| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Corresponding author: Yun Tao, Harvard University, 16 Divinity Ave., Cambridge, MA 02138., ytao{at}oeb.harvard.edu (E-mail)
Communicating editor: M. A. F. NOOR
 | ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
Hybrid male sterility (HMS) is a rapidly evolving mechanism of reproductive isolation in Drosophila. Here we report a genetic analysis of HMS in third-chromosome segments of Drosophila mauritiana that were introgressed into a D. simulans background. Qualitative genetic mapping was used to localize 10 loci on 3R and a quantitative trait locus (QTL) procedure (multiple-interval mapping) was used to identify 19 loci on the entire chromosome. These genetic incompatibilities often show dominance and complex patterns of epistasis. Most of the HMS loci have relatively small effects and generally at least two or three of them are required to produce complete sterility. Only one small region of the third chromosome of D. mauritiana by itself causes a high level of infertility when introgressed into D. simulans. By comparison with previous studies of the X chromsome, we infer that HMS loci are only 
40% as dense on this autosome as they are on the X chromosome. These results are consistent with the gradual evolution of hybrid incompatibilities as a by-product of genetic divergence in allopatric populations.
ALLOPATRIC speciation, or genetic divergence of subpopulations in geographic isolation, may be the most common mechanism of speciation (
Investigation of the genetic architecture of HI has a long history (e.g., 40% of the chromosome (
Comparison of the genetic architecture of HI loci on the X and autosomes is potentially very valuable because population genetic theories predict different evolutionary dynamics for genes on these two types of chromosome (
| MATERIALS AND METHODS |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
Introgression lines, markers, fertility assay, and single-fly genotyping:
A full account of the methods for constructing the introgression lines, developing allele-specific oligonucleotide (ASO) markers, assaying male fertility, and genotyping single flies has been given in the companion article (12–15 simB females in each generation.
Homozygous or trans-heterozygous males of the P[w+]-inserts (red-eyed) obtained from these lines were assayed for male fertility in a mating test and were also genotyped with ASO markers to detect possible recombination within the D. mauritiana segments (
Qualitative genetic mapping of HMS:
The crossing scheme in Fig 1A was used to construct the mapping population. Ten lines with introgressions that are not fully fertile when homozygous and that cover all of 3R (marked with an asterisk in Fig 2) were used to generate a total of 4461 progeny. For each line, several hundred red-eyed progeny (range of 311–509) were tested for fertility and genotyped for ASO markers. Males with increased fertility were usually recombinants. HMS loci were inferred by associating fertility shifts with genotypes, as described further in RESULTS. Additional evidence for HMS loci was obtained by comparing the positions of introgressed segments having different fertility levels (Fig 2). In this process we implicitly assume that D. mauritiana alleles always have a negative effect on fertility and that they interact with each other additively and synergistically but not antagonistically.
|
|
Both the sample sizes and progeny counts for most genotypes were small, thus rendering routine statistical methods (parametric or nonparametric) unsuitable to ascertain fertility differences among genotypes. We overcame this problem by a simple permutation test: Two samples of offspring counts were merged and then split randomly into two samples of the original sizes. The mean offspring difference of these random samples was calculated and compared to the observed difference. This procedure was repeated 1000 times and the empirical significance P
(one-tailed) was obtained. A one-tailed test was used because we assume that short introgressions derived from long introgressions by recombination are more fertile (Fig 1A).
QTL mapping:
The phenotypic and genotypic data for QTL mapping were collected from a total of 5025 male progeny from the two crossing schemes in Fig 1. Introgression lines covering the whole third chromosome were used. From the scheme in Fig 1A, 3801 males from 196 introgression lines were used. From the scheme in Fig 1B, 1224 males were generated from 87 crosses involving 58 introgression lines. Fig 3 shows the distribution of offspring counts in the males used for QTL mapping. Among the 5025 males tested, 2423 were sterile and 2602 sired at least one offspring.
|
Male flies for the qualitative mapping and for QTL mapping are from independent collections, although the introgression lines used are not. Eight of the 10 introgression lines (except 44.5 and 37.10) used in qualitative mapping were also used in QTL mapping, with a total of 279 males.
