Changing views of brain evolution.

Although brain studies began in ancient Egypt, speculations on vertebrate brain evolution occurred only much later, after the publication of Darwin's Origin of Species in 1859. Subsequently, views of brain evolution have been shaped by a complex interplay of theory and technique. Darwin's theory allowed the variation in brain size and complexity to be re-interpreted within an evolutionary context, albeit an erroneous pre-Darwinian context based on scala naturae. With the development of histological techniques, research shifted to descriptions of cellular structure, cellular aggregates and their putative interconnections. In spite of these technical advances, brain evolution continued to be viewed within the context of scala naturae. Following the publication of The Comparative Anatomy of the Nervous System of Vertebrates by Ariëns Kappers, Huber, and Crosby in 1936, there followed a period of stasis, after which biological views of evolution were radically altered by the confluence of genetics, paleontology, and systematics, termed the Evolutionary Synthesis. Against this background, the development of new experimental techniques for establishing neural connections resulted in a new flowering of comparative neuroanatomy. While comparative descriptive and experimental studies of brain organization continue, the rapprochement of embryology and genetics is fueling a new renaissance that promises to increase our understanding of brain evolution and its genetic basis.

Lateral line placodes are induced during neurulation in the axolotl.

In order to determine the time window for induction of lateral line placodes in the axolotl, we performed two series of heterotopic and isochronic transplantations from pigmented to albino embryos at different stages of embryogenesis and assessed the distribution of pigmented neuromasts in the hosts at later stages. First, ectoderm from the prospective placodal region was transplanted to the belly between early neurula and mid tailbud stages (stages 13-27). Whereas grafts from early neurulae typically differentiated only into epidermis, grafts from late neural fold stages on reliably resulted in differentiation of ectopic pigmented neuromasts. Second, belly ectoderm was transplanted to the prospective placodal region between early neurula and tailbud stages (stages 13-35). Normal lateral lines containing pigmented neuromasts formed in most embryos when grafts were performed prior to early tailbud stages (stage 24) but not when they were performed later. Our findings indicate that lateral line placodes, from which neuromasts originate, are already determined at late neural fold stages (first series of grafts) but are inducible until early tailbud stages (second series of grafts). A further series of heterochronic transplantations demonstrated that the decline of inducibility at mid tailbud stages is mainly due to the loss of ectodermal competence.

Distribution and innervation of taste buds in the axolotl.

Adult axolotls have approximately 1,400 taste buds in the epithelium of the pharyngeal roof and floor and the medial surfaces of the visceral bars. These receptors are most dense on the lingual surfaces of the upper and lower jaws, slightly less dense throughout lateral portions of the pharyngeal roof and floor, and more sparse within medial portions of the pharyngeal roof and floor, except for a median oval patch of receptors located rostrally between the vomerine tooth fields. Each taste bud is a pear-shaped organ, situated at the center of a raised hillock and averaging 80 and 87 microm in height and width, respectively. Each comprises 50 to 80 cells, which can be classified as basal, dark fusiform, or light fusiform, based on differences in their morphology. The distal ends of the apical processes of the fusiform cells reach the surface of each hillock, forming a single taste pore with an average diameter of 15 microm. Each apical process terminates in one of three ways: as short, evenly spaced microvilli; as long clustered microvilli; or as large, stereocilia-like microvilli. The pharyngeal epithelium and associated taste buds in axolotls are innervated solely by rami of the facial, glossopharyngeal and vagal nerves. Approximately, the rostral one half of the pharyngeal roof is innervated by the palatine rami of the facial nerve, whereas the caudal one half of the pharyngeal roof is innervated by the pharyngeal rami of the glossopharyngeal and vagal nerves. The lingual surface of the lower jaw is innervated by the pretrematic (mandibular) ramus of the facial nerve. The dorsal two-thirds of the visceral arches, and the ventral one-third of the visceral arches and the pharyngeal floor, are innervated by both the pretrematic and post-trematic rami of the glossopharyngeal and vagal nerves, respectively.

