Correlated variability of traits of the shape of the copulatory organ
To study the structure of correlations among the traits, and the latent factors responsible for a correlated variation, total variance over all samples was examined by a factor analysis. Seven factors were taken based on the scree plot and together accounted for 64.4% of the total variance (Table 1). Characteristics of the variation in the traits under study are shown in supplementary materials (Suppl. Table 1).
Table 1
Compositions and eigenvalues of factor loadings as revealed by the maximum likelihood method.
Factor 1 (2.674, 7.6) | Factor 2 (3.126, 8.9) | Factor 3 (5.326, 15.2) | Factor 4 (2.021, 5.8) |
29, 31*, 33 | 30*, 32*, 34*, β | 2, 4*, 6*, 14, 16*, 21, 25 | 11*, 12, 20* |
Factor 5 (3.084, 8.8) | Factor 6 (2.476, 7.1) | Factor 7 (3.412, 9.7) | Highly specific |
8, 10, 13, 15* | -2, α* | 5, 7*,9*, 17* | 3, 18, 19, 23, 24, 26, 27, 28 |
Traits included in factor loadings with weights higher than 0.5 are shown; |
* indicates the traits that had weights higher than 0.7. |
The eigenvalue and percent variance accounted for by a respective factor are shown in parentheses. |
Based on the traits with the greatest factor loadings, the factors were classified as follows.
Factor 1 characterized the variation in the width of the apodeme outline;
Factor 2 characterized the variation in the traits related to the apodeme bend;
Factor 3 included the most distinct species-specific traits and characterized the variation in (a) the shape of the ventral bend in the distal part of the aedeagus outline, (b) the position of the dorsal bend of the aedeagus outline relative to the central point, and (c) paramere lengths;
Factor 4 characterized the variation in traits related to the ventral bend in the proximal part of the aedeagus;
Factor 5 characterized the variation in traits related to the height of the dorsal bend of the aedeagus outline;
Factor 6 characterized the variation in the cook angle and the apical part of the aedeagus outline; and
Factor 7 characterized the variation in traits related to the dorsal bend in the distal part of the aedeagus outline and the lowermost point of the ventral bend in the same aedeagus part. IMP 3, 18, 19, 23, 24, 26, 27, and 28 were highly specific (Table 1, Highly specific).
Generalized Estimate of the Dependence of the Revealed Variability of the Phenotype of F1 and Fb Hybrids on the Genotype
RDA was carried out to relate the set of the variables that determine the factor structure values to the variables that characterize the genotype. In the analysis, the ordination of the genotypes and factors was performed in the space defined by correlations between factor weights, which were weighted sums of trait weights, and linear combinations of genotype scores, which were obtained for the genotypes characterizing the respective point in trait distribution. The center of gravity of the genotype distribution was brought into coincidence with the centers of gravity of the distribution of the factor structures. As is seen from Fig. 1, a two-dimensional space is defined by the orthogonal factor pair ML3–ML4 (X) and ML6, ML5–ML2, ML7 (Y). The genotypes showed the following arrangement in the two-dimensional factor space: +X: AAAH, ABAH, +Y(− X): BBBLu, AAHH, −Y(− X): ABHLu, BAHVi (genotype abbreviations are as in Table 5). Genotypes ABHH and AAAVi occur in the region of medium values. The angle between vectors on the plot and between genotype positions is proportional to the correlation between them. Therefore, the variation in traits of backcross males homozygous for the autosomes (with the D. virilis X chromosome) positively correlates with Factor 3 and negatively correlates with Factor 4. The finding indicates that the Y chromosome insignificantly affects the traits that determine the structures of the two factors. The contribution of the Y chromosome to the respective traits is somewhat greater in the backcross genotypes heterozygous for the autosomes but is still incomparable with the contributions that the Y chromosome makes to the traits in the other genotypes. Positive correlations with Factors 6 and 5 and negative correlations with Factors 2 and 7 were observed for traits of F1 males; opposite correlations, for traits of D. lummei males and backcross males with the D. virilis sex chromosomes. It is clear that combinations of the sex chromosomes in genotypes heterozygous or homozygous for the D. lummei autosomes and the male parent identity play a crucial role in the traits involved in the respective factor structures.
