Conserved Genetic Basis of a Quantitative Plumage Trait Involved in Mate Choice

See allHide authors and affiliations

Science  19 Mar 2004:
Vol. 303, Issue 5665, pp. 1870-1873
DOI: 10.1126/science.1093834


A key question in evolutionary genetics is whether shared genetic mechanisms underlie the independent evolution of similar phenotypes across phylogenetically divergent lineages. Here we show that in two classic examples of melanic plumage polymorphisms in birds, lesser snow geese (Anser c. caerulescens) and arctic skuas (Stercorarius parasiticus), melanism is perfectly associated with variation in the melanocortin-1 receptor (MC1R) gene. In both species, the degree of melanism correlates with the number of copies of variant MC1R alleles. Phylogenetic reconstructions of variant MC1R alleles in geese and skuas show that melanism is a derived trait that evolved in the Pleistocene.

The genetic basis of independent origins of the same phenotype is important to models of phenotypic evolution. There are few data, especially for vertebrates, because the loci underlying phenotypic evolution in natural populations are rarely known. The lesser snow goose (Anser c. caerulescens) and arctic skua (or parasitic jaeger, Stercorarius parasiticus) have prominent melanic plumage polymorphisms (Fig. 1) showing clinal variation in the frequency of melanic morph phenotypes across their arctic breeding ranges (1, 2). In both species, there is quantitative variation in the degree of melanism among adult individuals with the melanic phenotype (“blue” snow geese and “intermediate” and “dark” skuas) and discrete separation between these and the nonmelanic phenotypes (“white” geese and “pale” skuas).

Fig. 1.

Color morph phenotypes in lesser snow geese and arctic skuas. (A) Left, white (phase 1) lesser snow goose; right, blue (phase 6) lesser snow goose; behind is an immature bird. (B) Pale arctic skua. (C) Intermediate arctic skua. [(A), copyright A. Morris/VIREO; (B) and (C), copyright M. Lane/Alamy]

These polymorphisms influence mate choice. In snow geese, mate color preference follows parental color, leading to assortative mating. It is believed to result from juveniles learning their parents' color at an early age and using this information in mate choice decisions (3). In arctic skuas, both female preference for dark males and positive assortative mating have been documented (4, 5). Pale arctic skuas of both sexes begin breeding at a younger age than dark birds, whereas dark male skuas, including first-time breeders and experienced breeders mating with a new female, breed earlier than pale birds in their first year of breeding but not subsequently (2, 4). Life history differences among white and blue snow geese have yet to be demonstrated (6). We studied a candidate gene underlying these plumage differences in geese and skuas, the melanocortin-1 receptor (MC1R) gene, which is expressed in melanocytes in developing feather buds and is a key regulator of melanogenesis in vertebrates, including birds (711).

The frequencies of color morphs of lesser snow geese show an approximate east-west cline across their arctic breeding range from eastern Canada to eastern Siberia, with blue individuals common in the east and comparatively rare in the west. We sampled white and blue geese from three polymorphic eastern populations (Akimiski Island, Baffin Island, and Cape Henrietta) and white geese from two western populations (Queen Maude Gulf and Wrangel Island). A nonsynonymous point substitution (Val85→ Met85) in the MC1R gene was perfectly associated with the blue phenotype throughout the entire range of lesser snow geese across North America (12). All blue geese sampled (N = 91) were heterozygous or homozygous for the Met85 allele, whereas all white geese (N = 116), were homozygous for the Val85 allele (Fisher's exact test, P < 0.001). Furthermore, there was a strong correlation between the number of Met85 alleles and the degree of melanism (Fig. 2) in the two polymorphic populations examined. The degree of melanism in blue lesser snow geese was scored from phase 1 (palest) to phase 6 (darkest) (13). All phase 2, 3, and 4 individuals were heterozygotes, whereas phase 5 and 6 birds comprised increasing proportions of Met85 homozygotes.

Fig. 2.

Association between copy number of variant MC1R alleles and degree of melanism in lesser snow geese (A) and arctic skuas (B).

Blue and white populations probably only came into contact about a century ago (14), but population histories cannot account for the association of MC1R with phenotype, because if white and blue birds are considered as separate populations (that is, phase 1 birds versus phase 2 to 6 birds), differentiation at the MC1R locus among morphs at Baffin Island [fixation index (Fst) = 0.78, P < 0.01] and Akimiski Island (Fst = 0.56, P < 0.01) is far greater than that at other nuclear loci (microsatellite loci: Baffin Island, average over 10 loci, Fst = 0.02, P < 0.05; Akimiski Island, average over 9 loci, Fst = 0.01, NS) or in mitochondrial DNA (mtDNA) [Akimiski Island, fixation index (Φst) = 0.00 (15)].

These data show that the MC1R locus is the major determinant of melanism in lesser snow geese, and the Met85 allele is equivalent to the melanic allele hypothesized from modeling of transmission genetics in wild populations (13, 16). The Val85→ Met85 substitution may be directly responsible, because it is in the outer part of the second transmembrane domain of MC1R, which is known to control MC1R activity and amount of melanin (7, 8, 17), but the involvement of closely linked sites outside the region sequenced cannot be ruled out. There is no evidence that variation at other segregating MC1R sites sequenced contributes to the plumage phenotype. Variation among individuals of phases 2 to 6 that are Val/Met85 heterozygotes is not associated with variation at any of the eight other variable MC1R sites (18).

