Genetic Screens for Factors Involved in the Notum Bristle Loss of Interspecific Hybrids Between Drosophila melanogaster and D. simulans

Corresponding author: Toshiyuki Takano-Shimizu

Communicating editor: Z-B. ZENG

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Interspecific cross is a powerful means to uncover hidden within- and between-species variation in populations. One example is a bristle loss phenotype of hybrids between Drosophila melanogaster and D. simulans, although both the pure species have exactly the same pattern of bristle formation on the notum. There exists a large amount of genetic variability in the simulans populations with respect to the number of missing bristles in hybrids, and the variation is largely attributable to simulans X chromosomes. Using nine molecular markers, I screened the simulans X chromosome for genetic factors that were responsible for the differences between a pair of simulans lines with high (H) and low (L) missing bristle numbers. Together with duplication-rescue experiments, a single major quantitative locus was mapped to a 13F–14F region. Importantly, this region accounted for most of the differences between H and L lines in three other independent pairs, suggesting segregation of H and L alleles at the single locus in different populations. Moreover, a deficiency screening uncovered several regions with factors that potentially cause the hybrid bristle loss due to epistatic interactions with the other factors.

ARTIFICIAL selection for quantitative characters changes the phenotype far beyond the range of variation in the original base population in most cases (FALCONER 1960 ). Environmental stresses (phenocopy, WADDINGTON 1953 ; GIBSON and HOGNESS 1996 ) and mutant backgrounds (RENDEL 1959 ; RUTHERFORD and LINDQUIST 1998 ) also uncover a surprising amount of hidden variation, and selection for abnormal phenotypic characters in such conditions increases the frequencies of affected individuals. What is more, after several generations of selection, anomalous characters continue to be expressed without environmental stresses or in wild-type background conditions (WADDINGTON 1953 ; RUTHERFORD and LINDQUIST 1998 ). Thus, cryptic variation and selection may provide a scenario for discontinuous evolutionary changes in morphological characters.

Cryptic variation exists probably because of stabilizing selection and genetic buffering ability. Although these two factors can act independently, stabilizing selection may also contribute to development of genetic buffering ability. The ability to resist genetic and environmental perturbation is known as developmental canalization (WADDINGTON 1942 ). Canalization, in turn, largely depends on functional redundancy in several forms. Multiple regulatory sites in a gene or multiple genes may exert the same function individually, or cells may substrate for some others (e.g., developmental regulation and equivalence groups). Interactions among molecules, genes, and cells are clearly very common. While we are deepening our understanding of interactions and genetic networks (for Drosophila, SANCHEZ et al. 1999 ), the mechanisms underlying observable and hidden variation in populations and their roles in adaptive evolution and in morphological evolution are old issues that are still poorly understood.

Interspecific-hybrid analysis is another tool to reveal hidden between-species differences among closely related species. In Drosophila, postmating reproductive isolation, namely, sterility and inviability, has been an intensive focus of various studies (e.g., DOBZHANSKY 1970 for review of early works; WATANABE 1979 ; WU and BECKENBACH 1983 ; COYNE 1984 ; HUTTER et al. 1990 ; ORR 1992 ; PANTAZIDIS et al. 1993 ; PEREZ et al. 1993 ; TRUE et al. 1996B ). Morphological anomalies, the focus of this study, are also manifested in many hybrids (e.g., STURTEVANT 1920 ; WEISBROT 1963 ; DOBZHANSKY 1970 ; COYNE 1985 ; KHADEM and KRIMBAS 1991 ; PAPACEIT et al. 1991 ). One example is a bristle loss phenotype in the Drosophila melanogaster-D. simulans hybrids, although both the pure species have exactly the same pattern of bristle formation on the notum (STURTEVANT 1920 ; BIDDLE 1932 ; TAKANO 1998 ). This means that the genetic architecture of bristle formation can change in local populations without any obvious phenotypic alternation.

TAKANO 1998 found a large amount of genetic variability in the simulans populations with respect to the missing bristle numbers and significant simulans X chromosome effects in hybrids with melanogaster. I present here results of the screens of the simulans X chromosome for genetic factors affecting the bristle phenotype in melanogaster-simulans hybrids. Hybrid bristle loss anomaly is potentially caused by several distinct classes of genetic factors: (1) factors responsible for the within-simulans variation in the number of missing bristles, (2) fixed melanogaster-simulans differences involved in the hybrid bristle loss, and (3) factors that show dose-sensitive effects, especially in a hybrid background, without a between-species functional difference. Because the pattern of notum bristles is fixed within each species and identical between species, these factors do not cause the anomaly in the original pure species—only the combination of the factors in hybrids does. Thus, we have to make interspecific hybrids to screen for genetic factors responsible for the bristle loss phenotype. On the other hand, both male and female hybrids between the two species are completely sterile (but see DAVIS et al. 1996 for an exception), allowing only F₁ hybrid screening. In this article, I used two screening methods: (i) quantitative trait loci (QTL) mapping of X chromosome factors affecting the hybrid bristle number by using a pair of simulans lines with high (H) and low (L) missing bristle numbers and (ii) deficiency screening. The former is one method for identifying class 1 factors; the second method is based on phenotypic effects in melanogaster deficiency-carrying hybrids. It has the potential to identify factors of the three classes.

This QTL mapping, together with duplication rescue experiments, recovered a single major X chromosome QTL (or block of QTL). It is important to note that most differences between H and L simulans lines were attributed to the region including this QTL in three independent pairs of H and L lines. This suggests segregation of H and L alleles (but not necessarily the same alleles) at the same locus in different populations. On the other hand, a deficiency screening uncovered several regions with factors that potentially cause the hybrid bristle loss due to epistatic interactions with the other factors (class 2 or 3 factors).

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila strains:
Inbred and isofemale D. simulans lines used in this study are given in Table 1. I classified these lines into high missing bristle (H) lines and low missing bristle (L) lines on the basis of the numbers of missing bristles in the interspecific F₁ male hybrids with C(1)RM, y w^a [Basc/C(1)RM, y w^a, TT-35; TAKANO 1998] females of D. melanogaster. In the deficiency-screening experiments, I also used one D. mauritiana inbred strain, Mau-g76 (G20; provided by the Genetic Strain Research Center, National Institute of Genetics, Mishima, Japan), which had no bristle anomaly in the hybrids with TT-35 melanogaster females (0.1 ± 0.1).

