ARTICLE IN PRESS Deep-Sea Research I 56 (2009) 1003–1017 Contents lists available at ScienceDirect Deep-Sea Research I journal homepage: www.elsevier.com/locate/dsri Vision in lanternfish (Myctophidae): Adaptations for viewing bioluminescence in the deep-sea J.R. Turner a,, E.M. White a, M.A. Collins b, J.C. Partridge a, R.H. Douglas c a b c School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, UK British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 OET, UK Department of Optometry and Visual Science, City University, Northampton Square, London EC1V 0HB, UK a r t i c l e in fo abstract Article history: Received 10 June 2008 Received in revised form 5 January 2009 Accepted 15 January 2009 Available online 29 January 2009 The sensitivity hypothesis seeks to explain the correlation between the wavelength of visual pigment absorption maxima (lmax) and habitat type in fish and other marine animals in terms of the maximisation of photoreceptor photon catch. In recent years its legitimacy has been called into question as studies have either not tested data against the output of a predictive model or are confounded by the wide phylogeny of species used. We have addressed these issues by focussing on the distribution of lmax values in one family of marine teleosts, the lanternfish (Myctophidae). Visual pigment extract spectrophotometry has shown that 54 myctophid species have a single pigment in their retinae with a lmax falling within the range 480–492 nm. A further 4 species contain two visual pigments in their retinae. The spectral distribution of these visual pigments seems relatively confined when compared to other mesopelagic fishes. Mathematical modelling based on the assumptions of the sensitivity hypothesis shows that, contrary to the belief that deep-sea fishes’ visual pigments are shortwave shifted to maximise their sensitivity to downwelling sunlight, the visual pigments of myctophids instead seem better placed for the visualisation of bioluminescence. The predicted maximum visualisation distance of a blue/green bioluminescent point source by a myctophid was up to 30 m under ideal conditions. Two species (Myctophum nitidulum and Bolinichthys longipes) have previously been shown to have longwave-shifted spectral sensitivities and we show that they could theoretically detect stomiid far-red bioluminescence from as far as ca. 7 m. & 2009 Elsevier Ltd. All rights reserved. Keywords: Visual pigments Bioluminescence Lanternfish Myctophidae Spectral sensitivity Visualisation distance 1. Introduction The light environment of the mesopelagic zone of the ocean (200–1000 m depth) is dominated by the spectral attenuating and scattering properties of water (Lythgoe, 1979). With increasing depth not only does the intensity of sunlight fall in an approximately exponential manner, but the spectral quality of this light also changes, becoming increasingly restricted to a narrow waveband  Corresponding author. Tel.: +44 117 954 5911. E-mail address: jonny.turner@bristol.ac.uk (J.R. Turner). 0967-0637/$ - see front matter & 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr.2009.01.007 of light (470–490 nm). This was described by Beebe (1935), whose account conveys how the red, orange, yellow and green components of sunlight vanished from the background during his descent into the mesopelagic zone, leaving a predominately blue environment which eventually faded to complete darkness. The decrease in sunlight with depth is met with an increase in the relative importance of bioluminescent point sources produced by marine fauna. These are variable in intensity, having a radiant photon flux ranging from 107 to 1013 photons s1 (Warrant and Locket, 2004), spectral emission maxima (lImax) which largely falls in the range 450–510 nm (Herring, 1983), and in spatial and temporal distribution ARTICLE IN PRESS 1004 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 (Latz et al., 1987; Widder, 2002). The relative importance of bioluminescence also varies over the course of a day; at night it is the only light source available for vision within the mesopelagic environment. A well-established correlation in visual ecology describes the relationship between the absorption maxima of visual pigments (lmax) and the photic environment of fish and other marine fauna. Coincidently, the first references to this correlation occurred twice in the 1930s, just after Beebe’s account of the light and life in the mesopelagic was published: Clarke (1936) raised the question of ‘‘ythe possibility of a shift in the sensitivity of the eye of a deep-water fish toward the blue end of the spectrum’’, while Bayliss et al. (1936) postulated that ‘‘the most probable explanation [of the spectral distribution of visual pigments] would seem to be in terms of the depth or kind of water favoured by the species’’. However, the first pieces of evidence supporting these speculations, which became known as the ‘sensitivity hypothesis’, were not obtained until 20 years later (Denton and Warren, 1956, 1957; Munz, 1958). The rationale behind the hypothesis is that the visual environment of the ocean is photon-limited and therefore there will be selective pressure to maximise sensitivity to the available light. In terms of visual pigments this means shifting the lmax to a wavelength that closely matches the dominant wavelengths of the ambient environment. Over the last 50 years a wealth of visual pigment data have been accumulated and these have sometimes been interpreted as supporting the sensitivity hypothesis (see Crescitelli, 1991 for brief review). In recent years, however, the legitimacy of this hypothesis has been called into increasing question (Douglas et al., 1998a). Furthermore, it has been suggested that the distribution of visual pigments in deep-sea fish actually seems to place them better for the visualisation of bioluminescence rather than downwelling sunlight (Douglas et al., 1998a). Previous research is also potentially confounded by the wide phylogeny of fish studied and it is therefore impossible to determine whether the observations are an adaptation to the environment or the result of evolutionary history. The latter problem has been overcome for shallow water fish by focussing on a single taxon such as snappers (Lutjanidae; Lythgoe et al., 1994) and surfperch (Embiotocidae; Cummings and Partridge, 2001), but this approach has not been used with respect to deepsea fish. Here we examine the visual pigments of a single deep-sea teleost family, the lanternfish (Myctophidae), testing these data against the output of predictive models. The phylogeny of myctophids, based on morphological characters, is well described comprising about 230–250 species in 30–35 genera (Stiassny, 1996). They are widely distributed across the world’s oceans and have an estimated global biomass of 600 million tonnes (GjØsaeter and Kawaguchi, 1980) which, when considered alongside their distribution, both in global terms and their position in the water column (many being diel vertical migrators), means that myctophids play a prominent role in the ecology and energy transfer within the mesopelagic zone of the ocean (Collins et al., 2008). Despite their ecological significance and their large biomass remarkably little is known about the basic biology and ecology of lanternfish. In terms of testing hypotheses in visual ecology, it was important to choose a family that inhabited the mesopelagic zone because the greatest variation in the visual systems of deep-sea marine fauna is found in animals inhabiting this region of the ocean (Warrant and Locket, 2004). This is primarily because the mesopelagic has a depth-related gradient in light available for vision, being dominated (in daytime) by extended sources of light in the upper regions and bioluminescent point sources of light in the deepest parts (Warrant, 2000), with the nature of the visual environment and associated visual tasks (Partridge and Cummings, 1999) changing continuously between these two extremes. Vision clearly plays an important role in the life of myctophids, with all species possessing well-developed eyes, and light organs potentially used for conspecific identification, sexual signalling and counterillumination for camouflage (Edwards and Herring, 1977; Widder, 1999; Johnsen et al., 2004). Surprisingly, given their importance within the deep-sea ecosystem, the visual systems of myctophids have been investigated to a lesser degree than other families. The majority of the limited previous data suggest that most, like the majority of deep-sea fish, have a single visual pigment maximally sensitive in the blue/green part of the spectrum (Douglas and Partridge, 1997) approximately matching both the residual sunlight and the maximum emission of most deep-sea bioluminescence. However much short wavelengths predominate in the deep-sea, they are not the only source of illumination. Three genera of stomiid dragonfish have suborbital photophores with emission maxima beyond 700 nm. To enable them to see their own far-red bioluminescence these animals have longwave-shifted visual pigments within their retina (Aristostomias tittmanni, O’Day and Fernandez, 1974; Partridge and Douglas, 1995; Pachystomias microdon, Douglas et al., 1998b) which in the case of Malacosteus niger, are additionally coupled to a bacteriochlorophyll-derived photosensitiser (Douglas et al., 1998b, 1999). Interestingly, one myctophid species (Myctophum nitidulum) has recently been shown to have longwave shifted visual pigments compared to other lanternfish (Hasegawa et al., 2008). Another, Bolinichthys longipes, has also been shown to have a photostable pigment in its retina absorbing maximally around 670 nm, which is probably a chlorophyll derivative and could potentially act as a photosensitiser and, like in M. niger, confer longwave photosensitivity (Douglas et al., 2002, 2003). Since stomiids are sensitive to their own far-red bioluminescence, which other animals in the deep-sea with ‘conventional’ visual pigments cannot see, they have what could be regarded as a ‘private waveband’, which could be used for intraspecific signalling immune from detection by potential predators, or for the covert illumination of prey. As myctophids are predominant components of the diet of Aristostomias and Pachystomias (Sutton and Hopkins, 1996) it would be interesting to know whether M. nitidulum or B. longipes could potentially detect stomiid far-red bioluminescence, giving them a selective advantage over other myctophid species in terms of predator detection. ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 The aim of this study was to measure the distribution of lmax values of the visual pigments of myctophids. Prior to this study the visual pigments of 21 species, caught mainly in the North Atlantic, had been characterised. To reflect more accurately the pan-global distribution of this family we sampled a total of 40 species, the visual pigments of 36 of which had not been previously characterised, from the North and South Pacific as well as the South Atlantic. The distribution of their lmax values was analysed against predictions from models based on the assumptions of the sensitivity hypothesis and the visualisation of bioluminescence. One aspect of this modelling was to predict a theoretical maximum visualisation distance for a myctophid viewing a bioluminescent point source of a potential predator, prey item or conspecific. In the case of M. nitidulum and B. longipes the maximum visualisation distance of stomiid far-red bioluminescence was also modelled. Such predictions are interesting when contemplating general myctophid ecology and their interactions with other individuals within the mesopelagic environment. 1005 rescanned between each bleach. One final exposure for 1 min in white or green (501 nm) light was used to ensure complete bleaching of any remaining visual pigment. As shown by Knowles and Dartnall (1977), if the extract contains two or more visual pigments, a difference spectrum of the initial dark scan and the first longwave bleach reveals an absorption spectrum dominated by the more longwave-sensitive visual pigment, whilst the difference spectrum of the last shortwave bleach and the final bleach reveals an absorption spectrum dominated by the most shortwave-sensitive visual pigment. If only one visual pigment is present in the extract both difference spectra coincide. The difference spectra were best-fitted with the visual pigment templates of Govardovskii et al. (2000) using methods described by Hart et al. (2000) to determine the lmax of the visual pigments in the extracts. An example of a partial bleaching series, with associated difference spectra and template, is shown in Fig. 1. 2. Materials and methods Most of the data included in this study were obtained during three research cruises aboard the FS Sonne to: the Musician’s Seamount region of the Pacific Ocean north of Hawaii in 1999 (SO142 Hula II), the Pacific west of Costa Rica and Guatemala in 2003 (SO173-2) and the waters above the Kermadec and Tonga trenches in the south Pacific in 2007 (SO194). Further collections were made aboard the RRS James Clark Ross in 2004 (JR100) in the South Atlantic northwest of South Georgia. Animals were caught using a rectangular midwater trawl, usually fitted with a light-tight closing cod end. Individual specimens, generally dead on removal from the net, were immediately transferred to a darkroom in iced seawater. The eyes were removed and hemisected under dim far-red illumination. The retinae were dissected out and either placed in PIPES-buffered saline (PIPES 20 mM, NaCl 120 mM, KCl 19 mM, MgCl2 1 mM, CaCl2 1 mM; 300 mOsm/kg; pH 6.5) or TRIS-buffered saline (TRIS 50 mM, NaCl 150 mM, pH 7.6). Some retinae were used immediately, whilst others were stored in light-tight containers at 25 1C for later analysis at sea or on land. 2.1. Extract spectrophotometry Visual pigments were extracted in 1 ml PIPES-buffered or TRIS-buffered saline and 100 ml of 200 mM n-dodecyl b-D-maltoside, a mild detergent (De Grip, 1982), under dim far-red illumination, and subjected to partial bleaching (see Douglas et al., 1995 for details). Briefly, 5 ml of 1 M pH 6.5 hydroxylamine (NH2OH) was added to 150 ml of dark adapted extract and scanned either in a Shimadzu UV2101-PC or a USB2000 spectrophotometer. The sample was exposed to a series of bleaches using monochromatic light of decreasing wavelength, from a regulated AC light source combined with narrow band interference filters (10 nm bandwidth; B40 filters, Balzer, Liechtenstein) and Fig. 1. Absorbance spectra of a Diaphus longleyi retinal extract during a partial bleaching experiment. (a) Scans showing the effect of a series of bleaches by exposure to monochromatic light. In order of decreasing absorbance at 500 nm the absorbance spectra are: (i) initial measurement of the unexposed extract 10 min after the addition of 5 ml 1 M hydroxylamine to 150 ml of extract; (ii) 656 nm 20 min; (iii) 625 nm 15 min; (iv) 624 nm 12.5 min; (v) 595 nm 10 min; (vi) 585 nm 10 min; (vii) 575 nm 10 min; (viii) 560 nm 7.5 min; (ix) 560 nm 5 min; (x) 543 nm 3 min; (xi) 501 nm 1 min; (xii) white light 1 min. (b) Difference spectra calculated from data in (a): thick solid line ¼ (i)–(ii), lmax ¼ 491.6; dotted line ¼ (xi)–(xii), lmax ¼ 490.8; thin solid line ¼ template bestfitted to (i)–(ii) data as described in methods. The two difference spectra shown have very similar lmax values, therefore only one visual pigment was present in this extract. ARTICLE IN PRESS 1006 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 2.2. Further analyses The spectral transmission (300–700 nm) of lenses from seven species (Electrona antarctica, E. carlsbergi, Gymnoscopelus bolini, G. nicholsi, Protomyctophum bolini, P. choridon and P. gemmatum, Fig. 2) was measured according to methods of Douglas and McGuigan (1989) with the following modifications: spectral transmission was measured with a Shimadzu UV2101-PC spectrophotometer and the lenses were mounted in a hole drilled in an aluminium billet which fitted into the sample cuvette. Depth data used in analyses were taken from FishBase (Froese and Pauly, 2008). The mean deepest depth for each species was calculated from the tabulated occurrence datasets and used in all analyses. This was chosen for two reasons: firstly, as myctophids are diel vertical migrators we would expect the deepest depth to be representative of their daytime location when they will be viewing sunlight, and secondly the shallowest depth data is heavily skewed by zeros (as open nets are often recorded as fishing from the surface to their greatest depth) and so cannot be used as a reliable depth data source. A comparative analysis of the data, studying the intergroup variation amongst the myctophids, was performed as set out by Harvey and Pagel (1991). Briefly, this analysis creates a set of independent comparisons of the ecological and physiological data, but incorporating a known phylogeny. Appropriate statistical analyses can then be employed to determine whether the variation seen can be attributed to an adaptation or shared evolutionary history. 2.3. Mathematical modelling For the purposes of modelling various visualisation tasks, calculations were based on a single ‘standard myctophid’. The model outcome depends upon the size and shape of the eye. For simplicity we have based the models on a single eye size typical of myctophids: a pupil area of 10 mm2, giving a pupil, and therefore an approximate lens, diameter of 3.57 mm and a focal length of Fig. 2. Transmission spectra of the ocular lenses of seven species of myctophids, diameter 2.5 mm. All lines show the average of two transmission spectra, except for Protomyctophum choridon which is a single transmission spectrum. Species measured (left to right at 0.5 normalised transmission): Electrona antarctica, Protomyctophum gemmatum, P. bolini, E. carlsbergi, P. choridon, Gymnoscopelus nicholsi, G. bolini. 4.55 mm, the latter calculated according to Matthiessen’s Ratio (focal length divided by the lens radius, a ratio of 2.55 being typical for fish, Land and Nilsson, 2002). The importance of eye shape and size on the model’s outcomes will be examined in future studies. At all times the theoretical maximum visualisation distance is estimated in all modelling outputs. 