The winkle Littorina littorea (L.) has camera-type (simple) eyes. Light and electron microscopy were used to generate an accurate geometrical model of the eye, and this was used to predict the eye’s visual performance. The lens is spherical with a diameter of 112μm, and examination of images formed by isolated lenses indicates a mean focal length in water of 126μm. These images are crisp and apparently aberration-free. This, in addition to a shorter than expected focal length, implicates the presence of a gradient of refractive index in the lens. The lens has a focal length to lens-radius ratio of 2.3, which is close to the ‘Matthiessen ratio’ of 2.5 found in the aplanatic lenses of many other marine animals. The lens is predicted to focus images within the retina only when the eye is submerged in water: in air, images are focused distal to the retina. In the central retina, the inter-receptor angle is 1.8° and the rhabdom diameter is 4μm (5.5 times larger than the diffraction blur-circle), indicating a retinal array coarser than that necessary to sample all the information contained in the image. In addition, a low F-number (1.2) and the large possible angles of incidence of rays striking the retina (up to 40°) mean that the eye is likely to suffer substantial spreading of light between rhabdoms (which are unshielded) and further degradation of resolution. Possible behavioural roles for the eyes are discussed.
A well-developed camera-type (simple) eye is potentially a superb visual organ which, for the space it occupies, has the best information capacity of all optical designs. The presence of this kind of eye is usually accompanied by an impressive repertoire of visually guided behaviours. Exceptions to this rule are found in gastropod molluscs, many of which have simple eyes, apparently built for high resolution, but where visually guided behaviour seems surprisingly absent. Gastropods in general are slow moving compared to other animals with good vision, making any visually guided behaviour less obvious and easily overlooked.
One of the best examined gastropod groups are the Littorinidae, which inhabit sea shores and the intertidal zone. Some investigations (Burdon-Jones and Charles, 1958; Charles, 1961a,b; Evans, 1961; Hamilton and Winter, 1982) show that there is indeed some visually guided behaviour in the Littorinidae, but it remains questionable if this behaviour exploits the full information potential of the eyes.
An understanding of the eye, complete enough to explain its size and construction, can be reached only when the eye’s visual performance is known in sufficient detail. If we assume that eyes are matched to the visual needs of their bearers, then the visual performance becomes an informative guide for behavioural studies. Resolution and sensitivity are key parameters describing the performance of an eye, and both can be conveniently calculated from anatomical data. This approach has been used extensively to compare eyes from different animals and eyes of different designs (Land, 1981).
Anatomical data can also be used more specifically to estimate the quality of vision in a single species. Indeed, a morphological investigation of the eye of Littorina littorea by Newell (1965) concludes that the eye is focused when in air, but no attention is given to its resolution and sensitivity. In the present paper we carry the approach further, by creating an accurate geometrical model of the Littorina eye, which is then used to make estimates of both resolution and sensitivity. These values are used to explore the possible roles for the eye and to estimate the light intensities under which it is designed to operate. Measurements of the focal length are included to determine the refractive state of the eye, which is of particular interest in an amphibious animal like Littorina.
MATERIALS AND METHODS
Adult Littorina littorea were collected under water, among rocks and boulders close to the shoreline at the marine biological stations in Tjärnö, Sweden, and Bergen, Norway. The shells measured 17–19 mm in length. The animals were kept in aquaria with proper day-length lighting and a salinity adjusted to that of the catch site.
Histological procedures were performed during the day in normal laboratory lighting. Prior to dissection, the shells were carefully cracked with a small vice. The eyestalks were removed from the animals and fixed in a combined paraformal-dehyde/glutaraldehyde fixative in a sodium cacodylate buffer (Karnovsky, 1965). After fixation, the specimens were rinsed in buffer and post-fixed in 2% OsO4 for 1–2h, after which they were transferred to distilled water prior to dehydration. Dehydration was performed in dimetoxypropan (DMP) with one drop of HC1. The specimens were then put in absolute alcohol, and finally embedded in Epon resin.
Sections for light microscope studies (2–3 μm thick) were cut with glass knives and for electron microscope studies (EM) 50–60 nm thickness sections were cut with a diamond knife on an LKB3 ultrotome. The sections for light microscopy were stained with a 50% solution of Methylene Blue/Azure Blue and post-staining for EM was made with lead citrate/uranyl acetate.