Two approaches were taken to mapping QTL. In one approach, all males were analyzed and fertility was treated as a binary trait (z), with males characterized as either fertile (z = 1 for at least one offspring) or sterile (z = 0 for no offspring). In the other approach, only nonsterile males were analyzed and fertility was treated as a continuous trait (y = number of offspring produced in a mating test). In both cases, QTL mapping was performed by the method of multiple-interval mapping (MIM;
The binary-trait model:
A threshold model was used to account for z with an assumption that the ith male has a certain value of an underlying fertility score (x) that renders the male fertile when it reaches a threshold 
,
We further assume that the underlying fertility score is affected by m QTL according to the following model,
where exi N(0,
2x); 
r and 
r are additive and dominant effects on fertility score from the rth QTL; uir and vir are the effect variables corresponding to the genotype of the rth QTL for the ith individual (M, D. mauritiana; S, D. simulans),
the terms ()rs, ()rs, ()rs, and ()rs are epistatic effects (additive x additive, additive x dominant, dominant x additive, and dominant x dominant); and E1, E2, E3, and E4 are the numbers of pairs of QTL that show significant epistatic effects of the four kinds through a model selection process. See
The conditional probability of z given a QTL genotype specified by uir and vir is
and P(zi = 0|uir,vir) = 1 - P(zi = 1|uir,vir) = 
(( - µx - 
Ui)/x), where (·) is a cumulative normal density function and
In this notation, is a vector of QTL-effect parameters (r, r, ...), and Ui is a vector of effect variables (uir,vir, ... ). In this analysis, the threshold and QTL effects are estimated by assuming that µx = 0 and 2x = 1.
Of course, QTL genotypes are not observed, but we can calculate the probability of each possible QTL genotype for an individual, conditional on the observed marker genotypes, if we assume a specific model of QTL number and positions. There are three possible genotypes for each QTL and 3m possible joint genotypes for m QTL. Thus
where pij is the conditional probability of the jth multilocus QTL genotype given marker genotypes and a specific QTL model (see
The likelihood of the data (z) is
This is a multiple-interval mapping version (
The continuous-trait model:
For offspring counts y, we use the following model,
where eyi N(0, 2y) and ar, dr, (aa)rs, (ad)rs, (da)rs, and (dd)rs are additive, dominant, and epistatic effects of QTL on y. The likelihood of the data (y) is
where 
(·) is a normal density function and D is a vector of QTL-effect parameters (ar, dr, ... ).
Parameter estimates and hypothesis testing:
Maximum-likelihood parameter estimates were obtained with an expectation-maximization (EM) algorithm (as in 
0 and dr 0. The test statistic is
where L0 and L1 are the maximum likelihoods under H0 and H1, respectively, conditional on all other selected QTL effects.
Model selection: The initial model selection (IMS) was performed by regression of the trait value on marker genotypes (logistic regression for z and linear regression for y), using a backward stepwise selection process. The procedure is as follows:
After the initial model is selected (with the number of QTL nQ = nm), model selection is refined using MIM, as outlined in the following:
Here we use a LOD score of 2 as a threshold for QTL detection. In principle, a residual permutation test could be used to empirically estimate the threshold for model selection (
Epistasis: In this introgression experiment, only closely linked QTL segregate in a given cross (Fig 1), so only epistatic effects between close QTL are considered. Here "close" QTL mean pairs of QTL with no more than two other QTL between them. For each QTL pair considered, one of the pairwise epistatic interactions between additive and dominance effects (four interactions for each QTL pair) was either present or absent from the model.
Throughout this article, we use HMS and QTL to refer to the loci for hybrid male sterility mapped through the qualitative genetic and the QTL mapping procedures, respectively. When necessary, we use QTLx and QTLy to distinguish the QTL mapped from the binary or continuous traits, respectively.
Genetic marker map:
Genetic distances between ASO markers are required in the QTL mapping algorithm. These distances were obtained by analyzing data from the QTL-mapping population itself. For any two adjacent ASO markers on a maternally transmitted haplotype, the parental type (PT) is that of D. mauritiana alleles at both loci, whereas the nonparental type (NPT) is that of one D. mauritiana allele at one locus (proximal to P[w+]-insert) and a D. simulans allele at the other (distal to P[w+]-insert). The crossover rate is calculated as NPT/(NPT + PT), from which the genetic distance was calculated using the Haldane mapping function.
| RESULTS |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
The recombination rate within introgressed segments:
In this study, the map distance is 104.8 cM for the third chromosome when estimated from segments of D. mauritiana introgressed into the genetic background of D. simulans. Compared to the 211.3 cM estimated in a pure D. mauritiana background (50%. The introgression map is also much shorter than that measured in F1 hybrids (174.6 cM; 54.9 cM;
HMS loci identified on 3R by qualitative genetic mapping:
A total of 218 introgression lines covering the whole third chromosome have been tested for male fertility. There were 115 fertile, 25 subfertile, 30 quasi-sterile, and 48 sterile lines. The fertility class is based on the average number of offspring that one homozygous male can sire: sterile (0 per male), quasi-sterile (0–10), subfertile (10–45), and fertile (>45;
The existence of HMS factor(s) is inferred by a fertility shift among genotypes that differ in one or few markers (Fig 1). A fertility shift between two genotypes was declared by one of two criteria: They belong to different fertility classes, or the numbers of their offspring differ significantly as shown by the permutation test.