2.3.1. Visualisation of downwelling sunlight The spectrum of downwelling sunlight (Ez(l)) available at any depth (z) in the ocean was calculated using the following relationship: EzðlÞ ¼ E0ðlÞ eðK ðl;zÞ zÞ (1) where E0(l) is the spectral irradiance just below the surface of the water and K(l,z) is the depth-related spectral diffuse attenuation coefficient (m1). K(l,z) was calculated using the following relationship from Morel and Maritorena (2001, Eq. (3) and (5), respectively): K ðl;zÞ ¼ K wðlÞ þ K bioðl;zÞ g K bioðl;zÞ ¼ wðlÞ C ðzÞðlÞ (2a) (2b) where Kw(l) are the spectral diffuse attenuation coefficients for pure seawater, Kbio(l,z) the depth-dependent spectral diffuse attenuation coefficients describing the contributions of all the biogenic components of seawater and C(z) is the concentration (mg m3) of chlorophyll at depth z. The values of Kw(l), g(l) and w(l) are taken from Table 2 in Morel and Maritorena (2001). Using this model, different water types can be simulated by altering the depth-related chlorophyll profile and thus the value of C(z). For the purposes of this study chlorophyll profiles for the Sargasso Sea were taken from the 2003 website dataset of the Bermuda Bio-optics Project (BBOP). An average depth profile was calculated and subsequently fitted with a Gaussian function using CurveExpert v1.38 (Hyams, 2003), of the form 2 C ðzÞ ¼ aeðzbÞ =2c2 (3) Chlorophyll profiles of three other Seas of increasing attenuation were taken from the literature (Mediterranean, Antarctic and Celtic from: Estrada et al. (1993); Kimura and Okada (1997); Platt and Sathyendranath (1988), respectively). The coefficients a, b and c in Eq. (3) were altered to best-fit these profiles using CurveExpert v1.38 (Hyams, 2003). The coefficients for all four seas are shown in Table 1 and the resulting chlorophyll profiles shown in Fig. 3. For the purposes of this study different water types shall be referred to by name: Sargasso, Mediterranean, Antarctic and Celtic. Although naming four specific locations, these are representative water types that, when compared to SeaWIFS data (http://seawifs.gsfc.nasa.gov/SEAWIFS.html), encompass 80% of the world’s oceans as well as approximating to Jerlov’s classification (Type 1a, 1b, 2 and 3, respectively; Jerlov, 1976). At depths below the chlorophyll peak and where the concentration of chlorophyll (C(z))o0.001 mg m3 in ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 Table 1 Coefficients used in bio-optical modelling of four water types of increasing attenuation (Sargasso, Mediterranean, Antarctic and Celtic Seas). Coefficient Units Water type Sargasso Sea a b c g C mg m3 0.3 m 58 m 50 1 0.006 m 3 mg m 0.146 Mediterranean Sea Antarctic Sea Celtic Sea 0.7 75 40 0.0125 0.31 3.5 30 22 0.0325 1.35 6 17 15 0.036 2.79 Coefficients a, b and c are used in Eq. (3) to create chlorophyll profiles. Coefficient g used in Eq. (4b), when chlorophyll concentration o0.001 mg m3. Coefficient C, the average chlorophyll concentration for each water type, is used in Eq. (10). where d is the pupil diameter (for the purposes of this study taken to be 3.57 mm), f is the corresponding focal length of the eye (4.55 mm) and T(l) is the spectral transmission of the lens, here taken as an average of data in Fig. 2. Using any of the data or the mean from Fig. 2 does not affect the outcome of the model. Other ocular media (cornea, aqueous and vitreous humour) were assumed to be transparent down to 300 nm (Douglas and Thorpe, 1992). Visual pigment normalised absorbance spectra were created using the equations of Govardovskii et al. (2000). These were multiplied by the specific absorbance of rhodopsin (assumed to be 0.013 mm1, Partridge et al., 1988) and a pathlength of 50 mm, effectively the length of the photoreceptor outer segment typical of a myctophid rod (O’Day and Fernandez, 1976), to produce an absorbance spectrum (B(l)). A single pathlength value was used for simplicity. The importance of varying pathlengths on the model’s outcomes will be examined in future studies. This was converted to an absorptance spectrum (P(l)) using the following: P ðlÞ ¼ 1  10BðlÞ Fig. 3. Chlorophyll profiles used in the bio-optical modelling of four water types of increasing attenuation (Sargasso, Mediterranean, Antarctic and Celtic Seas). Profiles produced using Eq. (3) and the coefficients a, b and c in Table 1. Eq. (2a) and (2b), the model switches to: K ðl;zÞ ¼ K wðlÞ þ K gðlÞ (4a) where K gðlÞ ¼ ge0:014ðl440Þ p d2 4 f2 T ðlÞ (6) Such absorptance spectra were multiplied by the calculated values of Er(l) to give the spectral photon catch rate of an area of retina, total photon catch rate (ph s1 m2) being the integral of those values over the entire spectrum. The above model was iterated over a range of depths (0–1000 m in steps of 10 m) and a range of visual pigment lmax values (350–600 nm in steps of 1 nm) for each water type. At each depth the visual pigment with the highest summed photon catch was taken to be the most sensitive to the downwelling sunlight. The threshold depth at which a fish can see downwelling sunlight, in the clearest ocean water at midday in the tropics, has been taken to be ca. 1000 m (Denton, 1990). The calculated photon catch rate of a retina containing the most sensitive visual pigment at 1000 m in the Sargasso Sea was calculated as the visual threshold limit in this study (109 photons s1 m2). (4b) Eq. (4b) is taken from Morel and Maritorena (2001; Eq. (17) in Appendix B). The values of g for each water type are shown in Table 1. The g-value for the Mediterranean Sea was estimated using the Solver addin software in Microsofts Excel 2003 by minimising the RMS so that our estimated K(l,z) values were best-fitted to the diffuse attenuation coefficients measured in the Mediterranean by Riccobene et al. (2007). The resulting estimated g-value for the Mediterranean Sea was 0.00018 m2 mg1 times the integral of the Mediterranean chlorophyll concentration profile. The g-values for the other seas were taken to be the same proportion of the respective integrals of their depth-related chlorophyll profiles. Retinal irradiance (Er(l)) was calculated with the following relationship: ErðlÞ ¼ EzðlÞ 1007 (5) 2.3.2. Visualisation of bioluminescence Sixty-eight normalised bioluminescent emission spectra were taken from the literature (see Table 2), which covered a range of species inhabiting the mesopelagic region. The emission maxima (lImax) for these spectra cover the range 425–542 nm, with all but eight spectra falling within the range 450–510 nm. The latter range was used in this study. These emission spectra were subjected to a lmax//l transform, and the average was best-fitted with a Gaussian function using CurveExpert v1.38 (Hyams, 2003), of the form: 2 y ¼ eðð1xÞ =0:008969331Þ (7) where x is the lmax//l value for the bioluminescent emission maxima of interest. This model was used to create a normalised bioluminescence spectrum template with three important features typical of mesopelagic bioluminescence: a bell-shaped spectrum that has a ARTICLE IN PRESS 1008 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 longwave skew and an increased half-bandwidth with increasing lmax. See Fig. 4 for examples of bioluminescence spectra and the fitted templates. Total radiant photon flux for a bioluminescent point source in the ocean can vary between 107 and 1013 photons s1 (Warrant and Locket, 2004). In this study Table 2 Data on species and lImax of the bioluminescent spectra taken from the literature to create the bioluminescent template for modelling purposes. Genus Species Emission maxima (nm) Reference Abralia Abralia Abraliopsis Abylopsis Argyropelecus Argyropelecus Bargmannia Bargmannia Bathocyroe Bathocyroe Borostomias Ceratias Chaenophyrne Clytia Cranchia Cranchia Ctenopteryx Diaphus Edriolychnus Euphausia Euphausia Euplokamis Galiteuthis Gaussia Gonichthys Gonostoma Grammatostomias Halitrephes Hippopodius Isistius Japatella Leachia Meningodora (Mertensiid) Metridia Myctophum Ocyropsis Oncaea Oncaea Onychoteuthis Opisthoproctus Oplophorus Parapronoe Periphylla Photobacterium Photostomias Pleuromamma Pleuromamma Porichthys Pyrocystis Pyrosoma Pyrosoma Pyroteuthis Rhynchohyalus Scina Scina Selenoteuthis veranyi veranyi falco tetragona affinis spp. elongata elongata fosteri fosteri elucens holboelli spp. hemisphaericum scabra scabra siculus rafinesquei schmidti superba tenera stationis glacialis princeps coccoi spp. circularis valdiviae hippopus brasiliensis diaphana spp. vesca spp. pacifica punctatum spp. conifera conifera banksi grimaldii spinosus crustulum periphylla phosphoreum guernei abdominalis xiphias notatus noctiluca atlanticum spinosum margaritifera natalensis marginata rattrayi scintillans 485 494 462 488 (2) 485 (3) 469 440 492 463 488 475 471 476 441 505 430 425 471 476 468 469 462 510 478 476 509 470 502 450 457 505 500 454 508 490 469 480 469 428 505 469 475 470 470 476 469 484 486 (2) 501 473 (2) 495 487 480 479 435 440 461 7 7 4 9 4 5 9 9 9 9 3 3 3 9 4 3 3 5 3 3 2 9 3 3 3 2 3 9 3 3 3 3 3 9 8 1 3 8 3 3 5 3 3 3 3 3 6 6 4 3, 4 2 2 3 3 3 4 3 Table 2 (continued ) Genus Species Emission maxima (nm) Reference Sergestes Sergestes Sternoptyx Thalassicolla Vampyroteuthis Vibrio similis spp. diaphana spp. infernalis fischeri 470 466 476 450 456 542 4 3 3 3 3 4 Numbers in brackets next to the lImax value indicate occurrences of more than one similar spectrum per species. In this case the lImax is the mean from these spectra. Absence of brackets indicates occurrences of one spectrum per species. Where species are tabulated more than once the respective spectra were measured from different photophores. For creating the template all spectra were used (68 in total). References: 1: Nicol (1960), 2: Swift et al. (1977), 3: Herring (1983), 4: Widder et al. (1983), 5: Denton et al. (1985), 6: Latz et al. (1987), 7: Herring et al. (1992), 8: Herring et al. (1993), 9: Haddock and Case (1999). Fig. 4. Comparison of three bioluminescent template outputs against three real bioluminescent spectra taken from the literature. Solid lines ¼ template, dotted lines ¼ real spectra. Species used: Hippopodius hippopus, Ocyropsis spp. and Galiteuthis glacialis (lImax: 450, 480 and 510 nm, respectively). a value of 7  1010 photons s1 was used, typical of myctophids (Mensinger and Case, 1990). The spectral radiant photon flux was calculated by multiplying the total photon flux by the normalised spectral values described above and dividing by the integral of the normalised spectral values. It was assumed that the bioluminescent point source was situated on the body of the emitting animal against an entirely black background and thus the photons radiated outwards from the photophore over a hemisphere. The spectral radiant intensity of light emitted per unit solid angle over this hemisphere (I(l), photons s1 nm1 sr1) was calculated by dividing the bioluminescent emission spectrum by 2p steradians. The spectral photon flux entering the eye of an observer (F(l), photons s1 nm1) was derived from the following relationship: F ðlÞ ¼ IðlÞ A D2 (8) where