611
Historical hypotheses regarding segmentation of
the vertebrate head
R. Glenn Northcutt1
Neurobiology Unit, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA
92093-0201, USA; Department of Neurosciences, School of Medicine, University of California, San Diego,
La Jolla, CA 92093-0201, USA
Synopsis The morphology of the vertebrate head is extremely complex and comprises numerous iterative structures that
arise from each of the embryonic germ layers. The search for a fundamental plan uniting all of these serial structures
spans 200 years. The earliest attempt to identify a common plan was J. W. Goethe’s vertebral theory of skull
organization, in which the skull was interpreted as being formed by a series of trunk vertebrae. This theory was rejected
by T. H. Huxley in the 1858 Croonian Lecture and was replaced by the segmented mesodermal model of Francis Balfour,
which was elaborated subsequently by A. Marshall, Gavin de Beer, and Edwin Goodrich. This model assumes that the
head of the earliest vertebrates consisted of eight segments. It further assumes that each segment contained dorsal muscles
arising from the somitic mesoderm, and ventral muscles arising from lateral plate mesoderm, except for the first segment,
which lacked ventral muscles derived from the lateral plate mesoderm. The muscles of each head segment were believed
to be innervated by two pairs of cranial nerves, homologous to the dorsal and ventral spinal nerves of lampreys. The
validity of this theory, known as the Goodrich model, came into question, however, after the discovery that the
branchiomeric muscles associated with each pharyngeal arch do not arise from lateral plate mesoderm, as initially
proposed by Marshall and subsequently accepted by Goodrich and de Beer, but, rather, arise from paraxial mesoderm.
Furthermore, segmentation of the brain into some 14 neuromeres cannot be accommodated by any model involving eight
segments. Finally, there is also clear evidence that at least one, if not two, additional series of placodally derived sensory
nerves occurs in the head and has no counterpart in the trunk. At present, there is no theory of segmentation that can
account for all cephalic iterative structures.
Introduction
Our current view of the organization of the vertebrate
head is the result of a number of research agendas that
primarily span the period of 1790–1940. During this
period, a number of developmental topics were studied
intensively for the first time. They include (1)
formation of germ layers, (2) mesodermal differentiation, (3) cranial nerve development, (4) brain morphogenesis, (5) neural crest differentiation, and (6)
placodal development. Much of this research was
driven by two major questions: (1) What is the nature
of the vertebrate body plan? and (2) Is the vertebrate
head segmented? The data that resulted from these
studies led to a number of major hypotheses, including
the vertebral hypothesis, the mesodermal segmental
hypotheses, and the neuroepithelial hypotheses.
The history of these hypotheses has been touched
on in a number of brief reviews, including those by de
Beer (1937), Hall and Hanken (1985), Northcutt (1990,
1993), and Holland (2000), but the hypotheses have
not been re-evaluated in the context of much of the
developmental data generated in the past 10 years. The
present essay attempts to correct this situation by revisiting the 200-year-old question of whether or not the
vertebrate head is segmented.
A review of the literature on head segmentation
clearly indicates that there is no agreement on the
meaning of this term. In some cases, it implies that
the vertebrate head arose ancestrally from a series
of complete head segments (metameres), such as those
seen in annelids (Dohrn’s theory 1875) or arthropods
(Gaskell’s theory 1908). In other cases, however, segmentation in vertebrates has been viewed as the more
restricted division of one or more embryonic tissues,
such as the paraxial mesoderm (Balfour 1878), neurectoderm (Neal 1898), or endoderm (Kingsbury 1926;
Romer 1972), into a series of iterative structures in the
anterior to posterior axis. In this review, the vertebrate
head is viewed as segmented if there is developmental
From the symposium ‘‘Vertebrate Head Segmentation in a Modern Evo-Devo Context’’ presented at the annual meeting of the Society for
Integrative and Comparative Biology, January 2–6, 2008, at San Antonio, Texas.
1
E-mail: [email protected]
Integrative and Comparative Biology, volume 48, number 5, pp. 611–619
doi:10.1093/icb/icn065
Advance Access publication June 28, 2008
ß The Author 2008. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. All rights reserved.
For permissions please email: [email protected].
612
R. G. Northcutt
evidence that its structures arise from each of the three
germ layers and form a series of repeated elements, in
registry, in the anterior–posterior axis.
The vertebral hypothesis
According to de Beer (1937), Johann Wolfgang Goethe,
the German naturphilosopher best known as the
author of Faust, was the first person to hit upon the
idea that the vertebrate skull consisted of fused
vertebrae when his servant called his attention to a
partially disarticulated sheep’s skull in a Jewish
cemetery in Venice in 1790. However, Goethe did not
publish his idea until 1820 by which time the idea had
also occurred to a number of other people including
Oken (1807) and Geoffroy St.-Hillaire (1818). During
much of the 19th century, the vertebral hypothesis, in
various forms, was widely accepted and was most
dramatically illustrated by Owen (1848) in his
publication On the Archetype and Homologies of the
Vertebrate Skeleton. Initially, Owen viewed an archetype as a fundamental pattern on which a natural group
of animals or an organ system has been constructed,
but later he appears to have believed that it represented
a Platonic ideal. Various political and social reasons
have been suggested for this shift (Desmond 1982;
Rupke 1994). In any case, Owen described the head of
the vertebrate archetype as consisting of four segments,
with the skull being formed by the modification of four
vertebrae, which he identified rostrocaudally as the
nasal, frontal, parietal, and occipital vertebrae (Fig. 1).
