Summary and Keywords
The predatory stomatopod crustaceans, or mantis shrimp, are among the most attractive and dynamic creatures living in the sea. Their special features include their powerful raptorial appendages, used to kill, stun, or disable other animals (whether predators, prey, or competitors), and their highly specialized compound eyes. Mantis shrimp vision is unlike that of any other animal and has several unique features. Their compound eyes are optically triple, each having three separate regions that produce overlapping visual fields viewing certain regions of space. They have the most diverse set of spectral classes of receptors ever described in animals, with as many as 16 types in a single compound eye. These receptors are based on a highly duplicated set of opsin molecules paired with strongly absorbing photostable filters in some photoreceptor types. The receptor set includes six ultraviolet types, all spectrally distinct, many themselves tuned by photostable filters. There are as many as eight types of polarization receptors of up to three spectral classes (including an ultraviolet class). In some species, two sets of these receptors analyze circularly polarized light, another unique capability. Stomatopod eyes move independently, each capable of visual field stabilization, image foveation and tracking, or scanning of image features. Stomatopods are known to recognize colors and polarization features and evidently use these in predation and communication. Altogether, mantis shrimps have perhaps the most unusual vision of any animal.
Stomatopod crustaceans, commonly known as mantis shrimps, were familiar only to marine invertebrate researchers until recently. They are generally small animals (typically, a few centimeters in length, although a very few species can approach 0.5 m), all species inhabit seawater, and most are limited to the tropics (Schram et al., 2013). But many species are colorful, and nearly all have charismatic, active, and aggressive behavior, which involves a nasty spear or a ballistic strike that can—on occasion—break the glass walls of aquaria. This ability, coupled with their extremely unusual eyes, has brought them some fame in recent years, so that now they are familiar to many nonscientists.
Despite their common name, mantis shrimp are only distantly related to modern shrimp, crabs, or lobsters, having diverged from the main line of crustacean evolution some 400 million years ago and continuing on their own path ever since. Stomatopods have stalked compound eyes that are in basic design similar to those of many other crustaceans and insects. Compound eyes are constructed of multiple units (ommatidia), each ommatidium containing optical components and a light-sensitive structure called the rhabdom. In stomatopods, compound eyes are of the apposition optical design, meaning that each ommatidium is optically isolated from its neighbors. Its receptive field is a single unit, or pixel, sampling light from a small spot in space. The overall visual field of each eye and its spatial coverage are thus determined by the number and arrangement of ommatidia that form the entire compound eye (see Horridge, 1978; Land & Nilsson, 2012; Marshall & Land, 1993a, 1993b; Marshall et al., 2007).
In other animals with apposition compound eyes, the receptive field of each ommatidium is unique, sampling a visual pixel that is not shared with any other ommatidium. Uniquely, ommatidia in stomatopod eyes are laid out such that units in three separate parts of the field sample identical spatial locations. This is possible because the eye is triple in structure, being formed of three ommatidial groups: dorsal and ventral halves, known also as the hemispheres or the peripheral ommatidia, separated by a midband region formed by two to six parallel ommatidial rows (Cronin, 1986; Exner, 1891; Harling, 2000; Horridge, 1978; Manning et al., 1984a; Marshall, 1988; Marshall & Land, 1993a, 1993b; see Figure 1). Altogether, the three groups sample a nearly spherical view of space, but the dorsal half’s view extends down to just below the eye’s equator, the ventral half’s to just above this equator, and the midband inspects a narrow strip of space extending through the region of overlap of the hemispheres (Exner, 1891; Marshall, 1988; Marshall & Land, 1993a, 1993b; Schiff, 1996). Thus, when a stomatopod eye is examined or photographed from within the midband’s field of view, three pseudopupils (the dark spots that indicate regions directed at the viewer or camera) can be seen (Figure 1A-C).
It is thought that the shared views of hemispheres are used for depth ranging, perhaps to direct the raptorial strikes. The midband probably does not contribute to depth perception, but its overlapping field of view gives its ommatidia free reign to evolve high-level analyses of stimuli in their receptive fields. Most likely, the presence of the midband is the design feature that fostered the complexity of stomatopod vision, and in modern mantis shrimps its comparatively limited numbers of ommatidia contain all of the diverse color and ultraviolet (UV) receptors of the eye as well as specialized polarization–analysis receptors for linearly or circularly polarized light. In almost all stomatopod species, each compound eye has a region where ommatidial axes are separated by very small angles, which gives far more detail in the image seen by the eye as a whole (Horridge, 1978; Marshall & Land, 1993a). Such a region is called an acute zone, which is the compound eye equivalent of a vertebrate fovea. The acute zones of all three visual regions are aligned to sample a single spatial region, allowing both high-quality imaging and (probably) also improved analysis of color and polarization. Because of its improved resolution, the acute zone is thought to be important in depth ranging as well, and its ommatidia probably direct the predatory strikes.
Stomatopod Eye Design
There are several rather distinctive eye organizations among the Stomatopoda, but with one exception all involve two peripheral ommatidial arrays separated by a midband of two to six ommatidial rows, as described earlier. The exception is the eyes of a superfamily of deep-sea stomatopods known as the Bathysquilloidea (Harling, 2000); these animals have elongated eyes that lack either hemispheres or a midband, instead being built like those of other crustaceans, such as crabs, with apposition compound eyes. The taxonomic status of bathysquilloids is not established using molecular genetic techniques, but traditional taxonomic approaches suggest that these animals are derived from species with six-row midbands (Ahyong & Harling, 2000). In fact, the best available molecular phylogeny finds that the ancestors of all modern stomatopods had six ommatidial rows in the midband (Porter et al., 2010). The only sizable group of modern stomatopods with fewer rows is composed of members of the superfamily Squilloidea, all of which have two-row midbands (Harling, 2000; Manning et al., 1984a, 1984b; Marshall et al., 1991a, 1991b; Marshall & Land, 1993a, 1993b). Like the bathysquilloids, squilloids appear to have been derived from six-row ancestors (Ahyong & Harling, 2000; Porter et al., 2010).
