Journal of Comparative Physiology A

, Volume 190, Issue 10, pp 765–789 | Cite as

Vision in the dimmest habitats on Earth

Karl von Frisch Lecture

Abstract

A very large proportion of the world’s animal species are active in dim light, either under the cover of night or in the depths of the sea. The worlds they see can be dim and extended, with light reaching the eyes from all directions at once, or they can be composed of bright point sources, like the multitudes of stars seen in a clear night sky or the rare sparks of bioluminescence that are visible in the deep sea. The eye designs of nocturnal and deep-sea animals have evolved in response to these two very different types of habitats, being optimised for maximum sensitivity to extended scenes, or to point sources, or to both. After describing the many visual adaptations that have evolved across the animal kingdom for maximising sensitivity to extended and point-source scenes, I then use case studies from the recent literature to show how these adaptations have endowed nocturnal animals with excellent vision. Nocturnal animals can see colour and negotiate dimly illuminated obstacles during flight. They can also navigate using learned terrestrial landmarks, the constellations of stars or the dim pattern of polarised light formed around the moon. The conclusion from these studies is clear: nocturnal habitats are just as rich in visual details as diurnal habitats are, and nocturnal animals have evolved visual systems capable of exploiting them. The same is certainly true of deep-sea animals, as future research will no doubt reveal.

Keywords

Camera eye Compound eye Deep-sea vision Nocturnal vision Visual ecology 

Introduction

Humans have evolved to see well in bright daylight. We can distinguish a broad range of colours, and our acute spatial and temporal resolution is amongst the best found in the vertebrate world. But this is only true during the day. On a moonless night, when light levels may be more than 100 million times dimmer, our visual powers are significantly diminished. At light levels at which we are nearly blind, our cats are out stalking prey, and moths are flying agilely between flowers on our balconies. While we are sleeping, millions of other animals are relying on their visual systems to survive. The same is true of an enormous variety of animals inhabiting the eternal darkness of the deep sea, a living space more than 150 times larger than all terrestrial habitats combined. Despite its difficulties for us, the majority of the world’s animals are primarily active in dim light. Moreover, recent research has shown that many of them see remarkably well.

To see reliably, an eye must capture sufficient light. This obvious fact suggests that dim habitats pose a particular problem for vision. Indeed, in some of the dimmest—such as ocean depths beyond the penetration of daylight, or subterranean caves—many animals have dispensed with vision altogether. Fishes like the blind cave fish (Greenwood 1976) or the benthic hagfish (Warrant and Locket 2004) have degenerate or absent eyes. These and other visually impaired bottom-living fishes rely heavily on other senses to find food within the dark oozes. In the words of Walls (1942, p 398): “Life on the bottom is largely life in one plane, and the finding of food by touch and chemoreception is vastly easier. Go far enough along the bottom (if you’re a fish), and you’re bound to bump into something good to eat”. Nevertheless, even in the dimmest light, a sufficiently large and sensitive eye will be able to capture enough photons to support reliable vision.

It is now becoming apparent that the eyes of many nocturnal and deep-sea animals are indeed sufficiently sensitive to permit reliable vision in very dim light. Even in the vast darkness of the bathypelagic deep-sea, the “diminished” eyes of fishes and invertebrates are in fact exquisitely adapted for localising the bright points of bioluminescence produced by other animals. In nocturnal terrestrial habitats, some animals even have sufficient visual sensitivity to distinguish colours, to detect faint movements, to learn visual landmarks, to orient to the faint polarisation pattern produced by the moon and to navigate using the constellations of stars in the sky. Even though our visual powers are reduced at night, those of many other animals are remarkably heightened. In this essay, these impressive visual abilities will be reviewed, and the optical and neural strategies that animals have evolved to improve visual sensitivity in dim light will be explored. The conclusion is clear. The dim nocturnal world we avoid and even fear—not to mention the unfathomable darkness of the ocean depths beyond our reach—are much richer in visual information than we have previously given them credit for.

Nocturnal and deep-sea habitats

To understand the vision of nocturnal and deep-sea animals, it is necessary to understand the nature of the visual scenes they normally encounter.

The intensity and colour of natural daylight

Sunlight is the major source of light on earth, illuminating either directly as during the day, or indirectly by reflection from the moon at night. The spectrum of sunlight and moonlight is thus similar (Fig. 1a), although on nights with a full moon light levels are approximately one million times dimmer (Lythgoe 1979). On moonless nights, illumination is provided by starlight, a light considerably redder than that produced by our own sun (Clark et al., unpublished observations). The illumination provided by starlight on a clear night is typically about 108 times dimmer than that provided by full sunshine. The presence of clouds may further reduce starlight intensities by up to 100 times.
Fig. 1a–c

Light in terrestrial and aquatic habitats. a The irradiance spectra of sunlight and moonlight are similar, but moonlight is about a million times dimmer. Adapted from data given in Moon (1940) and Lythgoe (1979). b The relationship between depth (shown in metres) and the spectrum of downward irradiance in the Golfe du Lion. c The change in the radiance distribution of green light with depth (shown in metres) in Lake Pend Oreille. φ is the angle relative to vertical (0° = vertical, ±180°=horizontal). The distribution is skewed in the direction of the sun near the surface, but becomes more symmetric with increasing depth. In Lake Pend Oreille it becomes perfectly symmetric (asymptotic) at approximately 100 m. b and c adapted from Jerlov (1976)

In the open ocean, the intensity and spectrum of light changes rapidly with depth (Fig. 1b; Jerlov 1976; Warrant and Locket 2004). In a clear sea, the down-welling daylight becomes near-monochromatic blue in colour with increasing depth (Tyler and Smith 1970). Clear water is most transparent to blue light of 475-nm wavelength, and within the first 100 m the orange-red part of the spectrum (beyond 550 nm) is almost entirely absorbed. Ultraviolet (UV) light is also absorbed, but not quite as effectively: in the clearest oceans, biologically relevant intensities remain down to at least 200 m (Frank and Widder 1996; Losey et al. 1999). Due to the absorption of light by water and its narrowing spectrum, the intensity of light available for vision thus falls rapidly with depth. Within the first 100 m, it declines by about 2.6 log units (Jerlov 1976; Warrant and Locket 2004). Below this depth, presumably because the concentration of plankton and suspended organic matter is less and the spectrum of light is approaching its narrowest, light intensity declines less rapidly: about 1.5 orders of magnitude for every 100 m of depth. It reaches starlight levels (during the day) by approximately 600–700 m (Clarke and Denton 1962). Below 1,000 m almost no daylight penetrates, probably not enough to be seen by deep-sea animals (Denton 1990).

The spatial nature of visual scenes

The scenes viewed by terrestrial animals—whether diurnal or nocturnal—are said to be extended, that is, light reaches the eye from many different directions at once. Terrestrial animals also experience sources of light that are much smaller in spatial extent. Stars, for instance, are point sources. So too are the flashes of bioluminescence produced by nocturnal fireflies. The tiny dark silhouette of a female fly passing across a bright sky will also be point-like to a male fly in hot pursuit. However, it is within the depths of the sea that this distinction between extended sources and point sources has had its greatest influence on the visual systems of animals. In the shallower depths, where scattered daylight produces an even blue space light and where the sea floor may be clearly visible, visual scenes are extended in all directions. But at greater depths, where the space light is diminished, bioluminescent point sources also begin to appear, especially from below where the space light is up to 300 times dimmer than that coming from above. Upwards, and even laterally, the scene is still extended. But downwards the scene begins to be dominated by point sources. At still deeper levels, bioluminescent point sources can be seen in all directions. In these mesopelagic depths between 200 and 1,000 m, the scene is semi-extended, becoming less extended and more point-like as the space light diminishes with increasing depth. Below 1,000 m, where daylight no longer penetrates, the visual scene is entirely point-like in nature. As we shall see below, the eyes of deep-sea animals have evolved in response to this changing nature of visual scenes with depth, being optimised for dim extended daylight, or for the detection of bioluminescent point sources, or both (Warrant 2000; Warrant et al. 2003; Warrant and Locket 2004).

In addition to the gradual change from extended to point-like scenes, the spatial properties of light alter in yet another way: with increasing depth in a clear ocean, almost all of the daylight available for vision comes increasingly from above (Fig. 1c). In other directions the scattered space light is much dimmer. The distribution of daylight is dominated by the position of the sun in shallow water, but this dominance declines with depth, disappearing altogether below the so-called “asymptotic depth”. Below this depth—which is approximately 400 m in clearest ocean water—the radiance distribution is vertically symmetric (Jerlov 1976). This radiance distribution has had a profound influence on the evolution of deep-sea vision: the dim down-welling daylight provides a backdrop against which aquatic animals can spot animals floating above, or against which they themselves can be seen from below.

Polarised light in dim habitats

Due to the scattering of sunlight from particles in the atmosphere, the dome of the sky contains a circular pattern of polarised light centred on the sun, with the degree of polarisation being greatest for light emitted from regions of the sky lying on a circular locus 90° from the sun (for review, see Waterman 1981; Wehner 1981). This pattern moves with the sun during the course of the day. At sunset (or sunrise), when the sun is at the horizon, the polarisation pattern is very simple, with the full sky emitting light polarised in a single direction. The degree of polarisation is greatest across the zenith of the sky, up to 85% (Waterman 1981), the highest value attained during the day. Once the sun slips below the horizon, the degree of polarisation declines, reaching negligible values at astronomical twilight when the sun is 18° below the horizon (Rozenberg 1966).

For identical reasons, light from the moon also produces a circular pattern of polarised light, a fact we did not appreciate until very recently (Gál et al. 2001). Apart from its intensity—which is a million times dimmer—the pattern of polarised light formed around the full moon is identical in structure to that formed around the sun. When the moon is in its first or last quarter, the pattern’s intensity is a further ten times dimmer.

Polarised light is also present in the ocean. In still waters, the pattern of polarised light across the dome of the sky is visible through the water surface, but turbidity and the presence of waves can degrade the pattern significantly (Waterman 1954; Horvath and Varju 1995; Cronin and Shashar 2001). In lateral directions underwater, the space light is strongly polarised in the horizontal plane, but the degree of polarisation declines rapidly with depth (Waterman 1981), falling to a constant value of between 13 and 38% below the asymptotic depth.

The reliability of vision in dim light

The greatest challenge for an eye that views a dimly illuminated object is to absorb sufficient photons of light to reliably discriminate it (Laughlin 1990). Because the absorption of photons is stochastic, quantum fluctuations set an upper limit to the visual signal-to-noise ratio (SNR; Rose 1942; De Vries 1943): the greater the number of photons, the greater the signal relative to the noise and the more reliable is visual discrimination. As we shall see below, the eyes of animals living in the world’s dimmest habitats are usually adapted to capturing and absorbing as many photons as possible, thereby maximising the SNR.

The reliability of vision starts with the photoreceptors. Ever since the pioneering studies of Yeandle in the horseshoe crab Limulus in the late 1950s (Yeandle 1958), we have known that photoreceptors can respond to single photons with small but distinct electrical responses known as “bumps”, responses found in both vertebrates and invertebrates. This ability of photoreceptors seems to suggest a remarkable sensitivity, but exactly how ideal are they as photodetectors? To behave as an ideal detector of photons, a photoreceptor must absorb and transduce every photon incident upon it. It must also produce an identical electrical response, of fixed amplitude and duration, to each of them. Finally, an ideal photoreceptor must never produce an electrical response when photons are absent. In other words, the quantum efficiency of an ideal detector is 100%. Is this the case for photoreceptors?

The first condition—absorption and transduction of every photon—is not met by any photoreceptor, although efficiency is remarkably high nonetheless. Numerous studies have now established that there is a 1:1 relationship between transduced photons and bumps (e.g. Fuortes and Yeandle 1964; Lillywhite 1977): a single bump results from the absorption and transduction of no more than a single photon, and a single transduced photon leads to no more than a single bump. However, not all photons that are incident on a photoreceptor are actually absorbed and transduced, that is, not all incident photons lead to a bump. For a start, not all wavelengths of light are absorbed with equal efficiency: the resident rhodopsin molecule absorbs maximally at only a single wavelength, the absorption peak wavelength λmax. All other wavelengths are absorbed with lower efficiency. Moreover, absorption efficiency also depends on the angle relative to the photoreceptor waveguide axis that light is incident (Stavenga 2003). However, for axial light, and across the entire range of visible wavelengths, it has been estimated for flies that about 80% of the energy of the lens diffraction pattern is transferred to the photoreceptive waveguide (van Hateren 1984; Stavenga 2003), and that between 86 and 95% of these photons are absorbed (Stavenga 1976). Of these perhaps only 80% are actually transduced since the quantum capture efficiency of the photopigment is around 80% (Laughlin 1990). Thus, only around 55% of photons that are incident on a fly photoreceptor lead to a bump (0.80×0.86×0.80). Moreover, due to reflection at the cornea, only about 90% of photons that strike the corneal surface actually reach the fly’s retina. Thus, of all photons incident on the cornea of a fly, about half lead to the production of a bump (0.9×0.55), which is quite efficient. Other arthropods have a similarly high efficiency. In the shore crab Leptograpsus variegatus, 45% of corneal photons result in a bump (Doujak 1985), and in the locust Locusta migratoria the figure is 59% (Lillywhite 1977). In the cat the figure is somewhat less, probably below 25% (Barlow et al. 1971).

Photoreceptors also fail to meet the second and third conditions—that responses are identical and do not occur when photons are absent. Bumps vary in latency, duration and amplitude and the biochemical pathways responsible for their generation are occasionally activated in the absence of light. These two failures—known as “transducer noise” and “dark noise” respectively—reduce efficiency even further, as we will now see in more detail.

Sources of noise

Transducer noise—originating in the biochemical processes leading to signal amplification and represented by variations in the latency, duration and amplitude of bumps at low light levels—degrades the reliability of vision. At higher intensities, the quantum bumps fuse to form a continuous but noisy receptor potential which not only results from variations in bump waveform (transducer noise) but also from variations in the rate of photon absorptions (known as “photon shot noise”: see below). Careful experiments have shown that at low light levels the contributions of both types of noise are approximately equal (Lillywhite and Laughlin 1979).