In his view, an ideal vertebra consisted of a dorsal
neural arch, a centrally located centrum, and a ventral
hemal arch (Fig. 1). The neural arch was believed to be
formed by a pair of neurapophyses, capped by a neural
spine, whereas the hemal arch was believed to be
formed by a pair of dorsal pleurapophyses and a pair of
ventral hemapophyses, capped by a hemal spine. Owen
believed that the fused centra formed the floor of the
neurocranium, whereas the enlarged neurapophyses
and neural spines formed the walls and roof of the
neurocranium, respectively. Finally, he interpreted the
first and second hemal arches of the head as forming
the upper and lower jaws, respectively, with the third
and fourth hemal arches forming the branchial
skeleton. Thus, Owen was able to account for all
bones of the vertebrate ‘‘endocranium’’ as modified
elements of four vertebrae.
In 1858, Huxley dealt the death blow to all vertebral
hypotheses in his now famous Croonian Lecture before
the Royal Society. The paper (Huxley 1858) based on
his lecture is a masterpiece of scientific reasoning and
still well worth reading. Huxley realized that two
questions had to be answered in order to test these
Fig. 1 Owen’s archetype for the skeleton of the vertebrate head
and rostral trunk. Redrawn from Owen (1848). c, centrum
(black); df, dorsal fin; hp, hemapophysis; hs, hemal spine; lj, lower
jaw; np, neurapophysis (gray); ns, neural spine; pp, pleurapophysis;
uj, upper jaw; 1–4, first through fourth head segments.
hypotheses. The two questions were: (1) Are all
vertebrate skulls constructed on the same plan? and
(2) If such a plan exists, is it identical to that of the
vertebral column? He also realized that only two
methods could be used to answer these questions:
comparative adult anatomical studies and comparative
developmental studies. In the first case, Huxley
considered it necessary to examine an extensive series
of skulls and vertebral columns and in the second case
he believed it was necessary to compare the embryonic
development of skulls and vertebral columns. He stated
that comparative developmental evidence was the
stronger of the two, but did not list the reasons on
which he based this conclusion. It is possible that
Huxley believed that if two adult structures arise from
the same embryonic source and pass through the same
developmental stages, they must be homologous. It is
also possible, however, that he emphasized development because Owen lacked expertise in this area
(Rupke 1994).
For whatever reasons, Huxley concluded that all
vertebrate skulls do exhibit a common plan, but that
skulls and vertebrae arise from different embryonic
sources and develop through very different stages.
Huxley noted that only the caudalmost portion of
the skull includes the notochord, one of the main
embryonic sources of vertebral formation; the
notochord is totally absent from the rostral twothirds of the skull, which arises from trabecular
mesenchyme, a tissue unique to the skull.
In an elegant conclusion, Huxley stated:
‘‘I confess I do not perceive how it is possible,
fairly and consistently to reconcile these facts with
Head segmentation hypotheses
613
any existing theory of the vertebrate composition
of the skull, except by drawing ad libitum upon
the Deus ex machina of the speculator, - imaginary
‘‘confluences’’, ‘‘connations’’, ‘‘irrelative repetitions’’, and shiftings of position – by whose skilful
application it would not be difficult to devise half
a dozen very pretty vertebral theories, all equally
true, in the course of a summer’s day.’’
Although different vertebral hypotheses continued
to appear in some anatomical texts for a few years
after Huxley’s lecture, comparative studies shifted
heavily to a developmental paradigm.
Fig. 2 Lateral view of a reconstruction of the brain and cephalic
mesoderm (gray) of an early embryo of the spiny dogfish (Squalus
acanthias). Redrawn from Jarvik (1980). h, head; lp, lateral plate
mesoderm; not, notochord; nt, neural tube; otv, otic vesicle; ov,
optic vesicle; ps, pharyngeal slit; pv, Platt’s vesicle; s, somite;
t, trunk; 1–8, first through eighth head somites (head cavities).
Mesodermal segmental hypotheses
In 1878, Balfour published the first detailed description of the development of the shark Scyllium
(Scyliorhinus). Balfour claimed that the mesodermally
derived muscle plates of the trunk, now recognized
as somites, continued into the head as a series of
hollow epithelial sacs that he termed head cavities.
He recognized eight of these (Fig. 2) and suggested
that they were evidence that the head consisted of
eight segments, which he termed, rostrocaudally, the
preoral or premandibular, the mandibular, the hyoid,
and the first through the fifth branchial segments.
Shortly afterwards, Marshall (1881) confirmed
Balfour’s observations and suggested that the lateral
plate mesoderm, which forms smooth muscle in the
trunk, also continues into the head, where it is
divided into branchiomeres by the developing pharyngeal pouches. In contrast to the trunk, however,
the lateral plate mesoderm was believed to form
striated muscle. In 1891, Julia Platt looked at head
development in the spiny dogfish, Squalus acanthias,
and described an additional head cavity rostral to the
premandibular head cavity described by Balfour and
Marshall. The discovery of this additional head
cavity, which is known as Platt’s vesicle (Figs. 2
and 3), raised the possibility that there was a total of
nine head somites and thus nine segments, which
would have included two premandibular segments.
