The Growth of the Trout Retina

With two plates (Figs. 3 and 6)

THE retina of a fish goes on growing throughout its life. Conflicting accounts have been given of how the manifold increase in area is achieved. It is obvious that an increase in area implies either the spreading out of pre-existing elements or the addition of new elements ; or both processes may occur together or consecutively. The arrangement of the cones in regular patterns in the retina of a trout offers a convenient means of estimating the relative importance of the spreading out process at every stage of growth; and trout can readily be obtained in an age- and size-range which corresponds to a more than hundredfold change in the area of the retina. It is therefore a particularly suitable species for the investigation here described.

Adult trout (Salmo trutta) were obtained from Lake Bala and the surrounding streams, and young ones were reared in the laboratory and were also obtained from the Midland Hatchery. The body-length from snout to the fork of the tail was measured on a millimetre rule. The eyes were excised and their antero-posterior diameter measured with a micrometer eye-piece.

The measurements of the left eye were always recorded to eliminate any differences between the right and left eyes, although a number of comparisons of the dimensions of the two eyes did not show any differences between them. The measurements of cone-pattern size were obtained from tangential sections of the retinae with a micrometer eye-piece. The distance across ten pattern units was measured and from this the size of one pattern unit was calculated.

In describing the retinal layers I have referred to the surface nearest the chorioid as the external, or outer surface, and that nearest the lens as the inner surface. The region of the retina nearest the iris is described as peripheral and that near the point of entry of the optic nerve as central.

Most of the eyes were fixed in Bouin. The outer coats of the eyeball were removed immediately before embedding (except for very small eyes which were embedded whole), and only the retina was embedded in paraffin wax. It is important to keep the iris intact if one wishes to examine the peripheral growth zone of the retina as they are closely associated with each other and the growth zone is easily damaged when removing the retina from the eye. Serial sections 4−8 μ thick were cut in the tangential plane near the pole of the eye to study the cone pattern, and longitudinal sections were cut to examine the periphery of the retina and the distribution of the retinal cell-layers. The sections were stained with Heidenhain’s haematoxylin and eosin, Ehrlich’s haematoxylin and eosin, and Mallory.

The trout eye shows a considerable increase in size during growth, and this investigation covers a tenfold increase in eye diameter. I have used the body-length as a measure of the size of the fish for comparison with eyediameter, the larger fish being considered older than smaller ones. If eyediameter and body-length are plotted on a logarithmic scale, there is seen to be a direct relationship between them over most of the size range (fig. 1). The growth of the eye is negatively allometric in relation to body-length : this is seen most clearly if the relative eye-diameter (i.e. the ratio of eye-diameter to body-length expressed as a percentage) is plotted against body-length (fig. 2), and this shows that large trout have relatively smaller eyes than small ones.

The rate of increase of eye-diameter, and also of retinal surface area, is greatest in young trout, 15−50 mm long, and at this stage the enlargement of the retina is accompanied by developmental changes involving the relative proportions of retinal layers and cell densities. The vertebrate retina develops from the centre towards the periphery ; the innermost layer differentiates first and the visual cell-layer last. Walls (1942) states that the rods and cones differentiate when proliferation ceases. This generalization is not applicable to the whole retina, but, of any particular small region, it is true that differentiation of the visual cells only takes place when cell-divisions in that region have ceased.

The eyes of trout are well developed at the time of hatching and all the retinal layers can be distinguished in longitudinal sections. Fig. 3, A-D, shows four stages in the development of the retina. In the retina of a newly hatched trout (A) the visual cell-layer is represented by a thin protoplasmic-layer between the outer nuclear-layer and the pigment epithelium, and the individual rods and cones cannot be distinguished. The inner nuclear-layer is very thick and can be subdivided into two regions, the outer consisting of darkly stained, elongated nuclei closely packed together, and the inner region of paler, round nuclei. The inner fibre-layer is relatively thin at this stage and ‘he ganglion cell-layer has three or four layers of nuclei.

The most marked change during the early development of the trout retina is the great increase in thickness of the visual cell-layer. The rods and cones ‘h velop during the first week after hatching but their outer segments do not paite reach to the pigment-cells at this stage. The visual cells grow rapidly during the first month after hatching and then increase in size more slowly. The outer nuclear-layer increases in thickness during later growth. There are only two layers of rod nuclei in young retinae but in an adult retina there are three rows. The origin of these extra rod nuclei in the outer nuclear-layer is uncertain, but it seems probable that undifferentiated cells are present in the inner nuclear-layer of young retinae and that these migrate into the outer nuclear-layer where they differentiate to form rods. The inner nuclear-layer decreases in absolute thickness until about 6 weeks after hatching, after which it remains approximately the same but the nuclei become more spread out. The distinction between the two types of nuclei in this layer becomes less marked, but the outer nuclei are still distinguishable by their darker staining properties 3 months after hatching. The inner and outer fibre-layers increase in thickness as the retina grows and this is particularly noticeable with the inner fibre-layer during the first 2 months. The ganglion cell nuclei become more spread out as the retina grows ; 3−4 weeks after hatching there are only two layers of these nuclei in the central regions of the retina and after another 2 or 3 months they form a single layer and later become more widely dispersed. The increased thickness of the retina in an adult trout, compared with that of an alevin, is largely due to the greater development of the visual cell- and fibre-layers.

During very early development mitoses occur over the whole trout retina, but when differentiation commences near the centre of the retina, divisions in that region cease; and finally the whole retina differentiates except at the extreme edges where mitoses persist and constitute a peripheral growth zone. In young trout retinae mitoses still occur along the edges of the falciform process when they have disappeared from the surrounding areas. I have examined longitudinal sections of trout retinae of various sizes (1·2 mm to 11·6 mm eye-diameter) and I have found mitoses in the periphery throughout the size-range. The mitoses are usually situated immediately beneath the pigment epithelium in the undifferentiated peripheral region of the retina (fig. 3, E). In two or three retinae I have found a mitosis some distance from the edge of the retina in the outer nuclear-layer, in a region where differentiation is completed and the rods and cones have well-developed outer segments. These mitoses were found in the eyes of fish 1−2 years old, so a few nuclei in the outer nuclear-layer must retain the power to undergo mitosis.

I have generally cut each retina in only one longitudinal plane, but by cutting different retinae in different planes I have found cell-divisions in all radii, and mitoses must occur around the whole circumference of the growth zone. The mitoses are not uniformly distributed : in a series of longitudinal sections the divisions may be numerous in a number of sections and then entirely absent from the succeeding 12 or more sections. This irregular grouping of tire dividing cells makes comparisons of the number of mitoses in different retinae unreliable, and also comparisons between different regions of the same retina. If, however, two eyes of very different size are compared, there is no doubt that mitoses are more numerous in the smaller eye. The persistence of mitoses in adult trout retinae indicates that new cells are still being formed and, although there are relatively few divisions, the increase in the number of cells will contribute to the growth of the retina.

There is a regular arrangement of cones in trout retinae of all ages, and the cone pattern is of particular interest when considering the increase in retinal surface area during growth. In very young trout there is a regular arrange-ment of rods and cones which can be seen in tangential sections of the retina. The cone pattern consists of quadrilateral units formed by four double cones surrounding a central single cone, with an additional single cone at each corner (fig. 4). These units are arranged in rows forming parallel lines of cones intersecting at right angles. In very young trout retinae there are usually four rods arranged symmetrically around the central single cone of each pattern unit (fig. 5), but it is not always easy to distinguish the central single cone from the rods at this stage in tangential sections, as they have similar diameters in their central regions (fig. 6, A). This rod pattern is only present in very young trout retinae : the number of rods per pattern unit soon increases and there is an irregular grouping of rods around the central single cone. The average number of rods per pattern unit in trout with eye-diameters of 4 nun and 8 mm is 7 and 12 respectively. The rod arrangement in adult retinae is seen most clearly in tangential sections of dark-adapted retinae because in a light-adapted eye the myoids of the rods are usually too thin to be identified in sections through the cone inner segments.

The cone pattern in an adult trout retina is slightly different from that shown in fig. 4. The additional single cones disappear, so that the typical adult trout cone pattern is that shown in fig. 7. The additional single cones do not disappear simultaneously and, like the differentiation of the retina, the pattern change-over begins near the pole of the eye and progresses towards the edge. The loss of these single cones does not alter the arrangement of the other cones, which remains constant throughout subsequent retinal growth. The distance between two central single cones (a−b, fig. 7) has been used as a measure of the cone-pattern size, and this is not affected by the presence or absence of the additional single cones.

The cone-pattern size has been measured in many trout retinae from a wide size-range, and it is evident from the results that the dimensions of the pattern increase as the eye grows. Fig. 6, B and c, shows tangential sections from a young and a large adult trout retina respectively, and, in addition to the marked difference in the single-cone distance, these sections show that the individual cone elements also increase in size. The relationship between the cone-pattern size (as represented by the single-cone distance) and eyediameter is shown in fig. 8. The pattern enlarges as the eye-diameter increases and the rate of change of pattern size during the growth of the eye is nearly constant. Any deductions from the relationship between pattern size and eye-diameter, concerning the growth of the retina, are made on the assumptions that the single-cone distance is proportional to the square root of the surface area of one pattern unit, and that the eye-diameter is proportional to the square root of the total retinal surface area.

These assumptions involve certain inaccuracies in the calculations of retinal areas. In the first place, the area of one pattern unit is not a flat surface and therefore its surface area is not equal to the square of its single-cone distance, but the curvature of one unit is so small that the errors will be negligible when the area of one pattern unit is considered proportional to the single-cone distance. Much greater inaccuracies are associated with the assumption that the total retinal area is proportional to the square of the eye-diameter. For this to be true, the retina would have to retain the same shape throughout growth. In fact the retina is nearly hemispherical in large eyes, but in young trout the eyes are slightly elongated in the antero-posterior direction. The use of the eye-diameter instead of the retina-diameter probably causes the greatest inaccuracies, but a number of measurements of retinal diameters have been made from longitudinal sections, and the retina-diameter and eyediameter are directly proportional to each other. These values of retinadiameters suggest that the use of eye-diameters does not affect the general relationships between the change in cone-pattern size and the growth of the retina, although it may alter the apparent rates of change of pattern size.

