Microspectrophotometric evidence for cone monochromacy in sharks
- First Online:
- Cite this article as:
- Hart, N.S., Theiss, S.M., Harahush, B.K. et al. Naturwissenschaften (2011) 98: 193. doi:10.1007/s00114-010-0758-8
- 483 Views
Sharks are apex predators, and their evolutionary success is in part due to an impressive array of sensory systems, including vision. The eyes of sharks are well developed and function over a wide range of light levels. However, whilst close relatives of the sharks—the rays and chimaeras—are known to have the potential for colour vision, an evolutionary trait thought to provide distinct survival advantages, evidence for colour vision in sharks remains equivocal. Using single-receptor microspectrophotometry, we measured the absorbance spectra of visual pigments located in the retinal photoreceptors of 17 species of shark. We show that, while the spectral tuning of the rod (wavelength of maximum absorbance, λmax 484–518 nm) and cone (λmax 532–561 nm) visual pigments varies between species, each shark has only a single long-wavelength-sensitive cone type. This suggests that sharks may be cone monochromats and, therefore, potentially colour blind. Whilst cone monochromacy on land is rare, it may be a common strategy in the marine environment: many aquatic mammals (whales, dolphins and seals) also possess only a single, green-sensitive cone type. It appears that both sharks and marine mammals may have arrived at the same visual design by convergent evolution. The spectral tuning of the rod and cone pigments of sharks is also discussed in relation to their visual ecology.
KeywordsSharkColour visionMicrospectrophotometryConeVisual pigment
Elasmobranchs—or sharks and rays (including skates)—are the oldest living relatives of the first jawed vertebrates (Gnathostomata) and have well-developed eyes (Hart et al. 2006) and a large sensory brain area dedicated to the processing of visual information (Lisney and Collin 2006). Although it conforms to the basic vertebrate bauplan, the elasmobranch retina is relatively ‘simple’ in that it contains only rods and (where present) ‘single’ cone photoreceptors; it does not possess the paired ‘double’ or ‘twin’ cones seen in most teleost fish, reptiles, birds and marsupial mammals, the functions of which are still poorly understood. Combined with their extensive ecological diversity, including their adaptation to different light environments, elasmobranchs are a fascinating model system for the study of visual ecology, and in particular, the utility of colour vision in the aquatic environment.
The visual systems of very few of the more than 1,100 extant species of elasmobranch have been studied in detail, but it is evident that photoreceptor complement and the capacity for colour vision vary widely within the taxon. Skates (Raja spp.) appear to lack cone photoreceptors and have all-rod retinae containing a single visual pigment (Govardovskii and Lychakov 1977; Dowling and Ripps 1990); it is unlikely, therefore, that they are able to discriminate colour. On the other hand, many other species of ray have duplex retinae containing both rod and cone photoreceptors (Hart et al. 2006), and several lines of evidence indicate the presence of multiple cone visual pigments and the potential for colour vision. Using microspectrophotometry, two species of guitarfish (Aptychotrema rostrata and Glaucostegus typus) and one species of stingray (Neotrygon kuhlii) have been shown to possess three spectrally distinct cone visual pigments with maximal sensitivity to either short (blue), medium (green) or long (red) wavelengths of light and, therefore, have the potential for trichromatic colour vision (Hart et al. 2004; Theiss et al. 2007). These recent findings support earlier work on the ray retina showing colour-opponent electrical responses in the cone-specific (H3) horizontal cell sub-type of the red stingray Dasyatis akajei (Niwa and Tamura 1975; Tamura and Niwa 1967; Toyoda et al. 1978) and multiple peaks in the electroretinographically measured light-adapted spectral sensitivity function of the common stingray Dasyatis pastinaca (Govardovskii and Lychakov 1977).