A is the area of the pupil, taken to be 10 mm2, and D (m) the distance between the source and the observer’s eye. To calculate the spectral photon flux arriving at the retina (Fr(l), photons s1 nm1), accounting for the spectral attenuation properties of ocean water and the lens, the ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 1009 following relationship was used: F rðlÞ ¼ F ðlÞ eðcðlÞ DÞ T ðlÞ (9) where c(l) is the spectral beam attenuation coefficient (m1) for ocean water, which is the sum of the spectral absorption (a(l)) and scattering (b(l)) coefficients calculated using the following relationships from Morel and Maritorena (2001; Eqs. (16), (17) and (18) in Appendix B): aðlÞ ¼ ½awðlÞ þ ð0:06acðlÞ C 0:65 (10a) Þ þ ayðlÞ  where ayðlÞ ¼ ayð440Þ eð0:014ðl440ÞÞ ayð440Þ ¼ 0:2½awð440Þ þ ð0:06acð440Þ C (10b) 0:65 Þ (10c) and from Morel (1991), Eq. (19) bðlÞ ¼ bwðlÞ þ 550 l þ 0:3C 0:62 (10d) where aw(l) and bw(l) are, respectively, the spectral absorption and scattering coefficients of pure seawater (m1) taken from Smith and Baker (1981), and ac(l), the spectral absorption coefficients (m1) of phytoplankton pigments, taken from Prieur and Sathyendranath (1981). All datasets were interpolated to 1 nm intervals. The term C here refers to the average chlorophyll concentration (mg m3) for each water type in the depth range 0–200 m, the values of which are shown in Table 1. Visual pigment absorptance spectra were calculated as described above and multiplied by the calculated values of Fr(l) and summed to compute the photon catch per second for a photoreceptor (the bioluminescent point source assumed to be focussed to a point on the retina). The above model was iterated over a range of visualisation distances (0–100 m in steps of 1 m) and a range of visual pigment lmax values (350–600 nm in steps of 1 nm) for a range of modelled bioluminescent emission spectra (emission maxima (lImax) 450–510 nm in steps of 5 nm) in each water type. At each visualisation distance the visual pigment with the highest summed photon catch rate was taken to be the most sensitive visual pigment to the bioluminescent point source. The threshold for vision was varied in three steps over three orders of magnitude, with photon catch rates of 1, 10 and 100 photons s1. Studies on human observers have shown that the threshold of visualising a point source of light falls within these orders of magnitude (Hecht et al., 1942). The model was also run with the individual bioluminescent spectra used to create the template to validate its use. For the visualisation of far-red bioluminescence, a normalised emission spectrum for M. niger bioluminescence was taken from Widder et al. (1984) and a radiant photon flux of 2.5  1010 photons s1 was used (Mensinger and Case, 1990, Table 1, value for sub-orbital photophore of M. niger). These data were used for spectral radiant intensity, I(l) Eq. (8), in place of the bioluminescence emission template. The model was rerun over a range of visualisation distances (0–10 m in steps of 0.1 m) and a range of visual pigment lmax values (460–710 nm in steps of 1 nm) for each water type. As before, the photon catch Fig. 5. Partial bleach of a Bolinichthys longipes extract (taken from Douglas et al., 2003). Scans showing the effect of a series of bleaches under various durations of monochromatic light. In order of decreasing absorbance at 500 nm the absorbance spectra are: (i) initial measurement of the unexposed extract 10 min after the addition of 30 ml 1 M hydroxylamine to 600 ml of extract; (ii) 709 nm 2 min; (iii)–(v) 634 nm 10 min; (vi)–(xii) 609 nm 5 min; (xiii)–(xiv) 576 nm 1.5 min, 5 min; (xv) white light 2 min. Apart from the visual pigment which is bleached by successive light exposures, the extract also contained a photostable pigment absorbing maximally around 670 nm. This is assumed to be a chlorophyll-derived pigment that may act as a photosensitiser (Douglas et al., 2002, 2003). Line (xv) (500–800 nm range) was used for the modelling of the visualisation of stomiid red bioluminescence using the chlorophyll derivative. rates for each visual pigment at each visualisation distance were calculated down to the thresholds of 1, 10 and 100 photons s1. To calculate the potential sensitivity to far-red bioluminescence that the chlorophyll derivative of B. longipes could confer, the following changes were made to the above model. The absorption spectrum (500–800 nm) of the chlorophyll derivative (Fig. 5, line xv) was used as a substitute for the visual pigment absorption data. Normalised absorbance data were multiplied by the specific absorbance of rhodopsin (0.013 mm1) and the pathlength (50 mm), and the resulting absorbance values were used in Eq. (6) to calculate an absorptance spectrum for the chlorophyll derivative. This was then used in Eq. (8) and (9) to calculate the photon catch rate of the chlorophyll derivative. It was assumed that the chlorophyll occurs in the same density as the visual pigment (Fig. 5), and that photon absorption leads directly to photoreceptor activation with 100% efficiency (thus producing the theoretical maximum visualisation distance). 3. Results Table 3 lists the lmax values for all species of myctophid whose visual pigments have been studied to date (reproduced in Fig. 6). These are most likely all rod visual pigments as the majority of myctophids have entirely rod retinas (Ali and Anctil, 1976). The larvae of Myctophum punctatum (not used in this study) are known to possess both rods and cones in their retinas (Bozzano et al., 2007), but cones are lost on development to adulthood (Sabatès et al., 2003). Even so, if any cone pigments were present in our extracts they would have been destroyed by the use of hydroxylamine (NH2OH; Holcman and Korenbrot, 2005). ARTICLE IN PRESS 1010 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 Table 3 Visual pigment lmax values of 58 species of myctophid measured in this and previous studies. Subfamily Genus Species Pigment lmax (nm) References Lampanyctinae Bolinichthys Bolinichthys Bolinichthys Ceratoscopelus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Diaphus Gymnoscopelus Gymnoscopelus Gymnoscopelus Gymnoscopelus Lampadena Lampanyctus Lampanyctus Lampanyctus Lampanyctus Lampanyctus Lampanyctus Lampanyctus Lampanyctus Lobianchia Lobianchia Nannobrachium Nannobrachium Nannobrachium Nannobrachium Nannobrachium Nannobrachium Notoscopelus Stenobrachius Taaningichthys Triphoturus indicus longipes photothorax warmingii brachycephalus coeruleus dumerilii holti jenseni leutkeni longleyi lucidus metopoclampus pacificus perspicillatus problematicus rafinesquei regani termophilus theta bolini braueri fraseri nicholsi speculigera alatus crocodilus festivus nobilis omostigma parvicauda simulator tenuiformis dolfleini gemellarii achirus atrum idostigma nigrum regale ritteri resplendens leucopsarus bathyphilus mexicanus 489 488 488 468, 488 483 486 485 489 486 487 490 490 487 488 484 488 490 488 487 490 487 485 488 489 488 485 487 489 484 486 485 487 485 486 487 484 485 486 485 489 485 486 492 487 490 14 15 3 1 (9, 14) 3 3 14 14 (9) 3 3 3 9 1 2 3 3 14 (5, 10) 3 3 3 4 4 4 4 14 14 14 13 1 2 2 1 1 3 1 (10, 14) 1 (4) 14 (5, 11) 2 3 9 3 (9) 14 8 (12) 1 (11, 14) 6, 7 Myctophinae Benthosema Benthosema Diogenichthys Electrona Electrona Electrona Hygophum Krefftichthys Myctophum Myctophum Protomyctophum Protomyctophum Protomyctophum panamense suborbitale laternatus antarctica carlsbergi rissoi proximum anderssoni aurolaternatum nitidulum bolini choriodon parallelum 490 487 488 488 485 482 475, 502 482 485, 495 468, 522* 488 483 480 2 14 2 4 4 13 1 4 2 16 4 4 4 Where available we have used extract data obtained during this study or data generated by us in other studies using identical protocols (Douglas and Partridge, 1997; Hasegawa et al., 2008). Where such data were unavailable we have used either extract or microspectrophotometric (MSP) data from other studies. Measurements of visual pigment lmax by extract and MSP are not significantly different (Douglas and Partridge, 1997). References: Brackets indicate previous measurements on species. Only data outside of brackets were used in this study. lmax value highlighted with * indicates porphyropsin. All others are rhodopsins. 1: Sonne 1999 (SO142), 2: Sonne 2003 (SO173), 3: Sonne 2007 (SO194), 4: James Clark Ross 2004 (JR100), 5: Denton and Warren (1957), 6: Munz (1957), 7: Munz (1958), 8: O’Day and Fernandez (1976), 9: Fernandez (1978), 10: Partridge et al. (1988), 11: Partridge et al. (1989), 12: Crescitelli (1991), 13: Partridge et al. (1992), 14: Douglas and Partridge (1997), 15: Douglas et al. (2002, 2003), 16: Hasegawa et al. (2008). ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 1011 Fig. 6. Frequency distribution of lmax values for all myctophid species studied (n ¼ 58). Single-pigment species (n ¼ 54) shown as black bars, double-pigment species (n ¼ 4) shown as white bars. Grey bars show other single rhodopsin visual pigments of previously measured deep-sea species for comparison, data taken from Douglas et al. (1998a). In the single-pigment species, the lmax values are relatively confined in their distribution, with all species falling within the range 480–492 nm. The prevalence of double-pigment species is relatively low with only four out of the 58 species having either a non-paired rhodopsin/porphyropsin system (A1/A2) or two rhodopsin (A1/A1) pigments in their photoreceptors. Three of these are in the subfamily Myctophinae, of which two are within the same genus (Myctophum). An analysis of the subfamily lmax means (calculated from the genera lmax means) showed that the Myctophinae lmax values are not significantly different from those of the Lampanyctinae (one-way ANOVA F1,14 ¼ 3.5, P ¼ 0.083). Linear regression of the lmax values of single-pigment species (n ¼ 54) against their mean maximum depths (Fig. 7a) showed that the slope of the regression line was significantly different from zero, indicating that the lmax decreased as the depth increased (t ¼ 0.29; P ¼ 0.026; y ¼ 0.00281x+488.55; r2(adj.) ¼ 0.074). However, when this analysis is repeated with the genera means (n ¼ 16) the significance of the trend is greatly reduced (t ¼ 1.48; P ¼ 0.16; y ¼ 0.0038x+489.7; r2(adj.) ¼ 0.073; Fig. 7b). Even so, when the species within each genus are analysed a similar, although non-significant, relationship is also found (data not shown). The results of the comparative analysis were subjected to a linear regression forced through the origin (Harvey and Pagel, 1991). As shown i n Fig. 7c, there was no significant relationship (t ¼ 1.3; P ¼ 0.217; y ¼ 0.0035x; r2(adj.) ¼ 0.038), indicating that the variation of lmax values between the myctophid species is a result of their shared phylogeny rather than an adaptive within-group variation. The results of the modelling for the visualisation of downwelling sunlight show that the predicted values of the most sensitive lmax fall rapidly with increasing depth, reaching minima that depends on water type (Fig. 8). In the clearest water type, bright downwelling moonlight (which is ca. 6 orders of magnitude less than sunlight in total irradiance; Denton, 1990) can only penetrate down to 160 m, and in other waters considerably less deep. The lmax values for the single-pigment myctophid species are more longwave shifted when compared to the predictions for the visualisation of sunlight in the clearer water types Fig. 7. Linear Regression of the lmax values of the single-pigment species as a function of depth. (a) Individual species data (n ¼ 54) y ¼ 0.00281x+488.55, P ¼ 0.026, t ¼ 0.29, r2(adj.) ¼ 0.074. (b) Genera means (n ¼ 16) y ¼ 0.0038x+489.7, P ¼ 0.16, t ¼ 1.48, r2(adj.) ¼ 0.073. (c) An independent comparisons analysis of the myctophid genera, after the phylogeny of Paxton et al. (1984), cited in Stiassny (1996). Linear regression forced through the origin, y ¼ 0.0035x, P ¼ 0.217, t ¼ 1.3, r2(adj.) ¼ 0.038. The non-significant result suggests that the variation in lmax values is a result of shared evolutionary history rather than a specific adaptation to the environment. (Sargasso and Mediterranean), but are closer over the mesopelagic zone (200–1000 m) to the predictions for visualising sunlight in the more attenuating water types (Antarctic and Celtic). For the visualisation of bioluminescence the predictions obtained from using the bioluminescence template are very close to those produced from the real bioluminescent emission spectra (Fig. 9), therefore for clarity any further bioluminescent results shown will be produced from the template. The model predictions for the theoretical maximum visualisation distance ARTICLE IN PRESS 1012 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 Fig. 8. Model output for the visualisation of downwelling sunlight. The lines represent the model predictions for the lmax of the visual pigments most sensitive to downwelling sunlight in the four water types. The endpoints of the lines represent the maximum depth that sunlight can penetrate and still be seen; black diamonds represent the same endpoint for bright moonlight. The white diamonds represent the lmax data for single-pigment myctophid species plotted against their mean maximum depth. Fig. 9. Comparison of six predictions from the bioluminescent visualisation model using the bioluminescent templates and real spectra from Fig. 4. Filled points ¼ template predictions, open points ¼ real spectra predictions, squares: lImax ¼ 450 nm, triangles: lImax ¼ 480 nm, diamonds: lImax ¼ 510 nm. of bioluminescence show that the most sensitive visual pigments vary greatly with water type, wavelength of bioluminescent emission maxima (lImax) and distance from the source, creating a range of most sensitive pigments for varying situations (Fig. 10). These ranges (solid bars) are shown against the lmax distribution of the single-pigment species, with equivalent ranges for downwelling sunlight (dotted bars) shown for comparison (Fig. 11). The downwelling sunlight visualisation ranges for the clearer water types (Sargasso and Mediterranean) are shortwave shifted when compared to the lmax values. Whereas the equivalent ranges for the more attenuating water types (Antarctic and Celtic) seem more congruent it must be noted that downwelling sunlight can not penetrate far into the water column in these water types (350 m). Consequently, sunlight in a large portion of the myctophid depth range is below the visual threshold. The ranges of visual pigment lmax calculated for the visualisation of bioluminescence are very wide; however, this is largely due to the fact that they cover the lImax range 450–510 nm. The lImax of myctophid bioluminescence is Fig. 10. Model output for the visualisation of bioluminescence. a, b and c show the predicted visualisation distance for a point source of bioluminescence with emission maxima (lImax) of 450, 480 and 510 nm, respectively. The different lines represent the predictions for the four water types. Black diamonds on each line represent the maximum visualisation distance of the bioluminescent source at different thresholds of vision (from left to right on each line: 100, 10, and 1 photons s1, respectively). Note change in ordinate between subplots. relatively understudied, but the small amount of data shows it covers the range 469–474 nm (Nicol, 1960; Herring, 1983; Widder et al., 1983; Denton et al., 1985). The diamonds in Fig. 11 show the range of most sensitive visual pigments for visualising myctophid bioluminescence across all water types. As shown, they are more congruent with the myctophid lmax values than the equivalent ranges for downwelling sunlight. The results of the modelling for the visualisation of M. niger far-red bioluminescence are shown in Fig. 12a. All myctophid visual pigments studied so far (bar one, lmax ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 1013 Fig. 11. Comparison of the model outputs for the visualisation of downwelling sunlight and bioluminescence. The histogram shows the lmax distribution of the single-pigment myctophid species. The solid horizontal bars represent the ranges of the most sensitive visual pigments to a bioluminescent source covering the lImax range 450–510 nm at a threshold of 10 photons s1. The different solid bars represent the range for the four water types, bottom to top: Sargasso, Mediterranean, Antarctic and Celtic Sea. The dotted horizontal bars represent the range of most sensitive visual pigments for the visualisation of downwelling sunlight over the depth range of 200–1000 m (approximate daytime depth). The different dotted bars represent the range for the four water types, left to right: Sargasso, Mediterranean, Antarctic and Celtic Sea. The black diamonds represent the range of most sensitive visual pigments for the visualisation of known myctophid bioluminescence. 522 nm in M. nitidulum) are not sensitive at all to this far-red light, therefore myctophids are ‘blind’ to this particular predator’s ‘private waveband’. Interestingly, however, M. nitidulum could potentially detect the red bioluminescence of M. niger up to ca. 1 m away (Fig. 12b). It was assumed that the chlorophyll derivative, found in B. longipes, occurs in the same density as the visual pigment, and that photon absorption leads directly to photoreceptor activation with 100% efficiency. Under these conditions it could theoretically detect the far-red bioluminescence up to a maximum distance of ca. 7 m (Fig. 12b). Both would give a significant advantage to their respective species in terms of avoiding red-light-emitting stomiid predators when compared to other myctophid species. For comparison, Fig. 12b also shows the photon catch rate of two other visual pigments, a 580 nm rhodopsin and 655 nm porphyropsin. The 580 nm pigment represents the theoretical most longwave-sensitive rhodopsin (Allison et al., 2004) and the 655 nm pigment the theoretical porphyropsin equivalent, according to the equation of Whitmore and Bowmaker (1989). In this model the theoretical 655 nm porphyropsin is only slightly more sensitive than the chlorophyll derivative, which demonstrates just how strong an effect a non-visual pigment photosensitiser could have on the visual sensitivities, and thus behaviour, of an animal. 