Vision can also become unreliable at low light levels because of dark noise. Even in perfect darkness, the biochemical pathways responsible for transduction are occasionally activated (Barlow 1956). These activations are due either to spontaneous conversion of rhodopsin to metarhodopsin or to spontaneous activation of G-protein coupled steps in the transduction chain. Irrespective of their origin, these activations produce “dark bumps”, electrical responses that are indistinguishable from those produced by real photons, and these are more frequent at higher retinal temperatures. At very low light levels, this dark noise can significantly contaminate visual signals. In insects and crustaceans dark bumps are rare, only around one every 10 h at 25°C (Lillywhite and Laughlin 1979; Dubs et al. 1981; Doujak 1985). But in nocturnal toad rods the rate is much higher—360 h−1 at 20°C (Baylor et al. 1980)—and this sets the ultimate limit to visual sensitivity (Aho et al. 1988, 1993). The levels of dark noise in deep-sea animals are unknown, but low water temperatures (approximately 4°C) probably ensure that rates are rather low. But as we shall see below, these rates may still be sufficient to limit the visibility of bioluminescent flashes.

Even if transducer noise and dark noise is negligible, visual reliability will still be compromised by the random nature of photon arrival. Consider a photoreceptor that absorbs N photons during one integration time. Due to the random nature of photon arrivals (governed by Poisson statistics), the photoreceptor will experience an uncertainty—or shot noise—of \(\surd N\) photons associated with this sample, that is, \(N \pm \surd N\) photons (Land 1981; Warrant and McIntyre 1993). This noise reduces the reliability of intensity discriminations and thereby the ability of the eye to distinguish contrast details in a scene. Moreover, the relative proportion of noise \((\surd N/N = 1/\surd N)\) —and thus the unreliability of vision— will increase with decreasing light intensities (i.e. lower N).

Despite these peripheral sources of noise—and still further sources of noise originating within the circuitry of higher visual processing—vision is nonetheless remarkably efficient. Fewer than ten coincident photons are required for threshold perception of a light flash in humans (Hecht et al. 1942; Barlow 1956). The housefly Musca domestica is capable of detecting the movements of a large field of stripes when as few as two or three photons reach each photoreceptor every second (Scholes and Reichardt 1969; Dubs et al. 1981) and the shore crab L. variegatus can track the movement of a bright star when on average only one bump is produced in each photoreceptor every 3 s (Doujak 1985). As we will discuss below, it is likely that these arthropods employ spatial and temporal summation within and between ommatidia in order to achieve this sensitivity.

Resolution or sensitivity?

Ultimately, however, the ability to see well in dim light requires more than the simple threshold detection of uncomplicated light stimuli. The business of vision is to permit locomotion through a world of obstacles, to distinguish sources of food, to allow the selection of suitable mates, to find new habitats, to locate home and to detect the presence of danger. These tasks not only require good sensitivity to light but also adequate spatial and temporal resolution. They may also require the ability to distinguish colours and to analyse the polarisation pattern of the night sky.

An eye must reconstruct the spatial details of a scene from a matrix of intensity measurements, each measurement provided by an individual detector in the orderly matrix of visual detectors that make up a retina. Like the pixels of a digital camera, the density of detectors sets the finest spatial detail that can be reconstructed: more densely packed detectors can reconstruct finer details. The signals generated in neighbouring detectors must also be sufficiently different to distinguish the contrast differences inherent in these finer details. However, for any given eye radius, a greater density of smaller detectors also means that fewer photons are available for each. And because of noise, as we saw above, fewer photons result in a less reliable measurement of intensity. This leads in turn to fewer reliable levels of response to changes in intensity and thereby poorer contrast discrimination. Thus, even though a greater density of smaller detectors has the potential for higher spatial resolution, the detectors risk being too insensitive to achieve it. This becomes more and more true as light levels fall.

A nice way of quantifying this trade-off between resolution and sensitivity, and to determine how much information can be extracted from a visual scene, is to ask how many different “pictures” can be reconstructed by a matrix of visual channels (Snyder et al. 1977a, b). If there are p visual channels per unit solid angle of visual field, and each channel is capable of distinguishing i levels of intensity, then the maximum number of pictures that can be reconstructed is simply ip. The natural logarithm of this number is the maximum spatial information capacity H of the eye (with unit sr−1, Snyder et al. 1977a, b):
$$ H = {\text{ln}}i^p = p{\text{ln}}i $$
(1)

In bright daylight, many eyes capture so much light that the number of discriminable intensity levels (i) and the density of visual channels (p) can both be large. In other words, the amount of information that can be extracted from a bright scene is high. As light levels fall, each visual channel captures less light and the relative level of noise increases. The SNR falls, and with it the number of intensity levels that the channel can discriminate. The information capacity falls accordingly. One way of offsetting this loss is to somehow increase the amount of light reaching each visual channel. A common strategy is to decrease the density of channels and to widen their receptive fields, that is, to exchange spatial sampling density (reduced p) for an improved number of discriminable intensity levels (increased i).

As we shall see below, many animals that are active in dim light have eye designs that are constructed according to the predictions of this information argument, with fewer but more sensitive visual channels. As we shall see below, they may even have neural circuits that perform spatial summation, the sole purpose of which is to increase sensitivity by coupling visual channels together to produce larger “effective” channels of lower density (Snyder 1977; Laughlin 1981, 1990; Srinivasan et al. 1982; van Hateren 1993; Warrant 1999). Many animals that are active in a range of light intensities can even change this trade-off between sampling density and signal levels according to the ambient light intensity. Some arthropods dark-adapt by widening their rhabdoms and/or shortening their focal lengths at night (Williams 1982, 1983; Leggett and Stavenga 1981; Nilsson 1989). Some experience migrations of screening pigments within the eye that widen the receptive fields of photoreceptors at night and narrow them again during the day (Autrum 1981; Nilsson 1989; Land and Osorio 1990; Stavenga 2004a, b).

Of course, there is an unavoidable consequence of sacrificing spatial sampling density to improve the number of discriminable intensity levels in dim light: vision becomes coarser. That is, reliable contrast discrimination becomes confined to a decreasing range of courser image details as light levels fall, with all finer spatial details drowned by noise. The same information arguments apply to the resolution of contrasts in time. If a single photoreceptor views a moving object, then the photoreceptor’s response will rise and fall as brighter and darker details of the object pass through the visual field. As the object moves faster, the ability of the photoreceptor to reliably code these contrast details declines due to the photoreceptor’s finite response speed (Srinivasan and Bernard 1975): the response of the photoreceptor is eventually too slow to collect sufficient light to support reliable contrast discrimination. The problem worsens as ambient light levels fall. The “sufficient” amount of light can only be collected from progressively slower objects. To compensate, vision generally slows down. This increases the SNR and improves contrast discrimination by suppressing photon noise at temporal frequencies that are too high to be reliably resolved (van Hateren 1993). Just as in the spatial domain, reliable temporal contrast discrimination becomes confined to a decreasing range of slower image details as light levels fall, with all faster details drowned by noise. In other words, in terms of both space and time, the effects of noise confine vision to image details that are increasingly slower and coarser.

The sensitivity of eyes to extended scenes

The eyes of nocturnal and deep-sea animals, straining to see well in a dim extended world, are typically large relative to head size in order to maximise the area of the pupil and the sensitivity of the eye. In addition to a large pupil, sensitivity to a dim extended scene is also improved by having visual channels with wide receptive fields. The wider the receptive field, the larger the region of the scene that the visual channel views, and the greater the number of photons it captures. Of course, spatial resolution is consequently compromised (Warrant and McIntyre 1992), but this trade-off is typical of animals that view dim extended scenes.

These features of eyes are encapsulated in the Land sensitivity equation (Land 1981), which describes the optical sensitivity S of an eye (in units of μm2 sr) to an extended scene of broad spectral content (Warrant and Nilsson 1998), as found in terrestrial habitats: 1
$$ S = \left( {\frac{\pi } {4}} \right)^2 A^2 \left( {\frac{d} {f}} \right)^2 \left( {\frac{{kl}} {{2.3 + kl}}} \right)\quad (broad\;spectrum). $$
(2)
In deep-sea habitats, where daylight is essentially monochromatic, the following expression is more accurate:
$$ S = \left( {\frac{\pi } {4}} \right)^2 A^2 \left( {\frac{d} {f}} \right)^2 \left( {1 - {\text{e}}^{ - kl} } \right)\quad (monochromatic). $$
(3)

Thus, good sensitivity to an extended scene results from a pupil of large area (π A2/4), and photoreceptors each viewing a large solid angle (πd2/4f2 sr) of visual space and absorbing a substantial fraction of the incident light (1−ekl for monochromatic light, kl/(2.3+ kl) for broad spectrum light). Here A is the diameter of the pupil, f the focal length of the eye, and d and l the diameter and length of the photoreceptors, respectively. k is the peak absorption coefficient of the visual pigment. If all lengths have units of micrometers, then the unit of k is μm−1.

A slightly more useful demonstration of sensitivity is to estimate the number of photons N that are absorbed by a photoreceptor from an extended scene during one integration time Δt. By calculating the number of absorbed photons, we can obtain an impression of the reliability of vision \((N \pm \surd N),\) as well as the visual SNR \((\surd N).\)N is obtained by multiplying the optical sensitivity of the eye S (μm2 sr) with the intensity L (photons μm−2 s−1 sr−1) of the light spectrum being viewed (Snyder 1977, 1979; Warrant and Nilsson 1998; Warrant 1999; Kelber et al. 2002):
$$ N = 1.13{\left( {\frac{\pi } {4}} \right)}\;\Delta \rho ^{2} A^{2} \kappa \tau \Delta t{\int {(1 - {\text{e}}^{{ - kR(\lambda )l}} )\;I(\lambda )\;d\lambda } } $$
(4)

We have now replaced the solid angle of visual space viewed by a photoreceptor (πd2/4f2 steradians in Eqs. 2 and 3) with the solid angular subtense of its (assumed) Gaussian receptive field (πΔρ2/2.77=1.13Δρ2, where Δρ is the angular half-width of the receptive field (or “acceptance angle”) in radians: Snyder 1977; Laughlin et al. 1980). Other parameters in Eq. 4 include the quantum efficiency of transduction κ and the transmission of the optics τ. The integral term describes the number of photons that will be absorbed by a photoreceptor viewing a radiance spectrum of quantal intensity I(λ) with a resident visual pigment that has a normalised absorption spectrum R(λ), where λ is wavelength. 2.

Camera eyes

Wider pupils, larger photoreceptors, shorter focal lengths and wider receptive fields all improve photon capture (N) and increase sensitivity (S). These are common features of eyes in animals active in dim light, and are readily seen in nocturnal and deep-sea animals with camera eyes (Fig. 2a). Nocturnal tarsiers and owl monkeys are excellent examples: in the latter, pupils can reach 2 cm in diameter. Tarsiers, skilful hunters of nocturnal insects, have skulls dominated by enormous eye sockets (Fig. 3a). Because the eyes of nocturnal vertebrates are adapted to very low levels of light, their pupils tend to close very tightly during the day in order to protect the sensitive retina from over-exposure to bright daylight. The slit-pupils of many nocturnal hunters, like the domestic cat, can almost completely close during the day (Walls 1942; Ali and Klyne 1985).
Fig. 2a–c

Three common eye designs. a A camera eye. Light is focused by the cornea (air only) and lens to form an image on the retina. b A focal apposition compound eye. Light reaches the photoreceptors exclusively from the small corneal lens located directly above. This eye design is typical of day-active insects. c A refracting superposition compound eye. A large number of corneal facets and bullet-shaped crystalline cones collect and focus light—across the clear zone of the eye (cz)—towards single photoreceptors in the retina. Several hundred, or even thousands, of facets service a single photoreceptor. Not surprisingly, many nocturnal and deep-sea animals have refracting superposition eyes, and benefit from the significant improvement in sensitivity. Diagrams courtesy of Dan-Eric Nilsson

Fig. 3a–e

Optical mechanisms for improving sensitivity to a dim extended scene in camera eyes. a The skull of a tarsier (Tarsius sp.), with huge eye sockets relative to head size. b The two major eye designs of mesopelagic deep-sea fishes (from Warrant et al. 2003). The typical form (left), and the dorsally directed tubular form (right). 1 visual field of left eye, 2 visual field of right eye, 3 binocular visual field. Adapted from Munk (1980). c A rostral aphakic gap (arrow) in the eye of Bathytroctes microlepis. Together with sighting grooves in the snout, the rostral aphakic gaps entirely expose the lens of each eye in the frontal visual field. The temporally placed foveae that view the same frontal field are then assured of maximum light capture. From Munk (1980). d Schematic cross-sections through the eyes of the diurnal swan Cygnus olor (left), and the nocturnal owl Bubo bubo (right). Note the large tubular form of the owl’s eye and the more proximal position of the lens. Diagram adapted from Walls (1942). e The large posterior-medial eyes of the nocturnal net-casting spider Dinopis subrufus, which have an F-number of 0.6. Adapted from Sinclair (1985). Scales: 10 mm (a, d), 1 mm (e)

Compared to nocturnal vertebrates, the pupils of many deep-sea cephalopods are enormous. In this respect, the camera eyes of the giant deep-sea squid Architeuthis dux are undoubtedly among the most sensitive of all eyes. Those of one specimen, captured off the Irish coast in the late nineteenth century, were reported to be 37 cm in diameter (Land 1981). With pupils that can easily be wider than 10 cm (own unpublished data), these massive predators have the potential for extraordinary sensitivity at great depths, and formidable visual powers that could allow them to see prey long before being detected themselves. Even other cephalopods, of more modest size, have large and sensitive eyes. Octopus sp. has a sensitivity of around 4 μm2 sr (Land 1981) which does not sound extremely high until one realises that it is achieved with a spatial resolution that potentially approaches that of the much less sensitive dark-adapted human eye (S=0.9 μm2 sr).

Deep-diving and deep-sea fishes also have sensitive camera eyes. In tunas, thresher sharks and billfishes (marlins, spearfishes and swordfishes)—fishes that can dive to hundreds of metres—the eyes and pupils can be very large. In large swordfishes, tracked at up to 700 m below the surface, the eyes can have a diameter exceeding 10 cm and possess pupils 3 cm wide (E.J. Warrant and K.A. Fritsches, unpublished data). In the blue marlin, a surface-living fish capable of diving to 300 m, the eyes and pupils are respectively 8 and 2.5 cm wide, allowing them a high optical sensitivity of up to 5.6 μm2 sr (Fritsches et al. 2003). Ichthyosaurs, marine reptiles that lived between 90 and 250 million years ago, also had highly sensitive eyes, with diameters of up to 30 cm (Motani et al. 1999).