Distribution and innervation of lateral line organs in the channel catfish.

The lateral line system of the channel catfish is formed by mechanoreceptive neuromasts located within five pairs of cephalic and one pair of trunk canals, as well as superficial lines of neuromasts, termed accessory and/or pit lines. Five pairs of pit lines occur on the head, and three pairs of superficial lines occur on the trunk. In addition to these mechanoreceptors, which are found in most teleost fishes, catfish also possess a total of over 4000 electroreceptive ampullary organs scattered over the entire body. The lateral line receptors are innervated by five pairs of lateral line nerves whose rami are secondarily associated with facial and trigeminal fibers that innervate taste buds and the dermis of the skin, respectively. The neuromasts of the trunk canal and the ramules of the posterior lateral line nerve that innervate them seem to be organized in a segmental pattern. The same is true for the intervertebral ramules of the recurrent facial ramus, which innervate the external taste buds on the trunk. The fibers of the gustatory and lateral line systems may use the neural crest, the developing spinal nerves, or both, to establish this segmental pattern. In this context, it may not be surprising that there is an intimate relationship among each of the sensory systems in the trunk.

Succinic dehydrogenase histochemistry reveals the location of the putative primary visual and auditory areas within the dorsal ventricular ridge of Sphenodon punctatus.

In turtles, crocodilians, lizards and snakes, the dorsal ventricular ridge (DVR) is a nuclear cell mass that contains distinct visual and auditory thalamorecipient cell groups. In the tuatara (Sphenodon punctatus), the DVR is not organized into diverse cell groups but instead possesses a trilaminar cytoarchitecture resembling that characteristic of the telencephalic cortex in reptiles. To determine if visual and auditory fields might also be present in the DVR of Sphenodon punctatus, we used succinic dehydrogenase (SDH) histochemistry, which has been shown to delineate the visual and auditory fields of the DVR in turtles, crocodilians and lizards. We also used acetylcholinesterase (AChE) histochemistry to determine the boundary between the DVR and the basal ganglia in Sphenodon. We found an SDH-rich region in the neuropil ventral to the cell plate of the rostrolateral DVR and a slightly less intense SDH-rich zone in the neuropil deep to the cell plate of the ventromedial DVR. These SDH-rich zones appear to be located at the apical dendrites of the neurons of the adjacent cell plate. These SDH-rich zones were clearly located within the DVR and were distinct from the AChE-rich striatal part of the basal ganglia, which occupied the ventrolateral wall of the telencephalon. Based on findings in other reptiles, it seems likely that the SDH-rich zone in rostrolateral DVR represents the zone of termination of nucleus rotundus visual input to the DVR, whereas the zone in ventromedial DVR represents the zone of termination of nucleus reuniens auditory input. Because a trilaminar DVR such as that in Sphenodon might be the primitive DVR condition for reptiles, our results suggest that the cytoarchitecture of the DVR and the synaptic organization of its thalamic sensory input in the common ancestor of living reptiles might have been much like of the dorsal cortex.

Development of neurogenic placodes in Xenopus laevis.