Focal length measurements were carried out on isolated fresh lenses immersed in a microbasin which contained saline. The microbasin was constructed from a microscope slide upon which was waxed an O-ring and small washer (Fig. 1). An object of known size was then located on a cover slide and placed on the illuminated condenser of a microscope. The microbasin was placed on the microscope stage, and a Leitz 40× (n.a. 1.0) objective was lowered into the saline of the microbasin in order to observe the image formed by the lens. To calculate the focal length using the magnification of the image, the formula f=ui/o (Nilsson et al. 1988) was used, where/is the focal length, u is the object distance, i is the image size and o is the size of the object.
Each eye of Littorina is located on a small protuberance lateral to the base of each cephalic tentacle. The eye has the form of a closed vesicle containing a lens, a large vitreous body and the retina (Fig. 2). A model was constructed of the eye of L. littorea in order to show succinctly the morphological dimensions of the eye based on averages obtained from pairs of eyes in 12 individuals (Fig. 3).
The outer corneal layer of the eye consists of a clear layer of cuboidal cells, with an average thickness of 5–6 μm. The thickness of this epithelium in different eyes was quite variable in spite of the identical fixation techniques. In the periphery of the pupil there is a graded transition from this outer corneal epithelium to the normal 25 μm deep epidermal layer containing dark pigment and goblet cells.
Just proximal to the corneal epithelium lies a layer of transparent collagen fibres about 5–12 μm thick. The two structures above have been called ‘conjunctiva’ and ‘comea’, respectively (Newell, 1965), or collectively just termed ‘the cornea’ (Hamilton et al. 1983). For simplicity we use the latter terminology in this investigation. Newell (1965) states that a 6–10 μm thick layer of supporting cells devoid of pigment covers the pupil aperture, but electron micrographs show that this layer is equivalent to the layer of collagen fibres described by Hamilton et al. (1983), which also forms a basement membrane encircling the entire eye.
The large invariable pupil appears perfectly circular when observed in vivo. It is formed by the most distal parts of the retinal cup, which mainly consists of pigment cells. The aperture, which exposes almost the entire lens, is invariant and has a mean diameter of 108 μm, ensuring maximal photon capture. Just proximal to, and in close contact with, the cornea is the lens, perfectly transparent, semi-hard and approximately spherical. It has a mean diameter of 112 gm. In most eyes the lens was very slightly flattened such that the diameter along the eye’s optical axis was shorter than the perpendicular diameter.
Between the lens and the retina, a distance of approximately 70 μm, is a large vitreous body consisting of a clear jelly-like substance that appears structurally homogeneous in semi-thin sections as well as in electron micrographs. Hamilton et al. (1983) mention an acellular layer of gelatinous material between the membrane of the vitreous body and the retina. However, electron micrographs of the area clearly show that there is no vitreous body membrane visible and that the acellular layer consists of a material indistinguishable from the vitreous body, only somewhat less dense and partly filled by branching microvilli from the retina.
The retina consists mainly of two types of cells (Fig. 4), the photoreceptor cells (approximately 50μm long) and the pigment cells (approximately 40 μm long). In accordance with the findings of Newell (1965), the pigment cells completely surround the photoreceptor cells and are densely packed with pigment granules in their distal parts. Proximally, the pigment cells widen somewhat and have, compared to the photoreceptor cells, smaller and less densely stained nuclei.
Distally, each photoreceptor cell forms a rhabdom which is a high microvillar tuft. The rhabdoms are not shielded from each other by screening pigment. The entire retina is both thicker and denser than reported in earlier investigations (Newell, 1965; Hamilton et al. 1983). The distal part of the retinal cup, which forms the pupil, consists mostly of pigment cells and almost no photoreceptive microvilli are present. However, microvilli start to appear just a few micrometres proximal to the pupil aperture, and the microvilli layer increases in thickness to about 12 μm at 30-60° from the optical axis and further to a maximal thickness of about 20 μm in the centre of the retina. It seems rather unlikely that the unaligned microvilli of the rhabdom provide L. littorea with the ability to discriminate the plane of polarization of incident light (Nilsson et al. 1987).
In sections taken just below the microvillar layer, the photoreceptor cells have a diameter ranging from less than 4 μm in the central parts of the retina to about 7 μm in the periphery (Fig. 5). The inter-receptor distance varies accordingly from 4 μm in the central parts to about 8 μm. at the periphery. These values correspond approximately to those found by Hamilton et al. (1983), where the mean value of the photoreceptor diameter was found to be 3.4μm and the mean inter-receptor distance 4.4μm.
