Retinal topography in reef teleosts. I. Some species with well-developed areae but poorly-developed streaks

The retinal ganglion cell layer of five species of teleosts has been studied from Nissl-stained whole-mounts and the distribution of neuronal elements determined quantitatively. Isodensity contour maps of neurons in the ganglion cell layer revealed areas of high density (areae centrales) predominantly in the temporal retina, but other areae were also found in the nasal and dorso-nasal retina. Neuronal densities within the ganglion cell layer at the areae centrales ranged from 0.4 x 10(4) to 4.7 x 10(4) cells/mm2. Species that were found to lack a horizontal streak of high ganglion cell density appear to be those whose behaviour suggests they possess an interrupted view of the sand-water horizon and are 'enclosed' species. Concentric density contours around an area centralis seem to be associated with enclosed environments. The relationship between retinal topography and niche is also discussed.     

Quantitative analysis of the retinal ganglion cell layer and optic nerve of the barn owl Tyto alba

The visual capacity of the common barn owl (Tyto alba) was studied by quantitative analysis of the retina and optic nerve. Cell counts in the ganglion cell layer of the whole-mounted retina revealed a temporal area centralis with peak cell density of 12,500 cells/mm2 and a horizontal streak of high cell density extending from the area centralis into the nasal retina. Integration of the ganglion cell density map gave an estimated total of 1.4 million cells for the ganglion cell layer. Electron microscopy of a single, complete section of the optic nerve revealed a bimodal fiber diameter spectrum (modes at 0.3 and 0.9 microns; bin width = 0.2 microns), with diameters ranging from 0.15 microns (unmyelinated) to 6.05 microns (myelinated, sheath included). The total axon count for the optic nerve was estimated from sample counts to be about 680,000 axons (25% unmyelinated). Therefore, roughly half of the cells in the retinal ganglion cell layer do not send axons into the optic nerve. With certain assumptions, the data predict a visual spatial acuity for barn owls on the order of 8 cycles/degree, a value similar to the known behaviorally measured acuities of masked owls (10 cycles/degree) and domestic cats (6 cycles/degree).     

A quantitative analysis of the cat retinal ganglion cell topography.

A retinal ganglion cell distribution map has been prepared for the cresyl violet stained cat retina. It differs from previously published maps in revealing the visual streak to be more substantial and in showing a higher peak density of 9-10,000 ganglion cells/mm2 at the presumed visual pole. The map was used to obtain a minimum estimate of the retinal ganglion cell population as 217,000 cells, more than double the total previously reported. The problem of classifying the cells of the ganglion cell layer is discussed in detail and examples of criterion cells illustrated. The paper also includes an account of retinal mensuration (dimensions, area, etc.) and a discussion of the visual streak orientation.     

Development of ganglion cell topography in ferret retina

The adult ferret has approximately 90,000 retinal ganglion cells, arranged in a prominent area centralis and visual streak. The role of differential cell generation, cell death, and retinal growth in the control of adult retinal ganglion cell number and distribution was evaluated by examining basic aspects of retinogenesis, including growth in retinal area, developmental changes in the number, size, and distribution of retinal ganglion cells (identification aided by retrograde transport of HRP), and the incidence of degenerating cells in the ganglion cell layer. Retinal development in the ferret was also compared to retinal development in the cat (which has an even more differentiated area centralis) to determine what alterations of developmental parameters are most closely associated with this species difference in adult morphology. The area of the retina increases linearly from birth (12 mm2) to postnatal day 24 (54 mm2), reaching an eventual adult value of 64 mm2. Ganglion cell numbers peak at 155,000 (approximately twice the adult number) on postnatal day 3, and fall to adult numbers by postnatal day 6. The remaining cells of the ganglion cell layer, principally displaced amacrine cells, reach their peak number on postnatal day 10 (approximately 280,000), falling to 200,000 by adulthood. Degenerating cells are abundant in the ganglion cell layer in the immediate postnatal period. A difference in the incidence of degenerating cells in the presumptive area centralis versus that in the retinal periphery was not observed postnatally, though there were other striking spatial nonuniformities, suggesting that differential cell loss might contribute to other features of retinal topographic organization. Ganglion cell density is virtually uniform across the retina at birth. Cell density is first reduced in the dorsal retina, resulting in a dorsal-to-ventral gradient in cell density that persists until day 10, when ganglion cell number has stabilized. By postnatal day 24, an area centralis and visual streak has emerged, but not of adult magnitude. Because ganglion cell number has stabilized long before the area centralis and visual streak emerge, we conclude that differential retinal growth is the principal mechanism producing this feature of retinal topography. Comparison with the cat suggests that the proportionately greater nonuniform growth of the cat's eye accounts for the greater differentiation of its area centralis.     

