Light and vision in the deep-sea benthos: II. Vision in deep-sea crustaceans

The large eyes of deep-sea animals have been remarked upon since the first specimens were pulled up in trawl nets. In addition to large eyes, early structural studies on mesopelagic species (i.e. those living in the water column) demonstrated further adaptations for increasing sensitivity to the dim environmental light field, including larger apertures, longer photoreceptor cells, reductions in screening pigments and the presence of reflecting tapeta (reviewed in Warrant and Locket, 2004). Studies on visual pigment absorption maxima from mesopelagic species have shown a shift in peak sensitivity to shorter, ‘bluer’ wavelengths compared with those of shallow-water species [e.g. fish (reviewed in Douglas et al., 2003; Turner et al., 2009) and crustaceans (reviewed in Marshall et al., 2003)]. This shift to blue sensitivity is an adaptation to both downwelling light, which becomes more monochromatic and blue with increasing depth (Jerlov, 1976), and bioluminescence, the vast majority of which lies between 450 and 510 nm (Herring, 1983; Widder et al., 1983; Haddock and Case, 1999).

The only vision studies that have been conducted on benthic (bottom-dwelling) species are structural, and these show that there are similar structural adaptations for increasing light sensitivity, including reflecting tapeta [e.g. crustaceans (Hiller-Adams and Case, 1985; Eguchi et al., 1997) and fish (Denton and Shaw, 1963; Somiya, 1980)], enlarged photoreceptive membranes (Hiller-Adams and Case, 1985; Eguchi et al., 1997) and aphakic spaces [allowing light to hit the retina without going through the lens (Denton, 1990)]. However, in mesopelagic crustaceans, eye growth rates relative to body growth rates decline with depth, as does the eye size relative to body length (Hiller-Adams and Case, 1984; Adams and Case, 1988), whereas in benthic species, there is a depth-related increase in both parameters, resulting in the eyes of many benthic crustaceans being larger, relative to body length, than those of pelagic crustaceans (Hiller-Adams and Case, 1985). Similar trends are found among deep-sea fish: eyes decrease in size with increasing habitat depth among a variety of pelagic species, but many deep-sea benthic or near-benthic species retain large, well-developed eyes (Murray and Hjort, 1912; Locket, 1977). Many of these benthic species have depth ranges well below the depths at which there is sufficient daylight remaining to have any significance for vision (Denton, 1990), suggesting that these species may be adapted for seeing bioluminescence.

However, there is virtually no information on the visual pigments of benthic species because of the difficulty in collecting them without exposing their eyes to damaging light levels. Even minimal exposure to light (1 min of daylight) can cause irreparable damage to photoreceptors adapted for dim-light environments (Loew, 1976; Lindström et al., 1988). Photoreceptors of hydrothermal vent shrimp caught under the lights of a submersible also show significant light-induced damage, with disordered microvilli and disruption to photoreceptor membranes (O’Neill et al., 1995; Lakin et al., 1997), whereas those of their pelagic larvae, caught with a net that prevented exposure to light, show orthogonal layers of microvilli with intact membranes typical of crustacean rhabdoms (Gaten et al., 1998). The structural organization of the compound eye of the giant deep-sea isopod Bathynomus giganteus, collected with a dredge and brought up in bright sunlight, showed similar severe disruption of the microvilli, even after 2 months maintenance in constant darkness (Chamberlain et al., 1986). Therefore, it is imperative that these benthic species be collected without exposure to light. Deep-living pelagic species have successfully been collected alive and with intact eyes with the aid of insulated and light-tight containers attached to the end of trawl nets (Childress et al., 1978; Frank and Widder, 1994), but trawling in the benthic environment (other than sparsely populated muddy bottom) with such a device would seriously damage the environment and destroy the net.

Using new technologies developed for the Johnson-Sea-Link (JSL) submersible, the photoreceptor function of eight species of deep-sea benthic crustaceans were examined on a series of research expeditions funded by the NOAA Ocean Exploration program. These studies demonstrated unusual spectral sensitivity in two species of deep-sea crabs, in addition to extraordinarily low temporal resolution in a species of deep-sea isopod, indicative of an integration time so long that this species may not be able to track even slow-moving prey. Preliminary data from this study have been published in abstract form (Frank, 2006; Frank, 2008; Frank et al., 2010).