Evidence for colour vision in sharks is more equivocal, despite a popular belief that they have this ability. Colour-opponent responses could not be recorded from horizontal cells in the California horned shark Heterodontus japonicas (Tamura and Niwa 1967) and only a single spectral peak has been measured in the light-adapted (cone-dominated) electroretinograms of both the piked dogfish Squalus acanthias (Govardovskii and Lychakov 1977) and the lesser spotted dogfish Scyliorhinus canicula (Gačić et al. 2007). While the failure to find evidence for multiple spectral mechanisms was potentially due to the relatively low numbers of cones present in the retinae of these species, other studies on sharks with more abundant cones are also suggestive of cone monochromacy. Extracellular recordings of retinal ganglion cells in the lemon shark Negaprion brevirostris—which has a peak rod/cone ratio of 5:1 (Cohen 1980)—have shown that chromatic adaptation does not shift the wavelength of peak spectral sensitivity from that recorded under broad-spectrum (‘white’) light adaptation (Cohen and Gruber 1985), suggesting that the photopic spectral sensitivity of these cells is derived from a single spectral class of cone.
Behavioural tests of colour vision in sharks have provided conflicting results. Adult N. brevirostris have been trained to discriminate successfully between white and red visual targets in return for a food reward (Clark 1963). Moreover, juvenile blacktip reef sharks Carcharhinus melanopterus and grey reef sharks Carcharhinus amblyrhynchos have been trained to discriminate between grey and coloured targets using a negative reinforcement paradigm (electric shocks; Tester and Kato 1966). However, in both of these studies, the potentially confounding effects of differences in relative stimulus brightness were not well controlled, and so it is not clear whether the sharks were responding to differences in colour or in the relative achromatic brightness between the stimuli.
Stimulus brightness controls were used in another experiment, where a physically restrained N. brevirostris was conditioned to respond—by extending its nictitating membrane—when a coloured adapting light was silently substituted for another colour (Gruber 1975). Whilst the paradigm used rendered stimulus brightness an unreliable cue, a significant difference in conditioned responses was only obtained when changes in stimulus colour were accompanied by changes in stimulus brightness. Subsequent behavioural studies with N. brevirostris using a free-swimming, two-choice paradigm failed to demonstrate colour discrimination ability, despite over 4,000 trials (Cohen 1980).
Given the aforementioned difficulties in demonstrating colour vision in sharks with behavioural and electrophysiological techniques, we decided to use a technique that has been used successfully to identify multiple cone visual pigments in other elasmobranchs, i.e. single-receptor microspectrophotometry (Hart et al. 2004; Theiss et al. 2007). A previous attempt to measure cone visual pigment absorbance spectra using this technique in the brown smooth-hound shark (Mustelus henlei) and the leopard shark (Triakis semifasciata) was unsuccessful, perhaps due to the relative paucity of cones in the retinae of these species (Sillman et al. 1996). In this study, we succeeded in measuring cone absorbance spectra in seven of the 17 shark species investigated; rod spectral absorbance data are also presented and the spectral tuning of shark visual pigments is discussed in relation to habitat and ecology.
Sharks were obtained from commercial fisherman or caught under permits from Queensland Fisheries (Permit 55710), the Moreton Bay Marine Park (Permit QS2003/CVL625) and the Great Barrier Reef Marine Park Authority (Limited Impact Accreditation Number UQ003/2006). Experimental protocols were approved by the University of Queensland (UQ) Animal Ethics Committee (SBMS/613/08/ARC). Seven common blacktip sharks (Carcharhinus limbatus; total length (TL) 75–106 cm), one dusky shark (Carcharhinus obscurus; TL 105 cm), one nervous shark (Carcharhinus cautus; TL 135 cm), two Australian sharpnose sharks (Rhizoprionodon taylori; TL 64, 67 cm), three pigeye sharks (Carcharhinus amboinensis; TL 75–150 cm), two spot-tail sharks (Carcharhinus sorrah; TL 104, 125 cm), one tiger shark (Galeocerdo cuvier; TL 204 cm), seven spotted wobbegong sharks (Orectolobus maculatus; TL 80–132 cm), eight ornate wobbegong sharks (Orectolobus ornatus; TL 51–64 cm), two grey carpetsharks (Chiloscyllium punctatum; TL 17, 86 cm) and two bull sharks (Carcharhinus leucas; TL 177, 183 cm) were caught in Moreton Bay, Queensland, Australia. Four additional bull sharks (TL 81–88 cm) were caught from the freshwater and brackish reaches of the Brisbane River, Queensland. Two blacktip reef sharks (Carcharhinus melanopterus; TL 63, 73 cm), two sharptooth lemon sharks (Negaprion acutidens; TL 78, 80 cm) and three epaulette sharks (Hemiscyllium ocellatum; TL 55–63 cm) were caught from the reef shallows off Heron Island, near Gladstone, Queensland. Three western wobbegong sharks (Orectolobus hutchinsi; TL 91–105 cm), three dwarf spotted wobbegong sharks (Orectolobus parvimaculatus; TL 85–92 cm) and one Port Jackson shark (Heterodontus portusjacksoni; TL 67 cm) were caught in waters off Geraldton and/or Mandurah in Western Australia. Sharks were dark-adapted for 5–60 min before being euthanased by an overdose of tricaine methane sulfonate salt (MS222; 1:2,000) or by spinal section and pithing.