4. Discussion The wavelengths of maximum absorption (lmax) of single visual pigments within the photoreceptors of myctophids fall in the range 480–492 nm, which, compared to other mesopelagic teleosts, seems relatively Fig. 12. Model output predicting visualisation ranges of Malacosteus farred bioluminescence in the Sargasso Sea by visual pigments differing in lmax. (a) Histogram (left y-axis) shows all known myctophid lmax data (both single- and double-pigment species, n ¼ 58). Lines (right y-axis) represent visualisation ranges for different visual pigments, assuming visual thresholds of 1, 10 and 100 photons s1. (b) Photon catch rate of three visual pigments and the photostable pigment (probably a chlorophyll derivative) from Bolinichthys longipes. Line (i) lmax ¼ 522 nm (Myctophum nitidulum); line (ii) lmax ¼ 580 nm rhodopsin; line (iii) chlorophyll derivative; line (iv) lmax ¼ 655 nm porphyropsin. It was assumed that the chlorophyll derivative occurs in the same density as the visual pigment (see Fig. 5), and that photon absorption leads directly to photoreceptor activation with 100% efficiency, thus estimating an upper limit to the theoretical maximum visualisation distance. confined (Fig. 6). These pigments may be either optimised for detecting downwelling sunlight or moonlight or alternatively to bioluminescent emissions of other deepsea animals, or might not be related directly to the photic environment at all but simply be the result of phylogenetic constraints. Optimisation here refers to determining the visual pigment that catches the greatest number of photons, the output of the model being the lmax of the most sensitive visual pigment. In this model, deviating from the predicted most sensitive pigment by 1 nm corresponds to a decrease in photon catch rate efficiency in the order of 0.01%. The effects of such deviations on visual processes will be unnoticeable, but such small increments of increasing efficiency are the very basis for gradual evolutionary change (Nilsson and Pelger, 1994). In the photon-limited environment of the mesopelagic sensitivity to the available light is at a premium and adaptations to maximise the efficiency of this trait are to be expected. The sensitivity hypothesis, proposed over 70 years ago, suggested that the visual pigments of mesopelagic fish are ARTICLE IN PRESS 1014 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 maximally sensitive to the available ambient downwelling sunlight. As the spectral quality of this sunlight changes constantly with depth, becoming increasingly restricted to the shortwave part of the spectrum, a corresponding hypsochromic shift in the lmax of the rod photoreceptors should be encountered in the relatively deeper-living species. As predicted by this hypothesis, our data do show a significant hypsochromic shift with increasing depth when all species are treated as independent data points (Fig. 7a). A more conservative analysis on the genera means (Fig. 7b), however, shows that although the relationship is maintained, it is no longer significant. These relationships are similar to those found in previous studies but it is enlightening that when the data are analysed comparatively (Fig. 7c), no significant relationship is found. This indicates that the distribution of lmax values in this study is more due to a shared evolutionary history than adaptive variation. Although these data seem to offer some support for the sensitivity hypothesis, they are far from conclusive. When the observed myctophid lmax values are plotted against the predicted most sensitive visual pigment according to the assumptions of the sensitivity hypothesis and the visualisation of downwelling sunlight (Fig. 8) they are longwave shifted compared to the predicted optimum in the clearer water types (Sargasso and Mediterranean). However, in the more attenuating water types (Antarctic and Celtic) they do seem more congruent, but it must also be accepted that the threshold depth for visualising sunlight in the latter is relatively shallow (350 m), even when we assume tropical sunlight conditions at the ocean surface. Thus, it seems unlikely that myctophid visual pigments are an adaptation for detecting dim downwelling sunlight or moonlight. When the visualisation of bioluminescence is modelled for a range of bioluminescent emission spectra (lImax ¼ 450–510 nm) the predicted most sensitive visual pigments vary greatly with water type, bioluminescent lImax and with distance from the source (Fig. 10). The observed myctophid lmax values fall well within the range predicted for the visualisation of bioluminescence (Fig. 11), however, this is largely due to the fact that the range is very wide when compared to that of downwelling sunlight. The large ranges for lmax values predicted to be most sensitive to bioluminescence are the result of modelling over the lImax range of 450–510 nm. Although the majority of bioluminescence seems to peak in this range, it is possible that myctophids view a much more restricted range in their individual ecology. Moreover, an even more limited number of bioluminescent sources may have salience to myctophids. It can be postulated from looking at the ranges for sunlight and bioluminescence in Fig. 11 that myctophids are indeed tuned to visualising bioluminescence. Firstly, as described above, they seem too longwave shifted to be maximally sensitive to sunlight in the clearer water types (Sargasso and Mediterranean) and, although more congruent with the ranges in the more attenuating water types (Antarctic and Celtic), it must also be remembered that downwelling sunlight can only penetrate to 350 m in these latter water types. Secondly, the ranges for bioluminescence are wide and so give the opportunity for myctophids to tune specifically to bioluminescence over a large portion of the spectrum. As noted above, myctophids seem restricted in their lmax range when compared to other deep-sea fish (Fig. 6), which suggests that they could be adapted for seeing bioluminescence in this limited range rather than sunlight. This makes intuitive sense: not only do the myctophids feed on bioluminescent prey (Pakhomov et al., 1996; Pusch et al., 2004), they are all (bar one species, Taaningichthys paurolynchus, Craddock and Hartel 2002, not used in this study) characterised by their distinctive bioluminescent organs. The suggestion that they are tuned to visualising bioluminescence is further strengthened when only the range of myctophid bioluminescence is considered. The lImax values of their bioluminescence seems to fall within the range 469–474 nm (Nicol, 1960; Herring, 1983; Widder et al., 1983; Denton et al., 1985). To view this range the most sensitive visual pigments predicted from the model would fall within the lmax range of 483–499 nm across all water types (Fig. 11). When considering that within their individual ecology myctophids need to detect their conspecifics, potential prey, and potential predators then the range in Fig. 11 (black diamonds) indicates that at least they are spectrally tuned to be sensitive to their own bioluminescence. With greater knowledge of myctophid natural history, ecology and ethology, it might also be the case that they are well tuned to the bioluminescence of predators and prey. The predicted maximum visualisation distances for varying point sources of bioluminescence with an emittance of 7  1010 photons s1 show (Fig. 10) that for the clearest water type (Sargasso Sea), at the most sensitive threshold for vision (1 ph s1), the furthest a myctophid could detect the bioluminescent spot is ca. 30 m. In a more attenuating water type (Celtic Sea) this is reduced to ca. 10 m. These values, calculated for bioluminescence seen against otherwise complete darkness, are in good agreement with previous predictions on visualisation ranges (Denton, 1990; Warrant, 2000). Being able to see bioluminescence from a distance of tens of metres would be very useful for behaviours such as feeding, predator avoidance, or the detection and attraction of conspecifics. These predictions are useful in other ecological models such as those that focus on feeding rate or the interaction distances between individuals within the ocean. Most bioluminescence produced by mesopelagic organisms is blue/green in colour and the previous analysis suggests that myctophids are adapted for the detection of such illumination. However, some deep-sea animals, such as stomiid dragonfish, also produce far-red illumination (Douglas et al., 1998a). While myctophids do not themselves produce longwave bioluminescence, the present data suggest that some may, nonetheless, be more sensitive to it than most other mesopelagic fishes. Since myctophids are heavily preyed upon by red light producing stomiids (Sutton and Hopkins, 1996), enhanced longwave sensitivity would give them a degree of protection from one of their major predators. For example, it has previously been suggested (Hasegawa et al., 2008) that the longwave sensitive pigment in the retina of ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 M. nitidulum (lmax 522 nm), which is the most longwave sensitive visual pigment described in a deep-sea fish that does not belong to the red light producing stomiids, may be an adaptation for detecting stomiid longwave bioluminescence. Theoretical comparison to conventional blue/ green bioluminescence (Fig. 11) suggests its lmax lies outside the expected range for being maximally sensitive to such light, even though it could still detect it. However, similar modelling focussing on the visualisation of longwave bioluminescence from red light producing stomiids (Fig. 12) shows that M. nitidulum could potentially detect such light at a distance of ca. 1 m. Modelling of the photon catch rate of two other visual pigments, the most longwave possible rhodopsin (lmax 580 nm) and the corresponding porphyropsin (lmax 655 nm) in Fig. 12b shows that these pigments could detect such light at much greater distances, ca. 5 and 8 m, respectively. However, the apparent advantage of the porphyropsin pigment does not take into account the fact that porphyropsins tend to be noisier than rhodopsins (Ala-Laurila et al., 2007) and the apparent advantage of a porphyropsin is almost certainly overestimated in the figure. Previous studies have found that the retina of another myctophid, B. longipes, contains a substance similar to the photosensitising chlorophyll described in M. niger (Douglas et al., 2002, 2003) and a photosensitiser may have advantages over a longwave visual pigment. Although we have no evidence that the pigment in B. longipes actually functions as a photosensitiser to enhance longwave sensitivity, the data are highly suggestive of it by reference to the function of similar pigments in M. niger (Douglas et al. 1998b, 1999). Theoretical modelling (Fig. 12b) suggests that, assuming similar densities and photoactivation properties to that of a visual pigment, the chlorophyll derivative could endow B. longipes with the visual capabilities to detect red light producing stomiid bioluminescence up to ca. 7 m away. This is assuming, however, that the photosensitiser both absorbs the photon and transfers the energy to