SEARCH

SEARCH BY CITATION

Keywords:

  • adaptation;
  • craniates;
  • ecology;
  • evolution;
  • gene;
  • opsin

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

In craniates, opsin-based photopigments expressed in the eye encode molecular ‘light sensors’ that constitute the initial protein in photoreception and the activation of the phototransduction cascade. Since the cloning and sequencing of the first vertebrate opsin gene (bovine rod opsin) nearly 30 years ago (Ovchinnikov Yu 1982, FEBS Letters, 148, 179–191; Hargrave et al. 1983, Biophysics of Structure & Mechanism, 9, 235–244; Nathans & Hogness 1983, Cell, 34, 807–814), it is now well established that variation in the subtypes and spectral properties of the visual pigments that mediate colour and dim-light vision is a prevalent mechanism for the molecular adaptation to diverse light environments. In this review, we discuss the origins and spectral tuning of photopigments that first arose in the agnathans to sample light within the ancient aquatic landscape of the Early Cambrian, detailing the molecular changes that subsequently occurred in each of the opsin classes independently within the main branches of extant jawed gnathostomes. Specifically, we discuss the adaptive changes that have occurred in the photoreceptors of craniates as they met the ecological challenges to survive in quite differing photic niches, including brightly lit aquatic surroundings; the deep sea; the transition to and from land; diurnal, crepuscular and nocturnal environments; and light-restricted fossorial settings. The review ends with a discussion of the limitations inherent to the ‘nocturnal-bottleneck’ hypothesis relevant to the evolution of the mammalian visual system and a proposition that transition through a ‘mesopic-bottleneck’ may be a more appropriate model.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

All organisms must interact and respond to changes in their physical environment, such as the circadian rhythms in light, temperature and humidity. Thus, it would follow that genes encoding proteins that mediate the detection of these stimuli (e.g. intensity and spectral range of light) are under a high degree of selection, resulting in the maintenance of genomic variation (molecular adaptation), when the molecular change is ultimately advantageous to the survival of the organism within a particular ecological niche (Davies 2011).

For most craniates, the detection of light (photosensitivity) is of paramount importance in controlling behavioural responses, and a multitude of photosensitive tissues (e.g. eye, brain, pineal gland and dermal melanocytes) have evolved to mediate irradiance detection, although many of these systems have been lost in the mammalian lineage. The major exception is the eye (Peirson et al. 2009; Davies et al. 2010), which is broadly retained in all major vertebrate classes, except the nonvertebrate craniate hagfishes. Thus, despite the wide variation in both visual and nonvisual photosensitive mechanisms present in many vertebrates, the eye remains one of the most highly conserved and important sensory organs across all vertebrate classes.

It is well established that the basic design of the camera-like eye evolved prior to the divergence of the jawless (agnathan) and jawed (gnathostome) craniate lineages, over 540 Ma) (Lamb et al. 2007; Lamb 2009, 2011). For many decades, the retinae of most jawed vertebrates were described as duplex with the presence of two main types of photoreceptors (i.e. cones and rods). However, the discovery of a third novel photoreceptor system in the eye by Foster and colleagues in the 1990s demonstrated that ocular light detection was far more complex (Freedman et al. 1999; Lucas et al. 1999; Foster 2008; Provencio 2011). In addition to these photosensitive retinal ganglion cells (RGCs), it has now been shown that a multitude of other ocular photoreceptors expressing diverse photosensitive molecules exist (Jenkins et al. 2003; Foster & Bellingham 2004; Tomonari et al. 2005, 2007, 2008; Foster et al. 2007; Cheng et al. 2009; Lamb 2009; Peirson et al. 2009; Davies et al. 2010, 2011). Indeed, an in-depth study by Davies et al. (2011) showed that the melanopsin photopigment is expressed in all the major retinal photoreceptor types in the zebrafish, Danio rerio, with the conclusion that the eye of this model teleost may be broadly photosensitive (Davies et al. 2011). Thus, the so-called duplex retina of vertebrates may, in fact, be a multiplex photosensory organ. Based on historical data, it would appear that the duplicity designation may be restricted to the visual (image-forming) system alone, although evidence is emerging that suggests nonvisual photopigments (e.g. melanopsin) may play a role in visual tasks under bright-light conditions (Brown et al. 2010), and conversely, visual photoreceptors (e.g. rods and S cones) may be involved in nonimage-forming responses under dim-light conditions (Lall et al. 2010; Allen et al. 2011).

In vertebrates, vision is mediated via cones that are functional under bright-light conditions (photopic vision) and rods that detect low light intensities (dim-light or scotopic vision), although cones and rods are generally classed by a combination of features, such as morphology, cellular expression and electrophysiological photokinetics. Together, this duplex visual system processes visual information, which is ultimately sent to the brain via the optic nerve, where a coloured three-dimensional representation of the immediate environment is assembled (Fain et al. 2010). The key detectors of light are the visual photopigments that reside within retinal photoreceptors. These pigments are members of the G protein–coupled receptor (GPCR) superfamily and consist of an opsin protein moiety linked via a Schiff base to a retinal chromophore (Fig. 1). Within the literature, the term ‘rhodopsin’ is used to mean either any retinal-based photopigment, including the prokaryotic proton pumps (e.g. bacteriorhodopsin, proteorhodopsin and xanthorhodopsin), and the visual and nonvisual opsins of both invertebrates and vertebrates; or the pigment expressed in the rods of gnathostomes. Throughout this review, ‘rhodopsin’ will be used to describe photopigments that utilize a vitamin A1-derived chromophore, retinal, whereas those that use a vitamin A2-derived chromophore, 3,4-didehydroretinal, are referred to as ‘porphyropsins’ (Yokoyama 2000; Bowmaker 2008) (Fig. 1a). Whereas 11-cis retinal (mostly vitamin A1-based) and all-trans retinal are the predominant chromophores used in craniates, other retinoids (e.g. 3-hydroxyretinal and 4-hydroxyretinal) are commonly utilized in invertebrate rhabdomeres (Shichida & Matsuyama 2009). Visual opsins are seven transmembrane proteins that are trafficked to the disc membrane of a photoreceptor outer segment. The tertiary structure of an opsin is such that each transmembrane generates an internal chromophore-binding pocket, and it is the interaction between the chromophore and particular residues lining this pocket that determines the λmax of a pigment (Fig. 1b, c) (Yokoyama 2000; Davies 2011). Porphyropsin pigments are long-wavelength-shifted compared to rhodopsins, with the effect being more pronounced the longer the wavelength (Whitmore & Bowmaker 1989). Thus, the type of chromophore used may also affect the spectral sensitivity of photopigments.

image

Figure 1.  A diagram of the structure and function of craniate photopigments. (a) The initial step in phototransduction consists of photon (hv) absorption by 11-cis retinal, which photoconverts to all-trans retinal. Vertebrate photopigments are broadly divided into rhodopsins that utilize a vitamin A1-derived chromophore (black line) or porphyropsins that contain a vitamin A2-derived chromophore (3,4-didehydroretinal). In the latter case, the presence of a double (C=C) bond between C3 and C4 is shown as a dotted red line. (b) Side view and (c) mid-membrane section of a typical (opsin) photopigment, showing the presence of seven transmembrane domains (yellow), archetypal of the GPCR superfamily and their arrangement around the retinal chromophore (orange) (modified from Davies (2011)). The retinal attachment site (Lys296, black) and counterion (Glu113, pink) to the Schiff base are shown. Opsin residues that cluster around either the Schiff base or ionone ring of the retinal chromophore are coloured to highlight the amino acids involved in the spectral tuning of LWS (red), SWS1 (violet), SWS2 (blue) and RH2/RH1 (green) photopigments. Residues important for stabilizing the tertiary structure (e.g. disulphide bridge, amino-terminal (N) glycosylation sites) and the activation/deactivation of photopigments (e.g. carboxyl-terminal (C) phosphorylation sites), as well as membrane anchorage (e.g. palmitoylation sites), are also shown. TM, transmembrane; CL, cytoplasmic loop; EC, extracellular loop. The numbering is based on the bovine rod opsin (RH1) sequence.

Download figure to PowerPoint

Before further discussion, it should be noted that the number and type of cone opsin gene or pigment identified in any species is only indicative of the colour visual system that may be in use at a particular step of development or the life cycle. Thus, unless stated otherwise, the chromatic status of an organism refers to the potential for colour vision and not the actual case as determined by psychophysical or behavioural testing. Due to photoreceptor univariance, at least two cone photoreceptors are required for colour vision (colour opponency); thus, species in possession of a colour visual system must at least be dichromatic.

Five classes of visual pigments have been identified in craniates and are defined on the basis of their amino acid sequence, molecular phylogeny and spectral sensitivity. In addition to a single rod or RH1 class of opsin (with a λmax ∼ 500 nm), there are four cone opsin classes, comprising two short-wavelength-sensitive (SWS1 and SWS2) classes, a middle-wavelength-sensitive (RH2) class and a long-wavelength-sensitive (LWS) class. For rhodopsin pigments, the range of the respective λmax values is 500–570 nm for LWS, 480–530 nm for RH2, 400–470 nm for SWS2 and 355–445 nm for SWS1 (Yokoyama 2000; Bowmaker 2008; Davies 2011) (Fig. 2). In craniate visual pigments with λmax values >385 nm, the Schiff base formed between Lys296 of the opsin and the retinal chromophore is protonated, with a negatively charged residue at site 113 (usually Glu) acting as a counterion to stabilize the positive charge of the proton. Subsequent to photon capture, the chromophore of visual pigments photoisomerizes from 11-cis to all-trans, thus inducing a conformational change in the opsin protein that initiates the phototransduction cascade via a G-protein signalling mechanism.

image

Figure 2.  (a) A basic cladogram showing the relationship between each opsin class and its general spectral sensitivity. (b) Phylogeny of vertebrate visual pigment (opsin) gene families (LWS, SWS1, SWS2, RHB/RH2 and RHA/RH1). The tree was produced by applying Bayesian probability methodology with a Metropolis Markov chain Monte Carlo algorithm. A general time-reversal model was used with the percentage at the base of each node representing the posterior probability values. The scale bar indicates the number of nucleotide substitutions per site. The fruit fly Rh4 opsin gene sequence was used as an outgroup (modified from Collin et al. (2009) and Davies (2011)). Branch lengths do reflect evolutionary distances.

Download figure to PowerPoint

Phylogenetic analyses of the genetic sequences of the pigment molecules demonstrate that the rod (RH1) pigment evolved from a RH2 cone pigment (Okano et al. 1992; Yokoyama 2000; Collin et al. 2003b; Collin & Trezise 2004) (Fig. 2). The evolutionary position of the RH1 class therefore indicates that rod-based scotopic vision is a more recent evolutionary development (Okano et al. 1992; Davies et al. 2007b), with specific residues within the opsin protein being important to drive rod-like biochemical properties (Kuwayama et al. 2002; Imai et al. 2005). The primary adaptation of the scotopic visual system is the increased sensitivity of rods compared to cones, which is achieved through functional differences in the visual pigment and associated components of the phototransduction cascade (Ebrey & Koutalos 2001; Hisatomi & Tokunaga 2002).

Brightly lit aquatic environments and the origin of the craniate visual system

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Hagfishes and lampreys reside at the base of Craniata and are historically placed together as the sole survivors of the paraphyletic agnathan (jawless) lineage (Hardisty 1982). However, their phylogenetic relationships remain unresolved: some molecular studies separate them into two monophyletic groups that evolved sequentially from the main craniate branch, placing the lampreys phylogenetically closer to the gnathostomes (the so-called ‘vertebrate hypothesis’) (Rasmussen et al. 1998; Gursoy et al. 2000), while others reunite them into the Cyclostomata (known as the ‘cyclostome hypothesis’) (Stock & Whitt 1992; Mallatt & Sullivan 1998; Delarbre et al. 2002). Whichever case is correct, it will have important consequences for determining the origin of visual sensory systems in the craniates. Nonetheless, the basal phylogenetic location of either group is yielding important insights into the evolution of vertebrate photopigments and the selective pressures that have shaped them. Lampreys, or their very close relatives, were already present in the Lower Cambrian period around 540 Ma (Shu et al. 1999, 2003; Xian-guang et al. 2002). Northern hemisphere or holarctic lampreys form a single family, the Petromyzontidae (Renaud 1997), whereas the southern hemisphere lampreys are placed into either the Geotriidae or Mordaciidae (Potter 1980; Gill et al. 2003). The eyes of larval lampreys (known as ammocoetes) are similar to those of adult hagfishes and lie under the skin; in both cases, the retina is relatively undifferentiated (Dickson & Collard 1979; Rubinson & Cain 1989; Rubinson 1990). By contrast, the eyes of completely metamorphosed lampreys are well differentiated and follow closely the structure of ocular tissues found in gnathostomatous (jawed) fishes, with the presence of intra- and extra-ocular eye muscles, a lens that is multifocal and a complex multilayered retina (Collin et al. 1999; Gustafsson et al. 2008). Significantly, the retina of the southern hemisphere pouched lamprey, Geotria australis, an anadromous species, contains five morphologically distinct photoreceptor types that are all cone-like (Govardovskii & Lychakov 1984; Collin et al. 2003a; Collin & Trezise 2004), a complement of five different visual pigment genes that correspond to the SWS1, SWS2 and LWS cone classes found in the jawed gnathostomes, plus two copies of an RH-like opsin gene (named RHA and RHB) (Collin et al. 2003b; Davies et al. 2007b). Expression of these genes shows ontogenetic shifts in downstream and upstream adult migrants, from high expression of LWS in downstream migrants to high expression of RHA and RHB in upstream migrants; this would imply an enhanced sensitivity to longer wavelengths in the downstream phase (Davies et al. 2007b). The RHA and RHB genes encode photopigments that are spectrally similar with peak absorbances at 492 nm and 497 nm, respectively, and show a similar pattern of expression in downstream and upstream migrants (Davies et al. 2007b). Initial phylogenetic analysis of these two pigment genes identified both as cone-like (Collin et al. 2003b; Collin & Trezise 2004), with the duplication of the lamprey RH-like gene that resulted in RHA and RHB paralogues being an independent evolutionary event to that in the gnathostomes that led to the formation of both the cone RH2 and the ‘true’ rod RH1 gene lineages (Collin et al. 2003b). Thus, it would appear that true rod-based scotopic vision is not present in the Agnatha and evolved sometime later in the jawed vertebrates (Collin et al. 2003b). Subsequent analyses using a larger cohort of opsin sequences and different phylogenetic algorithms suggested that RHA and RHB genes cluster with the RH1 and RH2 gene classes of the jawed vertebrates, respectively (Pisani et al. 2006; Davies et al. 2009a), with the corollary that dim-light vision may, in fact, have arisen in lampreys (Pisani et al. 2006). Therefore, the question of when the molecules that mediate dim-light vision evolved remains controversial. Both studies offer opposing conclusions based on the phylogenetic position of the RHA/RH1 orthologue. The presence of an RH1 orthologue is, however, not in itself sufficient for true dim-light photoreception, and work is currently underway not only to identify the complement of the cone/rod phototransduction cascade genes present in the jawless lampreys, but also to determine the photokinetics of all photoreceptor types and how these compare to the photoreceptors of jawed vertebrates.