Retinal topography in the koala (Phascolarctos cinereus).

Nissl-stained retinal wholemounts were used to investigate the topographical organization of the ganglion cell layer of the koala (Phascolarctos cinereus); the visual resolution limit of this animal was subsequently estimated from retinal ganglion cell density data. Two types of cells could be differentiated on the basis of their size and staining characteristics: a subpopulation of presumed ganglion cells, consisting of medium to large cells with Nissl substance in the cytoplasm and pale uniformly staining nuclei, and a further subpopulation of small, densely staining cells. The latter group were presumed to be neuroglia and displaced amacrine cells. Iso-density contour maps were prepared from total cell counts and also counts of presumed ganglion cells; in all cases, the density of cells was greatest in the inferior retina where there was an area of peak density occurring as a poorly developed, horizontal streak that extended across the inferior retina. The inferior position of the streak in the koala contrasts with reports of the superior position of streaks in other marsupials. Peak cell densities of 2370 cells/mm2 and 1480 cells/mm2 were recorded for the total cell population and the presumed ganglion cell subpopulation, respectively. The latter value is equivalent to a visual resolution of 2.4 cycles/degree, based on sampling theory and a square packing paradigm, placing the koala close in visual performance to two other marsupials, the Australian Northern native cat and the American Virginia opossum.     

Ganglion cells density and retinal resolution in the sea otter, Enhydra lutris

The topographic distribution, density, and size of ganglion cells were studied in retinal wholemounts of the sea otter, Enhydra lutris. The cell distribution showed a well defined horizontal streak of higher cell density, and within this streak, a narrow area of the highest cell density. The peak cell density in this area ranged from 4050 to 4400 cells/mm(2), with a mean of 4225 cells/mm(2). The ganglion cell size ranged from 7 microm to 47 microm but the majority of cells were 7 to 30 microm. Cell size distribution revealed three size groups: 7-16, 17-28, and 29-47 microm. The highest-density area contained mainly small (7-16 microm) cells. The cell-density data predict a retinal resolution around 7' in water. Retinal organization in the sea otter exhibits more properties common with terrestrial rather than aquatic mammals, both in terms of ganglion cell characteristics and in terms of their topographic distribution.     

Topography of ganglion cells in the dog and wolf retina.

The topographical distribution of retinal ganglion cells in seven breeds of dog (Canis lupus f. familiaris) and in the wolf (Canis lupus) was studied in retinal wholemounts stained with cresyl violet or with a reduced silver method. A prominent feature of all wolf retinae was a pronounced "visual streak" of high ganglion cell density, extending from the central area far into both temporal and nasal retina. By contrast, either a pronounced or a moderate visual streak was found in dog retinae. It is hypothesized that a pronounced streak is an archetypal feature of Canis lupus, and that the moderate streak in some dogs is a corollary of breeding during domestication. Irrespective of the differences in streak form and retinal area, the estimated total number of ganglion cells was about 200,000 cells in the wolf and 115,000 in the dog. Ganglion cell density maxima in the central area of the wolf were about 12,000-14,000/mm2, and in the dog they ranged from 6,400/mm2 to 14,400/mm2. This implies individual differences in visual acuity. Alpha ganglion cells constituted 3-14% of all ganglion cells in the dog and 1-18% in the wolf, depending on retinal location. A distinct feature of all dogs and wolves was the absence of alpha cells in a substantial region of temporal peripheral retina. This has not been found in any other mammalian species and suggests corresponding functional deficits.     