the visual pigment with 100% efficiency. In reality this will not occur and the true efficiency is unknown. However, by assuming it is 100% efficient, an upper boundary to the predictions is given. If, given increased knowledge of the process, the efficiency was reduced by a certain proportion then the prediction given in Fig. 12b can be reduced by the same proportion to calculate the visualisation distance. Thus, in comparison to most other mesopelagic animals, some myctophids potentially have enhanced longwave sensitivity either through longwave shifted visual pigments, or the possession of a longwave absorbing photosensitiser, potentially protecting them from ‘covert’ red light mediated stomiid predation. It must be emphasised that the visualisation distances for the detection of far-red bioluminescence shown in Fig. 12 are dependent on other factors such as water type (assumed here to be clear Sargasso type) and changes in the brightness of the photophore, so in reality these values are almost certainly less. Even so, if either M. nitidulum or B. longipes could detect stomiid longwave bioluminescence up to ca. 1 m away, this is still a significantly large volume of space in which to react and escape, conferring an advantage in predator avoidance compared with other myctophid 1015 species, which will be essentially blind to stomiid farred bioluminescence. Certain caveats must, nevertheless, be accepted alongside these data. Firstly, the sensitivity hypothesis assumes that maximised sensitivity is conferred by the visual pigment that catches the greatest number of photons (maximising the signal). However, photoreceptors are inherently noisy (Barlow, 1956) and, in the photon-limited environment of the ocean, the greatest sensitivity to light will be the result of maximising the signal-to-noise ratio (S/N).Thus, as shortwave pigments tend to have less dark noise (Firsov and Govardovskii, 1990; Ala-Laurila et al. 2004a, 2004b), the hypsochromic shift in lmax seen here may be the result of visual pigments simply occupying a spectral location that maximises the S/N ratio. Secondly, although the tuning mechanism of visual pigments is reasonably well understood, the extent to which the lmax of a particular visual pigment can be tuned is unknown (Hunt et al., 2001). It is possible that the visual pigments in the photoreceptors of myctophids, which are based on the RH1 opsin (Hunt et al., 2001), are as shortwave shifted as they can possibly be. This argument, however, seems unlikely because, as shown in Fig. 6, the distribution of the visual pigments in myctophids is relatively confined and longwave-shifted when compared to other mesopelagic fish. This suggests that the myctophid visual pigments are tuned for the visualisation of bioluminescence, in all likelihood the bioluminescence of particularly salient signallers. However, the tuning of the bioluminescent emission spectra to exploit the spectral sensitivities of the intended receiver cannot be disentangled from the tuning of the visual pigment itself. Perhaps the most parsimonious explanation is to assume a coevolution of both traits to maximise the efficiency of communication within an environment whose spectral attenuation imposes the overriding constraints. In this respect myctophids are an ideal taxon for investigation, limited only by our knowledge of their natural history, ecology and ethology. Acknowledgements Much of this work was funded by grants from NERC and the Royal Society. We acknowledge BMBF for funding FS Sonne and the NERC Antarctic Funding Initiative for funding RRS JCR. We are also indebted to the following people for their scientific advice and practical support at sea: Steve Doolan and Lucy Peddar for help with visual pigments analysis, Innes Cuthill for help and advice on statistics, the masters and crews of the FS Sonne and RRS James Clark Ross and especially the principal scientists on the FS Sonne (Ernst Flüh, Willi Weinrebe and Jochen Wagner). References Ala-Laurila, P., Donner, K., Koskelainen, A., 2004a. Thermal activation and photoactivation of visual pigments. Biophysical Journal 86, 3653–3662. Ala-Laurila, P., Pahlberg, J., Koskelainen, A., Donner, K., 2004b. On the relation between the photoactivation energy and the absorbance spectrum of visual pigments. Vision Research 44, 2153–2158. ARTICLE IN PRESS 1016 J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 Ala-Laurila, P., Donner, K., Crouch, R., Cornwall, M., 2007. Chromophore switch from 11-cis-dehydroretinal (A2) to 11-cis-retinal (A1) decreases dark noise in salamander red rods. Journal of Physiology 585, 57–74. Ali, M.A., Anctil, M., 1976. Retinas of Fishes. An Atlas. Springer, Berlin, Heidelberg, New York. Allison, W.T., Haimberger, T.J., Hawryshyn, C.W., Temple, S.E., 2004. Visual pigment composition in zebrafish: evidence for a rhodopsin–porphyropsin interchange system. Visual Neuroscience 21, 945–952. Barlow, H., 1956. Retinal noise and absolute threshold. Journal of the Optical Society of America 46, 634–639. Bayliss, L.E., Lythgoe, R.J., Tansley, K., 1936. Some new forms of visual purple found in sea fishes, with a note on the visual cells of origin. Proceedings of the Royal Society of London B 816, 95–113. Beebe, W., 1935. Half Mile Down. John Lane, The Bodley Head, London. BBOP (Bermuda Bio-Optics Project): ftp://ftp.icess.ucsb.edu/pub/bbop/ BBOP_by_DATE/bbop2003. Bozzano, A., Pankhurst, P., Sabatès, A., 2007. Early development of eye and retina in lanternfish larvae. Visual Neuroscience 24, 423–436. Clarke, G.L., 1936. On the depth at which fish can see. Ecology 17, 452–456. Collins, M.A., Xavier, J.C., Johnston, N.M., North, A.W., Enderlein, P., Tarling, G.A., Waluda, C.M., Hawker, E.J., Cunningham, N.J., 2008. Patterns in the distribution of myctophid fish in the northern Scotia Sea ecosystem. Polar Biology 31, 837–851. Craddock, J., Hartel, K., 2002. Myctophidae. Lanternfishes. In: Carpenter, K. (Ed.), FAO species Identification Guide for Fishery Purposes. The Living Marine Resources of the Western Central Atlantic. Bony Fishes, Part 1 (Acipenseridae to Grammatidae), vol. 2, pp. 944–951. Crescitelli, F., 1991. Adaptations of visual pigments to the photic environment of the deep-sea. Journal of Experimental Zoology 5 (Suppl.), 66–75. Cummings, M., Partridge, J.C., 2001. Visual pigments and optical habitats of surfperch (Embiotocidae) in the California kelp forest. Journal of Comparative Physiology A 187, 875–889. De Grip, W., 1982. Thermal stability of rhodopsin and opsin in some novel detergents. Methods in Enzymology 81, 256–265. Denton, E.J., Warren, F.J., 1956. Visual pigments of deep sea fish. Nature 178, 1059. Denton, E.J., Warren, F.J., 1957. Photosensitive pigments in the retinae of deep sea fish. Journal of the Marine Biological Association of the UK 36, 651–662. Denton, E.J., Herring, P., Widder, E., Latz, M., Case, J., 1985. The roles of filters in the photophores of oceanic animals and their relation to vision in the oceanic environment. Proceedings of the Royal Society of London B 225, 63–97. Denton, E.J., 1990. Light and vision at depths greater than 200 metres. In: Herring, P., Campbell, A., Whitfield, M., Maddock, L. (Eds.), Light and Life in the Sea. Cambridge University Press, Cambridge, New York, pp. 127–148. Douglas, R.H., McGuigan, C.M., 1989. The spectral transmission of freshwater teleost ocular media—an interspecific comparison and a guide to potential ultraviolet sensitivity. Vision Research 29, 871–879. Douglas, R.H., Thorpe, A., 1992. Short-wave absorbing pigments in the ocular lenses of deep-sea teleosts. Journal of the Marine Biological Association of the UK 72, 93–112. Douglas, R.H., Partridge, J.C., Hope, A.J., 1995. Visual and lenticular pigments in the eyes of demersal deep sea fishes. Journal of Comparative Neurology 177, 111–122. Douglas, R.H., Partridge, J.C., 1997. On the visual pigments of deep-sea fish. Journal of Fish Biology 50, 68–75. Douglas, R.H., Partridge, J.C., Marshall, N.J., 1998a. The eyes of deep sea fish I: lens pigmentation, tapeta and visual pigments. Progress in Retinal and Eye Research 17, 597–636. Douglas, R.H., Partridge, J.C., Dulai, K.S., Hunt, D.M., Mullineaux, C.W., Tauber, A., Hynninen, P.H., 1998b. Dragon fish see using chlorophyll. Nature 393, 423–424. Douglas, R.H., Partridge, J.C., Dulai, K.S., Hunt, D.M., Mullineaux, C.W., Hynninen, P.H., 1999. Enhanced retinal longwave sensitivity using a chlorophyll-derived photosensitiser in Malacosteus niger, a deep-sea dragon fish with far red bioluminescence. Vision Research 39, 2817–2832. Douglas, R.H., Bowmaker, J.K., Mullineaux, C.W., 2002. A possible retinal longwave detecting system in a myctophid fish without far-red bioluminescence; evidence for a sensory arms race in the deep-sea. In: Stanley, P.E., Kricka, L.J. (Eds.), Bioluminescence and Chemilumi- nescence; Progress and Current Applications. World Scientific, New Jersey, pp. 391–394. Douglas, R.H., Hunt, D.M., Bowmaker, J.K., 2003. Spectral sensitivity tuning in the deep-sea. In: Collin, S.P., Marshall, N.J. (Eds.), Sensory Processing in Aquatic Environments. Springer, New York, pp. 323–342. Edwards, A., Herring, P., 1977. Observations on the comparative morphology and operation of the photogenic tissues of myctophid fishes. Marine Biology 41, 59–70. Estrada, M., Marrase, C., Latasa, M., Berdalet, E., Delgado, M., Riera, T., 1993. Variability of deep chlorophyll maximum characteristics in the Northwestern Mediterranean. Marine Ecology Progress Series 92, 289–300. Fernandez, H., 1978. Visual pigments of bioluminescent and nonbioluminescent deep-sea fishes. Vision Research 19, 589–592. Firsov, M.L., Govardovskii, V.I., 1990. Dark noise of visual pigments with different absorption maxima. Sensory Systems 4, 25–34 (in Russian). Froese, R., Pauly, D., 2008. FishBase. World