DNA markers on the X chromosome:
Single-strand conformation polymorphism (SSCP; ORITA et al. 1989 ) and length polymorphism markers on the simulans X chromosome were developed at the nine loci including three Berkeley Drosophila Genome Project sequence tagged sites (STSs): shaggy (sgg; 3B1, its genetic map position in D. melanogaster is 1.3 from FLYBASE 1999 ), no receptor potential A (norpA; 4B6, 6.5), spaghetti squash (sqh; 5D6, 13), G01498 [designed from a published sequence of accession no. G01498 in the GenBank, European Molecular Biology Laboratory, and DNA Data Bank of Japan sequence databases (STS name = Dm1977), 8A1-2], Protein phosphatase 1ß at 9 C (Pp1ß–9C; 9C1-2, 31), Dm1865 (10F2-11), Dm0478 (13A8-9), forked (f; 15F2-4, 56.7), and runt (run; 19E2, 65). Table 2 summarizes the primer sequences and the genotype-determining methods. The length polymorphisms in sgg and norpA were analyzed on 3% MetaPhor agarose gels (FMC, Rockland, ME) and that in sqh on 5–20% gradient polyacrylamide gels. SSCP analyses were done as in TAKANO-SHIMIZU 1999 .

Experimental design to study X chromosome effects (F₁ experiment):
Reciprocal G₀ crosses between Sim-5 (G20; abbreviated to S5) and Tananarive (G20; Ta) were done to produce F₁ males carrying the different X chromosomes but, on average, the same autosomes (S5 x Ta and Ta x S5; throughout this article, the first strain indicates the female parent). The interspecific crosses were made between about 20 pairs of TT-35 females of D. melanogaster and the above F₁ males as well as between TT-35 females and males of Sim-5 (G20) and Tananarive (G20) with six replications each. All parental flies were transferred to new vials after 3 days. This was done twice, and the transfer set was cleared after another 3 days (on the 9th day of cross). In sum, each parental set was allowed to lay eggs in three different vials. Five male progeny were sampled from each vial, making a total sample size of 15 males (5 males x 3 vials) per cross. The number of missing bristles for these hybrid males was studied for 13 pairs of macrochaetae on the notum and humeri as described in TAKANO 1998 (Fig 1).

View larger version (13K): In this window In a new window Download PPT slide   Figure 1. Diagram shows the macrochaete positions on a heminotum and humerus with their nomenclature. PS, presutural; uHU and lHU, humerals; aNP and pNP, notopleurals; aSA and pSA, supraalars; aPA and pPA, post-alars; aDC and pDC, dorsocentrals; and aSC and pSC, scutellars.

The same experiment was done simultaneously for a pair of T-6 (G20) and Sim-11 (G20) and a pair of A-1 (G20) and Tananarive (G20) with two replications for each of simulans parental males and with four replications for F₁ male progeny of each cross, except for one replication for Tananarive (G20) males in this experimental set and three replications for F₁ males of Sim-11 (G20) () x T-6 (G20) (). Because of a small number of replications, I made an effort to examine 10 males from each vial.

As mentioned above, each parental set was allowed to lay eggs in three different vials. Before pooling the data from different vials, a two-way analysis of variance for each male parent [e.g., individually for Sim-5 (G20) males, Tananarive (G20) males, the F₁ males of S5 x Ta cross, and the F₁ males of Ta x S5 cross] was done for the number of missing bristles of hybrids in a mixed model. When there were missing data, the sample sizes were reduced in such a way as to have an equal sample size for all subclasses. The model for the analysis is Y_ijk = µ + C_i + V_j + (CV)_ij + _k(ij), where C_i is the random effect of the ith cross (replication), V_j is the fixed effect of the jth vial, (CV)_ij is the cross-by-vial interaction, and _k(ij) is the residual. Only 3 of 33 tests [the cross-effect in F₁ males of Sim-5 (G20) () x Tananarive (G20) () (P < 0.05; this added variance component is 12% of the overall variation), the cross-effect (P < 0.05; 14% of the overall variation) in Sim-11 (G20) males, and the cross-by-vial interaction effect (P < 0.01, 43% of the overall variation) in Sim-11 (G20) males] were significant. Because of the low frequency of significant tests and the small effects of separate crosses (replication) and different vials except for one case, the data from different vials and crosses were pooled.

Using the pooling data, I made planned comparisons of two means of the missing bristle numbers for different types of simulans males individually in each pair of simulans males, because, as seen in Fig 2, an equality of variances was rejected in many comparisons. I did a t-test for two samples when there was no significant difference in their variances, but I did Welch's approximate t-test when there was a significant difference in variances of two samples (SOKAL and ROHLF 1995 ). The comparisons made are, for example, in the Sim-5 (G20)-Tananarive (G20) pair, the F₁ male progeny of the S5 x Ta cross vs. the F₁ males of Ta x S5, the Sim-5 (G20) males vs. the F₁ males of S5 x Ta cross, and the Tananarive (G20) males vs. the F₁ males of Ta x S5.

View larger version (54K): In this window In a new window Download PPT slide   Figure 2. Distributions of the number of missing bristles in melanogaster-simulans hybrids. (A) Bar diagram showing distributions of the number of missing bristles in hybrids of Sim-5 (G20), Tananarive (G20), F1 males of Sim-5 (G20) () x Tananarive (G20) (), and those of the reciprocal crosses with C(1)RM/Y females of D. melanogaster as well as those of hybrids between the F2 males [all F2 males, F2 males that carried the Sim-5 (G20) alleles both at Dm0478 and at f (SS F2), and F2 males with the Tananarive (G20) alleles both at Dm0478 and f (TT F2)] and C(1)RM/Y females in the QTL mapping experiment. The mean number of missing bristles (sample size) was 12.76 ± 0.41 (SEM; 90) for Sim-5 (G20); 0.22 ± 0.05 (90) for Tananarive (G20); 11.16 ± 0.47 (90) for F1 males of Sim-5 (G20) () x Tananarive (G20) (); 0.22 ± 0.08 (90) for F1 males of Tananarive (G20) () x Sim-5 (G20) (); 5.45 ± 0.24 (600) for all F2 males; 9.92 ± 0.45 (88) for SS F2; and 0.10 ± 0.03 (87) for TT F2. (B) Distributions of the number of missing bristles in hybrids of T-6 (G20) males, Sim-11 (G20) males, and the F1 males with C(1)RM/Y females. The mean number of missing bristles per fly (sample size) was 8.21 ± 0.61 (58) for T-6 (G20) males; 0.27 ± 0.08 (55) for Sim-11 (G20) males; 6.32 ± 0.36 (120) for F1 males of T-6 (G20) () x Sim-11 (G20) (); and 0.37 ± 0.08 (78) for F1 males of Sim-11 (G20) () x T-6 (G20) ().