Retinal ganglion cell topography in elasmobranchs

Retinal wholemounts are used to examine the topographic distribution of retinal cells within the ganglion cell layer in a range of elasmobranchs from different depths. The retina is examined for regional specializations for acute vision in six species of selachians, Galeocerdo cuvieri, Hemiscyllium ocellatum, Scyliorhinus canicula, Galeus melastomus, Etmopterus spinax, Isistius brasiliensis, one species of batoid, Raja bigelowi and one species of chimaera, Hydrolagus mirabilis. These species represent a range of lifestyles including pelagic, mesopelagic and benthic habitats, living from shallow water to the sea bottom at a depth of more than 3000 m. The topography of cells within the ganglion cell layer is non-uniform and changes markedly across the retina. Most species possess an increased density of cells across the horizontal (dorsal) meridian or visual streak, with a density range of 500 to 2,500 cells per mm(2) with one or more regional increases in density lying within this specialized horizontal area. It is proposed that the higher spatial resolving power provided by the horizontal streak in these species mediates panoramic vision in the lower frontal visual field. Only I. brasiliensis possesses a concentric arrangement of retinal iso-density contours in temporal retina or an area centralis, thereby increasing spatial resolving power in a more specialized part of the visual field, an adaptation for its unusual feeding behavior. In Nissl-stained material, amacrine and ganglion cell populations could be distinguished on the criteria of soma size, soma shape and nuclear staining. Quantitative analyses show that the proportion of amacrine cells lying within the ganglion cell layer is non-uniform and ranges between 0.4 and 12.3% in specialized retinal areas and between 8.2 and 48.1% in the peripheral non-specialized regions. Analyses of soma area of the total population of cells in the ganglion cell layer also show that the pelagic species possess significantly smaller soma (9-186 micrometer(2)) than benthic and/or deep-sea species (16-338 micrometer(2)), and that a number of different morphological classes of cells are present including a small population of giant ganglion cells.     

The number and distribution of ganglion cells in the retina of the brush-tailed possum, Trichosurus vulpecula

The distribution of ganglion cells in the retina of the adult brush-tailed possum was determined by light microscopy of Nissl stained retinal whole mounts. Qualitatively, the distribution in this marsupial retina shows features, such as an area centralis and a visual streak, which are found separately or together in eutherian mammals. The possum retina is avascular and the eye has a weak tapetum in the superior fundus.The retinal area is 260 mm2 and there are about 280,000 ganglion cells. The diameters of the ganglion cell somas range from 5 micrometer to 26 micrometer and the frequency distribution of soma size classes is skewed and unimodal (mean" 12.8 micrometer) with 62% of the cells falling in the class of diameters 7-13 micrometer. Maps of ganglion cell density were made for five retinae. These maps show that there is a band of high ganglion cell density (greater 2,000 cells mm-2) which extends across the retina about 0.6 mm above the optic disc in the tapetal region of the fundus and which lies in the plane of the animal's horizon when the eyes are in their primary position. By analogy with other species, this band is termed the visual streak. Near the temporal end of the visual streak, 2.9 mm from the optic disc, the ganglion cell density reaches a localized maximum of approximately 5,000 cells mm-2 thereby defining the centre of an area centralis (greater than 3,000 cells mm-2). The posterior nodal distance of the possum eye was estimated at 7.8 mm, which corresponds to a retinal magnification of 136 micrometer per degree of visual field. There are up to 30,000 glial cells which lie in, or slightly vitread to, the layer of the retinal ganglion cells.     

The retinal ganglion cell layer and visual acuity of the camel.

We examined the retinal ganglion cell layer of the dromedary camel, Camelus dromedarius. We have estimated that there are 8 million neurons in the ganglion cell layer of this large retina (mean area of 2,300 mm(-2)). However, only approximately 1 million are considered to be ganglion cells. The ganglion cells are arranged as two areas of high cell density, one in the temporal and one in the nasal retina. Densities of ganglion cells between these two high density regions is much lower, often less than 100 per mm(-2). In between these two high density regions, on the nasal side of the optic nerve head, is a unique and dense vertical streak of mostly non-ganglion cells; the function of this specialization is unknown. On the basis of ganglion cell density we estimate that the peak acuity in the dromedary camel is about 10 and 9.5 cycles per degree in the temporal and nasal high density regions respectively and falls to 2-3 cycles per degree in the central retina. Behavioral acuity was estimated for one bactrian camel and was found to be approximately 10 cyc deg(-1). The camel has a retina with a mean thickness of 104 microm, less than the 143 microm thickness that has previously been thought to be necessary for a retinal vasculature. Nevertheless, there is an extensive vitreal vasculature that does not appear to spare any retinal region.     