To investigate the refractive state of the eye, lenses were isolated and their focal lengths measured. Measurements of 12 lenses gave a mean focal length of 126μm (±4μm, S.D.). This equals 2.3 lens radii, which is very short considering that a homogeneous lens with a refractive index of 1.53 would have a focal length of 3.8 lens radii (Land, 1984). The short focal length indicates that the lens probably contains a gradient of refractive index. Further, the crisp images produced by the lenses imply that spherical aberration is much less than that expected from a homogeneous sphere. The graded staining of the lens in semi-thin sections (Fig. 2) provides yet further evidence in favour of a graded-index lens design.
Knowing the geometry of the L. Littorea eye, and the dimensions of its components, it is now possible to calculate the maximum possible resolution and to estimate the sensitivity of the eye. To do these calculations we need a value for the focal length: this will depend on whether the eye is focused in air or water.
An air or water eye?
L. littorea is found both below and above the water surface so, assuming activity in both media, its life style does not provide many clues about the refractive state of the eye. We can, however, compare the measured focal length of the isolated lens with the anatomical distance between lens and retina. The measured focal distance was 126 μm (distance from nodal point to image). The average anatomical separation of the lens centre (nodal point) and the distal retinal surface along the optical axis is about 120 μm. The depth of the rhabdom layer in the retina is approximately 20 μm, putting the image produced by the lens at a nearly ideal plane, safely within the retina. The additional refracting power of a cornea (added when the eye is in air) would shift the image distally towards the lens, probably outside the retina. In sea water (approximate refractive index 1.34), the refractive power of the corneal surface is neutralised. This strongly suggests that the eye is focused when in water and that the image-forming capabilities of the eye originate from refraction in the lens alone.
The sensitivity of the eye
The resolution of the eye
Matthiessen (1880, 1886) discovered that spherical lenses of several aquatic animals were, in fact, not homogeneous spheres but inhomogeneous, with a gradient of refractive index: the refractive index present in the central part of the lens decreases towards the periphery. Matthiessen also observed that the lenses of the aquatic animals he studied had a ratio of focal length to radius (f/r) of about 2.5 (now called the ‘Matthiessen ratio’), and thus an F-number (f/2r) of about 1.3.
A lens with these properties will have a much shorter focal length because rays will be bent continuously through the lens and not only at its surface, as in a homogeneous lens. In addition, such a lens will, if ideally constructed, correct spherical aberration, because the rays passing through the outer region of the lens will be bent less relative to those passing through more central parts (Pumphrey, 1961). Fletcher et al. (1954) investigated how a gradient of refractive index should be constructed to produce an aplanatic (aberration-free) lens with a focal length of 2.5r (the Matthiessen ratio). They found that such a lens could be obtained if it possessed a central refractive index of 1.52, whereas a homogeneous lens would need a refractive index of 1.66, which is significantly higher than the values possessed by the materials from which biological lenses are made (most biological lens materials have a refractive index between 1.50 and 1.55: Land, 1984). Lenses with these properties are found in several different animal groups, including fish (Matthiessen 1886), coleoid cephalopods (Sivak, 1982), prosobranch gastropods (Newell, 1965), pulmonate gastropods (Land, 1984), one family of polychaete worms (Hermans and Eakin, 1974) and at least one genus of copepod crustaceans (Land, 1976), all pointing towards a remarkable convergent evolution (Land, 1984). In L. littorea the lens has a ratio f/r of about 2.25 and an F-number of 1.16. This geometry, combined with a probable gradient of refractive index, endows L. littorea with an aplanatic lens and a superb optical image quality. Thus, spherical aberration is not suspected to be the limiting factor in the performance of the eye.
Diffraction remains as the only likely limitation to image quality. The size of the diffraction blur-circle (Airy-disc) produced by a point-source object is easily calculated if we assume the incident light (λ) is 500 nm and that the lens has an effective aperture (A) of 108μm and a focal length (f) of 126μm. The diameter of the Airy-disc is given by 2.44 f(λ/A), giving a blur circle diameter of 1.42μm. However, the receptor diameter in the central retina is 4μm, which means that the optical image focused by the lens is much finer than the retinal matrix is able to sample. Consequently, the eye is unable to utilize all the information passed by the optics and the coarseness of the array of rhabdoms in the retina becomes the critical limit to resolution.