Deficiency screening:
The deficiency and duplication stocks used in this study and their origins are summarized in Table 3 and Table 4. Dp(1;Y)shi⁺3 was given as a Dp(1;Y)shi⁺1 stock [Df(1)sd72a/Df(1)FM7 + Df(1)sd72a/Dp(1;Y)shi⁺1] by the Bloomington Drosophila Stock Center (Bloomington, IN), but it turned out to carry Dp(1;Y)shi⁺3. As in the other experiments, about 20 females heterozygous for a deficiency were crossed independently to about 20 males of three strains: two simulans [Sim-5 (G20) and A1 (G20)] and one mauritiana [Mau g76 (G20)] inbred strain. The experimental design was the same as that for the X chromosome-effect study and one cross for each combination of line and deficiency chromosome was made for most cases. I made an effort that the bristle numbers were counted for five hybrid females carrying a deficiency and five females carrying a balancer chromosome from each of three vials, making a total sample size of 30 females. Because the stocks were maintained over different balancer chromosomes and the background of different deficiency stocks was not isogenized, the absolute numbers of missing bristles could not be simply compared among the different deficiency stocks. As a simple indication of deficiency effects, the statistical tests for the number of missing bristles per fly were done between deficiency-carrying and balancer chromosome-carrying female hybrids. When the sample variances of these two classes were significantly different, Welch's approximate t-test was done (SOKAL and ROHLF 1995 ). As usual in many deficiency screenings, different origins of deficiency chromosomes require consistent significance of multiple independent deficiencies covering a region on bristle phenotype to prove its effect in hemizygous condition.

Besides the deficiencies, several duplications were studied for rescue action of the bristle loss. The compound-X females carrying duplications were crossed to the simulans males, and male hybrids were examined for the bristle number as in the other experiments. I used a molecular marker (Dm1944 from Berkeley Drosophila Genome Project STS, 14A3-5) to detect the presence of a duplication on the fourth chromosome, Dp(1;4)r⁺. The amplified fragments with these primers in D. melanogaster and D. simulans had different lengths.

Genetic map construction of the D. simulans X chromosome:
Using Sim-5 (G20) and Tananarive (G20), I studied crossover frequencies among the nine markers on the X chromosome described above. F₁ females from the G₀ crosses in the above experiment (F₁ experiment), S5 x Ta and Ta x S5, were backcrossed to the two parental stocks, making four different G₁ crosses (S5 x Ta) x S5, (S5 x Ta) x Ta, (Ta x S5) x S5, and (Ta x S5) x Ta. Each type of cross was done in 10 vials and each vial contained about 10 pairs of females and males. Genotypes at the nine loci in Table 2 were determined for 2 to 5 males from each vial (35 males each for four different crosses) except for sgg genotype of one male. Crossover frequencies were converted to map distances by the map function of FOSS et al. 1993 (with m = 4) as in TRUE et al. 1996A .

QTL mapping:
Twenty F₂ males from each of 40 G₁ crosses in the genetic map construction experiment were collected and crossed to 20 TT-35 females of D. melanogaster in a vial. The number of missing bristles on the notum and humeri for these hybrid males was studied as in the F₁ experiment. Six hundred hybrid males were examined for missing bristles [4 (G₁ different crosses) x 10 (replications) x 3 (vials) x 5 (males)]. Genotypes at the nine loci in Table 2 were determined for two or three hybrid males randomly selected from the first and second vials for each cross [in sum, 4 (G₁ crosses) x 2 (vials) x 24 males from 10 vials = 192 males].

The F₂ males from four different crosses between Sim-5 (G20) and Tananarive (G20) [(S5 x Ta) x S5, (S5 x Ta) x Ta, (Ta x S5) x S5, and (Ta x S5) x Ta] have the different autosomal composition: the F₂ males from (S5 x Ta) x S5 and (Ta x S5) x S5 carry, on average, larger amounts of the Sim-5 (G20) genome than the F₂ males from (S5 x Ta) x Ta and (Ta x S5) x Ta. To examine this autosomal effect, I did a t-test for the difference in the mean numbers of missing bristles between the F₂ males from (S5 x Ta) x S5 and those from (S5 x Ta) x Ta, and between the F₂ males from (Ta x S5) x S5 and those from (Ta x S5) x Ta.

I did the composite interval mapping (ZENG 1994 ) on two populations of different genetic backgrounds [96 F₂ males each from (S5 x Ta) x S5 and (Ta x S5) x S5 crosses (Sim-5 backcross) and from (S5 x Ta) x Ta and (Ta x S5) x Ta ones (Tananarive backcross)] separately. I used QTL Cartographer programs (BASTEN et al. 1996 ) to calculate the likelihood ratio (LR) test statistic for H₁/H₀ in Model 6 (with a window size of 5 cM) of Zmapqtl and to obtain estimates of additive effects, assuming a random-mating F₂ population. The likelihood test statistic is -2 ln(L₀/L₁). L₀ is the maximum likelihood under the null hypothesis (H₀) of a (additive effect of a putative QTL) = 0 and d (its dominance effect) = 0, and L₁ the maximum likelihood under the alternative hypothesis (H₁) of a 0 and d = 0. Because I studied only the male hybrids, the dominance effects could not be estimated. Using the FB method (BASTEN et al. 1996 ), I made the stepwise regression analyses, which selected four markers for the Sim-5 backcross (sqh, G01498, Pp1ß–9C, and f) and two markers for the Tananarive backcross (G01498 and f) to control genetic background. The 5% critical value of the likelihood ratio test statistic was calculated by ²_{0.05/8=0.00625}(d.f. = 2) (ZENG 1994 ).

I also examined the effect of the Dm0478 (13A8-9)–f (15F2-4) region on bristle loss, using three different pairs of H and L strains: Sim-16 and Sim-14; Sim-3 (G20) and Sim-12; and Shira-2 (G13) and Sim-11 (G20). The crosses were done in the same way as the crosses between Sim-5 (G20) and Tananarive (G20): 10 crosses of 20 TT-35 females with 20 F₂ males for each of 4 different G₁ crosses and 5 crosses between 20 TT-35 females and 20 males for each of two parental strains. In addition, 5 crosses were made between TT-35 females and males for each of two different F₁ crosses, Shira-2 (G13) x Sim-11 (G20) and Sim-11 (G20) x Shira-2 (G13). There were three vials of no progeny: one of the crosses of Sim-14, one of Sim-11 (G20), and one of F₂ males from [Shira-2 (G13) x Sim-11 (G20)] x Shira-2 (G13) with TT-35. All the crosses for each pair of H and L lines were done simultaneously. As in other experiments, each parental set (20 females and 20 males) was allowed to lay eggs in three different vials. I counted the number of missing bristles of five hybrid males from each vial and determined the genotype at Dm1911 (14B1-4) for two hybrids randomly chosen from the five hybrid males from the first vial of each cross. Thus, the data from only the first vial of crosses are given in Table 5.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

