In 1829, von Baer proposed four principles of development, the first of which states that the embryos of a large group of animals have most in common at early stages of development, the specialised features of different subgroups only emerging later. After the publication of Darwin's The Origin of Species in 1859, the relevance of von Baer's principles to evolution was clear to many biologists, who assessed evolutionary relationships on the basis of similarities and differences between early embryos. The validity of these comparisons, carried out more than a century ago, has been confirmed by recent molecular analyses. These new gene-based comparisons have in turn led to a renewed interest in the scientific literature of the first half of the 20th century, which documented in great detail comparative anatomical studies of living and fossil animals.

The past decade has seen a growing synergy between developmental biology and evolutionary studies. As a result, not only is an evolutionary perspective now integral to developmental studies, but palaeontologists have become deeply interested in embryos. This integration was the subject of the symposium of the Winter Meeting of the Anatomical Society, held in London in January 2001, entitled ‘The Evolution of Developmental Mechanisms’. It was a very stimulating and enjoyable meeting that addressed the theme at a number of levels, ranging from the major evolutionary changes resulting from gene duplication to the fine detail of insect wing vein evolution. These two ends of the spectrum of topics reflect Darwin's perspective on von Baer's principles, i.e. his recognition that homologies between different phyla are to be seen in early embryonic and larval structures, while adaptations that enable an organism to survive in its particular environment are the result of changes acting late in development.

The reviews brought together in this volume represent a thoughtful and thought-provoking synthesis of data collected over many years. They show the rewards of detailed and long-term study of embryos that in some cases are difficult to obtain, but represent living clues to the mechanisms underlying important evolutionary transitions. We feel privileged to have had the responsibility for editing this impressive issue of the Journal of Anatomy, which will, we have no doubt, be a significant landmark in the ongoing story of development and evolution, and will be widely cited.

Regeneration poses a distinctive set of problems for evolutionary biologists, but there has been little substantive progress since these issues were clearly outlined in the monograph of T. H. Morgan (1901). The champions at regeneration among vertebrates are the urodele amphibians such as the newt, and we tend to regard urodele regeneration as an exceptional attribute. The ability to regenerate large sections of the body plan is widespread in metazoan phylogeny, although it is not universal. It is striking that in phylogenetic contexts where regeneration occurs, closely related species are observed which do not possess this ability. It is a challenge to reconcile such variation between species with a conventional selective interpretation of regeneration. The critical hypothesis from phylogenetic analysis is that regeneration is a basic, primordial attribute of metazoans rather than a mechanism which has evolved independently in a variety of contexts. In order to explain its absence in closely related species, it is postulated to be lost secondarily for reasons which are not understood. Our approach to this question is to compare a differentiated newt cell with its mammalian counterpart in respect of the plasticity of differentiation.

The arrangement of Hox genes into physical clusters is fundamental to the patterning of animal body plans, through the phenomenon of colinearity. Other homeobox genes are often described as dispersed, implying they are not arranged into clusters. Contrary to this view, however, two clusters of non-Hox homeobox genes have been reported: the amphioxus ParaHox gene cluster and the Drosophila 93D/E cluster (referred to here as the NKL cluster). Here I examine the antiquity of these gene clusters, their conservation and their pattern of evolution in vertebrate genomes. I argue that the ParaHox gene cluster arose early in animal evolution, and duplicated in vertebrates to give the four clusters in human and mouse genomes. The NKL cluster is also ancient, and also duplicated to yield four descendent clusters in mammalian genomes. The NKL and Hox gene clusters were originally chromosomal neighbours, within an ancient and extensive array of at least 30 related homeobox genes. There is no necessary relationship between clustering and colinearity, although it is argued that the ParaHox gene cluster does show modified spatial colinearity. A novel hypothesis for the evolution of ParaHox gene expression in deuterostomes is presented.

