Drosophila yakuba and D. santomea are sister species that differ in their levels of abdominal pigmentation; D. yakuba shows heavily pigmented posterior abdominal segments in both sexes, whereas D. santomea lacks dark pigment anywhere on its body. Using naturally collected lines, we demonstrate the existence of altitudinal variation in abdominal pigmentation in D. yakuba but not in D. santomea. We use the variation in pigmentation within D. yakuba and two body-color mutants in D. yakuba to elucidate selective advantage of differences in pigmentation. Our results indicate that although differences in abdominal pigmentation have no effect on desiccation resistance, lighter pigmentation confers ultraviolet radiation resistance in this pair of species.

Many sister species of Drosophila show marked differences in abdominal pigmentation (Val 1977; Hollocher et al. 2000; Wilder and Hollocher 2003; Lachaise et al. 2000; Orgogozo and Stern 2009; Wittkopp and Beldade 2009; Wittkopp et al. 2009). In some cases, pigmentation appears to be an adaptive trait that confers thermal regulation (Watt 1969; Gibbs 1998; Gibert et al. 1996; Forsman 2011), resistance to parasitism (Dombeck and Jaenike 2004), protection against ultraviolet radiation (Jacobs 1974; Wang et al. 2008), and/or desiccation resistance (Kalmus 1941a-1941c; Kalmus 1945; Brisson et al. 2005; Parkash et al. 2008, 2009). In general, darkly pigmented Drosophila species have been found to be more resistant to water stress (Brisson et al. 2005; Parkash et al. 2008, 2011; Pool and Aquadro 2007; Gibbs et al. 2009; Singh et al. 2009; Gibbs and Rajpurohit 2010). However, the correlation between dark-pigmentation and desiccation resistance is not uniform across the entire genus Drosophila (Wittkopp et al. 2003, 2011), In other cases, pigmentation may have evolved as a response to sexual selection (i.e., mating preferences, Kopp et al. 2000; Wittkopp et al. 2003; Parkash et al. 2011), or simply by neutral genetic drift (David et al. 1985). The evolutionary forces affecting pigmentation in Drosophila are varied, and it is unclear whether particular intra- and interspecific pigmentation differences are caused by neutral divergence, sexual selection, and/or natural selection (da Cunha 1949; Machado et al. 2001; Wittkopp et al. 2003, 2011; Clusella-Trullas and Terblanche 2011). In particular, the adaptive value of pigmentation within the Drosophila melanogaster subgroup remains unresolved.

Sister species D. santomea and D. yakuba show a striking difference in pigmentation (Lachaise et al. 2000; Llopart et al. 2002; Carbone et al. 2005; Jeong et al. 2006; Rebeiz et al. 2009). In D. yakuba (yak), as well as four of the other eight species in the melanogaster species subgroup, the posterior abdominal segments of both sexes are heavily pigmented. Additionally, D. orena and D. teissieri females mostly lack pigmentation but males have dark abdomens. In contrast, D. santomea (san), a species endemic to the island of São Tomé (a volcanic island off the west coast of Africa) is unique in that both sexes completely lack dark pigment anywhere on their bodies (Llopart et al. 2002; Carbone et al. 2005).

The study of the genetic basis of pigmentation has made substantial progress, and it has been possible to identify some of the genes that cause interspecific pigmentation differences (including at least two loci in the D. santomea/D. yakuba species pair; Brisson et al. 2004; Jeong et al. 2006; Gibert et al. 2007; Ng et al. 2008; Rebeiz et al. 2009; Werner et al. 2010; Wittkopp et al. 2011). However, the fitness consequences of differences in abdominal pigmentation in the melanogaster group of species have not been fully identified. Here, we report geographical variation in abdominal pigmentation in D. yakuba and use pigmentation mutants in D. yakuba to test two hypotheses that have been proposed to explain differences in pigmentation levels: the possibilities that abdominal pigmentation confers differences in desiccation resistance (Wittkopp et al. 2003; Brisson et al. 2005) and that abdominal pigmentation levels are involved in resistance to ultraviolet radiation (Jacobs 1974; Wittkopp et al. 2003; Parkash et al. 2011).