Peak density, size and regional distribution of ganglion cells in the retina of the fur seal Callorhinus ursinus.

The total number, size, topographic distribution and peak density of ganglion cells were studied in retinal wholemounts of the fur seal, Callorhinus ursinus. The cell distribution showed a distinct zone of high ganglion cell density. It was located in the temporal retinal quadrant, near the horizontal meridian, 10-12 mm (25-31 degrees) from the optic disk. The peak cell density in this zone was 812-1332 cells/mm2 (mean 1053 cells/mm2), i.e. 125-205 cells/deg2 (mean 162 cells/deg2). These data predict a retinal resolution of 5.6-7.1 cycle/deg. The ganglion cell soma size ranged from 10 to 50 microns. Cell size histograms were bimodal in shape with modes below and above 30 microns.     

The retina of tyrant flycatchers: topographic organization of neuronal density and size in the ganglion cell layer of the great kiskadee Pitangus sulphuratus and the rusty margined flycatcher Myiozetetes cayanensis (Aves: Tyrannidae)

Tyrant flycatchers comprise the largest group of passerine birds of the Neotropical region but their retinal organization is unknown. The great kiskadee, Pitangus sulphuratus, is categorized as a supreme generalist and utilizes a variety of foraging strategies. The rusty margined flycatcher, Myiozetetes cayanensis, is partially frugivorous and captures insects in the air. Using retinal wholemounts, we described the topographic distribution of density and size of neurons lying in the retinal ganglion cell layer in those two species of tyrant flycatchers. Maps of neuron distribution showing isodensity contours revealed the presence of a pronounced central fovea and a temporal area in both species. Both retinal specializations were circumscribed by an inconspicuous horizontal visual streak. The highest foveal densities ranged from 48,000 to 55,000 cells/mm(2) for Pitangus sulphuratus and between 62,000 and 65,000 cells/mm(2) for Myiozetetes cayanensis. The peak density in the temporal area was around 40,000 cells/mm(2) for Pitangus sulphuratus and 46,000 cells/mm(2) for Myiozetetes cayanensis. At central, mid-peripheral and peripheral eccentricities, perikaryon size varied quite similarly in both species. A cohort of giant retinal ganglion cells with perikaryon size > 300 microm(2) was observed at the temporal periphery and defines an 'area giganto cellularis' described previously in procellariiform seabirds. This specialization is thought to be involved in movement detection and could aid the tyrant flycatchers to capture moving prey. Functionally, the presence of a fovea associated with a temporal area would allow high spatial resolution for capturing insects by the tyrant flycatchers. Nonetheless, even though both species exhibit different foraging strategies, they shared a similar topographic arrangement of neuronal density in the ganglion cell layer. This suggests that the retinal topography did not accompany changes in the foraging ecology throughout evolutionary history for these species of tyrant flycatchers.     

Retinal ganglion cell layer of the Caspian seal Pusa caspica: topography and localization of the high-resolution area

Retinal topography, cell density and sizes of ganglion cells in the Caspian seal (Pusa caspica) were analyzed in retinal whole mounts stained with cresyl-violet. The topographic distribution of ganglion cells displayed an area of high cell density located in the temporal quadrant of the retina and was similar to the area centralis of terrestrial carnivores. It extended nasally, above the optic disk, as a streak of increased cell density. In different whole mounts, the peak cell density in the high-density area ranged from 1,684 to 1,844 cells/mm2 (mean 1,773 cells/mm2). The cell density data predict a retinal resolution of around 8.5 cycles/degree in water. A distinctive feature of the Caspian seal's retina is the large size of ganglion cells and the low cell density compared to terrestrial mammals. The ganglion cell diameter ranged from 10 to 58 μm. Cell size histograms featured bimodal patterns with groups of small and large ganglion cells. The large cells appeared similar to α-cells of terrestrial mammals and constituted 7% of the total ganglion cell population.     