During insect development, morphological differences between segments are controlled by the Hox gene family of transcription factors. Recent evidence also suggests that variation in the regulatory elements of these genes and their downstream targets underlies the evolution of several segment-specific morphological traits. This review introduces a new model system, the larval oenocyte, for studying the evolution of fate specification by Hox genes at single-cell resolution. Oenocytes are found in a wide range of insects, including species using both the short and the long germ modes of development. Recent progress in our understanding of the genetics and cell biology of oenocyte development in the fruitfly Drosophila melanogaster is discussed. In the D. melanogaster embryo, the formation of this cell type is restricted to the first 7 abdominal segments and is under Hox gene control. Oenocytes delaminate from the dorsal ectoderm of A1-A7 in response to an induction that involves the epidermal growth factor receptor (EGFR) signalling pathway. Although the receptor itself is required in the presumptive oenocytes, its ligand Spitz (Spi) is secreted by a neighbouring chordotonal organ precursor (COP). Thus, in dorsal regions, local signalling from this component of the developing peripheral nervous system induces the formation of oenocytes. In contrast, in lateral regions of the ectoderm, Spi signal from a different COP induces the formation of secondary COPs in a homeogenetic manner. This dorsoventral difference in the fate induced by Spi ligand is controlled by a prepattern in the responding ectoderm that requires the Spalt (Sal) transcription factor. Sal protein is expressed in the dorsal but not lateral ectoderm and acts as a competence modifier to bias the response to Spi ligand in favour of the oenocyte fate. We discuss a recently proposed model that integrates the roles of Sal and the EGFR pathway in oenocyte/chordotonal organ induction. This model should provide a useful starting point for future comparative studies of these ectodermal derivatives in other insects.

Over the past 50 years and more, many models have been proposed to explain how the nervous system is initially induced and how it becomes subdivided into gross regions such as forebrain, midbrain, hindbrain and spinal cord. Among these models is the 2-signal model of Nieuwkoop & Nigtevecht (1954), who suggested that an initial signal (‘activation’) from the organiser both neuralises and specifies the forebrain, while later signals (‘transformation’) from the same region progressively caudalise portions of this initial territory. An opposing idea emerged from the work of Otto Mangold (1933) and other members of the Spemann laboratory: 2 or more distinct organisers, emitting different signals, were proposed to be responsible for inducing the head, trunk and tail regions. Since then, evidence has accumulated that supports one or the other model, but it has been very difficult to distinguish between them. Recently, a considerable body of work from mouse embryos has been interpreted as favouring the latter model, and as suggesting that a ‘head organiser’, required for the induction of the forebrain, is spatially separate from the classic organiser (Hensen's node). An extraembryonic tissue, the ‘anterior visceral endoderm’ (AVE), was proposed to be the source of forebrain-inducing signals. It is difficult to find tissues that are directly equivalent embryologically or functionally to the AVE in other vertebrates, which led some (e.g. Kessel, 1998) to propose that mammals have evolved a new way of patterning the head. We will present evidence from the chick embryo showing that the hypoblast is embryologically and functionally equivalent to the mouse AVE. Like the latter, the hypoblast also plays a role in head development. However, it does not act like a true organiser. It induces pre-neural and pre-forebrain markers, but only transiently. Further development of neural and forebrain phenotypes requires additional signals not provided by the hypoblast. In addition, the hypoblast plays a role in directing cell movements in the adjacent epiblast. These movements distance the future forebrain territory from the developing organiser (Hensen's node), and we suggest that this is a mechanism to protect the forebrain from caudalising signals from the node. These mechanisms are consistent with all the findings obtained from the mouse to date. We conclude that the mechanisms responsible for setting up the forebrain and more caudal regions of the nervous system are probably similar among different classes of higher vertebrates. Moreover, while reconciling the two main models, our findings provide stronger support for Nieuwkoop's ideas than for the concept of multiple organisers, each inducing a distinct region of the CNS.