X chromosome effects:
TAKANO 1998 revealed that melanogaster-simulans hybrid males carrying the simulans X chromosomes exhibited much larger numbers of missing bristles than hybrid males carrying the melanogaster X chromosomes. That study also showed that there exists much variation in the simulans strains at the level of hybrid bristle loss. Using Sim-5 (G20; S5) as a high missing bristle number line and Tananarive (G20; Ta) as a low missing bristle number line, I examined the effects of the simulans X chromosomes on the hybrid bristle loss. Fig 2A illustrates the distributions of the number of missing bristles in simulans X chromosome-carrying male hybrids. Although the mean number of missing bristles for F₁ males of S5 x Ta cross was significantly smaller than that for Sim-5 (G20) (P < 0.05, t = 2.6 with d.f. = 178), the difference [11.16 ± 0.47 for S5 x Ta and 12.76 ± 0.41 for Sim-5 (G20)] was small. The sample variances among hybrid flies were not significantly different from each other [19.9 for S5 x Ta and 15.2 for Sim-5 (G20)]. Indeed, the distribution for the F₁ males from the S5 x Ta crosses was similar to that for Sim-5 (G20). The distribution for the F₁ males from the Ta x S5 crosses was also very similar to that for Tananarive (G20). No difference existed in the mean number of missing bristles between F₁ males of Ta x S5 (0.22 ± 0.08) and Tananarive (G20; 0.22 ± 0.05). By contrast, the difference between the mean number of missing bristles for the F₁ males of the S5 x Ta crosses and that for the F₁ males of the Ta x S5 crosses was highly significant (P < 0.001, Welch's t' = 22.9 with d.f. = 89). The same findings were obtained from the crosses between T-6 (G20) as an H line and Sim-11 (G20) as an L line (Fig 2B) and the crosses between A-1 (G20) as an H line and Tananarive (G20) as an L line. In the latter case, the mean number of missing bristles per fly (sample size) was 5.57 ± 0.49 (60) for A-1 (G20) males, 0.05 ± 0.05 (20) for Tananarive (G20) males, 4.87 ± 0.28 (120) for F₁ males of A-1 (G20) () x Tananarive (G20) (), and 0.40 ± 0.08 (114) for F₁ males of Tananarive (G20) () x A-1 (G20) ().

These findings strongly suggest that most of the difference in the number of missing bristles in hybrids between the H and L lines was attributable to their X chromosomes. Therefore, the subsequent experiments focused on the X chromosome.

Deficiency screening:
I screened the melanogaster X chromosome for genes whose absence affects the bristle number of interspecific hybrid females. Females of melanogaster deficiency stocks were crossed to D. simulans and D. mauritiana males. Table 3 shows the results with the number of missing bristles in balancer chromosome-carrying hybrid females from the same vials as a control. Between-species differences were very clear. No deficiency mauritiana hybrid showed, on average, more than one missing bristle per fly except for Df(1)260-1, In(1)sc^8Lsc^4R, and Df(1)sc10-1. The great degree of simulans-mauritiana differences in the bristle numbers of hybrids with D. melanogaster is consistent with the previous study, in which the loss of bristles was examined for hybrid males carrying the X chromosomes of simulans or mauritiana (TAKANO 1998 ).

Many deficiency chromosomes exhibited significantly greater numbers of missing bristles than the balancer chromosomes. The numbers of missing bristles in deficiency-carrying hybrids were significantly greater than those in balancer-carrying hybrids for both the two simulans strains in the following five regions, assuming the minimum number of loci involved: (1A1; 1B4-6), (3A4-5; 3C2), (10F1; 10F7), (11D-E; 11F1-2), and (18A5; 18D). Particularly, two regions, (3A4-5; 3C2) and (10F1; 10F7), were covered by three different deficiency chromosomes each and showed the greatest numbers of missing bristles.

The deficiency chromosomes, Df(1)260-1, In(1)sc^8Lsc^4R, and Df(1)sc10-1, had significant effects in all three hybrids. The missing region contains the achaete-scute complex (AS-C). Many AS-C mutants exhibit loss of macrochaetae and microchaetae because of the failure of emergence of sensory mother cells (GARCIA-BELLIDO 1979 ). The important difference between the AS-C deficiencies and the other strong-effect deficiencies is that both the melanogaster-mauritiana and the melanogaster-simulans hybrids showed a significant reduction in the bristle number in AS-C-deletion heterozygotes. On the other hand, interspecific hybrids between wild-type strains of D. melanogaster and D. mauritiana did not show the bristle anomaly, but melanogaster-simulans hybrids did. It seems unlikely that the AS-C region contains the locus responsible for the mauritiana-simulans difference, and thus this region alone cannot explain the high degrees of bristle anomaly in melanogaster-simulans hybrids.

The three deficiencies, Df(1)64c18, Df(1)JC19, and Df(1)w^258-11, had great effects on the bristle numbers of the melanogaster-simulans hybrids, but not on those of melanogaster-mauritiana hybrids. To study the region affected by these three deficiencies in more detail, the bristle numbers were examined for the hybrid males carrying the following duplication chromosomes: Dp(1;Y)w⁺Y chromosome carrying a duplication of the 2D2; 3D3 region and Dp(1;3)w^vco carrying a duplication of the 2C1; 3C5 region. Table 4 shows the significant rescue action of these duplications.

Taken together, the deficiency screening identified several regions of factors that potentially cause the hybrid bristle loss due to interactions with the other factors.

Genetic map of simulans X chromosome:
Before the QTL mapping of genetic factors responsible for the bristle loss phenotype of melanogaster-simulans hybrids, I constructed a genetic map of the simulans X chromosome by using two simulans strains [Sim-5 (G20) and Tananarive (G20)] and nine markers (Table 2). The polytene X chromosome banding patterns of the two species are identical except for the small inversions (1E1-2; 1E3 and 3A1-2; 3A5) and the chromosome tip (HORTON 1939 ; LEMEUNIER and ASHBURNER 1976 ). The nine markers covered most of the X chromosome (cytogenetic location 1.32 at the sgg through the location of 65.8 at the run in D. melanogaster). The map distances between consecutive markers (those in D. melanogaster are given in parentheses) were as follows: 10.80 cM (5.2) between sgg and norpA, 7.14 (6.5) between norpA and sqh, 15.79 between sqh and G01498, 9.29 between G01498 and Pp1ß–9C, 6.43 between Pp1ß–9C and Dm1865, 10.01 between Dm1865 and Dm0478, 7.86 between Dm0478 and f, and 8.57 (8.3) between f and run. The total map length between the sgg and run loci was 75.89 cM, which was 1.2 times longer than the standard melanogaster map (65). The longer map of the simulans chromosome is consistent with previous reports that showed that the simulans map is about 1.3 times longer than the melanogaster map as a whole (STURTEVANT 1929 ; OHNISHI and VOELKER 1979 ; TRUE et al. 1996A ).

QTL mapping:
F₂ males from four different crosses between Sim-5 (G20) and Tananarive (G20) [(S5 x Ta) x S5, (S5 x Ta) x Ta, (Ta x S5) x S5, and (Ta x S5) x Ta] were crossed to TT-35 females of D. melanogaster. The distribution of the number of missing bristles in a total of 600 male hybrids is illustrated in Fig 2A. The mean number of missing bristles per fly was 5.5 ± 0.2. Significant differences among the four crosses existed. The average number of missing bristles per fly was 6.0 ± 0.5 for 150 males from the (S5 x Ta) x S5 cross, 4.5 ± 0.4 for 150 males from the (S5 x Ta) x Ta, 6.5 ± 0.5 for 150 males from the (Ta x S5) x S5, and 4.8 ± 0.4 for 150 males from the (Ta x S5) x Ta. Both the difference between the (S5 x Ta) x S5 and (S5 x Ta) x Ta and that between (Ta x S5) x S5 and (Ta x S5) x Ta were statistically significant at the 5% level (Welch's t' = 2.2 for the former comparison and Welch's t' = 2.5 for the latter), but no significant difference existed between (S5 x Ta) x S5 and (Ta x S5) x S5 and between (S5 x Ta) x Ta and (Ta x S5) x Ta. Thus, males carrying larger amounts of the Sim-5 (G20) genome had greater numbers of missing bristles in hybrids.