The presence of four cone pigment classes in G. australis, which are similar to those found in the gnathostomes, implies that the blueprint for craniate colour vision was first laid down before the agnathan/gnathostome split in the craniate lineage, more than 540 Ma (Fig. 2). Therefore, based on the classification of the RHA photoreceptor in G. australis as being morphologically and functionally more cone-like (Collin & Trezise 2004), this species (and probably the lamprey ancestor) possesses the potential for tetrachromatic or even pentachromatic vision (Collin et al. 2003b), the latter of which is unique amongst the founder members that reside at the base of all other vertebrate classes (e.g. ancestral teleosts would only have possessed four cone subtypes prior to the many duplications discussed below). Based on the finding of fossilized green algal deposits (Walker & Laporte 1970) alongside the fossilized remains of Cambrian lampreys, it would appear that these species evolved in a shallow, well-lit aquatic environment, where a broad-light spectrum may have been the driver for the evolution of multiple visual pigments.

Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

The four basal photopigment genes (LWS, SWS1, SWS2 and RHB/RH2) that encode the archetypal cone classes of opsin protein that first appeared in the agnathans passed without deletion or significant modification into the Osteichthyes, comprising both Actinopterygii and Sarcopterygii lineages, which first appeared in the Cretaceous period around 150 Ma. A duplication of the entire genome is thought to have occurred early in the evolution of the ray-finned fishes (Amores et al. 1998; Taylor et al. 2003; Jaillon et al. 2004; Meyer & Van de Peer 2005), but most likely subsequent to the divergence of the actinopterygian and the sarcopterygian lineages about 417 Ma (Tudge 2000), with the latter giving rise to the lungfishes, the coelacanths and all tetrapods. Even though it is clear that a fish-specific whole-genome duplication occurred in the ray-finned species (Amores et al. 1998; Taylor et al. 2003; Jaillon et al. 2004; Meyer & Van de Peer 2005), the lack of genome information for more basal actinopterygians means that it is presently very difficult to determine where this event took place within the stem lineage of the Actinopterygii. Nonetheless, it is estimated that the genome duplication occurred between 226 and 350 Ma (Meyer & Van de Peer 2005; Van de Peer et al. 2009). Many ray-finned fishes dwell in brightly lit and colourful aquatic environments (e.g. coral reefs and the upper water column of the ocean), and the presence of multiple copies of opsin genes in many actinopterygian fishes, mediated by the whole-genome duplication event, formed the basis for the evolution of additional spectrally distinct pigments. Therefore, the development of such a large inventory of visual photosensory receptors gave these species a selective advantage that allowed them to colonize a plethora of spectrally broad ecological niches and evolve a diverse palette of body coloration that forms the basis of complex sexual selection relationships. As a result, multiple duplicated opsin genes have become fixed in the genomes of teleosts and in many cases serve as an evolutionary mixing pot from which the photopigment repertoire may be tailored, through further gene duplication or loss, by selection pressures bestowed by ecological constraints (Trezise & Collin 2005).

The genomes of a number of teleosts have now been fully sequenced, allowing the full complement of opsin genes to be determined. The study of one of these model teleosts, the zebrafish (Danio rerio), a cyprinid and member of the superorder Ostariophysi, shows that, in addition to two rod RH1 genes [RH1.1 on chromosome 8 and RH1.2 on chromosome 11 (GenBank accession no. HM367062; W.I.L Davies, unpublished) (Morrow et al. 2011)], a total of eight cone opsin genes are present comprising two LWS genes (LWS1.1 and LWS1.2), four RH2 genes (RH2.1, RH2.2, RH2.3 and RH2.4), a single SWS2 gene and a single SWS1 gene (Chinen et al. 2003; Takechi & Kawamura 2005) (Fig. 3). Although the RH1 genes are found on different chromosomes and probably arose by gene duplication, both the LWS (which is syntenic with the SWS2 gene) and RH2 genes are located in tandem gene clusters (Chinen et al. 2003), which indicates that these gene duplications most likely did not arise from the initial whole-genome duplication, but from a mechanism of individual gene duplication (Fig. 3a), similar to the process that gave rise to the multiple LWS genes present within the primate LWS opsin gene array (Dulai et al. 1999). The peak sensitivities of the LWS1.1 and LWS1.2 pigments are at 558 and 548 nm, respectively, and the four RH2 pigments range from 467 to 505 nm (Chinen et al. 2003) (Fig. 3b). The genes show major differences in expression, with the two LWS genes and three of the RH2 genes (RH2.1, RH2.3 and RH2.4) showing significantly lower expression than RH2.2, SWS1 and SWS2 genes; however, this differential expression profile may be due to inconsistencies between the age of the fish being studied (2 months vs. 1 year) or the time of sampling postillumination (1.5 h vs. 7 h) (Chinen et al. 2003). For both the LWS and RH2 opsin gene classes, the longer-wavelength subtypes are found later in development and are restricted to the ventro-nasal and peripheral regions of the adult retina, whereas shorter-wavelength subtypes are expressed earlier and are confined largely to the central to dorsal retina (Takechi & Kawamura 2005). As yet, the ecological significance of these temporally regulated retinal mosaics remains unclear. The zebrafish is not unique in possessing duplicated LWS and RH2 genes as they are also found in many other species of teleosts (although only representative examples are detailed in this review), along with duplications of the SWS2 gene class (as discussed below). By contrast, multiple copies of the SWS1 and RH1 pigment genes are less common (Davies 2011), with duplications of the RH1 gene found, so far, only in the zebrafish (W.I.L. Davies, unpublished; Morrow et al. 2011), in a few other cyprinids [e.g. two further members of Danio spp., a member of the Epalzeorhynchos spp. (Morrow et al. 2011) and the common carp, Cyprinus carpio (Lim et al. 1997)], in the European eel, Anguilla anguilla (Cottrill et al. 2009), and Japanese eel, Anguilla japonica (Zhang et al. 2000) (order Anguilliformes), and in the deep-sea pearl-eye, Scopelarchus analis (Pointer et al. 2007), a member of the order Aulopiformes. Duplications of the SWS1 gene have been reported, so far, only in the smelt, Plecaglossus altivelis (Minamoto & Shimizu 2005), a member of the order Osmeriformes.

image

Figure 3.  (a) The genomic structure of the ten zebrafish visual opsin genes, showing their chromosomal locations and syntenic relationships (Chinen et al. 2003). The exons are coloured as follows: LWS (red), SWS1 (violet), SWS2 (blue), RH2 (green) and RH1 (black). The zebrafish LWS-activating region (LAR) (Tsujimura et al. 2010), a region similar to the locus control region (LCR) upstream of primate L opsin and M opsin genes, is highlighted in orange. The arrows indicate the orientation of each gene in a 5′ to 3′ direction. (b) Representative absorbance spectra (Govardovskii et al. 2000) for each visual pigment expressed in the retina of the zebrafish (colour-coded as in (a)), showing the range of peak spectral sensitivity values for each cone (solid lines) (Chinen et al. 2003) and both rod (dotted line) photopigments (Chinen et al. 2003; Morrow et al. 2011).

Download figure to PowerPoint

A number of teleosts possess LWS gene duplications, including the zebrafish discussed above; the blind cavefish, Astyanax fasciatus (order Characiformes), where the two forms encode spectrally different pigments (Parry et al. 2003); the medaka, Oryzias latipes (order Beloniformes), that has two LWS opsin genes alongside a single SWS1 gene, two SWS2 genes and three RH2 genes (Matsumoto et al. 2006); four LWS genes (and a duplication of the SWS2 gene) in the green swordtail, Xiphophorus helleri (order Cyprinodontiformes) (Watson et al. 2010); and either three copies of the LWS gene in the one-sided livebearer, Jenynsia onca (Windsor & Owens 2009), or four LWS genes in the four-eyed fish, Anableps anableps (Owens et al. 2009), where these two latter species (order Cyprinodontiformes) express a further six opsin genes (a single copy of the SWS1 and RH1 (rod) genes and two SWS2 and RH2 genes) (Owens et al. 2009). Initially, it was reported that the guppy, Poecilia reticulate, another cyprinodontiform species, may possess three spectrally distinct LWS cone classes that express at least six LWS transcripts (Hoffmann et al. 2007; Weadick & Chang 2007), in addition to four other cone opsins (SWS1, SWS2 and two RH2 genes) and a single rod (RH1) gene (Hoffmann et al. 2007; Laver & Taylor 2011). However, a subsequent study, using bacterial artificial chromosome (BAC) libraries, found that three LWS genes were present on a single chromosome, accompanied by a fourth intronless ‘retrogene’ (Watson et al. 2011). Thus, similar to the green swordtail and the four-eyed fish, the guppy also expresses four LWS genes, the most that has been identified to date (Owens et al. 2009; Watson et al. 2010, 2011). In the guppy, such a large complement of distinct genes may be functionally important as some have been shown to specify pigments with altered spectral absorbance peaks (Watson et al. 2011) and putative changes in G-protein activation (Hoffmann et al. 2007; Weadick & Chang 2007). Overall, the level of expression of these and other opsins is associated with both the sex and age of these fish (Laver & Taylor 2011).

Cichlid fish, members of the order Perciformes, have undergone a massive species radiation in the African Great Lakes that has resulted in over seven hundred species with diverse colour patterns (Smith & Kornfield 2002; Kocher 2004). Multiple opsin gene duplications are found in these species (Carleton & Kocher 2001; Carleton et al. 2005; Parry et al. 2005; Spady et al. 2005), with differential expression within the multigene complexes clearly associated with adaptations to different photopic environments (Parry et al. 2005; Hofmann et al. 2009, 2010). Species generally express three cone pigments that differ in peak sensitivity depending on the habitat. The Mbuna or rock-dwelling species that inhabit clearer water have double cones (with principal and accessory members—see Fig. 7 for an avian example) that peak at 535 and 488 nm and ultraviolet (UV)- or short-wavelength-sensitive (perceived as blue) single cones that peak at 370 nm or 420 nm, respectively. By contrast, non-Mbuna (also known as Haps, which derives from their original membership of the broad genus Haplochromis) or sand-dwelling species that inhabit more turbid waters possess double cones at 570 and 535 nm and single cones at 450 nm and, thereby, show reduced sensitivity to short-wavelength light. A striking feature of the opsin gene duplications that underlie these sensitivities is the presence of three RH2 genes and two SWS2 genes (Parry et al. 2005), with different subsets expressed in diverse species. Within the cichlids of Lake Victoria, male nuptial coloration varies from red to blue within different populations of fish derived from closely related Pundamilia species that occupy distinct visual environments from clear water to more turbid aquatic environments. Significantly, the LWS genes have mutated at specific ‘tuning sites’ to produce photopigments with peak sensitivities that differ between species, and these differences coincide with alterations in the coloration of the male fish body, with the red-coloured species having the more long-wavelength-shifted LWS pigments than the species coloured blue. Moreover, individuals that dwell in turbid water possess more long-wavelength-shifted cones (Carleton et al. 2005). The combined effect of these variations may therefore lead to changes in the perceived intensity and spectral quality of downwelling or reflected light that may account for the evolution of variants in male coloration as a step towards speciation (Seehausen et al. 2008).