In the following text, as well as in Fig 4, we describe the details of three examples of mapping HMS loci using the scheme of Fig 1A. The other mapping details for seven additional lines are summarized in the Appendix. The critical evidence for declaring the eight HMS loci is summarized in Table 1.
|
|
Introgression 28.3: The genotypes and the corresponding phenotypes are displayed in Fig 4A. Genotype A is for the parental, nonrecombinant homozygous type, which is classified as quasi-sterile. Genotypes B and C were tested as sterile, but their sterility may not really be different from A, given the small sample size. A fertility shift from quasi-sterile to fertile obviously occurs within genotype D, suggesting an HMS factor (1) between markers Rga and Antp. Additional evidence for HMS factor 1 is noted in Table 1. This factor is not necessarily a single gene. Here and throughout this article, we consider that each HMS "locus" may actually be a cluster of genes of small effect.
On the other end of the 28.3 introgression, recombinants between P28 and rdgC are all similar to the parental genotype, providing no evidence for a major factor in that region. However, because the sample size is small and sterile or quasi-sterile males are minimally informative, these data do not rule out the possibility of HMS factor(s) in that region. Indeed, a comparison with line 44.5 described below provides some evidence for the existence of at least one factor in that interval, hence the bracket marked as "?" in Fig 4A.
Introgression 44.5: This line and line 37.10 (see Appendix) were the first two lines used for qualitative mapping. Unfortunately the sterile males (A.1 in Fig 4B and Fig 3) from these two lines were not saved for genotyping. So, here we assume that most or all of the sterile males were of parental type (A.1). Some males of the same genotype (A.2) sired small numbers of offspring, so line 44.5 is classified as quasi-sterile. Interestingly, a single male (A.3) with the same genotype as A.1 and A.2 was very fertile. All markers were scored twice to ensure correct genotyping and nonvirginity of tester females is excluded, although mislabeling of the male cannot be excluded. One may invoke a double crossover between Rga and Antp to explain the data, but for such a narrow range (0.79 cM) this is unlikely (P = 0.03 with Bonferroni correction). A more likely scenario is that an HMS factor ("2") to the right of ninaE is present. The fertility of two flies of genotype B supports this interpretation and additional evidence is noted in Table 1. On the other end of the introgression, HMS 1 was confirmed (genotype C). Obviously, HMS 2 alone is not enough to cause sterility (C–G in Fig 4B).
A comparison between the two introgressions 28.3 and 44.5 is also informative. In the analysis of 44.5, it appears that both HMS 1 and HMS 2 are needed for quasi-sterility. Yet genotype A of 28.3 (Fig 4A), hosting HMS 1 but clearly not HMS 2, is quasi-sterile. This observation suggests an HMS factor(s) to the left of Rga ("?" in Fig 4A). If a factor is located between rdgC and Edg78, it appears to have a dominant effect when combined with HMS 1 (F, G, and H in Fig 4A). Still other evidence discussed below indicates that the factor "?" in Fig 4A may consist of more than two HMS factors, including HMS 9.
Introgression 32.4:
Genotypes A, B, C, D, and G were all quasi-sterile, but significant fertility differences were detected from between-group comparisons: A vs. B (P = 0.014) and A vs. G (P = 0.024; Fig 4C). HMS 2 or other factors may contribute to the A vs. B difference (also compare to line 44.5, Fig 4B). The obvious shift from quasi-sterile (genotypes C and D) to fertile (genotype E) again supports the localization of HMS 1. On the other end of the introgressed segment, the fact that A and G differ in their fertility suggests that there might be some factor(s) to the left of 5-Ht2 (marked by "?"). However, it is unclear on which side of the centromere this factor(s) would be located. We therefore do not declare an HMS locus on 3R here.
Below we show that a factor (HMS 9) is indicated between HMS 1 and 5-Ht2. As in the case of 28.3 (Fig 4A), the observations made in Fig 4C are consistent with this interpretation. Without another factor in this region, genotypes B, C, and D in Fig 4C, which clearly are not homozygous for HMS 2, would be fertile because we noted earlier that HMS 1 alone cannot cause quasi-sterility (see also A.3 and B in Fig 4B).
Similar mapping analyses from crosses involving seven other lines are detailed in the Appendix and the evidence from all 10 lines suggesting a total of eight HMS loci is summarized in Table 1.
