Sanchez, S., Tafforeau, P., Clack, J.A., Ahlberg, P.E., 2016. Life history of the stem tetrapod Acanthostega revealed by synchrotron microtomography. Nature 537 (7620),
408e411. https://doi.org/10.1038/nature19354.
Standen, E.M., Du, T.Y., Larsson, H.C.E., 2014. Developmental plasticity and the origin of tetrapods. Nature 513 (7516), 54e58. https://doi.org/10.1038/nature13708.
Trinajstic, K., Long, J.A., Sanchez, S., Boisvert, C.A., Snitting, D., Tafforeau, P., Dupret, V., Clement, A.M., Currie, P.D., Roelofs, B., Bevitt, J.J., Lee, M.S.Y., Ahlberg, P.E., 2022.
Exceptional preservation of organs in Devonian placoderms from the Gogo lagerstätte. Science 377 (6612), 1311e1314. https://doi.org/10.1126/science.abf3289.
Turko, A.J., Rossi, G.S., Wright, P.A., 2021. More than breathing air: evolutionary drivers and physiological implications of an amphibious lifestyle in fishes. Physiology 36 (5),
307e314. https://doi.org/10.1152/physiol.00012.2021.
Witmer, L.M., 1997. The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils. In: Thomason, J. (Ed.), Functional Morphology in Vertebrate
Paleontology. Cambridge University Press, Cambridge, pp. 19e33.
Zhu, Y., Li, Q., Lu, J., Chen, Y., Wang, J., Gai, Z., Zhao, W., Wei, G., Yu, Y., Ahlberg, P.E., Zhu, M., 2022. The oldest complete jawed vertebrates from the early Silurian of China.
Nature 609 (7929), 954e958. https://doi.org/10.1038/s41586-022-05136-8.
16 Fossil record of fishes and major evolutionary transitions