Diurnal birds belong to one of two classes of colour vision. These are distinguished by the maximum absorbance wavelengths of the SWS1 visual pigment sensitive to violet (VS) and ultraviolet (UVS). Shifts between the classes have been rare events during avian evolution. Gulls (Laridae) are the only shorebirds (Charadriiformes) previously reported to have the UVS type of opsin, but too few species have been sampled to infer that gulls are unique among shorebirds or that Laridae is monomorphic for this trait. We have sequenced the SWS1 opsin gene in a broader sample of species. We confirm that cysteine in the key amino acid position 90, characteristic of the UVS class, has been conserved throughout gull evolution but also that the terns Anous minutus, A. tenuirostris and Gygis alba, and the skimmer Rynchops niger carry this trait. Terns, excluding Anous and Gygis, share the VS conferring serine in position 90 with other shorebirds but it is translated from a codon more similar to that found in UVS shorebirds. The most parsimonious interpretation of these findings, based on a molecular gene tree, is a single VS to UVS shift and a subsequent reversal in one lineage.
There is a categorical, physiological difference in colour vision between groups of diurnal birds (Cuthill et al. 2000). These have short-wavelength sensitive type 1 (SWS1) cone-opsin-based visual pigments of either an ultraviolet sensitive (UVS) type with maximum absorbance wavelengths (λmax) between 355 and 380 nm or a violet sensitive (VS) type with λmax between 402 and 426 nm (reviewed by Ödeen et al. 2009). The VS type appears to be ancestral (Yokoyama 2002) and phylogenetically most widespread (Ödeen & Håstad 2003) in birds. Independent shifts to UVS may have occurred only three or four times (Håstad et al. 2009): in shorebirds (Charadriiformes), in a common ancestor of passerines and psittaciforms, in Rhea americana and possibly in trogons (Carvalho et al. 2007). Gulls are the only shorebirds reported to be UVS—demonstrated through the presence of cysteine in amino acid (aa) position 90 (bovine rod opsin numbering) in their SWS1 opsin gene and ocular media with spectral transmission similar to that of other UVS birds (Ödeen & Håstad 2003; Håstad et al. 2005, 2009).
There is an increased cost of UV transmittance owing to photo-oxidation of the retina (e.g. Boulton et al. 2001), making the presence of UV tuning in gulls rather puzzling. Unlike other UVS birds, no selective advantage of UV vision has been demonstrated in gulls, whether for detecting sexual signals (e.g. Bennett et al. 1997) or foraging (e.g. Church et al. 1998). It has been suggested to allow eavesdropping on UV communication in schools of fish swimming close to the surface (Ödeen & Håstad 2003) but evidence seems weak. All auks, waders and terns investigated so far have SWS1 opsins of the VS type (Håstad et al. 2005), and some, especially terns, are ecologically similar to gulls. However, too few species have been sampled to safely infer that no other charadriiforms have UV-tuned visual pigments, or that this character state does not vary within Laridae.
The SWS1 λmax pigments in avian retinae closely follows aa sequence variation in the SWS1 opsin gene. In vitro studies have demonstrated that λmax can be predicted on the basis of the opsin's aa sequence (Wilkie et al. 2000; Yokoyama et al. 2000; Carvalho et al. 2007). The cone sensitivity can then be estimated from the opsin sequence, owing to covariance between the visual pigment's λmax and the spectral absorptance properties of the cone oil droplets (Hart & Vorobyev 2005). As spectral tuning of the SWS1 single cone is under direct genetic control, it becomes possible to identify the type of short-wavelength sensitivity in a bird from a sample of genomic DNA. This method was conceived by Ödeen & Håstad (2003) and validated by Ödeen et al. (2009). We have used it here to search for gross differences in colour vision in a significantly widened sample of gulls and other shorebirds. The aim has been to assess how stable the SWS1 pigment sensitivity has been during the evolution of shorebirds in general and gulls in particular. As the basis for this analysis, we have inferred a phylogeny, mainly using GenBank sequence data.
2. Material and methods
We studied a broad selection of species in the order Charadriiformes (table 1) with special emphasis on including a representative sample of various Laridae taxa. We sequenced a fragment of the SWS1 opsin gene containing the aa residues of positions 84–94 in up to six species each of seven shorebird families (table 1). We estimated λmax from three key positions, as outlined by Ödeen et al. (2009).
For the phylogenetic analyses, we used GenBank sequences of three mitochondrial and three nuclear loci, plus own β fibrinogen (FGB7) and myoglobin sequences of Anous tenuirostris and Gygis alba. Phylogenetic trees were constructed by maximum likelihood for one locus at a time and by Bayesian inference in the following combinations: (i) mitochondrial CO1, ND2 and cytochrome b, (ii) nuclear RAG-1, FGB7 and myoglobin intron 2 (myoglobin), and (iii) all loci together. For details, see the electronic supplementary material.
(a) Opsin gene analysis
All gull species investigated in this and previous studies (table 1) carry the UVS-pigment-conferring Ser to Cys substitution at aa position 90 (Ser90Cys: Wilkie et al. 2000; Yokoyama et al. 2000), as specified by the codon TGC in the SWS1 opsin gene. We found this trait in four other shorebirds, namely A. minutus, A. tenuirostris, G. alba (Sternidae) and Rynchops niger (Rynchopidae). All other species examined carry Ser90, which is ancestral to Charadriiformes (Ödeen & Håstad 2003; Håstad et al. 2005), but translated from different synonymous codons. In general, VS shorebirds carry AGC; the exception is Sternidae (excluding Anous and Gygis), where all sequences read TCC instead. The distribution of these traits on the phylogeny is shown in figure 1 (and the electronic supplementary material, figures S1 and S2).
All analyses produced results largely in agreement with previous studies (Ericson et al. 2003; Paton et al. 2003; Paton & Baker 2006; Baker et al. 2007; Fain & Houde 2007), and identify three main clades corresponding to Lari, Scolopaci and Charadrii. Only the tree based on all loci (figure 1) is shown (combined nuclear and mitochondrial gene trees in the electronic supplementary material, figures S1 and S2). All trees recover a strongly supported clade comprising Laridae, Sternidae and Rynchops. Within this clade, in the tree based on all loci, Laridae, Rynchops and Sterna form a poorly supported clade, which is sister to an insufficiently supported clade comprising Anous and Gygis. The nuclear tree has the same topology, but even weaker support. In contrast, in the mitochondrial tree, Laridae, Rynchops and Anous form a moderately well-supported trichotomy that is sister to a strongly supported clade containing Gygis as sister to the rest of Sternidae (excluding Anous).
Our results agree with previous studies in showing that gulls have Cys90 in their SWS1 opsins, characteristic of the UVS class of colour vision. This trait has been conserved during gull evolution, as revealed by the denser taxon sampling of this study compared with previous ones (Håstad et al. 2005). Another new finding is that gulls are not unique among shorebirds in being UVS, as the terns A. minutus, A. tenuirostris and G. alba and the skimmer R. niger share the UVS opsin forming Ser90Cys substitution.
The evolutionary interpretation of our results hinges on the correct inference of the phylogeny. As the relationships in the Laridae/Sternidae/Rynchopidae clade are somewhat uncertain, different interpretations are possible. However, the most parsimonious reconstruction suggests a Ser90Cys substitution in the common ancestor to Laridae/Sternidae/Rynchopidae, and a subsequent reversal to Ser in the Sternidae lineage (except Anous and Gygis), in all formed by two consecutive point mutations in the codon at aa position 90, from AGC to TGC to TCC. This is true even if the uncertainty in the gene tree is taken into account by considering the Sternidae branch (except Anous and Gygis) in alternative positions. For example, the posterior probability that Sterna is outside the clade comprising Laridae, Rynchops, Anous and Gygis—which would support a single Ser90Cys substitution but no subsequent reversal—is lesser than 0.06. Accordingly, our favoured hypothesis based on the available data is a single shift from VS to UVS in the common ancestor to Laridae/Sternidae/Rynchopidae and a subsequent reversal in the Sternidae lineage (except Anous and Gygis).
All key aas we report in positions 86, 90 and 93 have already been described (Wilkie et al. 2000; Yokoyama et al. 2000; Ödeen & Håstad 2003), but the spectral tuning effects of some of them are still unknown. The effect of Cys86 in birds is probably negligible (Ödeen et al. 2009), but neither the effects of Thr86 or Ile86 nor Ile93 has been investigated in vitro. However, the λmax of SWS1 pigment of Anas platyrhynchos, which carries Ile93, is not significantly different from that of species with aas of known effect in this position (Ödeen et al. 2009).
In one aspect, avian colour vision is under simple genetic control: the VS/UVS pigment opsin character depends on the Ser90Cys substitution, which can be caused by a single nucleotide mutation. Still, this character is surprisingly conserved among birds, indicating strong stabilizing selection. This is a rare example of a system where two very similar and well-understood variants of the same gene, which have a significantly different effect on the phenotype and ecology of the carrier, have been equally successful in a large taxon.
Tissue and DNA samples were kindly provided by Burke Museum, University of Washington, Seattle; Zoological Museum, University of Copenhagen, Copenhagen; LSU Museum of Natural Science, Baton Rouge; Swedish Museum of Natural History, Stockholm; Department of Animal Ecology, Uppsala University; Fredrik Widemo, Jean-Marie Pons and Malcom Wilson. This study was funded by the Swedish Research Council Formas (A.Ö.). Two anonymous reviewers provided constructive criticism, which has improved this paper.
- Received October 23, 2009.
- Accepted November 27, 2009.
- © 2009 The Royal Society