Within the basic design, there are many variants (Figure 1). Species placed in different superfamilies tend to favor different eye plans, but there is no strict association between the overall eye shape and taxonomic placement. Many species, particularly members of the superfamily Gonodactyloidea, have ovoid, egg-shaped eyes in which the midband may be wrapped around the middle perpendicular to the egg’s long axis or encircle the short axis, giving the eye the appearance of a squashed globe. If such an eye is nearly spherical, the midband forms an equator (Figure 1A). On the other hand, the entire eye may be vertically elongated to become a tall ovoid or even bean-shaped structure, with the midband running around the midpoint of the vertical axis. These eyes are favored by many Squilloids or Lysiosquilloids (Figure 1B and 1C). Whatever their shape, all mantis shrimp eyes are mounted on movable stalks, with the spherical or ovoid eyes generally more mobile than the tall eye types. Stomatopod eye movements are discussed in more detail in a later section.
Ommatidial Organization in Stomatopod Eyes: The Hemispheres
Ommatidia in stomatopod compound eyes vary in their construction depending on their locations within the regions of the eye (Figure 2A). With minor variations, ommatidia that make up the hemispheres are built on the same fundamental plan across the stomatopods (Harling, 2000; Marshall et al., 1991a), being basically similar to those of other malacostracan crustaceans. The cornea, usually hexagonal in shape, forms the outer surface of each ommatidium, followed by a crystalline cone and then by the photosensitive rhabdom, built of microvilli contributed by eight (rarely seven, in some squilloids) retinular cells (Marshall et al., 1991a; Schönenberger, 1977). When present, the four-lobed eighth cell (R8) sits atop the retinular cell array at the junction with the crystalline cone (Figure 2A). Its microvilli are arranged into a rather short, square rhabdom and have mutually orthogonal axes, making the array as a whole polarization-insensitive (Cronin et al., 1994a; Marshall, 1988; Marshall et al., 1991a; Waterman, 1981). Retinular cells 1 to 7 (R1–R7) jointly form a long, square-sided rhabdom. Three of these cells (R1, R4, and R5) have mutually parallel microvilli that form layers intercalated between layers formed from the other four cells (R2, R3, R6, and R7), whose microvilli are aligned orthogonal to those of the three-cell layers. The entire array forms a two-axis linear polarization analyzer and is very similar to rhabdoms of many other crustaceans (Marshall et al., 1991a; Waterman, 1981). Rhabdoms in each hemisphere of stomatopod species with six-row midbands are rotated 45° with respect to those in the other hemisphere, potentially enhancing polarization analysis (Marshall et al., 1991a; Daly et al., 2016).
Ommatidial Organization in Stomatopod Eyes: The Midband
Compound eyes of squilloid stomatopods have only two midband rows. Retinular cells of ommatidia in their hemispheres have microvilli that extend parallel or perpendicular to the plane of the midband, and the two rows of midband ommatidia are essentially identical to those of the hemispheres. Subtle differences in ommatidial anatomy indicate that the dorsal half of the eye is mirror-symmetrical to the ventral half (Marshall et al., 1991a), while details of midband neuroanatomy suggest that the two midband rows are derived from ommatidia in the ventral eye half. Overall, compound eyes of squilloids are remarkably simplified compared to those of all other species (besides bathysquilloids). They have small or degenerate R8 cells (Marshall et al., 1991a; Schönenberger, 1977), all ommatidia are similar in construction, axes of microvilli of all ommatidia lie in the same two planes, and there is a simple dorsal/ventral symmetry. Given all this, the function of the midband is not obvious, as it appears to contribute little or nothing to visual analysis (although this might not apply to species of squilloids that have not yet been described).
Six-row midbands are far more complicated than this (Figure 2A). The two most ventral rows (Rows 5 and 6) appear to be derived from the two midband rows of squilloids but have R8 cells that are far more specialized. With the exceptions of the R8 cells, their rhabdoms are generally similar to those of the hemispheres. Their microvilli are oriented at ±45° to the plane of the midband, but the entire ommatidium in Row 6 is rotated 90° compared to Row 5. This rotation is most clearly seen in the shapes of the rhabdoms and orientations of microvillar axes in the R8 cells. Unlike all the R8 cells of all other ommatidia, R8s of Rows 5 and 6 produce unidirectional microvilli that extend across the short axes of rhabdoms which have ovoid cross sections. Microvilli of Row 5 are parallel to the plane of the midband, and Row 6’s extend perpendicular to this plane (Marshall, 1988; Marshall et al., 1991a). A later section will consider how this contributes to polarization sensitivity in the eye.
The dorsal four midband rows (Rows 1 to 4) are devoted to color vision. Their ommatidia incorporate some of the most exotic specializations ever found in photoreceptors in either invertebrates or vertebrates, including the division of a single rhabdom into several tiers, scrambling of microvilli, incorporation of intrarhabdomal filters, and formation of 12 different photoreceptor classes with sensitivities ranging from the deep UV to the far red (Cronin & Marshall, 1989a, 1989b, 2004; Cronin et al., 1993, 1994b, 1994c, 1994d, 2014; Marshall, 1988; Marshall & Oberwinkler, 1999; Marshall et al., 1991a, 1991b, 1994). The overall organization of this part of the midband was first described by Marshall (1988), who discovered both the tiering of receptors and the presence of the filters. Further research showed that the photoreceptive tiers are created by dividing the rhabdom into three successive photosensitive regions: the R8 level, the distal tier of the main rhabdom, and the proximal tier of the main rhabdom (Marshall et al., 1991a, 1991b). To create the distal and proximal tiers, the microvilli of the seven main retinular cells (R1–R7) are separated into a group of three (R1, R4, and R5) and a group of four (R2, R3, R6, and R7). These are in fact the same sets that form the polarization-sensitive interleaved layers of all other main rhabdoms, which in a sense preadapts them to work together in photoreception.
