Observations on the Compound Eyes of the Deep-Sea Ostracod Macrocypridina Castanea

Clear ocean water attenuates blue light by a factor of 10 for every 70 m, and other wavelengths more strongly (Jerlov, 1968). Thus, at a depth of 700m, at which Macrocypridina is commonly encountered, the intensity of penetrating sunlight is about 10¹⁰ times lower than at the surface. If the surface is sunlit, then this level corresponds approximately to the human absolute threshold for vision. Nevertheless, in these dim conditions, vision remains a major sense for many animals. The changes in the camera-type eyes of vertebrates and cephalopods with increasing depth are well-documented (e.g. Marshall, 1954; Locket, 1978). The ways that compound eyes achieve increased sensitivity are less well understood. Most deep-water arthropods have superposition eyes, based on lenses in the euphausiids and mirrors in the decapod shrimps (see Land, 1984). The superposition design with its wide effective pupil and large receptors is intrinsically more sensitive than the more usual apposition type of compound eye. Nevertheless, a few deep-sea arthropods (some isopods, hyperiid amphipods and ostracods) have retained apposition eyes, and it is interesting to determine how this type of eye has been ‘stretched’ to deal with very dim conditions. Some information is available for a deep-sea isopod, Cirolana (Nilsson and Nilsson, 1981) and for the hyperiid amphipods Phronima and Cystisoma (Land, 1981b 1989). This paper provides an unusual example from a third group, the myodocopid ostracods.

The ability of an eye to resolve under different light conditions can be estimated with reasonable accuracy from a knowledge of its anatomy (Kirschfeld, 1974; Snyder, 1979; Land, 1984). In an apposition compound eye the features that contribute most to the sensitivity are the facet diameter (D) and the rhabdom acceptance angle (Δ ρ). Other things being equal, the square of their product (DΔ ρ)² determines the number of photons each rhabdom receives from a scene of given luminance. In contrast, under bright conditions the resolution of an eye is determined by the inter-ommatidial angle (Δ ϕ). There are thus a number of measures by which the performance of eyes can be compared that do not require knowledge of either electrophysiological characteristics or behaviour. Macrocypridina eyes have several features that suggest very high sensitivity -most obviously enormous facets -and here we try to assemble the evidence that will make possible comparisons with the eyes of animals from similar dark environments and from much brighter ones.

In spite of the depth from which they were brought to the surface, specimens of Macrocypridina were often in good condition. They swam actively and, to our surprise, made quite complicated spontaneous eye movements. These were studied by video recording and frame-by-frame analysis. We believe that these are the only eye movements known from an ostracod, and they invite comparison with the well-described eye movements of cladocerans such as Daphnia (Frost, 1975) and Polyphemus (Young and Taylor, 1987).

Macrocypridina castanea are unmistakeable ‘chocolate drop’ ostracods with shells about 7 mm long. They were obtained during Cruise 168 of the RRS Discovery, at a depth of about 800m in the vicinity of 20°N 20°W in the N. Atlantic. The net used (8 m² rectangular mid-water trawl) could be opened and closed at preselected depth by an acoustic signal. The animals were kept in cooled sea water until used for histology or filming, and they usually remained active for several hours.

For histological work, eyes were excised into cold (4 °C) fixative with the following composition: 2 ·5 % glutaraldehyde; 3 % paraformaldehyde; 3 % sucrose and 10 mmol 1⁻¹ EGTA in 150 mmol 1⁻¹ sodium cacodylate buffer. After 2h the eyes were rinsed in buffer and postfixed for 1 h in OsO₄ before they were transferred to buffer solution and stored at 4 °C for 3 weeks. Dehydration was performed in an alcohol series followed by embedding in Araldite. Semi-thin sections (1 –2 μm) for light microscopy were cut on a glass knife and stained with Toluidine Blue. For electron microscopy (Jeol 1200 EX) ultrathin sections were cut on a diamond knife and stained with lead citrate and uranyl acetate.

Video films of eye movements were made using a portable camera (Sanyo VC 500) attached to a dissecting microscope, and a video recorder with single-frame facilities (Sanyo VTC7100). The animals were in water at about 20°C, considerably higher than the 5 –10°C of the water at 800 m depth.

