DISCUSSION
We found that CYP2J19 on chromosome 8 was within one of three regions associated with crown colour across a natural hybrid zone in tinkerbirds. Indeed, it was the only region of the genome that was significantly associated with both hue, determined from spectral reflectance, and plumage colour score. Only extoni individuals homozygous for the CYP2J19 -linked SNP exhibited a yellow forecrown, while all individuals homozygous for the pusillusallele had red forecrowns. All heterozygotes at theCYP2J19 -linked locus had either red, reddish or orange crowns.CYP2J19 is a cytochrome P450 enzyme, which has been implicated as encoding the ketolase that mediates red coloration in birds (Lopes et al., 2016; Mundy et al., 2016). Ketolases convert dietary yellow carotenoids into red ketocarotenoids in birds with red feathers (Toews et al., 2017; Norden & Price, 2018). CYP2J19 has also been associated with red retinal oil droplets influencing colour vision in birds (Mundy et al., 2016), proposed to be its ancestral function, with its role in red coloration having evolved in some but not other species (Twyman et al., 2018a). It has also been shown to function in red integumental coloration in turtles (Twyman et al., 2016).
CYP2J19 is not the only gene shown to function in red coloration, however, with recent work finding that a mutation in the beta-carotenoid oxygenase 2 gene (BCO2) functions in red bare part coloration in birds (Gazda et al., 2020). The CYP2J19 linked SNP on chromosome 8 was also not the only region associated with red forecrown colour in tinkerbirds, with chromosome 20 associated with crown score and chromosome 24 with crown hue. BC02 is on chromosome 24, and in addition to its association with crown hue, a single SNP on scaffold 70216 on chromosome 24 was also significantly associated with crown score. However, the scaffolds significantly associated with forecrown colour in tinkerbirds on chromosome 24 aligned to the interval from 5.8 to 6.0 Mbp on zebra finch chromosome 24, while BCO2 spans from 1.53 to 1.54, suggesting that the SNPs associated with forecrown colour are not linked to BCO2 . We found that forecrown colour could be intermediate between red and yellow, with some heterozygotes at theCYP2J19 locus displaying orange forecrowns. This suggests possible additive effects of the gene rather than dominance of one allele over the other, and/or interactions with genes at other regions we found associated with forecrown colour. There was a cluster of SNPs on adjacent scaffolds on chromosome 8 that overlapped with several genes in zebra finch, but to our knowledge there are presently no known functional associations of those genes with carotenoid-based coloration determination.
We found no significant association with crown chroma. Unlike in hue, the two species were shown to overlap in chroma between sympatric forms by Nwankwo et al. (2019), with both species showing much more variation in this measure of colour intensity than in the measures of hue and brightness. We did, however, find SNPs significantly associated with supercilium colour, but not the extent of yellow on the wing bar. The SNPs associated with supercilium colour were on a different chromosome to those associated with forecrown colour, indicating that a different pathway is involved in mediating carotenoid-based coloration in those traits. This is not unexpected. Those plumage patches varied between white and shades of yellow, while CYP2J19 has been shown to function in red feather colour determination (Lopes et al., 2016; Mundy et al., 2016).
In another pair of sister taxa of Pogoniulus tinkerbird, yellow-rumped (P. bilineatus ) and yellow-throated tinkerbird (P. subsulphureus ), which meet in the forests of west and central Africa, reproductive isolation is maintained through character differences such as throat colour and song, which have diverged in sympatry (Kirschel et al., 2009; Kirschel et al., 2020). By contrast, yellow-fronted and red-fronted tinkerbird hybridise extensively, and hybridisation is asymmetric (Nwankwo et al., 2019). The pattern of asymmetry in putative hybrids with extoni ancestry and predominantly pusillus Z chromosome suggests red-fronted males are much more likely to mate with yellow-fronted females than the other way around. Even when the reverse scenario does happen, based on the proportion of genotypically extoni individuals in the contact zone with red forecrowns (Nwankwo et al., 2019), a high proportion of genotypically extoni males mating with pusillus females might indeed sport reddish forecrowns.
Such asymmetry between mitochondrial haplotype and hybrid index of the Z chromosome could result from a genetic incompatibility in the reverse cross, i.e. pusillus haplotype females interbeeding withextoni males. Yet, such crosses do occur – two individuals in our sample with pusillus haplotype had extoni fathers (AR93115 and AR93163), and both individuals are females – the heterogametic sex - theoretically more likely to suffer from inviability and infertility (Haldane, 1922). Instead, we suggest the asymmetry is a consequence of female preference for males with red forecrowns inpusillus and potentially in extoni females (c.f. Baldassarre et al., 2014). The pusillus mitochondrial haplotype is less than one quarter as common as the extoni haplotype in the contact zone (8/37 individuals, see Nwankwo et al. (2019)). Yet femalepusillus appear to put the extra effort into findingpusillus males, outnumbered four-to-one by their extonicounterparts. By contrast, female extoni appear to breed with males of either genotype at rates equivalent to their frequency in the population. Mate choice experiments would be needed to test whether female extoni would choose male pusillus (or otherwise males with extoni ancestry but with introgressed red forecrowns) over yellow-fronted male extoni , when given a choice. But it does appear to be the failure of female extoni to mate assortatively with males of their own species and phenotype likely drives introgressive hybridisation between these species. Premating isolation between the two species remains weak in spite of 4 million years of divergence, according to their mitochondrial DNA phylogeny (Nwankwo et al., 2019). The choosiness of female pusillus to mate assortatively on the other hand might be the main factor maintaining a narrow tension zone that does not progress into panmixia and species collapse (e.g., Kleindorfer et al., 2014; Kearns et al., 2018).
Red feathers result from the conversion of dietary carotenoids to red ketocarotenoids and such converted carotenoids have been shown to be an honest indicator of fitness that could function in mate choice (Weaver et al., 2018). Here, female red-fronted tinkerbirds might only mate with males that display red forecrowns because they consider yellow-forecrowned males of lower quality. Greater extents of difference in plumage between related species have been proposed to play a role in premating isolation (Scordato et al., 2017). Here, though yellow-fronted females instigate asymmetric introgression by not mating assortatively with yellow-fronted males. If there was a sensory bias for red plumage coloration the two species would surely collapse into one if there is no loss of fitness in hybrids, and with heterozygotes at the CYP2J19locus displaying more reddish forecrowns, we would also expect red plumage to rapidly introgress across the population. Further work is needed to determine the extent to which species boundaries are maintained and the rate of introgression of red plumage across the contact zone.
Both quantitative (hue) and qualitative (crown score) metrics recovered the same major-effect locus. This might not be surprising if they are both measures of the same trait, but each method has its own inherent biases. Scoring from photographs is subjective and affected by each observer’s perspective of the effects of ambient light. While correlated, there was a resultant variation in their respective scores, hence the number of individuals with intermediate scores such as reddish, which resulted from one observer scoring the forecrown red and the other scoring it orange. Likewise, reflectance spectrometry is also susceptible to variation in feather and probe placement, though we did attempt to control for this with measurements in two orientations. Despite these potential biases in qualitative and quantitative metrics, the same major-effect locus was recovered. Furthermore, we showed that admixture mapping is feasible using RADseq with a sample size < 50, as was shown for forecrown hue, but only for major-effect loci. This was possible in spite of no prior genomic resources available for Pogoniulus tinkerbirds or any closely-related genera.