Eyes (usually both from each specimen) were removed and dissected under infrared light with the aid of infrared image converters (ElectroViewer 7215; Electrophysics Corporation, Fairfield, NJ, USA). Retinal tissue was prepared for microspectrophotometry as described in detail elsewhere (Hart et al. 2004; Theiss et al. 2007). Queensland species were analysed immediately using fresh, unfixed retinal tissue. Isolated retinas from O. hutchinsi, O. parvimaculatus and H. portusjacksoni were immersion fixed in 2.5% glutaraldehyde in phosphate buffered saline (PBS, 860 mOsmol kg−1; pH 7.0) for 30 s, washed in PBS for 30 s and stored in PBS in light-tight containers at 4°C until they were transported to UQ, where they were analysed 3–12 days after fixation. Through comparisons of fixed and unfixed tissue from O. ornatus, it is known that the λmax values of rod visual pigments measured using these two preparation techniques differ by less than 1 nm (Theiss 2009). In all cases, small pieces (ca. 2 × 2 mm) of retinal tissue were mounted between No. 1 glass coverslips in a drop of shark physiological saline solution (Anderson et al. 2005) containing 8% dextran (MW 282,000; Sigma D-7265). Retinal samples thus prepared were examined using a computer-controlled single-beam wavelength-scanning microspectrophotometer (Hart 2004). The spectral range of this machine has been extended recently (now 330–800 nm) by the addition of a Zeiss Ultrafluar ×32 0.4 NA glycerol-immersion condenser objective and a Zeiss Fluar ×100 1.3 NA oil-immersion collector objective.
Transverse absorbance spectra were measured from individual rod and cone outer segments. A sample scan was made by aligning the measuring beam (typical dimensions 1 × 3 μm for cones and 1 × 10 μm for rods) in the outer segment and recording the amount of light transmitted at each wavelength across the spectrum. In the case of rods, which were always measured at the periphery of the mounted tissue, baseline scans were made in an identical fashion to sample scans but from a cell-free area of the preparation adjacent to the outer segment. Intact cones were never found ‘floating’ in the mountant or at the periphery of the tissue and were always measured within the mass of retinal tissue at locations where the preparations had been ‘squashed’ sufficiently that adjacent photoreceptors did not overlap. Baseline scans for the cone outer segment absorbance spectra were made from nearby areas of the retina that did not contain photoreceptor outer segments. The transmittance (ratio of sample to baseline signal) of the outer segment was calculated at each wavelength and converted to absorbance to give a prebleach spectrum. Each outer segment was then bleached with white light for 2 min and subsequent sample and baseline scans made to create a postbleach spectrum. The creation of postbleach spectra (not presented here) confirmed that any putative visual pigments were photolabile.