I determined the genotypes at the nine marker loci for 192 hybrid males and did a QTL analysis using a composite interval mapping method (ZENG 1994 ). Given the significant background effects, the data were analyzed separately for the two different backcrosses [Sim-5 backcross includes (S5 x Ta) x S5 and (Ta x S5) x S5 crosses; and Tananarive backcross (S5 x Ta) x Ta and (Ta x S5) x Ta ones], and the results are graphically presented in Fig 3. The maximum LR scores lay between Dm0478 and f both in the two backcrosses. Their estimated additive effects and positions are 4.52 bristles at 65 cM position from the sgg for the Sim-5 backcross and 4.43 at 66 cM for the Tananarive backcross, showing a very good concordance of the two results.

View larger version (12K): In this window In a new window Download PPT slide   Figure 3. Results of QTL mapping on F2 males between Sim-5 (G20) and Tananarive (G20) for bristle loss phenotype in hybrids with D. melanogaster. The analysis was done separately for two populations of different genetic backgrounds [96 F2 males from (S5 x Ta) x S5 and (Ta x S5) x S5 crosses (Sim-5 backcross, thick line) and 96 F2 males from (S5 x Ta) x Ta and (Ta x S5) x Ta ones (Tananarive backcross, thin line)]. Likelihood ratio test statistic (ZENG 1994 ) is plotted at every 1-cM position. The likelihood ratio test statistic = 2 x (ln 10) x LOD (= 4.61 x LOD). We may take 2 (d.f. = 2) = 10.2 as the 5% critical value (dashed line, ZENG 1994 ).

The LR test statistic profile gave evidence for additional QTL; for instance, another QTL may exist between the f and run loci particularly in the Sim-5 background. However, because estimates and positions for putative QTL within adjacent intervals are interdependent, I was not able to further dissect the possible effects of QTL around f with the present amount of data. In addition, LR test statistic scores within the G01498 and Pp1ß–9C interval slightly exceeded the 5% critical value in the Sim-5 backcross, but the evidence for a QTL in this interval depended on the analytical model. When all the markers are used to control the genetic background (model I IN ZENG 1994 ), the maximum LR score in this interval was only 1.7. In sum, although more than one QTL may be involved, the current QTL analysis found a putative QTL between the Dm0478 and f loci. Its average additive effect was about 4.5 bristles, and this QTL explains 72% of one-half the difference in mean number between the parental strains. Indeed, the number of missing bristles per fly for 88 males with the Sim-5 (G20) alleles both at Dm0478 and at f (SS genotype) ranged from 2 to 22 with the mean of 9.9 ± 0.5; the number of missing bristles per fly for 87 males with the Tananarive (G20) alleles at Dm0478 and f (TT genotype) never exceeded 1 (the mean was 0.1 ± 0.0; Fig 2A).

As mentioned above, the comparisons of mean number of missing bristles between the two different backcrosses provided evidence for the autosomal effects of the two strains. The same tendency was also found in the F₂ males of SS genotype. The average number of missing bristles per fly was 10.93 ± 0.71 for 44 SS males from the Sim 5 backcross and 8.91 ± 0.52 for 44 SS males from the Tananarive backcross, and the difference was statistically significant at the 5% level (Welch's t' = 2.3). This difference was not due to a segregation bias of the Sim-5 (G20) and Tananarive (G20) genomes at the other loci on the X chromosome between the two backcrosses, because the frequencies of Sim-5 (G20) alleles in the Tananarive-backcross SS-genotype F₂ males were higher than or equal to those in the Sim-5-backcross SS males at all the seven loci excluding Dm0478 and f. Thus, the results are best explained by the autosomal effects. By contrast, no difference in the average number of missing bristles of TT-genotype males existed between the two backcrosses, 0.14 ± 0.05 for 44 TT males from the Sim 5 backcross and 0.07 ± 0.04 for 43 TT males from the Tananarive backcross. This suggests an epistatic interaction between the QTL in the Dm0478–f region and an autosomal putative QTL (or multiple QTL). The autosomal factor(s) on Sim-5 (G20) functioned as a dominant enhancer in the SS genotype hybrid males.

The average number of missing bristles of the SS genotype males from the Sim-5 backcross was 2.0 greater than that from the Tananarive backcross. The probability that the hybrid males from the former backcross carried the Sim-5 (G20) allele at the autosomal loci is 3/4 and that in the latter is 1/4. Consequently, the autosomal effects between Sim-5 (G20) and Tananarive (G20) was estimated as two times the difference between the two different backcrosses, that is, about 4. An alternative approach is to use the estimates from F₁ male data. The difference between the parental Sim-5 (G20) and the S5 x Ta males was 1.6 and this difference was statistically significant. This gave 3.2 as the estimate of the autosomal effects between the two parental strains. At any rate, no matter which estimate we take, the sum of the Dm0478–f QTL and autosomal effects could explain nearly all difference between the two parental lines, 12.5 (4.5 x 2 + 3.2 = 12.2).

The major QTL responsible for the difference between Sim-5 (G20) and Tananarive (G20) is not necessarily responsible for the variation in the bristle loss phenotype of hybrids of different simulans strains with D. melanogaster. Different loci may be involved in bristle losses of different hybrids. The effects of the Dm0478–f region on bristle loss phenotype was examined in three different pairs of H and L lines. In this experiment, I determined the genotypes only at Dm1911 (14B1-4), which lies halfway between Dm0478 (13A8-9) and f (15F2-4). Table 5 clearly shows the significant effects of Dm1911 genotypes on the number of missing bristles. However, in the Sim-3 (G20)-Sim-12 pair, the difference between two Dm1911 genotypes of the F₂ males was much smaller than that between the parental lines, although the mean number of missing bristles in Sim-3 (G20) males was not significantly larger than that in F₂ males carrying the Sim-3 (G20) allele at Dm1911. This result was largely due to two outliers in the TT-35 x Sim-3 (G20) crosses, namely, 12 and 16. The others ranged from 1 to 7, with a mean of 3.6 ± 0.4. The mean number of missing bristles for each parental line in Table 5 was obtained from five hybrids from the first of three vials in which each parental set was allowed to lay eggs. If we pool the data from the three vials, the average number of missing bristles for Sim-3 (G20) is only 3.3 ± 0.3, including the above two outliers. Thus, the Dm1911 genotypes accounted for most of the differences between the parental strains in all three cases. In sum, most of the variability in bristle loss phenotype in hybrids of different simulans strains with D. melanogaster may be attributed to a single locus.