Ontogenetic differences in expression are also present in these cichlid fishes, with only one of two copies of the SWS2 gene and two of three copies of the RH2 gene expressed in the adult retina, alongside single copies of the LWS and SWS1 genes, with the other SWS2 and RH2 duplicated genes expressed at earlier developmental stages. Subfunctionalization through differential ontogenetic expression may be a significant mechanism, therefore, for the conservation of particular opsin genes (Spady et al. 2006).

In the black bream, Acanthopagrus butcheri, a member of the order Perciformes, the relative proportions of the different cone photoreceptors have been shown to change during development (Shand et al. 2002), with larval fish having cones that are sensitive to shorter wavelengths, while both juveniles and adults possess more LWS cones. The cone opsin gene complement comprises a SWS1 gene, two SWS2 genes, two RH2 genes and a single LWS gene (Shand et al. 2008). This demonstrates once again that individual teleost species have undergone rapid genomic changes to yield diverse opsin gene profiles and the expression of these genes at different developmental phases correlates with the differing frequencies of cone classes. In the larvae, the SWS1 gene is expressed alongside the SWS2 gene encoding a photopigment with a shorter λmax. By contrast, the SWS1 gene appears to be switched off in the adult retina, accompanied by a switch to the other SWS2 gene encoding a pigment with a longer λmax and the commencement of LWS expression (Shand et al. 2008). The overall effect of these changes is to extend the sensitivity of adult fish to longer wavelengths of light, a shift that correlates with the ecological change in life cycle from relatively clear water to one that is stained by reddish-coloured tannins (Shand et al. 2008).

A developmental sequence of opsin gene activation is also found in the Pacific pink salmon, Oncorhynchus gorbuscha, where single cones shift from the expression of a UV-sensitive (UVS) pigment to a short-wavelength-sensitive pigment (perceived as blue) during development (Cheng & Flamarique 2004). It is now apparent that duplicated copies of different opsin gene families in many teleosts exhibit developmental changes in gene expression; such plasticity in the spectral sensitivities of photoreceptors may therefore operate to modify colour vision as the lifestyle of the fish species changes.

A particularly interesting example of a developmental progression occurs in the European eel, A. anguilla, a member of the order Anguilliformes, which spends different stages of its life cycle in diverse aquatic environments (Berry et al. 1972; Tesch 1977). After hatching in the Sargasso Sea at depths >200 m (van Ginneken & Maes 2005), the leptocephalous larvae drift to the European coasts, where they metamorphose into larval glass elvers that enter estuaries and rivers. The eels remain in freshwater for many years and undergo further metamorphic changes into elvers (juvenile) and sexually mature adult eels, before migrating back to the deep-sea breeding grounds (Tesch 1977). In this cycle, they move from clear surface oceanic water, through green coastal waters, into yellow or brown stained freshwater and then to the blue waters of the deep sea. As an adaptation to these differing photic environments, the European eel alters the λmax of its rod pigment at the time of migration from river to sea and at the start of migration to deep water–spawning sites in the Atlantic (Carlisle & Denton 1957). A change in the λmax from around 523 nm to near 501 nm is achieved by the replacement of the retinal chromophore (from one based on vitamin A2 to one derived from vitamin A1). This is followed by a shift in the λmax to around 482 nm as a result of a change in opsin gene expression (Wood & Partridge 1993). Unusually, two different rod (RH1) opsin genes exist in the genome, and the switch from the long-wavelength-shifted or ‘freshwater’ form to the short-wavelength-shifted or ‘deep-sea’ form occurs within existing rod photoreceptors, such that the deep-sea form progressively occupies more of the photoreceptor outer segment (Archer et al. 1995). Phylogenetic analysis shows that the RH1 gene duplication that produced these two forms occurred in the Anguilliformes lineage (Hope et al. 1998). Eels also possess two cone classes expressing an RH2 and SWS2 pigment, respectively, but throughout metamorphosis, these cones are virtually absent (Bowmaker et al. 2008).

Molecular changes in photopigment genes associated with deep-water environments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Although many fishes occupy brightly lit environments and possess complex eyes that express a multitude of spectrally tuned visual pigments, others have colonized the more dimly lit environments of deep lakes and oceans. The effect of the transmission of light through a body of water is not only an attenuation of intensity with a maximum limit for vision of about 1000 m in the clearest tropical oceans (Jerlov 1976; Denton 1990), but also a change in spectral composition, with maximum penetration occurring in the short-wavelength region of the spectrum around 480 nm (Fig. 4). In the deep water of clear, tropical oceans, the ambient light that penetrates the sea at high noon on bright days is therefore composed of dim-downwelling light that is generally restricted to shorter wavelengths at depths <1000 m (Marshall 1979), although the actual intensity and spectral composition of available downwelling light differs across the globe and is dependent on geographical location, oceanic salinity, the time of day and relative weather conditions (Douglas et al. 1998). Even though bioluminescence is found at all depths with variations in its frequency and intensity (Nicol 1969; Herring 1983), it is particularly important as a perceivable light source (e.g. below 1000 m), where vision is not dependent upon penetrable wavelengths of sunlight (Nicol 1969; Herring 1983).

image

Figure 4.  A schematic representation showing the effect of water depth on the transmission of oceanic and coastal sunlight (top panel), where the intensity and spectral composition of sunlight diminish with increasing depth. At 250 m, the only wavelengths detected are in the range of 470–480 nm, with little or no visibly relevant sunlight present at depths >1000 m in any ocean. Representative fishes that dwell at different depths are included to illustrate how visual systems adapt to differing photic environments. For example, (a) Geotria australis (1–200 m) possesses five cone-like photopigments that are sensitive to UV and longer wavelengths (Collin et al. 2003b; Davies et al. 2007b); (b) Callorhinchus milii (6–500 m) expresses three cone (in addition to a single rod) photopigments that are sensitive to both short- (but not UV) and long-wavelengths (Davies et al. 2009a); (c) Petromyzon marinus (1–2200 m) possesses two cone-like photopigments that are sensitive to middle- and long-wavelengths only (Davies et al. 2009b); and (d) Histiobranchus bathybius (1790–4790 m) that possesses only a short-wavelength-shifted (λmax at 477 nm) rod (RH1) photopigment (Hope et al. 1997).

Download figure to PowerPoint

As discussed above, the anadromous southern hemisphere lamprey, Geotria australis, lives in the brightly lit upper water column (1–200 m) of oceanic waters in its marine phase (Potter et al. 1979) and possesses five cone photoreceptor types, each expressing one of the five distinct cone-like visual pigments (Govardovskii & Lychakov 1984; Collin et al. 2003a,b; Collin & Trezise 2004) (Fig. 4a). By contrast, in the northern hemisphere sea lamprey, Petromyzon marinus, the complement of pigments is reduced to only two (LWS and RHA/RH1 genes) (Zhang & Yokoyama 1997; Davies et al. 2009b) (Fig. 4c), with the functional loss of SWS1, SWS2 and RHB/RH2 genes, although SWS1 and SWS2 are retained in the genome as nonfunctional pseudogenes (Davies et al. 2009b). Thus, it appears that the dimly lit and spectrally restricted light environment experienced by P. marinus at depths that reach 2200 m (Froese & Pauly 2009) has resulted in the retention of only limited colour vision that lacks pigments that are spectrally tuned to the extremes of the light spectrum (Davies et al. 2009b).

A number of members of the Chondrichthyes or cartilaginous fishes also dwell in deep oceanic waters; however, there appears to be an absence of chondrichthyan species at depths below ∼2500 m (Priede et al. 2006). By contrast, many actinopterygian representatives dwell at much greater depths, with the cuskeel, Abyssobrotula galatheae, holding the record as the deepest fish ever found (8370 m) (Nielson 1977). Cartilaginous fishes constitute the most basal and oldest extant group of the gnathostome lineage, where they share a common ancestor with all other jawed vertebrates at least 450 Ma (i.e. sarcopterygians and actinopterygians) (Sansom et al. 1996). Broadly divided into two lineages, these fishes comprise the holocephalans (chimaeras) and the elasmobranchs [sharks (Selachii), and the rays and skates (Batoidea)], with the chimaeras diverging about 374 Ma (Cappetta et al. 1993). Evidence that all four vertebrate cone classes survived into this lineage is, so far, lacking. The elephant shark, Callorhinchus milii, a chimaera that inhabits the continental shelf at 200–500 m (Last & Stevens 1994), possesses only two cone opsin gene classes, RH2 and LWS, with a duplication of the LWS gene to give two spectrally different LWS pigments (Davies et al. 2009a) (Fig. 4b). Like P. marinus, this cartilaginous fish has lost both SWS1 and SWS2 pigment genes but has retained the LWS and RH2 genes. The elephant shark spawns in the more brightly lit waters of estuaries and shallow bays (6–30 m), and the presence of cones expressing LWS and RH2 pigments will provide for photopic colour vision in these environments. The elephant shark has a second copy of the LWS gene, which encodes a pigment that is spectrally tuned towards shorter wavelengths; the expression of these three pigments in separate cones may therefore provide trichromatic vision (Davies et al. 2009a).

Although skates are reported to have a rod-only retina (Dowling & Ripps 1970; Szamier & Ripps 1983), many sharks and rays possess both cones and rods in various proportions (Hart et al. 2004). The retinae of three species of ray, Rhinobatos typus, Aptychotrema rostrata and Dasyatis kuhlii, all have three spectral classes of cones with λmax values for each class between 459 and 477 nm; 492 and 502 nm; and 552 and 561 nm, respectively (Hart et al. 2004; Theiss et al. 2007), with the potential for trichromatic colour vision. By contrast, a recent study of seventeen species of sharks demonstrated the presence of just a single cone class with λmax values ranging from 531 to 560 nm (Hart et al. 2011). These shark species are therefore cone monochromats and may be colour-blind (Hart et al. 2011), although some colour visual input via the rods may be perceived under mesopic conditions (discussed below). The molecular identities of the pigments expressed in the cones of the shark species studied by Hart et al. (2011) have yet to be determined, although the spectral peaks of these pigments would indicate the presence of an LWS photopigment. Consistent with this, an LWS opsin gene has been shown to be expressed in the retinae of another group of sharks, the wobbegong or carpet sharks (S.M. Theiss, W.I.L. Davies, N.S. Hart, S.P. Collin, & D.M. Hunt, unpublished).

Unlike the sparse molecular information available for cartilaginous fish photopigments, the visual systems of deep-sea teleosts are better defined and show many adaptations to a photon-restricted environment. These include complete cone photoreceptor loss, and hence loss of colour vision in the majority of species (Partridge et al. 1988, 1989), and the tuning of the λmax of the rod visual pigment to match the wavelengths of the available light (Partridge et al. 1988, 1989; Hunt et al. 2001); for example, Histiobranchus bathybius, a deep-sea eel that dwells at a depth of 1790–4790 m (Merrett et al. 1991a,b), expresses a rod pigment with a λmax at 477 nm (Hope et al. 1997) (Fig. 4d). Nevertheless, cone photoreceptors have been retained in a few species occupying the deep ocean. Examples are the lantern fish, Lampanyctus crocodilus, Benthosema glaciale and Myctophum punctatum (members of the superorder Scopelomorpha), that possess a rod-dominated retina but retain a few cones mainly distributed in the central retina (Bozzano et al. 2007). The spectral characteristics and underlying visual pigments in these cones remain, however, unknown. In fact, the only deep-sea species in which the corresponding cone opsin genes have been identified is the pearl-eye, Scopelarchus analis. Although occupying depths of 500–1000 m, it has retained an expressed RH2 gene, in addition to a duplicated rod (RH1) opsin gene (Pointer et al. 2007).

Changes in the retention and tuning of pigments are also seen in freshwater species of cottoid fish (order Scorpaeniformes) that are endemic to Lake Baikal in Eastern Siberia, providing a natural experiment of visual adaptations to a dim-light environment. Lake Baikal is one of the oldest (and clearest) lakes in the world and, with depths in excess of 1600 m, the deepest freshwater lake. Even at such depths, however, the oxygenation levels are at 75-80% of surface levels (Weiss et al. 1991). The lake contains a species flock of cottoid fish that occupy different depths; parallels with deep-sea fish are clearly evident in the deeper-dwelling Baikal cottoids where large photoreceptor size, loss of cone photoreceptors and the tuning of visual pigments to shorter wavelengths have been reported (Bowmaker et al. 1994; Hunt et al. 1996). The molecular mechanism underlying spectral shifts in rod visual pigments to shorter wavelengths involves largely sequential amino acid substitutions at four key tuning sites within the RH1 opsin protein (Hunt et al. 1996). The surface-dwelling species have retained LWS cones but these are lost in species inhabiting deeper water. Like the rod pigments, the cone photopigments are also sensitive to shorter wavelengths in deeper-dwelling species and, for the SWS2 pigment, the key substitutions have been identified (Cowing et al. 2002a).

Transition to land: a spectral challenge?