Only spectra that satisfied established selection criteria (Hart et al. 1998; Levine and MacNichol 1985) were retained for further analysis. Individual prebleach absorbance spectra were analysed as described elsewhere (Govardovskii et al. 2000; Hart 2002; MacNichol 1986) to provide an estimate of the wavelength of maximum absorbance (λmax) of each outer segment/visual pigment. The mean λmax of a given visual pigment was then calculated from these individual λmax values. For display purposes, a mean prebleach absorbance spectrum was calculated by averaging acceptable individual (non-normalised) absorbance spectra. To investigate the possibility that the photoreceptors of some species contained a mixture of visual pigment molecules utilising both the A1 and A2 chromophores, mean prebleach absorbance spectra (smoothed with a 21-nm unweighted running average) were fitted iteratively with mixed-chromophore pigment templates to find the combination of visual pigment λmax (pure A1) and A1/A2 ratio that gave the smallest sums of squares deviation between the real and modelled spectra between 0.7 and 0.3 normalised absorbance on the long-wavelength limb of the real spectrum (the same region used to estimate λmax; Temple et al. 2010). We made the assumption that only a single type of visual pigment opsin protein was present in each photoreceptor and used established A1 and A2 visual pigment templates (Govardovskii et al. 2000) and a known relationship between the λmax values of A1 and A2 visual pigment pairs (Parry and Bowmaker 2000).
Mean wavelengths of maximum absorbance (λmax) of rod and cone visual pigments in 17 species of shark
Rod λmax (nm)
Cone λmax (nm)
Carcharhinidae (Whaler sharks)
Common blacktip shark
505.6 ± 1.8 (n = 51)
531.8 ± 3.9 (n = 134)
518.4 ± 6.5 (n = 47)
554.4 ± 11.7 (n = 45)
Australian sharpnose shark
508.2 ± 2.7 (n = 18)
533.3 ± 8.9 (n = 6)
507.4 ± 5.2 (n = 24)
534.1 ± 8.9 (n = 8)
505.5 ± 4.4 (n = 20)
Blacktip reef shark
505.4 ± 3.9 (n = 27)
502.4 ± 2.1 (n = 5)
504.9 ± 3.3 (n = 10)
499.3 ± 5.3 (n = 15)
Sharptooth lemon sharka
513.5 ± 7.7 (n = 34)
Hemiscyllidae (Longtail carpetsharks)
499.5 ± 2.6 (n = 13)
499.6 ± 2.6 (n = 84)
531.8 ± 6.7 (n = 5)
484.4 ± 3.9 (n = 44)
552.8 ± 4.8 (n = 10)
498.4 ± 3.7 (n = 45)
560.5 ± 4.9 (n = 10)
495.6 ± 2.2 (n = 27)
Dwarf spotted wobbegongb
493.8 ± 3.2 (n = 28)
Port Jackson sharkb
496.3 ± 3.1 (n = 14)
Absorbance spectra from all species except C. leucas and N. acutidens were fitted well by a vitamin A1-based visual pigment (rhodopsin) template and, therefore, are assumed to contain a single chromophore type (retinal). However, the visual pigment absorbance spectra of C. leucas and N. acutidens were considerably broader (full-width at half maximum bandwidth) than a template based on the A1 chromophore alone and narrower than a template based solely on the vitamin A2 chromophore (3, 4-didehydroretinal; porphyropsin). Modelling shows that the shapes of the mean rod absorbance spectra in C. leucas and N. acutidens are explained best by visual pigments with λmax values (when combined with the A1 chromophore only) and an A1/A2 chromophore ratio of 509 nm at 30:70% and 507 nm at 47:53%, respectively. Assuming that the A1/A2 ratio in C. leucas cones is the same as it is in the rods, the λmax of the cone opsin when combined with pure A1 chromophore would be 535 nm, suggesting it belongs to the same opsin ‘class’ as the LWS cone pigments in the other shark species.
We have shown, using microspectrophotometry, that the retinae of sharks from three different families contain only a single class of rod and a single class of long-wavelength-sensitive (LWS) cone photoreceptor, based on the absorbance spectra of the visual pigments they contain. In contrast to other studies using this technique in different vertebrate groups, cones were only seen within the retinal tissue mass, never free-floating or at the periphery of the sample. This may explain the failure to find any cones at all in some species of shark, both in this study and previous investigations (Sillman et al. 1996), although on the basis of histological data we suspect that at least one of the species we studied (H. portusjacksoni) lacks cones altogether (Scheiber 2007). Of the shark species in which cones were measured, there is a possibility that less abundant spectral types of cone are also present but were missed during our analysis. However, although the numbers of cones measured in some species was relatively low, the higher numbers measured in C. leucas and C. limbatus give more confidence that a second spectral class of cone was not simply overlooked. The numbers of cones measured in these two species (Table 1) were considerably higher than those that have been required to identify multiple cone types in a variety of ray species (Hart et al. 2004; Theiss et al. 2007; Hart, unpublished data) and, on the basis of sample size calculations (Cohen 1988), should have been sufficient to detect a population of cones that were present at levels of 1.5% and 4.5% of the total cone population, respectively (α = 0.05, power = 0.8), providing strong evidence that these sharks do, in fact, possess only a single cone type.