Additional studies on Scyllium and other elasmobranches (van Wijhe 1882; Dohrn 1890; Froriep 1894)
indicated a variable number of head somites ranging
from 12 to 15. In an attempt to resolve these
differences, Goodrich (1918) re-examined head development in Scyllium. He recognized the premandibular,
mandibular, and hyoid head cavities of Balfour,
but he was also able to trace the head mesoderm
rostral to the premandibular head cavity, where it
ended rostral to the foregut as a sheet of cells that is
presently recognized as the prechordal plate. It is
within this plate in Squalus, that Platt described the
Fig. 3 Photomicrographs of horizontal sections through dorsal
(A) and ventral (B) levels of the head of an embryonic spiny
dogfish, Squalus acanthias, 7.5 mm in total length. The preotic
head cavities are particularly prominent as hollow epithelial sacs.
d, diencephalon; h, hyoid head cavity; hb, hindbrain;
m, mandibular head cavity; n, notochord; oc, optic cup; og,
otic ganglion; ov, otic vesicle; p, premandibular head cavity; pv,
Platt’s vesicle. Bar scale equals 200 mm.
development of her vesicle. In Scyllium, however,
Goodrich claimed that the prechordal plate disappears
without ever forming a vesicle. In Squalus, however,
Platt (1891) and Lamb (1902) believed that her vesicle
formed myoblasts before degenerating. In any case,
Goodrich (1918) concluded that Platt’s vesicle had no
phylogenetic significance and that only three preotic
614
head cavities or somites existed in the earliest
vertebrates.
Goodrich also attempted to resolve another
problem with head somites. In contrast to Balfour
(1878) and Marshall (1881), both of whom recognized eight head somites, van Wijhe (1882) attributed nine segments to the head, but believed that the
pharyngeal slit and branchiomere associated with the
fourth head somite had been lost so that the ‘‘facial’’
nerve innervated two somites. In re-examining head
development in Scyllium, Goodrich saw no evidence
for the loss of a pharyngeal slit and associated
branchiomere, but he did see evidence for the loss of
at least one, if not two, head somites. As the otic
capsule develops in Scyllium, it comes to lie over the
caudal portion of the third somite and most of
the fourth somite (Fig. 2). Goodrich believed that the
expansion of the otic capsule destroyed the caudal
portion of the third somite and all of the fourth
somite, so that no distinct myotome forms in this
region. He also believed that the fifth head somite
was affected by the otic capsule, so that only the
most caudal portion of the fifth somite formed
myoblasts, and he was unable to identify a nerve
innervating these myoblasts. Therefore, he concluded
that head somites four and five failed to form
somatic muscle. Similar conclusions were reached
regarding the fourth and fifth head somite in Squalus
by de Beer (1922), 4 years later.
R. G. Northcutt
In his 1918 paper, Goodrich published what may
have become the single most important morphological illustration of the 20th Century (Fig. 4), in
which he indicated the fate of myotomes and
branchiomeres, as well as their innervation in each
of the eight segments that he believed constituted the
primitive head of jawed vertebrates. He reprinted
this figure in his magnus opus (Goodrich 1930), as
did de Beer (1937) in his treatise on the vertebrate
skull. Goodrich’s model of the organization of the
vertebrate head has been widely adopted and has
been reproduced in almost all American textbooks of
comparative anatomy.
Many European textbooks of comparative anatomy, however, have adopted another segmental
mesodermal model based primarily on the morphological studies of Bjerring (1970, 1972, 1977, 1984)
and Jarvik (1980). These authors believed that Platt’s
vesicle gives rise to the inferior oblique muscle,
which was claimed by Holmgren (1940) to be
evidence that the head of the first jawed vertebrates
consisted of two premandibular segments, in addition to the remaining seven segments of Goodrich’s
model. Although they accepted van Wijhe’s claim
(1882) that a pair of dorsal and ventral nerves
characterizes each head segment; they suggested that
the terminal nerve is the dorsal nerve of the first
premandibular segment; they divided the oculomotor
nerve into two nerves for the first and second
Fig. 4 Schematic lateral view of a vertebrate head, illustrating Goodrich’s model of head segmentation (1918). Goodrich believed
that the head consisted of eight segments, and that each segment was composed of a dorsal somite and more ventral branchiomeres.
The first through the third somites were thought to form the extrinsic eye muscles, the fourth and fifth somites were believed to
degenerate, and the last three somites (6–8) were said to form the hypobranchial muscles. Branchiomeres were believed to form from
lateral plate mesoderm and, in turn, to form the jaw and gill muscles associated with each head segment, except for the first head
segment, which was believed to lack a branchiomere, as it was rostral to the mouth and pharynx. Goodrich adopted van Wijhe’s (1882)
conclusion that each head segment is characterized by a dorsal cranial nerve (V, VII, IX, X) that innervates a branchiomere and a ventral
cranial nerve (III, IV, VI, XII) that innervates a somite. From Goodrich (1918).