Previous mouse models have indicated that Otx1 and Otx2 play an important role in brain and sense organ development and, together with the Drosophila orthodenticle (otd) gene, they share a high degree of reciprocal functional equivalence. Interestingly, mouse models replacing the same region of the Otx2 locus with Otx1, otd or lacZ genes have revealed the existence of a differential post-transcriptional control between the visceral endoderm (VE) and epiblast cells. Indeed Otx1, otd or lacZ mRNA were transcribed in both tissues but translated only in the VE. Embryos lacking OTX1 or OTD proteins in the epiblast and derived tissues, such as the neuroectoderm and axial mesendoderm (AME), fail to maintain the anterior identity and result in a headless phenotype. This finding leads us to hypothesise that, during evolution, the specification of the vertebrate-type brain may have required epiblast cells to translate Otx2 mRNA in order to establish maintenance properties. The establishment of this regulatory control might have been reflected into a remarkable reorganisation of the rostral CNS architecture and might have represented an important event in the evolution of the vertebrate head. Current data suggest that the Otx2 replaced region and in particular the 3′ untranslated region (UTR), may contain regulatory element(s) necessary to translate and/or stabilise Otx2 mRNA in epiblast and its derivatives.

The epithalamus is a major subdivision of the diencephalon constituted by the habenular nuclei and pineal complex. Structural asymmetries in this region are widespread amongst vertebrates and involve differences in size, neuronal organisation, neurochemistry and connectivity. In species that possess a photoreceptive parapineal organ, this structure projects asymmetrically to the left habenula, and in teleosts it is also situated on the left side of the brain. Asymmetries in size between the left and right sides of the habenula are often associated with asymmetries in neuronal organisation, although these two types of asymmetry follow different evolutionary courses. While the former is more conspicuous in fishes (with the exception of teleosts), asymmetries in neuronal organisation are more robust in amphibia and reptiles. Connectivity of the parapineal organ with the left habenula is not always coupled with asymmetries in habenular size and/or neuronal organisation suggesting that, at least in some species, assignment of parapineal and habenular asymmetries may be independent events.

The evolutionary origins of epithalamic structures are uncertain but asymmetry in this region is likely to have existed at the origin of the vertebrate, perhaps even the chordate, lineage. In at least some extant vertebrate species, epithalamic asymmetries are established early in development, suggesting a genetic regulation of asymmetry. In some cases, epigenetic factors such as hormones also influence the development of sexually dimorphic habenular asymmetries. Although the genetic and developmental mechanisms by which neuroanatomical asymmetries are established remain obscure, some clues regarding the mechanisms underlying laterality decisions have recently come from studies in zebrafish. The Nodal signalling pathway regulates laterality by biasing an otherwise stochastic laterality decision to the left side of the epithalamus. This genetic mechanism ensures a consistency of epithalamic laterality within the population. Between species, the laterality of asymmetry is variable and a clear evolutionary picture is missing. We propose that epithalamic structural asymmetries per se and not the laterality of these asymmetries are important for the behaviour of individuals within a species. A consistency of the laterality within a population may play a role in social behaviours between individuals of the species.

Recent studies of protochordates (ascidian tunicates and amphioxus) have given insights into possible ancestors of 2 of the characteristic features of the vertebrate head: neural crest and placodes. The neural crest probably evolved from cells on either side of the neural plate–epidermis boundary in a protochordate ancestral to the vertebrates. In amphioxus, homologues of several vertebrate neural crest marker genes (BMP2/4, Pax3/7, Msx, Dll and Snail) are expressed at the edges of the neural plate and/or adjacent nonneural ectoderm. Some of these markers are also similarly expressed in tunicates. In protochordates, however, these cells, unlike vertebrate neural crest, neither migrate as individuals through embryonic tissues nor differentiate into a wide spectrum of cell types. Therefore, while the protochordate ancestor of the vertebrates probably had the beginnings of a genetic programme for neural crest formation, this programme was augmented in the earliest vertebrates to attain definitive neural crest. Clear homologues of vertebrate placodes are lacking in protochordates. However, both amphioxus and tunicates have ectodermal sensory cells. In tunicates these are all primary neurons, sending axons to the central nervous system, while in amphioxus, the ectodermal sensory cells include both primary neurons and secondary neurons lacking axons. Comparisons of developmental gene expression suggest that the anterior ectoderm in amphioxus may be homologous to the vertebrate olfactory placode, the only vertebrate placode with primary, not secondary, neurons. Similarly, biochemical, morphological and gene expression data suggest that amphioxus and tunicates also have homologues of the adenohypophysis, one of the few vertebrate structures derived from nonneurogenic placodes. In contrast, the origin of the other vertebrate placodes is very uncertain.