Rows 1 to 4 all have three photoreceptive tiers, but there are also photostable, intrarhabdomal filters between tiers in Rows 2 and 3 (Figure 2A). Depending on the taxonomic assignment of the species, these rows may have a single filter between the R8 tier and the distal R1-7 tier, or two filters, with the second one placed between the distal and proximal R1-7 tiers (Cronin et al., 1994b; Marshall, 1988; Marshall et al., 1991a, 1991b, 2007; Porter et al., 2010). The spectral properties of each type of filter also vary taxonomically, with filters in Row 2 being yellow, orange, or red (the filter below the R8 tier is always yellow, however), and those of Row 3 being pink, purple, red, or blue (Cronin et al., 1994b; Marshall et al., 1991b; Porter et al., 2010). Also, filters in individual animals can alter their spectral properties when the animal moves to a different spectral environment (Cheroske & Cronin, 2005; Cheroske et al., 2006; Cronin et al., 2001, 2014). The diversity of filter pigments used by mantis shrimps, together with their ability to change spectral properties when required, makes the system of intrarhabdomal filters a remarkably flexible one for adapting vision to the photic environment (Cronin et al., 1994b, 1994c, 2000, 2001, 2002, 2014). Also, both the filters and the visual pigments (described next) vary with habitat, allowing the complicated and spectrally broad color-analysis system to function in a range of environmental lighting conditions.
Stomatopod Photoreceptors and Visual Pigments
The photoreceptors of stomatopod eyes were first characterized by microspectrophotometry. Squilloids were found to have a single receptor class (Cronin, 1985; Cronin et al., 1993), but microspectrophotometric research on gonodactyloids and lysiosquilloids identified up to 12 spectral types of photoreceptors: two in the hemispheres, nine more in the four-tiered rows of the midband, and one more in the two ventral midband rows (Bok et al., 2014, 2015; Cronin & Caldwell, 2002; Cronin & Marshall, 1989a, 1989b; Cronin et al., 1993, 1994c, 1994d, 1996). Spectral modeling (based on photoreceptor absorption together with properties of the various photostable filters), combined with electrophysiological measurements, established the presence of 15 or 16 spectral sensitivity classes of photoreceptors (Figure 2A), five or six in the UV, and ten in the human-visible range (Bok et al., 2014; Cronin & Marshall, 2004; Marshall & Oberwinkler, 1999; Marshall et al., 2007). This spectral diversity exceeds the abilities of any other described visual system by a factor of at least three! The UV-sensitive system alone outperforms the entire spectral array of any other known animal (Bok et al., 2014, 2015; Marshall & Oberwinkler, 1999). The pattern of spectral classes is constant across stomatopod species with six-row midbands. For shortest-wavelength sensitive to longest, the four rows range in spectral order from Row 1 (which is violet/blue sensitive) through blue/bluegreen sensitivities in Row 4, yellow/orange in Row 2, and finally orange/red in Row 3 (Figure 2A).
As with other animals, stomatopod visual pigments are based on a retinal1 chromophore (Goldsmith & Cronin, 1993) linked to various opsin proteins. Given the known spectral properties of stomatopod photoreceptors, 12 opsins are expected. Thus, it has been a surprise to find that molecular genetic approaches have so far identified up to 33 opsins in a single species, and many more are probably present (Bok et al., 2014; Cronin et al., 2010; Porter et al., 2009, 2013). Even Squilla empusa, a squilloid species with a single spectral receptor class, expresses at least six different transcripts. At present, how this diversity of potential visual pigments contributes to the more limited number of spectral receptor types is not known. In contrast, only two opsins have been found in stomatopod UV receptors, where up to six spectral types are known (the difference is due to spectral filtering by mycosporine amino acids in the crystalline cones), so here at least the connection between visual pigments and receptor properties seems clear and relatively simple (Bok et al., 2014).
Color Vision in Stomatopod Crustaceans
Given the spectacular coloration of some species, the use of species-specific colored signals (Caldwell & Dingle, 1975, 1976), and the diversity of spectral receptor types in mantis shrimp retinas, it is no surprise that they are among the few marine invertebrates known to possess true color vision. Exhaustive behavioral tests show that all species tested (all with six-row midbands) learn the color of a light or pigmentary stimulus associated with a food reward and discriminate this color from similarly bright shades of gray (Marshall et al., 1996; Thoen et al., 2014). This appears to involve the four-tiered rows of main rhabdoms (Rows 1 to 4) in the dorsal midband (Marshall et al., 2014). Since photoreceptors here are separated into three-retinular-cell and four-retinular-cell layers, which is the same division that forms the layered rhabdoms of polarization-sensitive rhabdoms (Waterman, 1981), it seems likely that the processing of color involves a form of opponency between the two-tiered components of single ommatidia (see the later section on neuroanatomy). Surprisingly, however, color discrimination, as revealed by behavior that requires learning the color of a box concealing a food reward, is far worse than predicted by a simple opponent model. In initial experiments, for example, the animals did not discriminate blue from gray (Marshall et al., 1996). Indeed, further research demonstrated that stomatopod color discrimination is among the poorest of the animals that have been measured (Thoen et al., 2014). It is thus possible that color is processed differently in the peripheral nervous system than expected or that color is sorted into a relatively small number of categories of recognition in the central nervous system. This could provide a quick classification of color to permit rapid decision making during predation or aggressive behavior. Another possibility is that color vision is not primarily used in food choice but is more important for intraspecific signaling (Caldwell & Dingle, 1975, 1976; Chiao et al., 2000) or some other task. Besides the ability of mantis shrimps to choose among colors visible to humans, these animals are likely to have color vision in the UV as well, since they have up to six spectral classes of UV-sensitive photoreceptors (Bok et al., 2014). Whether they actually discriminate UV “colors” is unknown.
Using vision for signaling is problematical for marine animals that are as mobile and versatile as mantis shrimps because single species often inhabit a range of depths; in marine waters, light quality varies rapidly with depth and time of day. Furthermore, as already noted, the filters in photoreceptors of some species vary in their spectral absorption of light depending on the quality of light viewed by the eye, further compromising color constancy and signal recognition. Thus, in some species, mantis shrimps that live at different depths vary in their body pigmentation, and in some cases this could favor signal constancy in each individual photic environment (Cheroske & Cronin, 2005).
Light is made up of photons, each of which has an electromagnetic field that varies in the tilt angle (called the angle of polarization) of a single plane. Most natural sources of light emit photons with all possible polarization orientations, so the overall illumination is depolarized—as a whole, there is no favored angle of polarization. But polarized light is nevertheless abundant on earth because scattering of light (as in air or water) or reflection from most shiny surfaces (e.g., water, waxy leaves) converts depolarized light into partially polarized light (in other words, most photons have angles of polarization that are nearly parallel; see Cronin et al., 2014; Marshall & Cronin, 2011; Wehner, 2001).
