How Fish Got a Neck: The Evolution of the Shoulder Girdle

• Two classical theories of the origin of paired appendages
• A posterior pharyngeal origin, part 1: the articular facet and the sixth arch
• A posterior pharyngeal origin, part 2: the subclavian artery
• Objection 1: the subclavian artery is postbranchial, not branchial
• Objection 2: trunk nerves versus branchial nerves
• The alternative: repatterning of the head–trunk interface
• Supplementary evidence from embryology
• What is still open
• References

The post is based on the final essay for the course Vertebrate Evolution (Tetrapods) I took at Utrecht University.

The origin of the vertebrate neck and pectoral girdle marks one of the most consequential reorganizations of the vertebrate body plan. Decoupling the head from the trunk gave vertebrates cranial mobility, more efficient feeding strategies, and the ecological versatility that ultimately underpinned the success of jawed vertebrates. At the centre of that transition sits the pectoral girdle: the skeletal complex that anchors the pectoral fins or limbs to the body and defines the anatomical boundary between head and trunk.

Where this structure came from is still contested after more than a century of work. The classical literature splits between a trunk-based origin from ventrolateral fin folds and a derivation from a posterior branchial (gill) arch. The debate has recently reopened in a sharp form. Brazeau et al. (2023) argued from placoderm and osteostracan fossils that the girdle arose by incorporating the sixth branchial arch into the trunk. Miyashita et al. (2025) countered that the neurovascular anatomy of jawless stem gnathostomes places the pectoral attachment unambiguously in the trunk domain. Kuroda et al. (2024) added zebrafish fate-mapping data pointing the same way.

This article walks through those arguments in sequence and defends an integrated position: the pectoral girdle evolved through repatterning at the head–trunk interface, not through direct transformation of a branchial arch. That framework reconciles the fossil, anatomical, and embryological evidence, and explains how early vertebrates acquired a functional neck without violating conserved axial identities.

Two classical theories of the origin of paired appendages

The evolutionary origin of the vertebrate pectoral girdle and paired fins has long been framed by two classical hypotheses: the ventrolateral fin-fold hypothesis and the archipterygium hypothesis. Both seek to explain the emergence of paired appendages, but they differ fundamentally in which anatomical structures they treat as ancestral, and in the developmental logic they assume for limb evolution.

The fin-fold hypothesis, first articulated in the late nineteenth century (Balfour 1881; Thacher 1877) and later refined through comparative anatomy, proposes that paired fins evolved from continuous longitudinal folds of tissue running along the ventrolateral margins of the trunk. Early vertebrates on this view possessed extended fin folds that stabilized swimming, analogous to the median fin fold that gives rise to the dorsal and anal fins. Through evolutionary subdivision and regional specialization, these continuous folds were partitioned into discrete pectoral and pelvic fins.

The hypothesis has been regarded as particularly successful because it is broadly consistent with developmental genetic evidence showing shared patterning mechanisms between paired and unpaired fins (Freitas et al. 2006). Fossil evidence from early jawed vertebrates and their stem relatives supports it further, as several taxa exhibit ventrolateral fin structures that appear intermediate between continuous folds and discrete appendages (Gai et al. 2022).

Its limitation is scope. The fin-fold hypothesis largely addresses the origin of the fins themselves and offers little insight into the origin of the girdle — the skeletal complex that anchors the fins to the body and ultimately enables the formation of a neck.

The archipterygium hypothesis, originally proposed by Gegenbaur (1876), instead places the origin of paired appendages within the branchial apparatus. It argues that the pectoral girdle and fin endoskeleton derive from a modified posterior gill arch, with fin skeletal elements evolving from serially repeated branchial rays. The appeal of this model is that it explains the complex internal skeletal organization of paired fins by reference to an already segmented and articulated ancestral structure. It also supplies a mechanism for the emergence of the girdle itself: transforming an existing skeletal framework rather than inventing a new one. Historically, though, it has struggled to gain broad support, because direct fossil evidence linking gill arches to appendicular skeletons is rare.

Neither theory fully resolves how the pectoral girdle and neck evolved. But together they set the conceptual frame against which modern fossil discoveries and developmental data are still evaluated.