Cone monochromacy per se does not prohibit colour vision. If the rod and cone photoreceptors contain visual pigments with different spectral absorbance characteristics, and either the retina or brain possesses the neural machinery to compare the signals from rod and cone pathways, a rudimentary form of colour vision might exist. This is not without precedent: blue-cone monochromat humans (Alpern et al. 1971) and red-cone monochromat owl monkeys Aotus trivirgatus (Jacobs et al. 1993) have dichromatic colour vision based on a comparison of signals between their rods and cones. Moreover, despite a wealth of molecular genetic (Newman and Robinson 2005), immunohistological (Peichl et al. 2001) and electrophysiological (Crognale et al. 1998) data confirming the presence of only a single cone visual pigment, the results of some behavioural tests have implied that seals (Wartzok and McCormick 1978; Busch and Dücker 1987), sea lions (Griebel and Schmid 1992) and dolphins (Griebel and Schmid 2002) are also able to discriminate colours, presumably on the basis of rod–cone interactions (see also Crognale et al. 1998). (However, it should be noted here that other studies have failed to show wavelength discrimination ability in dolphins (Madsen 1976) and that the low threshold of brightness discrimination of harbour seals (Scholtyssek et al. 2008) may have confounded the interpretation of results from some of the early studies on colour vision that did not use test stimuli of equal subjective brightness). Consequently, it is possible that at least some of the positive outcomes from behavioural tests of colour vision in sharks (Clark 1963; Gruber 1975; Tester and Kato 1966) were the result of a colour vision system based on the comparison of signals from rods and cones.
In mammals, we might expect some sort of interaction between rods and cones as the rod pathway ‘piggybacks’ on the cone pathway: unlike cone bipolar cells, mammalian rod bipolars do not make direct synaptic contact with ganglion cells and instead pass signals indirectly via the AII amacrine cells (Kolb and Famiglietti 1974; Pang et al. 2007). Moreover, we might expect the neural machinery of colour vision to still exist in some form in LWS-cone monochromat marine mammals: although they lack the functional short-wavelength-sensitive (SWS) cones seen in some other aquatic and most terrestrial mammals, they appear to have either an intact SWS opsin gene that is not expressed in the retina or an SWS opsin pseudogene (Newman and Robinson 2005), suggesting that the loss of the SWS/blue-cone type from the retina is a relatively recent evolutionary event. However, the wiring of rod and cone pathways in sharks is unknown and they may be unable to ‘cheat’ at colour vision in this way.
It is not immediately obvious why sharks and some marine mammals have lost all but one cone type. Teleost fish with only a single spectral type of cone are rare, but in those with only two cone types that live at moderate depth in coastal waters, the most long-wavelength-sensitive of the two cone pigments often has a λmax between 520 and 545 nm (Lythgoe and Partridge 1991), similar to that seen in sharks. The spectral tuning of teleost visual pigments is thought to be driven primarily by habitat depth and water type. Both marine and freshwater species living at depth tend to have rod and cone pigments shifted towards shorter (bluer) wavelengths than those living near the surface (Bowmaker et al. 1994; Denton and Shaw 1963; Denton and Warren 1956; Munz 1957). Moreover, freshwater species tend to have visual pigments that are shifted towards longer (redder) wavelengths than coastal marine species, which in turn are more red-shifted than open water or reef species (Loew and Lythgoe 1978; Lythgoe et al. 1994). In general, this spectral tuning serves to maximise sensitivity by matching visual pigment spectral absorption to ambient illumination, which becomes increasingly blue with depth in the sea and increasingly green-red in freshwater (Levine and MacNichol 1982; Jerlov 1976).
C. leucas is one of the few elasmobranch species that lives in brackish and freshwater habitats for extended periods—juvenile sharks may remain in rivers for up to 5 years after birth (Last and Stevens 2009)—and the relatively red-shifted rod (λmax 518 nm) and cone (λmax 554 nm) pigments seen in the juvenile and sub-adult specimens examined in this study would be adaptive to maximise visual sensitivity in such habitats. C. leucas mature at about 220–230 cm in Australia, and the largest specimen we obtained was 183 cm. Although this sub-adult was taken from a marine habitat, it had the same visual pigment λmax values as the river-caught juvenile specimens studied and it is unknown whether adult C. leucas will have identical visual pigments or, like N. brevirostris, will show an ontogenetic shift in rod pigment sensitivity towards shorter wavelengths (λmax 501 nm) compared to that found in juveniles (λmax 522 nm; Cohen et al. 1990). As with N. brevirostris, the relatively red-shifted rod and cone visual pigment λmax values seen in N. acutidens and C. leucas are most likely caused by a mixture of A1 and A2 chromophores in the outer segment.
Amongst the carcharhinid sharks investigated, few are deep-living, with most species usually found in water less than 100–150 m in depth (Last and Stevens 2009). Nevertheless, there is considerable variation in rod visual pigment λmax from 499 nm in G. cuvier to 518 nm in C. leucas. At least some of this variation may be due to behavioural differences: bentho-pelagic species spending more time near the substrate (C. leucas, R. taylori, C. amboiensis and N. acutidens) have rod λmax values (range 508–516 nm) shifted towards longer wavelengths compared to those species that are more pelagic (G. cuvier, C. sorrah, C. obscurus, C. melanopterus and C. limbatus), which have λmax values in the range of 499–506 nm. Whilst the absolute difference in rod λmax between these two groups is small, a similar difference in rod λmax value between allopatric populations of the sand goby Pomatoschistus minutus is thought to confer an adaptive advantage given the differences in the spectral distribution of light available in the different habitats they occupy (Jokela et al. 2003). It may be that the rod visual pigments of bentho-pelagic carcharhinid sharks are tuned more to the radiance reflected from the substrate, which is relatively richer in long wavelengths compared to the upwelling radiance encountered at a similar depth in deeper water (McFarland and Munz 1975), than the pelagic species, which spend less time swimming close to the substrate and must scan the water below them for predators, a task that would benefit from a more blue-shifted rod visual pigment (Munz and McFarland 1973).
To some extent, this trend also appears to explain the spectral tuning of the cone pigments, which have a range of λmax from 532 to 561 nm. Shallow coastal waters, where these species were caught, have a peak spectral radiance between 525 and 575 nm, largely independent of the direction of view (Hart et al. 2004). Consequently, any cone pigment in this wavelength range would confer maximal visual sensitivity during the day. However, the strongly benthic wobbegong sharks (O. maculatus and O. ornatus) have red-shifted cone visual pigments (λmax 553–561 nm) compared to the other bentho-pelagic and pelagic shark species and these may be adaptive for detecting and ambushing prey items swimming close to the substrate.
All of the wobbegong sharks and H. portusjacksoni, H. ocelatum and C. punctatum have rod λmax values that are blue-shifted (range 484–500 nm) compared to the other (carcharhinid) species. Although O. maculatus and H. portusjacksoni have been observed at depths of up to 218 m and 275 m respectively, the rest of the species examined in this study are found in water less than 80–135 m in depth (Last and Stevens 2009). Depth alone, therefore, would not seem to be sufficient to cause the blue-shifted rod λmax values. Moreover, these species are predominantly benthic in habit (Last and Stevens 2009), which we have seen appears to be correlated with a more red-shifted rod λmax in the carcharhinids.