Are the eight HMS loci listed in Table 1 sufficient to account for male sterility observed in all introgression lines covering 3R? To answer this question, each of the introgression lines was examined for the presence or absence of the putative HMS loci. Lines informative for this purpose are displayed in Fig 2. Most of the sterility detected on 3R can be attributed to the eight loci. However, several lines seem to require at least two more HMS loci to account for the male sterility observed.
In the region around the centromere, at least one HMS factor seems to be in the Pka-Rga region as suggested in the legend of Fig 4A and Fig C. Additional evidence comes from comparisons among the lines 45.7, 45.8, 45.9, and 45.10. These lines were derived as recombinants from the same progenitor, and their right ends are very likely the same. An HMS factor ("9" in Fig 2) is implicated on the basis of the fact that line 45.10 was more fertile (16 progeny per male) than line 45.8 (4 progeny, P < 0.001).
To account for the sterility of lines 33.9, 33.10, and 45.9 and the quasi-sterility of lines 28.3, 32.4, 45.8, 29.5, 29.6, 32.9, and 29.7, another HMS factor ("10" in Fig 2) is implicated. This factor should fall within the region from the right end of line 27.2 (or 32.2 and 32.10) to the right end of the just-mentioned sterile or quasi-sterile lines.
An HMS equivalent consists of two or three minor HMS loci:
Fig 2 shows three regions on 3R (labeled S1, S2, and S3) that each can cause complete male sterility. Each region is by definition one "HMS equivalent" (
The S1 region contains HMS loci 1, 9, and 10. A combination of all three loci can account for almost all sterile or quasi-sterile lines, with one or two exceptions. HMS 9 and 1 together cannot cause significant sterility, while HMS 1 and 10 together may render a male subfertile (line 45.10). However, it is unclear what causes the fertility difference between the sterile and quasi-sterile lines covering the S1 region. Furthermore, line 45.5 is rather strange because it covers the whole S1 region and HMS 2, yet it is still not fully sterile (line 29.1 could also contain the whole S1 region but its right end may not contain HMS 10). Two ad hoc explanations are provided here to explain these observations, but further experimental evidence is needed to clarify these issues.
First, genetic background variation could underlie the fertility variations among different lines. Although these introgression lines were constructed deliberately in a genetic background as uniformly as possible, some variation cannot be avoided (
Second, it is possible that some D. mauritiana alleles act antagonistically, rather than synergistically, in causing sterility. In other words, two HMS factors can suppress each other. For example, the line 45.5 evidently covers the whole length of 45.9, yet the latter is fully sterile. It is possible that some loci on both ends of 45.5, not covered by 45.9, could be this kind of antagonistic gene. For example, it is possible that HMS 9 and 2 could suppress one another (lines 33.5 and 45.5). This notion of antagonistic interactions is supported by QTL mapping (see below).
In the S2 region, HMS 2 and 3 together can cause complete sterility (Fig 2, lines 41.3 and 41.5). A notable aspect of HMS 3 (called tmy) is that no fertile introgression lines ever cover it. Indeed, tmy is the only HMS locus that may contain a "major gene" effect. No other locus can cause significant sterility on its own, because the third chromosome is fully covered by fertile homozygous introgressions except in the region of tmy (
Distal to tmy, HMS 7 may also enhance the sterilizing effect of tmy. For example, line 37.10 is quasi-sterile while line 37.7 is subfertile. The latter may not contain HMS 7. It is unclear, however, whether HMS 8 has an enhancing effect because the mapping of HMS 8 is too rough to make any reliable inference.
In region S3, HMS 8, 7, and 5 together can cause complete sterility (Fig 2, lines 46.17, 46.7, 35.1, and 35.2). The effects of HMS 8 and 7 can be demonstrated best by lines 46.4, 46.15, and 46.9, as shown in Fig 4A Fig 6. In the S3 region, if HMS 8 is not included in the introgression, the fertility will recover a little but still be quasi-sterile (e.g., line 46.15 in Fig 4). If both HMS 7 and 8 are not included, the fertility will recover substantially (Fig 6, line 46.4).
The other two factors, HMS 4 and 6, have no apparent effect by themselves (Fig 2, 46.2, 49.11, etc.). If combined with HMS 5, they may cause significant fertility reduction but not full sterility (Fig 2, lines 46.4, 46.3, and 46.19).