One species of isopod, two species of caridean shrimp, four species of anomuran crabs and one species of brachyuran crab were collected on a series of research cruises in the Gulf of Mexico and in the Bahamas (Fig. 1, Table 1). Crustaceans were collected with the JSL submersible using several techniques. Baited traps were deployed in several locations, collected 6 to 24 h later under red light and placed into light-tight Bio-Boxes. These containers were constructed out of 1.5 cm thick black Plexiglas with lids fitted with O-ring seals, making the boxes light-tight, water-tight and thermally insulated. Pressure relief valves on the Bio-Boxes released excess pressure as the submersible ascended from the collection depths to the surface. In addition, animals attracted to a bait bag placed near the traps were collected under red light with the suction sampler on the JSL, and deposited directly into the Bio-Box. The red lights used in collections were two submersible lights, one filtered with a Schott OG 590 nm cutoff filter (transmission below 560 nm <0.001%; Dalton, GA, USA) and the other with a HOYA O-58 cut-off filter (transmission below 560 nm <0.1%; THK Photo Products Inc., Huntington Beach, CA, USA). Once back at the surface, the Bio-Boxes were removed from the submersible and brought into a light-tight cold room, where specimens were removed under dim red light, placed in individual light-tight chambers, and maintained at 7°C until used for experiments.

Electrophysiological recordings were made shipboard in a light-tight compartment constructed out of polyvinylchloride and black plastic sheeting, with all preparations carried out under dim red light. Animals were clamped into a holder of the appropriate size and suspended in a recording chamber containing chilled seawater (7°C). With this configuration, electroretinograms (ERGs) were recorded with subcorneal metal microelectrodes (FHC, Bowdoin, ME, USA) from live animals for 2–3 days. Signals were amplified with an X Cell-3 Microelectrode Amplifier (FHC) used in conjunction with a high-impedance probe to eliminate electrode polarization artifacts (Kugel, 1977). Low-frequency filters were set to minimal filtering (0.01–0.1 Hz) to minimize distortion of the AC-amplified signal. Recordings were digitized and stored for later analysis using a data acquisition program written in LabView (National Instruments, Austin, TX, USA).

Monochromatic [full-width half-max (FWHM)=5 nm] test flashes (CM110 monochromator, Spectral Products, Putnam, CT, USA) were conducted to the eye through one end of a bifurcated light guide composed of randomized fused silica fibers (EXFO, Quebec, QC, Canada). The light guide entered through a small slit in the light-tight compartment, and was taped in place with black tape to prevent any stray light from entering through the slit. The other branch of the light guide transmitted light filtered with the appropriate filter (see below) from an Ocean Optics LS-1 lamp (Ocean Optics, Dunedin, FL, USA) to the eye through another small baffled slit during chromatic adaptation experiments, ensuring that both the stimulus and adapting lights were impinging on the same receptors. Flash durations of 100 ms were controlled by a Uniblitz Shutter (Model VS14 S, Vincent Associates, Rochester, NY, USA) under computer control. The stimulus irradiance was controlled with a neutral-density wheel driven by a stepper motor under computer control and calibrated at 10 nm intervals with a UDT Optometer (UDT Instruments, San Diego, CA, USA) and calibrated radiometric probe.

Spectral sensitivity measurements were initiated when the amplitude of the response to a standard dim test flash of constant wavelength and irradiance had not changed for an hour. The irradiance of the stimulus light at each test wavelength (350–600 nm, every 10–20 nm) was adjusted such that a defined criterion response of 20 μV above background noise was obtained (Frank and Case, 1988). This criterion was chosen as it could reliably be measured against background noise (ranging from 10 to 20 μV), but was near the threshold sensitivity of the eye, so that the light flashes would not light-adapt the eye. Stimulus flashes were given at 1 min intervals, with a standard dim test flash presented throughout the experiment to ensure that the physiological state of the eye had not changed (Frank and Widder, 1999).

Chromatic adaptation experiments were conducted by filtering the white adapting light with a 400 nm bandpass filter (ESCO S914000, FWHM=10 nm) or a 486 nm bandpass filter (Melles Griot 03FIR022, FWHM=10 nm). The irradiance was adjusted with neutral density filters such that the sensitivity of the eye at the adapting wavelength was reduced by 1.5 to 2 log units. After the adapting light was turned on, another spectral sensitivity curve was recorded once the response amplitude to a standard test flash had not changed for 1 h.