In deep-sea fishes, eyes tend to become larger relative to body size with increasing depth down to the limit of daylight at 1,000 m (Warrant et al. 2003; Warrant and Locket 2004). Beyond this depth, they tend to become smaller again (although pupil size relative to eye size continues to increase). The eyes of deep-sea fishes, which characteristically have extremely clear optical media (Douglas and McGuigan 1989), come in two forms: (1) dorsal tubular eyes, which dominate the top of the head and view the dim down-welling daylight (found only above 1,000 m), and (2) more typical lateral eyes, placed on the side of the head and viewing the frontal and lateral visual field (Fig. 3b: Munk 1980). Dorsal tubular eyes—with their massive spherical lenses and wide pupils—have a very restricted dorsal field of view, but within this field they are extremely sensitive, and capable of distinguishing the faint silhouettes of animals above. A tubular form allows a portion of a larger (and thus more sensitive) spherical eye to fit onto a smaller head (Walls 1942). The two tubular eyes also effectively view the same region of space, thus collecting double the number of photons that would have been collected by a single eye. The lateral eyes of deep-sea fishes frequently improve sensitivity by having pupils that are significantly wider than the lens (Fig. 3c). Such pupils allow much more light to enter the eye, and despite the fact that much of the light is unfocussed, visual sensitivity is significantly improved (Fig. 2c; Munk 1980; Collin et al. 1997; Warrant et al. 2003; Warrant and Locket 2004).

On land, the camera eyes of nocturnal animals are frequently constructed with a highly curved and powerful cornea that allows the lens to be located closer to the retina than in a diurnal eye. To achieve a focused image on the retina, this implies that the focal length, f, is shorter in such an eye, and a shorter focal length leads to greater sensitivity (Eqs. 2 and 3). In nocturnal vertebrates, this often results in a tubular eye shape, like that found in owls (Fig. 3d; Walls 1942; Murphy et al. 1985; Martin 1994; Martin and Katzir 1999) and bush babies (Walls 1942). Just as in the tubular eyes of deep-sea fishes, even though the visual field can be restricted, the sensitivity of the eye is high. A short focal length (relative to lens diameter A) is also found in the eyes of many nocturnal spiders. The extremely sensitive eyes of the nocturnal net-casting spider Dinopis subrufus are an excellent example (Fig. 3e; Blest and Land 1977). The posterior-medial eyes of this skilful nocturnal hunter have lenses that can reach 1.4 mm in diameter, the largest known for a terrestrial arthropod. But despite this wide lens, the focal length is short, around 0.8 mm. Photographers frequently use the ratio between focal length and aperture diameter (f/A) – known as the “F-number” F—to describe the light-gathering capacities of photographic lenses. A lower F-number leads to a brighter image. Even though a shorter focal length primarily improves sensitivity by widening the receptive fields of the visual channels (see below), F-numbers are still a useful and easy metric for comparing the light-gathering capacities of different eyes (Warrant and McIntyre 1991). Dinopis has an F-number of less than 0.6 (0.8/1.4), a much lower value than in the AM eyes of the diurnal jumping spider Phidippus johnsoni, which have lenses of 0.38 mm diameter and an F-number of 2.0 (Land 1981). The dark-adapted human eye has an F-number of around 2.1. Thus, the eyes of Dinopis are clearly constructed for high sensitivity. So too are the eyes of the nocturnal oilbird Steatornis caripensis (F=1.07: Martin et al. 2004) and the tawny owl Strix aluco (F=1.30: Martin 1999). Despite their implications for high sensitivity, these low F-numbers also imply a wider cone of incident light rays on the retina: if the photoreceptors are not shielded from one another by pigments or a tapetal sheath, resolution is compromised (Warrant and McIntyre 1991; Stavenga 2003). The lowest F-number known is found in the curious mirror eyes of the deep-sea ostracod Gigantocypris mülleri (Land 1981). These eyes—measuring 3 mm across, and with reflectors that look like car headlamps—have large overlapping visual fields and an F-number of just 0.25. Even though these reflector eyes resolve poorly, they have impressive sensitivity, enabling them to detect dim bioluminescence in the dark bathypelagic world where they live.

As mentioned above, a short focal length widens the receptive fields of individual photoreceptors and improves sensitivity to a dim extended scene. So too do wider photoreceptors (d in Eqs. 2 and 3). Nocturnal photoreceptors are typically wide, and due to problems of inherent noise (Nilsson and E.J. Warrant, in preparation), they are also frequently rather short relative to their width. For instance, the rhabdoms of the dark-adapted nocturnal spider Dinopis are 20-μm wide by 55-μm long and are contiguous with each other (Blest and Land 1977). In contrast, the rhabdoms of the diurnal jumping spider Phidippus are well-spaced and only 2-μm wide by 23-μm long (Land 1981). These differences, together with their respective focal lengths (771 μm in Dinopis and 767 μm in Phidippus) mean that the rhabdoms of Dinopis view solid angular regions of space that are about 100 times larger than those viewed by the rhabdoms of Phidippus. These large fields of view, together with their much larger lenses, endow Dinopis with an optical sensitivity of 101 μm2 sr, compared to just 0.038 μm2 sr in Phidippus. Remarkably, even though the rhabdom’s field of view in Dinopis is much larger than in Phidippus, it is still quite narrow: electrophysiologically measured angular-sensitivity functions have a half-width (i.e. acceptance angle) of only 2.3° (Laughlin et al. 1980). This value is narrower than in nearly all nocturnal insects, which generally also have much lower sensitivity. The eyes of Dinopis are not only exquisitely sensitive, they also have excellent resolution, qualities that no doubt assist them during nocturnal hunting.

Nocturnal and deep-sea vertebrates—like deep-sea fishes, the oilbird S. caripensis (Martin et al. 2004) or the owl monkey Aotus trivirgatus (Wikler and Rakic 1990)—frequently have retinas dominated by rods, the vertebrate photoreceptor class responsible for vision in dim light. Cone photoreceptors—responsible for high-resolution colour vision during the day in diurnal vertebrates—are often reduced or absent in nocturnal vertebrates. Diurnal primates like ourselves, or the macaque Maccaca mulatta, have well-developed trichromatic colour vision based on three classes of cones. Aotus, on the other hand, has lost two of its three classes, presumably in the service of sensitivity. The single remaining class, with an absorption peak at 543 nm, implies that the owl monkey is probably a monochromat. Moreover, compared to Maccaca, the retina of Aotus possesses almost three times the density of rods (Wikler and Rakic 1990). Although the owl monkey has clearly sacrificed colour vision to achieve high sensitivity, this does not imply that nocturnal colour vision in other species is impossible. Indeed, as we will see below, it has recently been discovered in hawkmoths and geckoes.

The rod-dominated retinas of deep-sea fishes are exquisitely sensitive (Munk 1980; Warrant et al. 2003; Warrant and Locket 2004). A typical deep-sea rod outer segment is long, over 100 μm (compared to just 26 μm in humans), and has a visual pigment with a peak absorption coefficient k at least twice that found in rods from animals in brighter habitats (e.g. 0.064 μm−1: Partridge et al. 1989; Warrant and Nilsson 1998). Just as in nocturnal terrestrial animals, deep-sea fishes also frequently have a reflective layer—or tapetum—that lies behind the retina. By reflecting light back through the retina, the tapetum allows a second chance for absorption of light that has not been absorbed during its first passage through the rods, thus effectively doubling the lengths of the outer segments. Despite this improvement in sensitivity, the unconstrained reflection provided by a flat tapetum can degrade spatial resolution (Munk 1980; Nicol 1989; Warrant and McIntyre 1991).

In order to increase the pathlength of light travelling through the retina, and to absorb every last photon, the rods of many deep-sea fishes—and also curiously those of the nocturnal oilbird S. caripensis (Martin et al. 2004)—are stacked in layers (Fig. 4a). In the deep temporal foveae of the remarkable alepocephalid Bajacalifornia drakei, there are no less than 28 layers of rod outer segments through which the incoming light must pass, although it is still unclear whether all layers are functional (Locket 1985). The oilbird Steatornis has three rod layers (Martin et al. 2004). Other deep-sea fishes, like the scopelarchid Scopelarchus guntheri, have overcome the narrow receptive fields imposed by the slender rods, which are typically only 2 or 3 μm wide, by grouping large numbers of rod outer segments into round cups of retinal epithelium cells filled with reflective guanine crystals (Fig. 4b; Locket 1971, 1977). These reflective cups optically isolate the receptor groups from each other, effectively turning the rod group into a kind of “macroreceptor”, with the outer segments in close contact, so that light entering the cup is trapped and shared by all of them. This results in a much wider, and more sensitive, receptive field than achievable by a single rod. In S. guntheri, 23 rods, each approximately 2.5-μm wide, fill a cup that is 18-μm wide (Locket 1971, 1977). Spatial resolution is sacrificed to increase sensitivity by 50 times.
Fig. 4a–c

Retinal mechanisms for improving sensitivity to a dim extended scene in camera eyes (examples taken from deep-sea fishes). a A multibank retina consisting of three banks (1, 2, 3) of rods, an adaptation for increasing the path length for light absorption. os rod outer segment, d distal, p proximal. b The grouped retina of a deep-sea fish, seen schematically in tangential view at the level of the inner–outer segment junctions (left). Rod outer segments (os) are assembled as groups (seen in a longitudinal view, right) within cups of reflective guanine crystals (g). This spatial summation of rod signals significantly improves sensitivity to the dim extended space light. c The retina of the mesopelagic lantern fish Lampancytus macdonaldi, which lives at a depth of between 550 and 1,100 m, shown as a map of the distribution of retinal ganglion cells. The retina is shown as a flat mount, and densities are given in thousands of cells per square millimeter. The retina has a rather even and low density of ganglion cells, no doubt an adaptation for improved sensitivity to the dim extended space light. The large dot represents the exit of the optic nerve. T temporal (with frontal visual field), N nasal (with posterior visual field), D dorsal, V ventral. Adapted from Wagner et al. (1998)

Spatial summation of rods need not be as obvious as physically grouping them in a reflective pigment cup. Via the bipolar cells, rods in all vertebrate retinae converge in large numbers onto single ganglion cells, and the size of the converging rod pool sets the potential sensitivity of the receiving ganglion cell and the local visual resolution (Hughes 1977; Collin 1999). A smaller rod pool builds a smaller and less sensitive ganglion cell receptive field, and the density profile of ganglion cells across a retina reveals the trade-off between spatial resolution on the one hand, and sensitivity to an extended source on the other. A denser packing of ganglion cells indicates a region of higher resolution where sensitivity has been sacrificed. On the other hand, extensive rod summation leads to a lower density of ganglion cells and higher sensitivity, a situation found in the retina of the domestic cat (Hughes 1977). Cats have coarser vision than humans (we have less convergence), but are much more sensitive to light (Pasternak and Merigan 1981). Lower mesopelagic lantern fishes of the genus Lampanyctus (Collin and Partridge 1996; Wagner et al. 1998) also have a uniformly low density of ganglion cells, and high rod convergence, throughout the retina (Fig. 4c). With only 1,000–3,000 ganglion cells mm−2, visual resolution is no better than approximately 0.5° (which is much lower than in other deep-sea fishes with specialised retinae: see below). However, the eye is very sensitive for a vertebrate eye: S=247 μm2 sr (Eq. 2). This is approximately 100 times more sensitive than the eye of the nocturnal toad Bufo (Warrant and Nilsson 1998).

And finally, sensitivity to all sources of light—both extended and point-like—is improved by having a slower visual response. A longer response time in dim light, as we mentioned earlier, increases the SNR and improves contrast discrimination by suppressing photon noise at temporal frequencies that are too high to be resolved reliably (van Hateren 1993). For ease of modelling, Snyder (1977) represented this low pass filtering by a finite visual integration time Δt. Like the shutter time of a camera, a longer visual integration time improves the reliability of images in dim light (Eq. 4), but only at the expense of temporal resolution (Warrant 1999). Many nocturnal and deep-sea animals have long integration times. The nocturnal toad B. bufo has a remarkably long dark-adapted integration time of about 1.5 s (Donner 1989; Aho et al. 1993). Deep-sea mysids (Moeller and Case 1994, 1995) and the nocturnal spider Cupiennus salei (Pirhoffer and E.J. Warrant, unpublished data) also have very long visual integration times—up to 200 ms. This is very long for arthropods: day-active houseflies M. domestica have integration times more than 20 times shorter.