The paired eyes are situated dorsally, slightly anterior to the centre of the animal (Fig. 1). They are cylindrical structures 1.3 mm long and 0.7 mm in diameter. In life the long axis is inclined top-forwards at an angle of about 45 ° to the longitudinal axis of the body (taken as the longest diameter of the carapace), and the top is also tilted inwards towards the midline by about 25 °. In front of the eye and extending somewhat ventrally there is a clear circular region in the otherwise brown-pigmented carapace, and this is undoubtedly a window for the eye to see through (Fig. 1). Each eye contains 27 ommatidia in five rows in an approximately hexagonal lattice which covers only the lateral face of the eye (Fig. 2). The eyes from three individuals all showed the same pattern. The eyes are not symmetrical; the anterior two rows of lenses are much the largest (220 μm) and the posterior lenses the smallest (120 μm). The five most dorsal lenses are all small and belong less clearly to the ommatidial rows that make up the rest of the eye.

The directions of the axes of individual ommatidia were estimated from the axes of the pear-shaped lenses and the pigmented collars around their bases. Although this method is only accurate to a few degrees, it is clear that there are large differences in the inter-ommatidial angles (Δ ϕ) in different regions of the eye. They are smallest anteriorly (about 6 °) and greatest at the rear (about 20 °). Fig. 3 shows how the ommatidial axes of the right eye map onto a sphere around the animal, when the eye is in the ‘rest’ position. There is a definite concentration of axes, and hence greater resolution, in the anteroventral part of the field, with fair coverage of the dorsal hemisphere, and almost none in the posteroventral region. The direction of this ‘acute zone’, and the position of the window in the carapace, anteroventral to the eye, both suggest that this is the most important direction for vision. It is likely that in the sea this region points upwards, since this is the direction of the residual daylight and, if that is the case, the swimming orientation of the animal would be as shown in the lower inset of Fig. 3.

Some information is available for the eyes of two other species of myodocopid ostracods, Cypridina norvegica from a depth of about 400 m and Philomedes globosa which lives in the surface waters down to 50m (Andersson, 1979). In both species the eyes are generally similar to those of Macrocypridina but much smaller, less than 300 μm long, and the lenses are also smaller: 55 μm in diameter in male Cypridina and μm in male Philomedes. In both species there are only three rows of ommatidia, but the total number is 16 in Cypridina and 35 in Philomedes. There is no indication of the kind of anteroposterior asymmetry seen in Macrocypridina.

The entire eye is contained in a thin hypodermal envelope consisting of a cuticular sheath and a single layer of cells. The envelope is only loosely attached to the eye and there is no correlation between the arrangement of ommatidia and hypodermal cells. The cellular composition of the ommatidia is simple and not unlike that of other ostracod compound eyes (see Andersson, 1979).

There is only one class of pigment cell. These lie distally, contain a dark pigment (Figs 4 and 5) and occupy the space between the crystalline cones: one cell in each corner between three cones. The same type of pigment cell also screens the back of the eye where there are no ommatidia. The pigmentation gives the impression of a dark shell around the eye, penetrated only by the crystalline cones.

The shape of the cones depends on their size; the smaller cones are nearly spherical, whereas the larger ones are pear-shaped (Fig. 4). As judged by the staining density in both light and electron microscopy, the cones contain a refractive index gradient. In the smaller ommatidia this gradient constitutes the entire cone, but in the larger ommatidia the gradient is confined to a well-defined body, a lens, inside each cone (Fig. 4). The crystalline cone is composed of two cells, each forming one half. Their nuclei are found basally in the eye, below the rhabdom, and are connected to the cone by thin cytoplasmic threads which lie.embedded between folds of the rhabdom (Fig. 5).

Each ommatidium contains six retinula cells contributing equally to a very wide and short rhabdom. The enormous size of the rhabdoms allows only a thin peripheral layer of ordinary cytoplasm in the retinula cells (Fig. 5). This peripheral layer contains tiny, dark pigment granules which probably prevent any light passing between ommatidia. The rhabdom is irregularly arranged into lobes, sometimes partially separated by folds of the peripheral pigmented layer (Fig. 4). It even seems that in some cases these folds would shade parts of the rhabdom and thus impair the sensitivity. Another unusual feature of the Macrocypridina eye is that some retinula cell nuclei are displaced down the axon, giving the impression of a bipolar sensory cell (Fig. 5).

