The idea that the spectral sensitivities of visual pigments are adapted to an animal’s light environment is an old one, especially as it applies to fish. Vision underwater is limited by available light and there is a long held expectation in visual ecology that the visual pigments of fish should be attuned to the dominant wavelengths in the habitats in which they live in order to maximize visual sensitivity (Lythgoe 1979). Variation in absorption and scattering, related to water depth and geographical factors associated with the quantity and nature of suspended particulates, gives rise to substantial variation in the spectral composition of the light environments in aquatic habitats (Lythgoe 1979). Many studies have demonstrated associations between environmental light conditions and the spectral sensitivities of visual pigments in fish (e.g. Lythgoe 1984; Partridge et al. 1989; Lythgoe et al. 1994; Jokela-Määttäet al. 2007). Despite the well-established ecological framework for the evolution of spectral sensitivity in fish, support for the role of natural selection has been limited to phylogenetic and correlative evidence. In this issue of Molecular Ecology, Larmuseau et al. (2009a) present a study of local adaptation in the RH1 opsin gene that controls the spectral sensitivity of dim-light vision in the sand goby (Pomatoschistus minutus, Gobiidae; Fig. 1). This represents the first population genetic analysis of intraspecific variation in the RH1 opsin gene in natural populations and provides new evidence of selection acting on spectral sensitivity.
Vertebrates see via a battery of specialized photoreceptors: highly sensitive rod cells that are responsible for vision in dim light, and several classes of cone cell that are responsible for vision in bright light and the perception of colour. The process of photoreception, in both rods and cones, begins with the capture of light by the visual pigment molecules that are packed within their membranes. Visual pigments comprised a light-absorbing chromophore (usually 11-cis retinal) and a transmembrane opsin protein. The spectral sensitivities of rods and cones are largely determined by the narrow absorbance spectrum, characterized by a wavelength of maximum absorbance (λmax), of the visual pigment they contain. The visual pigment in rod cells, commonly known as rhodopsin, has λmax values ranging approximately from 480 to 525 nm in a diverse array of fish species (Yokoyama et al. 2008). Since the gene for rhodopsin, RH1, was first sequenced, comparative studies and in vitro assays of synthetic proteins have identified numerous functional mutations, mostly in the retinal binding pocket, that cause shifts in λmax (Hunt et al. 2001; Yokoyama et al. 2008). Through functional and molecular evolution studies of RH1 genes of fish and other vertebrates, a good understanding of the evolutionary history of dim-light spectral sensitivity has been attained (Yokoyama et al. 2008). While most studies of rhodopsin evolution have compared species and higher taxa, among-population variation in rhodopsin λmax has also been found within several fish species including the sand goby (Jokela et al. 2003; Jokela-Määttäet al. 2007). In the case of the sand goby, Jokela et al. (2003) determined that variation in rhodopsin spectral sensitivity is not associated with chromophore substitution, as it is in some species (Bowmaker 2008), and must therefore be driven by variation in the protein (opsin) component. While such intraspecific covariation of λmax with light environments suggests local adaptation, there has been, until now, no direct evidence of adaptive divergence of RH1 genes within species.
Sand gobies are coastal, bottom-dwelling fish that forage mostly at night, making dim-light vision a critical faculty. Larmuseau et al. (2009a) set out to test the hypothesis that sand goby populations are adapted at the RH1 locus to local light environments. Their approach was to perform multiple tests of neutral molecular evolution and to compare the distributions of nucleotide substitutions with the spectral environments of populations. In addition to microsatellite genotypes for 417 individuals, the authors obtained 868 bp of sequence for this single copy gene from 165 individuals. The seven populations sampled occurred in a range of light environments and represented three broad phylogeographic groups. Several rhodopsin haplotypes were shared among populations and among the historical units defined previously using mitochondrial DNA sequences (Larmuseau et al. 2009b). Larmuseau et al. (2009a) found no correlation between microsatellite-based FST-estimates and those based on single nucleotide polymorphisms (SNPs) in the RH1 gene. Furthermore, the SNPs in RH1 were clear FST-outliers, when compared with the microsatellite markers, confirming that the former are likely to have been influenced by non-neutral evolutionary processes.
Further evidence that natural selection plays a role in the functional evolution of the sand goby RH1 gene came from codon-based models of molecular evolution. Interestingly, four of the five amino acid substitutions in the sample have been shown to produce a shift in λmax in either rhodopsins or cone opsins of other fish (see references in Larmuseau et al. 2009a). The functional significance of these changes in the protein sequence makes this dataset an excellent candidate for studying the molecular evolution of rhodopsins. Although global comparisons of the frequencies of synonymous and nonsynonymous changes did not detect departures from neutrality, a likelihood ratio test of site- and branch-specific rates of evolution provided significant support for positive selection on sites within the gene and three sites were identified as probable targets of selection. These included Ala299Ser/Thr, a substitution that is known to cause a blue-shift of λmax in many species, probably due to its proximity to the Schiff base linkage between the protein and the chromophore. Also identified was a substitution at a site in the retinal binding pocket the functional importance of which is unknown except that it causes a shift in the absorbance spectrum of a cone opsin. The final candidate substitution is unusual in that it is located in a different part of the protein from the other substitutions.
Covariance of RH1 with the light environments of the populations supports local adaptation in this system and accounts for the incongruence of RH1 and neutral marker variation. The rhodopsin sequences of Iberian and North Atlantic sand gobies clustered together despite the genetic distinctness of these populations at other loci. These populations do, however, share similar light environments. The most striking similarity in RH1 sequences was between the Mediterranean and Baltic populations, which are divergent in both microsatellite and mitochondrial markers and the light transmittance measured using satellite images. However, the biology of the Mediterranean sand gobies suggests an explanation, as they spend much of their lives in coastal lagoons, which may have high turbidity in comparison with the ocean and thus have light environments dominated by longer wavelengths, similar to the Baltic Sea. More extensive measurement of the spectral qualities of sand goby habitats may shed light on the extent of local adaption in RH1. The direction of the expected shift in λmax produced by each substitution in RH1 is consistent with both the measured light environment and the λmax measurements made previously in Baltic, Atlantic and North Sea populations (see Jokela-Määttäet al. 2007 and references cited therein). As Larmuseau et al. (2009a) note, local adaptation implies that light environments affect the fitness of sand gobies and that they (and other fish) may be sensitive to anthropogenic changes in their light environment due to climate change, pollution or shipping, for example.
The study of Larmuseau et al. (2009a) is significant because it provides molecular evidence of local adaptation in the rhodopsin genes of fish. To date, our understanding of the evolutionary importance of visual pigments is most developed for the opsins related to colour vision in cichlids, where selection on colour vision has played a role in adaptive radiation (see Carleton et al. 2005; Terai et al. 2006; Seehausen et al. 2008). The application of population genetics to questions of adaptation marks a milestone in research on the spectral sensitivity of rod cells and dim-light vision, which were the original focus of visual ecology in fish. Nevertheless, functional testing of the candidate substitutions is needed to confirm the shifts in λmax produced in the sand goby RH1. Ultimately, this should be done using mutagenesis experiments, but since these substitutions typically vary within populations, microspectrophotometry and association mapping could provide alternative evidence of the link between genotype and phenotype. If this relationship is borne out, a simple molecular test of spectral sensitivity could be developed. A nondestructive test would greatly facilitate behavioural and evolutionary ecology research on vision in sand gobies.