QTL of hybrid male fertility:
An MIM procedure was used to carry out two analyses for identifying QTL in a population of introgression genotypes (Fig 3). In one case, 5025 males were assigned a binary trait classification (sterile or fertile), which was treated as a threshold trait with an underlying continuous fertility variable. In the other case, the offspring counts from a subsample of 2602 nonsterile males were treated as a continuous variable for QTL mapping. We use QTLx and QTLy to represent the results obtained from these two analyses, respectively (Fig 5). The additive, dominance, and epistatic effects for 19 QTLx are summarized in Table 2 and Table 3 and the corresponding effects for 18 QTLy are given in Table 4 and Table 5. Although variances of the estimates were not obtained in this analysis, we believe they are large because each estimate tends to depend on a small number of introgression lines. Therefore, these results provide only a rough estimate of the genetic architecture of hybrid male fertility. Nevertheless, some general patterns can be inferred from Table 2 Table 3 Table 4 Table 5.
|
|
|
|
|
First, the D. mauritiana allele (M) does not always have a negative effect on fertility (i.e., or a < 0) as might be expected for hybrid incompatibility factors. In fact, the mauritiana allele has a positive effect in 5 of 19 QTL for the binary sterility trait and in 8 of 18 QTL for the continuous fertility trait. This could mean that not all QTL detected are due to hybrid incompatibilities, but may be due to polymorphism of deleterious alleles that would have similar effects in both homo- and heterospecific genetic backgrounds.
Second, the dominance relationships between the M and S alleles are mixed, having no apparent trend. There are 8 QTLx and 8 QTLy, where the M allele is dominant over the S allele [i.e., (a) and (d) have the same sign], and 11 QTLx and 10 QTLy, where the M allele is recessive to the S allele [i.e., (a) and (d) have the opposite sign] (Table 2 and Table 4). Furthermore, overdominance of the M allele was estimated for 5 QTLx and 5 QTLy [|| < || or |d| < |a|, (a) and (d) have the same sign], while underdominance was estimated for 3 QTLx and 10 QTLy [|| < || or |d| < |a|, (a) and (d) have the opposite sign]. These results may suggest complex interactions among loci causing hybrid male sterility.
Third, epistatic effects are pervasive (Table 3 and Table 5). Ten out of 24 significant epistatic interactions for QTLx (and 5 out of 18 for QTLy) are positive. The signs of the epistatic effects cannot be predicted from the signs of the two effects [(a) and (d)] involved. These observations suggest both synergistic and antagonistic relationships among the loci causing hybrid male sterility.
Fig 5 shows the LOD profiles for QTL mapping of the two traits. There are several points of correspondence, but the high density of QTL, the pervasiveness of epistasis, and the large sampling variances of estimates make it difficult to identify one-to-one relationships with high confidence. Nevertheless, it is clear that each type of analysis identifies a similar number of QTL within the same regions and both analyses show that hybrid male sterility in these species is highly polygenic.
In the qualitative trait mapping, we implicitly assumed that the D. mauritiana allele at each locus confers a negative effect on fertility. However, we also mentioned an indication of antagonistic effects in the region around the centromere where longer introgressions such as 45.5 and 29.1 were actually more fertile than shorter ones such as 45.9 and 29.7 (Fig 2). The QTL mapping results are consistent with this indication, because QTLx 7 and 10 and QTLy 8 and 12 (near the centromere) have negative additive effects (i.e., positive effects on fertility of the D. mauritiana allele).
In another comparison between qualitative and QTL mapping, there is a puzzling disagreement between QTLx 16 and HMS 7, which occur at approximately the same position. HMS 7 was identified as a negative M effect by comparing lines 46.15 and 46.4 (Fig 4 and Fig 6). However, QTLx 16 has a very large positive —strongly suggesting that its M allele has a fertilizing effect. How to reconcile this contradiction? The additive effect of QTLx 16 has three significant negative interactions with other QTLx effects (Table 3). If we calculate the net effect of QTLx 16 as the value of plus half of each of these three epistatic effects, the value is actually -0.012, suggesting a sterilizing M allele. The interpretation of the additive effect is not straightforward when epistatic effects are present. This conclusion may apply to several other QTL with positive or a estimates.
Taking 19 as the number of HMS loci on the third chromosome, the density of HMS loci per polytene band is 40% of that on the X chromosome, where 9 HMS loci have been mapped on 40% of that chromosome (reviewed by 15 such equivalents (
| DISCUSSION |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
The genetic architecture of hybrid male sterility:
D. simulans and D. mauritiana are geographically isolated and diverged from a common ancestor 0.3 million years ago (
This polygenic view of reproductive isolation does not preclude an occasional incompatibility of large effect due to a single pair of allelic fixations (
In any case, the magnitude of effect of an incompatibility in hybrids is not necessarily related directly to the magnitude of effect on the primary trait within diverging subpopulations. For example, it is possible that different mutations fixed by selection in different subpopulations may have large effects on a primary trait within those subpopulations, but no incompatibility at all in hybrids. Similarly, it is possible that mutations of small positive effect on a trait within subpopulations could have very negative interactions in hybrids, producing an incompatibility of major effect.