ML1–7 are the variation vectors of Factors 1–7, which were obtained by the maximum likelihood method. The variables (traits) that constitute the X axis (in the order of decreasing factor loading) are: IMP6 (0.922), IMP4 (0.769), IMP16 (0.732), IMP11 (0.918), IMP20 (0.729). The variables (traits) that constitute the second Y axis are: IMP35 (0.890), IMP15 (0.832), IMP32 (1.039), IMP34 (0.9), IMP30 (0.843), IMP17 (0.831), IMP9 (0.739), IMP7 (0.718). The least significant factors are in italics; factor loadings are shown in parentheses. The genotypes are abbreviated as in Table 5. The chromosomes and paternal genotype are indicated in the following order: X chromosome, Y chromosome, autosomes, male parent identity.
The results are supported by the distribution of the genotypes on a scatter plot (Suppl. Figure 1) of the two first principal components, which are similar in loading structure to Factors 6 and 3. The parental genotypes are in opposite corners of the plot, having the lowest values in the case of D. virilis and the highest in the case of D. lummei. The cloud of D. virilis data overlaps the clouds of Fb males homozygous for autosomes regardless of the Y-chromosome status. The other genotypes are distributed between the D. lummei cloud and the cloud of D. virilis with homozygous Fb males; F1 males and Fb males heterozygous for autosomes differ in variation and form slightly overlapping clouds. Different genotypes differ in how close their variation parameters are to the parameters of the parental species at different principal component axes. Again, trait expression depends on the combination of the sex chromosome status, the autosome status, and the male parent identity as an epigenetic factor.
The Dominance of Parental Phenotypes by Quantitative Traits of the Shape of the Copulative Organ of D. virilis and D. lummei in Male Offspring of F1 and Fb
Two approaches were used to more precisely evaluate the effect that the sex chromosomes exert on trait expression in the shape of the male copulatory system. First, ANOVA was performed to compare the dominance of parental species-specific phenotypes in progenies from reciprocal crosses between D. virilis and D. lummei and backcrosses of F1 males with D. virilis females. All genotypes had the same autosome set and differed in sex chromosome combination and the identity of the male parent (the original parental species or the F1 hybrid). Second, ANOVA and post-hoc tests were performed using the genotype at the sex chromosomes, the genotype at the autosomes, and combinations of these factors, including the male parent identity, as independent variables.
The phenotypes of males obtained in direct and reciprocal crosses and backcross males heterozygous for all autosomes were compared with the phenotypes of males of the parental species. The results are summarized in Table 2. The logic of estimating the degree of dominance for a trait has been described previously [68]. Based on the distribution of the hybrid and parental genotypes over groups isolated by post-hoc comparisons, it is possible to identify the following variants: incomplete dominance, the dominance of the D. virilis or D. lummei phenotype, and lack of difference among all phenotypes. The resulting data clustering variants were ranged by the degree of phenotype dominance. All cases where a genotype in question appeared together with the parental genotypes in one group were considered to suggest no significant difference (ns) in evaluating the degree of dominance. The variants f1(b) < l < v, v < l < f1(b), l,f1(b) < v, v < f1(b),l, v ≤ f1(b),l, and l,f1(b) ≤ v suggested dominance of the D. lummei phenotype (DLu); the variants l < f1(b) < v, l ≤ f1(b) ≤ v, v ≤ f1(b) ≤ l, and v < f1(b) < l, intermediate dominance (ID); and f1(b) < v < l, l < v < f1(b), f1(b),v < l, l < f1(b),v, l ≤ v,f1(b), and f1(b),v ≤ l, dominance of the D. virilis phenotype (DVi). The Gabriel test and Tukey’s HSD test yielded similar results; only those of the Gabriel test are shown.
Table 2
Dominance of the copulative system shape-related traits as dependent on the sex chromosome composition in D. virilis/D. lummei hybrid males, heterozygous for the autosomes.