The vertebrate inner ear forms a highly complex sensory structure responsible for the detection of sound and balance. Some new aspects on the evolutionary and developmental origin of the inner ear are summarised here. Recent molecular data have challenged the longstanding view that special sense organs such as the inner ear have evolved with the appearance of vertebrates. In addition, it has remained unclear whether the ear originally arose through a modification of the amphibian mechanosensory lateral line system or whether both evolved independently. A comparison of the developmental mechanisms giving rise to both sensory systems in different species should help to clarify some of these controversies. During embryonic development, the inner ear arises from a simple epithelium adjacent to the hindbrain, the otic placode, that is specified through inductive interactions with surrounding tissues. This review summarises the embryological evidence showing that the induction of the otic placode is a multistep process which requires sequential interaction of different tissues with the future otic ectoderm and the recent progress that has been made to identify some of the molecular players involved. Finally, the hypothesis is discussed that induction of all sensory placodes initially shares a common molecular pathway, which may have been responsible to generate an ‘ancestral placode’ during evolution.

Here we present ideas connecting the behaviour of the cranial neural crest during development with the venerable, perhaps incorrect, view that gill-supporting cartilages of an ancient agnathan evolved into the skeleton of an early gnathostome's jaw. We discuss the pattern of migration of the cranial neural crest ectomesenchyme in zebrafish, along with the subsequent arrangement of postmigratory crest and head mesoderm in the nascent pharyngeal segments (branchiomeres), in diverse gnathostomes and in lampreys. These characteristics provide for a plausible von Baerian explanation for the problematic inside-outside change in topology of the gills and their supports between these 2 major groups of vertebrates. We consider it likely that the jaw supports did indeed arise from branchiomeric cartilages.

Marsupial mammals have a distinctive reproductive strategy. The young are born after an exceptionally short period of organogenesis and are consequently extremely altricial. Yet because they must be functionally independent in an essentially embryonic condition, the marsupial neonate exhibits a unique suite of adaptations. In particular, certain bones of the facial region, most cranial musculature and a few additional structures are accelerated in their development. In contrast, central nervous system structures, especially the forebrain, are markedly premature at birth, resembling an embryonic d 11 or 12 mouse. This review examines the developmental processes that are modified to produce these evolutionary changes. The focus is on the early development of the neural plate, neural crest and facial region in the marsupial, Monodelphis domestica, compared with patterns reported for rodents. Neural crest begins differentiation and migration at the neural plate stage, which results in large accumulations of neural crest in the facial region at an early stage of development. The early accumulation of neural crest provides the material for the accelerated development of oral and facial structures. The first arch region is massive in the early embryo, and the development of the olfactory placode and frontonasal region is advanced relative to the forebrain region. The development of the forebrain is delayed in marsupials relative to the hindbrain or facial region. These observations illustrate how development may be modified to produce evolutionary changes that distinguish taxa. Further, they suggest that development is not necessarily highly conserved, but instead may be quite plastic.