Photoreceptors with aligned microvilli, like those of stomatopods, respond most strongly when the angle of polarization is parallel to the microvilli. The main rhabdoms of all photoreceptors (e.g., those below the level of the R8 cell in the hemispheres) are thus polarization sensitive. Usually, the rhabdoms of the ventral hemisphere are rotated by 45º relative to those of the dorsal hemisphere (Marshall et al., 1991a). Since each rhabdom has two sets of mutually orthogonal microvilli, the two hemispheres respond best to polarization at four angles separated by 45º. The fields of view of the hemispheres overlap near the equator of the visual field as a whole, which in principle allows the animal to determine the exact angle and degree of polarization in any visual location sampled by ommatidia in the two hemispheres. This analysis probably involves rotations of the eyes to align some microvilli with polarization features in the visual world, as described later (Daly et al., 2016).
R8 cells throughout the hemispheres and in the four most dorsal rows of the midband are not sensitive to the properties of polarized light because each cell in these locations has two mutually perpendicular sets of microvilli and is thus polarization-insensitive (Cronin et al., 1994a; Marshall, 1988; Marshall et al., 1991a; Kleinlogel & Marshall, 2006). Rhabdoms of R8 cells of the two most ventral midband rows are unique, having unidirectional microvilli throughout their full length, giving them strong responses to light polarized parallel to the midband’s plane in the fifth row and perpendicular to it in the sixth (Cronin et al., 1994a; Marshall et al., 1991a; Kleinlogel & Marshall, 2009).
Retinular cells that form the distal tier of the second midband row are also polarization sensitive (Kleinlogel & Marshall, 2006), in this case in the yellow spectral region, but it is not obvious that this has a physiological or behavioral significance. Main rhabdoms of the ventral two rows of the midband have a similar organization, but since here the axonal wiring is like that of the hemispheres, they have orthogonal polarization sensitivity. These photoreceptors generally have a different spectral sensitivity compared to the hemispheric main rhabdoms, adding yet another spectral receptor class to the polarization system. However, their response to polarized light is affected by the overlying R8 cell microvilli. These are birefringent, acting as quarter-wave phase delay plates in some species (Chiou et al., 2008; Roberts et al., 2009) and thus converting circularly polarized light, in which the polarization angle rotates by a full 360º for each wavelength of light, to linearly polarized light (the type discussed to this point). Main rhabdoms (i.e., R1-7), of course, respond to linearly polarized light, and the conversion from circular to linear by the R8s makes them effectively responsive to circularly polarized light in the external world (Chiou et al., 2008). Mantis shrimps are the only animals known to detect circularly polarized light, which is rare in nature (Cronin & Marshall, 2011, 2014; Gagnon et al., 2015; Marshall& Cronin, 2014; Marshall et al., 2014).
As with color vision, stomatopods can be trained to choose among potential food-containing objects by their polarization properties alone. If containers are marked by attaching a fixed polarizer, the animals learn which angle of polarization corresponds to a reward (Marshall et al., 1999). More impressively, they also learn to discriminate targets marked with a right-handed circular polarizer (producing right circular polarization) from one that produces left circular polarization (Chiou et al., 2008; Gagnon et al., 2015). In fact, they are the only animals known to be capable of making this discrimination. Similarly to their color patterns, mantis shrimps are marked with patterns that reflect linearly or circularly polarized light, so it appears that one function of their polarization vision is to recognize conspecific signals (Chiou et al., 2008; Cronin et al., 2003, 2014; Gagnon et al., 2015; Marshall et al., 1999, 2014). For example, Haptosquilla is a genus whose species commonly have mouthparts (first maxillipeds) with structures that strongly scatter horizontally polarized light, and these structures are thought to be important for intraspecific signaling (Chiou et al., 2011; How et al., 2014b; Jordan et al., 2016). It appears that a preexisting bias in the functional properties of their polarization-sensitive receptor sets has fostered the use of horizontal polarization as a signal (How et al., 2014a, 2014b). Other proposed (but unproven) uses include navigation and orientation, prey detection, and contrast enhancement in the underwater environment.
Eye Movements of Stomatopods
As in many crustaceans, compound eyes in all adult stomatopods are mounted on stalks (the actual eye is called the eyecup, while the extension that connects to the anterior end of the rostrum is the eyestalk). The linkage includes a flexible membrane at the base of the stalk, giving the eyes remarkable freedom to move, being driven by six sets of oculomotor muscles that swing them on the three primary axes: pitch, yaw, and roll (Jones, 1994). The degree of actual mobility varies with species. Squilloids generally move their eyes very little, while hemisquilloids and lysiosquilloids have an intermediate degree of motion (using rolls more than pitch or yaw). Gonodactyloids, odontodactylids, and pseudosquilloids, in contrast, have extremely mobile and active eyes, with the freedom to rotate nearly 180º in yaw and pitch and at least 90º in roll.
The mobility also varies with eye type, being least in the tall eyes (e.g., Figure 1B and 1C). In members of the genus Odontodactylus, which have nearly spherical eyecups (Figure 1A), the cuticular extension that anchors the membrane to which the eyestalks are attached is near the center of the ommatidial array, so that the eyes move nearly completely in rotation on the primary axes (Jones, 1994; Land et al., 1990)—much like vertebrate eyes do. In most other gonodactyloids and in pseudosquilloids, the stalk is long, extending the eyecup out from the rostrum and thus producing large translations during pitch and yaw movements. These could be important for inducing motion parallax, but more likely the advantage is that it allows the animal to expose its eyes to survey its environment while being mostly protected within the burrow.
Eye movements in stomatopods are essentially independent, with each eye’s activity statistically unrelated to that of the other (Cronin et al., 1988, 1991; Land et al., 1990). The eyes may work together for some visual tasks, but this does not rule out the possibility of independent control systems, much as the eyes of individual spectators at Wimbledon move in concert with those of others when following the path of a tennis ball. Both smooth and saccadic movements are used, with the smooth movements occurring mainly during scans, while most other movements are saccadic (Cronin et al., 1988, 1991; Land et al., 1990; Marshall et al., 2014).