Factor | Sign | ♂Vi x ♀Lu | ♂Lu x ♀Vi | ♂F1(♂Lu x♀Vi) x ♀Vi | ♂F1(♂Vi x♀Lu) x ♀Vi |
F1 XLuYVi | F1 XViYLu | Fb XViYLu | Fb XViYVi |
Dx | P.-h. | Dx | P.-h. | Dx | P.-h. | Dx | P.-h. |
F1 | IMP33 | DVi | l < v,f1 | DVi | l ≤ f1,v | DVi | l ≤ fb,v | ns | fb,l,v |
F2 | IMP30 | DLu | f1,l < v | DLu | l,f1 ≤ v | DLu | fb,l < v | DLu | l,fb≤v |
F2 | IMP32 | DVi | f1,v ≤ l | ns | f1,v,l | DVi | fb≤ v,l | DVi | fb,v ≤ l |
F2 | IMP34 | DVi | f1 < v < l | ns | v,f1,l | DVi | fb,v < l | ns | v,fb,l |
F2 | beta | DLu | l,f1 < v | DVi | l < v,f1 | DLu | l,fb<v | ID | l ≤ fb<v |
F3 | IMP4 | ns | f1,l,v | DLu | f1,l < v | ns | fb,l,v | DLu | fb,l < v |
F3 | IMP6 | DLu | l,f1 < v | DLu | l,f1 < v | ID | l,fb<v | DLu | l,fb<v |
F3 | IMP14 | ID | l < f1 < v | ID | l < f1 < v | ID | l < fb<v | DLu | l,fb<v |
F3 | IMP16 | ID | l < f1 < v | ID | l < f1 < v | ID | l < fb<v | ID | l ≤ fb<v |
F3 | IMP21 | ns | l,f1,v | DVi | l ≤ f1,v | ns | l,fb,v | ns | l,fb,v |
F3 | IMP25 | ID | l < f1 < v | DVi | l < f1,v | ID | l < fb<v | DLu | l,fb<v |
F4 | IMP11 | DLu | f1,v < l | DVi | f1,v ≤ l | ns | v,fb,l | DLu | v ≤ l,fb |
F4 | IMP20 | DLu | v < f1,l | DVi | v,f1 < l | DVi | fb,v < l | ID | v ≤ fb≤l |
F5 | IMP8 | DVi | l < v,f1 | DVi | l < v,f1 | DVi | l < v,fb | ID | l ≤ fb≤v |
F5 | IMP10 | DLu | l,f1 < v | DLu | l,f1 < v | ID | l ≤ fb≤v | ID | l ≤ fb≤v |
F5 | IMP13 | DVi | l < v,f1 | DVi | l < v,f1 | ID | l ≤ fb≤v | DLu | l,fb<v |
F5 | IMP15 | DVi | l ≤ v,f1 | DVi | l ≤ v,f1 | DVi | l ≤ v,f1 | ns | fb,l,v |
F3,6 | IMP2 | DVi | f1,v < l | DVi | v,f1 ≤ l | ns | fb,v,l | ns | v,fb,l |
F6 | alpha | DVi | l < v < f1 | DVi | l < v < f1 | ID | l < fb<v | ID | l < fb<v |
F7 | IMP5 | DLu | f1,l ≤ v | DLu | f1,l ≤ v | DLu | fb,l < v | ns | l,fb,v |
F7 | IMP7 | DVi | v,f1 ≤ l | DVi | v,f1 ≤ l | DVi | v,fb≤l | DVi | fb,v ≤ l |
F7 | IMP9 | DVi | v,f1 ≤ l | ID | v ≤ f1 ≤ l | DVi | v,fb<l | DVi | v,fb<l |
F7 | IMP17 | DVi | f1,v < l | ID | v < f1 ≤ l | DVi | fb,v < l | DVi | v,fb<l |
CS | IMP3 | ID | v ≤ f1 ≤ l | DLu | v < f1,l | DLu | v < l,fb | DLu | v < l,fb |
CS | IMP18 | ID | l ≤ f1 ≤ v | DLu | f1,l < v | ID | l ≤ fb≤v | DVi | l < v,fb |
CS | IMP19 | ID | l ≤ f1 ≤ v | ID | l ≤ f1 ≤ v | ID | l ≤ fb≤v | DVi | l < fb,v |
CS | IMP23 | ns | l,f1,v | ns | l,v,f1 | ns | l,v,fb | DVi | l ≤ v,fb |
CS | IMP24 | ID | l ≤ f1 ≤ v | ID | l ≤ f1 ≤ v | DLu | l,fb<v | DLu | l,fb<v |
CS | IMP27 | DVi | l ≤ f1,v | DVi | l ≤ f1,v | DVi | l ≤ fb,v | DVi | l ≤ fb,v |
CS | IMP28 | DLu | f1,l ≤ v | DLu | l,f1 ≤ v | DLu | l,f1 ≤ v | DLu | l,fb≤v |
Traits are grouped according to their maximal weights in the factors extracted isolated by the maximum likelihood method (Table 5); CS, highly specific trait; Dx, dominance: DLu, dominance of the D. lummei phenotype; DVi, dominance of the D. virilis phenotype; ID, intermediate dominance; ns, a nonsignificant difference. Crosses and male genotypes are specified in the two uppermost rows. P.-h., results of post-hoc comparisons: f1, fb, l, v are the mean values of a respective trait in the F1 progeny, backcross progeny, and D. lummei and D. virilis parental males, respectively; the symbols are arranged in the order of increasing trait values. Symbols separated with a comma correspond to a group obtained by pooling samples homogeneous in variance. |