Notwithstanding phylogenetic constraints on visual pigment spectral tuning, it is possible that the degree of nocturnality may influence rod λmax. Heterodontus portusjacksoni, H. ocelatum, O. maculatus and O. hutchinsi (and probably other wobbegong species) are known to be strongly nocturnal (Compagno et al. 2005; Last and Stevens 2009; Chidlow 2003). Most of the carcharhinid species studied would best be described as crepuscular or cathemeral, with the possible exception of the tiger shark, which is considered to be quite nocturnal and also has a relatively blue-shifted rod λmax. However, at least in tropical marine teleosts, the degree of nocturnality per se does not appear to be strongly correlated with rod λmax (Munz and McFarland 1973) and the functional significance of a blue-shifted rod pigment for nocturnal vision shift is unclear. One possible explanation is that a blue-shifted visual pigment might be better for the detection of bioluminescent light sources—which frequently have a spectral emission peak in the blue-green region of the spectrum (for a recent review see Haddock et al. 2010)—than one matched to the spectral peak of the downwelling light (Frank and Widder 1999; Douglas et al. 1998). It is unknown how many of the prey items (predominantly bony fish) known to be consumed by wobbegongs (Huveneers et al. 2007) and H. portusjacksoni (Powter et al. 2010) are themselves sufficiently bioluminescent (if at all) for this to be used as a means of prey detection by the sharks at night, but it is also possible that the sharks might use the mechanically-stimulated bioluminescence of commonly occurring dinoflagellates (wavelength of peak emission approximately 479 nm, e.g. Esaias et al. 1973) to deduce the locations of (non-bioluminescent) moving prey, as is thought to be the case for some teleost fish that feed on zooplankton (Mensinger and Case 1992).
Cone pigment λmax values in monochromat marine mammals range from 522 nm in the harbour porpoise Phocoena phocoena to 548 nm in the harp seal Phagophilus groenlandicus, with the general trend that deeper-diving cetacean species have cone pigments that are blue-shifted compared to shallower-dwelling species (Newman and Robinson 2005). In each case, the cone pigment expressed in the retina is produced by the long-wavelength-sensitive (LWS) opsin gene that is found in other mammals. Possible reasons for the loss of SWS cones from the retina of these aquatic mammals include reducing the deleterious effects of chromatic aberration on visual performance and an evolutionary ‘bottleneck’ caused by the long-wavelength-dominated coastal habitat that marine mammals would have experienced as they returned to the sea (Peichl et al. 2001). As suggested for the sharks in this study, in coastal habitats a LWS cone type would confer greater visual sensitivity than a SWS cone due to the predominance of longer wavelengths, and at some stage during their evolution there may have been no selective pressure to retain SWS cones.
In most mammals, SWS cones comprise around 4–10% of the cone population regardless of rod/cone ratio (Peichl et al. 2001; Szél et al. 1994), which would imply that, whether on land or in water, the LWS cone type is more important for photopic vision. If multiple spectral cone types are not required for the purposes of colour vision, the retention of a duplex retina would serve only to extend the dynamic range of the visual system from brightly lit surface waters to the deep ocean. However, it is not immediately obvious why sharks should have only a single spectral cone type while rays (Hart et al. 2004; Theiss et al. 2007) and even one species of chimaerid holocephalan, the elephant shark Callorhinchus milii (Davies et al. 2009), should have up to three different cone pigments. Moreover, the all-rod retina of the skate (Raja spp.) appears to be perfectly capable of functioning over a wide range of light intensities through adaptive changes in rod photokinetics rather than employing dedicated photopic light receptors, i.e. cones (Ripps and Dowling 1990).
Nocturnality alone does not appear to be sufficient to cause cone monochromacy in mammals (Peichl et al. 2001) and recent evidence suggests that the relative importance of other sensory modalities may influence visual pigment complement (Zhao et al. 2009). By extension, the ambient light environment alone may not be the only determinant of visual system design in aquatic animals and the relative importance of other senses such as electroreception and the mechanosensory lateral line in different behavioural tasks may underlie differences in cone complement between elasmobranch families. The next important step in this story is to isolate and sequence the visual pigment opsin genes found in sharks. If such analyses support our finding that sharks are cone monochromats, this will have important implications not just for the design of all aquatic visual systems, but also for our understanding of the visual ecology and sensory-driven behaviours of sharks.
The authors would like to thank Scott Cutmore, Jeremy Ullmann, Clint Chapman, John Page, Alan Goldizen, Jamie Thornton, Darren Sapelli and Bob Stone for assistance in obtaining specimens, and three anonymous reviewers for their helpful comments on the manuscript. Funding was provided by the Australian Research Council and the Sea World Research and Rescue Foundation. This paper is dedicated to the late Dr. Julia Shand.