In gonodactyloids and pseudosquilloids, there are five major classes of movements: optokinesis, spontaneous saccades, targeting saccades, object tracking (either smooth or saccadic), and scans. All of these movements are probably driven by input from ommatidia in the hemispheres (Cronin et al., 1992). Atypically for crustaceans, optokinesis (the following of movements of the whole visual field, usually evoked in the laboratory by a pattern of moving bars surrounding the animal) is very irregular. It shows a weak expression of nystagmus (the characteristic sawtooth pattern in which a slow drift in the direction of the bars’ movement is interrupted by rapid returns to the starting point (Cronin et al., 1991), but overall, optokinesis starts and stops irregularly. Since mantis shrimps are bottom-living animals and move their eyes so irregularly and intermittently, visual field stabilization does not seem to be very important to them.
If a visual target is suddenly presented, gonodactyloids, pseudosquilloids, and lysiosquilloids often make very rapid saccades, up to several hundred degrees s−1, and fixate the acute zones of one eye, and commonly both, on the target’s location (Marshall et al., 2014). In the natural environment, such acquisitional saccades may suddenly shift visual attention from one object to another, giving a strong impression of alertness. Indeed, stomatopods are the only crustaceans known to make such movements, adding to their complex visual repertoire. Any object moving within the visual range of the eyes almost always encourages this type of saccade, in both the laboratory and the field (Cronin et al., 1988; Marshall et al., 2014). Acquisitional saccades are frequently followed by monocular or binocular tracking movements that serve to keep the eye fixated on a moving target, or by scanning movements, described later.
As with their other eye movements, stomatopod tracking movements are highly idiosyncratic, starting and stopping intermittently, and often switching between eyes in a single tracking session (Cronin et al., 1988). Most tracking responses are saccadic, as if the animal continuously refixates the position of the target, using one eye, the other, or both; it is unknown whether stomatopods make predictive tracking movements. Tracking is driven by input from hemispheric ommatidia (Cronin et al., 1992) and is most accurate when the tracked object is directly in front of the animal (Cronin et al., 1988).
The final class of stomatopod visual movements is composed of visual scans. These are necessary because stomatopod vision is faced with a strange compromise: the hemispheres have extended visual fields and can detect contrast, shape, and motion over a very extended area, but the midband samples essentially a planar slice of the visual field shared with some of ommatida of both hemispheres, no more than a few degrees across. Interommatidial angles vary across the eye to form the acute zones, but in the midband this variation leads only to improved resolution in the midband’s plane. Thus, to analyze color (and to some extent polarization) over a spatial region, the eye is obligated to move across (to scan) the object of interest. In doing this, the eye must move with an angular velocity that is slow enough to allow each point to be neurally registered within the overall image. All this is handled by the scans, which are slow movements (~40º s−1) of limited angular extent (~10 to 20º) made perpendicular to the midband’s plane (Land et al., 1990). In many gonodactyloid stomatopods, the favored resting orientation of each eye is at a rotation of about 30º from the horizontal plane, so most scans are directed upward and medially or downward and laterally. Very few types of animals make visual scans (Land & Nilsson, 2012) because they compromise spatial vision and interrupt other visual tasks such as tracking or fixation movements. In stomatopods, they are necessary because of the unique functional architecture of the eye, where extended spatial vision is separated from color and, to some extent, polarization vision.
To make best use of the complex polarization vision in their eyes, whether the eye is fixated on a target, tracking it, or scanning it, the eyes are best rotated to an angle at which polarization analysis occurs efficiently. Thus, stomatopod eyes rotate to specific angles when faced with a polarized object or a threatening polarization pattern (Daly et al., 2016). These rotations orient microvilli of one polarization-sensitive receptor group in one hemisphere parallel to the e-vector axis of the stimulus and of the other set of polarization receptors in the same hemisphere perpendicular to it. Since rhabdoms in one hemisphere are rotated by 45º to rhabdoms of the other one, this should maximize polarization contrast in ommatidia of one hemisphere’s view and minimize contrast in the other, presumably improving the quality and speed of interpreting the object’s polarization properties.
Neurobiology of Stomatopod Vision
The organization of the optic lobes of stomatopods follows the basic arthropod plan, with four successive layers of neuropil located within the eyestalk itself: the lamina, medulla, lobula, and lobula plate or lobula complex (Strausfeld & Nässel, 1981; Marshall & Cronin, 2014; Marshall et al., 2007; see Figure 2B). The general organization of these visual centers and their connections is fairly well understood, but the physiology and function of specific neurons and the way in which information flows to the central brain have yet to be worked out.
In general, the wiring from the retina at least to the level of the lobula is retinotopic. Neurons from the photoreceptors can be traced to the lamina and/or the medulla, and the connections between the medulla and lobula are also mapped back to the retina (reviewed in Marshall & Cronin, 2014; Marshall et al., 2007). As is common in arthropods, chiasmata exist between the lamina and the medulla and again between the medulla and lobula, but this is not thought to disrupt the retinotopic mapping. One very unusual feature of stomatopod neuroarchitecture is that the midband ommatidia project to specialized, expanded regions of the first three neuropils, called accessory lobes, within which the identity of the individual midband rows is maintained through the flow of retinal information (Kleinlogel et al., 2003; Marshall et al., 2007). Currently, the connections that permit exchange of information between the various midband rows, or between the midband and the hemispheres, and even between the dorsal and ventral hemisphere are not obvious. While further anatomical studies may clarify this, dye labeling of specific pathways and electrophysiological recordings from known neurons will be necessary to understand how the complex, spatially redundant information from the three retinal regions is brought together for higher analysis.
Even with the lack of direct physiological evidence, the general organization of retinal projections suggests an overall plan of function (Marshall et al., 2007). Axons of photoreceptors of single ommatidia project to individual cartridges in the lamina. Axons of main rhabdom photoreceptors (R1–R7) terminate here, while R8 axons continue through the cartridge to the medulla before terminating. Cartridges of the first four midband rows, those thought to foster color perception, are circular in shape, while all others are rectangular. As is not uncommon in crustaceans, axons of retinular cells R1, 4, and 5 terminate within a distal plexiform layer in the lamina, while those of R2, 3, 6, and 7 terminate slightly deeper in the proximal plexiform layer (Kleinlogel et al., 2003). In the hemispheres and the two ventral midband rows, these cell groups correspond to those forming microvilli in successive, orthogonal layers of the rhabdom, suggesting that the wiring fosters polarization analysis (most likely via opponency). Similarly, the same cell identities segregate into the proximal and distal tiers of rhabdoms of the four chromatic rows of the midband. These have spectrally distinct sensitivities, which indicates that processing of chromatic and polarizational information is handled by similar neuronal mechanisms.