The F₁ males between Shira-2 (G13) and Sim-11 (G20) were also crossed with TT-35 males, and the average numbers of missing bristles in hybrids were 8.8 ± 0.7 for F₁ males of Shira-2 (G13; as female parents) x Sim-11(G20; as male parents) and 0.4 ± 0.2 for those of the reciprocal cross. This finding indicates significant X chromosome effects again for this pair of lines, but also a significantly greater number of missing bristles for F₁s of Shira-2 (G13) x Sim-11 (G20) than that for the parental Shira-2 (G13) (P < 0.001, t = 3.7, d.f. = 48). Though the difference is not statistically significant, the Dm1911-Shira-2 (G13) F₂ males have a greater number of missing bristles than Shira-2 (G13) males (Table 5). This may be due to an epistatic interaction between the Shira-2 (G13) X chromosomal factor around Dm1911 and Sim-11 autosomal factor(s).

In conclusion, the QTL mapping analysis provided evidence for the disproportionately large effect of a single QTL (or block of QTL) and the epistatic interactions between X-linked and autosomal factors of D. simulans.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Limitation of deficiency- and duplication-chromosome study and QTL mapping:
The results of the deficiency screenings should be treated with caution because there was no good control to evaluate the effects of different deficiencies and duplications. Comparisons of the mean numbers of missing bristles between deficiencies and balancer chromosomes serve only the purpose of uncoupling X chromosome effects from autosomal contributions. Multiple independent deficiencies are required to test a region. In addition to the AS-C region, the two regions (3A4-5; 3C2) and (10F1; 10F7) met this criterion, where three different deficiency chromosomes showed great numbers of missing bristles (larger than 10 missing bristles per fly in most cases) for each region. On the other hand, many deficiencies showed relatively small numbers of missing bristles. For instance, 21 of 49 deficiency chromosomes studied had less than five missing bristles in hybrids with Sim-5 (G20). Though the small number of missing bristles by itself cannot rule out any effects of the region involved, the data clearly suggest the disproportionate effects of the above two regions. Importantly, the QTL mapping analyses did not find a QTL in these two regions.

Duplication study also requires multiple tests by independent duplications because of a lack of ability to control their genetic backgrounds. The recipient chromosomes of six duplications in Table 4, however, are the Y [Dp(1;Y)] or fourth [Dp(1;4)] chromosomes. Both of these chromosomes of D. melanogaster carry only very small numbers of genes except for the tandemly repeated copies of the ribosomal RNA genes on the Y of D. melanogaster. The copy number of the ribosomal RNA genes, however, cannot explain the great reduction of the bristle number in Sim-5 melanogaster hybrids (TAKANO 1998 ), and an accumulation of deleterious mutations are not allowed on the Y chromosome because of its hemizygosity except for the ribosomal RNA genes (the X chromosome also carries the copies of the ribosomal RNA genes). Therefore, the significant effects of several Dp(1;Y) and Dp(1;4) duplications are likely to be attributable to the duplication segments. At any rate, the duplication study complemented the deficiency screening and QTL mapping, and its results supported the presence of genetic factor in the (3A4-5; 3C2) region found in the deficiency screening and the QTL in the (13F; 14F) region (discussed below).

The composite interval mapping potentially has high power to detect a QTL under the condition of no epistatic interaction between QTL, and we may apply a joint mapping method to pool the two backcross populations. However, the small number of individuals studied in the current analysis and the large sampling variance in missing bristle numbers as seen in the hybrids of parental inbred Sim-5 (G20) males with D. melanogaster females did not allow a precise mapping of QTL, and the results of the QTL mapping alone seemed not to determine whether a major QTL (or block of QTL) lies in the Dm0478–f interval or in the f–run or whether a QTL lies in each of the two intervals. Hybrids of simulans L lines with D. melanogaster always showed no reduction in the bristle number; large sampling variances in missing bristle numbers seem to be a common character of H lines (TAKANO 1998 ). Therefore, a large sampling variance is likely to have existed in the hybrids of Sim-5 (G20)-Tananarive (G20) F₂ males with TT-35 females. To be more exact, hybrids of certain genotypes had large error variances and the others did not. The large sampling variance is problematic in the QTL mapping.

Alternatively, a cutoff point for the missing numbers may be set, say 2 or more vs. 0 or 1, to classify hybrids into H and L types. In total, 34 recombinants between Dm0478 and run (14 from the Sim-5 and 20 from the Tananarive backcross) were recovered from 192 hybrids studied. In a population of these recombinants, all 13 hybrids carrying the Sim-5 (G20) allele at f showed 4 or more missing bristles except for 1 hybrid (only 1 missing bristle), which carried the Tananarive (G20) Dm0478 allele and the Sim-5 (G20) run allele (a crossover between Dm0478 and f, TSS genotype). On the other hand, the number of missing bristles of the 21 hybrids carrying the Tananarive (G20) allele at f did not exceed 1 except in 2 hybrids. One exceptional hybrid carried the Sim-5 (G20) alleles both at Dm0478 and run (double crossovers, STS genotype) and showed 13 missing bristles. The other carried the Sim-5 (G20) allele at Dm0478 and the Tananarive (G20) allele at run (STT genotype) and lost 12 bristles. Despite the relatively low resolution of the QTL map for the hybrid bristle loss in the current analysis, the results strongly suggest the proximity of a QTL to the f locus, and the genotypes and the missing bristle numbers of these exceptional recombinants were consistent with the presence of a QTL between Dm0478 and f.

Factors involved in hybrid bristle loss:
Hybrid bristle loss anomaly is potentially caused by several distinct genetic factors. The Dm0478–f QTL is the factor responsible for the within-simulans variation in the number of missing bristles (class 1 factor). It is important to note that the origins of the Dm1911 (14B1-4) region accounted for most of the differences between H and L lines in the three other different pairs, suggesting that the within-simulans variation largely comes from a single major locus or block of loci.

The deficiency screening, however, failed to find significant effects of the Dm0478 (13A8-9)–f (15F2-4) region. There are several possible explanations. Not all the regions were surveyed in the deficiency screening. Particularly, the 14-15 region could not be studied completely because the strong Minute locus (haplo-insufficient locus and very low viability of the deficiencies including this region in heterozygotes), Minute(1)15D, existed. Alternatively, sex-dependent effects may explain the differences between the two screenings. The bristle loss was studied in hybrid females in the deficiency screening, and the bristle loss was studied in hybrid males in the QTL mapping. I previously showed that the male hybrids are more susceptible to bristle loss than are the female hybrids (TAKANO 1998 ). Therefore, I did the duplication screening in this region (Table 4), where the rescue action of duplications was studied in male hybrids carrying the simulans X chromosome and duplications on the Y or autosomes. Although there were large background effects among the different melanogaster strains, the Dp(1;Y)shi⁺3 chromosome had the greatest rescue effects on the bristle formation in hybrid males. Thus, the results supported the large effect of the Dm0478–f region and the failure of the deficiency analysis may be due to the sex specificity in genotypic effect.