A muscularised pharynx, with skeletal support, serving the dual functions of feeding and respiration, is a fundamental vertebrate characteristic. Embryologically, the pharyngeal apparatus has its origin in a series of bulges that form on the lateral surface of the embryonic head, the pharyngeal arches, whose development is complex. These structures are composed of a number of disparate embryonic cell types: ectoderm, endoderm, neural crest and mesoderm, whose development must be coordinated to generate the functional adult apparatus. In the past, most studies have emphasised the role played by the neural crest, which generates the skeletal elements of the arches, in directing pharyngeal arch development, but it has also become apparent that the endoderm plays a prominent role in directing arch development. Neural crest cells are not required for arch formation, their regionalisation nor to some extent their sense of identity. Furthermore, the endoderm is the major site of expression of a number of important signalling molecules, and this tissue has been shown to be responsible for promoting the formation of particular components of the arches. Thus vertebrate pharyngeal morphogenesis can now be seen to be a more complex process than was previously believed, and must result from an integration of both neural crest and endodermal patterning mechanisms. Interestingly, this also mirrors the fact that the evolutionary origin of pharyngeal segmentation predates that of the neural crest, which is an exclusively vertebrate characteristic. As such, the evolution of the vertebrate pharynx is also likely to have resulted from an integration between these 2 patterning systems. Alterations in the interplay between neural crest and endodermal patterning are also likely to be responsible for the evolutionary that occurred to the pharyngeal region during subsequent vertebrate evolution.

This review describes the evolutionary history of the mammalian skull vault as a basis for understanding its complex structure. Current information on the developmental tissue origins of the skull vault bones (mesoderm and neural crest) is assessed for mammals and other tetrapods. This information is discussed in the context of evolutionary changes in the proportions of the skull vault bones at the sarcopterygian-tetrapod transition. The dual tissue origin of the skull vault is considered in relation to the molecular mechanisms underlying osteogenic cell proliferation and differentiation in the sutural growth centres and in the proportionate contributions of different sutures to skull growth.

Teeth as a feeding mechanism in an oral cavity (mouth) are functionally and locationally linked with jaws. In fossils, teeth found in the oral cavity are usually linked with jaws, although mineralised structures with the same histology as teeth are known in fossils before jaws appeared. Denticles in the skin occur in both fossil and extant fish. Pharyngeal denticles also occur in both extant and fossil gnathostomes but in only a few fossil agnathans (thelodonts). Complex structures with dentine and enamel have been described in the earliest jawless vertebrates, conodonts. Such fossils have been used to suggest that teeth and jaws have evolved and developed independently. Our understanding of the developmental biology of mammalian tooth development has increased greatly in the last few years to a point where we now understand some of the basic genetic interactions controlling tooth initiation, morphogenesis and patterning. The aim of this review is to see what this developmental information can reveal about evolution of the dentition.

Although the 3 genera of living lungfish have different-shaped adult tooth plates, their larval stages have similar patterns of development. The sequence in the pattern of initiation of teeth and their modification through ontogeny in Neoceratodus hatchlings provides a developmental model for fossil hatchling tooth plates (smallest 1–2 mm) recovered as 3-dimensional dentitions from Andreyevichthys. This Late Devonian lungfish demonstrates that these also have a similar dentition pattern and suggests strongly conserved developmental processes. We postulate that a specific pattern of development, derived within lungfish, has been conserved in extant forms through evolution from the earliest known lungfish. The most basal early dipnoan, Diabolepis speratus, is also known from juveniles with tooth plates formed in this pattern.

The lungfish pattern is in marked contrast to the typical linear rows of teeth with lingual replacement for each tooth position, characteristic of most osteichthyan and chondrichthyan dentitions. Uniquely for lungfish, teeth are only added to the lateral ends of the radial rows in the palatal and lingual dentition and are consolidated into dental plates without loss through shedding. It is proposed that this tooth pattern is set up from primordial teeth at the patterning stage of the dentition, one in each dentate region of the larval jaws. Although in post-Devonian lungfish marginal dentate bones are absent in the adult, in both the fossil and extant hatchling, teeth are present and function on some of the marginal bones. This pattern of development and loss is described and we conclude that in both forms it is also based on a radial pattern of successive tooth initiation. We propose that this ontogenetic pattern constrained the phylogenetic pattern of adult form, through evolution of dipnoan dentitions from 360 MYBP until the present. The universality amongst dipnoans and the implications for such a conserved constraint in the developmental module for the dentition is discussed.

