A Different Form of Color Vision in Mantis Shrimp

See allHide authors and affiliations

Science  24 Jan 2014:
Vol. 343, Issue 6169, pp. 411-413
DOI: 10.1126/science.1245824


One of the most complex eyes in the animal kingdom can be found in species of stomatopod crustaceans (mantis shrimp), some of which have 12 different photoreceptor types, each sampling a narrow set of wavelengths ranging from deep ultraviolet to far red (300 to 720 nanometers) (13). Functionally, this chromatic complexity has presented a mystery (35). Why use 12 color channels when three or four are sufficient for fine color discrimination? Behavioral wavelength discrimination tests (Δλ functions) in stomatopods revealed a surprisingly poor performance, ruling out color vision that makes use of the conventional color-opponent coding system (68). Instead, our experiments suggest that stomatopods use a previously unknown color vision system based on temporal signaling combined with scanning eye movements, enabling a type of color recognition rather than discrimination.

Living Technicolor

Color vision is generally carried out through the number of photoreceptor types found in the retina. The mantis shrimps (stomatopods) can have up to 12 photoreceptors, far more than needed for even extreme color acuity. Thoen et al. (p. 411; see the Perspective by Land and Osorio) conducted paired color discrimination tests with stomatopods and found that their ability to discriminate among colors was surprisingly low. Instead, stomatopods appear to use a color identification approach that results from a temporal scan of an object across the 12 photoreceptor sensitivities. This entirely unique form of vision would allow for extremely rapid color recognition without the need to discriminate between wavelengths within a spectrum.

Stomatopods are benthic marine crustaceans that are generally found in tropical and temperate waters. Their compound eyes possess the largest number of photoreceptor types known in any animal [between 16 and 21 different receptors in some species (1, 3, 9)], allowing them to discriminate color (5) as well as both linear and circular polarized light (3, 10). Such retinal complexity is unrivaled in the animal kingdom, although papilionid butterflies may have up to eight spectral sensitivities (11). Theoretical approaches have predicted that between four and seven photoreceptor types are all that is needed to accurately encode the colors of the visible spectrum (1214). The four-channel (tetrachromatic) solution that birds and reptiles use to sample a spectral range from 300 to 700 nm is optimally arranged to encode the known colors within this range. Where the spectrum examined loses the ultraviolet (UV) or red end, three photoreceptors suffice, and trichromacy is the solution that many animals exhibit (12).

Our question was therefore, why do the stomatopods use 12 different photoreceptors to encode color? Before the experiments described here, Marshall et al. (5) demonstrated that stomatopods are capable of simple color discriminations based on color-card tests, similar to those devised by Carl von Frisch for bees and now used widely for a number of animals (15). We hypothesized two alternative mechanisms for color information processing in stomatopods: (i) a multiple dichromatic color-opponent system (as described below), or (ii) the binning of colors into 12 separate channels, without any between-channel comparisons (4, 16).

Like butterflies, stomatopods have a variety of colorful body patterns, even using fluorescence to enhance color display (17). Furthermore, many of these species inhabit shallow coral reefs, one of the most colorful environments on Earth. The stomatopod’s colors are thought to be involved in particularly complex communication systems, both between and within species (18), but little of this complexity requires a 12-dimensional color space to distinguish the colors available. Osorio et al. (19) speculated that stomatopods use their color sense to make reliable and quick judgements of color signals from conspecifics under changing light conditions, their steep-sided spectral sensitivities allowing particularly good color constancy. This would require comparison of the spectrally adjacent sharply tuned spectral sensitivities, and there is some anatomical evidence supporting this idea (6).