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Lungfishes (Dipnoi) have long been considered to be the link between aquatic and terrestrial vertebrates (Yokobori et al. 1994). Their visual systems are more similar to those of terrestrial vertebrates than that of other sarcopterygian fish (e.g. the coelacanth, Latimeria spp.), which suggests that lungfishes, and not the coelacanths, are the nearest extant sister group to the antecedents of the tetrapods. The correlation, however, between all early sarcopterygian fishes remains controversial despite new methods of phylogenetic analyses of nucleotide and amino acid sequences (Carroll 1997; Brinkmann et al. 2004; Takezaki et al. 2004). In these phylogenetic studies, the lungfishes appear to be more closely related to the tetrapods than to Latimeria chalumnae (Brinkmann et al. 2004), while other investigations support a trichotomy that remains unresolved (Takezaki et al. 2004). Three lungfish species remain extant and are geographically located in Africa, South American and Australia. The antipodean species, Neoceratodus forsteri, possesses several cone classes that may be distinguished by the brightly coloured oil droplets resident within the photoreceptor inner segments (Robinson 1994; Bailes et al. 2006), a feature that they share with many squamates, birds and some mammals. At the molecular level, a rod opsin and representatives of the four major vertebrate classes of cone pigments are present (Bailes et al. 2007) (Fig. 5a). In each case, the expressed sequences appear to be derived from a single copy of each gene, indicating that the sarcopterygian lineage separated prior to the whole-genome duplications that characterize the actinopterygian lineage. The presence of an intricate colour visual system in lungfishes, with multiple intracellular spectral filters and cone subtypes, suggests that the movement onto land did not significantly affect the photopigment repertoire of these terrestrial pioneers. However, a greater effect was observed regarding photoreceptor morphology and ocular optics. As several of these optical and cellular characteristics are also observed in terrestrial or secondarily aquatic vertebrates, it may be that they evolved in shallow water prior to the transition onto land.

image

Figure 5.  Representative absorbance spectra (Govardovskii et al. 2000) for visual photopigments expressed in the retinae of (a) a sarcopterygian fish, Neoceratodus forsteri; two amphibians, (b) the salamander, Ambystoma tigrinum, and (c) the caecilian, Typhlonectes natans; (d) a squamate reptile, Anolis carolinensis; (e) a bird, Serinus canaria; (f) a monotreme, Ornithorhynchus anatinus; (g) a marsupial, Sminthopsis crassicaudata; and two eutherian mammals (h) Mus musculus and (i) Homo sapiens. Peak spectral sensitivity values for both cone (solid line) and rod (dotted line) photopigments are shown. Spectra are colour-coded to illustrate the molecular basis for each photopigment as follows: LWS (red), SWS1 (violet), SWS2 (blue), RH2 (green) and RH1 (black).

Download figure to PowerPoint

The rod-dominated retinae of the coelacanth, Latimeria chalumnae, expressing only a single cone opsin encoded by a RH2 gene (Yokoyama et al. 1999), are more typical of a deep-sea fish (Millot & Carasso 1955; Locket 1973). The coelacanth dwells in a photon-restricted deep-sea ecology at depths of between 100 and 400 m in the Indian Ocean; the descent to this low-light, short-wavelength-shifted environment may be the major factor in the functional loss of the SWS1 gene (it has become a pseudogene) and the complete loss of the SWS2 and LWS genes, revealing a direct correlation between the spectral habitat and functional requirements of a species and the loss of particular opsin genes (Yokoyama et al. 1999).

Amphibians represent the next stage in the evolutionary transition onto land, and despite living in semi-dry terrestrial environments as adults, they must return to water to spawn. Phylogenetically, amphibians are grouped into the Anurans (toads and frogs), the Caudata (newts, salamanders and urodeles) and the burrowing, legless caecilians (Gymnophiona) (Tudge 2000). The caecilians generally possess rudimentary eyes and rod-only retinae (Mohun et al. 2010) (Fig. 5c), whereas duplex retinae are found in all other amphibians (Fig. 5b). Uniquely, anurans and some salamanders possess two spectral classes of rod-shaped photoreceptor: a majority class (90–95%) that expresses a rod RH1 pigment with a λmax at ∼500 nm (termed ‘red’ rods) and a minority class that expresses an SWS2 cone pigment gene (termed ‘green’ rods) (Hisatomi et al. 1999; Ma et al. 2001a; Darden et al. 2003). These two groups of amphibians also possess an LWS pigment that is expressed in double cones (Sherry et al. 1998; Makino et al. 1999; Rohlich & Szel 2000), with single cones expressing either a ‘UV-sensitive’ (SWS1) pigment or a ‘blue-sensitive’ (SWS2) pigment. The latter pigment is indistinguishable to that expressed in the ‘green’ rods (Harosi 1975; Deutschlander & Phillips 1995; Hisatomi et al. 1998; Ma et al. 2001a,b; Takahashi et al. 2001), where it is paired and interacts with the rod isoform of the G-protein transducin rather than the cone isoform found in short-wavelength-sensitive (SWS2) ‘blue’ cones (Ma et al. 2001b).

One class of cone pigment that is consistently missing from amphibians is that encoded by the RH2 gene (Fig. 5b, c). Colour vision is therefore at best trichromatic, with middle-wavelength cones in the newt containing a SWS2 pigment with a λmax value that is shifted to 474 nm, some 40 nm more long-wavelength-shifted than that found in other amphibians (Takahashi & Ebrey 2003).

Mammals return to the deep sea

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

As discussed below, mammals originated as species that possessed a trichromatic visual system based on photopigments encoded by SWS1, SWS2 and LWS opsin genes. However, like many fish that dwell in deep oceanic waters, the majority of aquatic mammals studied so far have lost a functional SWS1 gene (in addition to the loss of the SWS2 gene in the therian lineage), but retain an intact LWS gene; thus, these species are predicted to have dispensed with colour vision altogether (Figs 5 and 6, and Table 1). The majority of mammals that have returned to the sea may be grouped within the Cetacea (dolphins and whales) or the Pinnipedia (walruses, sea lions and seals). The evolution of the cetaceans followed an independent path from the Pinnipedia, which are in turn derived from the Carnivora, which is closely related to the Artiodactyla (Tudge 2000). A functional SWS1 gene has, however, been lost in both groups and as a result lacks short-wavelength-sensitive (blue) cones (Peichl & Moutairou 1998; Levenson & Dizon 2003; Newman & Robinson 2005; Levenson et al. 2006). Thus, these species appear to be colour-blind, retaining only a single cone subtype expressing a LWS gene and a RH1 gene expressed in the rod photoreceptors. Within the respective genomes of those animals studied thus far, the SWS1 gene is still recognizable as a pseudogene that has accumulated numerous deleterious mutations (Levenson & Dizon 2003; Newman & Robinson 2005; Levenson et al. 2006). This is consistent with a relaxation of selection pressure to maintain colour vision in aquatic mammals that spend a significant amount of time foraging in dimly lit, deep-sea environments. The loss of SWS1 gene function and the preferential retention of the LWS gene is presumably a reflection of the relative paucity of short-wavelength-sensitive cones and hence their lesser impact on acuity than the loss of LWS cones.

image

Figure 6.  A phylogeny of the mammalian lineage showing the presence of UV-sensitive (UVS) (purple) and violet-sensitive (VS) (blue) SWS1 pigments. The evolutionary course of replacements at site 86 is specified by their location on the tree. In general, the retention of Phe86 is invariably associated with the preservation of UVS photopigments, with primates being the only exception. Pseudogenes are denoted by a dagger. Branch lengths do not reflect evolutionary distances. Modified from Hunt et al. (2009a).

Download figure to PowerPoint

Table 1.   The residues present at five spectral tuning sites (180, 197, 277, 285 and 308), found in transmembrane helices IV, VI and VII, and the second extracellular loop of the LWS/MWS pigments expressed in a subset of mammals, known to influence spectral sensitivity (Yokoyama & Radlwimmer 2001). These sites are numbered based on the human L opsin protein sequence, where residues 180, 197, 277, 285 and 308 correspond to amino acids 164, 181, 261, 269 and 292 in the bovine rod opsin sequence, respectively. The peak spectral sensitivities (nm) of photopigments determined for each species are indicated
SpeciesResidue at tuning siteλmax (nm)
180197277285308
Human (Homo sapiens) (L opsin)SerHisTyrThrAla565*
Human (Homo sapiens) (M opsin)AlaHisPheAlaAla530*
Mouse (Mus musculus)AlaTyrTyrThrSer508
Rat (Rattus norvegicus)AlaTyrTyrThrSer509
Rabbit (Oryctolagus cuniculus)AlaTyrTyrThrSer509
Manatee (Trichechus manatus)SerHisTyrThrAla556§
Elephant (Loxodonta africana)AlaHisTyrThrAla552
Pilot whale (Globicephala melas)AlaHisTyrThrSer531§
Harbor porpoise (Phocoena phocoena)AlaHisTyrThrSer522§
Bottlenosed dolphin (Tursiops truncates)AlaHisTyrThrSer524**
Harbor seal (Phoca vitulina)SerHisPheThrAla548§
Harp seal (Phoca groenlandica)SerHisPheThrAla548§
Honey possum (Tarsipes rostratus)AlaHisTyrThrAla557††
Fat-tailed dunnart (Sminthopsis crassicaudata)AlaHisPheAlaAla530††
Quenda (Isoodon obesulus)AlaHisTyrThrAla551‡‡
Quokka (Setonix brachyurus)AlaHisTyrAlaAla538‡‡
Tammar wallaby (Macropus eugenii)AlaHisPheAlaAla530§§
Platypus (Ornithorhynchus anatinus)AlaHisTyrThrAla550¶¶

A third class of aquatic mammals, the Sirenia, are affiliates of the Mesaxonia and, as such, the elephants (Proboscidea) are their closest terrestrial relatives. Of the four extant species of the Sirenia, only the manatees have so far been studied. Unlike the cetaceans and pinnipeds, however, manatees have retained a functional SWS1 gene (Newman & Robinson 2006b) (Fig. 6) and have been shown to possess dichromatic colour vision (Cohen et al. 1982; Griebel & Schmid 1996), thereby enabling them to exploit shallow estuarine and coastal water habitats where they forage on immersed seagrasses.

Diurnality in terrestrial vertebrates

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Reminiscent of the Osteichthyes before them, the LWS, SWS1, SWS2, RHB/RH2 cone pigment classes have been conserved in many species of squamates and birds (Yokoyama 2000; Davies 2011) (Fig. 5).

The class Reptilia is divided into two subclasses: the first, Synapsida, gave rise to the therapsids and eventually the mammals and the second, Sauropsida, encompasses a further division into the Anapsida and Diapsida (Tudge 2000). Modern sauropids comprise the Testudines (turtles, tortoises and terrapins), the Sphenodontia (tuatara), the Squamata (snakes and lizards) and the Archosauromorpha, the latter of which is further divided into the relatively ancient Crocodilia (crocodiles and alligators) and the Aves (birds) (Tudge 2000). The anoline lizards are the most extensively studied, where they have been shown to possess a retina that consists solely of single and double cones, each containing a coloured oil droplet. These oil droplets contain carotenoid pigments that are spectrally matched to the absorbance properties of a particular cone class and act as cut-off filters to reduce noise and increase spectral discrimination. In the tetrachromatic green anole, Anolis carolinensis, four cone classes are present that are spectrally distinct with λmax values at 365, 455, 495 and 564 nm (Provencio et al. 1992; Kawamura & Yokoyama 1997, 1998; Loew et al. 2002) (Fig. 5d). There is also a low level of rod (RH1) expression (McDevitt et al. 1993; Kawamura & Yokoyama 1997) that encodes for a photopigment that peaks at 491 nm (Provencio et al. 1992; Kawamura & Yokoyama 1997, 1998; Loew et al. 2002). This raises the prospect that either a minor rod population is present or the RH1 opsin gene is co-expressed in cones alongside a cone pigment.

The retinae of turtles also contain spectrally distinct double and single cones. The LWS visual pigment is found in both members of the double cones but a coloured oil droplet is only contained within the principal inner segment (Lipetz 1984; Ohtsuka 1985a,b; Ohtsuka & Kawamata 1990; Loew & Govardovskii 2001). The two members are therefore spectrally distinct despite expressing the same LWS opsin gene. Separate to the double cones, four spectral classes of single cone are present, each containing a coloured oil droplet and therefore providing the potential for tetrachromatic colour vision.

Crocodilian reptiles are most closely related to the dinosaurs and birds (discussed below) (Janke & Arnason 1997). They have a rod-dominated retina with double and single cones that are devoid of oil droplets. Four spectral classes of cones are present in the Mississippi alligator, Alligator mississippiensis, with two classes of double cones that contain pigments with λmax values at 566 and 503 nm and two classes of single cones that have λmax values at 535 and 443 nm (Sillman et al. 1991). In the absence of a clear understanding of the role of double cones in this species, the presence of tetrachromacy remains uncertain. The same conclusion applies to the spectacled caiman, Caiman crocodiles, which possesses single cones that are spectrally similar to those of the alligator and double cones where either both members express a 535 nm pigment or one member expresses the 535 nm pigment and the other a 506 nm pigment (Govardovskii et al. 1988).

