The Genetic Basis of the Interspecific Differences in Wing Size in Nasonia (Hymenoptera; Pteromalidae): Major Quantitative Trait Loci and Epistasis

Corresponding author: J. Gadau, Universität Würzburg-Biozentrum, Am Hubland, 97074 Würzburg, Germany., jgadau{at}biozentrum.uni-wuerzburg.de (E-mail)

Communicating editor: R. HARRISON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

There is a 2.5-fold difference in male wing size between two haplodiploid insect species, Nasonia vitripennis and N. giraulti. The haploidy of males facilitated a full genomic screen for quantitative trait loci (QTL) affecting wing size and the detection of epistatic interactions. A QTL analysis of the interspecific wing-size difference revealed QTL with major effects and epistatic interactions among loci affecting the trait. We analyzed 178 hybrid males and initially found two major QTL for wing length, one for wing width, three for a normalized wing-size variable, and five for wing seta density. One QTL for wing width explains 38.1% of the phenotypic variance, and the same QTL explains 22% of the phenotypic variance in normalized wing size. This corresponds to a region previously introgressed from N. giraulti into N. vitripennis that accounts for 44% of the normalized wing-size difference between the species. Significant epistatic interactions were also found that affect wing size and density of setae on the wing. Screening for pairwise epistatic interactions between loci on different linkage groups revealed four additional loci for wing length and four loci for normalized wing size that were not detected in the original QTL analysis. We propose that the evolution of smaller wings in N. vitripennis males is primarily the result of major mutations at few genomic regions and involves epistatic interactions among some loci.

AN important question in evolutionary biology is whether adaptation involves the accumulation of many genetic changes with small phenotypic effects or, at least initially, few genes with large phenotypic effects (macromutations sensu ORR and COYNE 1992 ). An obvious way to test which hypothesis is correct is to determine the genetic basis of adaptive traits that vary between closely related species or different populations of the same species.

Most interspecific studies of the genetic basis of quantitative traits have been performed on Drosophila hybrids and demonstrated a polygenic basis of these characters (reviewed in COYNE and ORR 1998 ). These results are not necessarily incompatible with the view that major genes with large phenotypic effects (e.g., >10% explained phenotypic variance; TANKSLEY 1993 , Figure 5, p. 219) play a significant role in the evolution of adaptive traits. One hypothesis for the evolution of adaptive traits is that major mutations generate major phenotypic changes in the trait but simultaneously generate negative pleiotropic effects. Subsequent selection on modifying genes ameliorates the negative effects of major mutations (CLARKE 1997 ).

With the advent of complete genomic maps and new statistical methods for mapping quantitative trait loci (QTL), we can now estimate the minimum number of "genetic factors" (VIA and HAWTHORNE 1998 , p. 353) affecting specific traits and pinpoint their locations within the genome. Additionally, it is possible to directly estimate the effects of individual QTL on the phenotypic variance and the effects of different alleles at specific QTL. Several QTL studies of interspecific differences of adaptive quantitative traits have found QTL with large effects (LAURIE et al. 1997 ; MACDONALD and GOLDSTEIN 1999 ).

Epistasis is here defined as a nonadditive phenotypic effect of interacting genes. Although the role of epistasis in evolutionary and quantitative genetics has been of great theoretical interest, little is known about the relative importance of epistasis during speciation (e.g., COYNE et al. 2000 ; GOODNIGHT and WADE 2000 ) or the influence of epistasis on quantitative traits (CHEVERUD and ROUTMAN 1995 ). This is because of the inherent difficulties of measuring epistatic genetic variance using classical quantitative genetics (but see CHEVERUD and ROUTMAN 1995 ; WOLF et al. 2000 ). However, new analytical methods like QTL provide direct experimental access to estimates of epistatic interactions (e.g., LARK et al. 1995 ; YU et al. 1997 ). Detection of epistasis is easier in haploids because the number of possible genotypes is reduced (increasing statistical power). Therefore, whole genomes can be readily scanned for genetic interactions in haploids. For example, two-locus (pairwise) interactions can be detected by analyzing the phenotypic values of the four genotypes; epistasis is indicated by significant deviation from additive effects of individual loci (LARK et al. 1995 ; CHASE et al. 1997 ). The comparable analysis in diploids would involve nine genotypes, and recessive interactions would occur at low frequencies in the mapping population, making detection more difficult. We have taken advantage of the haploidy of males in our system to investigate both additive and epistatic effects on wing size in the haplodiploid insect genus Nasonia.

The Nasonia species complex consists of three closely related species: N. vitripennis, N. giraulti, and N. lonigicornis (DARLING and WERREN 1990 ). N. giraulti occurs in eastern North America where it parasitizes protocalliphora fly pupae in bird nests, and N. vitripennis is cosmopolitan and occurs microsympatrically with N. giraulti (DARLING and WERREN 1990 ) in parts of its range. It is estimated that the species have been separated for 0.8 millon years (CAMPBELL et al. 1993 ; J. H. WERREN, unpublished data), which is comparable to the youngest known Drosophila species pair, Drosophila mauritiana and D. sechellia (0.6–0.9 mya; KLIMAN and HEY 1993 ; MACDONALD and GOLDSTEIN 1999 ). Nasonia species are reproductively isolated, in part by infections with different strains of the bacterium Wolbachia, which cause cytoplasmic incompatibilities (BREEUWER and WERREN 1990 ; BORDENSTEIN et al. 2001 ). However, Nasonia that have been cured with antibiotics of their Wolbachia endosymbionts (BREEUWER and WERREN 1990 , BREEUWER and WERREN 1995 ) are capable of producing viable and fertile hybrid offspring. However, there is some mortality among the sons of F₁ females in crosses between N. giraulti and N. vitripennis due to recessive genetic incompatibilities (BREEUWER and WERREN 1995 ; GADAU et al. 1999 ). N. vitripennis males have small vestigial wings that are 40% the size of male wings in N. giraulti.

WESTON et al. 1999 conducted hybrid crosses with five eye color mutant strains of N. vitripennis (corresponding to the five chromosomes of Nasonia) to assess the effects of each chromosome on a wing-size trait of an F₂ hybrid. They found an effect of three chromosomes on wing size and tight linkage of one or more wing-size loci with the eye color mutation "or123" on linkage group (LG) IV. They also introgressed this region from N. giraulti into the genetic background of N. vitripennis. The introgressed region, containing either one or a few tightly linked genes, accounted for 44% of the species difference in normalized wing size between N. vitripennis and N. giraulti.