The fields of view of individual ommatidia were estimated from semi-thin sections. The refractive index gradient in the crystalline cones is the only lens in the ommatidium and, since this lens is nearly concentric and the eye is in water, the nodal point must be very close to the centre of the gradient. The field of view then the angle that the pigment aperture between cone and rhabdom subtends at the nodal point (Fig. 5). This type of estimate is reasonably accurate, and is not dependent on the actual focal length of the lens. Measurements made on several sections cut in different planes revealed a large variation depending on the size of the crystalline cones. The largest cones typically gave acceptance angles (Δ ρ) of 15 –20 ° and the smallest cones 50 –60 °. These follow the same pattern as the inter-ommatidial angles (Δ ϕ), but are about three times longer. These estimates of acceptance angle may be slightly larger than the real physiological values, because the most skew rays travel shorter distances through the rhabdom, but this is certainly not a large enough effect to explain the discrepancy between the Δ ϕ and Δ ρ values.

The eyes show considerable movement in the pitch plane, rotation occurring around a transverse axis through the optic nerve (Figs 2 and 6). Some of the movement is ‘accidental’ in that when the second antennae are withdrawn the muscles actually make contact with the eyes and knock them into a more dorsoventral orientation. However, not all movements are passive and, when the antennae are extended and out of contact with the eyes, bursts of spontaneous activity can be seen. An analysis of a video recording of such a burst is given in Fig. 6. The upper figure shows the full sequence, in which the eye moves through about 50 °, rotating top-backwards in the sagittal (pitch) plane. After 14 s there is a large movement caused by retraction of the antennae, followed by resumption of spontaneous movements. The ‘spikiness’ of the record is not due to measurement error, which is only about 1°. There is a real ‘tremor’ component, with a frequency of about 2 Hz and an amplitude of about 5°, superimposed on a series of slower movements of larger amplitude. In Fig. 6 these components have been separated by taking a running average of the data (averaging period 1 s), and then taking the difference between this and the original record. The former shows the slow movements and the latter the tremor.

Data for Macrocypridina and some comparable species are given in Table 1. It is clear that values for 5 are very high, and similar to the deep-water isopod Cirolana studied by Nilsson and Nilsson (1981) which also has an apposition eye. Both are comparable with that of the reflecting superposition eye of the deep-sea shrimp Oplophorus where S is 3300. These high values contrast with the diurnal shore crab Leptograpsus where 5 is only about 1 in the light-adapted eye. Both Macrocypridina and Cirolana owe their extraordinary sensitivity to a combination of very large facets (A) and large acceptance angles (Δ ρ). Limulus, which lives in fairly shallow water but is nocturnal in its habits, has 5 values not far short of those of the deep-water crustaceans.

The large acceptance and inter-ommatidial angles found in both Macrocypridina and Cirolana carry the penalty of poor resolution, since grating-like objects will not be resolved if they have periods less than Δ ρ or 2 Δ ϕ The eye of neither species achieves good resolution as well as high sensitivity, although there is a weak anterior acute zone in Macrocypridina. This contrasts with the situation in the deep-water amphipod Phronima, which is also encountered down to about 800m. Here the value of S is high compared with that of Leptograpsus, but not nearly as high as in Macrocypridina. However, in the dorsal eyes of Phronima the inter-ommatidial angles are very small indeed, less than 0 ·5 °, and this appears to be an adaptation to a predatory life style in which the function of the eye is to detect small objects that are slightly darker than the background provided by the sea surface (Land, 1981b, 1989). The key feature here is the devotion of a very large amount of eye surface to a very small visual angle. In Macrocypridina there is really no possibility that the eyes could be used in this way. Even if the animal could detect a target that subtended the smallest inter-ommatidial angle, this would still mean that it would only be able to detect an object as large as itself at a distance of 10 body lengths, or about 7 cm. This does not exclude a role for the eyes in the detection of living targets, but certainly means that they cannot detect prey at any distance.

Another comparison can be made between Macrocypridina and the freshwater cladocerans Daphnia and Polyphemus. The former is a suspension feeder and uses its eye principally to maintain body orientation and position in the water column (Young, 1981), whereas the latter is a predator and uses its eye to track prey (Young and Taylor, 1987, 1988). In Daphnia there are only 22 ommatidia, there is no acute zone, and the Δ ϕ is nearly 60 °. However, in Polyphemus there are 130 ommatidia, and in the centre of the pronounced acute zone Δ ϕ is only 2 ° (Nilsson and Odselius, 1982). Macrocypridina, with a total of 54 ommatidia and a minimum value for Δ ϕ of 6 ° in the acute zone, is thus intermediate, in terms of resolution. Again this does not exclude a role for the eyes in predation, but does not support it either.