The development of neurogenic placodes in Xenopus laevis from the time of neural fold closure to larval stages is described. Placodes were reconstructed from camera lucida drawings of serial sections, and the spatiotemporal pattern of placodal neurogenesis was analyzed using in situ hybridization for the genes X-NGNR-1, XNeuroD, X-MyT1, and X-Delta-1, all of which have been implicated in the regulation of neurogenesis. Olfactory, profundal, and trigeminal placodes, a series of dorsolateral placodes (otic placode and five lateral line placodes), a series of epibranchial placodes, and two hypobranchial placodes were identified. Earlier claims that all placodes in anurans develop from a common primordium could not be confirmed. Profundal and trigeminal placodes, however, are partially fused, and all lateral line placodes arise from a common precursor. Epibranchial and hypobranchial placodes develop ventral to other placodes and dorsal and ventral to the pharyngeal pouches, respectively. Hypobranchial placodes give rise to neurons that become intimately associated with the developing heart. All neurogenic placodes strongly express the neuronal differentiation gene XNeuroD. The neuronal determination gene X-NGNR-1, however, is expressed strongly in only some placodes and not in dorsolateral placodes, indicating that neurogenesis in the latter relies on other determination genes. X-Delta-1 is expressed not only in the neurogenic parts of the placodes but also in the primordia of the lateral lines. This suggests that Delta-Notch-mediated lateral inhibition may be involved not only in placodal neurogenesis, but also in the patterning of lateral line neuromasts.

The development of a biological novelty: a different way to make appendages as revealed in the snout of the star-nosed mole Condylura cristata.

The nose of the star-nosed mole Condylura cristata is a complex biological novelty consisting of 22 epidermal appendages. How did this new set of facial appendages arise? Recent studies find remarkable conservation of the genes expressed during appendage formation across phyla, suggesting that the basic mechanisms for appendage development are ancient. In the nose of these moles, however, we find a unique pattern of appendage morphogenesis, showing that evolution is capable of constructing appendages in different ways. During development, the nasal appendages of the mole begin as a series of waves in the epidermis. A second deep layer of epidermis then grows under these superficial epidermal waves to produce 22 separate, elongated epidermal cylinders embedded in the side of the mole's face. The caudal end of each cylinder later erupts from the face and rotates forward to project rostrally, remaining attached only at the tip of the snout. As a result of this unique 'unfolding' formation, the rostral end of each adult appendage is derived from caudal embryonic facial tissue, while the caudal end of each appendage is derived from rostral facial tissue. This developmental process has essentially no outgrowth phase and results in the reversal of the original embryonic orientation of each appendage. This differs from the development of other known appendages, which originate either as outgrowths of the body wall or from subdivisions of outgrowths (e.g. tetrapod digits). Adults of a different mole species (Scapanus townsendii) exhibit a star-like pattern that resembles an embryonic stage of the star-nosed mole, suggesting that the development of the star recapitulates stages of its evolution.

Loss of ectodermal competence for lateral line placode formation in the direct developing frog Eleutherodactylus coqui.

In the direct-developing frog Eleutherodactylus coqui neuromasts and ganglia of the lateral line system never develop. We show here that this absence of the lateral line system, which is evolutionarily derived in anurans, is due to very early changes in development. Ectodermal thickenings, which are typical of lateral line placodes, and from which neuromasts and ganglion cells of the lateral line originate, never form in E. coqui, although other neurogenic placodes are present. Moreover, although NeuroD is expressed in the lateral line placodes of Xenopus laevis, corresponding expression sites are lacking in E. coqui. Heterospecific transplantation experiments show that axolotl ectoderm can be induced to form lateral line placodes after transplantation to E. coqui hosts but that E. coqui ectoderm does not form lateral line placodes on axolotl hosts. This suggests that the loss of the lateral line system in E. coqui is due to the specific loss of ectodermal competence to form lateral line placodes in response to inductive signals. Our results (1) indicate that the competence for lateral line placode formation is distinct and dissociable from the competence to form other neurogenic placodes and (2) support the idea that the lateral line system acts as a module in development and evolution.

Field homology: a meaningless concept.

Most biological homologues involve comparison of single characters in two or more taxa. It is possible, however, to recognize homologous characters between two or more taxa that involve the transformation of one character into many characters or many characters into one character. This type of homology is recognized as field homology and it has been widely used in comparative neuroanatomy. The emergence of the cladistic analysis of embryonic stages in the development of neural characters, however, strongly suggests that field homology is a meaningless concept. When it appears necessary to recognize field homologues, it is because comparisons are being made at an inappropriate level within a given biological hierarchy. Furthermore, recognition of field homologues restricts evolutionary mechanisms to a single mechanism of parcellation as defined by Ebbesson.

The chemoarchitecture of the forebrain of lampreys: evolutionary implications by comparisons with gnathostomes.

The distribution of substance P-, leucine-enkephalin-, tyrosine hydroxylase-, and serotonin-like immunoreactivities was studied in the forebrain of the silver lamprey and compared to other vertebrate groups. In silver lampreys, substance P and leucine-enkephalin provide a clear distinction between the pallium and subpallium as they do for gnathostomes. The pallium consists of three subdivisions which show striking similarities to the medial, dorsal and lateral pallia of gnathostomes. Boundaries of the preoptic area, the hypothalamus and posterior tubercle can be well defined by histochemical characteristics. These data refute the existence of a dorsal hypothalamic subdivision in silver lampreys and suggest rostral and caudal divisions of the hypothalamus.

Brainstem neurons with descending projections to the spinal cord of two elasmobranch fishes: thornback guitarfish, Platyrhinoidis triseriata, and horn shark, Heterodontus francisci.

We studied two cartilaginous fishes and described their brainstem supraspinal projections because most nuclei in the reticular formation can be identified that way. A retrogradely transported tracer, horseradish peroxidase or Fluoro-Gold, was injected into the spinal cord of Platyrhinoidis triseriata (thornback guitarfish) or Heterodontus fransisci (horn shark). We described labeled reticular cells by their position, morpohology, somatic orientation, dendritic processes, and laterality of spinal projections. Nineteen reticular nuclei have spinal projections: reticularis (r.) dorsalis, r. ventralis pars alpha and beta, r. gigantocellularis, r. magnocellularis, r. parvocellularis, r. paragigantocellularis lateralis and dorsalis, r. pontis caudalis pars alpha and beta, r. pontis oralis pars medialis and lateralis, r. subcuneiformis, r. peduncularis pars compacta, r. subcoeruleus pars alpha, raphe obscurus, raphe pallidus, raphe magnus, and locus coeruleus. Twenty nonreticular nuclei have spinal projections: descending trigeminal, retroambiguus, solitarius, posterior octaval, descending octaval, magnocellular octaval, ruber, Edinger-Westphal, nucleus of the medial longitudinal fasciculus, interstitial nucleus of Cajal, latral mesencephalic complex, periventricularis pretectalis pars dorsalis, central pretectal, ventromedial thalamic, posterior central thalamic, posterior dorsal thalamic, the posterior tuberculum, and nuclei B, F, and J. The large number of distinct reticular nuclei with spinal projections corroborates the hypothesis that the reticular formation of elasmobranches is complexly organized into many of the same nuclei that are found in frogs, reptiles, birds, and mammals.

Coexistence of FMRFAMIDE-like and LHRH-like immunoreactivity in the terminal nerve and forebrain of the big brown bat, Eptesicus fuscus.

The coexistence of molluscan cardioexcitatory neuropeptide (FMRFAMIDE) and luteinizing hormone-releasing hormone (LHRH) was studied in the nervous system of the big brown bat, Eptesicus fuscus, with immunocytochemistry. Within mammals, this is the first report of the coexistence of these neuropeptides in the terminal nerve. In juvenile and adult bats, both neuropeptides are distributed identically throughout the terminal nerve (tn), and they coexist in many parts of the prosencephalon from the olfactory bulb as far caudally as the interpeduncular nucleus. Peripherally, on the basal surface of the forebrain, fibers and a few perikarya, which may belong to the tn, form a loose plexus. Within the brain wall, regions of maximal immunoreactivity (ir) are the habenula, medial preoptic area, arcuate nucleus, and the infundibulum. Whereas in most areas of the prosencephalon (e.g., stria terminalis and bed nuclei, amygdaloid complex) fibers show stronger immunoreactivity to FMRFAMIDE, labeling of fibers in the habenula and infundibulum is largely identical for both neuropeptides. The arcuate nucleus contains a large number of perikarya and is the major source of both FMRFAMIDE- and LHRH-ir within the forebrain. A number of fibers run along the ependyma of the ventricular system and seem to terminate here; this is particularly evident in the median eminence and infundibular stalk. In the big brown bat, there seems to exist a continuum of FMRFAMIDE- and LHRH-ir throughout the tn and those structures of the forebrain that are known to be engaged in the control of mating behavior, reproduction, and rhythmicity. Concerning the hypothalamo-hypophyseal-gonadal axis, the arcuate nucleus may serve as a central hub between the olfactory/terminal input and superior areas including the limbic system. In contrast to LHRH immunoreactivity, FMRFAMIDE-like ir extends throughout the brainstem and cervical spinal cord. This system may also be involved in the processing and modulation of autonomic input via the parabrachial and solitary nuclei, the rhombencephalic central gray, and its projection into the hypothalamus (paraventricular nucleus), thus facilitating feed-back of gonadotropic influences of the terminal nerve and prosencephalon.

Telencephalic connections in the Pacific hagfish (Eptatretus stouti), with special reference to the thalamopallial system.

The pallium of hagfishes (myxinoids) is unique: It consists of a superficial "cortical" mantle of gray matter which is subdivided into several layers and fields, but it is not clear whether or how these subdivisions can be compared to those of other craniates, i.e., lampreys and gnathostomes. The pallium of hagfishes receives extensive secondary olfactory projections (Wicht and Northcutt [1993] J. Comp. Neurol. 337:529-542), but there are no experimental data on its nonolfactory connections. We therefore investigated the pallial and dorsal thalamic connections of the Pacific hagfish. Injections of tracers into the pallium labeled many cells bilaterally in the olfactory bulbs. Other pallial afferents arise from the contralateral pallium, the dorsal thalamic nuclei, the preoptic region, and the posterior tubercular nuclei. Descending pallial efferents reach the preoptic region, the dorsal thalamus, and the mesencephalic tectum but not the motor or premotor centers of the brainstem. Injections of tracers into the dorsal thalamus confirmed the presence of reciprocal thalamopallial connections. In addition, these injections revealed that there is no "preferred" pallial target for the ascending thalamic fibers; instead, ascending thalamic and secondary olfactory projections overlap throughout the pallium. The mesencephalic tectum and tegmentum, which receive afferents from a variety of sensory sources, are interconnected with the dorsal thalamus; thus, ascending nonolfactory sensory information may reach myxinoid pallia via a tectal-thalamic-telencephalic route. A comparative analysis of pallial organization reveals that the subdivisions of the pallium in gnathostomes (i.e., medial, dorsal, and lateral pallia) cannot be recognized with certainty in hagfishes.

The diencephalon and pretectum of the white sturgeon (Acipenser transmontanus): a cytoarchitectonic study.

The cytoarchitecture of the diencephalon and pretectum of the white sturgeon, Acipenser transmontanus, was studied utilizing cresyl violet stained serial paraffin sections. The identified cell groups were assigned to the preoptic area, hypothalamus, thalamus and posterior tubercle, epithalamus, synencephalon and pretectum. The outlines of the diencephalic and pretectal nuclei were projected graphically onto a midsagittal section of the brain, thus providing a reconstruction of the relative positions of the major cell groups. This facilitated comparisons with the diencephalic and pretectal nuclei of other ray-finned fishes. Our cytoarchitectural analysis indicates that the diencephalon and pretectum of the white sturgeon is intermediate to that described for cladistians and neopterygians. The preoptic area in Acipenser is relatively conservative compared to other ray-finned fishes but lacks distinct subdivisions in the magnocellular periventricular preoptic nucleus and includes a unique migrated rostral accessory nucleus. The hypothalamic walls are rather thin, and due to the presence of extensive lateral and posterior recesses and the lack of migrated nuclei, they superficially resemble the condition seen in sharks. The dorsal and ventral thalamic nuclei do not exhibit much variation compared to other ray-finned fishes, except for the presence of a small lateral posterior thalamic nucleus, the absence of a distinct ventrolateral thalamic nucleus, and slight differences in the internal organization of the ventromedial thalamic nucleus. The posterior tubercle in Acipenser clearly comprises more migrated cell groups than that of Polypterus, however, these cell groups are considerably less well defined than in neopterygians. As in other nonteleost actinopterygians, the habenular nuclei are highly asymmetrical with the right side larger than the left side. The cytoarchitectonic complexity of the pretectum in Acipenser is intermediate to that observed in Polypterus and neopterygians in that a magnocellular component within the superficial pretectal nucleus is clearly present but cannot be delineated as a distinct magnocellular superficial pretectal nucleus. Also, the posterior pretectal nucleus which is absent in Polypterus but which has been identified as a small nucleus both in Lepisosteus and Amia is represented in Acipenser by a small group of scattered cells.

Amphibians provide new insights into taste-bud development.

Until recently, the predominant model of taste-bud development was one of neural induction: ingrowing sensory fibers were thought to induce taste-bud differentiation late in embryonic development. Recent experimental studies, however, show that the development of taste buds is independent of their innervation. In amphibian embryos, the ability to generate taste buds is an intrinsic feature of the oropharyngeal epithelium long before the region becomes innervated. These studies indicate that patterning of the oropharyngeal epithelium occurs during gastrulation, and suggest that taste buds or their progenitors play the dominant role in the development of their own innervation.

The role of innervation in the development of taste buds: insights from studies of amphibian embryos.

Amphibian embryos have long been model organisms for studies of development because of their hardiness and large size, as well as the ease with which they can be experimentally manipulated. These particular advantages have allowed us recently to test the role of innervation in the development of vertebrate taste buds using embryos of an aquatic salamander, the axolotl. The predominant model of taste bud genesis has been one of neural induction, in which ingrowing sensory neurites induce taste bud differentiation in the epithelium that lines the mouth and pharynx. However, when we prevented embryonic sensory neurons from contacting the oropharyngeal epithelium by using transplantation or tissue culture techniques, we found that taste bud differentiation was independent of nerve contact. Additionally, using similar types of experimental manipulations, we have recently shown that taste bud differentiation is not a result of interactions of the oropharyngeal epithelium with craniofacial mesenchyme. Surprisingly, we found that although taste bud genesis occurs very late in embryonic development, it is an intrinsic feature of the presumptive oropharyngeal epithelium extremely early, in fact as early as the completion of gastrulation. These data have prompted us to propose a new model for the development of amphibian taste buds: (i) The presumptive oropharyngeal epithelium is specified by the time gastrulation is complete; (ii) Subsequently, a distributed population of taste bud progenitors is set up within this epithelium via local cell-cell interactions. These progenitor cells give rise to taste buds, which are distributed throughout the mouth and pharynx. How widely applicable this model might be for the genesis of taste buds in other vertebrates remains to be seen. However, since it is likely that the taste system of axolotls more closely resembles the ancestral state from which both the amphibian and mammalian taste systems have evolved, it is possible that many of the same developmental mechanisms that give rise to amphibian taste buds are also used to generate the receptor organs in mammals.

Evolution of gnathostome lateral line ontogenies.