The objective of this study was to conduct a QTL analysis of wing-size differences in N. vitripennis x N. giraulti hybrid males, using a linkage map based on 91 randomly amplified polymorphic DNA (RAPD) markers (GADAU et al. 1999 ). This analysis permits us to more precisely map wing-size loci, to measure the magnitude of individual QTL effects, and to investigate epistatic interactions among loci affecting wing size. The results are compared to the previous work using five visible markers and an introgression strain (WESTON et al. 1999 ). The QTL analysis of the likely adaptive wing-size difference between two Nasonia species reveals the genetic architecture that underlies this difference.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Nasonia stocks and mapping population:
Two inbred and endosymbiont (Wolbachia) free strains of N. vitripennis (ASYMC) and N. giraulti (R16A) were used to generate 15 genetically identical hybrid F₁ females. These females produced 178 males that were used for constructing a genomic map and for QTL analysis (see GADAU et al. 1999 for the construction of a linkage map of Nasonia). The N. giraulti strain R16A is an introgression strain with a N. giraulti nuclear genome in a N. vitripennis cytoplasm (see BREEUWER and WERREN 1995 for details). R16A was used to avoid the known nuclear-cytoplasmic incompatibility between N. vitripennis and N. giraulti (BREEUWER and WERREN 1995 ).

Measurements and transformations:
Measurements were done on the forewings as described in WESTON et al. 1999 . Wing length is the distance from the distal end of the wing to the mass of dark connective tissue at the proximal end. Wing width is measured as the widest part, perpendicular to the length measurements. Interocular distance (WESTON et al. 1999 ) is the distance between the eyes at the first ocelli. It is correlated with relative body size in Nasonia (SKINNER 1983 ) and was used to normalize the wing-size measurements. All measurements are given in ocular units [1 unit = 0.02 mm; note the conversion given in WESTON et al. 1999 was reported incorrectly by a factor of 10].

Wing setae in Nasonia are small, slender hairs evenly distributed over the whole area of the forewing and are probably homologous to the setae of D. melanogaster described as bristles by DOBZHANSKY 1929 . Seta density in a region just below the end of the stigmal vein of the forewing was measured in the F₂ males by counting the number of setae in a 0.0156-mm² grid. Every seta with its base within the grid was counted, including those on the underlying side of the wing. Besides the two basic wing measurements, wing length and wing width, we also used a derived composite measurement—normalized wing multiple (wing length x wing width/interocular distance)—which was also used by WESTON et al. 1999 . Normalized wing multiple was introduced since this measurement eliminated the correlation between wing size and body size in both species (WESTON et al. 1999 and Table 2).

Linkage analysis:
The mapping population and linkage map used for the QTL analysis were the same as in GADAU et al. 1999 . This linkage map was based on the segregation of 91 RAPD markers in 178 males derived from 15 F₁ females. Linkage group designations were the same as in previously published maps (SAUL 1993 ; GADAU et al. 1999 ). The average distance between two markers in the linkage map was 8.4 cM. The relationship between physical distance and map unit is 0.41 Mb/cM (GADAU et al. 1999 ). Having haploids as a mapping population for QTL analysis has multiple advantages: (1) The effect of an allele is directly measurable because there are no dominance interactions among alleles of the same locus; (2) epistatic interactions between nuclear loci are easier to analyze because in two-way interactions only four genotypes are possible; and (3) linkage phase can be determined in each individual even if dominant markers like RAPDs are used.

QTL analysis:
MapQTL 4.0 (VAN OOIJEN et al. 1999 ) was used to identify QTL for all traits. First a standard interval mapping was done to identify the major QTL. Then multiple-QTL-model (MQM) mapping, implemented in MapQTL, was used to fit more than one QTL at a time. The MQM-mapping procedure uses markers closest to the QTL as cofactors to take over the role of the QTL. Thus, the cofactors will reduce the residual variance, increase the power in the search for other segregating QTL, and enhance the accuracy of QTL mapping (JANSEN 1993 , JANSEN 1994 ; ZENG 1993 , ZENG 1994 ; JANSEN and STAM 1994 ). VAN OOIJEN 1999 obtained statistical thresholds for QTL analysis by large-scale simulations. He distinguished between chromosome-wide (suggestive linkage) and genome-wide thresholds (significant linkage) that control the type I error. The statistical threshold for a suggestive QTL (controls for a chromosome-wide type I error at = 0.05) for our map was LOD = 1.9 (VAN OOIJEN 1999 ). Markers with LOD scores >1.9 in an interval mapping procedure were used as cofactors during the consecutive MQM mapping. If the LOD value for a QTL linked with the cofactor dropped below 1.9 during the MQM mapping it was removed from the cofactor list and MQM was run again. This procedure was repeated until the cofactor list remained stable. The genome-wide LOD thresholds for a significant QTL at the 5 and 1% false-positive rates in Nasonia according to VAN OOIJEN 1999 were 2.9 and 3.7, respectively. Additionally, suggestive and statistically significant QTL were statistically confirmed using the standard permutation test for interval mapping (CHURCHILL and DOERGE 1994 ) incorporated in MapQTL 4.0 (VAN OOIJEN et al. 1999 ).

Epistat (CHASE et al. 1997 ) was used to search for epistatic interactions of QTL. This program searches the whole genome for significant interactions between QTL and uses log-likelihood ratios to compare the likelihood of explaining the effects by null, additive, or epistatic models (for a detailed description of the underlying algorithms see CHASE et al. 1997 ). The program organizes genetic mapping data and quantitative trait values into graphic displays that illustrate the individual effects of single loci as well as the interactions between any two loci. The program is available for download at the following site: http://64.226.94.9/epistat.htm.