615
Head segmentation hypotheses
premandibular segments; and they claimed that they
had discovered an additional nerve, nervus tenuis,
said to supply their last preotic somite, which, unlike
Goodrich, they believed forms a posterior basiotic
muscle (Bjerring 1970).
A number of criticisms can be raised regarding the
models of Goodrich and Jarvik-Bjerring (Kingsbury
1926; Romer 1972; Northcutt 1990, 1993). Although
Kingsbury’s and Romer’s criticisms predate the
publication of the Jarvik-Bjerring model, their
criticisms nonetheless apply to it as well as to
Goodrich’s model. Kingsbury (1926) emphasized that
in order for the vertebrate head to be segmented,
each head segment should contain a number of
iterative structures, such as a neuromere, a pair of
dorsal and ventral cranial nerves, and a mesodermal
somite. Kingsbury accepted the earlier claims that
both the head somites and pharyngeal arches form
iterative series but believed there was no evidence
that they were in topographical registry. He noted
that segmentation of the paraxial mesoderm in the
head developmentally precedes formation of the
pharyngeal pouch, and he concluded that segmentation of the paraxial mesoderm (mesomerism) and
lateral plate mesoderm (branchiomerism) had
occurred independently.
Similar concerns were raised by Romer (1972),
who believed that the head of vertebrates consisted
of somatic and visceral units, which arose phylogenetically at different times. He argued that the head of
the earliest deuterostomes essentially consisted of a
visceral branchial basket used for feeding and that a
somatic series of head muscles was added later as
part of a locomotive apparatus. Thus, he concluded:
‘‘It may happen by chance that during development some one gill bar and its musculature may
lie below some one specific myotome and its
derived musculature. But there is no a priori
reason to think that the two segmental systems –
one basically mesodermal and related to the
‘‘somatic’’ animal, the other basically endodermal,
‘‘visceral’’ in origin – have any necessary relationship to one another.’’
The authors of all mesodermal segmental models, as
well as the earlier critics of these models (Kingsbury
1926; Romer 1972), accepted Marshall’s initial claim
(1881) that branchiomeric muscles arise from lateral
plate mesoderm. If this were true, however, branchiomeric muscles should consist of smooth muscle fibers,
not striated muscle fibers, and these muscles should be
innervated in the same manner as gut muscle, by first
and second order visceral motor neurons. Only Romer
(1972) addressed the discrepancy of muscle type by
arguing that branchiomeric muscles were striated due
to ‘‘the functional need for more efficient musculature
in the mouth and pharynx region.’’ Amazingly, the
differences in the innervation patterns of smooth and
striated muscle were already known (Gaskell 1886;
Langley 1893) at the time that Goodrich formulated
his head segmentation model in 1918, but no attempt
was ever made by any of the authors of the mesodermal
segmental models to account for this discrepancy in
innervation.
Fortunately, the discrepancy in muscle type was
resolved experimentally in 1983 when Noden transplanted segments of paraxial head mesoderm between
quail and chick embryos and demonstrated that
branchiomeric muscles arise from paraxial mesoderm, not lateral plate mesoderm. This seminal
observation not only substantially altered all mesodermal segmental models but also negated Romer’s ‘‘dual
animal’’ model and blunted much of Kingsbury’s
arguments concerning the independent segmentation
of paraxial and lateral plate mesoderm. Although
Noden’s experiments also explain the way in which
branchiomeric muscles are innervated, they do not
address the issue raised by Kingsbury (1926) and
Northcutt (1993) of the claimed alignment of cranial
nerves and paraxial mesoderm. This issue will be
considered in the next section on neuroepithelial
segmental hypotheses.
Neuroepithelial segmental hypotheses
The cephalic neural tube in vertebrate embryos exhibits
a series of transitory transverse ridges, termed neuromeres, which are separated by furrows. These neural
segments have been interpreted as additional evidence
of head segmentation (Locy 1895; Neal 1898, 1918;
Vaage 1969). After their initial discovery, and a brief
period in which neuromeres were extensively described
in many vertebrates, they were generally afforded less
significance and were believed to be artifacts, due to
histological fixation, or localized swellings of the neural
tube, due to stress points where nerve roots enter or
exit the brain, or a reflection of localized neuronal
migration. The lack of agreement on their number,
reported to be between 4 and 15 (Neal 1898), further
reinforced the notion that they were neural artifacts.
Thus, Goodrich (1930) and Bjerring (1984) did not
even mention neuromeres, and Jarvik (1980) dismissed
them as having no significance for the understanding of
head organization.
It was not until Lumsden and Keynes (1989) revived
Neal’s earlier (1898, 1918) claim that most cranial
motor nuclei arise from adjacent pairs of neuromeres,
and experimentally confirmed this claim, and the
616
discovery that hindbrain neuromeres, termed rhombomeres, are determined by the Hox genes (reviewed
by Wilkinson and Krumlauf 1990), that neuromeres
came to be regarded as critical elements in the
development of the brain and thus important to
our understanding of the organization of the head
(Noden 1991; Gilland and Baker 1993; Northcutt
1993). In addition, the discovery that the expression
patterns of an extensive series of developmental genes
allow recognition of midbrain and forebrain segments
(reviewed by Puelles and Rubenstein 1993, 2003) has
allowed a resolution of most of the discrepancies
regarding the number and extent of the more rostral
neuromeres.
