Fishing for jaws in early vertebrate evolution: A new hypothesis of mandibular confinement

– At what level and which time point are animal structures truly segmented?  Changing the underlying assumptions generates a new scenario for the origin of vertebrate jaws –

The origin of jaw is a major transition in vertebrate evolution

  • Jawed vertebrates originated approximately 450 million years ago and dominated – today, 99.8% of all living vertebrate species have a set of biting jaws.
  • Perhaps more remarkably, no jawed vertebrate lineage secondarily lost jaws.  Jaw skeleton may be modified and occasionally dysfunctional during ontogeny, but the skeleton is never lost altogether, in a clear contrast to frequent loss events of appendages and sensory structures.
  • These observations underwrite the significance of jaws in vertebrate evolution.

Showcasing the diversity of living vertebrates at the Royal Society's Commonwealth Science Conference (Dec. 2014 in Bengaluru, India)

Photography by the Royal Society, London


Why are the previous jaw-origin hypotheses incomplete?

  • Previous jaw-origin hypotheses attempted to explain the jaw as some kind of modification of an anterior pharyngeal segment. That is, they assumed that the vertebrate pharynx was ancestrally segmented, and patterned metameristically. Then an anterior segment acquired specialized function.
  • This assumption - somewhat akin to the view of appendages as serial homologues - is incompatible with fossil, anatomical, and developmental evidence. It necessitates a hypothetical ancestor with morphology never seen in the fossil record. No known vertebrate pharynx is perfectly segmented. In fact, the pharynx of a jawed vertebrate is closer to the predicted ancestral segmentation than that of a jawless vertebrate.

Graphic summary of three jaw origin hypotheses. A, Gill Arch Hypothesis (modified from de Beer, 1937); B, Velum Hypothesis (modified from Janvier 1996). C, Ventilation Hypothesis (modified from Mallatt, 1996).

From Miyashita (2016)


  • The sole exception was the Heterotopy Hypothesis advocated by Professor Shigeru Kuratani (see ref. here, here, and here).  This hypothesis does not view the mandibular arch in a series.  Instead, it considers how the premandibular region is patterned in early vertebrates, and postulates how the 'mandibular' domain is decoupled from the snout by the acquisition of paired nasal placodes.
  • The Heterotopy Hypothesis predicts character transitions that are corroborated by fossil evidence.  However, it does not consider the region posterior to the mandibular arch.  My aim was to generate a hypothesis that explains anatomical changes in and at the periphery of the mandibular arch across the jaw origin.

A schematic illustration of vertebrate pharyngeal patterning. Note that the mandibular region does not form a perfect series with the branchial arches.

From Miyashita (2016)


What is the Mandibular Confinement Hypothesis?

  • I began by questioning whether the jaw skeleton can be truly viewed as a modification of a segment or other way around – that is, the jaw skeleton evolved by expanding the gill-arch-like pattern onto what originally was a distinct mandibular region.
  • The jaw appears like a modified pharyngeal segment and as if in series with the branchial arches, NOT because it used to be a gill arch or some kind of serial structure, BUT because it originated by (secondarily) assimilating the patterning program of the gill arches.
  • This new way of looking at the jaw origin (as inspired by the Heterotopy Hypothesis) explains character distribution across jawless and jawed vertebrates well.  It requires no hypothetical ancestor whose anatomy is wholly different from known vertebrate lineages.

The highly specialized and structurally complex mandibular musculoskeletal system in cyclostomes is contrasted against the spatially confined and serially patterned counterpart in gnathostomes. 

A, hagfish (modified after Miyashita, 2012). B, lamprey (modified after Miyashita, 2012; Johnels, 1948). C, spiny dogfish (modified after Mallatt, 1997). D, chondrocranium of a 5-day old zebrafish larva (modified after Hernández, Barresi & Devoto, 2002). E, the skull and cranial musculature of an adult zebrafish (modified after Diogo, Hinits & Hughes, 2008).

From Miyashita (2016)


  • This new hypothesis, Mandibular Confinement Hypothesis, predicts an ancestor in which the region homologous to the jaw-bearing arch is a distinct region forming much of the orofacial structures. Like modern hagfish and lampreys (and ostracoderms), the ancestral mandibular region spread across the premandibular to hyoid to hypobranchial domains, forming specialized feeding and ventilatory structures at each interface.
  • To generate a jaw skeleton via assimilation of the gill-arch pattern, the specialized 'mandibular' structures were gradually replaced by the surrounding domains. Accepting the Heterotopy Hypothesis, the origin of paired nasal placodes decoupled it from the premandibular domain. At the interface with the hyoid arch (first gill arch), the velum was lost, and the pouch between the mandibular and hyoid domains became a firm boundary. The origin of trunk-derived hypobranchial muscles replaced the specialized feeding apparatus derived from the mandibular domain, which sits in the pharyngeal floor in hagfish, lampreys, and likely ostracoderms.
  • Thus, the widespread mandibular region was gradually 'confined' by the surrounding domains, with clearly delineated interfaces.

Schematic summary of mandibular confinement.  The originally distinct, differentially patterned mandibular region loses specialized peripheral structures, and becomes spatially confined by the surrounding domains.   Finally, it acquires the branchial patterning program to become a jaw apparatus.

Illustrations by Tetsuto Miyashita


  • As a result of this spatial confinement, progenitors in the mandibular domain - which would otherwise develop into the specialized feeding and ventilation structures - acquired new developmental fates by acquiring expression of the gill-arch patterning program (i.e., dorsoventrally nested expression of Dlx genes).
  • The jaw evolved as a phenocopy of a gill arch in once a distinct domain of the vertebrate pharynx. As a result, its dorsoventrally patterned skeleton is morphologically similar to the gill arches, and share the same developmental basis.
  • This explains why the mandibular arch develops as part of the pharyngeal series in jawed vertebrates. Not due to the ancestral segmentation, but thanks to serial assimilation that transformed a spatially confined, distinct mandibular domain of the pharynx.

Phylogenetic distribution of characters used to determine degree of confinement of the mandibular musculoskeletal system. Partial confinement occurred in independent lineages of jawless vertebrates, but only jawed vertebrates have full state of the mandibular confinement for each interface.

From Miyashita (2016)


How can the Mandibular Confinement Hypothesis be tested?

  • Because the Mandibular Confinement Hypothesis aims to explain why the jaw skeleton develops as part of the serial patterning program of the pharynx, the fact that the mandibular development is serial with the branchial arches in any jawed vertebrate does not refute the hypothesis – rather, it reinforces the hypothesis.
  • This hypothesis can be falsified via multiple predictions –– if: a) an ancestor of jawed vertebrates turns out to have already had a serial mandibular skeleton, prior to the origin of jaw; b) developmental evidence clearly points to the ancestrally serial nature of development of the mandibular region in cyclostomes; or c) immediate outgroups of jawed vertebrates provide unequivocal evidence that character transformations predicted by the Mandibular Confinement Hypothesis is out of order.

Metaspriggina walcotti has dorsoventrally bipartite pharyngeal skeleton (modified from Conway Morris and Caron, 2014). The Mandibular Confinement Hypothesis can be rejected if compelling evidence shows: (a) the most anterior pair is derived of the Arch I; and (b) such polyisomeric state is shared across stem gnathostomes up to the origin of the jaw.  


  • For potential falsification scenario a), the Cambrian stem vertebrate Metaspriggina appears to have serially patterned pharyngeal skeleton. However, this animal is far removed from the gnathostome stem. Additionally, there is no evidence to suggest that its most anterior arch of the pharyngeal skeleton is derived from the region homologous to the gnathostome mandibular arch. It is simply the most anterior skeletal arch, and does not rule out any distinct domain to the front. If the first arch is homologous to the mandibular arch, and if the morphology seen in Metaspriggina is the general condition across the gnathostome stem, the Hypothesis would be falsified.
  • For potential falsification scenario b), ostracoderms (including jawless stem gnathostomes) are typically reconstructed with insights from the anatomy of living cyclostomes. Is this to assume that cyclostomes are primitive? It is other way around. These animals share a set of correlates that can be used to make inferences about the ostracoderm anatomy. The prevalence of such anatomical correlates reflects how the cyclostome pattern reflects a general theme across early evolution of jawless vertebrates.

Two alternative mechanisms giving rise to a serial pattern.  Polyisomeric ancestral state represents a series of identical or similar units and assumes differential specialization to derive the existing serial pattern.

Anisomeric ancestral state represents distinctly patterned regions and assumes serial assimilation to derive the existing serial pattern.

Illustrations by Tetsuto Miyashita


Resources