Mapping procedure of conditional QTL:
First an automated search option was used to find all conditional QTL for all traits exceeding a predetermined threshold [the default settings of the program were used: i.e., 5.0 log-likelihood ratio (LLR) of an additive model vs. a nonadditive model for the null threshold and 6.0 LLR for the additive threshold; minimal group size was 10 (CHASE et al. 1997 )]. Then, we discarded all interactions between linked markers because linkage confounds detection of epistatic interactions in this program. The remaining interactions were analyzed with a Monte Carlo program implemented in Epistat (CHASE et al. 1997 ) to test for the statistical significance. Each Monte Carlo simulation was specific to a given trait and pair of loci. Subpopulations were chosen randomly without replacement from the total population. Log-likelihood ratios for the additive model were calculated on the basis of these distributions. A total of 1,000,000 trials were done for each interaction. We transformed the P value found in the Monte Carlo simulation (the probability that a single trial exceeded the observed value) into 1 - (1 - P)⁵ to account for searching through five linkage groups (LARK et al. 1995 ). We did not correct for the number of markers (n = 91) because we considered only epistatic interactions between markers of different linkage groups to avoid problems with confounding effects of linkage between markers of the same linkage group.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Morphometrics of wing- and head-size measurements:
A comparison between the wing-size measurements of males from both parental strains (ASYMC and R16A; Fig 1) and F₂ hybrid males showed that the average values of the F₂ males were intermediate between the two parental phenotypes (Table 1). The phenotypic correlations between the different traits in the 178 F₂ males are given in Table 2. Surprisingly the phenotypic correlations between the different wing traits were rather low (Table 2), which could indicate that these traits are determined by different sets of genes. The nonsignificant correlation coefficient (0.03) of head size and normalized wing multiple indicates that we effectively removed the influence of body size on wing size by using normalized wing multiple.

View larger version (40K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Comparison of forewing size of the two parental species Nasonia vitripennis (line ASYMC) and N. giraulti (line R16A). Note the reduced wing venation typical for species of the hymenopteran family Pteromalidae and the differences in seta density.

QTL mapping:
Significant QTL (genome-wide type I error P < 0.01; Table 3) were detected for each of the four wing trait measurements: wing length (two), wing width (one), normalized wing multiple (three), and seta density (five). Additionally, we detected epistatic interactions for all traits except wing width (Table 4). All epistatic interactions were conditional. Conditional QTL have no significant effect individually but show a significant phenotypic effect if a particular allele is present at a second unlinked locus. Hence, the effect of a "conditional QTL" is conditional on the genetic background of an individual.

View this table: [in this window] [in a new window]   Table 3. Significant QTL using MapQTL (VAN OOIJEN et al. 1999 )

Contrary to expectations based on the phenotype of the parental species, we found some QTL in which the vitripennis allele was associated with the larger wing phenotype (Table 3, footnote a). Similar results (the "high" allele comes from the "low" line) have been found repeatedly in QTL mapping studies (HUNT et al. 1995 ; LARK et al. 1995 ; PAGE et al. 2001 ) and are normally thought to result from fixation of low alleles in the high line during selection or speciation. They are referred to as "transgressive alleles" (TANKSLEY 1993 ).

Wing-size QTL:
Overall, the initial QTL search for the three wing traits (wing width, wing length, and normalized wing multiple) revealed a few genomic regions (QTL) of large effect. The LOD scores for these QTL are typically well above the 0.01 genome-wide acceptance threshold. Furthermore, permutation tests indicate that we could detect QTL with effects of phenotypic variance of 4–6%. For example, our threshold value for detection of wing width QTL is LOD 2.3 (Table 3) and allows detection of a QTL explaining as little as 5.5% of the phenotypic variance. However, only one QTL was detected, which had a LOD score of 17.63 (38% of phenotypic variance explained). Therefore, we can confidently say that there is a region of very large effect on wing width and no evidence of intermediate magnitude QTL for this trait.

A QTL search for wing length revealed two QTL (LOD scores 3.98 and 5.56, respectively) whereas QTL with LOD scores as low as 3.4 could be detected with P < 0.01 probability. One of these two QTL, occurring on LG II, apparently involves transgressive alleles; the vitripennis allele has a wing length significantly greater than that of the giraulti allele. Two QTL on LG III and IV explained 13.5 and 11.8% of the phenotypic variance of wing length, respectively, in our mapping population. The QTL for wing length on LG IV explained 38.1% of the phenotypic variance for wing width and 22% of the phenotypic variance of normalized wing multiple (Table 3). As seen in Fig 2 and Table 3, there also appears to be a three-way epistatic interaction affecting wing length among regions on LG II (marker 407-1.01), LG IV (tightly linked markers 323-0.98 and A20-1.5), and LG V (marker P4-1.46). The two markers on LG IV map very closely to the major wing length QTL (all markers map within a 2.9-cM region; 1 cM = 0.41 Mb) and therefore possibly represent the same locus or tightly linked loci.

View larger version (42K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Epistatic interactions between QTL for wing size superimposed on a linkage map based on the 178 hybrid males used for the QTL analysis (for details on the linkage map see GADAU et al. 1999 ). Symbols indicate interacting loci; for details on the effect see Table 4. All but one interaction (#) are two-way interactions. QTL detected by interval mapping are shown in boldface type and interacting loci in italics.

The normalized wing multiple was used to measure overall wing size normalized for body size. For this trait, one QTL of large effect was found on LG IV (LOD 11.39, 22.0% phenotypic variance explained, see also Fig 3) and two QTL of smaller effect (LOD 4.00 and 3.32) were found on LG I and LG III, closer to the detection threshold (P < 0.01, LOD 3.2). The QTL for normalized wing multiple on LG IV corresponds to the region of large effect linked to the visible marker or123, described by WESTON et al. 1999 ; or123 was previously mapped to the same region (Fig 1, marker 320-2.1f, LG IV; GADAU et al. 1999 ). This region was introgressed from N. giraulti into N. vitripennis and found to account for 44% of the wing-size difference between the species.

View larger version (17K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Distribution of the LOD scores on LG IV for four wing traits (wm, normalized wing multiple; wi, wing width; le, wing length; set, seta density). (A) Results of a QTL analysis using the interval mapping algorithm of MapQTL (VAN OOIJEN et al. 1999 ). (B) Same data as in A analyzed with the multiple-QTL-model (MQM) mapping algorithm of MapQTL. Cofactors for the calculation were the markers listed in Table 3. The peaks for wing length, wing width, and seta density in B are at the same position (marker 306-0.75f) but were slightly shifted for better visibility. MQM significantly increased the possibility to localize a QTL and the result indicated a single significant QTL on LG IV for each trait and two different QTL for wing size, one influencing wing length, wing width, and seta density and a second affecting the normalized wing multiple only. The horizontal line indicates the threshold for a 1% genome-wide error rate. Note the y-axis in A ends at LOD 20 whereas in B it ends at LOD 15.