An outgroup analysis of multiple ontogenies provides the most robust approach to understanding phylogeny. Such an analysis of the lateral line system among extinct and extant gnathostomes reveals that lateral line placodes constitute the basic ontogenetic unit responsible for the development of this system. Six pairs of lateral line placodes appear to have existed in the earliest gnathostomes, and eight stages (stages A-H) can be recognized in their differentiation. Terminal truncation (heterochronic changes) in the primitive sequence of placodal development has occurred in one or more placodes in each gnathostome radiation, with the most extensive truncations occurring in arthrodire placoderms, lepidosirenid lungfishes and extant amphibians. The most extensive nonterminal changes in the primitive sequence of placodal development involve the failure of electroreceptors to form within the lateral zones of the elongatiang sensory ridges of the placodes. This nonterminal change appears to have occurred independently in ancestral neopterygian bony fishes, in many amphibians and, possibly, in the extinct acanthodians. At least two teleost radiations, osteoglossomorphs and ostariophysines, have re-evolved electroreceptors which may represent additional nonterminal changes in placodal patterning or, possibly, a change in the embryonic source of these receptors.

Taste buds develop autonomously from endoderm without induction by cephalic neural crest or paraxial mesoderm.

Although it had long been believed that embryonic taste buds in vertebrates were induced to differentiate by ingrowing nerve fibers, we and others have recently shown that embryonic taste buds can develop normally in the complete absence of innervation. This leads to the question of which tissues, if any, induce the formation of taste buds in oropharyngeal endoderm. We proposed that taste buds, like many specialized epithelial cells, might arise via an inductive interaction between the endodermal epithelial cells that line the oropharynx and the adjacent mesenchyme that is derived from both cephalic neural crest and paraxial mesoderm. Using complementary grafting and explant culture techniques, however, we have now found that well-differentiated taste buds will develop in tissue completely devoid of neural crest and paraxial mesoderm derivatives. When the presumptive oropharyngeal region was removed from salamander embryos prior to the onset of cephalic neural crest migration, taste buds developed in grafts and explants coincident with their appearance in intact control embryos. Similarly, explants from neurulae in which movement of paraxial mesoderm had not yet begun also developed taste buds after 9-12 days in vitro. We conclude that neither cranial neural crest nor paraxial mesoderm is responsible for the induction of embryonic taste buds. Surprisingly, the ability to develop taste buds late in embryonic development seems to be an intrinsic feature of the oropharyngeal endoderm that is determined by the completion of gastrulation.

Afferent and efferent connections of the lateral and medial pallia of the silver lamprey.

Injections of the carbocyanine dye, DiI, into the lateral pallium of the silver lamprey reveal that this pallial region receives bilateral inputs from the olfactory bulbs, dorsomedial telencephalic neuropil, and the habenular nuclei, and ipsilateral inputs from the septum, preoptic area, medial pallium, thalamus, and, possibly, the striatum. The efferent projections of the lateral pallium form dorsal (olfacto-habenular tract of Heier) and ventral (olfacto-thalamic and hypothalamic tracts of Heier) bundles. The dorsal bundle terminates ipsilaterally in the dorsal pallium, medial pallium, habenular nuclei, and pretectum and contralaterally in the habenular nuclei and, possibly, the dorsal pallium. The ventral bundle terminates ipsilaterally in the septum, striatum, and preoptic areas and bilaterally within the hypothalamus. Injections of DiI into the medial pallium reveal bilateral inputs to this pallial formation from the olfactory bulbs, the dorsomedial telencephalic neuropil, septum, habenular nuclei, thalamic nuclei, preoptic area and hypothalamus, as well as ipsilateral inputs from the lateral pallium, dorsal isthmal grey and midbrain tegmentum. The efferent projections of the medial pallium form dorsal, ventral and descending bundles. The dorsal bundle terminates ipsilaterally in the dorsal and lateral pallia and in the olfactory bulb. The ventral bundle terminates ipsilaterally in the dorsal pallium and bilaterally within the lateral pallium and in preoptic and hypothalamic areas. The descending bundle terminates bilaterally in thalamic and hypothalamic areas and in the pretectum and optic tectum. These data support a number of earlier hypotheses concerning pallial homologues in lampreys and other vertebrates but suggest that the earlier hypothesis of an olfactory origin of the telencephalon of craniates should be rejected.