IMPs 12, 22, 26, 29, and 31 did not significantly depend on the genotype and are not included in Table 4. The IMPs that showed significant correlations were grouped. We describe the most general results of the analysis of variance. First, traits at which the D. virilis phenotype dominated (43) prevailed over traits with dominance of the D. lummei phenotype (32) and traits with intermediate dominance (26) in all four samples. Dominance of the same phenotype regardless of the sex chromosome combination was observed for only 5 (IMPs 7, 16, 27, 28, and 30) out of the 30 IMPs included in the analysis, suggesting a substantial role of the autosome combination and the identity of the male parent in trait expression. The effect of the male parent identity was evaluated by comparing F1 males with genotype XViYLu and backcross males with the same genotype, given that crossing over is absent in males and the males in question were entirely identical in chromosome composition. Differences in parental phenotype dominance were observed for 16 out of the 30 traits, and the dominance character changed to the opposite one in the case of apodeme declination angle. A substitution of the D. lummei Y chromosome for its D. virilis counterpart in backcross males similarly changed the dominance character in 16 traits. Of these, ten traits, which mostly loaded on Factors 1, 3, 4, and 5, showed a change to dominance of the opposite species-specific phenotype relative to the species origin of the Y chromosome. Phenotypic comparisons of F1 progenies showed that simultaneous substitution of the sex chromosomes changed the character of dominance at 12 traits, of which three (IMP 11, IMP20, and β) changed their dominance status to the opposite one, according to the species origin of the X chromosome. Other changes were less distinct and included firstly, intermediate dominance was shifted toward dominance of one of the parental species. Secondly, the phenotype was changed relative to that of the parental species so that a homogeneous variance group within one of the parental species and offspring with a particular genotype was replaced with a group that included offspring with an alternative genotype and both of the parental species (Table 2, category “ns”). It is of interest to note that the greatest difference in the total number of traits showing a dominance of one of the parental phenotype was observed in backcross flies heterozygous for the autosomes. Thus, Fb XViYVi males had eight traits at which the D. virilis phenotype dominated and ten traits at which D. lummei phenotype dominated, while Fb XViYLu males displayed an opposite pattern and had ten and five such traits, respectively.
As expected, the number of traits at which the D. virilis phenotype dominated increased in backcross males homozygous for the D. virilis autosomes (Suppl. Table 2); the set included 19 traits in Fb XViYVi males and 23 traits in Fb XViYLu males. A substitution of the Y chromosome changed the character of dominance at nine traits, of which seven again demonstrated a negative effect of the species identity of the Y chromosome on the phenotype.
Factorial MANOVA was carried out to evaluate the effects of the sex chromosomes, autosomes, paternal genotype, and their interactions on the phenotypic traits. Significant effects were observed for each of the four factors taken alone and the interaction of the Y chromosome and autosomes (Suppl. Tabl. 3).