A QTL for normalized wing multiple on LG III appears to be transgressive. That is, males with the giraulti allele actually have wings significantly smaller than those of males with the vitripennis allele. This involves the same marker showing a transgressive effect for wing length, which may explain the effect. The QTL for normalized wing multiple on LG I and III explain 7 and 5.4%, respectively, of the observed phenotypic variance. Interestingly, the region on linkage group I that explained 7% of the observed variance of normalized wing multiple had no effect on wing length or wing width. Furthermore, the region on LG III that influenced wing length had no effect on wing width or normalized wing multiple.

Using the program Epistat (CHASE et al. 1997 ) we screened for pairwise epistatic (nonadditive) interactions occurring between markers on different linkage groups. As previously explained, haploid males facilitate detection of epistatic interactions between loci. The analysis for wing length revealed a region on LG II that interacts with the primary QTL on LG IV (Table 4). The vitripennis allele at this locus significantly increases wing size relative to the giraulti allele, but only when combined with the giraulti allele at the primary QTL on LG IV. The locus on LG II has no effect on wing length when combined with the vitripennis allele at the primary QTL. Thus, it can be described as a "conditional" epistatic interaction or a conditional QTL.

Two significant epistatic interactions were detected for normalized wing multiple. One involves regions on LG V and LG IV (Table 4). The LG V region shows a much larger wing size for its giraulti allele in combination with the giraulti allele on LG IV, but a weak effect on wing size when the LG IV region has the vitripennis allele. The conditional nature of these QTL may explain the failure to detect their effect in the original primary QTL analysis (see DISCUSSION for details). The LG IV marker (A20-1.5) maps 24.1 cM from the major QTL on LG IV, and it is possible that it is the same locus given the uncertainties involved in map locations of QTL. To investigate this possibility, we determined the nonadditive interaction between the major QTL on LG IV (76-1.03f) and the LG V marker; no significant epistatic interactions were found between these two loci (LLR = 0.58, P >> 0.05). Therefore, we conclude that these represent two different loci on LG IV: one that interacts epistatically with a locus on LG V and one that has an effect on its own.

The epistatic interactions between regions on LG I (76-0.42) and LG IV (76-1.29) also probably represent two new wing-size loci. The marker on LG IV is >50 cM from the major QTL on this linkage group, and the major QTL on LG I does not show a significant epistatic interaction with the major QTL on LG IV (LLR 4.26, LLR > 6.0 statistical threshold). This epistatic interaction appears to be synergistic: The giraulti allele at each locus increases wing size, but giraulti alleles at both loci have a markedly larger effect on wing size. It is possible that the locus on LG IV (A20-1.5) involved in this interaction is the same as that involved in the interaction between LG IV and LG V: a three-way epistasis. However, these two regions are >50 cM apart, arguing against this interpretation. Taken together, the data suggest six to seven loci affecting wing size, with one region of large effect (22% of explained phenotypic variance) and two additional sets of epistatically interacting loci.

We also conducted QTL analyses for the wing width/wing length (ww/wl) ratio and normalized wing width and wing length (each divided by interocular distance). These analyses gave fundamentally the same results. QTL for ww/wl ratio were found on LG IV and LG III, mapping near the QTL for normalized wing size. Normalized wing width and normalized wing length yielded the same QTL as for the unnormalized trait. Therefore, the location of wing-size QTL is insensitive to various normalization procedures.

Seta density:
The density of setae on the wing is strongly correlated with the wing cell size in Drosophila and presumably also for Nasonia. Seta density also increases with the measurements of body size in Nasonia, indicating that larger males have larger wing cells. We, therefore, investigated QTL for seta density in hybrid males. We found significant QTL for seta density on all five LGs (Table 3) and detected three epistatic QTL (Table 4; Fig 2). In each case, the giraulti allele significantly decreases seta density (i.e., increases wing cell size) relative to the vitripennis allele. Highly significant QTL were found on LG IV (LOD 11.38, 14.8% explained phenotypic variance) and LG III (LOD 17.61, explaining 22.2% of the phenotypic variance). The major QTL on LG IV is the same marker identified for the major QTL of normalized wing size and is tightly linked to the QTL marker for wing width (Table 3; Fig 3). The LG III major QTL (22.2% of phenotypic variance) maps 19.2 cM from the transgressive QTL for wing length and normalized wing multiple. However, the seta density QTL is not transgressive but rather the giraulti allele is associated with a large reduction in seta density (presumed increase in cell size). It is perplexing why a QTL that has such a large effect on seta density (and therefore presumably on wing cell size) would not have any effect on wing size. Possible explanations for this observation are discussed later.

Three significant epistatic interactions were detected (Table 4; Fig 2). In each case, a double dose of vitripennis alleles greatly increases seta density. Two interactions involved a marker on LG IV (209-1.05) that is 21.0 cM from the major primary QTL for seta density on the same linkage group. The region on LG IV (209-1.05) interacts with the regions on LG V and LG I. The marker (209-1.05) is only 4.1 cM from the epistatic QTL for wing multiple on LG IV (A20-1.05) that interacts epistatically with a region on LG V. Furthermore, the respective markers on LG V (307-0.77 and P4-1.46) are only 15.0 cM apart, and the epistatic interactions for seta density and wing multiple are both in the same directions. Therefore, the most likely explanation is that these conditional QTL for seta density and wing multiple represent the same QTL and these QTL interact epistatically to affect wing size through their effects on wing cell size. The second seta density epistatic interaction also involves the same marker on LG IV and a region on LG I. Thus, there is likely a three-way interaction affecting seta density (Fig 2). Although wing multiple also shows a LG IV-LG I epistatic interaction, the respective markers on LG IV are unlinked, as are those on LG I, arguing against involvement of the same loci. Finally, there is a third epistatic interaction for seta density, between a LG V region (315-0.53) and a LG I region (N16-0.8). The two LG V regions are unlinked as are the LG I regions, indicating different loci. In summary, the analysis of seta number reveals a region of very large effect plus a complex web of epistatic interactions; some loci also affect wing size.

Interocular distance:
Head width was measured as the distance between the eyes at the ocellar region of the head, which was used as an index of body size for normalization of the wing multiple. We therefore investigated QTL involved in interocular distance. Four LGs contained significant QTL, each of which explained between 6.2 and 10.6% of the phenotypic variance among the hybrid males (Table 2). Three of these four QTL showed the effect where the allele of N. vitripennis was associated with the larger head size. We could also detect two epistatic interactions for interocular distance (results not shown).