On the other hand, a few regions, especially 3A4-5; 3C2 and 10F1; 10F7 regions, manifested significant effects on the bristle number in melanogaster deficiency-carrying hybrid females. Moreover, the duplications, Dp(1;Y)w⁺Y covering 2D2; 3D3 and Dp(1;3)W^VCO covering 2C1; 3C5, rescued the bristle development significantly. These regions are likely to include fixed melanogaster-simulans differences involved in the hybrid bristle loss (class 2) or factors that show dose-sensitive effects, especially in a hybrid background, without a between-species functional difference (class 3). Thus, they do not necessarily have a between-species difference. It is not easy to distinguish between these two possibilities without molecular cloning of the factor even if a duplication of the region restores bristle formation.

Special attention may be paid to the AS-C gene because the insertion polymorphism in this region significantly affects the number of sternopleural bristles (MACKAY and LANGLEY 1990 ). However, the AS-C deficiencies in heterozygous condition significantly increased the number of missing bristles not only in the melanogaster-simulans hybrids but also in the melanogaster-mauritiana ones. In addition, this complex is located on the tip of the X chromosome, and the within-species DNA variation in this region is extremely low in D. simulans (BEGUN and AQUADRO 1991 ; MARTIN-CAMPOS et al. 1992 ). Taken together, functional melanogaster-simulans differences at the AS-C, if any, are presumably categorized as class 2 not 1, though there is no compelling evidence of between-species differences at this complex so far.

The estimate of Dm0478–f QTL homozygous effect (9.0) was about two times larger than that of the autosomal effects (3.2–4) for the Sim-5 (G20) and Tananarive (G20) pair. However, because of the "homozygosity effects" of the X chromosomes, the X chromosomal effects cannot be compared with the autosomal effects (WU and DAVIS 1993 ; TRUE et al. 1996B ). The autosomal factors in homozygous condition may have the same levels of effect as the X factor. Interestingly, I found epistatic interactions in the hybrid bristle loss phenotype between the simulans X and autosomal factors for two pairs of H and L lines. A significant effect of Sim-5 (G20) autosomes in the SS genotype and a lack of autosomal effects in the TT genotype were observed for the Sim-5 (G20)-Tananarive (G20) pair. Additionally, I found a significantly greater number of missing bristles of F₁ males between Shira-2 (G13) and Sim-11 (G20) than the parental Shira-2 (G13) males. These findings suggest the presence of within-simulans variation on the autosome(s). Because the heterozygotes between H and L strains have the normal bristle pattern (TAKANO 1998 ), hybrid bristle loss must further depend on a melanogaster autosomal factor(s) that may be a fixed difference(s) (class 2).

Variations in pure species and hybrid backgrounds:
A comparison of within-species and between-species sequence variations is a powerful method to gain insights into the evolutionary processes and has been used to distinguish between selective and neutral changes (HUDSON et al. 1987 ; SAWYER et al. 1987 ; MCDONALD and KREITMAN 1991 ; AKASHI 1995 ; EYRE-WALKER 1999 ). This method is applicable to quantitative characters as well, if they can be described in molecular terms.

Unfortunately, only a few studies have examined within-species variation for interspecific-hybrid traits (JOHNSON et al. 1993 ; WADE et al. 1997 ; TAKANO 1998 ). WADE et al. 1997 found significant within-species variation in family size, sex ratio, and morphological anomalies in hybrids of two flour beetle species, but not a significant correlation in those traits between interspecific hybrids and conspecific offspring. This suggests that the factors responsible for the variation in hybrids are different from those in the pure species background or that the same factors act differently in the two backgrounds.

With respect to the number of missing bristles in hybrids with D. melanogaster, TAKANO 1998 showed a large amount of between-population genetic variability in the simulans populations: very low levels of anomaly in the simulans strains collected in Madagascar and the Seychelles and a wide range of degrees of bristle defects in the strains from Japan, the United States, and Africa (Zimbabwe and Congo). As mentioned above, conspecific crosses of the simulans strains produce flies with the normal bristle number, so that the within-simulans variation found in interspecific hybrids is hidden at the level of macrochaetae number in natural populations. On the other hand, sternopleural and abdominal bristle numbers in D. melanogaster have long been used as a model for study of the genetic basis of quantitative variation and of its maintenance mechanisms in natural populations (MACKAY 1995 for review). DNA polymorphisms at candidate loci uncovered significant associations with naturally occurring phenotypic variation in bristle number (MACKAY and LANGLEY 1990 ; LAI et al. 1994 ; LONG et al. 1998 ). The QTL involved in the within-simulans variation in notum-bristle number of hybrids are not known to have an effect on the other bristle numbers in the pure simulans background.

The hybrid bristle defect does not lie in cell fate decisions during bristle development, but in the maintenance of neural fate or differentiation of the descendants of sensory mother cells (SMCs) or both (TAKANO 1998 ). The loci known to affect the sternopleural and abdominal bristle numbers, the AS-C, scabrous, and Delta (MACKAY and LANGLEY 1990 ; LAI et al. 1994 ; LONG et al. 1998 ), are all involved in the emergence and regulation of sensory mother cells, suggesting variations in the SMC numbers for sternopleural and abdominal bristles. Thus, like flour beetles, the genetic basis of the hybrid bristle loss may be different from that of the bristle-number variation in pure species background.

By contrast, sex specificity of genotypic effects are interestingly common both in the QTL affecting sternopleural and abdominal bristle numbers in flies of pure melanogaster background (GURGANUS et al. 1999 ) and in the factors responsible for the hybrid bristle defect (TAKANO 1998 ). Sex-specific action of the factors probably plays a much bigger part in shaping within- and between-species variation than has been thought.

Factors of large effects:
Quantitative traits are genetically dissected in terms of the number of genes involved and their effects. The classical model of the genetic basis of within- and between-species quantitative variation has assumed a large number of genes with small effects (FISHER 1930 ; LANDE 1981 , LANDE 1983 ; CHARLESWORTH et al. 1982 ). However, this view is questioned by recent results of theoretical (ORR 1998 ) and experimental studies (e.g., PATERSON et al. 1988 ; DOEBLEY and STEC 1991 ; HILBERT et al. 1991 ; JACOB et al. 1991 ; LAI et al. 1994 ; BRADSHAW et al. 1995 ; LONG et al. 1995 ; PATERSON et al. 1995 ; LIU et al. 1996 ). In agreement with these findings, this study also identified the single major QTL on the simulans X chromosome. This means that the number of factors involved in the hybrid bristle loss presumably is not so large. In other words, genetic interactions of a small number of factors caused a great degree of bristle anomaly in hybrids. Thus, the defect in the maintenance of neural fate or differentiation of the descendants of SMCs is likely to be attributed to failures of activation of some key genes but not to an accumulation of small effects like slight reductions of gene activities.