It is often stated that superposition optics are a better solution to the problems of dim-light vision than apposition optics, and yet both Macrocypridina and Cirolana have apposition eyes with sensitivities (S) comparable with the most sensitive superposition eyes. Is there some other way that these eyes are less effective than a superposition equivalent? In particular, would eyes of the two types with the same size and sensitivity have the same resolution? To answer this, like must be compared with like, and this is not straightforward because the design principle of a single superposition eye does not allow the range of values of Δ ϕ and Δ ρ found in Macrocypridina. A fair comparison would be with a double superposition eye, like those found in many mid-water euphausiids (Land et al. 1979). If the Macrocypridina eye is thought of as consisting of two roughly spherical halves each with a diameter of 0·7 mm, representing the higher and lower resolution regions, then a comparison can be made between these and two model part-spherical superposition eyes with S-values of 1851 and 5031, respectively (see Table 1).

The first eye, representing the acute zone, need only have a narrow field, like the dorsal part of a divided euphausiid eye. In this case the eye radius could be the full 0 ·7 mm, and allowing 350 pm for the rhabdom length, the distance (f) from cornea to image would also be 350 μm. The F-number (f/A ′) for a superposition eye can be as low as 0.5 (Oplophorus, Land, 1976), giving an effective pupil diameter (A ′) of 700 μm, the full width of the eye and clearly the largest possible. Compared with the largest cone diameter of 220 μm in Macrocypridina, the pupil diameter of the model eye is 3.2 times larger, and from equation 2 this means that the angular field of view (Δ ρ) of each receptor can be 3.2 times smaller without reducing the 5-value of 1851. Keeping the ratio of Δ ρ: Δ ϕ the same in the model as in Macrocypridina, the angular receptor separation A ϕ can also be 3.2 times smaller, and the spatial resolution better by the same factor. The other part of the eye, with a larger field, must be modelled as a spherical superposition eye with a radius of μm, and an image formed at half this depth from the cornea, i.e. 175 μm. With an F-number of 0 ·5 the effective aperture would be 350 μm, which is 2 ·9 times larger than the small (120 μm) posterior cones of Macrocypridina (Fig. 2). By the same logic as for the first eye, this means that the resolution can be 2 ·9 times greater without loss of sensitivity.

Degradation of the image quality of dim-light superposition eyes, which is often caused by optical aberrations (Bryceson and McIntyre, 1983; Land et al. 1979; McIntyre and Caveney, 1985), does not invalidate this particular comparison. These defects will tend to make the acceptance angle Δ ρ large in relation to Δ ϕ; but in Macrocypridina the ratio of Δ ρ: Δ ϕ of about 3:1 is itself already high. This ratio is similar to that estimated for the meal-moth Ephestia, which results from image aberrations (Cleary et al. 1977), and also to the ratio measured electro-physiologically in dark-adapted crayfish by Bryceson and McIntyre (1983). We conclude that, if Macrocypridina had a divided superposition eye with the same size, sensitivity and Δ ρ: Δ ϕ ratio as its actual apposition eye, the resolution would improve by a factor of three.

It is interesting to compare the eye movements of Macrocypridina with those of Daphnia, since both are somewhat similar bivalved crustaceans with small movable eyes. Frost (1975) distinguished four types of movement in Daphnia, (i) a high-frequency tremor of 3 –4 ° at 16 Hz, (ii) a slower rhythmic scanning of 5 –6 ° at 4Hz, (iii) large spontaneous ‘saccades’, and (iv) optokinetic nystagmus driven by a rotating pattern. Young (1981) also found a 4Hz, 7 ° movement, which he described as a tremor, and noted that a number of other cladocerans had a tremor of similar frequency and amplitude. It looks as though Macrocypridina shows a tremor analogous to the 4 Hz tremor of Daphnia although a little slower (Fig. 6), plus some other movements perhaps like the saccades described by Frost. We made no study here of stabilizing (optokinetic) movements, nor of possible tracking movements like those seen in Polyphemus (Young and Taylor, 1987). Our observations lend support to the idea that low-amplitude tremor is important in the operation of the compound eyes of lower crustaceans, but they do not tell us what its function is.

We are grateful to Dr Peter Herring of the Institute of Oceanographic Sciences, Wormley, UK, for inviting us to take part in the Discovery cruise, and to the NERC for funding it.