The R8 axons follow a different pathway. In the hemispheres and in the dorsal four rows of the midband, they send out lateral neurites into the laminar cartridges. This suggests that they interact in some way with inputs from R1–7 cells before continuing to their final terminations in the medulla. R8 axons of the two ventral rows (i.e., Rows 5 and 6) pass through the cartridges without forming synapses. These axons lead from the only polarization-sensitive UV cells in the retina, so bypassing the lamina could be a way to preserve polarization integrity at middle wavelengths in those cartridges segregating UV linear polarization sensitivity from middle-wavelength linear or circular polarization analysis. Presumably, their terminations in the medulla enable analysis of UV polarized light, but the mechanisms involved are unknown.
In recent decades, many surprising features of the visual systems of stomatopod crustaceans have emerged. As noted at the outset, stomatopod vision is immensely complex, with many photoreceptor classes based on an extraordinary opsin diversity, unusual ocular anatomy, exceptional polarization vision in both linear and circular polarization, distinctive eye movements, and a very complex visual neuroanatomy. Given the pugnacious nature of stomatopods, it seems likely that these specializations evolved in parallel with the ability to identify and strike predators, prey, or competitors accurately and nearly instantaneously, which requires an advanced sense of space, shape, and coloration. The specializations vary among species and among habitats. Few of these features are well understood, and no doubt other oddities in stomatopod vision are yet to be revealed. While being in many ways like other large crustaceans, these fascinating animals nevertheless are outstanding exemplars of the creative powers of evolution.
Ahyong, S. T., & Harling, C. (2000). The phylogeny of the stomatopod Crustacea. Australian Journal of Zoology, 48, 607–642.Find this resource:
Bok, M. J., Porter, M. L., & Cronin, T. W. (2015). Diversity, ecology, and evolution of ultraviolet filters in stomatopod crustaceans. Journal of Experimental Biology, 218, 2055–2066.Find this resource:
Bok, M. J., Porter, M. L., Place, A., & Cronin, T. W. (2014). Biological sunscreens tune polychromatic ultraviolet vision in mantis shrimp. Current Biology, 24, 1636–1642.Find this resource:
Caldwell, R. L., & Dingle, H. (1975). Ecology and evolution of agonistic behavior in stomatopods. Naturwissenschaften, 62, 214–222.Find this resource:
Caldwell, R. L., & Dingle, H. (1976). Stomatopods. Scientific American, 234(1), 80–89.Find this resource:
Cheroske, A. G., Barber, P. H., & Cronin, T. W. (2006). Evolutionary variation in the expression of phenotypically plastic color vision in Caribbean mantis shrimps, genus Neogonodactylus. Marine Biology, 150, 213–220.Find this resource:
Cheroske, A. G., & Cronin, T. W. (2005). Variation in stomatopod (Gonodactylus smithii) color signal design associated with organismal condition and depth. Brain, Behavior and Evolution, 66, 99–113.Find this resource:
Chiao, C. C., Cronin, T. W., & Marshall, J. (2000). Eye design and color signaling in a stomatopod crustacean, Gonodactylus smithii. Brain, Behavior and Evolution, 56, 107–122.Find this resource:
Chiou, T.-H., Kleinlogel, S., Cronin, T. W., Caldwell, R. L., Lofle, B., Siddiqi, A., Goldizen, A., & Marshall, J. (2008). Circular polarisation vision in a stomatopod crustacean. Current Biology, 18, 429–434.Find this resource:
Chiou, T.-H., Marshall, N. J., Caldwell, R. L., & Cronin, T. W. (2011). Changes in light reflecting properties of signaling appendages alter mate choice behaviour in a stomatopod crustacean, Haptosquilla trispinosa. Marine and Freshwater Physiology and Behaviour, 44, 1–11.Find this resource:
Cronin, T. W. (1985). The visual pigment of a stomatopod crustacean, Squilla empusa. Journal of Comparative Physiology, 156, 679–687.Find this resource:
Cronin, T. W. (1986). Optical design and evolutionary adaptation in crustacean compound eyes. Journal of Crustacean Biology, 6, 1–23.Find this resource:
Cronin, T. W., Bok, M. J., Marshall, N. J., & Caldwell, R. L. (2014). Filtering and polychromatic vision in mantis shrimps: Themes in visible and ultraviolet vision. Philosophical Transactions of the Royal Society B, 369, 20130032.Find this resource:
Cronin, T. W., & Caldwell, R. L. (2002). Tuning of visual function in three mantis shrimp species that inhabit a range of depths. I: Visual pigments. Journal of Comparative Physiology A, 188, 179–186.Find this resource:
Cronin, T. W., Caldwell, R. L., & Erdmann, M. (2002). Tuning of visual function in three mantis shrimp species that inhabit a range of depths. II: Filter pigments. Journal of Comparative Physiology A, 188, 187–197.Find this resource:
Cronin, T. W., Caldwell, R. L., & Marshall, J. (2001). A deeper shade of red: colour vision tuned to habitat in a mantis shrimp. Nature, 411, 547–548.Find this resource:
Cronin, T. W., Johnsen, S., Marshall, N. J., & Warrant, E. J. (2014). Visual ecology. Princeton, NJ: Princeton University Press.Find this resource:
Cronin, T. W., & Marshall, N. J. (1989a). Multiple spectral classes of photoreceptors in the retinas of gonodactyloid stomatopod crustaceans. Journal of Comparative Physiology A, 166, 267–275.Find this resource:
Cronin, T. W., & Marshall, N. J. (1989b). A retina with at least ten spectral types of photoreceptors in a stomatopod crustacean. Nature, 339, 137–140.Find this resource:
Cronin, T. W., & Marshall, J. (2004). The unique visual world of mantis shrimps. In F. Prete (Ed.), Complex worlds from simpler nervous systems (pp. 239–268). Cambridge, MA: MIT Press.Find this resource:
Cronin, T. W., & Marshall, J. (2011). Patterns and properties of polarized light in air and water. Philosophical Transactions of the Royal Society of London B, 366, 619–626.Find this resource:
Cronin, T. W., Marshall, N. J., & Caldwell, R. L. (1993). Photoreceptor spectral diversity in the retinas of squilloid and lysiosquilloid stomatopod crustaceans. Journal of Comparative Physiology A, 172, 339–350.Find this resource:
Cronin, T. W., Marshall, N. J., & Caldwell, R. L. (1994d). The retinas of mantis shrimps from low-light environments (Crustacea; Stomatopoda; Gonodactylidae). Journal of Comparative Physiology A, 172, 607–619.Find this resource:
Cronin, T. W., Marshall, N. J., & Caldwell, R. L. (1994b). The intrarhabdomal filters in the retinas of mantis shrimps. Vision Research, 34, 279–291.Find this resource:
Cronin, T. W., Marshall, N. J., & Caldwell, R. L. (1996). Visual pigment diversity in two genera of mantis shrimps implies rapid evolution. Journal of Comparative Physiology A, 179, 371–384.Find this resource:
Cronin, T. W., Marshall, N. J., & Caldwell, R. L. (2000). Spectral tuning and the visual ecology of mantis shrimps. Philosophical Transactions of the Royal Society of London B, 355, 1263–1267.Find this resource:
Cronin, T. W., Marshall, N. J., Caldwell, R. L., & Shashar, N. (1994c). Specialization of retinal function in the compound eyes of mantis shrimps. Vision Research, 34, 2639–2656.Find this resource:
Cronin, T. W., Marshall, N. J., & Land, M. F. (1991). Optokinesis in gonodactyloid mantis shrimps (Crustacea; Stomatopoda; Gonodactylidae). Journal of Comparative Physiology A, 168, 233–240.Find this resource:
Cronin, T. W., Marshall, N. J., & Land, M. F. (1994b). Vision in mantis shrimps. American Scientist, 82, 356–365.Find this resource:
Cronin, T. W., Marshall, N. J., Quinn, C. A., & King, C. A. (1994a). Ultraviolet photoreception in mantis shrimp. Vision Research, 34, 1443–1452.Find this resource:
Cronin, T. W., Nair, J. N., Doyle, R. D., & Caldwell, R. L. (1988). Visual tracking of rapidly moving targets by stomatopod crustaceans. Journal of Experimental Biology, 138, 155–179.Find this resource:
Cronin, T. W., Porter, M. L., Bok, M. J., Wolf, J. B., & Robinson, P. R. (2010). The molecular genetics and evolution of colour and polarization vision in stomatopod crustaceans. Ophthalmic and Physiological Optics, 30, 460–469.Find this resource:
Cronin, T. W., & 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. Journal of Experimental Biology, 204, 2461–2467.Find this resource:
Cronin, T. W., Shashar, N., Caldwell, R. L., Marshall, J.Cheroske, A. G., & Chiou, T.-H. (2003). Polarization vision and its role in biological signaling. Integrative and Comparative Biology, 43, 549–558.Find this resource:
Cronin, T. W., Yan, H. Y., & Bidle, K. D. (1992). Regional specialization for control of ocular movements in the compound eyes of a stomatopod crustacean. Journal of Experimental Biology, 171, 373–393.Find this resource:
Daly, I. M., How, M. J., Partridge, J. C., Marshall, N. J., Cronin, T. W., & Roberts, N. W. (2016). Rolling their eyes: dynamic polarization vision in mantis shrimps. Nature Communications, 7, 12140.Find this resource:
Exner, S. (1891). Die physiologie der facettirten augen von krebsen und insecten. Leipzig Wien: Deuticke.Find this resource:
Gagnon, Y. L., Templin, R. M., How, M. J., & Marshall, N. J. (2015). Circularly polarized light as a communication signal in mantis shrimp. Current Biology, 25, 3074–3078.Find this resource:
Goldsmith, T. H., & Cronin, T. W. (1993). The retinoids of seven species of mantis shrimp. Visual Neuroscience, 10, 915–920.Find this resource:
Harling, C. (2000). Re-examination of eye design in the classification of stomatopod crustaceans. Journal of Crustacean Biology, 20, 172–185.Find this resource:
Horridge, G. A. (1978). The separation of visual axes in apposition compound eyes. Philosophical Transactions of the Royal Society of London, 285, 1–59.Find this resource:
How, M. J., Christy, J., Roberts, N. W., & Marshall, N. J. (2014a). Null point of discrimination in crustacean polarisation vision. Journal of Experimental Biology, 217, 2462–2467.Find this resource:
How, M. J., Porter, M. L., Radford, A., Feller, K. D., Temple, S., Caldwell, R. L., . . . Roberts, N. W. (2014b). Out of the blue: The evolution of horizontally polarized signals in Haptosquilla (Crustacea, Stomatopoda, Protosquillidae). Journal of Experimental Biology, 217, 3425–3431.Find this resource:
Jones, J. (1994). Architecture and composition of the muscles that drive stomatopod eye movements. Journal of Experimental Biology, 188, 317–331.Find this resource:
Jordan, T. M.,Wilby, D., Chiou, Feller, K. D., Caldwell, R. L., Cronin, T. W., & Roberts, N.W. (2016). A shape-anisotropic reflective polarizer in a stomatopod crustacean. Scientific Reports, 6, 21744.Find this resource:
Kleinlogel, S., & Marshall, N. J. (2006). Electrophysiological evidence for linear polarization sensitivity in the compound eyes of the stomatopod crustacean Gonodactylus chiragra. Journal of Experimental Biology, 209, 4262–4272.Find this resource:
Kleinlogel, S., & Marshall, N. J. (2009). Ultraviolet polarisation sensitivity in the stomatopod crustacean Odontodactylus scyllarus. Journal of Comparative Physiology A, 195, 1153–1162.Find this resource:
Kleinlogel, S., Marshall, N. J., Horwood, J. M., & Land, M. F. (2003). Neuroarchitecture of the color and polarization vision system of the stomatopod Haptosquilla. Journal of Comparative Neurology, 467, 326–342.Find this resource:
Land, M. F., Marshall, N. J., Brownless, D., & Cronin, T. W. (1990). The eye-movements of the mantis shrimp Odontodactylus scyllarus (Crustacea: stomatopods). Journal of Comparative Physiology A, 167, 155–166.Find this resource:
Land, M. F., & Nilsson, D.-E. (2012). Animal eyes. Oxford: Oxford University Press.Find this resource:
Lythgoe, J. N. (1979). The ecology of vision. Oxford: Clarendon.Find this resource:
Manning, R. L., Schiff, H., & Abbott, B. C. (1984a). Eye structure and the classification of the stomatopod Crustacea. Zoologica Scripta, 13, 41–44.Find this resource:
Manning, R. L., Schiff, H., & Abbott, B. C. (1984b). Cornea shape and surface structure in some stomatopod crustacea. Journal of Crustacean Biology, 4, 502–513.Find this resource:
Marshall, N. J. (1988). A unique colour and polarisation vision system in mantis shrimps. Nature, 333, 557–560.Find this resource:
Marshall, J., & Cronin, T. W. (2011). Polarization vision. Current Biology, 21: R101–R105.Find this resource:
Marshall, J., & Cronin, T. W. (2014). Polarisation vision in crustaceans. In G. Horvath (Ed.), Polarized light and polarization Vision in animal sciences (2d ed., pp. 171–216). Berlin: Springer-Verlag.Find this resource:
Marshall, N. J., Cronin, T. W., & Kleinlogel, S. (2007). A review of stomatopod eye structure and function. Arthropod Structure and Development, 36, 420–448.Find this resource:
Marshall, J., Roberts, N., & Cronin, T. W. (2014). Polarisation signals. In G. Horvath (Ed.), Polarized light and polarization Vision in animal sciences (2d ed., pp. 407–442). Berlin: Springer-Verlag.Find this resource:
Marshall, N. J., Cronin, T. W., & Shashar, N. (1999). Behavioural evidence for polarization vision in stomatopods reveals a potential channel for communication. Current Biology, 9, 755–758.Find this resource:
Marshall, N. J., Jones, J. P., & Cronin, T. W. (1996). Behavioural evidence for color vision in stomatopod crustaceans. Journal of Comparative Physiology A, 179, 473–481.Find this resource:
Marshall, N. J., & Land, M. F. (1993a). Some optical features of the eyes of stomatopods. I. Eye shape, optical axis and resolution. Journal of Comparative Physiology A, 173, 565–582.Find this resource:
Marshall, N. J., & Land, M. F. (1993b). Some optical features of the eyes of stomatopods. II. Ommatidial design, sensitivity and habitat. Journal of Comparative Physiology A, 173, 583–594.Find this resource:
Marshall, N. J., Land, M. F., & Cronin, T. W. (1994). The six-eyed stomatopod. Endeavour, 18, 17–26.Find this resource:
Marshall, N. J., Land, M. F., & Cronin, T. W. (2014). Shrimps that pay attention: Saccadic eye movements in stomatopod crustaceans. Philosophical Transactions of the Royal Society of London B 369, 20130042.Find this resource:
Marshall, N. J., Land, M. F., King, C. A., & Cronin, T. W. (1991a). The compound eyes of mantis shrimps (Crustacea, Hoplocarida, Stomatopoda). I. Compound eye structure: The detection of polarized light. Philosophical Transactions of the Royal Society of London B, 334, 33–56.Find this resource:
Marshall, N. J., Land, M. F., King, C. A., & Cronin, T. W. (1991b). The compound eyes of mantis shrimps (Crustacea, Hoplocarida, Stomatopoda). II. Colour pigments in the eyes of Stomatopod crustaceans: Polychromatic vision by serial and lateral filtering. Philosophical Transactions of the Royal Society of London B, 344, 57–84.Find this resource:
Marshall, N. J., & Oberwinkler, J. (1999). The colourful world of the mantis shrimp. Nature, 401, 873–874.Find this resource:
Porter, M. L., Bok, M. J., Robinson, P. R., & Cronin, T. W. (2009). Molecular diversity of visual pigments in Stomatopoda (Crustacea). Visual Neuroscience, 26, 255–266.Find this resource:
Porter, M. L., Speiser, D. I., Zaharoff, S., Caldwell, R. L., Cronin, T. W., & Oakley, T. H. (2013). The evolution of complexity in visual systems of stomatopods: insights from transcriptomics. Integrative and Comparative Biology, 53, 39–49.Find this resource:
Porter, M. L., Zhang, Y., Desai, S., Caldwell, R. L., & Cronin, T. W. (2010). Evolution of anatomical and physiological specialization in the compound eyes of stomatopod crustaceans. Journal of Experimental Biology, 213, 3473–3496.Find this resource:
Roberts, N. W., Chiou, T.-H., Marshall, N. J., & Cronin, T. W. (2009). A biological quarter-wave retarder with excellent achromaticity in the visible wavelength region. Nature Photonics, 3, 641–644.Find this resource:
Schiff, H. (1996). Influence of different eye regions on striking, size discrimination, and habituation in mantis shrimps. Italian Journal of Zoology, 63, 139–148.Find this resource:
Schönenberger, N. (1977). The fine structure of the compound eye of Squilla mantis (Crustacea, Stomatopoda). Cell and Tissue Research, 176, 205–233.Find this resource:
Schram, F. R., Ahyong, S. T., Patek, S. N., Green, P. A., Rosario, M. V., Bok, M. J., … Abellό, P. (2013). Subclass Hoplocarida Calman, 1904; Order Stomatopoda Latreille, 1817. In J. C. von Vaupel Klein, M. Charmantier-Daures, & F. Schram (Eds.), Treatise on zoology—anatomy, taxonomy, biology, vol IV/A: The Crustacea (pp.179–356). Leiden: E. J. Brill.Find this resource:
Strausfeld, N. J., & Nässel, D. R. (1981). Neuroarchitecture of brain regions that subserve the compound eyes of crustacea and insects. In H. Autrum (Ed.), Handbook of sensory physiology, vol VII/6A (pp. 357–444). Berlin: Springer-Verlag.Find this resource:
Thoen, H. H., How, M. J., Chiou, T. H., & Marshall, J. (2014). A different form of color vision in mantis shrimp. Science, 343, 411–413.Find this resource:
Waterman, T. H. (1981). Polarization sensitivity. In H. Autrum (Ed.), Handbook of sensory physiology (Vol. VII/6B; pp. 283–469). Berlin: Springer-Verlag.Find this resource:
Wehner, R. (2001). Polarization vision—a uniform sensory capacity? Journal of Experimental Biology, 204, 2589–2596.Find this resource: