Research Article

Intraspecific Polymorphism to Interspecific Divergence: Genetics of Pigmentation in Drosophila

See allHide authors and affiliations

Science  23 Oct 2009:
Vol. 326, Issue 5952, pp. 540-544
DOI: 10.1126/science.1176980


Genetic changes contributing to phenotypic differences within or between species have been identified for a handful of traits, but the relationship between alleles underlying intraspecific polymorphism and interspecific divergence is largely unknown. We found that noncoding changes in the tan gene, as well as changes linked to the ebony gene, contribute to pigmentation divergence between closely related Drosophila species. Moreover, we found that alleles linked to tan and ebony fixed in one Drosophila species also contribute to variation within another species, and that multiple genotypes underlie similar phenotypes even within the same population. These alleles appear to predate speciation, which suggests that standing genetic variation present in the common ancestor gave rise to both intraspecific polymorphism and interspecific divergence.

Similar phenotypes that vary within and between species may or may not be caused by similar genetic mechanisms. Quantitative trait mapping shows that loci contributing to polymorphism and divergence of a single character map to the same region of the genome approximately half of the time (table S1). These overlapping quantitative trait loci (QTLs) may or may not result from changes in the same genes, and most studies lack the power to distinguish between these possibilities. To determine whether the same genes (and potentially even the same alleles of these genes) contribute to phenotypic diversity within and between species, one must resolve intra- and interspecific QTLs to individual genes, localize functionally divergent sites within these genes, and then compare specific alleles within and between species.

ebony and tan QTLs contribute to pigmentation divergence. To investigate the relationship between intraspecific polymorphism and interspecific divergence, we examined the genetic basis of pigmentation differences within and between a pair of closely related Drosophila species, D. americana and D. novamexicana. These two species are sister taxa within the Drosophila virilis species group that diverged about 300,000 to 500,000 years ago (1, 2) (Fig. 1A). D. novamexicana has a derived light yellow body color, whereas other members of this group (including D. americana) retain an ancestral dark brown body color (3) (Fig. 1B). In the laboratory, these species can mate and produce fertile offspring. Genetic mapping showed that a region of the second chromosome containing the ebony gene contributes to pigmentation divergence between D. novamexicana and D. americana (4). This gene is required for pigmentation in D. melanogaster (5). Three other autosomal regions, as well as an unidentified region of the X chromosome, also contribute to pigmentation divergence, although none of these regions were linked to other pigmentation genes tested (i.e., yellow, dopa-decarboxylase, optomotor blind, and bric-a-brac).

Fig. 1

D. novamexicana yellow body color is derived. (A) Phylogenetic relationships among members of the virilis phylad within the virilis group of Drosophila are shown with estimated divergence times (1) at each node (numbers denote millions of years ago). (B) Dorsal body pigmentation is shown for D. virilis (vir), D. americana (amer), and D. novamexicana (nova). D. lummei (not shown) has pigmentation similar to that of D. virilis and D. americana (3).

Recently, the X-linked pigmentation gene tan was cloned in D. melanogaster (6). To test whether this gene might contribute to pigmentation differences between D. americana and D. novamexicana, we crossed D. americana females to D. novamexicana males, backcrossed F1 hybrid females to D. novamexicana males, and scored 495 backcross progeny for body color (fig. S1). All of the lightest male offspring (n = 10) inherited the D. novamexicana allele of tan, whereas all of the darkest male offspring (n = 24) inherited the D. americana allele of this marker. These data show that sequences linked to tan contribute to pigmentation divergence (P = 8 × 10−9; Fisher’s exact test). The previously described pigmentation QTL linked to ebony and the lack of a pigmentation QTL linked to yellow (4) were also reconfirmed in this population (P = 3 × 10−8 and 0.7, respectively; Fisher’s exact test).

To determine the phenotypic effects of QTLs linked to ebony and tan, we created lines of D. novamexicana in which genomic regions containing these genes were replaced with orthologous sequences from D. americana. These genotypes were constructed by marker-assisted introgression, moving ebony and tan alleles from D. americana into D. novamexicana. F1 hybrid females were backcrossed to D. novamexicana males, and a single female inheriting the D. americana tan (or ebony) allele was randomly selected and backcrossed to D. novamexicana males again. This process was repeated for 10 generations (fig. S2), with females carrying the D. americana ebony or tan allele selected randomly in each generation without regard to pigmentation. Introgressed D. americana sequences linked to either tan or ebony darkened pigmentation relative to wild-type D. novamexicana (Fig. 2, A to C), with sequences linked to ebony (Fig. 2C) causing darker pigmentation than sequences linked to tan (Fig. 2B). Digital quantification of pigmentation showed that, when combined, the introgressed tan and ebony regions recapitulated 87% of the pigmentation difference between species (Fig. 2, A, D, and E).

Fig. 2

QTLs linked to tan and ebony account for the majority of pigmentation divergence between species. Dorsal abdominal cuticle is shown from segments A4 and A5 of 7- to 10-day-old adult females. (A to C) Relative to D. novamexicana (A), introgression of alleles linked to tan (B) or ebony (C) led to darkened pigmentation. (D and E) Together, the introgressed regions produced even darker pigmentation (D), although these flies were not as dark as wild-type D. americana (E). Numbers indicate intensity of grayscale images, where 0 = black and 255 = white. Panels (B), (C), and (D) are all heterozygous for the introgressed region(s).

ebony and tan affect pigmentation development. Studies of pigmentation in D. melanogaster suggest that ebony and tan may themselves be responsible for these interspecific QTLs. Loss-of-function mutations in ebony darken pigmentation (5), whereas loss-of-function mutations in tan lighten it (7). Biochemically, Ebony catalyzes the conversion of dopamine into N-β-alanyl-dopamine (NBAD), which is a precursor for (yellow) sclerotin, and Tan catalyzes the reverse reaction, converting NBAD back into dopamine, which is a precursor for (brown) melanin [reviewed in (8)] (Fig. 3A). Ectopic expression of Ebony induces yellow pigmentation (9) (Fig. 3D), whereas ectopic expression of Tan induces brown pigmentation (6) (Fig. 3E). Ectopic expression of both proteins simultaneously results in pigmentation intermediate to that caused by ectopic expression of either protein alone (Fig. 3F), showing that the balance between Ebony and Tan enzymatic activity affects pigmentation. Genetic and biochemical pathways controlling pigment synthesis are highly conserved among insects (10), which suggests that the D. americana and D. novamexicana tan and ebony genes function similarly to their D. melanogaster orthologs. Consistent with this prediction, the Ebony protein is more abundant in epidermal cells of the yellowish D. novamexicana during late pupal stages than in the darker D. americana (4).

Fig. 3

Ebony and Tan have reciprocal effects on pigmentation development. (A) A simplified melanin biosynthesis pathway is shown. For a more complete pathway, see (6). The gene(s) controlling each enzymatic step are shown in italics. P.O. indicates genes encoding phenol oxidase proteins. Branches with two consecutive arrows include multiple enzymatic steps that are not well defined. (B to F) Dorsal abdominal cuticle (segments A3 to A5) is shown for various D. melanogaster genotypes. (B) Canton-S, a wild-type strain of D. melanogaster, shows the striped dorsal abdominal pigment pattern typical of this species. (C) Expression of UAS-GFP (green) shows that pnr-Gal4, the driver used to ectopically express Ebony and Tan in (D) to (F), activates gene expression in a stripe along the dorsal midline during late pupal development. (D) Ectopic expression of UAS-Ebony caused increased yellow pigmentation. (E) Ectopic expression of UAS-Tan caused increased brown pigmentation. (F) Simultaneous expression of both UAS-Tan and UAS-Ebony resulted in an intermediate phenotype. Cuticle is from 3- to 5-day-old females in all panels except (C), in which cuticle is from a female pupa just before eclosion (stage P15).

Noncoding changes in tan contribute to pigmentation divergence. The above results are consistent with changes in ebony and tan contributing to pigmentation divergence, but they cannot be used to distinguish divergence affecting these genes from divergence affecting linked loci. This is particularly concerning for ebony because it is located in a part of the genome that is inverted between species (4, 11). Inversions effectively suppress recombination, precluding genetic dissection of the region. Nonetheless, differences in Ebony protein expression between D. americana and D. novamexicana (4) strongly suggest that this gene is involved in pigmentation divergence.

Unlike ebony, tan is in a freely recombining region of the genome. This allowed us to use fine-scale genetic mapping to separate the effects of tan from neighboring genes and to determine whether tan contributes to the altered pigmentation observed in the tan introgression line (Fig. 2B). A 2.7-kb region of tan was identified that contributes to pigmentation divergence (fig. S3) and contains 57 single-nucleotide differences and 19 insertions or deletions (indels) (fig. S4). All of these changes affect noncoding sequences, and the region includes the entire first intron (fig. S3). Differences located 3′ of this region must also affect pigmentation, however, because the recombinant fly inheriting D. americana tan sequence only in this region was not as dark as flies inheriting D. americana sequence for the full tan gene (fig. S3). Within tan, this 3′ region includes many noncoding differences as well as two nonsynonymous differences that affect amino acids 190 and 267.

tan expression correlates with pigmentation differences. Given the absence of coding changes in the 2.7-kb mapped region of tan, we expect that divergent sites in this region affect pigmentation by altering tan expression. Because of its darker pigmentation, we hypothesized that D. americana has higher levels of tan expression than D. novamexicana. In situ hybridization showed that tan is expressed throughout each dorsal abdominal segment (“tergite”) in both species during the P14 and P15 pupal stages (12) when pigmentation develops (fig. S5, A and B). This expression pattern correlates with the distribution of pigments in adult D. americana and D. novamexicana tergites, and is distinct from the patterns of tan expression in Drosophila species with other pigment patterns (13). Differences in tan expression detected with in situ hybridization correlate with pigmentation divergence in these other species (13), yet we saw no obvious expression differences between D. americana and D. novamexicana during the same developmental stages with this technique (fig. S5, A and B).

To quantitatively compare levels of tan expression, we measured the relative abundance of tan transcripts in stage P14 and P15 pupae of each species with Pyrosequencing (14). We observed an average of 34% more tan transcripts in D. americana females than in D. novamexicana females (n = 4 samples, each containing six flies; t = 3.7, P = 0.03; t test) (fig. S5, C and D), consistent with the darker pigmentation of D. americana. To determine whether this expression difference results from cis-regulatory divergence of tan, we compared transcript abundance of D. americana and D. novamexicana tan alleles in F1 hybrid females during the same pupal stages with Pyrosequencing (14). Surprisingly, no significant differences in allele-specific expression were observed (n = 5 samples, each containing six flies; t = 0.72, P = 0.51; t test) (fig. S5, E and F). Divergent expression levels may therefore be caused by differences in trans-regulatory factors and/or differences in the number of tan-expressing cells between species (15). The noncoding differences we identified by fine-scale genetic mapping may alter fine-scale temporal control (16) and/or posttranscriptional regulation (17) of tan. It is also possible that these noncoding differences may affect transcriptional regulation of a neighboring gene that is also involved in pigmentation (13) or a cryptic, small, noncoding RNA encoded by the tan intron (18).

Phenotypic consequences of tan divergence revealed in transgenic flies. To determine whether evolutionary changes in the tan gene itself are sufficient to affect pigmentation, we inserted transgenes carrying the D. americana and D. novamexicana tan alleles into the D. melanogaster genome (19). Both transgenes were integrated at the same site, allowing us to compare pigmentation of flies whose genomes differed only for divergent sites within the transgenes. Both the D. americana and D. novamexicana alleles of tan rescued pigmentation in a D. melanogaster tan null mutant (Fig. 4, A to D), indicating that the transgenes were expressed in D. melanogaster and that Tan protein function is (at least largely) conserved. Flies carrying the D. americana tan allele had darker pigmentation than flies carrying the D. novamexicana tan allele (Fig. 4, C and D; F = 26.94, P < 0.0001 for abdominal segments A3 and A4, and F = 6.51, P = 0.03 for the darker A5 segment). This is consistent with the darker pigmentation of D. americana relative to D. novamexicana.

Fig. 4

The D. americana allele of tan causes darker pigmentation than the D. novamexicana allele of tan. Transgenes containing tan alleles from D. americana and D. novamexicana were transformed into D. melanogaster, D. novamexicana, and D. americana. In D. melanogaster, transgenes were crossed into a genetic background homozygous for null mutations in yellow (y-) and tan (t-). The yellow mutation was used to lighten pigmentation, making the effects of tan transgenes easier to see. (A) D. melanogaster yellow (y-) mutant, which is wild-type for tan. (B) D. melanogaster yellow, tan (y-, t-) double mutant. (C) D. melanogaster yellow, tan mutant carrying the D. novamexicana tan transgene (D. nova t+). (D) D. melanogaster yellow, tan mutant carrying the D. americana tan transgene (D. amer t+). (E) Wild-type D. novamexicana carrying the D. novamexicana tan transgene. (F) Wild-type D. novamexicana carrying the D. americana tan transgene.

We also compared the phenotypic effects of D. americana and D. novamexicana tan alleles in D. americana and D. novamexicana themselves by randomly inserting both tan transgenes into the genomes of both species. Two independent insertions were recovered for each transgene in each species. In D. americana, we were unable to detect a difference in pigmentation between transformed and untransformed flies, presumably because of the already dark pigmentation of this species (see Fig. 1B). In D. novamexicana, however, transformant flies carrying the D. americana tan transgene (Fig. 4F) were visibly darker than flies carrying the D. novamexicana tan transgene (Fig. 4E).

ebony and tan QTLs underlie variable pigmentation within D. americana. Pigmentation of D. americana is always distinct from that of D. novamexicana (3), but the intensity of dark pigmentation varies within D. americana. This variation is geographically structured, with D. americana captured in the eastern United States visibly darker than those captured from the western part of the species range (3). These pigmentation differences remain visible after rearing flies under common environmental conditions in the laboratory (Fig. 5A), indicating a genetic basis for the pigmentation cline. Sequence variation within D. americana at putatively neutral loci shows no population structure (2, 2024), which suggests that this cline is due to local adaptation.

Fig. 5

ebony and tan QTLs also contribute to polymorphism. (A) Dorsal abdominal cuticle from D. americana isofemale lines (table S2) is shown. Eastern populations are darker than western populations. (B) Phenotypes, genotypes, and statistical significance for interspecific QTL mapping experiments. Dorsal abdominal pigmentation of each of isofemale line is shown in the color column; the middle columns show the proportion of male backcross progeny genotyped from the lightest (light) and darkest (dark) pigmentation classes with D. americana tan (tanA) and ebony (ebonyA) alleles. P values are from 2×2 Fisher’s exact tests of the genotype count data. (C and D) Neighbor-joining trees of tan [(C), 7625 base pairs] and ebony [(D), 1136 base pairs] from D. americana (black), D. novamexicana (red), and D. virilis (blue) are shown with bootstrap values >75% (n = 1000). Branch lengths are to scale. (E) Fixed differences within the 2.7-kb candidate region of tan. Sites 889 to 981 and 3521 to 3616 are exons 1 and 2, respectively. The D. virilis allele is from the 2005 assembly of the D. virilis genome sequence (29). Positions refer to alignment of GQ457336 through GQ457353. Alleles shared between D. novamexicana and A01 are red; the derived subset, relative to D. virilis, is boxed.

To determine whether sites linked to ebony and/or tan contribute to this intraspecific polymorphism, we used D. novamexicana alleles as a reference to compare the phenotypic effects of ebony and tan QTL alleles among lines of D. americana. Genetic mapping was performed using four isofemale lines of D. americana (A01 collected from Poplar, Montana; DN2, DN4, and DN12 collected from Duncan, Nebraska) with lighter pigmentation than the line of D. americana (A00) used previously. D. americana females from each strain were crossed to D. novamexicana males, F1 hybrid females were backcrossed to D. novamexicana males, and molecular markers in ebony and tan were genotyped in flies from the lightest and darkest pigmentation classes (fig. S6). Despite their lighter pigmentation, the DN12 and DN4 lines of D. americana produced similar mapping results to A00: Both ebony and tan showed highly significant linkages to loci affecting pigmentation (Fig. 5B). Genotyping 101 males from the DN4 backcross population and fitting their genotypes and phenotypes to a linear model (19) showed that ebony (E) and tan (T) both had significant additive effects on pigmentation (FE = 132.98, PE < 0.0001; FT = 160.85, PT < 0.0001) with no significant epistatic interaction between them (F = 3.02, P = 0.09). These additive effects explained 76% of the pigmentation variance in the DN4 backcross population. For DN2, sites linked to tan contributed to pigmentation differences between species, but sites linked to ebony did not (Fig. 5B). The converse was true for A01: Sites linked to ebony contributed to pigmentation differences between species, whereas sites linked to tan did not (Fig. 5B).

Taken together, these data reveal three distinct genotypes among D. americana lines with a light-pigmentation phenotype. DN2 has alleles linked to ebony that appear to be functionally equivalent to those found in D. novamexicana; A01 has alleles linked to tan that appear to be functionally equivalent to those found in D. novamexicana; and DN4 and DN12 have alleles linked to tan and ebony that appear to be functionally distinct from those found in D. novamexicana. It remains to be seen whether the DN12 and DN4 alleles of these QTLs have the same effect on pigmentation as each other or as the alleles from the darker A00 line of D. americana. Two of the three D. americana lines (DN2 and DN4) were collected during the same year and DN12 the following year (table S2), which suggests that genetic heterogeneity for pigmentation exists within this local population. This heterogeneity may be caused by gene flow among populations and/or balancing selection within the population.

Shared pigmentation alleles contribute to polymorphism and divergence. The functional similarity observed for the A01 and D. novamexicana alleles linked to tan, as well as for the DN2 and D. novamexicana alleles linked to ebony, may result from shared ancestry (i.e., alleles that are identical by descent) or from convergent evolution. Sequences of tan and ebony from multiple lines of D. americana and D. novamexicana show that the functional similarity most likely reflects shared ancestry, as the A01 tan sequence is more similar to D. novamexicana alleles than to other D. americana alleles (Fig. 5C) and the DN2 ebony sequence is more similar to D. novamexicana alleles than to other D. americana alleles (Fig. 5D). The DN2 line of D. americana also has the same arrangement of the ebony-containing inversion [“In(2)b” in (11)] as D. novamexicana (fig. S7) (25), further suggesting that pigmentation alleles linked to ebony in DN2 and D. novamexicana have a common origin.

Sequence variation identifies candidate sites for divergent pigmentation. Sequence similarity between A01 and D. novamexicana was found to be highest beginning in the first intron of tan and extending 3′ of tan (fig. S8). Within the 2.7-kb region identified by fine-scale mapping (fig. S3), we observed 13 fixed single-nucleotide differences and two fixed indels between D. americana (excluding A01) and D. novamexicana (Fig. 5E). The A01 allele of D. americana contains the same sequence as D. novamexicana at nine of these 13 divergent sites and shares one of the two indels (Fig. 5E, red). Only four of the shared substitutions are derived changes relative to D. virilis (Fig. 5E, boxed). Because D. virilis has pigmentation similar to D. americana (Fig. 1B), we consider these four noncoding changes to be the best candidates for divergent function in this region. Derived changes outside of this region that are also unique to A01 and D. novamexicana tan may contribute to pigmentation divergence as well. The two amino acid differences between alleles used for fine-scale mapping are polymorphic and are thus unlikely to contribute to fixed differences between species.

A model of pigmentation evolution. Our data reveal the relationship between intraspecific polymorphism and interspecific divergence by showing that the same alleles contribute to pigmentation differences within and between species (fig. S9). These alleles may have been present in the common ancestor of D. americana and D. novamexicana, or they may have arisen in D. novamexicana and subsequently introgressed into D. americana after hybridization. Distinguishing between these two scenarios is notoriously difficult (2628), yet haplotype sharing between D. americana and D. novamexicana has been postulated to be due to shared ancestral variation (2). Our data are consistent with this interpretation: Sequences of D. americana alleles that appear to have the same function as D. novamexicana alleles are basal to these D. novamexicana alleles in gene trees (Fig. 5, C and D); there is evidence of recombination (or gene conversion) within the D. americana A01 tan haplotype (fig. S5A), which argues against a recent introgression event; and D. novamexicana is thought to have evolved from a peripheral population of the common ancestor shared with D. americana (2). Therefore, we propose that light-pigmentation alleles segregating in this common ancestor became fixed in D. novamexicana, contributing to its yellow body color, and continue segregating in D. americana, contributing to clinal variation. Additional D. novamexicana–like alleles of D. americana are needed to further evaluate this model.

Supporting Online Material

Materials and Methods

Figures S1 to S9

Tables S1 to S3


References and Notes

  1. See supporting material on Science Online.
  2. P. Mena, B. McAllister, personal communication.
  3. We thank G. Kalay, X. Heng, M. Weisel, A. Ratnala, and E. Larimore for technical assistance; P. Mena and B. McAllister for sharing unpublished data (including the images shown in fig. S7) and lines of D. americana; C. Dick for use of the Mixer Mill, M. Rebeiz for sharing an unpublished in situ protocol; N. Rosenberg, G. Coop, and other colleagues for discussions of incomplete lineage sorting and introgression; and J. Gruber, J. Coolon, A. Cooley, G. Kalay, D. Yuan, B. McAllister, A. Kopp, B. Prud'homme, and N. Gompel for comments on the manuscript. Supported by NSF grant DEB140640485, the Margaret and Herman Sokol Endowment for Faculty and Graduate Student Research Projects, and the University of Michigan. P.J.W. is an Alfred P. Sloan Research Fellow. DNA sequences are available from GenBank (accession numbers GQ457336 to GQ457453).
View Abstract

Navigate This Article