Spectral sensitivity curves were generated by plotting the inverse of irradiance (photons cm⁻² s⁻¹) required to produce the criterion response at each wavelength. When data from several animals of the same species were available, data from individuals were normalized to the peak wavelength responses before being combined to produce the final curve. These averaged data were best-fit to an A1-based visual pigment absorbance template (Govardovskii et al., 2000), using the Solver function of Excel (Microsoft, Redmond, WA, USA) to determine the lowest residual sum of squares fit to the data. Absorbance was calculated instead of absorptance because of the lack of information on photoreceptor rhabdom path length and visual pigment optical density.

Eyes from the crabs Eumunida picta Smith 1883, Gastroptychus spinifer (A. Milne Edwards 1880) and Bathynectes longipes (Risso 1816) were excised shipboard, quick-frozen with a cryogenic spray (SHUR/Freeze, Triangle Biomedical Sciences, Durham, NC, USA) and stored in a freezer until they could be shipped to a shore-based laboratory for microspectrophotometry (MSP) analysis. Frozen eyes were sectioned at 14 μm on a cryostat, and sections were transferred to microscope slide cover slips in a drop of marine crustacean Ringer’s solution (Cavenaugh, 1956), pH 7.0, containing 1.25% glutaraldehyde to enhance photobleaching of the visual pigment. Photoreceptors were selected under dim, far-red light and scanned from 400 to 700 nm at 1 nm intervals in a single-beam microspectrophotometer using a beam placed in each rhabdom. Rhabdoms were scanned when fully dark-adapted and after full photobleaching under 2 min of bright white light. Absorbance spectra of the rhodopsin in the dark-adapted receptor were taken as difference spectra between the appropriate scans and the scan of the bleached receptor. These difference spectra were fitted to standard rhodopsin templates using a least-squares procedure (Cronin et al., 2002; Cronin and Frank, 1996).

Temporal resolution of the eye was determined in two ways. Firstly, the critical flicker frequency (CFF; also known as the critical flicker fusion frequency) in response to square pulses of light with a constant 0.5 duty cycle (50:50 light:dark ratio) was measured. CFF is the highest stimulus rate at which the eye can produce electrical responses that remain in phase with a flickering light of a certain irradiance. CFF increases as the irradiance of the stimulating light increases (Bröcker, 1935; Crozier and Wolf, 1939; Crozier et al., 1939), until a plateau is reached, at which point further increases in irradiance do not result in further increases in CFF, and may actually result in a decrease (Glantz, 1968; Frank, 1999). The maximum CFF (CFF_max), which is the highest flicker rate that the eye is capable of following at any irradiance, was used to compare temporal resolution between different species. The wavelength of light used for these experiments was 490 nm, at an initial irradiance that generated a 100 μV response to a single flash. Flash duration was controlled by the Uniblitz shutter and the duration of the pulse trains was 2 s. The response to a dim test flash was monitored between every pulse train, and subsequent flickering stimuli were not given until the response to the test flash had recovered to the dark-adapted level. At the brightest flickering stimuli, recovery took up to 1 h. After every pulse train, the output was immediately analyzed visually to determine whether the eye was able to produce a modulated electrical signal that remained in phase with the flickering light for 0.5 s. Flicker stimulus rate was increased by 1–2 Hz for every pulse train until fusion occurred. At this point, irradiance was increased by 0.5 log units, and the flicker rate was increased until fusion was again achieved. The CFF_max was defined as the point at which three successive irradiance increases did not result in a faster flicker frequency. The critical duration (r), which is the time period over which light is summed in the eye (i.e. the integration time of the eye), was estimated as the inverse of CFF_max (Matin, 1968).

Temporal resolution was also determined by measuring the response latency, defined as the time elapsed between the onset of the stimulus flash and the onset of the photoreceptor response. Because response latency decreases as stimulus intensity increases (Moeller and Case, 1995), latencies were calculated for flashes yielding response amplitudes of 50% of the maximum amplitude of the V/logI curve (Frank, 1999; Frank, 2003).

Eight species of deep-sea crustaceans were examined in this study (Table 1) and usable spectral sensitivity curves were obtained from seven of them (Figs 1, 2). Spectral sensitivity curves were judged usable if replicates were available (five species), or the preparation was viable enough to conduct chromatic adaptation experiments in addition to dark-adapted spectral sensitivity, and chromatic adaptation did not shift the spectral peak (two species). Based on these criteria, the spectral sensitivity data from Munidopsis erinacea were not usable. The spectral sensitivity peaks of the seven remaining species were in the blue region of the visible spectrum, ranging from 470 to 497 nm. The spectral sensitivity peaks of the two shrimp species were the longest of all the species tested.