Quantitative analysis of the retinal ganglion cell layer in the ostrich,Struthio camelus

The total number, distribution and peak density of ganglion cells were evaluated in the Nissl-stained retina of the ostrich (Struthio camelus). The mean (n = 4) total number of retinal ganglion cells (RGC) was estimated at 2,274,128 (s.d. = 273, 152). The ostrich retina exhibited a prominent horizontal visual streak along which a central area located nasal to the pecten had a peak density of 9,500 cells/mm2. A high concentration of cells with a peak density of 2,646 cells/mm2 was also observed in the temporal retina, slightly dorsal to the visual streak. The results further showed that the ostrich eye has a 15-mm pupil entrance diameter, its mean axial length is 39.81 mm, the estimated retinal magnification factor is 0.4075 mm/deg and the maximum visual acuity along the well-defined visual streak was estimated to be 19.32 cycles/deg. The latter component of the retina might subserve vision along the horizon while the temporal region mediates binocular processing. The data also showed that the degree of retinal illumination in this bird could be comparable to that noted in some nocturnal species. The findings in this study suggest that the ostrich might not be restricted to diurnal activity.     

Retinal structure and visual acuity in a polyprotodont marsupial, the fat-tailed dunnart (Sminthopsis crassicaudata)

The visual system of the fat-tailed dunnart (Sminthopsis crassicaudata), a small polyprotodont marsupial, has been examined both anatomically and behaviourally. The ganglion cell layer was examined in cresyl-violet stained wholemounts and found to contain a mean of 81,400 ganglion cells (SD +/- 3,360); the identification of ganglion cells was supported by a correspondence to optic axon counts. Ganglion cells were distributed as a mid-temporally situated area centralis, embedded in a pronounced visual streak. Localised implants of horseradish peroxidase into retinal wholemounts revealed both A-type and B-type horizontal cells. Sections of the outer retina showed it to be rod-dominated, with a rod-to-cone ratio of 40:1 at the area centralis; cones were found to contain oil droplets but double cones were not a prominent feature. The retinal pigment epithelium consisted of squamous cells. Visual acuity, estimated from counts of peak ganglion cell density (8,300/mm2, SD +/- 1,180) and measurements of posterior nodal distance (2.9 mm), was found to be 2.30 cycles per degree. The value was close to that of 2.36 cycles per degree estimated by behavioural tests using a Mitchell jumping stand; values were similar at low, intermediate and high light levels. Our findings are discussed in relation to the lifestyle of the dunnart.     

Retinal ganglion cell topography and spatial resolving power in the white rhinoceros (Ceratotherium simum)

This study sought to determine whether the retinal organization of the white rhinoceros (Ceratotherium simum), a large African herbivore with lips specialized for grazing in open savannahs, relates to its foraging ecology and habitat. Using stereology and retinal wholemounts, we estimated a total of 353,000 retinal ganglion cells. Their density distribution reveals an unusual topographic organization of a temporal (2,000 cells/mm²) and a nasal (1,800 cells/mm²) area embedded within a well‐defined horizontal visual streak (800 cells/mm²), which is remarkably similar to the retinal organization in the black rhinoceros. Alpha ganglion cells comprise 3.5% (12,300) of the total population of ganglion cells and show a similar distribution pattern with maximum densities also occurring in the temporal (44 cells/mm²) and nasal (40 cells/mm²) areas. We found higher proportions of alpha cells in the dorsal and ventral retinas. Given their role in the detection of brisk transient stimuli, these higher proportions may facilitate the detection of approaching objects from the front and behind while grazing with the head at 45 °. Using ganglion cell peak density and eye size (29 mm, axial length), we estimated upper limits of spatial resolving power of 7 cycles/deg (temporal area), 6.6 cycles/deg (nasal area), and 4.4 cycles/deg (horizontal streak). The resolution of the temporal area potentially assists with grazing, while the resolution of the streak may be used for panoramic surveillance of the horizon. The nasal area may assist with detection of approaching objects from behind, potentially representing an adaptation compensating for limited neck and head mobility. J. Comp. Neurol., 525:2484–2498, 2017. © 2017 Wiley Periodicals, Inc.     