In eyes of low F-number (like those of L. littorea), the cone of light focused in the retina (from a point source object) is usually of much greater angular width than that which the rhabdoms themselves can contain by total internal reflection (Warrant and McIntyre, 1990, 1991). Measurements made on fly rhabdomeres (Kirschfeld and Snyder, 1975) indicate a rhabdomeric refractive index of around 1.36, which is only marginally greater than that of the surrounding medium (1.34). This refractive index difference would allow a cone of incident light of rays to be totally internally reflected in the rhabdom if the cone has an angular half-width less than about 10° (Warrant and McIntyre, 1990). Cones of greater width than this result in the spreading of light to neighbouring rhabdoms, which consequently absorb some of the light and thus degrade resolution. Many species with eyes of low F-number have tapetal or pigment shielding between the rhabdoms to prevent this problem (Warrant and McIntyre, 1991), but L. littorea is not one of them. In the distal parts of the retina of L. littorea are tufts of microvilli which form the rhabdoms. In the central parts of the retina these are approximately 20 μm high and 4μm in diameter and have no barriers of screening pigment between them. With an aperture of 108μm, the focused cone of incident light in the retina will have an angular width of around 81°. If we assume that the cone of light is focused in the middle of the depth of the retina, it will be absorbed not only by the target rhabdom (the rhabdom for which the light was intended), but also to some extent by its six nearest and six next nearest neighbour rhabdoms. This extent of retinal spreading implies that the eye is not constructed for high spatial resolution, but rather for higher sensitivity (Warrant and McIntyre, 1991).
An interesting comparison can be made with the nocturnal spider Dinopis subrufus (Blest and Land, 1977), whose principal eyes have a diameter of 1.4 mm, making them among the largest invertebrate simple eyes. In several respects the principal eyes of Dinopis are quite similar to those of L. littorea: the focused cone of incident light has an angular width of 78° and there is no retinal shielding between rhabdoms. However, unlike L. littorea, the rhabdoms of Dinopis are only 2.5 times longer than they are wide (50 μm ×20 μm) so that spreading of light in the retina extends only as far as the immediate neighbouring six rhabdoms. By having a very long focal length (771 μm), Δ ϕ is only 1.5°. Thus, spatial resolution may be much better in Dinopis than in L. littorea: indeed, electrophysiological measurements of receptive fields from Dinopis retinula cells indicate half-widths (acceptance angles) of slightly more than 2° (Laughlin et al. 1980), which is quite remarkable for an eye with such a low F-number (0.6). In addition, Dinopis has a very high sensitivity with S=225 μm2 (equation 1) so, with regard to both resolution and sensitivity, the principal eyes of Dinopis are likely to be better than those of L. littorea. Dinopis however, hunts prey by night, so this superior performance is not surprising.
The possible role of the eye
The findings in this paper indicate that the eye of L. littorea is constructed for vision under water during the day. At best, the anatomical resolution, if it is exploited by the nervous system, allows L. littorea to resolve an object of its own size at approximately 72cm (assumes a shell length of 20mm and Δ ϕ =1.8°). Hamilton et al. (1983) concluded that the eye of another species, L. irrorata, is designed for vision in air. This agrees with the fact that this species is primarily active at low tide, above the water surface. The habitat preferences for L. irrorata and L. littorea are somewhat different: L. irrorata inhabits the upper half of the intertidal zone, whereas L. littorea inhabits the lower half (Newell, 1958, 1965). L. littorea, therefore, seems to be active mainly when submerged, and it is thus not surprising to find that the eyes of this species are designed for vision in water rather than in air. It is not entirely clear if the focal length measurements made by Newell (1965) were conducted with the objective immersed in the saline or if there was air present between the objective and the saline surface over the lens. In the latter case, corrections to the measured value would have to be made to compensate for the differences in refractive index of the object medium (saline) and the objective medium (air). This could possibly explain the differences in focal lengths found in the present investigation compared to those found by Newell (1965). An investigation by Stoll (1973) on Lymnaea stagnalis showed that this basommato-phoran pulmonate has eyes quite similar to those of Littorina. An optically important difference, however, is that the retina of Lymnaea is situated much closer to the lens, making it less probable that the lens is able to focus a crisp image on the retina. Although Stoll states that the role of the eyes remains uncertain, he suggests that they are used for monitoring light intensity and for orientational purposes. The present data suggest that the eyes of L. littorea are good enough to subserve the basic needs of finding food, retracting from predators and locating mates. The eyes of L. littorea show no specializations that would indicate more specific uses, but behavioural studies may clarify this. The results of this study reveal the conditions (in water and daylight) under which these investigations should best be made.
I wish to thank Dr Dan-E. Nilsson for his encouraging supervision and Dr Eric J. Warrant who carefully read the manuscript and suggested many improvements. I also wish to thank Mrs Rita Wallen and Ms Lina Hansen, for their great patience and technical assistance, and Ms Inger Norling for the micrograph prints in this paper.