Wide Web electronic publication. /www.fishbase.orgS, version (01/2008). GjØsaeter, J., Kawaguchi, K., 1980. A Review of the World’s Resources of Mesopelagic Fish. FAO Fisheries Technical Paper, vol. 193. FAO, Rome, pp. 1–151. Govardovskii, V.I., Fyhrquist, N., Reuter, T., Kuzmin, D.G., Donner, K., 2000. In search of the visual pigment template. Visual Neuroscience 17, 509–528. Haddock, S.H.D., Case, J., 1999. Bioluminescence spectra of shallow and deep-sea gelatinous zooplankton: ctenophores, medusae, and siphonophores. Marine Biology 133, 571–582. Hart, N., Partridge, J.C., Bennett, A.T.D., Cuthill, I.C., 2000. Visual pigments, cone oil droplets and ocular media in four species of estridid finch. Journal of Comparative Physiology A 186, 681–694. Harvey, P., Pagel, M., 1991. The Comparative Method in Evolutionary Biology. Oxford University Press, Oxford. Hasegawa, E., Sawada, K., Abe, K., Watanabe, K., Uchikawa, K., Okazaki, Y., Toyama, M., Douglas, R.H., 2008. The visual pigments of a deep-sea myctophid fish Myctophum nitidulum (Garman); an HPLC and spectroscopic description of a non-paired rhodopsin–porphyropsin system. Journal of Fish Biology 72, 937–945. Hecht, S., Shlaer, S., Pirenne, M., 1942. Energy, quanta and vision. Journal of General Physiology 25, 819–840. Herring, P., 1983. The spectral characteristics of luminous marine organisms. Proceedings of the Royal Society of London B 220, 183–217. Herring, P., Widder, E., Haddock, S.H.D., 1992. Correlation of bioluminescence emissions with ventral photophores in the mesopelagic squid Abralia veranyi (Cephalopoda: Enoploteuthidae). Marine Biology 112, 293–298. Herring, P., Latz, M., Bannister, N., Widder, E., 1993. Bioluminescence of the Poecilostomatoid Copepod Oncaea conifera. Marine Ecology Progress Series 94, 297–309. Holcman, D., Korenbrot, J., 2005. The limit of photoreceptor sensitivity: molecular mechanisms of dark noise in retinal cones. Journal of General Physiology 125 (6), 641–660. Hunt, D.M., Dulai, K.S., Partridge, J.C., Cottrill, P., Bowmaker, J.K., 2001. The molecular basis of spectral tuning of rod visual pigments in deep sea fish. Journal of Experimental Biology 204, 3333–3344. Hyams, D., 2003. CurveExpert v1.38: /http://home.comcast.net/ curveexpert/S. Jerlov, N.G., 1976. Marine Optics. Elsevier, Amsterdam. Johnsen, S., Widder, E., Mobley, C., 2004. Propagation and perception of bioluminescence: factors affecting counterillumination as a cryptic strategy. Biological Bulletin 207, 1–16. Kimura, N., Okada, Y., 1997. Estimation of vertical profile of chlorophyll concentration around the Antarctic peninsula derived from CZCS images by the statistical method. Polar Biology 10, 66–76. Knowles, A., Dartnall, H.J.A., 1977. The characterisation of visual pigments by absorption spectroscopy. In: Davson, H. (Ed.), The Eye. Academic Press, London, pp. 53–100. Land, M.F., Nilsson, D.-E., 2002. Animal Eyes. Oxford University Press, Oxford. Latz, M., Frank, T., Bowlby, M., Widder, E., Case, J., 1987. Variability in flash characteristics of a bioluminescent copepod. Biological Bulletin 173, 489–503. Lythgoe, J., 1979. The Ecology of Vision. Clarendon Press, Oxford. Lythgoe, J., Muntz, W.R.A., Partridge, J.C., Shand, J., Williams, D., 1994. The ecology of the visual pigments of snappers (Lutjanidae) on the Great Barrier Reef. Journal of Comparative Physiology A 174, 461–467. Mensinger, A., Case, J., 1990. Luminescent properties of deep sea fish. Journal of Experimental Marine Biology and Ecology 144, 1–15. ARTICLE IN PRESS J.R. Turner et al. / Deep-Sea Research I 56 (2009) 1003–1017 Morel, A., 1991. Light and marine photosynthesis: a spectral model with geochemical and climatological implications. Progress in Oceanography 26, 263–306. Morel, A., Maritorena, S., 2001. Bio-optical properties of oceanic waters: a reappraisal. Journal of Geophysical Research 106, 7163–7180. Munz, F.W., 1957. Photosensitive pigments from the retinas of deep-sea fishes. Science 125, 1142–1143. Munz, F.W., 1958. Photosensitive pigments from the retinae of certain deep sea fishes. Journal of Physiology 140, 220–225. Nicol, J.A.C., 1960. Spectral composition of the light of lantern-fish, Myctophum punctatum. Journal of the Marine Biological Association of the UK 39, 27–32. Nilsson, D.-E., Pelger, S., 1994. A pessimistic Estimate of the time required for an eye to evolve. Proceedings of the Royal Society of London B 256, 53–58. O’Day, W., Fernandez, H., 1974. Aristostomias scintillans (Malacosteidae): a deep-sea fish with visual pigments apparently adapted to its own bioluminescence. Vision Research 14, 545–550. O’Day, W., Fernandez, H., 1976. Vision in the lantern fish, Stenobrachius leucopsarus (Myctophidae). Marine Biology 37, 187–195. Pakhomov, E., Perissinotto, R., McQuaid, C., 1996. Prey composition and daily rations of myctophid fishes in the Southern Ocean. Marine Ecology Progress Series 134, 1–14. Partridge, J.C., Archer, S., Lythgoe, J., 1988. Visual pigments in the individual rods of deep-sea fishes. Journal of Comparative Physiology A 162, 543–550. Partridge, J.C., Shand, J., Archer, S., Lythgoe, J., van Groningen-Luyben, W., 1989. Interspecific variation in the visual pigments of deep-sea fishes. Journal of Comparative Physiology A 164, 513–529. Partridge, J.C., Archer, S., van Oostrum, J., 1992. Single and multiple visual pigments in deep-sea fishes. Journal of the Marine Biological Association of the UK 72, 113–130. Partridge, J.C., Douglas, R.H., 1995. Far-red sensitivity in the deep-sea dragon fish Aristostomias tittmanni. Nature 375, 21–22. Partridge, J.C., Cummings, M., 1999. Adaptations of visual pigments to the aquatic environment. In: Archer, S., Djamgoz, M., Loew, E., Partridge, J.C., Valerga, S. (Eds.), Adaptive Mechanisms in the Ecology of Vision. Kluwer Academic Publishers. Platt, T., Sathyendranath, S., 1988. Oceanic primary production: Estimation by remote sensing at local and regional scales. Science 241 (4873), 1613–1620. Prieur, L., Sathyendranath, S., 1981. An optical classification of coastal and oceanic waters based on the specific absorption curves of phytoplankton pigments, dissolved organic matter, and other particulate materials. Limnology and Oceanography 26, 671–689. Pusch, C., Schnack-Schiel, S., Mizdalski, E., von Westernhagen, H., 2004. Feeding ecology of three myctophid species at the Great Meteor Seamount (North-east Atlantic). Archive of Fishery and Marine Research 51, 251–271. Riccobene, G., Capone, A., Aiello, S., Ambriola, M., Ameli, F., Amore, I., Anghinolfi, M., Anzalone, A., Avanzini, C., Barbarino, G., Barbarito, E., Battaglieri, M., Bellotti, R., Beverini, N., Bonori, M., Bouhadef, B., 1017 Brescia, M., Cacopardo, G., Cafagna, F., Caponetto, L., Castorina, E., Ceres, A., Chiarusi, T., Circella, M., Cocimano, R., Coniglione, R., Cordelli, M., Costa, M., Cuneo, S., D’Amico, A., De Bonis, G., De Marzo, C., De Rosa, G., De Vita, R., Distefano, C., Falchini, E., Fiorello, C., Flaminio, V., Fratini, K., Gabrielli, A., Galeotti, S., Gandolfi, E., Grimaldi, A., Habel, R., Leonora, E., Lonardo, A., Longo, G., Lo Presti, D., Lucarelli, F., Maccioni, E., Margiotta, A., Martini, A., Masullo, R., Megna, R., Migneco, E., Mongelli, M., Montaruli, T., Morganti, M., Musumeci, M., Nicolau, C.A., Orlando, A., Osipenko, M., Osteria, G., Papaleo, R., Pappalardo, V., Petta, C., Piattelli, P., Raffaelli, F., Raia, G., Randazzo, N., Reito, S., Ricco, G., Ripani, M., Rovelli, A., Ruppi, M., Russo, G.V., Russo, S., Russo, S., Sapienza, P., Sedita, M., Schuller, J.-P., Shirokov, E., Simeone, F., Sipala, V., Spurio, M., Taiuti, M., Terreni, G., Trasatti, L., Urso, S., Valente, V., Vicini, P., 2007. Deep seawater inherent optical properties in the Southern Ionian Sea. Astroparticle Physics 27, 1–9. Sabatès, A., Bozzano, A., Vallvey, I., 2003. Feeding pattern and the visual light environment in myctophid larvae. Journal of Fish Biology 63, 1476–1490. Smith, R.C., Baker, K.S., 1981. Optical properties of the clearest natural waters (200–800 nm). Applied Optics 20, 177–184. Stiassny, M., 1996. Basal ctenosquamate relationships and the interrelationships of the myctophiform (scopelomorph) fishes. In: Stiassny, M., Parenti, L., Johnson, G. (Eds.), Interrelationships of Fishes. Academic Press, San Diego, CA. Sutton, T., Hopkins, T., 1996. Trophic ecology of the stomiid (Pisces: Stomiidae) fish assemblage of the eastern Gulf of Mexico: strategies, selectivity and impact of a top mesopelagic predator group. Marine Biology 127, 179–192. Swift, E., Biggley, W., Seliger, H., 1977. The bioluminescence emission spectra of Pyrosoma atlanticum, P. spinosum (Tunicata), Euphasia tenera (Crustacea) and Gonostoma sp. (Pisces). Journal of the Marine Biological Association of the UK 57, 817–823. Warrant, E., 2000. The eyes of deep-sea fishes and the changing nature of visual scenes with depth. Philosophical Transactions of the Royal Society of London B 355, 1155–1159. Warrant, E., Locket, N.A., 2004. Vision in the deep sea. Biological Review 79, 671–712. Whitmore, A., Bowmaker, J.K., 1989. Seasonal variation in cone sensitivity and short-wave absorbing visual pigments in the rudd, Scardinius erythrophthalmus. Journal of Comparative Physiology A 166, 103–115. Widder, E., Latz, M., Case, J., 1983. Marine bioluminescence spectra measured with an optical multichannel detection system. Biological Bulletin 165, 791–810. Widder, E., Latz, M., Herring, P., Case, J., 1984. Far red bioluminescence from two deep-sea fishes. Science 225, 512–514. Widder, E., 1999. Bioluminescence. In: Archer, S., Djamgoz, M., Loew, E., Partridge, J.C., Valerga, S. (Eds.), Adaptive Mechanisms in the Ecology of Vision. Kluwer Academic Publishers. Widder, E., 2002. Bioluminescence and the pelagic visual environment. Marine and Freshwater Behaviour and Physiology 35, 1–26.