Avian colour vision is thought to be a major player in the evolution of species with highly coloured plumage and the complex networks of mate selection that many species display. All major cone opsin gene classes are retained within the avian lineage, providing the potential for tetrachromacy. Unlike many squamates, however, the LWS pigment of diurnal birds is expressed in both halves of their double cones as well as in single cones. The other three cone opsin genes, RH2, SWS2 and SWS1, are only found in single cones (Hart & Hunt 2007). In avian species, about 50% of the total population of cone photoreceptors comprises double cones. Of the remaining cone types, about 40% are single cones expressing either RH2 or LWS opsin genes, with the remaining 10% consisting of single cones that contain either SWS1 or SWS2 photopigments (Bowmaker 2008). Avian photoreceptors contain coloured oil droplets, a feature they share with many squamates (Fig. 7); a large pale yellow (P-type) droplet is found in the principal member of double cones and functions as a cut-off filter to wavelengths shorter than 460 nm, whereas diffuse pigment or a small oil droplet may be observed in the inner segment of the accessory member. A red (R-type) droplet is found in single cones expressing the LWS gene and serves to cut off wavelengths at about 560 nm, whereas those photoreceptors expressing the RH2 gene (with a yellow Y-type droplet) or the SWS2 gene (with a C-type droplet) possess oil droplets that cut off at about 505 nm and 410–440 nm, respectively (Hart & Hunt 2007; Bowmaker 2008). SWS1 cones that are either UV-sensitive (UVS) or violet-sensitive (VS) have a transparent (T-type) droplet that freely transmits wavelengths above 350 nm (Bowmaker et al. 1997; Hart et al. 2000). Except for the SWS1 pigments, which vary from UVS to VS, the peak sensitivities of avian pigments are generally conserved, although notable exceptions are found in the LWS pigments of the Humboldt penguin, Spheniscus humboldti, which peaks at 543 nm (Bowmaker & Martin 1985), and the tawny owl, Strix aluco, which peaks at 555 nm (Bowmaker & Martin 1978). However, the molecular basis (via substitutions within the opsin proteins) for these short-wavelength shifts has yet to be determined.

image

Figure 7.  A summary diagram illustrating the complement of single and double cones (including oil droplets) and rods in the avian diurnal retina. Depending on the bird species, the SWS1 pigments peak either in the UV (360–380 nm) or in the violet (400–420 nm) region of the light spectrum. Oil droplets are labelled as follows: T (transparent), C (clear or colourless), Y (yellow), R (red) and P (pale yellow). Note that values for the spectral cut-off filter of P-type oil droplets are ill-defined and the accessory member of avian double cones usually has a regressed or absent oil droplet.

Download figure to PowerPoint

All avian UVS pigments appear to be derived from an ancestral avian VS pigment, which was in turn derived from an ancestral vertebrate UVS pigment: therefore, spectral sensitivity in the UV range in birds has been ‘re-invented’. All nonavian UVS SWS1 pigments possess Phe at site 86 in the opsin protein (Cowing et al. 2002b; Hunt et al. 2004), and this is replaced in avian VS pigments by Ser (Carvalho et al. 2007). The shift to UV sensitivity in avian SWS1 pigments is achieved by a second change with Cys replacing Ser at site 90 (Wilkie et al. 2000; Yokoyama et al. 2000). This change is found in avian UVS pigments that derive from several orders of bird (Odeen & Hastad 2003); their phylogenetic relationships suggest that UVS pigments have evolved from the ancestral avian VS pigment on a number of separate occasions (Fig. 8).

image

Figure 8.  The phylogenetic relationships between representative members of the Aves, showing the presence of either ultraviolet-sensitive (UVS) or violet-sensitive (VS) SWS1 pigments. The ancestor to all birds is likely to have had a VS photopigment that derived from a UVS SWS1 pigment that arose in the early vertebrates. Subsequently, a few species of bird (e.g. members of the Passeriformes, Psittaciformes, Trogoniformes, Ciconiiformes and Struthioniformes) have ‘reinvented’ SWS1 photopigments that possess λmax values in the UV range of the visible spectrum via a Ser90Cys substitution (modified from Hunt et al. (2009a)). Branch lengths do reflect evolutionary distances.

Download figure to PowerPoint

Crepuscular and nocturnal adaptation to dim-light environments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

The diverse group of squamate reptiles includes the geckos, iguanid lizards, true chameleons, skinks, monitor lizards and the snakes. Lifestyles vary from diurnal to crepuscular and nocturnal, and it has been proposed that some species have evolved from diurnal to nocturnal and back again, with consequent transformation of cones into rods, before returning to a more cone-like morphology, as suggested by Walls in his transmutation theory (Walls 1934, 1942).

The transmutation theory is also applicable to the photoreceptors of geckos (Gekkonidae). Based on the gross morphology of the eye, Walls advocated that nocturnal geckos evolved from diurnal lizards that utilized a pure-cone retina, followed by the reversion to diurnality. However, on closer inspection, photoreceptor outer segments of some nocturnal geckos, although rod-like in their size and general morphology, exhibit ultrastructural characteristics of cones, such as a continuous connection between the disc and the plasma membrane over the entire course of the outer segment, and the presence of oil droplets (Roll 2000). The structural characterization of cones in nocturnal geckos is supported by the expression of opsin genes that belong to three of the four cone classes, LWS, RH2 and SWS1 (Kojima et al. 1992; Yokoyama & Blow 2001), retained in diurnal geckos (Ellingson et al. 1995; Taniguchi et al. 1999).

The retinae of snakes range from rod-dominated to cone-only. The more ‘basal’ henophidian snakes (e.g. pythons and boas) possess a rod-dominated retina with two classes of single cones that are photosensitive with spectral peaks at around 550 and 360 nm (Sillman et al. 1999, 2001). Boas and pythons hunt for prey under mesopic and nocturnal conditions and yet retain the potential for dichromatic colour vision. The identity of the pigments that are expressed in these two cone classes has come from a study of the sunbeam snake, Xenopeltis unicolor, and the python, Python regius (Davies et al. 2009c). In both species, distinct UVS SWS1 (λmax = 361 nm) and LWS (λmax = 550 nm) pigments were found, in addition to a rod (RH1) pigment with a spectral absorbance peak at 497 nm (Fig. 9). By contrast, some more ‘advanced’ colubrid snakes (e.g. the diurnal garter snake, Thamnophis sirtalis) have pure-cone retinae with both double and single cones, the latter of which comprise separate populations of large and small photoreceptors (Jacobs et al. 1992; Sillman et al. 1997). The double and large single cones contain the same LWS pigment with a peak at about 554 nm, and the small single cones have λmax values at 482 nm or 360 nm. The molecular identity of these pigments remains undetermined, but if the three retinal photoreceptors of colubrid snakes share a similar opsin complement to the henophidian species studied so far, then a rod (RH1) pigment gene may be expressed in a cone (Davies et al. 2009c).

image

Figure 9.  A diagram illustrating the three main extant branches of the mammalian lineage (monotremes, marsupials and eutherians) that derive from a reptilian ancestor, superimposed with the evolution of the visual photopigment genes. Approximate dates of speciation are shown in million years.

Download figure to PowerPoint

Recently, microspectrophotometry, light microscopy and scanning electron microscopy have been used to characterize the retinal photoreceptors of the spine-bellied (Lapemis curtus) and horned (Acalyptophis peronii) sea snakes (Hart et al. 2012). Both species possess three types of single cone-like photoreceptors sensitive to short-, middle- or long-wavelengths, with λmax values at 428–430 nm, 496 nm or 555–559 nm, respectively. They also possess a double cone–like photoreceptor subtype, where a LWS photopigment may be found in both principal and accessory members. Conventional rods were not observed, although the middle-wavelength-sensitive photoreceptor may be a ‘transmutated’ rod (Hart et al. 2012).

There is relatively little information available regarding the complement of cones present in nocturnal birds. In those species studied, a rod-dominated duplex visual system is found, but cones have been reduced to only 10–20% of the total photoreceptor number (Bowmaker & Martin 1978; Rojas et al. 2004). Cone spectral sensitivities are only available for the tawny owl, Strix aluco. In this species, three classes are present with λmax values at 555, 503 and 463 nm that are most likely expressing the LWS, RH2 and SWS2 opsin genes, respectively (Bowmaker & Martin 1978). The study of additional species will be required to determine whether this complement of cones is common to nocturnal species and whether the absence of an SWS1 pigment is also a consistent feature.

Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Like some squamate reptiles that are thought to have switched from diurnality to nocturnality and back again, it has been proposed that early mammals also experienced a similarly reduced photic environment during their evolution. During this so-called nocturnal-bottleneck (c. 150–200 Ma), it is thought that ancestral mammalian species switched from a diurnal to a more nocturnal regime as a result of archosaur domination during the day (Walls 1942; Young 1981) and, in response, lost all but two cone pigment genes (Yokoyama 2000; Davies 2011).

The mammalian lineage derived from insectivorous or omnivorous reptile-like animals (therapsids) (Kemp 2006) between 267 and 310 Ma (Young 1981; Tudge 2000), with modern mammals first appearing in the fossil record about 225 Ma (Lucas & Lou 1993). Extant mammals are divided into three main branches, namely the monotremes (subclass Prototheria), the marsupials (subclass Metatheria) and the placentals (subclass Eutheria) (Fig. 9). The retinae of most mammals are rod-dominated with up to 3% cones in nocturnal species and 5–30% cones in diurnal species (Peichl 2005). The proportion of cone subtypes is only >50% in a very small number of species, for example, the ground squirrel, Spermophilus beecheyi (with ∼85% cones) (Kryger et al. 1998), and the tree shrew, Tupaia glis (with ∼95% cones) (Muller & Peichl 1989). Consistent with their reptilian ancestry, double cones are found in the mammalian lineage, although they are limited to the monotremes and marsupials, and are not found in eutherian mammals.

In marsupials and eutherians, only the SWS1 and LWS opsin genes remain, with the RH2 and SWS2 pigment genes having been lost after the divergence of the protherian and therian lineages (Fig. 9). In mammals, the SWS1 photopigments exhibit spectral absorbance peaks that range from UV (∼360 nm) [e.g. in the mouse, Mus musculus (Yokoyama et al. 1998)] to violet wavelengths (∼440 nm) [e.g. in the tree squirrel, Sciurus carolinensis (Carvalho et al. 2006)] (Fig. 6). By contrast, the spectral sensitivities of mammalian LWS pigments are less varied and range from about 530 to 565 nm. A notable exception is the LWS pigment utilized by rodents, such as the rat (Rattus norvegicus) and mouse (M. musculus), and by the rabbit (Oryctolagus cuniculus), where the λmax is short-wavelength-shifted to about 510 nm (Sun et al. 1997; Radlwimmer & Yokoyama 1998) (Fig. 5h and Table 1).

The egg-laying monotremes, the platypus (Ornithorhynchus anatinus) and the echidna (Tachyglossus aculeatus), diverged from the marsupial/placental mammal lineage around 200 Ma (Woodburne et al. 2003) and have preserved both SWS2 and LWS genes but lost the SWS1 gene (Davies et al. 2007a; Wakefield et al. 2008) (Figs 5f and 9, Table 1). In the platypus, the intact pigments have spectral absorbance peaks at 451 nm and 550 nm (Fig. 5f); the monotreme LWS gene that encodes a pigment with its λmax at 550 nm is therefore similar to its marsupial and eutherian orthologues. Similarly, the SWS2 absorbance peak at 452 nm in the platypus is only marginally more long-wavelength-shifted than the violet-sensitive SWS1 pigments of many eutherian mammals (Fig. 5i), even though the evolutionary routes adopted in the generation of VS photopigments in these species are markedly different. The presence of the SWS2 gene in monotremes infers that ancestral mammals prior to the division into the three subclasses must have retained both SWS1 and SWS2 genes, which, when accompanied by the LWS gene, would have provided the foundation for a trichromatic colour visual system (Fig. 9).

Except for primates (discussed below), eutherians possess only two cone classes that express the LWS and SWS1 opsin genes (Figs 5h and 9). Some Australian marsupials, however, appear to possess a third class of cones that express a middle-wavelength-sensitive (MWS) pigment (Arrese et al. 2002, 2005) (Fig. 5g), and may therefore exhibit trichromatic colour vision (Arrese et al. 2006). This is not the case in the South American counterparts, which appear to be dichromatic under bright-light conditions (Palacios et al. 2010). Marsupial trichromacy has been reported from behavioural testing in the fat-tailed dunnart, Sminthopsis crassicaudata (Fig. 5g), the honey possum, Tarsipes rostratus (Arrese et al. 2002), the quokka, Setonix brachyurus, and the quenda or bandicoot, Isoodon obesulus (Arrese et al. 2005). As these species are drawn from three different marsupial orders, it would imply that trichromacy is widespread in marsupials, yet there is no supporting evidence for it in other species (Hunt et al. 2009b; Ebeling et al. 2010). Moreover, all attempts to identify the molecular basis for the MWS cone pigment in the genome of the fat-tailed dunnart (Sminthopsis crassicaudata) have failed (Strachan et al. 2004; Cowing et al. 2008), and there is no evidence for it even as a pseudogene in the sequenced genomes of the grey short-tailed opossum, Monodelphis domestica (Hunt et al. 2009b), or the tammar wallaby, Macropus eugenii. A duplication of the rod RH1 opsin gene is present in the genome of the fat-tailed dunnart (Sminthopsis crassicaudata), raising the possibility that the MWS cones are expressing a rod (RH1) pigment gene (Cowing et al. 2008) (Fig. 9), although opsin antibody labelling of cones in the fat-tailed dunnart has failed to confirm this (Ebeling et al. 2010).