The visual pigments’ absorption peaks were determined via MSP for three species, and are very close to the sensitivity peaks measured electrophysiologically. For G. spinifer, the MSP peak was at 468 nm versus 470 nm measured electrophysiologically; for E., 492 nm versus 490 nm; and for B. longipes, 483 nm versus 478 nm (Fig. 2).

Chromatic adaptation experiments were conducted on all seven species with the 486 nm bandpass filter. This blue chromatic adaptation had no effect on spectral sensitivity in the isopod Booralana tricarinata, the carideans Eugonatonotus crassus and Heterocarpus ensifer, or two of the crab species, Munidopsis tridentata and B. longipes. Chromatic adaptation reduced overall sensitivity by 1.5 to 2 log units, but did not affect the shape or peak of the spectral sensitivity curve. The curves for B. longipes are given in Fig. 3A as a representative example for these five species, plotted on a log scale so that both curves could be plotted on the same graph. However, the effects of blue chromatic adaptation on the photoreceptors of E. picta and G. spinifer revealed the presence of secondary peaks in the ultraviolet portion of the spectrum – at 363 and 383 nm, respectively (Fig. 3B,C). To demonstrate that the effects of light adaptation were specific to blue adaptation, spectral sensitivity curves under chromatic adaptation with a 400 nm bandpass filter (FWHM=20 nm) were also recorded. In both species, adaptation with this violet light had no wavelength-specific effects on spectral sensitivity: overall sensitivity decreased but the wavelength of the peak response remained the same.

Response waveforms to short wavelength (370 nm) and long wavelength (570 nm) stimuli were identical in the dark-adapted eyes of all the species (Fig. 3D,F,H). Blue chromatic adaptation had no wavelength specific effects on the five species whose spectral sensitivity was not affected by this adaptation light. However, in G. spinifer and E. picta, the two species in which blue chromatic adaptation revealed a UV sensitivity peak, a wavelength-specific effect on the response waveforms is also evident (Fig. 3G,I). In G. spinifer, under blue chromatic adaptation, the mean (±s.e.m.) time to peak for responses to short-wavelength light (360–390 nm) was 193±3 ms (N=4) compared with 180±4 ms (N=5) for long-wavelength light (560–590 nm). For E. picta, the times to peak were 215±11 ms (N=6) and 176±7 ms (N=5) for short- and long-wavelength light, respectively. These differences were statistically significant for both species (P=0.042 and 0.009, respectively). In contrast, adaptation with violet light did not affect the response waveforms in either species.

Because temporal resolution depends upon temperature, the following data were obtained from animals that had been maintained at 7°C throughout the experiment. In addition, temporal resolution also depends on the irradiance of the test flash. While the irradiance of the light source was calibrated, there was no way to ensure, when working on live animals suspended in a chilled water bath, that the tip of the light guide was in the identical position for each preparation. Therefore, only CFF_max, which is the highest flicker rate that the eye is capable of following at any intensity, was used for species comparisons. Flicker fusion data could not be obtained from M. tridentata. As shown in Table 2, the CFF_max was highest for the two caridean shrimp species, ranging from 16 to 24 Hz (Fig. 4A). The crabs, both anomuran and brachyuran, had somewhat lower temporal resolutions, with CFF_max ranging from 10 to 14 Hz. The isopod B. tricarinata had a CFF_max of 4 Hz, the lowest ever measured in a marine crustacean (Fig. 4B).

Response latency, an indication of the integration time of the eye, is also a function of irradiance, so response latencies were determined for responses whose amplitudes were 50% of the V_max (as determined by plotting the V/logI curves; Fig. 4C). V_max is the largest response that the eye is capable of generating at any irradiance; once V_max is reached, a plateau is seen in the V/logI curves. V_max was not reached in the shrimp H. ensifer or the crab M. tridentata, and therefore the 50% V_max could not be determined for these species. As seen in Table 2, response latency mirrored the CFF_max results: the species with the highest CFF_max (the two shrimp species) also had the shortest response latencies.