These results may reveal alleles assorting for head shape or possible alleles affecting overall body size among the hybrid males. A subset of hybrid males was smaller than typical for the rearing conditions, indicating that negative epistatic interactions in some hybrid males could contribute to reduced body size. Consistent with this interpretation, one epistatic interaction showed significantly smaller head size for the two recombinant genotypes (vitripennis-giraulti and giraulti-vitripennis) and the other showed a significantly smaller head size for one recombinant genotype (vitripennis-giraulti). However, because there is no correlation between head width and wing multiple (Table 1), such effects are unlikely to confound the identification of the wing-size QTL.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

All QTL reported here were highly significant and exceeded the 0.01% genome-wide statistical threshold determined with two different methods: (1) a standard permutation test for interval mapping (CHURCHILL and DOERGE 1994 ) and (2) an alternative method described by VAN OOIJEN 1999 . Both methods gave approximately the same statistical results. The method of Van Ooijen was slightly more conservative in that the thresholds for controlling for a genome-wide type I error were more stringent (0.1–0.2 LOD units).

The analyses of the genetic basis of seta density, wing shape, and wing size in D. melanogaster have a long history (DOBZHANSKY 1929 ; MILKMAN 1970 ) and have recently been revived by the use of new quantitative genetic techniques (WEBER et al. 1999 ; ZIMMERMAN et al. 2000 ). The results of these investigations indicated that wing size and shape in Drosophila are determined by multiple loci. WEBER et al. 1999 estimated a minimum of 11 QTL on LG III in D. melanogaster while ZIMMERMAN et al. 2000 estimated a minimum of 23 QTL for two intervein regions of the anterior compartment of the wing. Few of the QTL for wing size or wing shape had large phenotypic effects >1 SD or explained >10% of the phenotypic variance (WEBER et al. 1999 ; ZIMMERMAN et al. 2000 ).

In contrast to that finding, WESTON et al. 1999 showed that one or a few genes in a single region of LG IV explained 44% of the differences in wing size between the two hymenopteran sibling species, when introgressed from N. giraulti into N. vitripennis. Our QTL analysis of wing size and other wing characteristics provides a more detailed analysis of the genetic architecture of wing-size differences between the two Nasonia species. We corroborate the WESTON et al. 1999 finding of a region of large effect on LG IV near the visible marker or123. Additionally, we were able to dissect the genetic basis of these traits in more detail, determine the magnitude of the effect of single QTL, and detect epistatic interactions affecting wing size and seta density on wings (Table 3 and Table 4). In addition, both the coarse scale analysis using one visible marker per linkage group (WESTON et al. 1999 ) and our finer scale QTL analysis revealed a major QTL on LG I.

The epistatic analysis also helped us to explain a conflicting result with the WESTON et al. 1999 study. This study found a significant phenotypic effect of LG V in their initial cross between N. vitripennis and N. giraulti (Table 2 in WESTON et al. 1999 ), whereas we could not detect a QTL for the normalized wing multiple on that LG using normal interval mapping or MQM mapping procedures. There was a significant phenotypic effect of LG V on normalized wing multiple in our mapping population [t-test on phenotypic distribution at marker 315-0.53 (LG V), t = 2.959, P = 0.004]. This failure to map a QTL on LG V was probably due to the increased statistical threshold of interval mapping compared to the simple t-test performed by WESTON et al. 1999 . Additionally, the additive phenotypic effect of the QTL on LG V was "masked" by an epistatic interaction between LG IV and LG V (Table 4). In this interaction the effect of the conditional QTL for normalized wing multiple on LG V was conditional on the presence of the giraulti allele on LG IV (Table 4, normalized wing multiple). Therefore, about one-half of the individuals in our mapping population were not informative for the interval QTL mapping (54% of the individuals at the primary locus at LG IV had the vitripennis allele; Fig 2). However, by controlling for two-way epistatic interactions, we were able to reveal a conditional QTL on LG V. Although WESTON et al. 1999 were able to detect that LG V has a phenotypic effect on normalized wing multiple, only the QTL mapping and subsequent search for epistatic interactions revealed the mechanism of the genetic effect of LG V on normalized wing multiple.

Studies from a variety of plants and animals show that experimental manipulations of cell size result in a compensatory change in cell number (and vice versa), maintaining overall size of the tissue (DAY and LAWRENCE 2000 ). This suggests a regulatory mechanism that adjusts cell size and cell number to keep tissue (e.g., wing) size relatively constant. Wing setae in Nasonia are small, slender hairs evenly distributed over the whole area of the forewing. Therefore, it appears that the Nasonia seta densities are comparable to those of Drosophila, where seta densities (sensu DOBZHANSKY 1929 ; see also FRISTROM and FRISTROM 1993 ; ZWAAN et al. 2000 ) are correlated with cell size.

One mechanism of wing-size reduction in N. vitripennis seems to be a reduction in wing cell size because seta density in N. vitripennis is 2.9 times higher than that in N. giraulti and wing area of N. giraulti is 2.5 times larger than that of N. vitripennis. Therefore, our data suggest that the major QTL affecting male wing size act primarily by regulating wing cell size. Furthermore, several of our QTL appear to affect both wing size and seta density. However, we did map a QTL on LG III with large effect (LOD 17.61, 22% explained phenotypic variance) on seta density (the vitripennis allele has much higher seta densities than the giraulti allele) that did not show a corresponding QTL for wing width. Furthermore, a relatively weaker QTL on LG III for wing length and normalized wing multiple (Table 3) actually has the opposite effect expected on the basis of seta density—the vitripennis alleles increase wing size. Given the large effect of the seta QTL on LG III, an explanation is needed. One explanation, that seta density is simply a correlate of body size, is not supported by the correlation analysis (Table 2). A second possibility is pseudolinkage. LG III shows a general bias toward recovery of vitripennis alleles, and two sets of recessive hybrid lethal interactions occur on this linkage group (GADAU et al. 1999 ). Therefore, it is possible that the occurrence of an apparent seta density effect on this linkage group is actually due to accumulated effects of QTL on other linkage groups linked to interacting recessive hybrid lethal loci. This warrants further investigation and emphasizes the need for follow-up genetic analysis to confirm results of QTL studies, particularly when performed in interspecies crosses.