Although many mammals display nocturnal behaviour, cones are frequently retained in an otherwise rod-dominant retina (as discussed above). In fully nocturnal species, activity occurs at light levels that are largely below the sensitivity of cones; in such species, a total loss of colour vision may occur, although in some cases, short-wavelength-sensitive (S) cones may be preferentially lost, as found in two nocturnal carnivores, the kinkajou (Potos flavus) and the raccoon (Procyon lotor). These species do, however, retain LWS (L) cones and, as such, are described as L cone monochromats (Jacobs & Deegan 1992). A similar phenomenon is seen in the squirrel family of rodents; most are strictly diurnal with retinae that are rich in cones (e.g. the grey squirrel, Sciurus carolinensis) (Long & Fisher 1983; Blakeslee et al. 1988; Kryger et al. 1998), but nocturnal flying squirrels (e.g. the Siberian flying squirrel, Pteromys volans, and the Northern flying squirrel, Glaucomys sabrinus) have also dispensed with a functional SWS1 gene, although a pseudogene with several deletions is retained in the genome (Carvalho et al. 2006) (Fig. 6). This loss of colour vision in flying squirrels would appear to be associated with the change from diurnality in their immediate tree squirrel ancestor to nocturnality (Carvalho et al. 2006). Nonetheless, this does not seem to be a universal mechanism, as many other rodents exhibit nocturnal behaviour, but still possess two intact cone opsins. Examples include the rat (R. norvegicus), mouse (M. musculus), Mongolian gerbil (Meriones unguiculatus), cururo (Spalacopus cyanus) (Jacobs & Tootell 1979; Govardovskii et al. 1992; Jacobs & Deegan 1992; Deegan & Jacobs 1993; Chavez et al. 2003; Peichl et al. 2005; Williams et al. 2005), and the pocket gopher (Thomomys bottae). Curiously, Syrian hamsters show both patterns with the Siberian dwarf hamster, Phodopus sungorus, possessing cones that express both SWS1 and LWS pigments, whereas in the Syrian golden hamster, Mesocricetus auratus, only L cones are present (Calderone & Jacobs 1999).

Bats fall into two major groups, the Megachiroptera and Microchiroptera. The former are crepuscular (i.e. active at twilight) and possess a well-developed visual system, whereas the latter rely more on acoustic orientation (echolocation) than vision as they are nocturnal. Most megabats have retained dichromacy with two cone classes expressing LWS and SWS1 pigments (Wang et al. 2004), although a few species have lost S cones and appear to be L cone monochromats (Muller et al. 2007). Surprisingly, nocturnal microbats (Wang et al. 2004) also express both LWS and SWS1 pigments.

Amongst the primates, the platyrrhine owl monkey, Aotus trivirgatus, is typically described as nocturnal (Levenson et al. 2007) with no S cones (Wikler & Rakic 1990) and a pseudogene with deleterious mutations in the SWS1 gene (Jacobs et al. 1996b). Molecular studies have demonstrated that similar pseudogenes are also found in several prosimian species, notably in all Lorisiformes studied so far, and in some members of the Cheirogaleidae, the latter of which belongs to the Lemuriformes (Kawamura & Kubotera 2004; Tan et al. 2005) (Figs 6 and 10). This group includes the greater dwarf lemur (Cheirogaleus major) and the fat-tailed dwarf lemur (Cheirogaleus adipicaudatus), but their related cousins, Coquerel’s mouse lemur (Mirza coquereli) and the grey lemur (Microcebus murinus), still maintain an intact SWS1 gene. Surprisingly, the strictly nocturnal aye-aye (Daubentonia madagascariensis) has retained a functional SWS1 opsin gene, displaying purifying or stabilizing selection (Perry et al. 2007), which indicates that dichromatic colour vision continues to be important for the aye-aye despite its nocturnal activity.

image

Figure 10.  The phylogeny of different primate species showing variation in the amino acid found at site 86 in SWS1 opsins, together with the corresponding codon sequences. The effects on nucleotide substitution of the alternative hypotheses of ancestral (1) Phe86 or (2) Tyr86 are shown (Carvalho et al. 2012). Substitutions at codon 86 are shown in lower case, and a dash represents any nucleotides that are not changed at this site. Solid lines (green) are lineages with functional SWS1 pigments, and dotted lines (red) indicate species where SWS1 pseudogenes are present. Branch lengths do not reflect evolutionary distances. The tree was modified from Perelman et al. (2011) and Carvalho et al. (2012).

Download figure to PowerPoint

Reversion to a diurnal lifestyle—a primate perspective

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Primates comprise three major groups, the Old World or catarrhine primates from Africa and Asia, the New World or platyrrhine monkeys from Central and South America and the prosimians. The latter group consists of five families of lemur (confined to Madagascar) and loris that are endemic to Asia and Africa, and a single family of tarsiers found exclusively in Indonesia and the Philippines (Tudge 2000). The potential for trichromacy is commonly found across all three groups although not in all species of a particular group (Nathans et al. 1986b; Bowmaker et al. 1991; Ibbotson et al. 1992; Dulai et al. 1999). The selective advantages conferred by trichromacy are unclear, and many controversial ideas have been proposed: the first suggested that improved red/green colour discrimination would allow for the detection and evaluation of ripe fruits (Mollon 1989; Osorio & Vorobyev 1996; Sumner & Mollon 2000a,b; Regan et al. 2001) and newly emerging succulent leaves (Dominy & Lucas 2001), although recent studies have suggested that achromatic contrast and odours are likely to be predominantly important (Hiramatsu et al. 2008, 2009) or that dichromacy may be more advantageous than trichromacy (Morgan et al. 1992; Saito et al. 2005; Melin et al. 2007). The situation is obviously very complex and may be based on finding a balance between the intrinsic properties of the colour visual system and the task undertaken, resulting in highly polymorphic vision in different primate populations (Hiwatashi et al. 2010). The molecular mechanism that mediates this trichromacy differs, however, between the platyrrhines and catarrhines, with prosimians showing similarity to the former group (Table 2). In Old World monkeys and the great apes, trichromacy is achieved by the presence of an autosomal SWS1 gene and a duplication of the X-linked LWS gene, with the duplicated copies of the LWS gene encoding spectrally distinct pigments that peak either around 535 nm (M pigment) or around 560 nm (L pigment) (Table 2). The duplication of the LWS gene must have arisen early in the evolution of the catarrhine lineage (Nathans et al. 1986a,b) as it is found throughout the catarrhines with an identical opsin gene array on the X chromosome with an L gene that is located upstream to one or more M genes (Nathans et al. 1986a,b; Drummond-Borg et al. 1989; Feil et al. 1990).

Table 2.   The allelic variation found in the LWS/MWS (L/M cone opsin) genes of New World monkeys and prosimians
FamilyGenusCommon nameNumber of M and L genesVariants per geneλmax (nm)
Platyrrhini
 Atelidae Alouatta Howler monkey21530, 558
Ateles Spider monkey12538–551, 553–562
Lagothrix Woolly monkey12548, 563
 Pitheciidae Callicebus Titi monkey15530, 536, 542, 551, 562
Pithecia Saki monkey13535, 550, 562–565
 Cebidae Cebus Capuchin monkey13532–536, 543–549, 561–563
Samiri Squirrel monkey13532–537, 545–550, 561–565
Aoutus Owl monkey11539–545
Leontopithecus Cotton-top tamarin13546, 557, 563
Leontopithecus Saddle back tamarin13543, 555, 564
Callithrix Marmoset13539–545, 553–559, 561–567
Prosimians
 Lemuridae Lemur Ring-tailed lemur11547
Eulemur Sclater’s lemur12543, 558
Hapalemur Lesser bamboo lemur11558
Varecia Red-ruffed lemur12543, 558
Varecia Black and white-ruffed lemur12545, 558
 Cheirogaleidae Cheirogaleus Greater dwarf lemur11543
 Galagidae Otolemur Greater galago11539
 Indriidae Propithecus Coquerel’s sifaka12543–545, 558

By contrast, trichromacy in platyrrhines and prosimians is achieved by a single polymorphic X-linked LWS gene paired with an autosomal SWS1 gene (Table 2). The different allelic copies of the LWS gene that are found in most species of platyrrhines (Neitz et al. 1991; Williams et al. 1992) and some species of prosimians specify pigments with λmax values of 535–565 nm. Only females with two X chromosomes can possess two copies of this gene, so trichromacy is restricted to heterozygous females that are in possession of different allelic forms of the LWS gene. Dichromacy is therefore found in all males and homozygous females (Mollon et al. 1984). Polymorphic LWS genes have also been found in three diurnal (Jacobs et al. 2002) and a single cathemeral species of prosimian (Veilleux & Bolnick 2009) (Table 2).

Interestingly, one platyrrhine species, the howler monkey, Alouatta spp. (Jacobs et al. 1996a; Dulai et al. 1999), follows the Old World model with a LWS gene duplication that encodes pigments with λmax values of 530 and 558 nm. As in Old World primates, both sexes of howler monkey benefit from full trichromacy, although the duplication event is quite distinct and more recent than that found in the catarrhine lineage.

The spectral differences between primate L and M pigments are largely due to modifications at only three sites (Neitz et al. 1991) at positions 180, 277 and 285 in the opsin molecule (Table 1; numbering based on the human L opsin amino acid sequence). In all cases, Ser180, Tyr277 and Thr285 (all polar) are present in the L opsin and Ala180, Phe277 and Ala285 (all nonpolar) in the M opsin, respectively (Asenjo et al. 1994). New World primates from the family Cebidae (e.g. capuchins) utilize the same amino acid substitutions at sites 180, 277 and 285 to ‘spectrally tune’ their allelic variant LWS pigments (Neitz et al. 1991; Williams et al. 1992), which differ by up to 27 nm. By contrast, site 277 is invariant in the family Callitrichidae (e.g. marmosets), so the spectral shift between L and M pigments is limited to only 19–20 nm. For many New World primates, two or more different variants of the LWS gene are present within the population, thereby leading to a range of spectral sensitivity phenotypes amongst different individuals. The L and M pigments of the howler monkey again show replacements in the residues present at these same three sites (Jacobs 1996; Jacobs et al. 1996a) and, as such, generate pigments with λmax values at 530 and 558 nm.

In those prosimians with a polymorphic LWS gene (Tan & Li 1999), the different alleles vary at only site 285, with Thr in the L opsin and Ala in the M opsin sequence. The λmax values for these pigments were predicted to be at 543 and 558 nm (Tan & Li 1999) and subsequently confirmed by electrophysiological analysis (Jacobs et al. 2002). Where only a single pigment is found, the majority of species have the M opsin (Ala285) rather than the L opsin (Thr285) subtype, indicating that trichromacy may have existed in the common ancestor but only retained by a subset of present-day species.

Amongst nonprimate mammals, the peak absorbance values for LWS pigments are relatively conserved. These photopigments are generally anion-sensitive (Kleinschmidt & Harosi 1992), a biochemical property that appears to be unique to LWS pigments and results from the formation of an active binding site for chloride ions between His197 and Lys200 (Wang et al. 1993). An exception is found in the mouse, Mus musculus, where the λmax is short-wavelength-shifted by around 28 to ∼510 nm due to the loss of the chloride-binding site (Sun et al. 1997; Davies et al. 2012b) (Table 1). A similar mechanistic shift is also observed in other small mammals, such as the rat (Rattus norvegicus) and rabbit (Oryctolagus cuniculus), where the λmax of the LWS pigment peaks at 509 nm (Radlwimmer & Yokoyama 1998). Replacement of His197 with Tyr in the mouse pigment abolishes the binding of chloride ions and the corresponding spectral shift to longer wavelengths (Sun et al. 1997; Davies et al. 2012b). The bottlenose dolphin (Tursiops truncates), an L cone monochromat, has a LWS pigment with a λmax that also shows a short-wavelength shift, and site-directed mutagenesis has shown that the key change is the presence of Ser at site 308 (Fasick & Robsinson 1998; Fasick et al. 1998). Amongst the mammals, Ser308 is only found in cetaceans (Newman & Robinson 2005), but it is also found in the LWS pigment of the elephant shark, Callorhinchus milii (Davies et al. 2009a), where its role in generating a short-wavelength shift via the inactivation of an intact chloride-binding site has been demonstrated by site-directed mutagenesis and the in vitro expression of mutant pigments (Davies et al. 2009a).

The loss of UV sensitivity has already been described in birds (Fig. 8), but a similar loss was also a major evolutionary change in the development of the visual system in mammals. As found in the Aves, the basis of this loss is a switch from SWS1 pigments that peak in the UV to ones that peak in the violet region of the spectrum (Hunt et al. 2007) (Fig. 6). This shift cannot be fully accounted for by the attenuation of the intensity and spectral composition of light as mammals passed through the ‘nocturnal-bottleneck’ (Walls 1942), because ancestral UVS pigments are widespread in marsupials and to a lesser extent in eutherian mammals (Hunt et al. 2007) (see discussion below). Thus, it is possible that the selective pressures for this change may be associated with the protection of the retina from UV light damage, or the enhancement in image quality or both. The amino acid substitution responsible for these long-wavelength shifts is the replacement of Phe at site 86 (based on bovine rod opsin numbering) in UVS pigments by either Tyr, Ser or Val in mammalian VS pigments (Cowing et al. 2002b; Fasick et al. 2002; Parry et al. 2004; Yokoyama et al. 2005). This single change is found in both placental and marsupial lineages and therefore represents several examples of convergent evolution (Hunt et al. 2007) (Fig. 6).

For the tuning of VS pigments in humans, it has been proposed that substitution of residues at seven sites was responsible for the shift from UV to violet sensitivity (Yokoyama & Shi 2000). However, a more recent study that included a number of prosimian species has concluded that residues at only two positions are important (Carvalho et al. 2012), namely site 86, where the replacement of Phe is also responsible for the shift to VS in birds (Carvalho et al. 2007) and nonprimate mammals (Hunt et al. 2007), and site 93, which is occupied by Thr in UVS pigments (Carvalho et al. 2012). In all primate VS pigments, Thr93 is replaced by Pro, but site 86 may be occupied by one of several different residues that include Tyr, Leu, Val, Cys and Ser, and in one case, the aye-aye (Daubentonia madagascariensis), a highly endangered primate endemic only to Madagascar, by Phe. An examination of the phylogeny of these changes (Fig. 10) indicates that the initial shift to violet sensitivity at the base of the primate lineage was achieved either by a Thr93Pro substitution, which consequently allowed other changes at site 86 to occur, or by a Phe86Tyr substitution, with the role of Tyr86 in maintaining spectral sensitivity in the violet range being subsequently superseded by a Thr93Pro change. With this latter scenario, the presence of Phe86 in the aye-aye would be the result of a back mutation (Carvalho et al. 2012).

Fossorial adaptation of the vertebrate visual system

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

In species that have adopted a fossorial or subterranean habitat, cone loss may again be apparent. In the subterranean blind mole rat (Spalax ehrenbergi), a functional SWS1 gene has been lost with only L cones and rods still present (David-Gray et al. 1998, 2002; Janssen et al. 2000). Nonetheless, the blind mole rat eye is atrophied and the animal lacks any ability to react to visual stimuli (Cooper et al. 1993a,b). The loss of S cones is not, however, a consistent finding in subterranean rodents, with S cone retention (along with L cones) in the retinae of the bathyergid mole rats, Cryptomys anseelli, Cryptomys mechowi and Heterocephalus glaber, where S cones are substantially greater in number than L cones (Nemec et al. 2004; Peichl et al. 2004); in the European mole, Talpa europaea, a small-eyed subterranean species; and in the Chilean curoro, Spalacopus cyanus, where up to 20% of the cones in the ventral retina are S cones (Peichl 2005; Peichl et al. 2005). It is of particular note that these adaptations are not restricted to small mammals, as atrophied eyes are also a feature of fossorial caecilian amphibians, but in the latter case, no cone photoreceptors are present (Mohun et al. 2010) (Fig. 5c).

The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

There is overwhelming evidence to link the complement and spectral tuning of photoreceptors that mediate colour vision, and the photopigments that they express, to the distinct ecological niches occupied by each species. Building upon the ‘transmutation’ theory to explain photoreceptor differences in diurnal and nocturnal squamates (Walls 1934), Walls proposed that a similar correlation may exist to explain the morphological transition of photoreceptor cells found in squamate reptiles and birds to those of the mammalian lineage (Walls 1942). Such cone photoreceptor adaptations that include a loss of coloured oil droplets, an increase in size, a shift to a more rod-like morphology (Crescitelli 1972; Roll 2000) and gross changes in the eye (e.g. a larger aperture, a shorter focal length, and the development of a light-reflective tapetum) suggest that early mammals may have evolved from mammal-like reptiles that exhibited nocturnal behaviour (Walls 1942). Since its proposal, this suggestion has become known as the ‘nocturnal-bottleneck’ hypothesis (Heesy & Hall 2010). During this evolutionary phase, it is thought that mammals shifted from day to night to avoid predation from co-existing archosaurs (Walls 1942; Young 1981), a transition that was probably underpinned by the evolution of a complex thermoregulatory system (endothermy) in mammals and, thereby, permitted activity in the colder temperatures experienced at night (Grigg et al. 2004).

Throughout the circadian (daily) cycle, the intensity of solar radiation reaching the Earth changes dramatically from bright sunlight (∼1 × 105 lux at noon when the sun is at its zenith) to starlight (∼1 × 10−5 lux) (Lythgoe 1979; Turner & Mainster 2008), a staggering 1 × 1010-fold difference in illumination (Stockman & Sharpe 2006) (Fig. 11a). The craniate eye is limited in the spectral range of environmental light it can detect and generally perceives a broad range of wavelengths (300–700 nm) during the day (Lythgoe 1979). Although data are limited, it is possible to interpolate from human-based studies that both the intensity and the spectral composition of environmental light are restricted during the night, with a reduction at both the short- (i.e. UV) and long-wavelength (i.e. perceived as ‘red’) ends of the light spectrum, but enriched at middle-wavelengths (i.e. perceived as blue-green light) (Thorne et al. 2009).

image

Figure 11.  (a) An illustration showing the relative light levels (denoted as a logarithmic scale) in natural environments throughout the day (values taken from Turner & Mainster (2008)), highlighting the mesopic phase, where both cone and rod photoreceptors are active and confer colour information in humans. (b) Spectral sensitivity of photopic (λmax = 555 nm), mesopic (an estimate of the λmax at 528 nm is given, but the actual peak is difficult to determine and will lie within the range flanked by bright- and dim-light vision) (green shading), scotopic (λmax = 506 nm) and circadian (λmax = 480 nm) photoreception in humans (Rea et al. 2004; Turner & Mainster 2008), compared with the range of λmax values for photopigments encoded by LWS (red), SWS1 (violet), SWS2 (blue), RH2 (green) and RH1 (black) opsin gene classes.

Download figure to PowerPoint

Indirect evidence for the effect of environmental spectral differences during the transition from day into night and vice versa stems from studies of circadian photoentrainment and photoperiodism. It is well established that the melanopsin photopigment (OPN4) is the critical light sensor that measures the photic input for synchronizing many internal physiological rhythms (e.g. sleep, circadian clock, melatonin production) to the external environmental lighting conditions (Rollag et al. 2003; Bailes & Lucas 2010; Davies et al. 2010; Hatori & Panda 2010; Provencio 2011). As found in both mammals and nonmammals, this physiological system is maximally active at dawn and dusk, with an action spectrum that peaks at ∼480 nm (Lucas et al. 2001; Berson et al. 2002; Hankins & Lucas 2002; Hattar et al. 2003; Dacey et al. 2005; Davies et al. 2010), an adaptation that maximizes photon capture at the dominant shorter wavelengths that are enriched when the sun is close to the horizon at twilight (Davies et al. 2010). Therefore, it is not surprising that the molecular light detector, melanopsin, is also spectrally tuned to similar wavelengths for many chordate species [e.g. teleosts (Davies et al. 2011), birds (Torii et al. 2007) and rodents (Newman et al. 2003; Walker et al. 2008)]. Similar dawn and dusk oscillators regulate photoperiodism and seasonal breeding in birds (Foster 1998; Foster & Kreitzman 2009), and both the photopigment involved, vertebrate ancient (VA) opsin in this case, and the physiological response are maximally ‘tuned’ to 490 nm (Foster & Follett 1985; Foster et al. 1985; Davies et al. 2012c). Taken together, these studies support the conclusion that the spectral composition of environmental light at dawn and dusk through the enrichment of short-wavelengths, can have significant effects on the evolution of photopigments and the physiological systems they regulate.

Cones are sensitive to bright-light environments [e.g. intensities from ∼0.1 lux to ∼1 × 105 lux in humans (Turner & Mainster 2008)], whereas rods have evolved to function under dim-light conditions [e.g. in humans, they are sensitive to ∼1 × 10−6 lux to 50 lux (Turner & Mainster 2008)], and as such, cone function predominates during daylight (photopic vision), with rods being constantly bleached due to the high intensity of illumination. Conversely, rods function alone at night (scotopic vision) as the amount of illumination is below the threshold for cone activation (Fig. 11a). Therefore, it is not unexpected to find that the absolute sensitivities and peak spectral absorbances for photopic (e.g. λmax = 555 nm in humans) vs. scotopic (e.g. λmax = 506 nm in humans) photoreception, and the accompanying Purkinje shift (Purkinje 1823; Barlow 1957), correlate strongly with the spectral composition and intensity of light during the day and night (Turner & Mainster 2008) (Fig. 11b). The transition of day into night is not immediate, but instead follows a gradual diminution in light intensity, such that under the mesopic conditions experienced at dawn and dusk (10–100 lux) (Turner & Mainster 2008), both cones and rods may be active (Reitner et al. 1991; Stabell & Stabell 1994; Buck 1997) (Fig. 11a). In humans, a mesopic luminous efficiency function with a λmax at 528 nm has been calculated (Rea et al. 2004) (Fig. 11b). It is, however, experimentally difficult to accurately measure the λmax of sensitivity in response to mesopic stimuli (Stockman & Sharpe 2006), and as a consequence, the spectral maximum will be expected to lie between the sensitivity ranges of both cone- and rod-based vision, albeit rapidly coming closer to the scotopic peak with decreasing levels of illumination due to the nonlinearity behaviour of the Purkinje shift (Stockman & Sharpe 2006).

The spectral difference between day and night, as discussed above, is reminiscent of the restriction of available wavelengths of light observed at increasing depths in oceanic waters (Jerlov 1976; Denton 1990) (Fig. 4). Therefore, it may be predicted that terrestrial species, much like their light-restricted deep-sea fish counterparts, that dwell in dim-light conditions would lose photopigments that have peak spectral sensitivities <400 nm or >550 nm, with completely nocturnal species possessing rod-only retinae that are sensitive to ∼480 to 500 nm and as such, are colour-blind. Diurnal, land-based, nonmammalian vertebrates possess up to four cone photopigments (tetrachromatic) (Fig. 5d, e), whereas in many mammals these have been reduced to only two photopigments (dichromacy), as is found in all three branches of the mammalian lineage (Yokoyama 2000; Bowmaker & Hunt 2006; Bowmaker 2008; Davies 2011). One of these photopigments has been secondarily lost to leave a single class of opsin (monochromacy) in most marine mammals (Fasick & Robinson 2000; Newman & Robinson 2005) and a number of nocturnal species (Hunt et al. 2009a), or a gene duplication event that has given rise to three opsin types in primates (Jacobs et al. 1996a; Hunt et al. 1998; Dulai et al. 1999) and possibly in some Australian marsupials (Arrese et al. 2002, 2005) (Fig. 9).

If the ‘nocturnal-bottleneck’ hypothesis (Walls 1942; Heesy & Hall 2010) is correct, it would follow that present-day mammals would largely possess rod-only retinae, or if cones were present, then they would be sensitive to wavelengths that peak around 480–500 nm (perceived as blue-green). Such a situation would appear to apply to the process of photoentrainment, where the pigment melanopsin is most sensitive during the transitional periods between day and night and is spectrally tuned to a λmax at ∼480 nm (Hankins et al. 2008; Davies et al. 2010). However, the mammalian radiation evolved from ancestral species that were at least trichromatic (Davies et al. 2007a), with the majority of present-day mammals possessing typically two cone classes with spectral maxima at opposite ends of the light spectrum (Yokoyama 2000; Bowmaker & Hunt 2006; Bowmaker 2008; Davies 2011). In marsupials and eutherian mammals, these cone photopigments are encoded by the SWS1 and LWS opsin genes (Yokoyama 2000; Bowmaker & Hunt 2006; Bowmaker 2008; Davies 2011), whereas in the monotremes, the SWS1 gene has been lost in favour of the SWS2 gene (Davies et al. 2007a; Wakefield et al. 2008). Parsimonious comparison of the peak spectral sensitivities of photopigments from all three branches of the mammalian lineage suggests that early mammals expressed three cone opsins (Davies et al. 2007a), SWS1, SWS2 and LWS, that encode photopigments with spectral maxima at 360, 440 and 560 nm, respectively (Fig. 12). Subsequently, the middle-wavelength-sensitive SWS2 pigment was lost in the therians, resulting in a dichromatic visual system that was maximally sensitive at the outer limits of the visible spectrum.

image

Figure 12.  A schematic of the ‘mesopic-’ or ‘twilight-bottleneck’ hypothesis as described in text and figures. Absorbance spectra (Govardovskii et al. 2000) for putative visual photopigments inferred to be present at key stages in the evolution of mammals are shown. Peak spectral sensitivity values are included for both cone (solid line) and rod (dotted line) photopigments. Spectra are colour-coded to illustrate the molecular basis for each photopigment as follows: LWS (red), SWS1 (violet), SWS2 (blue), RH2 (green) and RH1 (black).

Download figure to PowerPoint

Photoreceptors can only respond to the number of photons absorbed and cannot distinguish between the wavelength and intensity of the incoming photic stimulus (univariance); therefore, at least two ‘opposing’ photoreceptors are required for wavelength discrimination (colour opponency). In a dichromatic system, photopigments that have peak spectral sensitivities that are far apart will possess a spectral overlap that is minimal or absent altogether. Thus, a photostimulus at a particular wavelength is more likely to activate the photoreceptor that maximally absorbs at that wavelength only and colour information is not perceived. In such a situation, it may be predicted that the selection pressure to maintain the function of both cones would be relaxed, leading to the acquisition of deleterious mutations and the eventual loss of opsin genes from the genome. Under extremely bright-light (unphysiological) conditions, it is possible that both photoreceptors would be activated, even with a small spectral overlap. However, to discriminate the intensity and wavelength of a stimulus within the ‘normal’ range of light, and especially under dim-light conditions, a pair of photoreceptors would need to share a relatively high degree of overlap in their spectral profiles for colour perception to be achieved.

Given the scenario of dichromacy with pigments that are maximally sensitive to the extremes of the visible spectrum, as discussed above, it would follow that the potential for wavelength discrimination that is essential for colour vision would be severely limited or nonexistent under the reduced photic environment of continuous darkness (i.e. night), thus predicting that most if not all cone opsin genes would be lost. Clearly, the transition through a ‘nocturnal-bottleneck’ did not have this effect, which raises the question as to what selection mechanisms were at play in early mammals to explain the evolution of their colour visual systems as derived from diurnal members of Reptilia.

An important clue to answering this paradox may lie in what is known about snake colour vision. Like many mammals, it has been shown that snakes possess duplex image-forming retinae and are dichromatic with the presence of two cone photopigments that spectrally peak at 360 and ∼555 nm, respectively (Sillman et al. 1999, 2001). Molecular analysis of the opsins expressed in these serpentine species has revealed that only SWS1, LWS and RH1 genes are present (Davies et al. 2009c). Despite the apparent restriction in colour vision due to the presence of two cone photopigments with a limited degree of spectral overlap, many snakes are active under mesopic conditions (Conant & Collins 1998). Colour vision would be substantially enhanced if both cone and rod photoreceptors were functional under such conditions (Reitner et al. 1991; Stabell & Stabell 1994; Buck 1997); these species may therefore be conditional trichromats (Davies et al. 2009c). As this complement of opsin classes present in snakes is shared with members of the mammalian lineage, it is possible that the evolutionary origins of colour vision in the ancestor to all mammals were shaped by a similar transition from a diurnal lifestyle to one that was mesopic, and not nocturnal, in nature (Fig. 12).

The inherent plasticity of the visual system is well known, and the interchange between different classes of cone and rod photopigments does not appear to be detrimental to photoreceptor function. Examples include the expression of intermediary rod-like (RHA) pigments in cone photoreceptors of the pouched (Geotria australis) and sea (Petromyzon marinus) lampreys (Collin & Trezise 2004; Davies et al. 2007b, 2009b), and rod (RH1) photopigments in the green anole (Anolis carolinensis) (Kawamura & Yokoyama 1997) and the garter snake (Davies et al. 2009c); the expression of the SWS2 gene in the ‘green’ rods of the bullfrog (Rana catesbeiana) (Hisatomi et al. 1998, 1999); and the development of trichromacy in a genetically engineered mouse that is usually dichromatic (Smallwood et al. 2003; Jacobs et al. 2007). These cases suggest that the neural pathways that mediate colour vision are likely to function whether the input is derived from cones, rods or a combination, as long as at least two spectrally distinct, functioning photoreceptors remain for wavelength discrimination (Lythgoe & Partridge 1989).

Taken together, a ‘mesopic-bottleneck’ or ‘twilight-bottleneck’ model may be proposed (Fig. 12) that essentially consists of two components: (i) the effect of spectral changes in environmental light on cone pigments during the transition from day to night and (ii) the dual activity of rod photoreceptors under mesopic conditions as a luminance detector and a middle-wavelength colour channel. How both parts may be unified into a single driving factor to explain the adaptations observed in the visual photopigments of mammals is discussed below. There is a substantial body of evidence to support the postulate that the first mammal-like reptiles possessed orthologues of all four cone opsin gene classes (SWS1, SWS2, RH2 and LWS), in addition to a rod (RH1) opsin gene (Yokoyama 2000; Davies 2011), which would collectively be sensitive to a broad range of wavelengths (with λmax values from 360 to 560 nm for photopigments utilizing a vitamin A1-based chromophore). We propose that as these mammalian ancestors gradually extended their behaviour towards dusk (and dawn) phases of the day to synchronize feeding with the maximal activity of their prey (e.g. insects, annelids, and other small vertebrates), as well as avoid predation from generally larger, diurnal archosaurs, their visual system would be subjected to dim-light, mesopic conditions that would result in the activation of both cone- and rod-based vision. Under such mesopic conditions, the spectral composition of the environmental light would be dominated by short- and middle-wavelengths (i.e. perceived as blue-green) between 480 and 500 nm (see above for discussion).

As rods are more sensitive than cones and their peak spectral sensitivity is close to that of the RH2 pigment (with λmax values that range between 467 and 535 nm, therefore giving an average λmax at 501 nm) (Yokoyama & Tada 2010), we propose that the RH1 photopigment (with a λmax close to 500 nm) may offer a selective advantage under low light levels. Under prolonged periods of sleep/wake phases dictated by mesopic circadian triggers, the RH2 gene may have become redundant, incurred loss-of-function mutations and was eventually lost from the mammalian genome. Due to the enrichment of wavelengths towards the shorter wavelengths (i.e. perceived as blue) of the visible spectrum (i.e. ∼480 nm) under such light conditions, and the redundancy inherent with overlapping spectra (i.e. RH1 vs. RH2), the SWS2 photopigment (with λmax values that range between 440 and 470 nm) may have been initially maintained. Such a scenario is reflected in the monotremes (Davies et al. 2007a; Wakefield et al. 2008), although the subsequent loss of the SWS1 gene and the long-wavelength shift in the spectral tuning of the SWS pigment in these species (Fig. 5f) are likely to be a recent adaptation to a semi-fossorial lifestyle (Shimer 1903).

The increased sensitivity and spectral tuning of the rod pigment in mesopic conditions may have offered an additional selective advantage over the SWS2 photopigment, resulting in its subsequent and eventual loss. Such a scenario is reflected in many marsupials and eutherian mammals, where the loss of SWS2 and RH2 genes results in dichromacy (or conditional trichromacy under mesopic conditions) based on the retained expression of SWS1 and LWS photopigments that are sensitive at the UV or violet and long-wavelength extremes of the light spectrum. As the therian radiation unfolded, new species frequently infiltrated unique ecological niches, where their opsin genes were ‘fine-tuned’; examples include the shift of primate SWS1 photopigments from UV sensitivity to violet sensitivity (Carvalho et al. 2012), the loss of SWS1 in a number of nocturnal rodents (Carvalho et al. 2006) and marine mammals (Fasick & Robinson 2000; Newman & Robinson 2005) and the duplication event that generated primate L and M opsins (Jacobs et al. 1996a; Hunt et al. 1998; Dulai et al. 1999).

The potential contribution of rod photoreceptors to wavelength discrimination in ancestral mammals has wide-ranging implications, not only for the evolutionary origins of mammalian colour vision, but also for present-day species. Despite the diurnal activity pattern of many species that indicates a continuing selection pressure to maintain both functional SWS1 and LWS genes, some mammalian species appear to be conditional trichromats [e.g. the cat (Daw & Pearlman 1969, 1970) and the mouse opossum, Thylamys elegans (Palacios et al. 2010)] or dichromats [e.g. the owl monkey, Aotus trivirgatus (Jacobs et al. 1993), and many marine mammals (Griebel & Peichl 2003)] under mesopic light conditions, at least during specific crepuscular phases of the circadian light cycle. Where studied, the involvement of rods to spectral sensitivity has been shown by electrophysiological means; however, behavioural or psychological testing will be required to determine whether rods convey spectral information, in addition to a measure of luminance alone. Such carefully executed experiments would allow middle-wavelengths to be discriminated from either longer or shorter wavelengths under both mesopic conditions, where both cones and rods would be active, compared to bright-light scenarios, where all the rods would be bleached and refractory. Interestingly, a number of Australian marsupials would appear to be intrinsically trichromatic (Arrese et al. 2002, 2005), yet possess only two cone pigments (Cowing et al. 2008). Most species will be exposed to mesopic light conditions between their circadian behavioural phases, especially those exhibiting an arrhythmic lifestyle (e.g. humans or horses). Therefore, if the ‘mesopic/twilight-bottleneck’ hypothesis is confirmed, ‘conditional multichromacy’ might be more widespread than presently appreciated, where species with monochromatic vision that are thought to be colour-blind or those that possess some colour vision at very low levels of light (e.g. in some insects, such as moths and bees) may, in fact, detect and perceive colour information.

Concluding remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

Despite the overwhelming evidence to support Darwin’s revolutionary ideas on the origins of life and subsequent change by natural selection, many critics use the development of the eye, as a representative example of an apparently perfect, yet complex, organ that cannot be explained in terms of subtle modification from one generation to the next. Comparative neuroecologists, however, are now able to re-evaluate the visual system at the molecular level to better understand the interplay between vision and behavioural ecology. In this review, these principles have been illustrated in extant craniates active in bright- and dim-light environments, spanning species from the last common ancestor of the agnathans to those of the present day, constituting a period of over 540 Myr. We discuss the changes that have taken place in both photoreceptor and opsin gene complement in species from a diverse range of ecological niches (e.g. shallow water, deep sea, terrestrial and fossorial terrains), where a wide range of disparate light environments may be experienced.

We conclude that cone-based tetrachromacy, accompanied by rod-mediated scotopic vision, first evolved from a complement of visual pigments present in the lampreys and has been maintained at each major node of the vertebrate radiation, except for members of the mammalian lineage (Fig. 13). Furthermore, as species moved into previously unexplored niches for shelter, predator avoidance and improved foraging success, each opsin photopigment was either retained or lost as a consequence of adaptation to new spectral environments. Subsequently, each of the remaining opsins evolved distinct, yet partially overlapping, molecular tuning mechanisms independently in each major vertebrate radiation.

image

Figure 13.  A summary cladogram showing the evolutionary progression of each opsin class, mediated by gene conservation, loss and duplication, within the main branches of the craniate radiation. The presence of five opsin genes (LWS, SWS1, SWS2, RHB/RH2 and RHA/RH1), which first evolved in the last common ancestor of the Agnatha and the Gnathostomata, is conserved at each major node (yellow star), except for the mammals, where gene loss occurred early and throughout the mammalian lineage. A black cross (X) refers to gene loss, while small arrows (↓ or ↘) indicate the presence of opsin gene duplications. A question mark (?) shows that the presence or absence of a particular opsin gene is unknown. The following footnotes are included: #greatly simplified; +most marine mammals (e.g. dolphins) have lost the SWS1 gene to become L cone monochromats, the exception being the manatee where both LWS and SWS1 genes remain intact; *most nonprimate eutherians are dichromatic; however, some nonprimates (e.g. flying squirrels) have lost the SWS1 gene to become L cone monochromats; and ‡primate trichromacy has evolved by different mechanisms (see text). Modified from Davies (2011).

Download figure to PowerPoint

Stemming from this comprehensive account of molecular ecology and adaptation of the craniate visual system, we are now able to make more accurate predictions of both the spectral characteristics of a diverse range of ecological niches and the visual systems that have evolved to operate in these conditions. For mammals, a ‘nocturnal-bottleneck’ that resulted in the marginalization of colour vision in favour of improved dim-light sensitivity in the nocturnal realm has been previously proposed. We postulate, however, that a ‘mesopic-‘ or ‘twilight-bottleneck’ may offer a better explanation for the cellular and molecular changes that occurred in the transition from a reptile-like ancestor to the three main branches of the mammalian lineage observed today.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

The authors wish to thank Jim Bowmaker, Ian Potter, Byrappa Venkatesh, Nathan Hart, Jill Cowing, Sue Wilkie, Livia Carvalho, Helena Bailes and Susan Theiss for assistance in previous research forming part of this review and Mark Hankins for insightful discussion. In addition, the authors are indebted to the four anonymous reviewers for their detailed comments and helpful suggestions. This work was supported by grants from the Biotechnology and Biological Sciences Research Council (BBSRC), UK, and the Australian Research Council (ARC), Australia.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Brightly lit aquatic environments and the origin of the craniate visual system
  5. Adaptations of jawed aquatic vertebrates to broad-spectrum, bright light
  6. Molecular changes in photopigment genes associated with deep-water environments
  7. Transition to land: a spectral challenge?
  8. Mammals return to the deep sea
  9. Diurnality in terrestrial vertebrates
  10. Crepuscular and nocturnal adaptation to dim-light environments
  11. Adaptation of the mammalian visual system through the ‘nocturnal-bottleneck’
  12. Reversion to a diurnal lifestyle—a primate perspective
  13. Fossorial adaptation of the vertebrate visual system
  14. The twilight zone and the evolution of colour vision in mammals: an alternative hypothesis
  15. Concluding remarks
  16. Acknowledgements
  17. References

W.I.L.D. is a molecular biologist with broad interests in functional genomics and the evolution of large gene families, as well as the mechanisms that regulate gene expression. Specifically, he investigates the evolutionary origins, molecular ecology, function and spectral tuning of photopigments that mediate visual and non-visual light detection and phototransduction in chordates. S.P.C. is a comparative neurobiologist interested in the evolution, development and adaptation of the vertebrate visual system with a particular interest in early vertebrates. However, he is also investigating all of the known senses in various species of cartilaginous fishes in order to assess the relative importance of each sensory modality and the selection pressures acting on their survival in different ecological niches. D.M.H. has a long-term interest in the evolution of colour vision and the spectral tuning of visual pigments, with studies on a diverse number of species that range from cephalopods to primates. He is also involved in the study of inherited retinal disease in humans and has identified a number of gene mutations that are responsible for blinding disorders. Current work is this latter field and is focused on understanding disease mechanisms, largely through the study of model systems.