This study extends our understanding of SWS1 opsin function and evolution by investigating evolutionary changes that occurred in avian SWS1 genes. The SWS1 opsin of the great bowerbird C. nuchalis, a basal passerine bird, was expressed along with a series of spectral tuning mutants and ancestral passerine SWS1 pigments allowing us to investigate spectral tuning mechanisms and identify the evolution of UV/violet sensitivity in early passerines and parrots. The C. nuchalis SWS1 opsin was found to be a VS pigment, with a maximal absorbance of 403 nm, which is in agreement with previous MSP studies identifying a λmax of 404 nm . However, our experimentally recreated passerine ancestral SWS1 pigments were also found to be VS, addressing a longstanding issue of ancestral passerine SWS1 spectral tuning in previous studies [25, 28, 41, 55].
Evolution of UV/violet vision in passerines and parrots
Our finding that the passerine ancestor had a violet-type SWS1 reaches slightly different conclusions in comparison with a recent study suggesting that the passerine ancestor was UVS , which was the first paper examining avian SWS1 evolution that used a phylogeny in which passerines and parrots were specified sister orders. Not only are the predicted ancestral sequences different, but a VS-type λmax in ancestral pigment was experimentally confirmed in our study. While it is not entirely clear why our study reached such different conclusions, there are a number of important differences. Our analysis included additional outgroup sequences, and used maximum likelihood reconstruction methods (as opposed to parsimony). Furthermore, in our study the ancestral pigments were experimentally recreated and functionally assayed. Finally, our phylogeny is based on the current understanding of phylogenetic relationships among landbirds that includes a recent revision of the relationships among higher lineages [56–58], and therefore is somewhat different from that of Odeen et al. . However, we did not find any differences in our reconstructions of the ancestral passerine SWS1 when we used a tree with the relationships among higher passerines arranged similar to their phylogeny, suggesting that the difference in our findings from previous studies are probably due to methodological differences, such as the use of maximum likelihood reconstruction methods and/or the use of additional outgroup lineages. (Odeen et al.  did note that the inclusion of additional outgroup sequences resulted in an ambiguous reconstruction of the passerine ancestor even in their analyses.) Our results support earlier studies that investigated the evolution of UV/violet sensitivity in birds suggesting the passerine ancestor had a VS type SWS1 [25, 28, 55], but these early studies do not place passerines and parrots as sister orders. Because the parrots are now thought to be closer to the basal passerines than before, our results are more robust than they would be if based upon the older tree.
Our findings, that UVS in passerines and parrots evolved from VS ancestors, and that this occurred independently in at least two lineages, are rather unusual with respect to other vertebrate groups. The ancestral vertebrate state is thought to have been UVS, with VS pigments evolving independently in various lineages within fish, mammals, and amphibians [16, 22–24, 28, 29, 50]. Birds are believed to be an interesting exception where a switch to VS is thought to have occurred in the ancestral avian pigment with some descendants subsequently re-evolving UVS [24, 50]. Our identification of VS type pigments in both passerine and parrot/passerine ancestors confirm this hypothesis, and our ancestral reconstruction results provide a more precise prediction of where these spectral sensitivity shifts occurred. The re-evolution of UVS from a VS type pigment has not previously been predicted elsewhere in the vertebrate phylogeny. The reasons why bird SWS1 pigments are an exception remain largely unknown, but may be related to their unique spectral tuning mechanisms among vertebrates.
Spectral tuning in C. nuchalis SWS1
The C. nuchalis VS pigment possesses an unusual residue combination at the two spectral tuning sites known to be most important in specifying UVS or VS in vertebrates: C86/S90. This residue combination has been found in a few passerine SWS1 opsins in past sequence-based surveys [41, 42], but its spectral relevance has not been examined using mutagenesis experiments, which thus far have only dealt with VS-type pigments with S86/S90, in pigeon and chicken, [18, 28] and UVS type with either A86/C90 or C86/C90, in budgerigar and zebra finch, respectively [17, 18, 59]. Past mutagenesis studies of vertebrate SWS1 pigments have shown the magnitude of λmax shift caused by a given amino acid change can differ significantly among pigments due to synergistic interactions within and between transmembrane regions I-VII [19, 50, 60, 61]. Characterization of C. nuchalis SWS1 mutants was therefore carried out, as it may provide new clarification of the mechanisms contributing to the naturally occurring variation in avian SWS1 pigment spectral sensitivity, particularly among the VS type pigments. These mutants can also help clarify patterns of evolution between VS and UVS visual systems in birds.
Our results showing that S90C shifts the C. nuchalis SWS1 into the UV is consistent with previous studies where similar shifts have been documented in the chicken, pigeon, and the reverse in zebra finch, and budgerigar [17, 18, 28]. In C. nuchalis, the effect of the double mutant C86S/S90C was identical to that of the single S90C mutant. Thus, in the presence of C90, C86 has no additional effect on sensitivity. In other avian pigments, substitutions at known spectral tuning sites also do not change λmax if expressed with C90 [17, 28]. Others have suggested that the effect of C90 is so strong it prevents detection of any subtler effects other residues might have . In birds, all in vitro expressed pigments, whether wild type or mutant, with C90 have λmax ~360 nm. The exception is in chicken where S90C only shifts λmax to 369 nm .
The mutation C86F in C. nuchalis also shifts λmax into the UV. Unlike C90, which, as far as we know only has a functional role in avian SWS1 opsins, F86 is an important spectral tuning site across vertebrates where it confers UVS in most pigments in which it occurs [16, 19, 29, 30], the exception being the aye-aye, which is VS despite the presence of F86 . It is, in fact, believed to be the ancestral vertebrate state and substitutions from F86 are responsible for the loss of UVS in many mammalian lineages [16, 19, 22, 23, 29, 30], and in ancient birds . In C. nuchalis, C86 therefore plays an important role in maintaining sensitivity in the violet range, as the replacement of C86F shifts λmax into the ancestral UV state. F86 is also interesting because it has been suggested to be a second mechanism by which birds achieve UVS: It is found in the SWS1 genes of some birds including the trogon, paleognaths and a few sandgrouses and motmots [25–27], is capable of UV shifting VS pigments of pigeon and chicken , and is responsible for UVS in fish and most mammals [19, 29, 62]. Correspondingly, our mutagenesis results support the hypothesis that extant birds with F86 are UVS, and, therefore, the supposition that there are at least two mechanisms determining UVS in birds . The expression of a wild type pigment with F86 would be needed to confirm this hypothesis.
In contrast to the previous mutants, C86S did not affect λmax in the C. nuchalis SWS1. This mutation was previously suggested as contributing to the broad spectral variation observed among VS type pigments [55, 59], which in birds range from 388 nm (pigeon) to 420 nm (chicken) . Site 86 is an important spectral tuning site in other vertebrate SWS1 pigments, and S86C is capable of shifting λmax into the UV in a hypothetical ancestral avian SWS1 . As with C. nuchalis SWS1, S86C barely shifts λmax in the pigeon SWS1 , and mutation to serine at site 86 has no effect on the budgerigar SWS1 . Therefore the residues responsible for this large variation in λmax among VS pigments remain unknown. Altogether, these studies indicate that the role of site 86 in avian SWS1 pigments depends not only on the residue at that site, but also on the background in which it is expressed. This is particularly true of mammalian SWS1 pigments where the variation at site 86 is better characterized: in most mammalian pigments the presence of F86 dramatically shifts λmax, into the UV [16, 19, 29, 30], but this is not always the case .
Implications for behavioural ecology
While higher passerine lineages with UV type pigments are known to use UV signals in communication [9–11], current evidence indicates no link between colouration and spectral sensitivity in bowerbirds . Here we have shown that despite the fact males display UV reflecting feathers and objects during courtship [3, 63, 64], C. nuchalis does not possess a UV type SWS1 visual pigment. These findings would seem to contradict evidence demonstrating a strong link between spectral tuning and signal colouration in other vertebrate groups, [65, 66], and the belief that UV type pigments offer a dramatic advantage by improving sensitivity in this short wave range .
The general correlation between colouration and sensitivity remains because birds with VS pigments can perceive UV; SWS1 visual pigments absorb strongly over most of the UV visible range , cone oil droplets are effectively transparent to light in this range  and, in most species, avian ocular media transmit most short wavelength light . The difference in UV sensitivity between UVS, VS and the blue shifted bowerbird VS is just a matter of degree. Nevertheless, while UV colouration might be perceived by bowerbirds, its importance in communication is not well understood. In the satin bowerbird (Ptilonorhynchus violaceus) plumage UV reflectance is correlated with factors such as the intensity of infection from blood parasites, feather growth rate, and body size , but it is unrelated to mating success .
Given that C. nuchalis and other bowerbird ocular media transmit more UV wavelengths than most other species with VS-type visual pigments, they might represent a transitional link in the evolution from a VS to a UVS visual system . This hypothesis is supported by the comparatively blue shifted SWS1 found in bowerbirds, which further augments UV sensitivity. Given the similarly blue shifted λmax of the ancestral SWS1 pigments, this hypothesized transitional state might have originated in the ancestral passerine, and be shared among other basal passerines as well. This could also explain the unusually high number of shifts from VS to UVS in this order. Further investigation into the evolutionary history of ocular transmission would be useful to clarify this possibility.
If an organism with a blue shifted VS pigment, like the great bowerbird, has sufficient UV sensitivity, then the adaptive advantage of a switch to UVS might not be as large as it would be if it could perceive little UV or only had the ancestral VS pigment. Aside from λmax, there are a number of other structural and functional differences between VS and UVS opsins that may be related to a deprotonated Schiff base linkage to the chromophore [48, 51, 70–74]. These differences may have important consequences for the evolution of UVS in birds and other vertebrates. Therefore, it is possible that the wavelength difference between UVS and VS type pigments might not be the only, or the most important, functional difference between them. Further biochemical and mutagenesis studies would be necessary to refine the functional differences between these two opsin subtypes.