To study the effect on particular phenotypic traits for each of the factors, MANOVA was performed with forced incorporation of all four predictors and the effect of the ChrY*Aut interaction. The results are summarized in Suppl. Table 4. The majority of the traits showed an association with the identities of the X chromosome and autosomes at a significance level p < 0.05; half of the traits were affected by the Y chromosome and male parent identities; and one-third of the traits, by the interaction of the Y chromosome and autosomes. A cooperative effect of these variables was mostly observed for the groups of traits extracted by the factor analysis. An analysis of the factor values confirmed the effect in the cases where the majority of the traits determining the respective factor structure significantly depended on the given variable. Note that a dependence on the hereditary factors was not confirmed for the apodeme width-related traits (F1). The apodeme declination and curvature (F2) depend on the identities of the Y chromosome and male parent. Distinct species-specific traits of the aedeagus and parameres (F3) depend on the X chromosome, autosomes, and male parent identity. The shape of the ventral bend in the proximal part of the aedeagus outline (F4) is determined by the interaction between the Y chromosome and autosomes. Traits related to the height of the dorsal bend in the aedeagus outline (F5) and the shape of the cook and the outline bend over the cook (F6) showed a similar dependence on the Y chromosome, autosomes, and male parent identity; an additional dependence on the X chromosome is specific to F5-related traits. Traits describing the shape of the distal part of the outline (F7) depend on the autosomes, male parent identity, and their interaction.
The effect of the male parent identity is possibly mediated through the epigenetic marking of chromosomes in interspecific hybrids during gametogenesis and a subsequent effect of the resulting signatures on the ontogeny of offspring. As mentioned above, an independent effect of the Y chromosome is expected to be insignificant based on the composition and functional activity of its coding sequences. It is possible to assume that interactions of the two factors play a crucial role in the phenotype.
The role of the components of variability and their combinations in the inheritance of traits of the shape of the male copulative organ
To detail the strength and direction of the effects exerted by the hereditary factors and their combinations on the dominance of the D. virilis phenotype, post-hoc tests were used to evaluate the probability for groups homogeneous in trait variance to form according to a published model [69] (Suppl., Table 5). The results obtained for each particular trait are provided in Supplementary; data on latent and highly specific traits are summarized in Table 3.
Table 3
Paired permutation test with the Bonferroni correction to evaluate the optimal genotype partitioning into groups homogeneous in latent variables and highly specific variables (up to 100 000 iterations in each case).
F# | | X→Y (epist) [A1] | AUT (add) [A2] | P→AUT(add) + P→Y [A3] | Y→AUT (rec.epist) + X→AUT (rec.epist) [A4] | Y + Y→AUT(dom.epist) [A5] | P→AUT(add) + P→X[A6] | X→AUT (dom.epist) + X + AUT (dom) [A7] |
F1 | Prob. | 0.493 | 0.476 | 0.302 | 0.638 | 0.136 | 0.473 | 0.304 |
Dom. | n.s. | n.s. | n.s. | n.s. | n.s. | n.s. | n.s. |
F2 | Prob. | 0.113 | 0.207 | 0.000 | 0.057 | 0.038 | 0.000 | 0.000 |
Dom. | n.s. | n.s. | *(X−chr) | n.s. | + | +sDLuX−chr | +*(sDViX−chr) |
F3+ | Prob. | 0.001 | 0.000 | 0.000 | 0.000 | 0.285 | 0.000 | 0.000 |