Retinal ganglion cell topography and spatial resolving power in African megachiropterans: Influence of roosting microhabitat and foraging

Megachiropteran bats (megabats) show remarkable diversity in microhabitat occupation and trophic specializations, but information on how vision relates to their behavioral ecology is scarce. Using stereology and retinal wholemounts, we measured the topographic distribution of retinal ganglion cells and determined the spatial resolution of eight African megachiropterans with distinct roosting and feeding ecologies. We found that species roosting in open microhabitats have a pronounced streak of high retinal ganglion cell density, whereas those favoring more enclosed microhabitats have a less pronounced streak (or its absence in Hypsignathus monstrosus). An exception is the cave‐dwelling Rousettus aegyptiacus, which has a pronounced horizontal streak that potentially correlates with its occurrence in more open environments during foraging. In all species, we found a temporal area with maximum retinal ganglion cell density (～5,000–7,000 cells/mm²) that affords enhanced resolution in the frontal visual field. Our estimates of spatial resolution based on peak retinal ganglion cell density and eye size (～6–12 mm in axial length) range between ～2 and 4 cycles/degree. Species that occur in more enclosed microhabitats and feed on plant material have lower spatial resolution (～2 cycles/degree) compared with those that roost in open and semiopen areas (～3–3.8 cycles/degree). We suggest that the larger eye and concomitant higher spatial resolution (～4 cycles/degree) in H. monstrosus may have facilitated the carnivorous aspect of its diet. In conclusion, variations in the topographic organization and magnitude of retinal ganglion density reflect the specific ecological needs to detect food/predators and the structural complexity of the environments. J. Comp. Neurol. 525:186–203, 2017. © 2016 Wiley Periodicals, Inc.     

Retinal ganglion cell topography and spatial resolving power in the river hippopotamus (Hippopotamus amphibius)

The river hippopotamus (Hippopotamus amphibius), one of the closest extant relatives to cetaceans, is a large African even‐toed ungulate (Artiodactyla) that grazes and has a semiaquatic lifestyle. Given its unusual phenotype, ecology, and evolutionary history, we sought to measure the topographic distribution of retinal ganglion cell density using stereology and retinal wholemounts. We estimated a total of 243,000 ganglion cells of which 3.4% (8,300) comprise alpha cells. The topographic distribution of both total and alpha cells reveal a dual topographic organization of a temporal and nasal area embedded within a well‐defined horizontal streak. Using maximum density of total ganglion cells and eye size (35 mm, axial length), we estimated upper limits of spatial resolving power of 8 cycles/deg (temporal area, 1,800 cells/mm²), 7.7 cycles/deg (nasal area, 1,700 cells/mm²), and 4.2 cycles/deg (horizontal streak, 250 cells/mm²). Enhanced resolution of the temporal area toward the frontal visual field may facilitate grazing, while resolution of the horizontal streak and nasal area may help the discrimination of objects (predators, conspecifics) in the lateral and posterior visual fields, respectively. Given the presumed role of alpha cells to detect brisk transient stimuli, their similar distribution to the total ganglion cell population may facilitate the detection of approaching objects in equivalent portions of the visual field. Our finding of a nasal area in the river hippopotamus retina supports the notion that this specialization may enhance visual sampling in the posterior visual field to compensate for limited neck mobility as suggested for rhinoceroses and cetaceans.     

Müller cells in adult rabbit retinae: morphology, distribution and implications for function and development

We describe the morphology and distribution of Müller cells in wholemounts of rabbit retinae labelled with either monoclonal antibodies (anti-Vimentin, 3H3, 4D6, and 4H11), or intracellular horseradish peroxidase. Several new features of Müller cell organization are noted. First, Müller cells appear to compose a single morphological class and their morphology varies systematically with retinal thickness. Second, in contrast to other retinal glia, Müller cells have a neuronlike distribution, with a peak density of 10,700-15,000 cells per mm2 at the visual streak and a minimum density of 4,400-6,000 per mm2 at both the superior and inferior retinal edges. There are 4.2 +/- 0.5 x 10(6) Müller cells per retina. Third, unlike in other species, rabbit Müller cells do not contact blood vessels, suggesting that they do not participate in the transfer of metabolites or in the blood:retinal barrier. Fourth, each Müller cell has a vitread endfoot about 20-40 microns in diameter composed of numerous fimbriae. The fimbriae from a single Müller cell generally contact several axon fascicles in the nerve fibre layer, and at each point along its length each fascicle is enclosed by the overlapping fimbriae from several Müller cells. Fifth, in the inner and outer plexiform layers, numerous filamentous branchlets extend 20 microns or more from the radial trunk, interweaving with branchlets from nearby Müller cells to form dense and continuous strata. In the ganglion cell layer and outer nuclear layer, Müller cell processes completely wrap neuronal somata, whereas in the inner nuclear layer they partially wrap somata. We discuss the functional and developmental implications of these observations.