Compound eyes

Many of same kinds of mechanisms used in camera eyes to improve sensitivity to a dim extended scene are also found in compound eyes. In apposition compound eyes (Fig. 2b), each ommatidium is isolated from its neighbours by a sleeve of light absorbing screening pigment, thus preventing light reaching the photoreceptors from all but its own small corneal lens. This tiny lens—typically a few tens of micrometres across—represents the pupil of the apposition eye, and not surprisingly this eye design is typical of insects and crustaceans living in bright habitats. Remarkable exceptions do exist, including nocturnal mosquitoes (Land et al. 1997, 1999) and the nocturnal tropical halictid bee Megalopta genalis that we will discuss below. Another is the deep-sea benthic isopod Cirolana borealis, with immensely sensitive apposition eyes. Compared to the ommatidia of relatives living in much shallower water (Fig. 5a), those of Cirolana are gigantic, with huge corneal lenses (150 μm wide) of short focal length (100 μm), and huge photoreceptors (90 μm wide×90 μm long) sitting in cups of reflective pigment (Nilsson and Nilsson 1981). These features endow Cirolana with record high sensitivity for an apposition eye: 5,091 μm2 sr. This very high sensitivity comes only at the cost of resolution. In Cirolana the ommatidia have visual fields of approximately 45°, but for the scavenging lifestyle that this animal leads, sensitivity is probably of greater importance.
Fig. 5a–c

Optical and neural mechanisms for improving sensitivity to a dim extended scene in compound eyes. a Ommatidial structure in the benthic isopod Cirolana borealis (left) and the coastal crab Callinectes ornatus (right). Receptive fields, apertures and rhabdoms are much larger in the deep-living Cirolana borealis, an indication of the greater sensitivity of this eye (which is reinforced by the presence of a reflective pigment tapetum, Re). C corneal facet lens, CC crystalline cone, Rh rhabdom. Scale bar: 100 μm. After Land (1984), using diagrams from Nilsson and Nilsson (1981) and Waterman (1981). b The size of the superposition aperture in three species of onitine dung beetles, the nocturnal Onitis aygulus (upper panel), the crepuscular Onitis alexis (middle panel) and the diurnal Onitis belial (lower panel). The circular superposition aperture is indicated in white on the surface of each eye. The dashed circles refer to “effective apertures”, theoretically-derived apertures in which each facet contributes light equally. In reality, facets near the edge of the aperture contribute less light than those near the centre. Note how the superposition aperture is smaller in beetles from brighter habitats. Scale bar: 0.5 mm. From McIntyre and Caveney (1998). c Schematic transverse sections through the retinas of the nocturnal dung beetle Onitis aygulus (upper panel) and the diurnal dung beetle Onitis belial (lower panel). In O. aygulus, the large flower-shaped rhabdoms (hatched) are almost contiguous, whereas in O. belial they are much smaller, widely separated and surrounded by retinular screening pigments. Scale bar for both panels: 5 μm. Adapted from Warrant and McIntyre (1991)

It is, however, the other major class of compound eyes—the superposition eyes (Fig. 2c)—that are better known for their high sensitivity. In this eye design, typical of nocturnal insects and deep-sea crustaceans, the pigment sleeve is withdrawn, and a wide optically transparent area, the clear zone (cz in Fig. 2c), is interposed between the lenses and the retina. This clear zone—and specially modified crystalline cones—allows light from a narrow region of space to be collected by a large number of ommatidia (comprising the “superposition aperture”) and to be focussed onto a single photoreceptor. Unlike the crystalline cones of most apposition eyes, those of superposition eyes have evolved refractive index gradients (in refracting superposition eyes), or reflecting surfaces (in reflecting superposition eyes), or a combination of both (in parabolic superposition eyes). These optical modifications allow as many as 2,000 lenses to collect light for a single photoreceptor (as in some nocturnal moths). The width of this superposition aperture (A in Eqs. 2, 3, 4) is much larger than the width of a single corneal facet lens and represents a massive improvement in sensitivity (Eq. 2).

The size of the superposition aperture—and thus the sensitivity of the eye—is adapted to the light intensity that the eye normally encounters. This can be seen in the superposition eyes of dung beetles from the single genus Onitis (McIntyre and Caveney 1998: Fig. 5b). Individual species fly in search of dung at different times of day. The superposition apertures of nocturnal species (width A=845 μm in O. aygulus) are considerably larger than those of crepuscular species (A=655 μm in O. alexis). These in turn are more than twice as large as those of diurnal species (A=309 μm in O. belial). Moreover, the nocturnal species O. aygulus has huge contiguous rhabdoms (13 μm wide×86 μm long) compared to the diurnal species O. belial where they are small (6.5×32 μm) and widely spaced (Fig. 5c). And unlike O. aygulus, diurnal species like O. belial also have sheaths of screening pigment around their rhabdoms, which cuts down light flux even more (Warrant and McIntyre 1991). These differences are reflected in the sensitivities (S) of their eyes (Eq. 2): S=59 μm2 sr in O. aygulus but only 1.5 μm2 sr in O. belial.

Of course, there are many deep-sea superposition eyes that also have exquisite sensitivity. The shrimp Oplophorus (Land 1976, 1981), an animal which lives at a depth of about 500 m, has an F-number of just 0.38, a reflective tapetum, and enormous rhabdoms (32×100 μm). These eyes thus attain a very high sensitivity—3,300 μm2 sr, one of the highest values known in the animal kingdom. This high sensitivity comes only at the cost of spatial resolution: Oplophorus would not be able to distinguish point light sources closer than about 15° apart (Land 1976).

Spatial resolution, however, is not always sacrificed. In many nocturnal superposition eyes—particularly dung beetles (Warrant and McIntyre 1990) and moths (Kelber et al. 2002)—a reasonably sharp image is still produced despite their greatly improved sensitivity. In fact, the superposition eyes of the nocturnal European hawkmoth Deilephila elpenor are so sensitive (S=69 μm2 sr) that this moth is able to see colour at night, the first animal known that can do so (see below). Despite this, Deilephila has quite reasonable spatial resolution: photoreceptor angular-sensitivity functions have a half-width (i.e. acceptance angle) of only 3–4°, a value also found in nocturnal and crepuscular dung beetles (Warrant and McIntyre 1990). This angular-sensitivity function is only about twice as broad as those found in many diurnal insects with apposition eyes, such as blowflies (Hardie 1979).

Higher neural enhancement of sensitivity: spatial and temporal summation

Even with the most sensitive optical construction allowable, each visual channel may still not collect enough light from a dim extended scene to subserve reliable vision. In this case, there is still one more strategy available. This strategy—which resides in the neural circuits processing the incoming visual signal—involves neural summation of light in space and time (Snyder 1977; Snyder et al. 1977a, b; Laughlin 1981, 1990; Warrant 1999). We have already discussed summation in time above: when light gets dim, the visual systems of nocturnal and deep-sea animals can improve visual reliability by responding more slowly, either by having slower photoreceptors, or by neurally integrating signals at a higher level in the visual system. Temporal summation only comes at a price: it can drastically degrade the perception of fast-moving objects. Too much temporal summation could be disastrous for a fast-flying nocturnal animal that needs to rapidly judge the presence of approaching obstacles! Not surprisingly, slowly moving animals like toads are more likely to employ temporal summation.

Eyes can also improve image quality by summing photons in space. Instead of each visual channel collecting photons in isolation (as in bright light), the transition to dim light could activate specialised lateral neurons which couple the channels—defined by ganglion cells in vertebrate camera eyes or the ommatidia in arthropod compound eyes—together into groups. We now have evidence of such neurons in the first optic ganglion (lamina ganglionaris) of the nocturnal bee Megalopta genalis (Greiner et al. 2004a). Laterally spreading monopolar cells have also been found in nocturnal cockroaches (Ribi 1977), fireflies (Ohly 1975) and hawkmoths (Strausfeld and Blest 1970), and these have been previously interpreted as an adaptation for spatial summation (Laughlin 1981). Each summed group—themselves now defining the channels—could collect vastly more photons over a much wider visual angle, that is, with a greatly enlarged receptive field. This “spatial summation” results in a simultaneous and unavoidable loss of spatial resolution. Despite being much brighter, the image becomes necessarily coarser.

There is considerable evidence—especially from insects—that spatial summation is likely to be widespread. Dubs et al. (1981) measured behaviourally the threshold optomotor response in tethered flies viewing a wide-field grating stimulus. In parallel they also recorded the rates of bump production at the same threshold intensity, both in the photoreceptors and in the first-order interneurons to which they connect. Using a point source centred in the field of view, the interneuron bump rate was found to be six times that of the photoreceptors—exactly the ratio expected, since six photoreceptors synapse onto one interneuron. However when the point source was exchanged for the dim extended grating stimulus at threshold intensity, the interneuron bump rate increased to between 18 and 20 times the photoreceptor rate, implying that signals from several neighbouring ommatidia were being summed at the interneuron (possibly via presynaptic summation between receptors). Spatial summation has also been found in the motion pathways that process the optomotor response in flies (Dvorak and Snyder 1978). In bright light, the elementary motion detectors calculate motion by using signals generated in neighbouring ommatidia. But as light levels fall, the elementary motion detectors calculate motion by comparing signals generated in successively more distant neighbours, up to two, three or even four ommatidia apart (Pick and Buchner 1979). This increase in spatial summation is accompanied by a decrease in lateral inhibition (Srinivasan and Dvorak 1980).

Even though summation compromises spatial and temporal resolution, the gains in photon catch are so enormous that vision in dim light can be greatly improved. This is especially true in small eyes like those of arthropods. A locust employing summation optimally should be able to see reliably at light intensities up to 100,000 times dimmer than those in which they would normally become blind (Warrant 1999). The same is true of the honeybee A. mellifera. The Africanised race, Apis mellifera scutellata, and the closely related south-east Asian giant honeybee A. dorsata, both forage during dusk and dawn, and even throughout the night, if a moon half-full or larger is present in the sky. This is despite the fact that the honeybee apposition eye should in theory be blind by mid-dusk (Warrant et al. 1996). Behavioural experiments show, however, that even the strictly day-active European honeybee is capable of seeing course habitat features, like large pale flowers, at moonlight intensities. This ability can be explained only if bees optimally sum photons over space and time (Fig. 6: Warrant et al. 1996). According to theory, an Africanised bee could forage slowly in moonlight if it possessed a visual “exposure time” of around 120 ms and summed signals from groups consisting of no more than seven ommatidia, both requirements being rather modest and within the capacity of an insect visual system. As I discuss below, we now believe that summation strategies like these can explain the impressive vision of many other nocturnal insects.
Fig. 6

Dynamic spatial and temporal summation improves sensitivity to a dim extended scene. Spatial resolution—measured behaviourally as the finest spatial frequency detectable (νmax)—is shown for the European honeybee as a function of light intensity (open circles). If light is collected by the optics of isolated ommatidia in the bee’s apposition eye (no spatial or temporal summation: -o-), spatial resolution is predicted to decline with intensity faster than the data, with bees becoming blind (i.e. νmax = 0) by about mid-dusk. With optimum spatial and temporal summation (−s−), spatial resolution is predicted to decline less rapidly with intensity, a prediction that fits the data quite well below mid-dusk intensities. From Warrant et al. (1996)

The sensitivity of eyes to point sources

To accurately localise a narrow point of bioluminescence in the darkness of the deep sea requires another type of eye design than that which is optimal for a dim extended scene. The image of a point source on the retina is, by definition, also a point of light (assuming that aberrations and diffraction do not blur the image too much). For a visual detector to collect all the light from this point, its receptive field need not be any larger than the image itself. Receptors viewing large solid angles of space or performing spatial summation—so important for improved sensitivity to a dim extended scene—are useless for improving detection of a point source. In fact, one would predict that eyes built to see point sources of light against a dark background should have (1) a wide and sensitive pupil to collect sufficient photons to detect the point source, and (2) good spatial resolution to then accurately localise it. This is precisely what one sees in the eyes of deep-sea fishes as one goes deeper in the ocean, from the dim extended world of the mesopelagic zone to the dark point-source world of the bathypelagic zone (Fig. 7a; Warrant 2000; Warrant et al. 2003; Warrant and Locket 2004).
Fig. 7a–d

Strategies for improving the detection of bioluminescent point sources in deep-sea fishes. a The smallest separation of ganglion cells found in a survey of 20 species of deep-sea fishes living at different depths, showing mean separation (in arc min) ±SD. Deeper-living fishes tend towards sharper retinae, a reflection of the increasing dominance of bioluminescent point-source illumination with depth. From Warrant (2000). b Rouleina attrita has a retinal design that is common in bathypelagic fishes, with deep convexiclivate temporal foveae containing densely packed ganglion cells. This design is ideal for localising bioluminescent point sources in the frontal visual field. All conventions as in Fig. 4c. From Wagner et al. (1998). c A transverse light microscope section through the deep convexiclivate fovea of the alepocephalid Conocara macroptera, a fish living in the bathypelagic zone down to 2.2 km. RFT radial fibre layer, INL = inner nuclear layer, R rod photoreceptors. From Locket (1977). d The visibility of bioluminescent point sources in the ocean for bathypelagic fishes of pupil diameter A. The furthest distance r that a fish can see a given point source is plotted as a function of A, for flashes of different intensities (solid lines, 107–1013 photons; dashed line, 1010 photons, an average intensity). The arrow marks an average bathypelagic pupil diameter of 7.3 mm. From Warrant and Locket (2004)

Consider a plot of the smallest angular separation of ganglion cells as a function of depth for some 20 species of deep-sea fishes (Warrant 2000, using data given in Wagner et al. 1998). A smaller separation of ganglion cells results in a greater anatomical resolution. Two trends are obvious. Firstly, the eyes of fish on average become sharper with depth, with the eyes of bathypelagic fish being the sharpest, typically having the potential to resolve details subtending just 5 min of arc (Fig. 7a). This is perfect for detecting point-source bioluminescence, the only light source at these depths. Secondly, the variation across species in ganglion cell separation (and thus resolution) is large in the brighter upper levels (Fig. 7a: error bars), but gradually declines with depth, with minimal variation in the bathypelagic zone (separation = 4.8±2.9 arc min). The small variation in the bathypelagic zone is easy to understand: here the only light sources are point sources and the best strategy involves little summation and high resolution. The large variation in the mesopelagic zone reflects its semiextended nature, with some species adapted to point sources, some to extended sources, and others to both (Warrant 2000). The bathypelagic Rouleina attrita (Fig. 7b) well exemplifies the trend towards sharp vision at depth. Living near sea floors between 1.4 and 2.1 km below the sea surface (Wagner et al. 1998), Rouleina has sharp frontally directed deep convexiclivate foveae (Fig. 7c) possessing up to 27,000 ganglion cells mm−2, giving them a resolution of around 5 min of arc. The same arguments apply to dorsally directed eyes designed to accurately detect the small dark silhouettes of other animals in the dim down-welling daylight. The large apposition eyes of hyperiid amphipods (Land 1989, 2000), and the superposition eyes of euphausiids (Land et al. 1979), become increasingly dorsal in their field of view – and increasingly well resolved – with increasing depth.

In the bathypelagic darkness, a bright pinpoint of bioluminescent light will be highly visible to any eye with a large pupil area. If the point source is located r m from an eye of pupil diameter A, and contains a total of E photons at source, then the number of photons N that enter the eye, and are absorbed by a photoreceptor, is (Warrant 2000):
$$ N = \frac{{EA^2 }} {{16r^2 }}{\text{e}}^{ - \alpha r} (1 - {\text{e}}^{ - kl} ). $$
(5)

The first exponential term describes the attenuation of the bioluminescent flash due to the scattering and absorption of light by water, and α is the combined attenuation coefficient (Lythgoe 1979). For clear water and blue light, α=0.05 m−1 (Denton 1990). The bracketed term is the fraction of incident monochromatic light that is absorbed by a photoreceptor (as in Eq. 3). According to Eq. 5, smaller pupils, or bioluminescent flashes that are further away, will reduce the photon catch and decrease the reliability of vision.