A third possibility is that this is a locus affecting seta density independent of wing size. Most of the QTL for seta density are probably best interpreted as QTL influencing cell growth in wings. Therefore, the genes underlying these QTL may be of interest as regulators of cell size. Likely candidate genes are those involved in the insulin-dependent pathway (WEINKOVE et al. 1999 ; COELHO and LEEVERS 2000 ). Some mutants affecting cell size in this pathway have been shown to also affect wing size in Drosophila (e.g., chico and S6K). However, other mutations in Drosophila alter cell size or number without changing the ultimate size of the wing (FLYBASE 1999 ). This and other evidence (e.g., surgical addition or removal of cells in imaginal disks) suggest a regulatory mechanism that adjusts cell size and cell number to keep wing size relatively constant (COELHO and LEEVERS 2000 ). The major seta density QTL on LG III could be in this cell size-cell number regulation pathway (hence explaining lack of an effect on wing size) whereas the normalized wing multiple QTL are not.

Our results demonstrate that epistasis occurs between QTL for nearly all analyzed traits (Table 4). One would expect epistasis to be quite common due to the ubiquity of gene interactions at the molecular level, e.g., pleiotropic gene action, gene regulatory pathways, and signal transduction pathways. However, surprisingly few significant epistatic interactions were found in other insect QTL studies (e.g., WEBER et al. 2001 ; but see LONG et al. 1995 , LARK et al. 1995 , and YU et al. 1997 for a different situation in plant breeding). This lack of epistatic interactions may have two explanations: (1) If looked for at all, many QTL studies just looked for epistatic interactions between significant QTL (e.g., FRY et al. 1995 ; for a review of the Drosophila literature see MACKAY 1996 ); and (2) using diploid organisms like Drosophila, one needs greater power for detecting pairwise epistatic interactions, both genetically and statistically (MACKAY 1996 ), than for using a haploid mapping population like hymenopteran males. Therefore, haploid and haplodiploid organisms seem to be good model systems to study epistasis. Our study shows clearly that epistatic interactions can be readily detected in haplodiploids.

What is the adaptive significance of the male wing-size differences between the Nasonia species? One (N. vitripennis) of the three species in the genus Nasonia has lost its ability to fly due to a significant reduction of wing size (J. GADAU, unpublished data). We assume that the evolution of smaller wings in N. vitripennis was an active selection process rather than a process of accumulating loss-of-function mutations in an unused structure for several reasons. First, reduction in wing size in Nasonia is male specific. Therefore, one would have to argue that mutational degeneration in wing size in N. vitripennis involved only wing-size genes that were expressed in a sex-specific fashion, which seems unlikely. Second, wing size in Nasonia males can affect several aspects of fitness. Wing-size-dependent vibrations and movements play a role in the male courtship behavior (VAN DEN ASSEM and WERREN 1994 ; WESTON et al. 1999 ). These courtship differences could act as a prezygotic isolating mechanism under field conditions in sympatric populations. Additionally, wings in N. vitripennis are involved in male-male aggressive displays. Furthermore, the reduction in wing size in male N. vitripennis is associated with a reduction in flight muscle volume [t-test, P = 0.001; mean ± SD of normalized (by interocular distance) volume of the longitudinal flight muscle of males: N. vitripennis, 0.011 ± 0.003, n = 5; N. giraulti, 0.025 ± 0.003, n = 5]. Therefore it is likely that reduced wing size in N. vitripennis has resulted in increased resources for other functions during pupal development. However, the fitness consequences of smaller wing sizes have not yet been determined.

We believe that smaller wings and the associated loss of flight in N. vitripennis have evolved as a reaction to an increase in local male-male competition in this species. Evidence suggests that N. vitripennis experiences more male competition than does N. giraulti (DRAPEAU and WERREN 1999 ). There are several examples in which males of species with wing-size polymorphism that retain their flight capabilities suffer from a decrease in their success in obtaining mates if they compete with males that have lost or reduced their flight ability (CRNOKRAK and ROFF 1995 ; FAIRBAIRN and PREZIOSI 1996 ). One model of the genetic basis of adaptive traits assumes that mutations with major effects are selected early in the evolution of adaptive traits and that subsequently modifying genes are selected that ameliorate the negative side effects of these major mutations (CLARKE 1997 ). Our epistatic QTL could be such modifying genes. For example, epistatic QTL for normalized wing multiple depend on the presence of a giraulti allele at both loci to show a significant phenotypic effect (Table 4). Our results are not compatible with the infinitesimal model for the genetic basis of quantitative traits (ROFF 1997 , p. 7) because we found QTL with large phenotypic effects for nearly all of our traits. Since we have now independently confirmed major phenotypic effects of a single region on chromosome IV with three different methods [hybrid crosses with mutant markers and introgression of major wing region (WESTON et al. 1999 ) and QTL analysis (this study)], it is very unlikely that this result is an artifact of QTL analysis or small sample size. Additionally, we found multiple loci that interacted nonadditively. This also violates one of the basic assumptions in classical quantitative genetics for the estimation of the number of genes underlying a quantitative trait (WRIGHT 1968 ; ROFF 1997 ).

As mentioned, a complication of this QTL analysis is the cosegregation of recessive hybrid lethal loci in our mapping population. Approximately 50% of F₂ haploid males in our cross die before adulthood (BREEUWER and WERREN 1995 ). Four pairwise lethal interactions have been identified and mapped that contribute to the majority of this mortality (GADAU et al. 1999 ). The majority of these are asymmetric: Only one allele combination is lethal. Presence of these recessive interactions has not prevented detection of major QTL (and confirmation of one by introgression) for wing size. However, the F₂ QTL analysis may (1) not detect wing QTL linked to hybrid lethal alleles (although it did detect the major QTL for wing size on LG IV, which is linked to a hybrid lethal) or (2) be subject to spurious appearance of QTL that are linked to the viable allele at a hybrid lethal locus that interacts with a lethal locus at another LG. This latter effect is referred to as pseudolinkage (WESTON et al. 1999 ). As mentioned, pseudolinkage could explain the apparent strong QTL for seta density on LG III, since this region also contains two sets of pairwise hybrid lethals that interact with the regions containing other setae QTL.

We view this QTL study as a starting point for a more detailed genetic dissection of wing size in Nasonia. Taking advantage of the haploid genetics and short generation time of this organism, we are now introgressing the different wing-size loci from N. giraulti into N. vitripennis, for fine-scale mapping and a more complete genetic analysis of wing-size evolution. Our long-term goal is positional cloning of the major wing-size genes in this system to investigate how divergent selection for a naturally evolving morphological trait acts at the molecular genetic level.