In the study reported here, the QTL approach to genetic analysis of incompatibilities has provided strong evidence for complex patterns of epistasis. Such a result may not be surprising, since hybrid incompatibilities are, by definition, negative interactions between alleles at different loci. However, there could be multiple systems of incompatibility that do not interact with each other. For example, an HMS locus A in this study may represent a D. mauritiana allele at locus i (within an introgression) that has a negative interaction with a D. simulans allele at locus j (outside of the introgression). Another HMS locus B may consist of a negative interaction between a D. mauritiana allele at locus k and a D. simulans allele at locus l. The locus pair i, j does not necessarily interact with the locus pair k, l and the effects of the two incompatibilities may be additive with respect to a quantitative measure of fertility.
One view of the genetic basis of hybrid incompatibility is that it is due to numerous loci distributed throughout the genome with effects that are additive and interchangeable (
D. simulans and D. mauritiana are very similar morphologically. The only distinctive difference is in the shape of the posterior lobe of the genital arch in males (
The introgression approach to genetic analysis of HMS:
The use of introgression in the genetic analysis of quantitative traits has a long history. A classic technique was to isolate a chromosomal region of interest in an isogenic background and then apply conventional mapping (
In the analysis of hybrid sterility, the use of introgressions has been a necessity because, in many species pairs, the large number of genes contributing to sterility and the high level of sterility make backcross or F2 designs unworkable. The study reported here and earlier studies of the species pair D. simulans and D. mauritiana (reviewed by
First, the fact that the detection of HMS loci is contingent on the genetic background, while sometimes an advantage, can also be a weakness. Some loci may escape detection if the background level of underlying fertility is too low or too high. Also, if complex epistasis is present, there is no way to detect a particular HMS locus if it interacts with other loci only in a particular way or with loci that are not closely linked within an introgressed segment that is being dissected genetically. Therefore, the current estimate of 19 HMS loci on the third chromosome could be considerably lower than the actual number.
Second, chromosomal rearrangements involving introgressed segments may suggest incompatibilities that do not exist. For example, transposition or translocation of a gene in one species, but not the other, could result in different genic content of an introgressed segment relative to the segment it is replacing. This rearrangement could result in the deletion of an essential gene in the introgression genotype relative to both parental species. Although the polytene band maps of D. simulans and D. mauritiana are nearly identical (
|
|
| FOOTNOTES |
|---|
2 Present address: Monsanto Company, 800 N. Lindbergh Blvd., N2NA, St. Louis, MO 63167. 
| ACKNOWLEDGMENTS |
|---|
We thank Lynn Stam, Kristen Inoue, Jermaine Fuller, Lixia Jiang, and Jing Zhang for technical support during this study and C. Meiklejohn and J. Blumenstiel for critical reading of various drafts of this article. We also thank two anonymous reviewers for their suggestions on improving this manuscript. Y.T. thanks R. Fehon, D. Kiehart, J. Mercer, M. Rausher, and M. Uyenoyama for guidance. Y.T. was a recipient of a Catherine Stern Dissertation Year Fellowship of Duke University. This work was supported by National Institutes of Health grants 47292 and 65169.
Manuscript received January 22, 2003; Accepted for publication March 21, 2003.
| APPENDIX |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
Fig 1A Fig 7 provide additional details about the qualitative mapping of HMS loci on chromosome 3. Each figure describes the results of recombination analysis of a particular introgression. Also see the legend of Fig 4 for explanation of symbols.
Sterile introgression 42.4 is summarized in Fig 1. One recombinant in the Akt1 and Su(Hw) interval was fertile, indicating HMS factor 3, which was named too much yin (tmy) and mapped to a region of < 80 kb (75%) progeny ( = 0.873) or G (P = 0.849) and D.2 was very fertile. The position of P42 relative to HMS factor 2, which was implicated in Fig 4B and previously named broadie (
The sterile introgression 40.8 (Fig 2) also provides evidence for HMS 3 (genotype D). Males homozygous for tmy alone are semi-sterile, but fully sterile if coupled with broadie (
Line 37.10 is summarized in Fig 3. Also shown is the proportion of females (k) in the offspring. The 245 males in A.1 were not genotyped but are assumed to be of parental genotype (A.1 and A.2). HMS factor 3 (tmy) is implicated once again with fertility and sex ratio shifts within genotype D (P = 0.04 for comparing fertility in D.1 and D.2). D.1 is homozygous for tmy while D.2 is heterozygous. At the other end of the introgression, it appears that recombination to the right of P37 may reduce sterility because only 14.9% of all males in A sired any progeny whereas the three individuals of B and C were all fertile. Some sterility factor(s) might be in the region to the right of P37 (marked as "?"), but the data here are not adequate to clarify this point. Indeed, later evidence supports at least one HMS factor in the Odh-Crc region (see Fig 4 and Fig 6).