Materials and Methods


We collected females and established isofemale lines (i.e., the progeny of a singly collected gravid female) from two different islands off the coast of the Gulf of Guinea (São Tomé Island and Bioko). To study whether the altitude at which each line was collected influences abdominal pigmentation, we studied the pigmentation of 39 isofemale lines per transect collected at different altitudes. In the case of D. yakuba, we included lines from two different altitudinal transects, São Tomé and Bioko, whereas for D. santomea, we included only one (São Tomé, where the species is endemic). The two transects spanned similar altitude ranges (0 to approximately 1500 m above the sea level). São Tomé is a volcanic island approximately 400 km off Gabón and reaches an altitude of 2024 m at its highest point. Bioko (Equatorial Guinea) is also a volcanic island, and is 32 km off Cameroon; its highest point, Pico Basile, reaches 3012 m. Drosophila yakuba ebony and yellow mutants were isolated as spontaneous appearances in wild cultures. Isofemale lines have been kept in laboratory conditions with at least three replicates (i.e., vials with at least 35 females and 35 males) since the establishment of the isofemale lines (July 2009). Pigmentation and desiccation measurements were performed upon collection (within five generations of establishing the isofemale line) and after 80 generations after collection.


Individuals were collected as virgins and kept in groups of 20 until they were four days old. To estimate the intensity of pigmentation on whole flies, we used a 0 to 4 point visual scale ranging from 0 (unpigmented areas) to 4 (dark and shiny black areas), with intermediate numbers representing intermediate degrees of pigmentation (David et al. 1990; Carbone et al. 2005). We also measured the proportion of the area of each tergite that was pigmented (estimated in 10 increments per tergite). To obtain the overall pigmentation score of each individual, we multiplied the percentage of the area of each tergite covered by each shade of pigmentation by the intensity of that pigmentation, and then summed these values across all three tergites (A4, A5, and A6; Carbone et al., 2005). Total pigmentation scores thus ranged between 0 (completely unpigmented) and 1200 (fully pigmented). All the scoring was done within 10 generations of establishing the isofemale lines. The scoring was done blindly; that is, the scorer did not know either the species or genotype.


To study the possible influence of abdominal pigmentation on desiccation and ultraviolet resistance (protocols described below) while controlling for differences caused by the genetic background, we introgressed the two D. yakuba mutants (ebony-like and yellow) into a common background (Täi18) by repeated backcrossing (10 generations, Wittkopp et al. 2009). Ebony-like is semidominant; to introgress it into a Täi18 background, we crossed virgin Täi18 females to males of the mutant stock and selected the progeny with the darkest abdominal pigmentation. We repeated this scheme for 10 generations (i.e., each generation we collected males who showed dark pigmentation and crossed them to virgin Täi18 females). To select for the introgression of yellow, an X-linked allele, we collected males that showed the yellow phenotype and crossed them to Täi18 females also for 10 generations. In all crosses, we mated three virgin Täi18 females (from the Täi18 stock) to five males from the introgression crosses. After 10 generations, we collected virgin females and males from the introgression crosses and performed brother–sister matings for five generations to isogenize the mutants. To control for spontaneous mutation, which can also give rise to darker or lighter flies, we crossed males from the introgressed mutant stocks to homozygous mutant females and scored the pigmentation of the mutants assuming that if the correct mutations were introgressed, then the phenotypes (either ebony-like or yellow) would not be complemented (data not shown).

Desiccation resistance

Desiccation resistance was measured by placing 20 four-day-old virgin females or males in empty 8-dram vials, which in turn were placed in a glass desiccator with 200 g of Drierite and kept at 21°C (Hoffman 1991; Hoffman and Harshman 1999). The relative humidity was kept under 20% and was measured with a hygrometer. Flies were checked hourly and the time of death recorded for each fly. To measure differences in desiccation resistance at higher temperatures, we followed the same protocol but desiccators were kept at 24°C or 28°C, respectively.

Differences between genotypes for each of the species in desiccation resistance were analyzed by fitting a nested analysis of variance (ANOVA; Sokal and Rohlf 2012) for the median time to death in each vial with the main effects of genotype (the number of isofemale lines or mutants studied for each species), line (nested within genotype), sex, temperature, and the interactions between these factors. We took the median longevity (in hours) per vial as the response in the model.

Ultraviolet resistance

We collected virgin males and females under CO2 anesthesia and kept them for three days in single-sex groups of 20 flies. On day four, flies of both sexes were lightly anesthetized and irradiated at different ultraviolet-B light dosages (seven different intensities; 100, 300, 500, 1000, 3000, 5000, and 6000 Joules/m2) using a ultraviolet Stratalinker 2400 (Stratagene, La Jolla, CA) according to the method proposed by Aguilar-Fuentes et al. (2008). The flies were scored for pigmentation and were transferred into fresh food-containing vials in groups of 20 individuals. Every vial was scored once daily to determine how many flies were still alive, with flies transferred to fresh food as needed (usually every two to three days). The experiment continued until all flies had died.

Heritability measurements

We collected virgin females and males from 20 D. yakuba lines (10 from Bioko and 10 from Sao Tomé) after 94 generations of starting the isofemale lines. On day four after eclosion, each single female was mated to a single male and the progeny of each cross was raised by standard fly husbandry methods. All progeny was collected upon eclosion and housed in sex-specific vials for four days. Twenty females and 20 males per cross were scored for pigmentation (see above). For each isofemale line, we scored 20 crosses (replicates). Because the mean and variance of abdominal pigmentation differs between sexes, we calculated the heritability of the trait for each sex. We calculated the slope of the regression of the mean phenotypic trait value of the progeny (either daughters or sons) to the mean value of the parent (mother or father, respectively). Sex-specific heritability was calculated as twice the slope of each regression (Falconer 1960).



To study the geographical variation in abdominal pigmentation in D. yakuba and D. santomea, we studied 39 lines of D. santomea and 78 lines of D. yakuba collected at different altitudes. In the case of D. yakuba, we included lines from two different altitudinal transects (São Tomé Island and Bioko Island, n = 39 for each transect), whereas for D. santomea, we included one transect (São Tomé Island, n = 39; this species is not found on Bioko). The two D. yakuba transects spanned similar altitude ranges (0 meters above the sea level to approximately 1500 m). To test for differences in pigmentation among collection sites, we fitted a nested ANOVA (sites nested within altitude, Sokal and Rohlf 2012) for each sex for each species (there were four different ANOVAs). In the case of D. yakuba, the ANOVAs showed that there is heterogeneity in the levels of abdominal pigmentation in both sexes and in both transects (Table 1). In the four cases, levels of abdominal pigmentation showed a positive relationship with the altitude at which individuals were collected (all regression coefficients were significantly higher than 0—P < 1 × 10−10—but did not differ among themselves, Table 1). The trend did not change when the two independent transects for D. yakuba were analyzed jointly. These results indicate that although the variation in pigmentation in D. yakuba is subtle, lines collected at low altitudes were generally lighter than those collected at high altitudes (Figs. 1 and 2, panels A–F, Table 1). In D. santomea, in contrast, we found that altitude had no effect in the abdominal pigmentation levels of females or males (Table 1; Fig. S1).

Table 1. Abdominal pigmentation varies within Drosophila yakuba depending on the altitude at which flies are collected, but not within D. santomea. The abdominal pigmentation cline disappears after 80 generations of keeping the lines in laboratory conditions. Estimates of the significance of the heterogeneity in abdominal pigmentation were assessed with a nested ANOVA. We also calculated the regression coefficient of abdominal pigmentation to altitude
  Upon collectionAfter 80 generations
  F12,1911 F26,1911  F12,1911 F26,1911  
D. yakubaBioko females52.833<1 × 10−1027.548<1 × 10−100.6846.543<1 × 10−101.0730.3650.053
 Bioko males165.149<1 × 10−1092.708<1 × 10−100.7065.5502.03×10−91.3200.1300.050
 São Tomé females159.641<1 × 10−1085.254<1 × 10−100.4142.5242.67 × 10−31.0210.4340.029
 São Tomé males195.692<1 × 10−1086.338<1 × 10−100.6974.8416.76 × 10−81.4610.0630.048
 Joint females95.078<1 × 10−1045.546<1 × 10−100.5383.6122.49 × 10−80.9920.4910.033
 Joint males198.690<1 × 10−1091.600<1 × 10−100.7056.187<1 × 10−101.5040.0110.053
D. santomeaSão Tomé females0.7590.66831.2850.1460.0220.7590.66831.2850.1460.022
 São Tomé males0.5720.838381.4040.0780.0231.7080.07351.3890.0850.028
Figure 1.

Drosophila yakuba shows heterogeneity in the levels of abdominal pigmentation in both sexes and in both transects (São Tomé and Bioko). The altitude at which the isofemale lines were collected (measured in meters above the sea level) is a significant factor in explaining the among-line heterogeneity (see text for details) and shows a positive association with altitude. Red: D. yakuba females from São Tomé; blue: D. yakuba males from São Tomé; dark red: D. yakuba females from Bioko; dark blue: D. yakuba males from São Tomé. The legend shows the regression coefficient for each of the four datasets.

Figure 2.

Pigmentation varies in the altitudinal transect in Pico de São Tomé. The pigmentation levels of three isofemale lines collected at different altitudes (A) female, (B) male, isofemale 17.1: collected at 0 m. (C) female, (D) male, isofemale 5.71: collected at 550 m. (E) female, (F) male, isofemale 1.23: collected at 1200 m.


Associations between pigmentation and environmental humidity and ultraviolet exposure have been reported in several pairs of Drosophila species. In some previous reports, flies with high levels of melanization have been found in high altitude, cool habitats (Munjal et al. 1997; Wittkopp et al. 2003; Pool and Aquadro 2007). In other cases, darker flies have been collected in warmer and dryer environments (Brisson et al. 2005). Given that D. yakuba is a species that is mainly found in open habitats and at low altitudes (Lachaise et al. 2000; Matute et al. 2009), we predicted that abdominal melanization played a protective role against ultraviolet irradiation and/or desiccation in nature. If so, dark individuals of D. yakuba should exhibit lower death rates than light individuals of the same species when exposed to high levels of ultraviolet or to desiccation conditions. We performed ultraviolet resistance assays in the set of isofemale lines of D. yakuba that showed the most extreme pigmentation values (both low and high; n = 12 isofemale lines). Figure S2 shows the dose or response curve for each of the 12 D. yakuba isofemale lines. We measured the correlation between the harmonic mean of longevities at all ultraviolet intensities of each line and the median abdominal pigmentation. To our surprise, we detected a negative correlation between abdominal pigmentation and longevity after ultraviolet exposure (r = −0.810, P < 0.001). When the data were pooled together, we detected a significant negative correlation between pigmentation and resistance to ultraviolet at all intensities tested, but found no difference in survival for flies at both pigmentation extremes that were not exposed to ultraviolet (Table 2). (No significant correlation was detected between longevity and pigmentation when flies were not exposed to ultraviolet: r = −0.174, P = 0.588.) These results suggest that when exposed to ultraviolet irradiation, lighter D. yakuba flies have a longer survival time than darker flies, which constitutes a possible advantage in open environments where the lighter (and not darker) flies live, and where ultraviolet might be a substantial selective pressure.

Table 2. Pearson's correlation coefficients between abdominal pigmentation and ultraviolet resistance at nine different intensities of ultraviolet irradiation in Drosophila yakuba
Spearman's correlation (abdominal pigmentation and ultraviolet resistance)
300−0.4050.014− 0.563<0.001
Not exposed−0.3170.110−0.1760.379

We also measured desiccation resistance in a set of isofemale lines (n = 26) from D. yakuba that showed differences in pigmentation to determine if there was any correlation between these two traits. The results from these analyses indicate that there is no discernable correlation between pigmentation and desiccation resistance (females: r = −0.129, P = 0.260; males: r = 0.055, P = 0.633; Spearman's correlation test) in D. yakuba, suggesting that differences in pigmentation do not confer substantial protection against desiccating conditions.


We isolated two pigmentation mutants that spontaneously appeared in wild cultures of D. yakuba: yellow (y) and ebony-like (yaky and yakebony-like, respectively; Fig. S3). yaky mutants were lighter than wild-type stocks, but the effect was subtle (mean pigmentation yaky: 423.95; mean pigmentation D. yakuba Täi18: 496.95; two-way ANOVA with sex and genotype as fixed effects followed by an honestly significant difference (HSD) Tukey test, P = 0.022; Fig. S1). yakebony-like, on the other hand, was significantly darker than D. yakuba wild-type individuals from the Täi18 line (mean pigmentation D.yakubaebony-like: 657.50; two-way ANOVA with sex and genotype as fixed effects followed by a HSD Tukey test, P < 0.001). These two pigmentation mutants allowed us to test a second prediction: if abdominal pigmentation affects ultraviolet resistance in D. yakuba, then the increase (or reduction) of abdominal pigmentation by a single locus and in a controlled genetic background should lead to a change in ultraviolet resistance. To test this hypothesis, we introgressed the two D. yakuba mutants (yellow and ebony-like) into a D. yakuba Täi18 background; introgression into an isofemale line allowed us to test the effects of the mutations in otherwise healthy flies, as mutant stocks are usually highly inbred and show poor fitness. The introgression scheme we followed was to take advantage of the fact that ebony-like is semidominant and could be selected based on the pigmentation phenotype; the introgression of yellow was selected by isolating F1 males, as yellow is X-linked (Wittkopp et al. 2009).

We compared the longevity of mutant stocks from D. yakuba (introgressed into a Täi18 background) with that of Täi18 after irradiation with different doses of ultraviolet. Our hypothesis was that the yellow mutant allele would confer a protective effect. To analyze the data, we fitted a full-factorial ANOVA in which ultraviolet resistance (i.e., longevity after radiation measured in days) was the response and ultraviolet intensity, sex, and genotype (yellow, ebony-like or wild-type) were the fixed effects. This ANOVA showed that there are significant differences in the ultraviolet resistance of these mutants (genotype effect: F2,132 = 11.33, P = 2.872 × 10−5), with yellow flies indeed showing greater longevity than Täi18 flies (P = 9.59 × 10−3, Tukey HSD test) but no difference in ultraviolet resistance between ebony-like and Täi18 flies (P = 0.912, Tukey HSD test). Ultraviolet intensity (F1,132 = 82.389, P = 1.387 × 10−16), and the interaction between genotype and ultraviolet intensity (F2,132 = 9.16, P = 1.89 × 10−4) also were significant causes of heterogeneity in longevity after ultraviolet irradiation.


We compared the ultraviolet resistance of isofemale (wild-type) lines of D. yakuba (n = 12 isofemale lines, six per island) and D. santomea (n = 12 isofemale lines from São Tomé). We fitted a full factorial linear model with ultraviolet resistance as the response; species, ultraviolet intensity, sex, and line as fixed effects. This model showed that species (F1, 892 = 53.697, P < 1 × 10−4), intensity (F8,892 = 242.233, P < 1 × 10−4), sex (F1,892 = 35.598, P < 1 × 10−4), and line (nested within species, F1,892 = 9.890, P < 1 × 10−4) are significant factors that affect the longevity of individuals once irradiated with ultraviolet (unsurprisingly, ultraviolet intensity also has a significant effect). When all the radiationxbrk intensities are pooled together, radiation reduced adult longevity by 63.5% in D. santomea and by 68.3% in D. yakuba. These results suggest that there is substantial variation within species in terms of ultraviolet resistance, but also that D. santomea is more resistant to ultraviolet irradiation than D. yakuba, which is consistent with the hypothesis that the lighter pigmentation of D. santomea might confer a protective effect against ultraviolet radiation, just as lighter pigmentation within D. yakuba does.

We also measured desiccation resistance in isofemale lines from both species (39 D. yakuba lines and 39 D. santomea lines) to determine whether either species was more resistant to desiccation at different temperatures. We fitted a fully factorial model with species, temperature, and line (nested within species) as the fixed effects, and desiccation resistance as the response. All four effects were significant (all effects P < 1 × 10−15), suggesting variation in desiccation resistance between and within species. Contrary to our expectations based on pigmentation and altitudinal distributions, D. santomea is more resistant to desiccation than D. yakuba at 21°C (time to desiccation in D. santomea females: 9.95 h vs in D. yakuba 9.15 h; P < 0.001) and 24°C (average time to desiccation in D. santomea females: 9.79 h vs D. yakuba 8.99 h; P < 0.001). At 28°C, desiccation resistance showed no differences between species (average time to desiccation in D. santomea females: 9.41 h vs D. yakuba 9.46 h; P = 0.81). This last result, however, might be an artifact caused by the low fitness of D. santomea at high temperatures (Matute et al. 2009). Finally, we compared the desiccation resistance of the introgressed D. yakuba ebony-like and yellow mutants to wild-type stocks with the same genetic background. We fitted a one-way ANOVA in which the median desiccation resistance of each replicate was the response and genotype was the only effect. These comparisons showed that neither pigmentation mutant changes the desiccation resistance in D. yakuba (F2,297 = 2.330, P = 0.213). Although D. santomea is slightly more resistant to desiccation conditions than D. yakuba, pigmentation mutants within D. yakuba show the same levels of desiccation resistance than wild-type flies. Therefore, we find no evidence that there is a causal relationship between the lighter pigmentation of D. santomea and its greater desiccation resistance.


We next explored how pigmentation changed over time in laboratory conditions in isofemale lines. Light pigmentation seems to be a costly trait in laboratory conditions. Eighty generations after being collected, the isofemale lines showed more subtle differences in pigmentation between collection sites, and the differences within collection sites are no longer significant for either sex (Table 1, Figs. S4 and S5). The results are similar when the two transects are analyzed jointly (Table 1). After 80 generations of laboratory conditions, the correlation coefficients between the altitude at which lines were collected and the pigmentation level were significantly lower than upon collection for both sexes (Fisher's transformation applied to both correlation coefficients, all P-values < 0.0001).

In an effort to characterize the amount of genetic variance for pigmentation within isofemale lines, we calculated heritability of abdominal pigmentation at generation 94 for females and males independently (Table S1). Although these comparisons do not fully reveal the initial amount of genetic variance in these lines, they do provide a picture of the amount of genetic variance remaining in the isofemale lines after being in laboratory conditions for 94 generations. All the estimates of heritability for 20 isofemale lines are shown in Table S1 and Figures S6 and S7. The heritability estimates ranged from 0% to up to 33% depending on the isofemale line and the sex, but heritability did not differ between sexes (paired Wilcoxon signed-rank test, V = 87, P-value = 0.522). These results suggest that there are significant differences in the heritability of pigmentation among lines and thus suggests that there are significant differences in the amount of genetic variance remaining in these lines, which might partially explain the disappearance of the cline.

We also measured the ultraviolet resistance of the 12 isofemales lines we initially characterized but only at 500 J/m2 (see above). The results indicate that after 80 generations of being in the laboratory, there is still significant heterogeneity in the ultraviolet resistance levels in males (F11,24 = 6.218 P = 9.432 × 10−5), but not in females (F11,24 = 1.988, P = 0.077). In both cases, ultraviolet resistance is still negatively correlated with abdominal pigmentation even in spite of the evolution of the two traits (males: r = −0.392, P = 0.018; females: r = −0.334, P = 0.046; Spearman's correlation test). These results suggest that pigmentation, like many other characters, changes under laboratory conditions (e.g., Sgró and Partridge 2000; Hoffmann et al. 2001; Frankino and Myers 2012, among many others).


This study describes how abdominal pigmentation levels in D. yakuba and D. santomea are associated with differences in resistance to ultraviolet radiation. These results demonstrate that ultraviolet radiation can significantly diminish the longevity of flies from these species, but indicate that not all species are affected in the same way. Drosophila santomea is more resistant to ultraviolet radiation and desiccation than D. yakuba, although its abdomen is lighter. We find these results to be surprising for two reasons. According to results from other species, we expected darker flies to be more resistant to water stress conditions and to environments in which ultraviolet exposure is a common and limiting factor. In D. melanogaster, for example, lighter flies are generally more susceptible to desiccation and ultraviolet damage than darker flies (Parsons 1979; Davidson, 1990; Pool and Aquadro 2007 but see Gibert et al. 1998). Second, D. santomea is found in the rainforest (i.e., high altitudes of São Tomé) where humidity and sheltering are higher than in the more open habitats where D. yakuba is commonly found (Lachaise et al. 2000; Llopart et al. 2005; descriptions of the island habitats can be found in Exell 1944 and Figueiredo and Gascoigne 2001). Our observation that D. santomea is slightly more resistant to desiccation than D. yakuba contradict the generality of dark species being more desiccation resistant than light species.

Interspecific differences in desiccation resistance have been studied in multiple species of Drosophila (Stanley et al. 1980; Gibbs and Matzkin 2001; Gibbs 2002; Gibbs et al. 2003; Hoffmann 2010; Kellerman et al. 2012). Testing six of the nine species in this group (not including D. santomea), Stanley et al. (1980) found that D. yakuba and D. erecta were most sensitive to cold, heat, and desiccation stress for both survivorship and male sterility. On the other hand, the cosmopolitan species D. melanogaster and D. simulans had high fitness in stress conditions, implying that invasive species may expand their range of temperature tolerance as they expand their habitat (Stanley et al. 1980; Gibbs and Matzkin 2001; Hoffman 2010; Mitchell and Hoffman 2010). Kellerman et al. (2012) found that D. yakuba was slightly more resistant to desiccation than D. santomea. Their experiments, however, aimed to cover a broad phylogenetic spectrum and had small sample sizes per species. Our data indicate that D. yakuba, a species that is commonly found in open habitats including semiarid areas, savannas, montane grassland, and semidomestic habitats (notably coffee and cacao plantations) does not show an expanded desiccation resistance range even when it is compared to D. santomea, a species endemic and restricted to humid habitats.

It is worth noting that other species-specific factors besides levels of melanization can influence the rate of water loss (e.g., kind of cuticular hydrocarbons, and the cellular concentration of glycogen and lipids; Gibbs et al. 1997; Gibbs 1998; Albers and Bradley 2006; Blomquist and Bagneres 2010). All these factors remain to be measured within and across species. There are two additional factors that remain unknown. First, the environmental conditions of São Tomé have not been fully characterized. A major question in the future will be to connect our results with measurements of ambient humidity and incidence of solar radiation in different habitats of São Tomé. This connection will reveal whether these interspecific differences are truly adaptive. Second, it remains unknown what degree of the phenotypic variance observed in pigmentation in D. yakuba is caused by genetic differences and what portion is caused by plasticity, especially given that abdominal pigmentation is highly responsive to the environment, as the adaptation of D. yakuba to laboratory conditions demonstrates (see also David et al. 1990; Gibert et al. 1996; Gibert et al. 2007). Heritability of pigmentation within each line varies (Table S1), but may also be affected by laboratory adaptation and loss of the clinal variation. Heritability measurements for pigmentation of newly collected strains from São Tomé and Bioko will be a high priority in the future.

The nature of potentially adaptive interspecific differences can be addressed by the study of physiological stress. Ultraviolet resistance and desiccation resistance are traits that show interspecific differences between D. santomea and D. yakuba and should be amenable to genetic dissection. However, caution will be needed as both abdominal pigmentation and ultraviolet resistance show rapid changes in laboratory conditions. The genetic basis of ultraviolet resistance in Drosophila has been previously studied, and several genes have been found to have a major influence on the trait (Boyd and Harris 1987; Yildiz et al. 2004; Aguilar-Fuentes et al. 2008). phr, Dmp8/TTDA, and mei-9 mutants have a strong photo-repair activity suggesting that these genes are involved in ultraviolet resistance (Boyd and Harris 1987; Yildiz et al. 2004; Aguilar-Fuentes et al. 2008). Whether these genes are involved in the interspecific difference between D. yakuba and D. santomea is a possibility that remains to be explored.


The authors would like to thank J. Ayroles, K. L. M. Gordon, E. Hungate, B. Krinsky, R. Mallarino, M. F. Przeworski, E. Scordato, and S. Restrepo for scientific discussions and comments at every stage of the manuscript. We would also like to thank I. A. Butler, J. Gladstone, and M. Ford for their technical assistance. Finally, we would like to thank the Bioko Biodiversity Protection Program and the Ministry of Environment, Republic of São Tomé and Príncipe for permission to collect and export specimens for study. DRM is funded by a Chicago Fellowship.