The genetic basis of the difference in wing size between the two Nasonia species, N. vitripennis and N. giraulti, seems to be few genes each with large effect. The demonstration of epistatically interacting QTL for all but one of our traits demonstrated that epistasis is a significant factor in the determination of wing size in N. vitripennis. However, the difference in wing size between N. vitripennis and N. giraulti is not determined simply by a few genes with large effects. Instead, a more realistic model might be that the evolution of smaller wings in N. vitripennis involved multiple genes with large phenotypic effects (defined as explaining >10% of the phenotypic variance), some with minor effects, so that we could not even detect them, and a significant proportion of nonadditive interactions within the genome.

	ACKNOWLEDGMENTS

We acknowledge Nida Meednu for useful comments during the development of this article. Assistance was provided by Helene Chan, Celina Kennedy, Seth Bordenstein, Berend-Jan Velthuis, Patrick O'Hara, and Jon Chen. J.H.W. and R.E.P. thank the Alexander von Humboldt Foundation Senior Scientist Awards, which led to the collaborations resulting in this work. The National Science Foundation is thanked for research support to J.H.W. (DEB-9981634) and R.E.P. J.G. thanks the Humboldt Foundation for a Feodor-Lynen Stipendium. Additional financial support for this project was provided for J.G. by the Deutschen Forschungsgemeinschaft SFB 554 (TP B-1 and GA-661). R.E.P. and J.G. acknowledge the Santa Fe Institute (SFI) for providing the opportunity to discuss these issues in a working group of the SFI.

Manuscript received June 5, 2001; Accepted for publication March 6, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BORDENSTEIN, S. R., F. P. O'HARA, and J. H. WERREN, 2001  Wolbachia-induced incompatibility precedes other hybrid incompatibilities in Nasonia.. Nature 409:707-710[Medline].

BREEUWER, J. A. J. and J. H. WERREN, 1990  Microorganism associated with chromosome destruction and reproductive destruction and reproductive isolation between two insect species. Nature 376:762-765.

BREEUWER, J. A. J. and J. H. WERREN, 1995  Hybrid breakdown between two haplodiploid species: the role of nuclear and cytoplasmic genes. Evolution 49:705-717.

CAMPBELL, B. C., J. D. STEFFEN-CAMPBELL, and J. H. WERREN, 1993  Phylogeny of the Nasonia species complex (Hymenoptera, Pteromalidae) inferred from an internal transcribed spacer (ITS2) and 28S rDNA sequences. Insect Mol. Biol. 2:225-237[Medline].

CHASE, K., F. R. ADLER, and K. G. LARK, 1997  Epistat: a computer program for identifying and testing interactions between pairs of quantitative trait loci. Theor. Appl. Genet. 94:724-730.

CHEVERUD, J. M. and E. J. ROUTMAN, 1995  Epistasis and its contribution to genetic variance components. Genetics 139:1455-1461[Abstract].

CHURCHILL, G. A. and R. W. DOERGE, 1994  Empirical threshold values for quantitative trait mapping. Genetics 138:963-971[Abstract].

CLARKE, G. M., 1997  The genetics and molecular basis of developmental stability: the Lucilia story. Trends Ecol. Evol. 12:89-91.

COELHO, C. M. and S. J. LEEVERS, 2000  Do growth and cell division rates determine cell size in multicellular organisms? J. Cell Sci. 113:2927-2934[Abstract].

COYNE, J. A. and H. A. ORR, 1998  The evolutionary genetics of speciation. Philos. Trans. R. Soc. Lond. B 353:287-305[Medline].

COYNE, J. E., N. H. BARTON, and M. TURELLI, 2000  Is Wright's shifting balance process important in evolution? Evolution 54:306-317[Medline].

CRNOKRAK, P. and D. A. ROFF, 1995  Fitness differences associated with the calling behaviour in the two wing morphs of male sand crickets, Gryllus firmus.. Anim. Behav. 50:1475-1481.

DARLING, D. C. and J. H. WERREN, 1990  Biosystematics of Nasonia (Hymenoptera: Pteromalidae): two new species reared from bird's nests in North America. Ann. Entomol. Soc. Am. 83:352-370.

DAY, S. J. and P. A. LAWRENCE, 2000  Measuring dimensions: the regulation of size and shape. Development 127:2977-2987[Abstract].

DOBZHANSKY, T., 1929  The influence of the quantity and quality of chromosomal material on the size of the cells in Drosophila melanogaster.. Arch. Entwicklungsmech. Org. 115:363-379.

DRAPEAU, M. and J. H. WERREN, 1999  Differences in mating behavior and sex ratio between three sibling species of Nasonia.. Evol. Ecol. Res. 1:223-234.

FAIRBAIRN, D. J. and R. PREZIOSI, 1996  Sexual selection and the evolution of sexual dimorphism in the waterstrider, Aquarius remigis.. Evolution 50:1501-1511.

FLYBASE,, 1999  The FlyBase database of the Drosophila genome projects and community literature. Nucleic Acids Res. 27, available from http://flybase.bio.indiana.edu/.

FRISTROM, D., and J. W. FRISTROM, 1993 The metamorphic development of the adult epidermis, pp. 843–897 in The Development of Drosophila melanogaster, Vol. 2, edited by M. BATE and A. M. ARIAS. Cold Spring Harbor Laboratory Press, Plainview, NY.

FRY, J. D., K. A. DERONDE, and T. F. C. MACKAY, 1995  Polygenic mutation in Drosophila melanogaster. Genetic analysis of selection lines. Genetics 139:1293-1307[Abstract].

GADAU, J., R. E. PAGE, JR., and J. H. WERREN, 1999  Mapping of hybrid incompatibility loci in Nasonia. Genetics 153:1731-1741[Abstract/Free Full Text].

GOODNIGHT, C. J. and M. J. WADE, 2000  The ongoing synthesis: a reply to Coyne, Barton, and Turelli. Evolution 54:317-324[Medline].

HUNT, G. J., R. E. PAGE, JR., M. K. FONDRK, and C. J. DULLUM, 1995  Major quantitative trait loci affecting honey bee foraging behavior. Genetics 141:1537-1545[Abstract].

JANSEN, R. C., 1993  Interval mapping of multiple quantitative trait loci. Genetics 135:205-211[Abstract].

JANSEN, R. C., 1994  Controlling the type I and type II errors in mapping quantitative trait loci. Genetics 136:871-881.

JANSEN, R. C. and P. STAM, 1994  High resolution of quantitative traits into multiple loci via interval mapping. Genetics 136:1447-1455[Abstract].