Introgression 46.15 is summarized in Fig 4. It is best to examine this and two other introgressions, 46.9 (Fig 5) and 46.4 (Fig 6), together. At the right end of introgression line 46.15 (Fig 4), there was a significant increase in fertility from genotypes B to C (P = 0.03; a shift from A to C is also significant with P = 0.002). A sterility factor between Ald and jan is suggested. The mean fertility between C and E was different (P = 0.03), implying an HMS factor in the region P46-Mlc1. These three regions that may contain HMS factor(s) are marked with "?" to the right of P46 in Fig 4. It is clear that males were progressively more fertile when larger regions distal to P46 were made heterozygous (A–E in Fig 4). This suggests that several HMS factors that dispersed distal to P46 are responsible. Alternatively, a simpler explanation would be that one factor between Ald and Mlc1 ("4") is responsible for the fertility change observed. This latter explanation is supported by more mapping data (see Fig 6).
At the left end of introgression 46.15, two phenotypic shifts can be detected. Compared to F, G became fairly fertile, implying an HMS factor in the region of Crc to hb ("?" to the left of P46 in Fig 4). However, only two individuals were in F, so this interpretation may not be reliable. There was also a significant increase in fertility from G to H (P = 0.002), implying another HMS factor between Tub85E and Rox8 ("5" in Fig 4). HMS factor 7 is inferred by comparing this mapping with that of line 46.4 (see Fig 6).
Introgression 46.9 is summarized in Fig 5. This line is almost fully sterile and is definitely more sterile than line 46.15, as shown by comparing genotype A in Fig 4 and Fig 5 (P < 0.001). Since the introgression of line 46.15 is inclusive of that of line 46.9 at the right end, the sterility must be due to factor(s) at the left end and must be proximal to Crc. This region ("?" in Fig 5) is responsible for the phenotypic shift from A to B, where B had more fertile males (P < 0.001; Fig 5). A parsimonious explanation is that it is HMS 8, narrowed down to the Mst-Odh region, as suggested by comparing this mapping with the two in Fig 4 and Fig 6.
Introgression 46.4 is summarized in Fig 6. Two significant fertility increases were at the right end: from A to B (P = 0.004) and from C to D (P < 0.001). Two factors may explain the fertility changes. One is between jan and the telomere of 3R ("6" in Fig 6), and the other is between Ald and jan ("?" in Fig 6). The latter factor is most parsimoniously inferred to be HMS 4, which was previously implicated in Fig 4. Proximal to P46, HMS 5 is implicated again in a significant shift in mean fertility from H to I (P = 0.04), but there is insufficient information to localize it precisely.
It is interesting to compare the mapping data from the three introgressions, 46.4, 46.9, and 46.15 (Fig 4A Fig 6). First, a comparison of genotype A in Fig 4 and Fig 6 (P < 0.001) suggests that line 46.15 hosts a factor that line 46.4 does not (HMS 7). However, it is unlikely that line 46.15 hosts HMS 6 at the right end. Second, a combination of HMS factors 4, 5, and 7 can cause severe sterility (Fig 4A in 46.15), yet 46.9 is more sterile than 46.15 even though 46.9 does not host HMS factors 4 and 6 (Fig 5A vs. Fig 4A; P < 0.001). This phenotypic difference suggests that at least one more HMS factor ("8" in Fig 5) must be proximal to HMS factor 7 but distal to Mst, because the introgressed D. mauritiana segment in 46.9 does not extend beyond this point.
Introgression 38.1 is summarized in Fig 7. The mapping data for this introgression are consistent with previous analyses (Fig 4A Fig 6). As in the case of 46.9, a combination of HMS factors 4, 5, 7, and 8 confers full sterility (e.g., E in Fig 7). Recombinants implicate an HMS factor ("5"), which is localized to the region between hb and Rox8.
| LITERATURE CITED |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXLITERATURE CITED |
|---|
BREESE, E. L. and K. MATHER, 1957 The organization of polygenic activity within a chromosome of Drosophila. I. Hair characters. Heredity 11:373-395.
CABOT, E. L., A. W. DAVIS, N. A. JOHNSON, and C.-I WU, 1994 Genetics of reproductive isolation in the Drosophila simulans clade: complex epistasis underlying hybrid male sterility. Genetics 137:175-189.[Abstract]
CHARLESWORTH, B., J. A. COYNE, and N. BARTON, 1987 The relative rates of evolution of sex chromosomes and autosomes. Am. Nat. 130:113-146.