In the greater snow goose (Anser c. atlanticus) and Ross' goose (Anser rossii), which are close relatives of the lesser snow goose, blue morph individuals occur only occasionally. White morphs of these taxa were homozygous for the Val85 allele, whereas the single blue Ross' goose examined was a Val85/Met85 heterozygote. Most MC1R haplotypes in Ross' goose were shared with snow geese (Fig. 3). Reconstruction of the evolutionary history of the Met85 allele in a haplotype network (Fig. 3) shows that it is derived from the Val85 allele and that it occurs in a single haplotype group (haplotype group 13). The derived position and low number of Met85 haplotypes (2) as compared to Val85 haplotypes (26) suggest a relatively recent origin of Met85 haplotypes, and hence melanism. Coalescent simulations (19) estimate the age of the Val85→Met85 mutation to be 380,000 ± 188,000 (SD) years. These results are consistent with suggestions that melanism arose and became fixed in an isolated eastern population of lesser snow geese in the Pleistocene (14). Another proposed Pleistocene vicariance event that occurred in the common ancestor of snow and Ross' geese, based on the presence of two divergent mtDNA clades in both taxa (20, 21), must have pre-dated the Val85→Met85 mutation, because the two mtDNA clades are present in similar frequency in both blue and white geese.

Fig. 3.

MC1R haplotype networks in (A) geese and (B) skuas, obtained using TCS software (30, 31). Thick lines represent nonsynonymous substitutions; thin lines are synonymous substitutions. Networks are rooted using chicken MC1R. Numbers indicate haplotype groups (snow and Ross' geese) or haplotypes (arctic skuas). Haplotype groups in geese are haplotypes that were defined ignoring variation at two hypervariable synonymous sites in MC1R (nucleotide sites 378 and 408) [see methods (12) for details]. The inset in (A) shows haplotypes sampled in snow geese and Ross' geese, with the area of the circles approximately proportional to haplotype frequency. Black and white circles correspond to melanic and nonmelanic haplotypes, respectively.

Color morph frequencies in the arctic skua show a latitudinal cline, with the proportion of pale birds increasing northward in the circumpolar breeding range (2). A separate nonsynonymous point substitution in the MC1R gene (Arg230→ His230) is associated with melanic plumage in adult arctic skuas. All melanic birds sampled (N = 16) in the polymorphic population at Slettnes, Norway, were heterozygous or homozygous for the His230 allele. Pale birds (N = 12) from the same polymorphic population in Slettnes and a monomorphic population at Komi, northeast Russia, were homozygous for the Arg230 allele (Fisher's exact test, P < 0.01). The degree of melanism correlates with copy number of the Arg230 allele (Fig. 2), with all intermediate birds being heterozygous at site 230, whereas a high proportion (83%) of dark birds were His230 homozygotes. Genetic differentiation among pale and intermediate to dark skuas is high at MC1R (Fst = 0.56, P < 0.02) and absent in mtDNA [Φst = 0.01, NS (22)]. Together these data strongly suggest that MC1R is the major locus determining plumage coloration in arctic skuas that had been predicted from genetic modeling (4). The Arg230→ His230 mutation may be causative, because a histidine at the homologous site in the third intracellular loop of MC1R is also associated with melanism in rock pocket mice (Chaetodipus intermedius) (23, 24).

A different nonsynonymous substitution at the same amino acid site (Arg230→Cys230) was found in the relatively dark great skua (Catharacta skua), a species representing the second lineage within the skua family (Stercorariidae). Thus, melanism may have evolved independently in the great skua lineage since it shared a common ancestor with arctic and long-tailed skuas (Stercorarius longicaudus). This hypothesis is supported by phylogenetic studies indicating that the common ancestor of today's skua lineages may have been Stercorarius-like, with barred juvenile and white-breasted alternate adult plumage similar to that of gulls (25). The His230 allele in arctic skuas is present in a derived position on a haplotype network (Fig. 3), suggesting that ancestral arctic skuas were pale. Coalescent simulations estimate an upper bound for the age of the Arg230→His230 mutation in arctic skuas at 340,000 ± 248,000 (SD) years, indicating a Pleistocene origin for the polymorphism.

The precise effects of MC1R variation on plumage coloration and patterning are surprisingly diverse. In snow geese, the dosage of variant MC1R alleles affects patterning: Color phases differ in the amount of the body covered in melanic feathers. In contrast, melanic arctic skua phases show graded differences in the amount of melanin in individual feathers over the neck, breast, and belly. In bananaquits (Coereba flaveola), the Glu92→Lys92 variant MC1R allele acts as a melanic switch, so that all individuals with one or two copies of this allele have completely melanized feathers throughout their plumage (10). MC1R variation is also associated with patterning effects in domestic chickens (8, 26), but not with small-scale feather tip melanization across species of Phylloscopus warblers (27). In addition, snow geese, arctic skuas, and bananaquits show that MC1R frequently controls variation between eumelanin production and the absence of melanin, whereas in chickens and most mammals it controls relative amounts of eumelanin and red/yellow phaeomelanin (8, 9, 26).

Variation in plumage color in geese and skuas provides a rare example where the major molecular genetic determinant of a quantitative trait has been identified in wild populations. The association of MC1R variation with naturally occurring melanism in three divergent avian lineages (Anseri-forms, snow geese; Charadriiforms, arctic skuas; and Passeriforms, bananaquits) reveals a conserved mechanism of plumage color evolution through many tens of millions of years of avian history (28). The repeated involvement of MC1R is surprising, because over 100 loci are known to affect pigmentation in vertebrates (29). This presumably reflects some combination of a high mutability to functionally novel MC1R alleles, a relative absence of deleterious pleiotropic effects of these alleles, and the visibility of dominant or codominant melanic MC1R alleles to natural selection. Our results provide strong support for the notion that, at least in the case of melanism in birds, evolution is driven by mutation rather than selection on existing standing genetic variation.

Supporting Online Material

Materials and Methods

Tables S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article