In conclusion, although distinct factors potentially cause the bristle loss due to their epistatic interactions, this study identified the single small X chromosomal region that accounted for most of the differences between the H and L simulans lines. This finding is encouraging, and there is reason to investigate the molecular basis of this hidden variation because it may become a future target of natural selection (WADDINGTON 1953 ; RUTHERFORD and LINDQUIST 1998 ).

	ACKNOWLEDGMENTS

I thank Y. Ishii and Y. Yamashita for their technical assistance, and Leah Gilner and two anonymous reviewers for improving the manuscript. I also thank the Bloomington Drosophila Stock Center, the Genetic Strain Research Center in National Institute of Genetics, the European Drosophila Stock Center at Umeå, T. K. Watanabe, C. C. Laurie, J. Modolell, R. S. Hawley, B. Williams, and C. W. Bazinet for fly stocks. This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan and the Sumitomo Foundation.

Manuscript received November 24, 1999; Accepted for publication May 25, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKASHI, H., 1995  Inferring weak selection from patterns of polymorphism and divergence at "silent" sites in Drosophila DNA. Genetics 139:1067-1076[Abstract].

BASTEN, C. J., B. S. WEIR and Z-B. ZENG, 1996 QTL Cartographer: A Reference Manual and Tutorial for QTL Mapping. Department of Statistics, North Carolina State University, Raleigh, NC.

BEGUN, D. J. and C. F. AQUADRO, 1991  Molecular population genetics of the distal portion of the X chromosome in Drosophila: evidence for genetic hitchhiking of the yellow-achaete region. Genetics 129:1147-1158[Abstract].

BIDDLE, R. L., 1932  The bristles of hybrids between Drosophila melanogaster and Drosophila simulans.. Genetics 17:153-174[Free Full Text].

BRADSHAW, H. D., JR., S. M. WILBERT, K. G. OTTO, and D. W. SCHEMSKE, 1995  Genetic mapping of floral traits associated with reproductive isolation in monkeyflowers (Mimulus). Nature 376:762-765.

CHARLESWORTH, B., R. LANDE, and M. SLATKIN, 1982  A neo-Darwinian commentary on macroevolution. Evolution 36:474-498.

COYNE, J. A., 1984  Genetic basis of male sterility in hybrids between two closely related species of Drosophila.. Proc. Natl. Acad. Sci. USA 81:4444-4447[Abstract/Free Full Text].

COYNE, J. A., 1985  Genetic studies of three sibling species of Drosophila with relationship to theories of speciation. Genet. Res. 46:169-192[Medline].

DAVIS, A. W., J. ROOTE, T. MORLEY, K. SAWAMURA, and S. HERRMANN et al., 1996  Rescue of hybrid sterility in crosses between D. melanogaster and D. simulans.. Nature 380:157-159[Medline].

DOBZHANSKY, T., 1970 Genetics of the Evolutionary Process. Columbia University Press, New York.

DOEBLEY, J. and A. STEC, 1991  Genetic analysis of the morphological differences between maize and teosinte. Genetics 129:285-295[Abstract].

EYRE-WALKER, A., 1999  Evidence of selection on silent site base composition in mammals: potential implications for the evolution of isochores and junk DNA. Genetics 152:675-683[Abstract/Free Full Text].

FALCONER, D. S., 1960 Introduction to Quantitative Genetics. Oliver and Boyd, Edinburgh and London.

FISHER, R. A., 1930 The Genetical Theory of Natural Selection. Clarendon Press, Oxford.

FLYBASE,, 1999  The FlyBase Database of the Drosophila Genome Projects and Community Literature. Nucleic Acids Res. 27:85-88. Available from http://flybase.bio.indiana.edu/[Abstract/Free Full Text].

FOSS, E., R. LANDE, F. W. STAHL, and C. M. STEINBERG, 1993  Chiasma interference as a function of genetic distance. Genetics 133:681-691[Abstract].

GARCÍA-BELLIDO, A., 1979  Genetic analysis of the achaete-scute system of Drosophila melanogaster.. Genetics 91:491-520[Abstract/Free Full Text].

GIBSON, G. and D. S. HOGNESS, 1996  Effect of polymorphism in the Drosophila regulatory gene Ultrabithorax on homeotic stability. Science 271:200-203[Abstract].

GURGANUS, M. C., S. V. NUZHDIN, J. W. LEIPS, and T. F. C. MACKAY, 1999  High-resolution mapping of quantitative trait loci for sternopleural bristle number in Drosophila melanogaster.. Genetics 152:1585-1604[Abstract/Free Full Text].

HILBERT, P., K. LINDPAINTNER, J. S. BECKMANN, T. SERIKAWA, and S. SOUBRIER et al., 1991  Chromosomal mapping of two genetic loci associated with blood-pressure regulation in hereditary hypertensive rats. Nature 353:521-529[Medline].

HORTON, I. H., 1939  A comparison of the salivary gland chromosomes of Drosophila melanogaster and D. simulans.. Genetics 24:234-243[Free Full Text].

HUDSON, R. R., M. KREITMAN, and M. AGUADÉ, 1987  A test of neutral molecular evolution based on nucleotide data. Genetics 116:153-159[Abstract/Free Full Text].

HUTTER, P., J. ROOTE, and M. ASHBURNER, 1990  A genetic basis for the inviability of hybrids between sibling species of Drosophila.. Genetics 124:909-920[Abstract].

JACOB, H. J., K. LINDPAINTER, S. E. LINCOLN, K. KUSUMI, and R. K. BUNKER et al., 1991  Genetic mapping of a gene causing hypertension in the stroke-prone spontaneously hypertensive rat. Cell 67:213-224[Medline].

JOHNSON, N. A., H. HOLLOCHER, E. NOONBURG, and C.-I. WU, 1993  The effects of interspecific Y chromosome replacements on hybrid sterility within the Drosophila simulans clade. Genetics 135:443-453[Abstract].

KHADEM, M. and C. B. KRIMBAS, 1991  Studies of the species barrier between Drosophila madeirensis and Drosophila subobscura. II. Genetic analysis of developmental incompatibilities in hybrids. Hereditas 114:189-195[Medline].

LAI, C., R. F. LYMAN, A. D. LONG, C. H. LANGLEY, and T. F. C. MACKAY, 1994  Naturally occurring variation in bristle number and DNA polymorphisms at the scabrous locus of Drosophila melanogaster.. Science 266:1697-1702[Abstract/Free Full Text].

LANDE, R., 1981  The minimum number of genes contributing to quantitative variation between and within populations. Genetics 99:541-553[Abstract/Free Full Text].

LANDE, R., 1983  The response to selection on major and minor mutations affecting a metrical trait. Heredity 50:47-65.

LEMEUNIER, F. and M. ASHBURNER, 1976  Relationships within the melanogaster species subgroup of the genus Drosophila (Sophophora). II. Phylogenetic relationships between six species based upon polytene chromosome banding sequences. Proc. R. Soc. Lond. Ser. B. 193:275-294[Medline].