Fig. 8 shows that for a five-fold increase in eye-diameter the single-cone distance is only doubled; it follows that at the end of this period less than 20% of the total assembly of double cones and central single cones were present at the beginning and more than 80% have been added during growth, f ig. 9 shows that the relative single-cone distance (i.e. single-cone distance / eye-diameter%) is not constant in trout retinae and that it decreases as the eye-diameter increases. The increase in cone-pattern size is therefore’ not sufficient to account for the increased retinal area and new cells must be added, particularly in the young eyes where there is a marked decline in the relative single-cone distance (fig. 9). In the larger eyes, where the slope of the graph (fig-9) is nearer the horizontal (and therefore the number of pattern units is approaching constancy), the increase in retinal area must be largely due to the increase in cone-pattern size.

An increase in surface area is the most noticeable feature of post-embryonic growth in trout retinae. During the first few months of development, the increase in surface area is accompanied by conspicuous changes in the relative proportions of the different retinal layers, but these changes do not affect the regular arrangement of visual cells which has been studied in relation to the increased retinal area. Post-embryonic growth in teleosts has been studied in Micropterus by Shafer (1900), and in more detail by Müller (1952) in Lebistes. Shafer studied the cone-pattern size in two different-sized specimens of Micropterus and, from a comparison of the ratio of the squares of the pattern size with the ratio of the squares of the eye-diameters, he concluded that the enlargement and spreading out of the cones was sufficient to account for the increased retinal area and that no new elements were added. In view of the work reported here, based on a large number of measurements on a wide and continuous size-range, it is probable that Shafer was mistaken. It is clear that in trout at least, both addition and spreading contribute to the increase in retinal area, but that the latter contributes proportionately more in older fish. It would therefore have been possible to reach a conclusion similar to Shafer’s if measurements had been made on two trout only, neither of which was very young.

Müller determined that the growth of the eye of Lebistes was negatively biometric and that the retina grew by the addition of cells from a peripheral growth zone. Lebistes has a cone pattern similar to that found in young trout retinae and, although he did not measure the size of the pattern in different fish, he observed that the cones were more spread out in the larger eyes. The growth of the retina in Lebistes is thus due to the addition of new cells and to the spreading out of existing cells, which agrees with my observations in trout retinae, and a combination of these two factors is probably typical of retinal growth in teleosts.

I have found that dividing cells are present at the edge of trout retinae throughout growth, but the presence of mitoses and a definite growth zone have not always been recognized, even in the early developmental stages of teleost retinae. Wunder (1925) could find no region of increased nuclear division during the development of trout and carp retinae, but he believed that multiplication of the cells must nevertheless occur. Mitoses in trout retinae are confined to the peripheral growth zone, except in very young eyes, and new cells must be added at the edge. The peripheral position of the growth zone does not necessarily signify that all new cells differentiate at the edge, as some cells may migrate into more central regions before differentiating. The uniformity of the cone pattern in trout retinae shows that an increase in the number of cones cannot take place by the addition of new cells throughout the retina, as this would upset the regular arrangement ; therefore new cones must differentiate to form new pattern units at the edge of the retina. The result of this centripetal growth of the trout retina is that each part becomes more central as the eye enlarges. Kolmer (1936) claims that retinal growth in teleosts is appositional, and although this is probably true of most retinal cells in trout, the addition of new rods is not confined to the edge and takes place throughout the retina. The number of rods per pattern unit increases as the retina grows and, since the mitoses are almost entirely confined to the edge of the retina, the additional rods cannot be formed by division of cells in the differentiated regions of the retina. The migration of undifferentiated cells into the outer nuclear layer is probably responsible for the increase in the number of rods.

The measurements of the cone-pattern size in trout retinae show conclusively that the pattern increases in size as the eye grows and this will contribute to the increased area of the retina. It has been shown that the rate of increase of the cone-pattern size is almost constant throughout growth, but during early growth this accounts for only a small fraction of the increased area and the addition of new cells from the peripheral growth zone contributes rnore to the growth of the retina. This has been deduced from figs. 8 and 9 and is also confirmed by the greater number of mitoses found in young ‘etinae. During later growth, when the mitotic rate has decreased, the spreading out of the cones is more significant. The spreading out of the visual cells decreases the cone density but this is compensated by growth of the lens and other parts of the eye so that the size of the image on the retina is also increased.

I am very grateful to Professor R. J. Pumphrey, F.R.S., for his invaluable assistance with this work, which was carried out while I held a D.S.I.R. maintenance grant. I should also like to thank Dr. J. W. Jones for providing the trout from Bala and Mr. W. Irvine for taking the photographs.