Stomatopod eyes are made up of a dorsal and ventral hemisphere, divided by a region of distinct ommatidia (optical units) termed the midband (Fig. 1). Within two superfamilies of stomatopods (Gonodactyloidea and Lysiosquilloidea), this midband consists of six separate rows of ommatidia, each with different functionalities (2). Rows 1 to 4 are involved in color processing; rows 5 and 6 mediate the detection of circular or linear polarized light (3). A total of 12 cell types, each with different spectral sensitivities, are found within rows 1 to 4, with four UV-sensitive retinular cells located distally in the retina and by convention termed R8 cells (3). Beneath the R8 cells, the remaining seven retinular cells (R1 to R7) are further divided into two tiers (2). Comparison of the secondary R1 to R7 cell tiers would yield a set of highly tuned dichromatic mechanisms, sampling the 400- to 700-nm part of the spectrum in four bins as per hypothesis (i) above (3, 6, 19) (Fig. 1). Hypothesis (ii) would require a recognition of the pattern of excitation over the entire spectrum.

Fig. 1 (A) Spectral sensitivities of H. trispinosa.

Spectral sensitivity curves obtained from intracellular electrophysiological recordings. The figure shows smoothed data (four neighbors on each side, second-order polynomial), normalized to 100% (see table S1). (B) Eye of H. trispinosa. Showing the dorsal hemisphere (DH) and ventral hemisphere (VH), divided by the midband (MB) containing the color receptors in the four top rows (CV).

To distinguish between the two proposed hypotheses, we tested the ability of stomatopods to distinguish between different hues [spectral discrimination (Δλ) functions] using the Gonodactyloid stomatopod species Haptosquilla trispinosa. When two narrow-band spectral stimuli are presented simultaneously, they can only be discriminated when the difference between them is over a certain threshold, giving the minimum discriminable difference. Spectral discrimination curves obtained from other animals usually exhibit certain minima in areas between the spectral sensitivity curves (20). Therefore, dichromats usually exhibit one such minimum, trichromats have two minima, tetrachromats have three, and so on. Our goal was therefore to test the color discrimination abilities of mantis shrimp by using a two-way choice test (Fig. 2 and fig. S1) in which the animal is trained to a specific wavelength by means of food rewards. The wavelength stimuli were presented to the animal with a pair of optical fibers, and a choice was recorded when the animal grabbed or tapped the end of the fiber. Test colors were presented together with the trained colors at varying wavelength intervals to determine at what point the animal could no longer discriminate between the two stimuli (i.e., when the success rate dropped to 50%).

Fig. 2 Examples of correct choice data from a two-way choice test of H. trispinosa.

Curves are plotted as mean ± SEM. The horizontal dashed lines indicate the 50% (chance) and 60% (discrimination) criteria. (A) Choices of animals trained to 470 nm (n = 7) and tested toward longer wavelengths. (C) Choices of animals trained to 570 nm (n = 4) and tested toward shorter wavelengths. The number above each point indicates the tested wavelength interval. (B and D) Examples of animals making a choice.

Animals were trained successfully to 10 different color wavelengths: 400, 425, 450, 470, 500, 525, 570, 578, 628, and 650 nm. When the test wavelength was 50 to 100 nm from the trained wavelength, the success rates were between 70 and 80%, indicating that they discriminated well between the two stimuli. However, when the interval between the trained and test wavelengths was reduced to between 25 and 12 nm, the success rates dropped to around 50%, and it was clear that the stomatopods could no longer distinguish test from trained stimuli. An example of success rates is given in Fig. 2, and further results are shown in tables S2 and S3. The discrimination threshold was chosen to be at a 60% success rate, in accordance with previous studies on animal discrimination thresholds (20). Using the interpolated points at 60%, we determined a relative spectral discrimination curve of Δλ/λ (Fig. 3). The resulting values of Δλ were all in the region between 12 and 25 nm, and the prominent dips, or minima, usually associated with the points between spectral sensitivity maxima in other studies were not clearly defined by spectral overlap regions (Fig. 3).

Fig. 3 Spectral discrimination curves (Δλ/λ).

The spectral discrimination curve from behavioral testing of H. trispinosa is shown by a thick black line, and the modeled spectral discrimination curve is shown by a thick dashed line. [The figure is modified from Koshitaka et al. (20).]