K, the irradiance required to generate a response that is 50% of the V_max, has been used an indicator of absolute sensitivity in intracellular studies on insects (Laughlin, 1976; Autrum, 1981), and an indicator of relative sensitivity when utilizing the ERG (Eguchi and Horikoshi, 1984; Frank and Case, 1988). As expected, the isopod B. tricarinata, which had the slowest eye, had half a log unit greater sensitivity than any of the other species (Table 2). The crabs had considerably less sensitive eyes with concurrent faster response latencies. The apparent anomaly was the caridean shrimp E. crassus, which had the fastest eye with the fastest response latency, but was second in sensitivity to the isopod.

The dark-adapted spectral sensitivity data from all the benthic species in this study showed a sensitivity maximum in the blue region of the spectrum, between 470 and 497 nm. For the three species in which the visual pigments were analyzed via microspectrophotometry, the absorbance peak of the visual pigment was remarkably close to the spectral sensitivity peak, suggesting that there was not much pre-retinal filtering occurring. The hypsochromatic shift to bluer wavelengths, compared with the 490–550 nm visual pigments found in shallow-water crustaceans (reviewed in Goldsmith, 1972; Cronin, 1986), is similar to what has been found in most species of mesopelagic crustaceans (Frank and Case, 1988; Frank and Widder, 1999; Cohen and Frank, 2007) and fish (reviewed in Douglas et al., 2003). This sensitivity shift to bluer wavelengths would confer greater sensitivity to both the remaining downwelling irradiance as well as bioluminescence, which is considered by some to be the major visual stimulus in the deep sea (Beebe, 1935; Clarke and Hubbard, 1959; Jerlov, 1976; Widder, 1999; Johnsen, 2005).

Blue chromatic adaptation had no effect on spectral sensitivity in five of the species in this study: the isopod B. tricarinata, the caridean shrimp E. crassus and H. ensifer, and the crabs M. tridentata and B. longipes. These results indicate that a single, blue-sensitive visual pigment is present in these species, similar to what has been found for most species of deep-sea pelagic crustaceans (Frank and Case, 1988; Frank and Widder, 1999; Cohen and Frank, 2007; Frank et al., 2009) as well as the only two deep-sea benthic crab species that have been studied to date via MSP [Geryon quinquedens (Cronin and Forward, 1988) and Bythograea thermydron (Jinks et al., 2002)]. Although the results of experiments involving a sample size of one (H. ensifer and M. tridentata) should be discussed with caution, in both cases, the animals were alive at the end of the experiment, and the adapting light depressed sensitivity by over 1.5 to 2 log units, which is sufficient to reveal the presence of a secondary sensitivity maximum if one is present (Frank and Case, 1988; Frank and Widder, 1999).

Unexpected responses to blue chromatic adaptation were discovered in the other two crab species, E. picta and G. spinifer. Blue chromatic adaptation revealed a secondary sensitivity peak at 363 and 383 nm, respectively. In addition, violet adaptation had no wavelength-specific effects on spectral sensitivity. These results can be explained if, as in deep-sea pelagic crustaceans with two visual pigments, the blue-sensitive cells vastly outnumber the ultraviolet-sensitive cells (Frank and Case, 1988; Cronin and Frank, 1996; Gaten et al., 2003). In this situation, blue chromatic adaptation would diminish the contribution of the blue-sensitive cells to the ERG response, unmasking the presence of the minority ultraviolet-sensitive cells. Conversely, violet chromatic adaptation would only diminish the already small contribution of the ultraviolet-sensitive cells to the ERG response, and no change in spectral sensitivity would be visible.

The conclusion that there are two classes of receptor cells is supported by the wavelength-specific effects of blue adaptation on the response waveforms. In the dark-adapted eyes of all species, the time to peak response at all wavelengths was identical within a species. The five species that showed no spectral sensitivity shift in response to blue adaptation also had identical times to peak at all wavelengths under blue adaptation. However, for the two species in which blue adaptation uncovered a secondary sensitivity peak at the ultraviolet portion of the spectrum, the time to peak was significantly longer at the shorter wavelengths than at the longer wavelengths. Again, if two classes of receptor cells are present in different quantities, blue chromatic adaptation would diminish the contribution of the blue-sensitive receptor cells to the ERG response, and the contribution of the ultraviolet-sensitive cells would become visible in response to short-wavelength light. Single cells do not respond differentially to different wavelengths of light (Graham and Hartline, 1935; Naka and Rushton, 1966; Stark and Wasserman, 1974), and if several different receptors with different time courses contribute to the ERG, equal amplitude responses at all wavelengths can never be matched (Chapman and Lall, 1967). Therefore, differences in response waveforms to short- versus long-wavelength light in crustaceans can only be attributed to two different populations of receptor cells with different membrane properties (Wald, 1968).