KLIMAN, R. M. and J. HEY, 1993  DNA sequence variation at the period locus within and among species of the Drosophila melanogaster complex. Genetics 133:375-387[Abstract].

LARK, K. G., K. CHASE, F. ADLER, L. M. MANSUR, and J. H. ORF, 1995  Interactions between quantitative trait loci in soybean in which trait variation at one locus is conditional upon a specific allele at another. Proc. Natl. Acad. Sci. USA 92:4656-4660[Abstract/Free Full Text].

LAURIE, C. C., J. R. TRUE, L. JIANJUN, and J. M. MERCER, 1997  An introgression analysis of quantitative trait loci that contribute to a morphological difference between Drosophila simulans and D. mauritiana.. Genetics 148:339-348.

LONG, A. D., S. L. MULLANEY, J. D. REID, C. H. FRY, and J. LANGLEY et al., 1995  High resolution mapping of genetic factors affecting abdominal setae number in Drosophila melanogaster. Genetics 139:1273-1291[Abstract].

MACDONALD, S. J. and D. B. GOLDSTEIN, 1999  A quantitative genetic analysis of male sexual traits distinguishing the sibling species Drosophila simulans and D. sechellia.. Genetics 152:1683-1699.

MACKAY, T. F. C., 1996  The nature of quantitative genetic variation revisited: lessons from Drosophila setae. BioEssays 18:113-121[Medline].

MILKMAN, R., 1970  The genetic basis of natural variation in Drosophila melanogaster. Adv. Genet. 55:100-114.

ORR, H. A. and J. A. COYNE, 1992  The genetics of adaptation: a reassessment. Am. Nat. 140:725-742.

PAGE, R. E., JR., M. K. FONDRK, G. J. HUNT, E. GUZMAN-NOVOA, and M. A. HUMPHRIES et al., 2001  Genetic dissection of honeybee (Apis mellifera L.) foraging behavior. J. Hered. 91:474-479[Abstract/Free Full Text].

ROFF, D. A., 1997 Evolutionary Quantitative Genetics. Chapman & Hall, New York.

SAUL, G. B., 1993 Gene map of the parasitic wasp Nasonia vitripennis (= Mormoniella vitripennis) 2N = 10, pp. 3277–3280 in Genetic Maps, Ed. 6, edited by S. J. O'BRIEN. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SKINNER, S. W., 1983 Extra chromosomal sex ratio factors in the parasitoid wasp Nasonia (= Mormoniella) vitripennis. Ph.D. Thesis, University of Utah, Salt Lake City.

TANKSLEY, S. D., 1993  Mapping polygenes. Annu. Rev. Genet. 27:205-233[Medline].

VAN DEN ASSEM, J. and J. H. WERREN, 1994  A comparison of the courtship and mating behavior of three species of Nasonia (Hymenoptera, Pteromalidae). Insect. Mol. Biol. 2:225-237.

VAN OOIJEN, J., 1999  LOD significance thresholds for QTL analysis in experimental populations of diploid species. Heredity 83:613-624.

VAN OOIJEN, J., M. P. BOER, R. C. JANSEN and C. MALIEPAARD, 1999 MapQTL (tm) Version 4.0: Software for the Calculation of QTL Positions on Genetic Maps. CPRO-DLO, Wageningen, The Netherlands.

VIA, S., and D. J. HAWTHORNE, 1998 The genetics of speciation, pp. 352–364 in Endless Form, edited by D. J. HOWARD and S. H. BERLOCHER. Oxford University Press, London.

WEBER, K., R. EISMAN, L. MOREY, A. PATTY, and J. SPARKS et al., 1999  An analysis of polygenes affecting wing shape on chromosome 3 in Drosophila melanogaster. Genetics 153:773-786[Abstract/Free Full Text].

WEBER, K. E., R. EISMAN, S. HIGGINS, L. MOREY, and A. PATTY et al., 2001  An analysis of polygenes affecting wing shape on chromosome 2 in Drosophila melanogaster. Genetics 159:1045-1057[Abstract/Free Full Text].

WEINKOVE, D., T. P. NEUFELD, T. TWARDZIK, M. D. WATERFIELD, and S. J. LEEVERS, 1999  Regulation of imaginal disc cell size, cell number and organ size by Drosophila class IA phosphoinositide 3-kinase and its adaptor. Curr. Biol. 9:1019-1029[Medline].

WESTON, R. F., I. QURESHI, and J. H. WERREN, 1999  Genetics of a wing size difference between two Nasonia species. J. Evol. Biol. 12:586-595.

WOLF, L. L., W. T. STARMER, M. POLAK, and J. S. F. BARKER, 2000  Genetic architecture of a wing size measure in Drosophila hibisci from two populations in eastern Australia. Heredity 85:521-529[Medline].

WRIGHT, S., 1968 Evolution and the Genetics of Population, Vol. 1: Genetic and Biometric Foundations. University of Chicago Press, Chicago.

YU, S. B., J. X. LI, C. G. XU, Y. F. TAN, and Y. J. GAO et al., 1997  Importance of epistasis as the genetic basis of heterosis in an elite rice hybrid. Proc. Natl. Acad. Sci. USA 94:9226-9231[Abstract/Free Full Text].

ZENG, Z., 1993  Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc. Natl. Acad. Sci. USA 90:10972-10976[Abstract/Free Full Text].

ZENG, Z., 1994  Precision mapping of quantitative trait loci. Genetics 136:1457-1468[Abstract].

ZIMMERMAN, E., A. PALSSON, and G. GIBSON, 2000  Quantitative trait loci affecting components of wing shape in Drosophila melanogaster. Genetics 155:671-683[Abstract/Free Full Text].

ZWAAN, B. J., B. R. A. RICARDO, C. J. AVIS, J. VAN LAND, and L. PARTRIDGE, 2000  Cellular basis of wing size variation in Drosophila melanogaster: a comparison of latitudinal clines on two continents. Heredity 84:338-347.

This article has been cited by other articles:

				O. Niehuis, A. K. Judson, and J. Gadau Cytonuclear Genic Incompatibilities Cause Increased Mortality in Male F2 Hybrids of Nasonia giraulti and N. vitripennis Genetics, January 1, 2008; 178(1): 413 - 426. [Abstract] [Full Text] [PDF]

				V. Orgogozo, K. W. Broman, and D. L. Stern High-Resolution Quantitative Trait Locus Mapping Reveals Sign Epistasis Controlling Ovariole Number Between Two Drosophila Species Genetics, May 1, 2006; 173(1): 197 - 205. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal