We believe that our study is among the most comprehensive genomic analyses of opsins in vertebrates, and particularly in birds. We have been able to characterize losses, gains, and selective evolution that are correlated with lineage-specific traits. Below we discuss the implications of the key highlights for each family of opsins.
The loss of PARA and PARIE in mammals and birds
The apparent independent loss of PARA and PARIE in mammals and birds is intriguing. In non-avian and non-mammalian vertebrates PARA and PARIE pineal opsins genes are expressed in the parietal organ, which is a part of the epithalamus . The parietal organ, along with the pineal organ, forms the parietal eye (or third eye), which functions as a proper photoreception organ, regulating circadian rhythmicity and hormone production for thermoregulation . In birds and mammals the parietal organ degenerates completely [12, 31, 32], which is very likely associated with the loss of PARA and PARIE. Taking into consideration the thermoregulatory function of the parietal organ in vertebrates  and that mammals and birds are endothermic, we consider that the independent loss of these genes in these lineages could be related with a change in the mechanisms regulating body temperature. As mammals and birds became less reliant on external sources of energy to maintain body temperature (evolution of endothermy ), the parietal organ would have degenerated (accompanied by the related signaling pathways).
Adaptive evolution of avian opsins
Apart from the PARA and PARIE opsins, most modern birds maintained the vertebrate repertoire of opsins, suggesting that birds never became specifically adapted to limited photic conditions that might have led to the extensive pseudogenization of opsins. Additionally, since we find no global events of opsin loss during the early modern avian species radiation, birds appeared to have possessed tetrachromatic vision (RH2, OPN1sw1, OPN1sw2 and OPN1lw) for most of their evolutionary history. This suggests that birds relied on a visual system specialized for discriminating different light qualities, particularly useful in complex photopic environments, where birds likely diversified.
In birds, 11 of the 15 avian opsins evolved in a non-neutral manner: RH1, OPN1sw1, OPN1sw2, OPN1lw, TMT, OPN3, PIN, VA, RGR, RRH and OPN4x. This suggests that visual and non-visual adaptive strategies have been imperative during avian evolution, validating the importance of the visual sense in birds. Among the visual opsins, sw1, sw2 and lw have been more conserved in birds, while for RH1 positive selection was found in birds and negative selection in mammals. Visual conopsins perform image-forming functions , and are particularly important for photopic animals, like birds. The patterns of purifying evolution in the sw1, sw2 and lw would have permitted fine adjustment of the spectral sensitivities of these opsins, ensuring elevated photic acuity throughout avian evolution. In addition, the higher site-specific adaptive rates of sw1 and lw opsins in mammals relative to birds is consistent with our preliminary analyses in Zhang et al. (2014) , where we have found that the sw1 ω-ratio is lower in birds than in mammals (0.16/0.21 respectively). This suggests that the mechanism to maintain optimal color discrimination is more stringent in birds than in mammals.
RH1 has the key function of conferring monochromatic vision in low light environments , and thus it is not surprising that it diversifies in birds that were mostly photopic-adapted. In contrast, in mammals the negative selection is consistent with their nocturnal habits and the anatomical features of the mammalian eye that are congruent with nocturnal ancestry . More specifically, our findings that site 217 of RH1 evolved under positive selection in birds is consistent with findings of positive-selection on this site in other vertebrates [26, 37]; however, it has been reported that different amino acids at these sites do not seem to cause spectral shifts in RH1 . Instead, we found an association between the T and M/A amino acid residues and the evolution of land and water neoaves clades – similar ecological condition also influenced the evolution of the olfactory receptor subgenomes . Possible contributions of these specific amino acid substitutions to water and land adaptations can be tested through in-vitro experiments to verify the role of this site in RH1 spectral tuning. Overall, the contrasting evolutionary signatures in the visual opsins between mammals and birds are consistent with their contrasting photic needs in that scotopic-adapted animals need to maximize the amount of light collected, while photopic-adapted animals require enhanced visual acuity.
OPN1sw1 evolution and VS/UVS vision
Our analysis suggesting that the ancestral bird possessed a VS OPN1sw1 opsin is consistent with other analyses of amino acid variation of sw1 spectral tuning sites that concluded that VS was the probable ancestral condition of birds [16, 39]. The ecological role of the OPN1sw1 opsin in birds is not well understood, but is likely to be broad, as it has been associated with coloration pattern recognition , social signaling, hunting, nectar localization and mate-choice [6–8]. Others also suggest a role in non-visual processes such as circadian rhythm regulation [40, 41]. Species in six avian orders, Pteroclidiformes, Charadriiformes, Coraciiformes, Trogoniformes, Psittaciformes and Passeriformes, have been shown to possess UVS sensitivity , of which the later four belong to core landbirds. However, there are no clear patterns of coloration, breeding behaviors, activity patterns and feeding habits among the species in these groups that would explain the acquisition of UVS (or the retention of VS).
The use of UVS is most clear among Passeriforms, which is consistent with evidence of strong positive selection found in the branch leading to the passerine group for the sw1 gene . However, some Passeriformes species are also VS, which is reflected in a relative higher ω-lineage in the golden-collared manakin and American crow. These two species have been reported as cases of recent adaptation to the VS vision . The contrasting root-to-tip ω-lineages appears to be an efficient methodology to study ecological adaptation scenarios in phylogenetic contexts, as it is sensitive enough to detect episodes of reversal evolution.
Another species associated with a high ω-lineage for the VS class is the paleognath tinamou. The tinamou sw1 amino acid sequences had F, C and M residues at sites 86, 90 and 93, which corresponds with the amino acid conformation found by Ödeen et al. (2013) in the closely-related ostrich (Struthio camelus) . The sw1 sensitivity in paleognaths has been somewhat controversial, because while microspectrophotometry analysis of the sw1 in ostrich suggests VS , the 86 F and 90C residues support UVS . Our results, linking the tinamou with a high ω-lineage in the VS class suggests that the UVS → VS shift was relatively recent. This is congruent with the amino acid sequence similarity with UVS sw1 and also with the microspectrophotometry analysis. Therefore, we presume that there are other amino acids in addition to 86 and 90 that likely changed the spectral sensitivity of the sw1 photopigment, such as 93 M as suggested by Ödeen et al. (2013) .
The adaptive association between the MC1R gene and the RH2 visual opsin in avian species suggest that plumage colorations have been photic mediated. The MC1R receptor is involved in melanin-based coloration in vertebrates and is found primarily in melanocyte cells where it controls the deposition of melanin in tissues . Activation of MC1R leads to increased synthesis of black/brown eumelanin, whereas low MC1R activity leads to increased synthesis of red/yellow phaeomelanin . First cloned from chickens , several studies have found that MC1R is closely associated with plumage coloration [46–49]. MC1R adaptive evolution has also been correlated with the degree of sexual dichromatism in galliform birds, suggesting that MC1R may be a key link in the interaction of sexual selection and plumage colour . Sexual selection would also be a reasonable explanation for the MC1R/RH2 co-evolution. Indeed, one would expect that plumage coloration patterns would only be important for birds if they were associated with a visual system capable of “read” plumage coloration cues. Indeed, it has been shown that tetrachromatic vision, a process which requires the RH2 photopigment, enhances plumage discriminability in birds . In addition, Bloch et al. (2015) have suggested that rapid evolution of RH2 in Setophaga birds (a genus of Passeriformes) is linked to sexual selection, given their exceptional plumage color diversification . In situations in which sexual selection evolves in association with plumage coloration patterns and color discriminability, then there should be strong associations among MC1R and visual opsins. Further tests on the sw1, sw2 and lw opsins would be welcome.
More-recent photic adaptations in birds: the barn owl
The barn owl has very distinctive photoreceptive features relative to other birds [53, 54]. Due to a recent nocturnal adaptation, barn owls have frontally placed eyes and anatomical adaptations that improve perception of photic stimuli in low light environments; this includes an elongated eye and a high ratio between the eye and corporal sizes [53, 54]. The pseudogenization of RH2 and the lineage-specific acceleration of PIN are consistent with these adaptive changes.
RH2 is sensitive to the green photo spectrum from about 480–535 nm  and has undergone rapid gene loss and gain in other vertebrate lineages (reviewed in ). In addition, RH2 was lost in placental mammals during the nocturnal bottleneck [3, 36]. There is evidence that some owls have a photoreceptor that is sensitive to the 503 nm spectrum, which would be consistent with a RH2-type photopigment . If confirmed that this photopigment is from RH2, the pseudogenization event reported here for the barn owl would likely be lineage-specific.
PIN is a blue-sensitive pigment (~470 nm) expressed in the pineal gland (determined in chicken) that has a role controlling the circadian pacemaker and the rhythmic production of melatonin [57, 58]. The accelerated evolutionary rate observed in the barn owl PIN is appreciable (0.548/0.146, i.e. 3.7 times faster than the avian trend) and includes several non-synonymous mutations. Although it is not known if the PIN opsin is fully functional in the barn owl pineal gland, owls possess a rudimentary pineal with the pinealocytes having rudimentary photoreceptive features .The PIN protein may continue to have a role in circadian tasks associated with a nocturnal lifestyle or the degeneration of the pineal gland may have permitted the unconstrained molecular evolution of PIN.
Other genes that are likely to be involved in the adaptation of birds to a nocturnal life style include sw1, sw2 and lw visual conopsins, which we were not found in the barn owl genome. Zhao et al. (2009a) performed phylogenetic analysis in the lw and sw1 photopigments in nocturnal bats and affirmed the importance of the sw1 in the species’ sensory ecology . In particular, it would be important to determine if the sw1 photopigment in birds is UV/UVS sensitive, as UVS vision has been associated with nocturnal habits in mammals [60, 61].
More-recent photic adaptations in birds: the penguins
Penguins possess specialized and unique optic adaptations, including an approximately-spherical lens and a flat cornea that augment their vision when underwater . At the molecular level, evidence of gene pseudogenization and positive selection in phototransduction genes have been associated with the aquatic lifestyle of the Adélie and emperor Antarctic penguins . Penguin specializations include cone visual pigments tuned towards the blue-green range of the visual spectrum, presumably related with the spectral composition of their aquatic environment . The retuning of the RH2 in penguins could be linked with non-synonymous mutations in the D83, Q122, A164, A207 and S222 amino acids of RH2 (site identification based on bovine rhodopsin homolog sites) [55, 64]. Although we did not find any variation in the two penguins for these sites, we found site deletions at S295 and K296. S295 is responsible for the red shift in RH2  while 296 K is an important retinal binding site . It is not known if this causes RH2 to be non-functional in penguins, which would require functional experimental tests. However, these indels were shared by the two penguin species, which shared a common ancestor ~23.0 mya , meaning that if they were deleterious (or significantly compromise the RH2 function) we would expect the RH2 to pseudogenize over that time period, which was not observed in any of the penguin lineages. Most likely, these indels are evidence that penguins have adapted a unique mechanism to perform the molecular interactions that mediate the photon absorption in RH2 that is better suited for underwater environments.
Additionally, we have found evidence of accelerated evolution of RH1, OPN4x and OPN5 in the penguin lineage. OPN4x is only present in non-mammalian vertebrates and is associated with non–image–forming light responses, including circadian entrainment [66–68]. The accelerated evolution of OPN4x suggests that penguins may have evolved new circadian responses to cope with the seasonal particularities of Antarctica, including the dramatic daily light changes and hourly differences. OPN5 is UV-sensitive, is expressed in the chicken retina and pineal gland, and plays a role in the assistance of an 11-cis-retinal-supplying system . The role of the OPN5 in penguin vision is less obvious.
RH1 has been associated with nocturnal/diurnal terrestrial lifestyles, but some studies have shown that RH1 underwent shifts in spectral tuning in marine mammals . Evidence from aquatic mammals are in congruence with the accelerated evolution of RH1 in the ancestral lineage leading to penguins. Zhao et al. (2009b)  reported evidence of positive selection in the cetacean and pinniped aquatic clades, suggesting that RH1 evolution were related with the turbid condition of aquatic environments. In addition, the occurrence of P in the 194 spectral tunning site was also verified in cetaceans, particularly in the sowerby's beaked whale (Mesoplodon bidens) . Changes in RH1 molecular features are compelling evidence that it contributed to penguin’s unique adaptive strategies to aquatic environments.
Similarly as for the barn owl, the emperor penguin also showed lineage-specific changes in PIN. However, we were unable to determine if is this is a case of adaptive or unconstrained evolution. Nevertheless, evidence suggests that the pineal organ of the Antarctic penguin (Pygoscelis papua) lacks typical photoreceptor elements , which as observed by Li et al. (2014), is the likely cause of the accelerated evolution/pseudogenization of the PIN opsin in penguins .