Dom. | + | +DLu | + PY, DLuAut | *(sDViAut) | n.s. | +* | +*sDLuX−chr |
F4 | Prob. | 0.697 | 0.489 | 0.000 | 0.738 | 0.000 | 0.000 | 0.000 |
Dom. | n.s. | n.s. | +PY | + n.s. | + | +* | +*sDLuX−chr |
F5 | Prob. | 0.000 | 0.000 | 0.001 | 0.000 | 0.025 | 0.043 | 0.051 |
Dom. | + | +* | +* DLuAut | + ID | + | n.s. | n.s. |
F6 | Prob. | 0.478 | 0.000 | 0.000 | 0.857 | 0.000 | 0.000 | 0.000 |
Dom. | n.s. | + DVi | + DViAut (X−chr) | n.s. | + | + DViAut | + ID |
F7 | Prob. | 0.542 | 0.000 | 0.002 | 0.028 | 0.031 | 0.045 | 0.001 |
Dom. | n.s. | + sDVi | +* DViAut | +* | + | n.s.(sDLuX−chr) | + DViAut |
FDR | 0.05 | 0.021 | 0.043 | 0.036 | 0.036 | 0.043 | 0.029 | 0.021 |
0.01 | 0.004 | 0.009 | 0.007 | 0.004 | 0.006 | 0.006 | 0.004 |
Imp 3 | Prob. | 0.000 | 0.000 | 0.000 | 0.000 | 0.009 | 0.811 | 0.123 |
Dom. | + | + DLu | + DLuAut | + DViX−chr | + | n.s. | n.s. |
Imp 18 | Prob. | 0.000 | 0.000 | 0.009 | 0.000 | 0.008 | 0.222 | 0.085 |
Dom. | + | + DLu | + PAut | + DViX−chr | + | n.s. | n.s. |
Imp 19 | Prob. | 0.000 | 0.000 | 0.002 | 0.000 | 0.000 | 0.302 | 0.009 |
Dom. | + | + ID | + DViAut | + ID | + | n.s. | + DViAut |
Imp 23 | Prob. | 0.385 | 0.004 | 0.002 | 0.000 | 0.412 | 0.033 | 0.002 |
Dom. | n.s. | + DVi | +* DViAut | +* | n.s. | n.s. | + ID |
Imp 24 | Prob. | 0.010 | 0.004 | 0.004 | 0.003 | 0.000 | 0.000 | 0.016 |
Dom. | + | + DVi | + DViAut | + ID | + | + DViAut | +* sDViAut |
Imp 26 | Prob. | 0.055 | 0.000 | 0.145 | 0.000 | 1.000 | 0.239 | 0.120 |
Dom. | n.s. | + DLu | n.s. | + DViX−chr | n.s. | n.s. | n.s. |
Imp 27 | Prob. | 0.306 | 0.000 | 0.004 | 0.000 | 0.253 | 0.054 | 0.002 |
Dom. | n.s. | + DLu | +*X−chr | +* sDViX−chr | n.s. | n.s. | + DLuX−chr |
Imp 28 | Prob. | 0.020 | 0.538 | 0.002 | 0.042 | 1.000 | 0.000 | 0.067 |
Dom. | + | n.s. | + DLuAut | n.s. | n.s. | + DLuX−chr | n.s. |
Designations of the factors - as in Table 5. + - a significant effect of the hereditary factor associated with a linear relationship with the given classes of genotypes; + * - a significant effect of the hereditary factor associated with non-linear dependence with the given classes of genotypes; (s) DVi (Lu) – dominance (overdominance) of the D. virilis (D. lummei) phenotype in genotype classes with an intermediate value of the indicator variable; PY (Aut) - epigenetic effect of the father origin factor on the Y chromosome (autosomes); Aut, X-chr, Y-chr - autosomes, X- or Y-chromosome, which determine the dominance of the phenotype. F# - the number of the latent trait, according to Table 5; F#* - permutation test values according to the values of the factors (latent traits). CH - signs with a high characteristic. FDR - false discovery rate for the latent traits. (X-chr) - unaccounted influence of the X chromosome on the genotype of males F1 XLuYVi under the influence of the hereditary factor P→Y + P→AUT(add) (explanation in the text); (sDViX-chr (Aut)) - changes in the phenotype of a latent trait in classes with an intermediate and maximum values of indicator variables coincide, but are statistically significant only for the former. Prob. – probability, Dom. – domination. |
The model implies that each hereditary factor exerts a discrete effect on the traits in question, depending on the genotype. The expected effect of each factor was analyzed using the indicator variables listed in Table 5. |
Factor X→Y(epist) has two categories for the A1 indicator variable, the effect being present or absent. A significant difference in these categories between samples confirms the significant effect of the interaction between the sex chromosomes. The effect of the conspecific D. virilis sex chromosomes is examined versus the effects of all other combinations in this case.