The potential spectral discrimination ability, based on previously known color vision systems, was also modeled to allow us to make predictions about the stomatopod color processing system. The sensitivities of the photoreceptors in H. trispinosa were measured using intracellular electrophysiology (9) (table S1). Eight sensitivity maxima were found in the visible part of the spectrum, and another three were found in the UV {the fourth UV cell often proving hard to locate and record from [(9), Fig. 1]}. We modeled the stomatopod spectral discrimination curve using the Vorobyev/Osorio noise-limited model (21) (which determines color thresholds using photoreceptor noise levels) for a serial dichromatic system with comparison between each adjacent spectral sensitivity [mechanism (i) above (note S1)]. This system predicts very fine discrimination between 1 to 5 nm throughout the spectrum, with few peaks of coarser discrimination as seen in other animals. Such fine spectral discrimination would be expected in a color vision system that made analog comparisons between adjacent spectral sensitivities. The behavioral results presented here (Fig. 2) suggest that such analog comparisons are not made. Instead, stomatopod color vision is remarkably coarse (Fig. 3).

The results from our experiments suggest that the stomatopods do not use a processing system of multiple dichromatic comparisons as previously hypothesized based on assumed neural connections (16). Instead, we provide evidence that scanning eye movements (22) may generate a temporal signal for each spectral sensitivity, enabling them to recognize colors instead of discriminating them. (Fig. 4) (3, 4). In such a system, the 12 sensitivities (including the UV, not analyzed here but with its multiple sensitivities a good fit to the system envisaged) would be converted into a temporal pattern when scanned across an object, which the animal could recognize as color. This system is comparable to the spectral linear analyzers (termed ”push-broom” analyzers because of the arrangement of the sensors and the flight direction) used in remote sensing systems (23) and is a unique way for animals to encode color. Although this system does not have the ability to discriminate between closely positioned wavelengths (and results in spectral “discrimination” defined by the distances between sensitivity peaks, seen when comparing Fig. 1 and 3), it would enable the stomatopod to make quick and reliable determinations of color, without the processing delay required for a multidimensional color space. Without the comparison of spectral channels, color constancy would not function in the way we currently understand it in other animals. Instead, identification of a color pattern by the mid-band and luminance by the hemispheres might function as a “panchromatic” method to discount illuminance (23). The eye is optically skewed so that both midband and hemispheres examine the same areas in space, which lends support to this idea (3). However, the details of the neural processing from the receptors remain unknown.

Fig. 4 Proposed processing mechanism.

(A) Idealized spectral reflectance from stomatopod body parts. WL, wavelength. (B) Spectral sensitivities throughout the spectrum divided into separate bins. (C and D) Excitation patterns of each spectral sensitivity when looking at the blue (C) and red (D) reflectance spectra.

Stomatopods live a rapid-fire lifestyle of combat and territoriality, so possessing a simple, temporally efficient color recognition ability may be critical for survival (24, 25). As with many invertebrate information-encoding solutions, the actual processing of the problem is dealt with at the periphery, in this case by an array of detectors seen in animals and unconsciously duplicated by remote-sensing engineers. What remains for us to discover is the nature of the information and its importance in the biological decisions these engaging crustaceans make.

Supplementary Materials

Materials and Methods

Figs. S1 and S2

Tables S1 to S3

References (2629)

References and Notes

  1. Acknowledgments: This work was supported by grants from the Asian Office of Aerospace Research and Development, the Air Force Office of Scientific Research, the Australian Research Council, the Lizard Island Research Foundation, and a Doctoral Fellowship (2013) from the Lizard Island Research Station, a facility of the Australian Museum. We thank the reviewers for their great comments and W.-S. Chung, M. Bue, and R. Bedford for technical help. All data described in this manuscript are presented in either the main manuscript or the supplementary materials.
View Abstract

Navigate This Article