How far away can a bathypelagic fish with a certain pupil size perceive a bioluminescent flash of given intensity? If it were not for the presence of dark noise in the photoreceptors, eyes could theoretically detect flashes infinitely far away. Even low rates of dark noise will eventually swamp the real signals generated by photons arriving from an increasingly distant flash. If the dark noise rate is X “false photons” per second, then following the logic of Land (1981), N photons entering the eye from a bioluminescent flash can be distinguished from the dark noise with 95% reliability when
$$ N \geq 1.96\sqrt {2X} . $$
(6)
For threshold detection, Eqs. 5 and 6 can now be equated:
$$ \frac{{EA^2 }} {{16r^2 }}{\text{e}}^{ - \alpha r} (1 - {\text{e}}^{ - kl} ) = 1.96\sqrt {2X} . $$
(7)

Even though rates of dark noise are not known in any deep-sea fish, a realistic value in the very cold (4°C) waters of the deep sea is X=0.0001 s−1 (see Warrant and Locket 2004). Let us also assume that the photoreceptor absorbs all the light that is incident upon it (i.e. the bracketed term in Eq. 5 equals 1). Imagine the fish has an average bathypelagic pupil and sees a blue bioluminescent flash of average intensity (A=7.3 mm, E=1010 photons, and α=0.05 m−1: averaged for several species—Warrant 2000). With these values, Eq. 7 can be solved numerically to obtain r=97 m. In other words, a fish with a 7.3-mm-wide pupil will see the flash at distances up to 97 m away. Of course, these calculations are based on an assumed rate of dark noise, and moreover we have ignored other forms of noise including photon shot noise and transducer noise. In reality, the maximum visible distances for bioluminescent flashes are probably a lot shorter than those calculated here. Nevertheless, brighter flashes can be seen further away, but even for the brightest flashes and the largest pupils, the maximum range of visibility, with the assumed rate of dark noise, is approximately 230 m (Fig. 7d). In reality, for weak and undernourished bathypelagic fishes, point sources at such distances are probably beyond reach: by the time these slowly moving fishes reach the flash, expending considerable precious energy, the animal that produced the flash is probably long gone. Paradoxically, smaller pupils are therefore an advantage since they are more likely to lack the sensitivity required to detect distant flashes. Far from their reputation as being reduced and regressed, the small eyes of many bathypelagic fishes ideally restrict point source detection to ecologically meaningful distances (Warrant 2000).

How well do nocturnal animals see?

So far we have discussed the various optical and neural adaptations that are available to animals to see well in dim light, but how well do they actually see? Are they able to see as well as diurnal animals can see in bright light? We are now starting to obtain answers to these questions by studying the visual performance of nocturnal animals using behavioural and physiological methods. The results so far suggest that nocturnal animals can see extremely well, and that the visual worlds of colour, polarisation, motion and landmarks are accessible not only to diurnal animals, but to nocturnal ones as well. I will illustrate this with several case studies from our own, and other, recent work.

Navigation using terrestrial landmarks in nocturnal bees

In the tropical jungles of Central and South America are several remarkable groups of bees and wasps that have become nocturnal, despite having apposition compound eyes. The advantages of becoming nocturnal are obvious. The nectar resources of nocturnally flowering plants can be exploited with relatively few competitors at a time of day when there is a lower risk of predation. Despite having apposition eyes, one of these insects—the halictid bee Megalopta genalis—can visually learn and use landmarks for homing at night, an impressive feat considering the size and design of its eyes.

Megalopta is a facultatively social bee, with females living in groups of up to ten in long bored-out sticks (Janzen 1968; Arneson and Wcislo 2004; Wcislo et al. 2004). Each day, bees emerge twice from the nest to forage, each trip lasting 1–36 min. The first foraging trip begins up to an hour before dawn, and the second ends about 40 min after sunset, when light levels under the thick rainforest canopy are similar to starlight levels above (Warrant et al. 2004). Using infrared cameras, we observed that bees leaving the nest turn in mid-air and begin fly backwards in ever increasing arcs while surveying the nest and its surrounds. This behaviour is very similar to the “orientation flights” observed in diurnal honeybees (Becker 1958; Zeil et al. 1996; Lehrer 1996; Capaldi and Dyer 1999). Orientation flights are used by honeybees to learn the arrangement of landmarks around the hive entrance before departure, and these are used to recognise the hive entrance upon return. Could it be that Megalopta uses orientation flights to visually learn the arrangement of landmarks around the nest entrance at night? To test this possibility, we arranged five nest sticks beside one another on a small stand in the rainforest (Warrant et al. 2004: Fig. 8a). Of these, only the middle nest was occupied (marked by a stars in Fig. 8). The bee left its nest at 18.48 hours (16 min after sunset), performed an orientation flight for a few seconds (presumably learning the spatial arrangement of the five nests), and then left (Fig. 8a, upper panel). While the bee was away, the positions of the bee’s nest and an empty nest were swapped (Fig. 8a, lower panel). Upon return at 18.58, the bee flew without hesitation into the central unoccupied nest—the “spatially correct” nest—but after a couple of seconds flew out again, the aroma or some other feature of the nest being repellent. After flying backwards and forwards in front of the five nests, the bee entered the central nest a second time, and again flew straight back out again. This behaviour continued until the bee’s actual nest was again returned to the central position, after which the bee remained in its nest. This experiment strongly suggests that Megalopta visually learns landmarks to locate its nest. A more telling test is to attach a visually obvious landmark—in this case a white square card—over the entrance to the nest (Fig. 8b). In this experiment the occupied nest remained in the central position—only the landmark was moved. Prior to the bee’s departure, the landmark was placed over the entrance of the central, occupied nest (Fig. 8b, upper panel). The bee departed its nest at 18.40 hours, performed an orientation flight, and left. While the bee was away, the white card was placed over the entrance of a neighbouring unoccupied nest. The bee returned at 18.58 hours and flew directly into the landmarked unoccupied nest (Fig. 8b, lower panel). Again the bee flew out almost immediately. After a flight inspection of the nests, the bee returned to the landmarked nest, only to fly out within a couple of seconds. Once again, this behaviour continued until the white card was returned to the central nest.
Fig. 8a, b

Nocturnal landmark orientation in the nocturnal halictid bee Megalopta genalis. Bees leaving for a foraging trip learn the position of their nest relative to others (a), or learn the presence of a white square card attached to their nest (b). Upon return, bees enter the nest marked by the landmarks they have previously learned, not their actual nests (which are marked by stars). Times and light intensities at departure (upper panels) and return (lower panels) are also shown. From Warrant et al. (2004)

Despite having apparently insensitive apposition eyes, the nocturnal bee Megalopta is capable of learning landmarks at light intensities where human observers see almost nothing. How do they achieve this? An inspection of the compound eye (Greiner et al. 2004b) reveals very large facets (up to 36 μm, compared to 20 μm in the diurnal honeybee A. mellifera) and huge rhabdoms (8×350 μm, compared to 2×320 μm in Apis). The effects of Megalopta’s very wide rhabdom are revealed in electrophysiological measurements of its receptive field (Warrant et al. 2004), which are very wide for a bee (the acceptance angle Δρ is 5.6°, compared to 2.6° in Apis). Wider receptive fields are more sensitive to light, although they do compromise spatial resolution. All these differences confer an optical sensitivity of 2.7 μm2 sr (Eq. 2) in Megalopta, compared to just 0.1 μm2 sr in Apis (Greiner et al. 2004b). Is this 27 times greater sensitivity enough to allow Megalopta to learn landmarks at night? First, we can ask how many photons are absorbed at night, during one integration time, by a green-sensitive photoreceptor, from a typical foliage light spectrum. Using Eq. 4, we can calculate that while foraging in its rainforest habitat at night, Megalopta absorbs an average of just 0.15 photons during every visual integration time. This is way too little light to sustain vision, in fact about 100 times lower than required to reliably distinguish the dark entrance to its nest (Warrant et al. 2004). Thus, nocturnal landmark discrimination cannot be mediated by the optics of the eyes alone. As in honeybees (Warrant et al. 1996), spatial and temporal summation of visual signals must be occurring at a higher level in the visual system. Indeed, we have now found laterally spreading first-order interneurons in Megalopta’s first optic ganglion (the lamina ganglionaris) that are possible neural candidates for spatial summation (Greiner et al. 2004a), although it remains to be seen whether they do in fact mediate it.

Navigation using nocturnal celestial landmarks in birds, alligators and insects

Terrestrial landmarks are not the only landmarks available to nocturnal animals. The night sky is also rich in landmarks. There are three main celestial landmarks that animals can potentially employ for nocturnal orientation: the constellations of stars, the bright disk of the moon and the newly discovered (Gál et al. 2001) circular pattern of polarised light produced around the moon. Of these three, the moon’s disk does not require a specialised nocturnal visual system to use as an orientation cue. Probably all animals—both nocturnal and diurnal—have little difficulty seeing the large and brightly shining moon, and indeed many of them use it as a landmark during orientation (reviewed in Dacke et al. 2004). The other two landmarks—the constellations of stars and the moon’s polarisation pattern—are sufficiently dim that many animals, especially those with small eyes, require special visual adaptations to detect them.

Stars are point sources that vary considerably in their intensity from star to star. Eyes with greater sensitivity for point sources, that is, with wider pupils (Eqs. 5, 6, 7), will see many more stars in the night sky than those with less sensitivity. The large 8 mm pupil of a dark-adapted human observer permits many thousands of stars to be seen, while for brown bats, with much smaller eyes and pupils, the number of visible stars falls to just a few hundred (Childs and Buchler 1981). The apposition eyes of the shore crab—with corneal facet lenses 45 μm wide—will be limited to seeing the sky’s 12 brightest stars (Doujak 1985). Arthropods with superposition eyes, such as nocturnal moths, have the potential to see more stars, but as with the shore crab, the spatial resolving power of compound eyes may not be sufficient to accurately distinguish constellations. Nonetheless, the presence of a bright star, or a bright group of stars, within the visual field may act as a simple landmark during migration, as indeed seems to be the case for yellow underwing moths (Sotthibandhu and Baker 1979). Of course, as the stars move throughout the night, so too does the moth’s landmark. This results in a curved flight path, but since the major goal of the moth’s migration is to simply move to a new habitat, rather than to arrive at a specific place, the nightly movement of the stellar landmark is immaterial.

Many vertebrates, with their large pupils and high resolving power, definitely do have the potential to see the spatial arrangements of stars, and many have now been shown to use stellar constellations as landmarks during nocturnal navigation. Even without accounting for the nightly movement of stars across the skies, buntings (Emlen 1970, 1975), garden warblers (Wiltschko et al. 1987) and alligators (Murphy 1981) are able to determine geographic north—and thereby maintain a fixed migratory bearing—using constellations of stars (like the Big Dipper) that rotate around the stationary north star, Polaris. As long as a centre of stellar rotation can be identified, a compass bearing is assured. Buntings raised under a planetarium night sky, where the centre of stellar rotation was deliberately altered to a direction other than north, set their migration directions according to the artificial centre of rotation (Emlen 1970, 1975). Thus, despite having small eyes of limited sensitivity, migrating birds like the bunting can nevertheless discriminate sufficient numbers of stars, with sufficient spatial accuracy, to set a compass bearing during migration.

A different kind of compass bearing is that provided by the dim polarisation pattern produced around the moon. Even though it has an intensity up to ten million times lower than the polarisation pattern produced around the sun, the moon’s pattern still has the potential to act as a compass bearing for nocturnal orientation in the same well-documented way that the sun’s pattern does for diurnal animals (Waterman 1981; Wehner 2001). The only difference is that to be seen, the nocturnal polarisation pattern demands a significantly more sensitive eye.

Such eyes—in this case superposition eyes—are found on the dorsal head surface of the crepuscular-nocturnal African dung beetle Scarabaeus zambesianus (Fig. 9a; Dacke et al. 2003a, b, 2004). Like all dung beetles, Scarabaeus must compete ferociously for a rare and precious food resource: a large elephant dung pad can be a feeding frenzy for literally thousands of beetles. Balls must be created swiftly and efficiently, and then rolled away from the dung pad as rapidly as possible. Fights, and subsequent thefts of balls, are commonplace, so an accidental return to the dung pad is a major mistake. A straight-line course in any direction from the dung pad will ensure the safest and most efficient route away from the strife. We now know that in day-active dung beetles this straight-line course is attained by setting a compass bearing based on the sun’s celestial polarisation pattern (Dacke et al. 2004; Byrne et al. 2003). Specialised polarisation-sensitive regions (or “dorsal rim areas”, DRA) in the dorsal halves of the two dorsal eyes analyse the sky’s polarisation pattern, allowing the beetle to set its course. In S. zambesianus, the two dorsal superposition eyes are sensitive enough to perform this task even at dusk (Fig. 9b), when the sun’s polarisation pattern has become quite dim (Dacke et al. 2003b). We discovered that on nights with a visible moon, Scarabaeus continues to forage—and to roll dung balls in straight lines—long after astronomical twilight when the sun’s polarisation pattern had disappeared. Could this beetle be using the even dimmer pattern of polarised light formed around the moon? To test this, we observed the rolling behaviour of Scarabaeus on nights with (Fig. 9c) and without (Fig. 9d) the moon present in the sky (on moonlit nights we obscured the moon’s disc to ensure that it was unable to act as an orientation cue). Straight-line rolling behaviour was observed only when the moon was present (Fig. 9c), which strongly suggested that polarised light was the cue (Dacke et al. 2003a). To be sure, we then allowed beetles to roll in straight lines on moonlit nights. During their rolling, a UV-transmitting polariser, with its transmission axis oriented perpendicularly to the dominant polarisation direction of the moonlit night sky, was placed directly over the beetle. If the beetle used polarised light to set its course, it should now turn by 90° (to the left or to the right). This is precisely what we observed (Fig. 9e, f). Thus, Scarabaeus’ small but very sensitive dorsal superposition eyes allow it to analyse the moon’s dim polarisation pattern, and to orient with respect to it, the first animal known that can (Dacke et al. 2003a). Many other animals, notably nocturnally migrating insects and birds, may have a similar ability, and utilise a celestial resource that until quite recently remained unknown. Nocturnal crickets, for instance, are prime candidates. Crickets have exquisite sensitivity to polarised light, with behavioural response thresholds occurring at even lower intensities than those of polarised moonlight (Herzmann and Labhart 1989; Labhart et al. 2001).
Fig. 9a–f

Polarisation of moonlight and orientation in the dung beetle Scarabaeus zambesianus. a Scarabaeus rolling a ball of dung. b A false-colour scanning electron micrograph of the left dorsal and ventral eyes. A canthus (can), here cut open to allow easier orientation, separates the two eyes. The blue region in the upper half of the dorsal eye indicates the dorsal rim area (DRA), the region of the eye whose rhabdoms are specialised for the analysis of polarised light. In other parts of the dorsal eye, and throughout the ventral eye, the rhabdoms are unable to process polarised light (eye regions coloured green). ant anterior. Scale bar: 500 μm. c On a moonlit night beetles orient along straight paths. d On moonless nights beetles orient along random paths. e The change in direction (right turn, +70°) taken by a single rolling beetle when a perpendicularly polarising filter is placed over the beetle at the position indicated by the dot. The beetle resumes its direction of travel on exposure to the open sky. The circle represents the extent of the 42-cm-diameter filter. f Average angles of turn made by 22 beetles when covered by the same filter as in E (filled circles, binned in 5° intervals): left turns: −77±14.7°; right turns: 87.9±9.3°. Under a parallel polarising filter, beetles did not deviate from their original direction (open circles). B adapted from Dacke et al. (2003b), all other panels adapted from Dacke et al. (2003a)

The perception of motion and colour in nocturnal hawkmoths and geckoes

A common but seldomly seen visitor to European gardens in summer is the beautiful pink elephant hawkmoth Deilephila elpenor, which feeds on nectar during the darkest hours of the night. This insect has well-developed superposition eyes (Fig. 2c) that it uses for locating flowers. Once located, Deilephila hovers in front of the flower, locates the entrance of the nectar reservoir and then sucks it, a visually demanding task even in bright light. Deilephila has evolved exquisite visual abilities to detect and fixate flowers at night. Our recent work has revealed that this moth is capable of distinguishing motion and colour at light intensities where the human visual system breaks down.

Many different flying insects, including Deilephila, have a specialised region of the optic lobe, the “lobula plate”, devoted to the detection of motion and the analysis of flow field information. The lobula plate of blowflies—arguably the best-studied motion neuropile in the animal kingdom—possesses neurons known as “horizontal” (H) and “vertical” (V) cells that respond to wide field motion. Some cells apparently prefer upward or downward motion, others leftward or rightward, but their actual role is to detect different classes of optic flow (such as pitch, roll and yaw), as beautifully shown by Krapp and Hengstenberg (1996). These cells are also found in the lobula plate of Deilephila, and as in flies, they respond vigorously to sinusoidal patterns of black-and-white stripes (called “gratings”) that move in a certain “preferred direction”. If the stripes are made finer and finer, there comes a point where the insect can no longer resolve the stripes and can thus no longer see them move. When this occurs the cell ceases to respond. When recordings from horizontal-motion-sensitive cells are made in Deilephila, and for comparison also in the blowfly Calliphora (with apposition eyes), it is possible to determine Deilephila’s ability to discriminate motion in dim light (Fig. 10, Warrant 2001; E.J. Warrant et al., unpublished observations). We chose two moving grating patterns, one with broad stripes and one with fine stripes, and by recording from lobula plate cells, we tested the ability of the moth and the fly to see these gratings at different light intensities (early dusk, mid-dusk, moonlight and starlight: Fig. 10). Response (spike) histograms show that the hawkmoth is unable to see the fine grating, even at the brightest intensity. Flies, on the other hand, can see it, which shows that the fly apposition eye is better resolved in bright light than the hawkmoth superposition eye. By mid-dusk, the fly apposition eye is unable to collect enough light to see the movements of the finer grating. By the time intensities have dropped to moonlight levels, it has also lost its ability to see the coarser grating. At some intensity between mid-dusk and moonlight the fly becomes blind. Even though the hawkmoth is unable to respond to movements of the fine grating at any intensity, they maintain a strong response to the coarse grating at all intensities, even starlight.
Fig. 10

The spatial properties of horizontal-motion-sensitive optomotor cells in the hawkmoth Deilephila elpenor and the blowfly Calliphora erythrocephala as a function of light intensity. Post-stimulus time histograms (PSTHs) of cellular responses to sinusoidal gratings (contrast 40%) moving in the preferred direction for hawk moths (left two columns) and flies (right two columns). PSTHs show the time course of the cellular action potential rate in response to a moving grating (horizontal bar indicates movement). Even at times when a movement stimulus is absent, a low spontaneous background level of cellular activity can be seen in the displayed PSTHs. For each insect, PSTHs are shown for four monitor intensities (from the upper row: early-dusk, mid-dusk, moonlight and starlight) for each of two gratings, one coarse (0.05 cycles/deg, left column), and one fine (0.20 cycles/deg, right column). Flies can respond to movements of the fine grating, but only at the brightest intensity, and fail to respond to either grating when intensity falls below mid-dusk levels. Whilst being unable to respond to movements of the fine grating at any intensity, moths maintain a strong response to the coarse grating at all light intensities. From Warrant (2001)

This sensitivity to motion probably allows Deilephila to fixate and follow a breeze-tossed flower in the middle of the night. Remarkably, Deilephila is equally sensitive to colour, a sensitivity it no doubt uses for detecting flowers at night. Using behavioural experiments, Kelber et al. (2002) discovered that Deilephila possesses true trichromatic colour vision at starlight intensities, the first nocturnal animal known to do so.

Like all hawkmoths so far investigated, Deilephila has three different spectral classes of photoreceptors, centred in the ultraviolet (350 nm), the violet (440 nm) and the green (525 nm) parts of the spectrum (Höglund et al. 1973; Schwemer and Paulsen 1973). Deilephila’s day-active cousin, the hummingbird hawkmoth Macroglossum stellatarum, uses these three spectral classes to obtain excellent trichromatic colour vision (Kelber and Hénique 1999; Kelber et al. 2003a). The obvious question was whether Deilephila, despite being nocturnal, also has trichromatic colour vision. Using methods developed by von Frisch, Deilephila was trained to associate a sugar reward with a blue disc (Kelber et al. 2002). We next removed the sugar reward, and placed the blue disk within a series of other disks of equal size but of various shades of grey (Fig. 11a, upper left two panels), or of different colours (Fig. 11a, lower left panel). These discs were placed on a uniform field of pale grey. With regard to the grey tones, these were deliberately made both lighter (upper panels in Fig. 11) and darker (middle panels in Fig. 11) in absolute brightness than the learned blue in order to test whether Deilephila uses achromatic contrast cues, rather than colour, to discriminate the learned disc.
Fig. 11a, b

Nocturnal colour vision in the hawkmoth Deilephila elpenor. a Choice frequencies in tests after training to blue. Each stacked bar diagram gives the choice frequencies to each test colour (as indicated by coloured regions of each bar), at a specific light intensity (see abscissa), for Deilephila elpenor (left, number of choices n given above each bar) or six human subjects (right). Three tests with five colours each (see colour patches) were performed. Moths were able to discriminate blue from all shades of grey at all intensities (two upper panels) but not from brighter and darker shades of blue (lower panel). Humans could not discriminate blue from darker shades of grey under scotopic intensities (10−3 and 10−4 cd m−2). b Choice frequencies in tests after training to yellow. Details as for A. Deilephila chose yellow more frequently than all shades of grey at all light intensities, but did not discriminate between different shades of yellow. Humans were incapable of discriminating yellow from lighter shades of grey at the lowest light level. From Kelber et al. (2002)

In a test situation, how well could a trained Deilephila distinguish this blue disk from the others? We illuminated the test arena with five intensities of broad-spectrum white light, ranging from 1 cd m−2 (mid-to-late dusk) to 10−4 cd m−2 (starlight). At all intensities tested, Deilephila discriminated the blue disk from all shades of grey with a choice frequency of at least 80%, even at starlight intensities (Fig. 11a, upper left two panels: the choice frequencies for each colour are shown as the height of the histogram segment bearing the same colour). When human observers sat in the experimental apparatus and attempted to choose the correct disk from the darker shades of grey, they succeeded only down to full-moon intensities (10−2 cd m−2: Fig. 11a, middle right panel). At dimmer intensities, they were incapable of distinguishing the trained blue from grey, implying that light levels had fallen too low for human colour vision to function. However, for brighter shades of grey (Fig. 11a, upper right panel), humans have no problem distinguishing the blue target at all light levels. This target is however considerably darker than the pale greys, and the human visual system relies on achromatic contrast cues to correctly discriminate the blue disc. This ability to use achromatic contrast cues is seen when humans attempt to discriminate the trained blue from lighter and darker shades of the same colour (Fig. 11a, lower right panel). Humans perform this task rather well. Deilephila, on the other hand, does not: while easily discriminating the trained blue from green or yellow, the moth only poorly discriminates this blue from the lighter and darker shades (Fig. 11a, lower left panel). This implies they do not use achromatic contrast cues for discrimination, and rely instead on colour. When we swapped the training colour to yellow, the results—both for moths and humans—were similar (Fig. 11b), although this time the trained yellow was considerably brighter than the darker shades of grey and easily discriminated by humans at dimmer light levels on this basis alone (Fig. 11b, middle right panel).

The nocturnal hawkmoth Deilephila is clearly capable of discriminating colours at starlight levels of illumination. Thus, Deilephila has colour vision at light levels more than 100 times dimmer than the dimmest in which the human visual system is just able to distinguish colour. Moreover, Deilephila’s colour vision is colour-constant, that is, correct discrimination of colour is not affected by moderate shifts in the spectrum of illumination (Balkenius and Kelber 2004), a feature of all advanced colour vision systems (Kelber et al. 2003b). How is this impressive ability to see colour at night—clearly absent in the large and advanced human visual system—mediated by a small compound eye? Again with the help of Eq. 4, we can calculate the number of photons N absorbed by the green-, blue- and UV-sensitive photoreceptors per integration time, when viewing the coloured targets at the dimmest illuminating intensity (10−4 cd m−2). Even though Deilephila has a very sensitive superposition eye, with an optical sensitivity over 25 times that of the nocturnal bee Megalopta, the photoreceptors only absorb between 1 and 16 photons per integration time for the darker grey shades and the various shades of blue (Kelber et al. 2002). This represents a SNR of between only 1 and 4 ( \({\text{SNR}} = \surd N,\) see above), which is insufficient to distinguish colour reliably (Land 1981). Again, we postulate that Deilephila must use spatial and temporal summation to process colour information at night.

Nocturnal colour vision may actually be quite common (Kelber and Warrant 2004). Large numbers of nocturnal insects may possess colour vision, and indeed even some nocturnal vertebrates. Many nocturnal geckoes, for instance, have three spectral classes of cones, but interestingly no rods. In those species studied (Loew 1994; Loew et al. 1996), the three cone classes have absorption maxima near 365 nm (UV), 460 nm (blue) and 525 nm (green). Recent behavioural experiments, similar to those performed on Deilephila, show that the all-cone retina of the helmet gecko Geckonia chazaliae is capable of discriminating a blue training pattern from a pattern of grey shades at dim moonlight intensities (0.002 cd m−2) using colour cues alone (Roth and Kelber 2004). The helmet gecko is the first nocturnal vertebrate known to have colour vision.

The world is as equally colourful at night as during the day, and the usefulness of colour for object discrimination (Kelber et al. 2003b) has clearly led to the evolution of nocturnal colour vision in both vertebrates and invertebrates. The primary prerequisites are the possession of at least two spectrally distinct classes of photoreceptors and a visual system that can provide sufficient sensitivity both optically and neurally. Even cones can mediate nocturnal colour vision, which is quite a surprise. However, some animals, like nocturnal frogs and toads, have retinas containing two spectral classes of rods (Liebman and Entine 1968). These may also subserve nocturnal colour vision, but this has yet to be established.

Conclusions

Our own rather limited ability to see at night, coupled to unwarranted feelings of sensory superiority, have clouded our expectations concerning the visual powers of other animals. In a sense this is quintessentially human, and it has happened before. When von Frisch—in whose name I have had the honour of writing this review—presented his discovery of trichromatic colour vision in the humble honeybee, and worse, that one of the chromatic channels involved was ultraviolet, his learned colleagues didn’t believe him. It was preposterous, they claimed, that such a lowly creature could have visual powers rivalling those of humans. But that was in 1914, and today of course we know that von Frisch (1914) was right . One can only guess what his colleagues would have said had they known that this trichromatic colour vision is even found in nocturnal insects. Even today the possibility of nocturnal colour vision is difficult for humans to swallow, not helped by the fact that our nearest nocturnal relative, the owl monkey, has dispensed with colour vision in favour of sensitivity. But the reality is, many nocturnal animals do see extremely well, and this is no doubt true of animals inhabiting the vast depths of the sea as well. This is the triumph of nature and evolution, that even the limitations and difficulties imposed by the world’s dimmest habitats have been overcome by the never-ending urgency to survive and reproduce.

Footnotes

  1. 1.

    As recently pointed out by Stavenga (2003), the Land sensitivity equation, despite its great usefulness, does have some limitations, especially for photoreceptors behaving as waveguides and for certain eye designs of lower F-number.

  2. 2.

    This integral is calculated between two wavelength limits: λ1 and λ2 (Warrant and Nilsson 1998). λ1 is set at 280 nm, the lowest wavelength likely to be seen by any animal (because the relative intensity of daylight below this wavelength is very low, and the internal structures of the eye absorb all wavelengths that are shorter). λ2 is the wavelength at which the spectral sensitivity R(λ) falls to 1% of its maximum at its long wavelength end. R(λ) is given by the rhodopsin template of Stavenga et al. (1993). In this template λ2 = 1.231 λmax, where λ max is the absorbance peak wavelength of the visual pigment.

Notes

Acknowledgements

Much of the work presented in this review would not have been possible without the intelligence, generosity and friendship of so many students and colleagues with whom I have been privileged to collaborate, both in Lund and abroad. It is impossible to name them all, but in Lund it is equally impossible to omit naming seven: Marie Dacke, Almut Kelber, Dan-Eric Nilsson, and our past and present students Anna Balkenius, Rikard Frederiksen, Anna Gislén, and Birgit Greiner. With their laughter, genius and friendship, they have made the work outlined in this review a sheer pleasure to be involved in. I am also greatly indebted to the generosity of Professor Friedrich Barth and the Austrian Academy of Sciences who graciously invited me to deliver the Karl von Frisch Lecture upon which this review is based, and for promoting research undertaken on whole organisms. I am also very grateful to Mike Land and Doekele Stavenga for carefully reviewing the manuscript and suggesting many improvements. As always, I am very grateful to the many institutions that have supported our work over the years, including the Swedish Research Council, the Royal Physiographic Society of Lund, the Crafoord Foundation, the Wenner-Gren Foundation, the Swedish International Development Agency (SIDA) and the University of Lund. And finally, I wish to express my sincerest thanks to the man for whom this review is dedicated, Professor Rüdiger Wehner. Apart from his astonishing achievements in unravelling the secrets of polarised light navigation in animals—the scientific thrill from this alone is worth a standing ovation of thanks—Rüdiger Wehner has been an inexhaustible source of inspiration and support, not only for myself personally, but for the entire Vision Group in Lund. It is difficult to believe that such a youthful and vital man is nearing his retirement. I daresay it will pass as casually as any other date—the cataglyphid ants of the world have not divulged all of their secrets quite yet.

References

  1. Aho A-C, Donner K, Hydén C, Larsen LO, Reuter T (1988) Low retinal noise in animals with low body temperature allows high visual sensitivity. Nature 334:348–350CrossRefPubMedGoogle Scholar
  2. Aho A-C, Donner K, Helenius S, Larsen LO, Reuter T (1993) Visual performance of the toad (Bufo bufo) at low light levels: retinal ganglion cell responses and prey-catching accuracy. J Comp Physiol A 172:671–682PubMedGoogle Scholar
  3. Ali MA, Klyne MA (1985) Vision in vertebrates. Plenum Press, New YorkGoogle Scholar
  4. Arneson L, Wcislo WT (2004) Dominant-subordinate relationships in a facultatively social, nocturnal bee, Megalopta genalis (Hymenoptera: Halictidae). J Kansas Entomol Soc (in press)Google Scholar
  5. Autrum H (1981) Light and dark adaptation in invertebrates. In: Autrum H (ed) Handbook of sensory physiology, vol VII/6C. Springer, Berlin Heidelberg New York, pp 1–91Google Scholar
  6. Balkenius A, Kelber A (2004) Colour constancy in diurnal and nocturnal hawkmoths. J Exp Biol 207:3307–3316CrossRefPubMedGoogle Scholar
  7. Barlow HB (1956) Retinal noise and absolute threshold. J Opt Soc Am 46: 634–639PubMedGoogle Scholar
  8. Barlow HB, Levick WR, Yoon M (1971) Responses to single quanta of light in retinal ganglion cells of the cat. Vision Res 11:87–101CrossRefGoogle Scholar
  9. Baylor DA, Matthews G, Yau K-W (1980) Two components of electrical dark noise in toad retinal rod outer segments. J Physiol (Lond) 309:591–621PubMedGoogle Scholar
  10. Becker L (1958) Untersuchungen über das Heimfindevermögen der Bienen. Z Vergl Physiol 41:1–25Google Scholar
  11. Blest AD, Land MF (1977) The physiological optics of Dinopis subrufus: a fish lens in a spider. Proc R Soc London Ser B 196:197–222PubMedGoogle Scholar
  12. Byrne M, Dacke M, Nordström P, Scholtz C, Warrant EJ (2003) Visual cues used by ball-rolling dung beetles for orientation. J Comp Physiol A 189:411–418CrossRefGoogle Scholar
  13. Capaldi EA, Dyer FC (1999) The role of orientation flights on homing performance in honeybees. J Exp Biol 202:1655–1666PubMedGoogle Scholar
  14. Childs SB, Buchler ER (1981) Perception of simulated stars by Eptesicus fuscus (Vespertilionidae): a potential navigational mechanism. Anim Behav 29:1028–1035Google Scholar
  15. Clarke GL, Denton EJ (1962) Light and animal life. In: Hill MN (ed) The sea. Wiley, New York, pp 456–468Google Scholar
  16. Collin SP (1999) Behavioural ecology and retinal cell topography. In: Archer SN, Djamgoz MBA, Loew ER, Partridge JC, Vallerga S (eds) Adaptive mechanisms in the ecology of vision. Kluwer, Dordrecht, pp 509–535Google Scholar
  17. Collin SP, Partridge JC (1996) Fish vision: retinal specialisations in the eyes of deep-sea teleosts. J Fish Biol 49A:157–174CrossRefGoogle Scholar
  18. Collin SP, Hoskins RV, Partridge JC (1997) Tubular eyes of deep-sea fishes: a comparative study of retinal topography. Brain Behav Evol 50:335–357PubMedGoogle Scholar
  19. Cronin TW, Shashar N (2001) The linearly polarized light field in clear, tropical marine waters: spatial and temporal variation of light intensity, degree of polarization and e-vector angle. J Exp Biol 204:2461–2467PubMedGoogle Scholar
  20. Dacke M, Nilsson D-E, Scholtz CH, Byrne M, Warrant EJ (2003a) Insect orientation to polarized moonlight. Nature 424:33CrossRefGoogle Scholar
  21. Dacke M, Nordström P, Scholtz CH (2003b) Twilight orientation to polarised light in the crepuscular dung beetle Scarabaeus zambesianus. J Exp Biol 106:1535–1543CrossRefGoogle Scholar
  22. Dacke M, Byrne M, Scholtz CH, Warrant EJ (2004) Lunar orientation in a beetle. Proc R Soc Lond B 271:361–365CrossRefPubMedGoogle Scholar
  23. De Vries H (1943) The quantum character of light and its bearing upon threshold of vision, the differential sensitivity and visual acuity of the eye. Physica 10:553–564Google Scholar
  24. Denton EJ (1990) Light and vision at depths greater than 200 metres. In: Herring PJ, Campbell AK, Whitfield M, Maddock L (eds) Light and life in the sea. Cambridge University Press, Cambridge, pp 127–148Google Scholar
  25. Donner K (1989) Visual latency and brightness: an interpretation based on the responses of rods and ganglion cells in the frog retina. Vis Neurosci 3:39–51PubMedGoogle Scholar
  26. Douglas RH, McGuigan CM (1989) The spectral transmission of freshwater teleost ocular media—an interspecific comparison and a guide to potential ultraviolet sensitivity. Vision Res 29:871–879CrossRefPubMedGoogle Scholar
  27. Doujak FE (1985) Can a shore crab see a star? J Exp Biol 166:385–393Google Scholar
  28. Dubs A, Laughlin SB, Srinivasan MV (1981) Single photon signals in fly photoreceptors and first order interneurons at behavioural threshold. J Physiol (Lond) 317:317–334PubMedGoogle Scholar
  29. Dvorak D, Snyder AW (1978) The relationship between visual acuity and illumination in the fly Lucilia sericata. Z Naturforsch C 33:139–143PubMedGoogle Scholar
  30. Emlen ST (1970) Celestial rotation: its importance in the development of migratory orientation. Science 170:1198–2011Google Scholar
  31. Emlen ST (1975) The stellar-orientation system of a migratory bird. Sci Am 233:102–111PubMedGoogle Scholar
  32. Frank TM, Widder EA (1996) UV light in the deep sea: in situ measurements of downwelling irradiance in relation to the visual threshold sensitivity of UV-sensitive crustaceans. Mar Fresh Behav Physiol 27:189–197Google Scholar
  33. Frisch K von (1914) Der Farbensinn und Formensinn der Biene. Zoologische Jahrbücher. Abt Allg Zool Physiol Tiere 35:1–188Google Scholar
  34. Fritsches KA, Marshall NJ, Warrant EJ (2003) Retinal specialisations in the blue marlin—eyes built for sensitivity to low light levels. Mar Fresh Res 54:1–9CrossRefGoogle Scholar
  35. Fuortes MGF, Yeandle S (1964) Probability of occurrence of discrete potential waves in the eye of Limulus. J Gen Physiol 47:443–463CrossRefPubMedGoogle Scholar
  36. Gál J, Horváth G, Barta A, Wehner R (2001) Polarization of the moonlit clear night sky measured by full-sky imaging polarimetry at full moon: comparison of the polarization of moonlit and sunlit skies. J Geophys Res 106:22647–22653CrossRefGoogle Scholar
  37. Greenwood PH (1976) A new and eyeless cobitid fish (Pisces, Cypriniformes) from the Zagros Mountains, Iran. J Zool Soc Lond 180:129–137Google Scholar
  38. Greiner B, Ribi WA, Warrant EJ (2004a) Neuronal organisation in the first optic ganglion of the nocturnal bee Megalopta genalis. Cell Tissue Res (in press)CrossRefPubMedGoogle Scholar
  39. Greiner B, Ribi WA, Warrant EJ (2004b) Retinal and optical adaptations for nocturnal vision in the halictid bee Megalopta genalis. Cell Tissue Res 316:377–390CrossRefPubMedGoogle Scholar
  40. Hardie R (1979) Electrophysiological analysis of fly retina. I: comparative properties of R1–6 and R7 and 8. J Comp Physiol A 129:19–33Google Scholar
  41. Hateren JH van (1984) Waveguide theory applied to optically measured angular sensitivities of fly photoreceptors. J Comp Physiol 154:761–771Google Scholar
  42. Hateren JH van (1993) Spatiotemporal contrast sensitivity of early vision. Vision Res 33:257–267CrossRefPubMedGoogle Scholar
  43. Hecht S, Shlaer S, Pirenne (1942) Energy, quanta and vision. J Gen Physiol 25:819–840CrossRefGoogle Scholar
  44. Herzmann D, Labhart T (1989) Spectral sensitivity and absolute threshold of polarization vision in crickets: a behavioral study. J Comp Physiol A 165:315–319Google Scholar
  45. Höglund G, Hamdorf K, Rosner G (1973) Trichromatic visual system in an insect and its sensitivity control by blue light. J Comp Physiol 86:265–279Google Scholar
  46. Horvath G, Varju D (1995) Underwater refraction-polarization patterns of skylight perceived by aquatic animals through Snell’s window of the flat water surface. Vision Res 35:1651–1666CrossRefPubMedGoogle Scholar
  47. Hughes A (1977) The topography of vision in mammals of contrasting life style: comparative optics and retinal organisation. In: Crescitelli F (ed) Handbook of sensory physiology, vol VII/5. Springer, Berlin Heidelberg New York, pp 613–756Google Scholar
  48. Janzen DH (1968) Notes on nesting and foraging behavior of Megalopta (Hymenoptera: Halictidae) in Costa Rica. J Kansas Entomol Soc 41:342–350Google Scholar
  49. Jerlov NG (1976) Marine optics. Elsevier, AmsterdamGoogle Scholar
  50. Kelber A, Hénique U (1999) Trichromatic colour vision in the hummingbird hawkmoth, Macroglossum stellatarum. J Comp Physiol A 184:535–541CrossRefGoogle Scholar
  51. Kelber A, Warrant EJ (2004) Colour vision in dim light: more common than we think. News Physiol Sci (in press)Google Scholar
  52. Kelber A, Balkenius A, Warrant EJ (2002) Scotopic colour vision in nocturnal hawkmoths. Nature 419:922–925CrossRefPubMedGoogle Scholar
  53. Kelber A, Balkenius A, Warrant EJ (2003a) Colour vision in diurnal and nocturnal hawkmoths. Integr Comp Biol 43:571–579Google Scholar
  54. Kelber A, Vorobyev M, Osorio D (2003b) Animal colour vision—behavioural tests and physiological concepts. Biol Rev 78:81–118CrossRefPubMedGoogle Scholar
  55. Krapp HG, Hengstenberg R (1996) Estimation of self-motion by optic flow processing in single visual neurons. Nature 384:463–466CrossRefPubMedGoogle Scholar
  56. Labhart T, Petzold J, Helbling H (2001) Spatial integration in polarization-sensitive interneurones of crickets: a survey of evidence, mechanisms and benefits. J Exp Biol 204:2423–2430PubMedGoogle Scholar
  57. Land MF (1976) Superposition images are formed by reflection in the eyes of some oceanic decapod crustacea. Nature 263:764–765PubMedGoogle Scholar
  58. Land MF (1981) Optics and vision in invertebrates. In: Autrum H (ed) Handbook of sensory physiology, vol VII/6B. Springer, Berlin Heidelberg New York, pp 471–592Google Scholar
  59. Land MF (1984) Crustacea. In: Ali MA (ed) Photoreception and vision in invertebrates. Plenum, New York, pp 401–438Google Scholar
  60. Land MF (1989) The eyes of hyperiid amphipods: relations of optical structure to depth. J Comp Physiol A 164:751–762Google Scholar
  61. Land MF (2000) On the functions of double eyes in midwater animals. Philos Trans R Soc London Ser B 355:1147–1150CrossRefGoogle Scholar
  62. Land MF, Osorio DC (1990) Waveguide modes and pupil action in the eyes of butterflies. Proc R Soc London Ser B 241:93–100Google Scholar
  63. Land MF, Burton FA, Meyer-Rochow VB (1979) The optical geometry of euphausiid eyes. J Comp Physiol 130:49–62Google Scholar
  64. Land MF, Gibson G, Horwood J (1997) Mosquito eye design: conical rhabdoms are matched to wide aperture lenses. Proc R Soc London Ser B 264:1183–1187CrossRefGoogle Scholar
  65. Land MF, Gibson G, Horwood J, Zeil J (1999) Fundamental differences in the optical structure of the eyes of nocturnal and diurnal mosquitoes. J Comp Physiol A 185:91–103CrossRefGoogle Scholar
  66. Laughlin SB (1981) Neural principles in the peripheral visual systems of invertebrates. In: Autrum H (ed) Handbook of sensory physiology, Vol VII/6B. Springer, Berlin Heidelberg New York, pp 133–280Google Scholar
  67. Laughlin SB (1990) Invertebrate vision at low luminances. In: Hess RF, Sharpe LT, Nordby K (eds) Night vision. Cambridge University Press, Cambridge, pp 223–250Google Scholar
  68. Laughlin SB, Blest AD, Stowe S (1980) The sensitivity of receptors in the posterior median eye of the nocturnal spider Dinopis. J Comp Physiol 141:53–65Google Scholar
  69. Leggett LMW, Stavenga DG (1981) Diurnal changes in angular sensitivity in crab photoreceptors. J Comp Physiol 144:99–109Google Scholar
  70. Lehrer M (1996) Small-scale navigation in the honeybee: active acquisition of visual information about the goal. J Exp Biol 199: 253–261PubMedGoogle Scholar
  71. Liebmann P, Entine G (1968) Visual pigments of frog and tadpole (Rana pipens). Vision Res 8:761–775CrossRefPubMedGoogle Scholar
  72. Lillywhite PG (1977) Single photon signals and transduction in an insect eye. J Comp Physiol 122:189–200Google Scholar
  73. Lillywhite PG, Laughlin SB (1979) Transducer noise in a photoreceptor. Nature 277:569–572Google Scholar
  74. Locket NA (1971) Retinal anatomy in some scopelarchid deep-sea fishes. Proc R Soc London Ser B 178:161–184PubMedGoogle Scholar
  75. Locket NA (1977) Adaptations to the deep-sea environment. In: Crescitelli F (ed) Handbook of sensory physiology, vol VII/5. Springer, Berlin Heidelberg New York, pp 67–192Google Scholar
  76. Locket NA (1985) The multiple bank rod foveae of Bajacalifornia drakei, an alepocephalid deep-sea teleost. Proc R Soc London Ser B 224:7–22Google Scholar
  77. Loew ER (1994) A third, ultraviolet-sensitive, visual pigment in the tokay gecko (Gekko gekko). Vision Res 34:1427–1431CrossRefPubMedGoogle Scholar
  78. Loew ER, Govardovskii VI, Röhlich P, Szél A (1996) Microspectrophotometric and immunocytochemical identification of ultraviolet photoreceptors in geckos. Vis Neurosci 13:247–256PubMedGoogle Scholar
  79. Losey GS, Cronin TW, Goldsmith TH, Hyde D, Marshall NJ, McFarland WN (1999) The UV visual world of fishes: a review. J Fish Biol 54:921–943CrossRefGoogle Scholar
  80. Lythgoe JN (1979) The ecology of vision. Clarendon Press, OxfordGoogle Scholar
  81. Martin GR (1994) Form and function in the optical structure of bird eyes. In: Davies MNO, Green PR (eds) Perception and motor control in birds: an ecological approach. Springer, Berlin Heidelberg New York, pp 5–34Google Scholar
  82. Martin GR, Katzir G (1999) Visual field in short-toed eagles Circaetus gallicus and the function of binocularity in birds. Brain Behav Evol 53:55–66CrossRefPubMedGoogle Scholar
  83. Martin GR, Rojas LM, Ramirez Y, McNeil R (2004) The eyes of oilbirds (Steatornis caripensis): pushing at the limits of sensitivity. Naturwissenschaften 91:26–29CrossRefPubMedGoogle Scholar
  84. McIntyre PD, Caveney S (1998) Superposition optics and the time of flight in onitine dung beetles. J Comp Physiol A 183:45–60CrossRefGoogle Scholar
  85. Moeller JF, Case JF (1994) Properties of visual interneurons in a deep-sea mysid, Gnathophausia ingens. Mar Biol 119:211–219Google Scholar
  86. Moeller JF, Case JF (1995) Temporal adaptations in visual systems of deep-sea crustaceans. Mar Biol 123:47–54Google Scholar
  87. Moon P (1940) Proposed standard solar-radiation curves for engineering use. J Franklin Inst 230:583–617CrossRefGoogle Scholar
  88. Motani R, Rothschild BM, Wahl W (1999) Large eyeballs in diving ichthyosaurs. Nature 402:747CrossRefGoogle Scholar
  89. Munk O (1980) Hvirveldyrøjet: Bygning, funktion og tilpasning. Berlingske, CopenhagenGoogle Scholar
  90. Murphy PA (1981) Celestial compass orientation in juvenile American alligators (Alligator mississippiensis). Copeia 3:638–645Google Scholar
  91. Murphy CJ, Evans HE, Howland HC (1985) Towards a schematic eye for the great horned owl. Fortschr Zool 30:703–706Google Scholar
  92. Nicol JAC (1989) The eyes of fishes. Oxford University Press, OxfordGoogle Scholar
  93. Nilsson D-E (1989) Optics and evolution of the compound eye. In: Stavenga DG, Hardie RC (eds) Facets of vision. Springer, Berlin Heidelberg New York, pp 30–73Google Scholar
  94. Nilsson D-E, Nilsson HL (1981) A crustacean compound eye adapted for low light intensities (Isopoda). J Comp Physiol 143:503–510Google Scholar
  95. Ohly KP (1975) The neurons of the first synaptic regions of the optic neuropil of the firefly, Phausius splendidula L. (Coleoptera). Cell Tissue Res 158:89–109PubMedGoogle Scholar
  96. Partridge JC, Shand J, Archer SN, Lythgoe JN, Groningen-Luyben WAHM van (1989). Interspecific variation in the visual pigments of deep-sea fishes. J Comp Physiol A 164:513–529PubMedGoogle Scholar
  97. Pasternak T, Merigan WH (1981) The luminance dependence of spatial vision in the cat. Vision Res 21:1333–1339CrossRefPubMedGoogle Scholar
  98. Pick B, Buchner E (1979) Visual movement detection under light- and dark-adaptation in the fly, Musca domestica. J Comp Physiol A 134:45–54Google Scholar
  99. Ribi WA (1977) Fine structure of the first optic ganglion (lamina) of the cockroach Periplaneta americana. Tissue Cell 9:57–72PubMedGoogle Scholar
  100. Rose A (1942) The relative sensitivities of television pickup tubes, photographic film and the human eye. Proc Inst Radio Eng New York 30:293–300Google Scholar
  101. Roth LSV, Kelber A (2004) Nocturnal coulour vision in geckos. Proc R Soc London Ser B, in pressGoogle Scholar
  102. Rozenberg GV (1966) Twilight: a study in atmospheric optics. Plenum Press, New YorkGoogle Scholar
  103. Scholes JH, Reichardt W (1969) The quantal content of optomotor stimuli and the electrical responses of receptors in the compound eye of the fly Musca. Kybernetik 6:74–80PubMedGoogle Scholar
  104. Schwemer J, Paulsen R (1973) Three visual pigments in Deilephila elpenor (Lepidoptera, Sphingidae). J Comp Physiol 86:215–229Google Scholar
  105. Sinclair S (1985) How animals see. Facts on File Publications, New YorkGoogle Scholar
  106. Snyder AW (1977) Acuity of compound eyes: physical limitations and design. J Comp Physiol 116:161–182Google Scholar
  107. Snyder AW (1979) Physics of vision in compound eyes. In: Autrum H (ed) Handbook of sensory physiology, vol VII/6A. Springer, Berlin Heidelberg New York, pp 225–313Google Scholar
  108. Snyder AW, Stavenga DG, Laughlin SB (1977a) Spatial information capacity of compound eyes. J Comp Physiol 116:183–207Google Scholar
  109. Snyder AW, Laughlin SB, Stavenga DG (1977b) Information capacity of eyes. Vision Res 17:1163–1175CrossRefPubMedGoogle Scholar
  110. Sotthibandhu S, Baker RR (1979) Celestial orientation by the large yellow underwing moth, Noctua Pronuba L. Anim Behav 27:786–800Google Scholar
  111. Srinivasan MV, Bernard GD (1975) The effect of motion on visual acuity of the compound eye: a theoretical analysis. Vision Res 15:515–525CrossRefPubMedGoogle Scholar
  112. Srinivasan MV, Dvorak DR (1980) Spatial processing of visual information in the movement-detecting pathway of the fly. J Comp Physiol 140:1–23Google Scholar
  113. Srinivasan MV, Laughlin SB, Dubs A (1982) Predictive coding: a fresh view of inhibition in the retina. Proc R Soc London Ser B 216:427–459PubMedGoogle Scholar
  114. Stavenga DG (1976) Fly visual pigments: difference in visual pigments of blowfly and dronefly peripheral retinula cells. J Comp Physiol 111:137–152Google Scholar
  115. Stavenga DG (2003) Angular and spectral sensitivity of fly receptors. II. Dependence on facet lens F-number and rhabdomere type in Drosophila. J Comp Physiol A 189:189–202Google Scholar
  116. Stavenga DG (2004a) Angular and spectral sensitivity of fly receptors. III. Dependence on the pupil mechanism in the blowfly Calliphora. J Comp Physiol A 190:115–129CrossRefGoogle Scholar
  117. Stavenga DG (2004b) Visual acuity of fly photoreceptors in natural conditions—dependence on UV sensitizing pigment and light-controlling pupil. J Exp Biol 207:1703–1713CrossRefPubMedGoogle Scholar
  118. Stavenga DG, Smits RP, Hoenders BJ (1993) Simple exponential functions describing the absorbance bands of visual pigment spectra. Vision Res 33:1011–1017CrossRefPubMedGoogle Scholar
  119. Strausfeld NJ, Blest AD (1970) Golgi studies on insects. I. The optic lobes of Lepidoptera. Philos Trans R Soc London Ser B 258:81–134Google Scholar
  120. Tyler JE, Smith RC (1970) Measurement of spectral irradiance underwater. Gordon and Breach, New YorkGoogle Scholar
  121. Wagner H-J, Fröhlich E, Negishi K, Collin SP (1998) The eyes of deep sea fish II. Functional morphology of the retina. Prog Ret Eye Res 17:637–685CrossRefGoogle Scholar
  122. Walls GL (1942) The Vertebrate eye and its adaptive radiation. The Cranbrook Press, Bloomfield HillsGoogle Scholar
  123. Warrant EJ (1999) Seeing better at night: life style, eye design and the optimum strategy of spatial and temporal summation. Vision Res 39:1611–1630CrossRefPubMedGoogle Scholar
  124. Warrant EJ (2000) The eyes of deep-sea fish and the changing nature of visual scenes with depth. Philos Trans R Soc London Ser B 355:1155–1159CrossRefGoogle Scholar
  125. Warrant EJ (2001) The design of compound eyes and the illumination of natural habitats. In: Barth FG, Schmid A (eds) Ecology of sensing. Springer, Berlin Heidelberg New York, pp 187–213Google Scholar
  126. Warrant EJ, Locket NA (2004) Vision in the deep sea. Biol Rev 79:671–712CrossRefPubMedGoogle Scholar
  127. Warrant EJ, McIntyre PD (1990) Limitations to resolution in superposition eyes. J Comp Physiol A 167:785–803Google Scholar
  128. Warrant EJ, McIntyre PD (1991) Strategies for retinal design in arthropod eyes of low F-number. J Comp Physiol A 168:499–512Google Scholar
  129. Warrant EJ, McIntyre PD (1992) The trade-off between resolution and sensitivity in compound eyes. In: Pinter RB, Nabet B (eds) Nonlinear vision: determination of neural receptive fields, function, and networks. CRC Press, Boca Raton, pp 391–421Google Scholar
  130. Warrant EJ, McIntyre PD (1993) Arthropod eye design and the physical limits to spatial resolving power. Prog Neurobiol 40:413–461CrossRefPubMedGoogle Scholar
  131. Warrant EJ, Nilsson D-E (1998) Absorption of white light in photoreceptors. Vision Res 38:195–207CrossRefPubMedGoogle Scholar
  132. Warrant EJ, Porombka T, Kirchner WH (1996) Neural image enhancement allows honeybees to see at night. Proc R Soc London Ser B 263:1521–1526Google Scholar
  133. Warrant EJ, Collin SP, Locket NA (2003) Eye design and vision in deep-sea fishes. In: Collin SP, Marshall NJ (eds) Sensory processing of the aquatic environment. Springer, Berlin Heidelberg New York, pp 303–322Google Scholar
  134. Warrant EJ, Kelber A, Gislén A, Greiner B, Ribi W, Wcislo WT (2004) Nocturnal vision and landmark orientation in a tropical halictid bee. Curr Biol 14:1309–1318CrossRefPubMedGoogle Scholar
  135. Waterman TH (1954) Polarisation patterns in submarine illumination. Science 120:927–932Google Scholar
  136. Waterman TH (1981) Polarization sensitivity. In: Autrum H (ed) Handbook of sensory physiology, vol VII 6B. Springer, Berlin Heidelberg New York, pp 281–469Google Scholar
  137. Wcislo WT, Arneson L, Roesch K, Gonzalez V, Smith A, Fernandez H (2004) The evolution of nocturnal behaviour in sweat bees, Megalopta genalis and M. ecuadoria (Hymenoptera: Halictidae): an escape from competitors and enemies? Biol J Linnean Soc (in press)Google Scholar
  138. Wehner R (1981) Spatial vision in arthropods. In: Autrum H (ed) Handbook of sensory physiology, vol VII 6C. Springer, Berlin Heidelberg New York, pp 287–616Google Scholar
  139. Wehner R (2001) Polarization vision—a uniform sensory capacity? J Exp Biol 204:2589–2596PubMedGoogle Scholar
  140. Wikler KC, Rakic P (1990) Distribution of photoreceptor subtypes in the retina of diurnal and nocturnal primates. J Neurosci 10:3390–3401Google Scholar
  141. Williams DS (1982) Ommatidial structure in relation to turnover of photoreceptor membrane in the locust. Cell Tissue Res 225:595–617PubMedGoogle Scholar
  142. Williams DS (1983) Changes of photoreceptor performance associated with the daily turnover of photoreceptor membrane in the locust. J Comp Physiol 150:509–519Google Scholar
  143. Wiltschko W, Daum P, Fergenbauer-Kimmel A, Wiltschko R (1987) The development of the star compass in Garden Warblers, Sylvia borin. Ethology 74:285–292Google Scholar
  144. Yeandle S (1958) Evidence of quantized slow potentials in the eye of Limulus. Am J Ophthalmol 46:82–87Google Scholar
  145. Zeil J, Kelber A, Voss R (1996) Structure and function of learning flights in bees and wasps. J Exp Biol 199:245–252PubMedGoogle Scholar

Copyright information

© Springer-Verlag 2004

Authors and Affiliations

  1. 1.Vision Group, Department of Cell and Organism BiologyUniversity of LundLundSweden

Personalised recommendations