Although a second visual pigment was not found during the MSP analysis of G. spinifer and E. picta photoreceptors, the same situation occurred for the initial discovery of a short-wavelength-sensitive visual pigment in several species of mesopelagic shrimp in the family Oplophoridae. Electrophysiological studies indicated the presence of two receptor classes (Frank and Case, 1988), whereas the initial MSP analysis found only a single visual pigment (Hiller-Adams et al., 1988). It took a re-examination of fresh-frozen material (which was not available in the present study) to detect the presence of the second visual pigment in Systellaspis debilis (Cronin and Frank, 1996).

Sacrificing retinal space for a UV photoreceptor suggests that UV sensitivity plays an important role in the visual ecology of these species, and at these depths (400–600 m), UV sensitivity is most likely related to bioluminescence. As mentioned above, several species of mesopelagic oplophorid shrimp have both UV-sensitive and blue-sensitive visual pigments. Those species with the dual visual pigment system possess both light-emitting organs called photophores, which are used for counterillumination and hypothesized to play a role in sexual signaling (reviewed in Herring, 2007) and a bioluminescent spew (i.e. vomit), which probably serves a defensive function (Herring and Barnes, 1976). Further analysis by Cronin and Frank (Cronin and Frank, 1996) indicated that the absorption maxima of the two pigments may be optimized for distinguishing between the broader emission spectrum of the spew and the narrower emission spectrum of the photophores. The other species in this family possess the bioluminescent spew but lack photophores, and those that have been studied to date have a single blue-sensitive visual pigment (Frank and Case, 1988), supporting the hypothesis that the dual visual pigment system is used to distinguish between different bioluminescent ‘colors’.

This cannot be the case for the deep-sea crabs, as neither species is bioluminescent. However, at the Little Bahamas Bank Lithoherms, both species of crabs were frequently seen on pennatulaceans (sea whips) or zoanthids (colonial anemones known as gold corals) (Fig. 5A,B), both of which emit a greenish bioluminescence (Johnsen et al., 2012). In a remarkable long exposure image taken by S.J. from the front sphere of the JSL submersible, the green bioluminescence of the zoanthid (formerly Gerardia sp.) is clearly distinguishable from the blue luminescence of the bioluminescent plankton striking it (Fig. 5C). A series of observations (under both red and light white) and videos (under white light) made from the JSL indicate that the crabs are stationary on these benthic structures for long periods of time, and then periodically use their claws to pick organisms off the structure and bring them to their mouths. Eumunida picta has also been observed in association with other benthic structures that would impose a similar impediment to plankton flowing by in the water column. In the Mississippi Canyon in the northern Gulf of Mexico, 81% of the >100 E. picta observed during the study occurred on the deep coral Lophelia pertusa (Kilgour and Shirley, 2008). Although there are clearly other possibilities to explain the relationship between the crabs and stationary benthic structures that project up into the water column, one intriguing possibility is that these crabs may be using their putative dual visual pigment systems to distinguish the green benthic bioluminescence of their preferred substrate, which they apparently do not eat, from the blue bioluminescence of the planktonic organisms impacting these structures, which may be their preferred food source. This hypothesis is supported by measurements of emission spectra of gelatinous zooplankton, the most likely organisms to be impacting benthic structures. These emission spectra are considerably bluer than the bioluminescence from benthic animals, with a mean (±s.e.m.) λ_max of 473.8±2.9 nm for medusae (N=34), 486.1±1.6 nm for ctenophores (N=41), and two modes for siphonophores, centered at 450.5±1.3 nm (N=16) and 486±2.3 nm (N=9) (Haddock and Case, 1999). Lophelia pertusa is not bioluminescent (Johnsen et al., 2012), but would still provide a raised substrate for the plankton to impact against, and a dual visual pigment system may help enhance contrast between the bioluminescence and the remaining bluish downwelling light as well.