A posterior pharyngeal origin, part 1: the articular facet and the sixth arch

Brazeau et al. (2023) proposed a renewed and more explicitly fossil-based version of the archipterygium hypothesis, arguing that the vertebrate pectoral girdle originated through the incorporation of a posterior branchial arch into the trunk during early gnathostome evolution. Their argument rests on a detailed anatomical reinterpretation of placoderm and osteostracan fossils, with particular emphasis on cranial articulation surfaces, vascular topology, and the positional relationship between the branchial skeleton and the shoulder girdle.

A central observation concerns a distinct articular facet at the posterior end of the placoderm headshield, especially well preserved in Kolymaspis and Brindabellaspis. This facet occurs on the craniospinal process of the braincase and is interpreted as the articulation point for a now-lost endoskeletal element that connected the skull to the pectoral girdle. Brazeau argues that this articular surface represents a structural remnant of a gill-arch-derived element rather than a purely trunk-derived shoulder connection. The location of the facet — posterior to the branchial region but still contiguous with it — is taken to mark the ancestral head–shoulder boundary.

Crucially, this hypothetical endoskeletal element is identified as serially homologous with a branchial arch, most plausibly the sixth. Two considerations support the inference. First, placoderms appear to possess no more than five functional gill arches, suggesting that a more posterior arch could have been co-opted for a non-respiratory role. Second, incorporating the sixth arch into the pectoral girdle offers an elegant explanation for the otherwise puzzling constraint of five gill arches in most crown gnathostomes, in contrast to the much higher number seen in jawless vertebrates. In this framework the sixth arch ceases to function as a respiratory element and instead becomes a structural buttress supporting the posterior wall of the pharynx and the emerging shoulder girdle.

A posterior pharyngeal origin, part 2: the subclavian artery

Brazeau further supports the model by analysing the course of the subclavian artery, the main arterial supply to the pectoral fin. In osteostracans, the subclavian artery originates near the posterior branchial circulation and runs along the interbranchial ridge between the sixth and seventh branchial arches.

This vascular association is interpreted as evidence that the pectoral fin attachment — and by extension the pectoral girdle — was ancestrally positioned within the posterior pharyngeal domain. On that basis, Brazeau proposes that a posterior portion of the pharyngeal skeleton was incorporated into the trunk, forming the earliest pectoral girdle.

Synthesizing these observations, the hypothesis is that the pectoral girdle evolved as a composite structure derived from a posterior branchial arch, establishing a discrete head–shoulder separation for the first time in vertebrate evolution. On this view the origin of the girdle is inseparable from reorganization of the pharynx and the emergence of a kinetic cranio-thoracic joint, linking paired appendages, ventilation, and feeding mechanics in early gnathostomes.

Objection 1: the subclavian artery is postbranchial, not branchial

Miyashita et al. (2025) directly challenge the posterior pharyngeal origin model, arguing that the pectoral girdle and fin attachment in jawless stem gnathostomes belong unequivocally to the trunk domain, not to the branchial skeleton. They do not dispute the fossil observations themselves; they re-evaluate the anatomical interpretation using a combined analysis of vascular topology, neural innervation, and positional relationships relative to the head–trunk interface.

The first objection targets the vascular argument above. In Norselaspis, the subclavian artery arises posterior to the last efferent branchial arteries, and therefore belongs anatomically to the postbranchial trunk region rather than to the pharyngeal circulation. Blood exiting the seventh branchial pouch first enters the midventral artery, from which the subclavian artery then branches to supply the pectoral fin. The subclavian artery also irrigates the extrabranchial cavity located behind the posterior branchial arch.

These relationships show that the artery’s proximity to the branchial system reflects expansion of the branchial chamber in osteostracans, not a branchial identity of the pectoral girdle. Vascular topology, on this reading, provides no support for placing the shoulder within the branchial series.

Objection 2: trunk nerves versus branchial nerves

The second objection is the more decisive one, and it concerns neural innervation.

Brazeau’s interpretation implicitly requires either that the pectoral attachment be governed by branchial nerves, or that a homeotic shift occurred in which cranial nerves acquired trunk-like functions. Miyashita shows that neither is necessary or supported by the fossil anatomy.

In Norselaspis, the pectoral fenestra is penetrated by the most anterior spinal nerve, which possesses a distinct root immediately posterior to the vagus nerve (CN X). Because there is no evidence in Norselaspis for the accessory or hypoglossal nerves (CNs XI and XII), this nerve is a true trunk nerve, not a branchial nerve, and it innervates the pectoral region directly. The posterior branchial arches, by contrast, retain their typical vagal innervation and branchial arterial drainage.

The clean separation between trunk and branchial innervation demonstrates that the shoulder attachment is controlled by the trunk nervous system. Accepting a branchial origin for the pectoral girdle would require multiple independent homeotic transformations of nerves and vessels — an assumption Miyashita regards as unnecessary.

The alternative: repatterning of the head–trunk interface

On the basis of these anatomical relationships, Miyashita proposes an alternative model. Rather than being derived from a posterior gill arch, the shoulder girdle emerged through repatterning at the head–trunk interface.

In osteostracans, that interface included a postbranchial wall separating the branchial chamber from the trunk, with the pectoral fins attached posterolaterally to the dermal shield. The absence of neck and hypobranchial muscles — and of the accessory and hypoglossal nerves that innervate them in crown gnathostomes — further indicates that osteostracans represent a stage prior to true head–shoulder decoupling.

On this view the pectoral girdle originated as a trunk structure positioned adjacent to, but not derived from, the pharynx. Subsequent gnathostome evolution then intercalated new muscles and nerves between the head and the shoulder, producing a functional neck without altering the axial identities of pre-existing branchial structures.

Supplementary evidence from embryology

Recent embryological work by Kuroda et al. (2024) provides independent and mechanistically robust support for this interpretation. Using high-resolution lineage tracing and genetic fate-mapping in zebrafish, the study addresses a question fossils alone cannot settle: from which embryonic tissues do the individual elements of the pectoral girdle arise?

The dermal and endoskeletal components of the pectoral girdle turn out to have distinct and multiple embryonic origins. The scapulocoracoid, the principal endoskeletal element, derives predominantly from trunk lateral plate mesoderm (LPM). This is critical, because LPM is a canonical trunk tissue and does not contribute to the branchial arches, which form largely from cranial neural crest cells and cardiopharyngeal mesoderm (CPM). The trunk origin of the scapulocoracoid therefore contradicts any model in which the shoulder girdle is homologous to, or transformed from, a posterior gill arch.

The cleithrum, a major dermal element, shows a composite origin instead. Fate-mapping reveals contributions from trunk LPM, CPM, and neural crest–derived connective tissues. The cleithrum develops precisely at the fin field positioned along the head–trunk boundary, where anterior somites, CPM, and trunk mesoderm overlap. This mosaic origin explains how the girdle can maintain close anatomical proximity to the pharyngeal region without being derived from the branchial skeleton — it reflects positional integration rather than serial homology.

Taken together, the embryology supplies a developmental mechanism that reconciles the fossil evidence emphasized by Miyashita with the long-standing difficulty of demonstrating true branchial homology for the shoulder girdle.

What is still open

The debate is far from settled. Competing interpretations of the same fossil structures expose the inherent limits of the fossil record with respect to soft tissues, developmental boundaries, and functional connectivity. Progress will require additional well-preserved fossils that capture neurovascular and musculoskeletal relationships, plus expanded comparative embryology across a broader range of taxa. Integrating fossil reconstructions with experimentally grounded developmental data remains essential for testing hypotheses of homology and positional identity.

One methodological point deserves emphasis, and it is arguably the most transferable lesson here: cranial nerve organization is remarkably conservative. Root positions and target domains are stable across jawless and jawed vertebrates. Inferences based on neural innervation — whether a structure is supplied by cranial or by trunk nerves — are therefore likely to be more reliable indicators of axial identity than gross morphological similarity. Skeletal morphology, by contrast, is evolutionarily plastic and prone to functional convergence.

Where neuroanatomical evidence is available, it should be weighted heavily. Continued integration of fossil neuroanatomy with developmental and genetic data looks like the most promising path to resolving this question.

References