Vertebrate somitogenesis has been shown to be associated with a molecular oscillator, the segmentation clock, whose periodicity matches that of the process of somitogenesis. The existence of such a clock in presomitic mesoderm (PSM) cells was originally proposed in theoretical models such as the ‘clock and wavefront’. Molecular evidence for the existence of this clock in vertebrates has been obtained on the basis of the periodic expression of several genes, most of which are related to the Notch signalling pathway. These genes are expressed in a dynamic sequence which appears as a wave sweeping caudo-rostrally along the whole PSM once during each somite formation. Notch-pathway mouse and fish mutants lose the dynamic expression of the cycling genes, indicating that Notch signalling is required for their periodic expression, or is required to coordinate the oscillations between PSM cells. Therefore Notch signalling is either part of the mechanism of the oscillator itself or acts as a cofactor required for cycling gene expression. A further potentially important role for the segmentation clock is to periodically activate Notch signalling in the rostral presomitic mesoderm, thereby generating the periodic formation of somite boundaries.

The backbone or vertebral column is the defining feature of vertebrates and is clearly metameric. Given that vertebrae arise from segmented paraxial mesoderm in the embryo, this metamerism is not surprising. Fate mapping studies in a variety of species have shown that ventromedial sclerotome cells of the differentiated somite contribute to the developing vertebrae and ribs. Nevertheless, extensive studies in amniote embryos have produced conflicting data on exactly how embryonic segments relate to those of the adult. To date, much attention has focused on the derivatives of the somites, while relatively little is known about the contribution of other tissues to the formation of the vertebral column. In particular, while it is clear that signals from the notochord induce and maintain proliferation of the sclerotome, and later promote chondrogenesis, the role of the notochord in vertebral segmentation has been largely overlooked. Here, we review the established role of the notochord in vertebral development, and suggest an additional role for the notochord in the segmental patterning of the vertebral column.

The bagpipe-related homeobox-containing genes are members of the NK family. bagpipe (bap) was first identified in Drosophila and there are three different bagpipe-related genes in vertebrates. Only two of these are found in mammals, the Nkx3.1 and the Bapx1 (Nkx3.2) gene. The targeted mutation in the mouse Bapx1 gene shows a vertebral phenotype in which the ventromedial elements are lacking; these are the centra and the intervertebral discs. In addition, a region of gastric mesenchyme is abnormal. This mesenchyme surrounds the posterior region of the presumptive stomach and duodenum, and in the mutant fails to support normal development of the spleen. In Drosophila, bagpipe has a role in gut mesoderm and the mutant embryos have no midgut musculature. Thus bap related genes in mouse and Drosophila have roles in patterning gut mesoderm; however, neither of the mammalian genes has a discernible role in the gut musculature. In contrast, both mammalian genes have roles in developmental processes that have appeared recently in evolution. The Bapx1 gene found in fish, amphibians, birds and mammals appears to have derived vertebrate specific functions sometime after the split between the jawless fish and gnathostomes.

The pectoral girdle articulates the forelimb with the axial skeleton in all vertebrates with paired anterior appendages. The structure of the pectoral girdle and its position along the axial skeleton has changed significantly during vertebrate evolution. These morphological changes have been well described, but there is little comparative embryology to indicate how these changes may have occurred. It is equally obscure how the muscles that connect the head with the pectoral girdle have maintained appropriate attachments even though these 2 structures have become separated. Here I review the changes in the pectoral girdle across different vertebrate taxa, indicating, where known, the developmental mechanisms underlying these changes. I also suggest how the muscular connections between the head and pectoral girdle have been maintained between these once adjacent bones, displaced during vertebrate evolution.

Emerging developmental studies contribute to our understanding of vertebrate evolution because changes in the developmental process and the genes responsible for such changes provide a unique way for evaluating the evolution of morphology. Endoskeletal limbs, the locomotor organs that are unique to vertebrates, are a popular model system in the fields of palaeontology and phylogeny because their structure is highly visible and their bony pattern is easily preserved in the fossil records. Similarly, limb development has long served as an excellent model system for studying vertebrate pattern formation. In this review, the evolution of vertebrate limb development is examined in the light of the latest knowledge, viewpoints and hypotheses.