The temporal resolution of most of the species was within the range of what one would anticipate for deep-sea species, based on previous studies utilizing the CFF_max techniques (reviewed in Marshall et al., 2003). All of the species had relatively low temporal resolution compared with shallow species, with that of the more mobile shrimp approximately equal to that of mesopelagic decapod shrimp (Frank, 1999), at ~20 Hz. The 10–14 Hz found in the crabs is considerably lower than that of the shrimp, which could be a function of their less mobile lifestyle, and fits well with Autrum’s hypothesis that the response dynamics of the retina match the habitat and lifestyle of the organism (Autrum, 1958). This is similar to the situation demonstrated in the only other physiological study of deep-sea benthic species, which used intermediate light levels (Johnson et al., 2000). Because temporal resolution depends so strongly on the irradiance of the stimulus light (Bröcker, 1935; Crozier and Wolf, 1939; Crozier et al., 1939; Frank, 2000), results from that study cannot be directly compared with results from the studies utilizing the invariant CFF_maxtechnique, but within the present study, the same correlations were seen. The eyes of the slower-moving scavenger crab species had slower response dynamics than the eyes of the deeper-living crab species whose stomach contents indicated it was an active fast-moving predator.

The real surprise was the isopod B. tricarinata. With a CFF_max of only 4 Hz, it has the slowest eye measured thus far in a crustacean, with the next slowest being found in the deep-sea crabs in the present study. The integration time, the time period over which incoming photons are summed, could not be directly measured using the ERG technique, but the critical duration, which is the inverse of the CFF_max (Matin, 1968), can be directly related to integration time, such that a long critical duration indicates a long integration time (de Souza and Ventura, 1989). The critical duration was 250 ms for these isopods, more than twice that of the crabs and shrimps, with ranges from 42 to 105 ms (Table 2). A further indication of the very long integration time is the response latency, as a long response latency is an indication of a long integration time. The response latency for the isopods was more than twice as long as those of the other species in this study, as well as those of deep-sea pelagic crustaceans that have been examined using the same methodology (Frank, 2003; Cohen and Frank, 2006; Cohen and Frank, 2007). This long integration time would give this eye very high sensitivity, but likely limited ability to track moving prey or separate fast visual stimuli, such as rapid flashes of bioluminescence (Warrant, 1999). One intriguing possibility for the adaptive significance of this extremely slow eye has to do with bioluminescence and food preferences of isopods. Lampitt et al. (Lampitt et al., 2001) suggested that bacterial luminescence may be present on aggregations of organic matter deposited on the deep-sea floor, based on previous observations that luminous bacteria are present on crustacean and fish carcasses (Wada et al., 1995) and that fecal pellets in sediment traps have been observed to luminesce (Andrews et al., 1984). This colonization of organic matter by bioluminescent bacteria may produce a background glow that could be used by deep-sea scavengers (Nishida et al., 2002), if their eyes were sensitive enough to see it. Deep-sea isopods are thought to be scavengers (reviewed in Barradas-Ortiz et al., 2003), and an eye with an extremely long integration time might be able to visualize dimly glowing detritus, aiding the animal in finding a target initially tracked via chemoreception.

An interesting anomaly is seen in the caridean shrimp E. crassus, which has the shortest response latency amongst the benthic species, generally indicative of a less sensitive eye, but this species was second in sensitivity to the isopod. However, sensitivity is a function not only of temporal resolution, but also of other factors such as size of the eye, interommatidial angle, facet size, length of the photoreceptor, amount of visual pigment present, etc. (reviewed in Land and Nilsson, 2002). This species had the largest stalked eye of all the species in the present study, and further studies are currently underway to determine whether there are other optical adaptations present to increase light capture.

In summary, this study of only a few of the denizens of the deep-sea benthic environment demonstrated some unexpected physiological adaptations to the deep-sea light environment. Although the number of benthic bioluminescent species is relatively rare in the location occupied by many of the crustaceans in this study (Johnsen et al., 2012), evidence suggesting that at least two species of deep-sea anomuran crabs possess both UV- and blue-sensitive visual pigments indicates that bioluminescence still may have been a driving force behind their visual adaptations. Further studies are needed to determine whether UV sensitivity is common in the deep-sea Chirostyloidea, the superfamily to which E. picta and G. spinifer belong, as well as other deep-sea ‘stalk sitters’. Because so little work has been done on deep-sea benthic bioluminescence, there are, in all probability, many unknown sources of benthic bioluminescence yet to be discovered, including the intriguing possibility of dimly glowing mats of marine snow and detritus on the seafloor bottom, as suggested by the remarkably slow eye of the isopod B. tricarinata.

We thank the captain and crew of the RV Seward Johnson, and the pilots and crew of the deep-sea submersible Johnson-Sea-Link II. We also thank two anonymous reviewers for comments on earlier drafts of the manuscript.