COYNE, J. A., 1992 Genetics and speciation. Nature 355:511-515.
COYNE, J. A. and B. CHARLESWORTH, 1986 Location of an X-linked factor causing sterility in male hybrids of Drosophila simulans and D. mauritiana. Heredity 57:243-246.
COYNE, J. A. and B. CHARLESWORTH, 1989 Genetic analysis of X-linked sterility in hybrids between three sibling species of Drosophila. Heredity 62:97-106.
COYNE, J. A. and H. A. ORR, 1998 The evolutionary genetics of speciation. Philos. Trans. R. Soc. Lond. B 353:287-305.[Medline]
DAVIS, A. W. and C.-I WU, 1996 The broom of the sorcerer's apprentice: the fine structure of a chromosomal region causing reproductive isolation between two sibling species of Drosophila. Genetics 143:1287-1298.[Abstract]
DOBZHANSKY, TH., 1936 Studies on hybrid sterility. II. Localization of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21:113-135.
DOBZHANSKY, TH., 1937 Genetics and the Origin of Species, Ed. 1. Columbia University Press, New York.
DOBZHANSKY, TH., 1975 Analysis of incipient reproductive isolation within a species of Drosophila. Proc. Natl. Acad. Sci. USA 72:3638-3641.
ESHED, Y. and D. ZAMIR, 1995 An introgression line population of Lycopersicon pennellii in the cultivated tomato enables the identification and fine mapping of yield-associated QTL. Genetics 141:1147-1162.[Abstract]
FRARY, A., T. C. NESBITT, A. FRARY, S. GRANDILLO, and E. VAN DER KNAAP et al., 2000 fw2.2: a quantitative trait locus key to the evolution of tomato fruit size. Science 289:85-88.
FRIDMAN, E., T. PLEBAN, and D. ZAMIR, 2000 A recombination hotspot delimits a wild species quantitative trait locus for tomato sugar content to 484 bp within an invertase gene. Proc. Natl. Acad. Sci. USA 97:4718-4723.
HOLLOCHER, H. and C.-I WU, 1996 The genetics of reproductive isolation in the Drosophila simulans clade: X vs. autosomal effects and male vs. female effects. Genetics 143:1243-1255.[Abstract]
JIANG, C. and Z-B. ZENG, 1997 Mapping quantitative trait loci with dominant and missing markers in various crosses from two inbred lines. Genetica 101:47-58.[Medline]
KAO, C.-H. and Z-B. ZENG, 1997 General formulations for obtaining the MLEs and the asymptotic variance-covariance matrix in mapping quantitative trait loci when using the EM algorithm. Biometrics 53:653-665.[Medline]
KAO, C.-H. and Z-B. ZENG, 2002 Modeling epistasis of quantiative trait loci using Cockerham's model. Genetics 160:1243-1261.
KAO, C.-H., Z-B. ZENG, and R. D. TEASDALE, 1999 Multiple interval mapping for quantitative trait loci. Genetics 152:1203-1216.
KLIMAN, R. M., P. ANDOLFATTO, J. A. COYNE, F. DEPAULIS, and M. KREITMAN et al., 2000 The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156:1913-1931.
LEMEUNIER, F. and M. ASHBURNER, 1984 Relationships within the melanogaster species subgroup of the genus Drosophila (Sophophora). IV. The chromosomes of two new species. Chromosoma 89:343-351.
LIU, J. J., J. M. MERCER, L. F. STAM, G. C. GIBSON, and Z-B. ZENG et al., 1996 Genetic analysis of a morphological shape difference in the male genitalia of Drosophila simulans and D. mauritiana. Genetics 142:1129-1145.[Abstract]
MAYR, E., 1942 Systematics and the Origin of Species. Columbia University Press, New York.
MULLER, H. J., 1942 Isolating mechanisms, evolution, and temperature. Biol. Symp. 6:71-125.
NAVEIRA, H. F., and X. R. MASIDE, 1998 The genetics of hybrid male sterility in Drosophila, pp. 330–338 in Endless Forms: Species and Speciation, edited by D. J. HOWARD and S. H. BERLOCHER. Oxford University Press, New York.
ORR, H. A., 1998 The population genetics of adaptation: the distribution of factors fixed during adaptive evolution. Evolution 52:935-949.
ORR, H. A. and A. J. BETANCOURT, 2001 Haldane's sieve and adaptation from the standing genetic variation. Genetics 157:875-884.
ORR, H. A. and J. A. COYNE, 1992 The genetics of adaptation: a reassessment. Am. Nat. 140:725-742.