At present, the most widely accepted neuromeric
model (Puelles and Rubenstein 2003) recognizes that
the hindbrain consists of an isthmus (r0), seven distinct
additional rhombomeres, and two or three pseudorhombomeres, which have many of the cellular and
histochemical patterns of the more rostral rhombomeres but are not delineated by distinct sulci (Fig. 5).
The midbrain appears to consist of a single neuromere,
termed the mesomere (Fig. 5). The forebrain consists of
four segments: a pretectum (p1), a dorsal thalamus
(p2), a ventral thalamus or prethalamus (p3), and a
prechordal segment, the secondary prosencephalon,
which consists of the hypothalamus, the optic vesicle,
and the telencephalon (Fig. 5).
The neuromeric origin of the motor neurons of the
cranial nerves has been experimentally determined in a
number of species (Lumsden and Keynes 1989; Gilland
1992; Gilland and Baker 1993), and the segmental
locations of these efferent neurons are generally
R. G. Northcutt
conserved. The oculomotor and trochlear motor
neurons arise in the mesomere and the isthmic
rhombomere (r0), respectively (Fig. 5). The origin of
the abducent motor neurons is varied, as these neurons
may arise only in rhombomere 5 (mammals), only in
rhombomere 6 (elasmobranchs), or in both rhombomeres 5 and 6 (teleosts and birds). In contrast, the
motor neurons of the hypoglossal cranial nerve appear
to arise within pseudorhombomere 8 in all vertebrates
that have been examined. The motor neurons of the
branchiomeric cranial nerves generally have been
described as arising from neuromeric pairs. Thus, the
motor neurons of the trigeminal and facial cranial
nerves arise from rhombomeres 2 and 3, and 4 and 5,
respectively in most vertebrates that have been
examined. Similarly, the motor neurons of the
glossopharyngeal cranial nerve arise from rhombomeres 6 and 7 (Fig. 5). In mammals, however, both the
trigeminal and facial motor nuclei each arise from three
neuromeres (Gilland and Baker 1993). The trigeminal
motor neurons arise from caudal rhombomere 1, as
well as rhombomeres 2 and 3, and the facial motor
nucleus arises from rhombomeres 4 and 5, as well as
rhombomere 6. There is also variation in the
neuromeric origin of vagal motor neurons. These
neurons arise from rhombomeres 7 and 8 in most
vertebrates, but in Squalus they appear to arise only
from rhombomere 8.
Clearly, both the cephalic neural tube and the
cephalic paraxial mesoderm form iterative units, but
are the two series aligned? Neal (1918) concluded
that they were, and he believed there was embryological evidence that both the brain and the paraxial
Fig. 5 Schematic lateral view of the head of an embryonic spiny dogfish, Squalus acanthias, illustrating the relative positions of the
neuromeres, the embryonic origin of the motor nuclei of the cranial nerves, and the derivatives of the paraxial mesoderm (gray).
Modified from Neal (1918), de Beer (1922), Gilland (1992), and Gilland and Baker (1993). The boundaries of the prosomeres are only
approximate and are based on the Puelles and Rubinstein (2003) model. bt1, branchiomere 1; gl, glossopharyngeal mesoderm; hy, hyoid
mesoderm; m, mesomere; ma, mandibular mesoderm (head cavity); p, premandibular mesoderm (head cavity); ps, pharyngeal slit;
pv, Platt’s vesicle; p1–3, prosomeres 1–3; r0–8, rhombomeres 0–8; sp, secondary prosencephalon; s3, third somite; III, oculomotor
nerve; IV, trochlear nerve; V, trigeminal nerve; VI, abducent nerve; VII, facial nerve; IX, glossopharyngeal nerve; X, vagal nerve.
Head segmentation hypotheses
mesoderm were divided into seven segments, which
were aligned. He believed that the forebrain arose from
the first neuromere and was aligned with the
premandibular somite (head cavity). The midbrain
was believed to arise from the second neuromere and to
be aligned with the mandibular somite (head cavity).
Neal’s third neuromere was subdivided to form
rhombomeres 1 and 2, which were aligned with the
hyoid somite (head cavity). Finally, neuromeres 4
through 7 (rhombomeres 3 through 6) were aligned
with his somites 4 through 7. Neal’s model is not supported, however, by the neuromeric model of Puelles
and Rubinstein (2003), which is based heavily on the
expression of a number of developmental genes for
recognition of individual neuromeres and their
boundaries. This new model suggests that the brain is
divided into at least 14 neuromeres, which can not be
aligned with the recognized paraxial mesodermal
subdivisions (Fig. 5) or with Goodrich’s model of
eight head segments (Fig. 4). In fact, in amniotic
embryos the paraxial mesoderm is not divided at all.
Whereas the cephalic paraxial mesoderm in many
anamniotic embryos is clearly divided into a series of
epithelial head cavities, the cephalic paraxial mesoderm
in amniotic embryos is only partially constructed,
at best, into units that have been termed somitomeres
(see Jacobson 1993, for an extensive review). Those
researchers who have recognized somitomeres have
reported them in varying numbers in different
vertebrate groups. What is notable is that somitomeres,
if they exist at all, do not exist in sufficient number to
be in registry with 14 neuromeres.
Furthermore, the origin and distribution of the
cranial nerves do not support van Wijhe’s (1882)
contention that these nerves correspond to the dorsal
and ventral spinal nerves of lampreys, a position that
was also adopted in Goodrich’s model (1918, 1930). In
1882, van Wijhe examined the development of the
spinal and cranial nerves of Scyliorhinus and concluded
that the spinal nerves, as well as most of the cranial
nerves, initially develop as separate dorsal and ventral
nerves. Subsequently, the spinal nerves were said to
fuse, except in lampreys, where separate dorsal and
ventral spinal nerves were said to be retained. Based on
these observations, van Wijhe concluded that most of
the cranial nerves constitute pairs of dorsal and ventral
nerves that do not fuse, and he argued that all cranial
nerves can be grouped into three series: (1) special
sensory nerves of the paired sense organs; (2) a dorsal
series of nerves (profundal, trigeminal, facial, glossopharyngeal, and vagal); and (3) a ventral series of
nerves (oculomotor, trochlear, abducent, and hypoglossal). Over the next 40 years, the distribution of the
sensory and motor components of these nerves was
617
established (Strong 1895; Cole 1896; Allis 1897; Herrick
1899; Norris and Hughes 1920), and it was noted that
many, if not all, of the dorsal cranial nerves contain
sensory fibers that innervate lateral line and gustatory
organs. These fibers were interpreted as specialized
somatic components (innervating the lateral line
organs) and specialized visceral components (innervating the gustatory organs), which were retained in
the dorsal cranial nerves but lost in the dorsal spinal
nerves (Strong 1895). Although this view became
predominant (Goodrich 1930; Romer 1970; Jarvik
1980), at least two researchers (Pollard 1892 and
Cole 1896) noted that lateral line organs are innervated
by sensory fibers whose cell bodies form ganglia
distinctly separate from those of the dorsal cranial
nerves and proposed that the cranial nerves innervating lateral line organs should be considered a separate
series of nerves. Furthermore, comparative embryological studies (Landacre and Conger 1913; Stone 1922;
Northcutt and Brändle 1995; O’Neill et al. 2007) have
demonstrated that these nerves and their organs
arise from a series of cephalic dorsolateral placodes.
Today, there is substantial evidence that six pairs of
lateral line nerves are derived from these placodes
(Song and Northcutt 1991; Northcutt and Bemis 1993;
Northcutt and Brändle 1995) and form an additional
series of cranial nerves.
The sensory ganglia of spinal nerves arise only from
the neural crest, whereas many sensory ganglia of the
‘‘dorsal’’ cranial nerves arise from epibranchial placodes as well as the neural crest. The epibranchial
placodes are a series of ectodermal thickenings located
dorsally adjacent to the pharyngeal slits, which are
induced by the pharyngeal endoderm (Begbie et al.
1999; Matz and Northcutt 1999) and are believed to
innervate taste buds (Gross et al. 2003). Sensory
neurons that arise from these placodes contribute to
the geniculate ganglion of the facial cranial nerve and
form the separate distal sensory ganglia of the
glossopharyngeal and vagal nerves. While the glossopharyngeal nerve exhibits only a single distal, epibranchially derived sensory ganglion, the vagal nerve is
frequently characterized by four distal, epibranchially
derived sensory ganglia. It is possible that these
gustatory sensory ganglia, like the sensory ganglia of
the lateral line nerves, are evidence that vertebrates
originally possessed a separate series of six gustatory
cranial nerves that have secondarily become associated
with the neural crest derived sensory ganglia of the
‘‘dorsal’’ cranial nerves (Northcutt 1990; Butler and
Hodos 1996). If so, most anamniotic vertebrates
actually possess 25 pairs of cranial nerves, (the terminal
nerve, the 12 traditionally recognized cranial nerves,
6 lateral line nerves, and 6 gustatory nerves) a number
618
R. G. Northcutt
that cannot be accommodated by any single existing
segmental model of the vertebrate head.
Bjerring HC. 1972. The nervus rarus in coelacanthiform
phylogeny. Zool Scr 1:57–68.
Conclusions
Bjerring HC. 1977. A contribution to structural analysis of the
head of craniate animals. Zool Scr 6:127–83.
Research over the past 200 years has resulted in a
wealth of information on the organization of the
vertebrate body plan. Many of the evolutionary steps
leading to the emergence of this body plan—which is
impacted by the neural crest, neurogenic placodes,
and the muscularization of lateral plate mesoderm—
have been elucidated, and during the past 20 years,
there has been remarkable progress in understanding
the genetic framework for the developmental basis of
this body plan. Unfortunately, all segmental hypotheses for explaining this plan have proven inadequate
so far. There is no question that the vertebrate head
consists of a large number of iterative structures,
involving the brain, neurogenic placodes, and paraxial mesoderm, but there is no obvious alignment of
these cephalic serial structures. Such alignment must
exist for there to be any possibility that the
vertebrate head is segmented (i.e., metameric). Even
if an alignment were discovered, this would not
necessarily demonstrate segmentation. Such demonstration would also require the discovery of genetic
mechanisms that underlie and conserve the registration of neuromeres, paraxial subdivisions, pharyngeal
pouches, and neurogenic placodes. To paraphrase
Huxley (1858), no existing theory, and perhaps no
future theory, will, even after many summer days,
demonstrate head segmentation in vertebrates.
Acknowledgments
I thank Shigeru Kuratani and Thomas Schilling for
inviting me to participate in their symposium on
vertebrate head segmentation. I also thank Jo
Griffith for assistance with the illustrations, Susan
Commerford for literature retrieval and word processing, and Mary Sue Northcutt for assistance with
numerous phases of the research and preparation of the
manuscript. This research was supported, in part, by
the National Science Foundation (IBN-0236018).
References
Allis EP. 1897. The cranial muscles and cranial and first spinal
nerves in Amia calva. J Morphol 12:487–808.
Balfour FM. 1878. A Monograph on the development of
elasmobranch fishes. London: Macmillan and Co.
Begbie J, Brunet J-F, Rubenstein JLR, Graham A. 1999.
Induction of the epibranchial placodes. Development
126:895–902.
Bjerring HC. 1970. Nervus tenuis, a hitherto unknown cranial
nerve of the fourth metamere. Acta Zool 51:107–14.
Bjerring HC. 1984. Major anatomical steps toward craniotedness: a heterodox view based largely on embryological data.
J Vertebr Paleontol 4:17–29.
Butler AB, Hodos W. 1996. Comparative vertebrate neuroanatomy. Evolution and adaptation. New York: Wiley-Liss.
Cole FJ. 1896. On the cranial nerves of Chimaera monstrosa
(Linn): with a discussion of the lateral line system and of
the morphology of the chorda tympani. Trans R Soc Edinb
38:631–80.
de Beer GR. 1922. The segmentation of the head in Squalus
acanthias. Q J Microsc Sci 66:457–74.
de Beer GR. 1937. The development of the vertebrate skull.
Oxford: Oxford University Press (reprinted 1985, Chicago:
University of Chicago Press).
Desmond A. 1982. Archetypes and ancestors: paleontology in
Victoria London, 1850–1875. London: University of
Chicago Press.
Dohrn A. 1875. Der Ursprung der Wirbeltiere und das Prinzip des
Functionswechsels. Leipzig (Germany): Genealogische Skizzen.
Dohrn A. 1890. Neue Grundlagen zur Beurteilung der
Metamerie des Kopfes. Mitt Zool Stat Neapel 9:330–434.
Froriep A. 1894. Entwickelung des Kopfes. Ergeb Anat
Entwicklungsgesch 3:391–459.
Gaskell WH. 1886. On the structure, distribution, and
function of the nerves which innervate the visceral and
vascular systems. J Physiol 7:1–80(+ 4 plates).
Gaskell WH. 1908. The origin of vertebrates. London: Green
& Co.
Geoffroy St.-Hillaire E. 1818. Philosophie anatomique. Paris:
MBquignon-Marvis.
Gilland E. 1992. Morphogenesis of segmental units in the
chordamesoderm and neuroepithelium of Squalus acanthias
[dissertation]. Cambridge, MA: Harvard University. Available
through University Microfilms International, Ann Arbor, MI.
Gilland E, Baker R. 1993. Conservation of neuroepithelial and
mesodermal segments in the embryonic vertebrate head.
Acta Anat 148:110–23.
Goethe JW. 1820. Zur Naturwissenchaft überhaupt, besonders
zur Morphologie. 1. Stuttgart (Germany): Cotta.
Goodrich ES. 1918. On the development of the segments of
the head in Scyllium. Q J Microsc Sci 63:1–30.
Goodrich ES. 1930. Studies on the structure and development
of vertebrates. London: Macmillan (reprinted 1986,
Chicago: University of Chicago Press).
Gross JB, Gottlieb AA, Barlow LA. 2003. Gustatory neurons
derived from epibranchial placodes are attracted to, and
trophically supported by, taste bud-bearing endoderm
in vitro. Dev Biol 264:467–81.
Hall BK, Hanken J. 1985. Forward. In: de Beer GR, editor.
The development of the vertebrate skull. Chicago and
London: University of Chicago Press. p. vii–xxviii.
Head segmentation hypotheses
Herrick CJ. 1899. The cranial and first spinal nerves of
Menidia: a contribution upon the nerve components of the
bony fishes. J Comp Neurol 9:153–455.
619
Northcutt RG. 1993. A reassessment of Goodrich’s model of
cranial nerve phylogeny. Acta Anat 148:71–80.
Holland PWH. 2000. Embryonic development of heads,
skeletons and amphioxus: Edwin S. Goodrich revisited.
Int J Dev Biol 44:29–34.
Northcutt RG, Bemis WE. 1993. Cranial nerves of the
coelacanth,
Latimeria
chalumnae
[Osteichthyes:
Sarcoptergii: Actinistia], and comparisons with other
Craniata. Brain Behav Evol 42 Suppl 1:1–76.
Holmgren N. 1940. Studies on the head in fishes.
Embryological, morphological and phylogenetic researches.
Acta Zool 21:51–267.
Northcutt RG, Brändle K. 1995. Development of branchiomeric and lateral line nerves in the axolotl. J Comp Neurol
355:427–54.
Huxley TH. 1858. On the theory of the vertebrate skull. Proc
R Soc Lond 9:381–457.
Oken L. 1807. Uber die Bedeutung der Schadelknochen. Jena:
JCG Göpferdt.
Jacobson AG. 1993. Somitomeres: mesodermal segments of
the head and trunk. In: Hanken J, Hall BK, editors. The
Skull. Vol. 1. Development. Chicago and London:
University of Chicago Press. p. 42–76.
O’Neill P, McCole RB, Baker CVH. 2007. A molecular
analysis of neurogenic placode and cranial sensory ganglion
developmeant in the shark, Scyliorhinus canicula. Dev Biol
304:156–81.
Jarvik E. 1980. Basic structure and evolution of verebrates. 2
Vols. London: Academic Press.
Owen R. 1848. On the archetype and homologies of the
vertebrate skeleton. London: John Van Voorst:
Paternoster Row.
Kingsbury BF. 1926. Branchiomerism and the theory of head
segmentation. J Morphol 42:82–109.
Lamb AB. 1902. The development of the eye muscles in
acanthias. Am J Anat 1:185–202.
Landacre FL, Conger AC. 1913. The origin of the lateral
line primordia in Lepidosteus osseus. J Comp Neurol
23:575–633.
Langley JN. 1893. Preliminary account of the arrangement
of the sympathetic nervous system based chiefly on
observations upon pilomotor nerves. Proc R Soc Lond
52:547–56.
Locy WA. 1895. Contribution to the structure and development of the vertebrate head. J Morphol 11:497–594.
Lumsden A, Keynes R. 1989. Segmental patterns of neuronal
development in the chick hindbrain. Nature 337:424–8.
Marshall AM. 1881. On the head cavities and associated
nerves of elasmobranches. Microsc J 21:72–97.
Matz Stuart P, Glenn Northcutt R. 1999. Pharyngeal endoderm,
not chordamesoderm, induces epibranchial placodes in
axolotls to form neurons. Soc Neurosci Abstr 25:1545.
Neal HV. 1898. The segmentation of the nervous system in
Squalus acanthias. Bull Mus Comp Zool Harv 31:146–294
(þ Nine Plates).
Pollard HB. 1892. The lateral line system in siluroids. Zool
Jahrb Anat 5:525–51.
Platt JB. 1891. A contribution to the morphology of the
vertebrate head, based on a study of Acanthias vulgaris.
J Morphol 5:79–112.
Puelles L, Rubinstein JLR. 1993. Expression patterns of
homeobox and other putative regulatory genes in the
embryonic mouse forebrain suggest a neuromeric organization. Trends Neurosci 16:472–9.
Puelles L, Rubinstein JLR. 2003. Forebrain gene expression
domains and the evolving prosomeric model. Trends
Neurosci 26:469–76.
Romer AS. 1970. The vertebrate body. 4th ed. Philadelphia,
PA: Saunders.
Romer AS. 1972. The vertebrate as a dual animal–somatic and
visceral. Evol Biol 6:121–56.
Rupke NA. 1994. Richard Owen: victorian naturalist. New
Haven, CT: Yale University Press.
Song J, Northcutt RG. 1991. Morphology, distribution
and innervation of the lateral-line receptors of the
Florida Gar, Lepisosteus platyrhincus. Brain Behav Evol
37:10–37.
Neal HV. 1918. Neuromeres and metameres. J Morphol
31:293–315.
Stone LS. 1922. Experiments on the development of the
cranial ganglia and the lateral line sense organs in
Amblystoma punctatum. J Exp Zool 35:421–96.
Noden DM. 1983. The embryonic origins of avian cephalic
and cervical muscles and associated connective tissues. Am
J Anat 168:257–76.
Strong OS. 1895. The cranial nerves of amphibia. J Morphol
10:101–231.
Noden DM. 1991. Vertebrate craniofacial development: the
relation between ontogenetic process and morphological
outcome. Brain Behav Evol 38:190–225.
Vaage S. 1969. The segmentation of the primitive neural tube
in chick embryos (Gallus domesticus). A morphological,
histochemical and autoradiographic investigation. Ergeb
Anat Entwicklungsgesch 41:3–87.
Norris HW, Hughes SP. 1920. The cranial, occipital, and
anterior spinal nerves of the dogfish, Squalus acanthias.
J Comp Neurol 31:293–401.
van Wijhe JW. 1882. Ueber die Mesodermsegmente und die
Entwicklung der Nerven des Selachierkopfes. Verh K Akad
Wetenschappen 22:1–50.
Northcutt RG. 1990. Ontogeny and phylogeny: a re-evaluation of conceptual relationships and some applications.
Brain Behav Evol 36:116–40.
Wilkinson DG, Krumlauf R. 1990. Molecular approaches to
the segmentation of the hindbrain. Trends Neurosci
13:335–9.