Factor Y + Y→AUT(dom.epist) has two similar categories for the A5 indicator variable. The independent effect of the Y chromosome alone on the traits is close to zero. Therefore, a significant difference in the categories between samples confirms the significant effect of the interaction between the Y chromosome and autosomes. The effect of the D. virilis Y chromosome and autosomes is examined versus the effects of all other combinations in this case.
With all subsequent factors, a significant difference between the extreme variants confirms that the respective factor significantly affects the trait in question. A significant difference between intermediate and other genotype groups in the absence of differences between the extreme groups indicates that the effect in question depends nonlinearly on the genotype.
Factor AUT(add) has three categories of the A2 indicator variable: two opposite effects of the D. virilis and D. lummei homozygous autosomes and an intermediate effect of the heterozygous autosomes. An intermediate effect suggests a decrease in additive interactions of divergent genes, and the lack of significant difference from one of the homozygous genotypes indicates that dominant alleles mostly contribute to the trait expression.
Factor Y→AUT(rec.epist) + X→AUT(rec.epist) has three categories of the A4 indicator variable: an effect of interactions between the homozygous D. virilis autosomes and the D. virilis sex chromosomes, a partial effect of interactions between the homozygous D. virilis autosomes and the X chromosome, and all variants with the heterozygous autosomes or homozygous D. lummei autosomes. A significant difference between the extreme variants indicates that the sex chromosomes interact with recessive autosomal genes, leading to species-specific distinctions. Lack of a significant difference from one of the extreme variants suggests a predominant contribution of epistatic interactions with one of the sex chromosomes, depending on the genotype combination (dominance of the D. virilis phenotype suggests a leading role for the X chromosome; dominance of the D. lummei phenotype, for the D. virilis Y chromosome, which is absent in 1.9.30 males, of the D. lummei Y chromosome present in the given genotype).
Factor P→X + P→AUT(dom) has three categories of the A6 indicator variable: the minimal and maximal values correspond, respectively, to the absence or presence of interactions of the D. virilis X chromosome and autosomes with the homozygous paternal genotype, and an intermediate value corresponds to interactions of dominant genes of the D. virilis autosomes with the D. virilis homozygous paternal genotype. A grouping of genotypes having intermediate values with those having one of the extreme values maximizes the role of the autosomes or the X chromosome, depending on the group composition (dominance of the D. virilis phenotype suggests a maximal role for the autosomes; dominance of the D. lummei phenotype, for the D. virilis X chromosome, which is absent in (PVi)F1 XLuYViAViALu males).
Factor X→AUT(dom.epist) + X + AUT(dom) has three categories of the A7 indicator variable. The minimal and maximal values correspond to expression of the D. virilis phenotype due to the effects of dominant autosomal genes and the D. virilis X chromosome and their epistatic interactions. The intermediate value corresponds to a sole effect of dominant autosomal genes. Dominance of the D. virilis phenotype (a grouping of genotype XLu/YVi AutVi/AutLu with genotypes XVi/Y*_AutVi/Aut*) suggests a predominant effect for D. virilis dominant autosomal genes; dominance of the D. lummei phenotype, for the D. lummei X chromosome, D. lummei autosomes, or their combination.
Factor P→AUT(add) + P→Y has four categories of the A3 indicator variable. The extreme values define the effects that the paternal genotype exerts on the expression of the D. virilis phenotype under the influence of recessive autosomal alleles and the Y chromosome. The two alternative extreme values accordingly belong to the genotypes of the D. virilis and D. lummei parental strains. The intermediate values of indicator variables are defined by the epigenetic effect that the D. virilis male parent identity exerts exclusively on the D. virilis Y chromosome (genotype (PVi)F1 XLuYViAViALu, the value is 1) and by lack of this effect on the genotypes that are heterozygous for the autosomes and have the Y chromosome originating from a heterozygous male or a D. lummei male (genotypes (PLu)F1 XViYLuAViALu, (PVi/Lu)Fb XViYViAViALu, (PVi/Lu)Fb XViYViAViAVi, (PVi/Lu)Fb XViYLuAViALu, and (PVi/Lu)Fb XViYLuAViAVi; the value is 0). Lack of a significant epigenetic effect on the autosomes will lead to the clustering of the D. lummei genotype with genotype (PVi)F1 XLuYViAViALu and the D. virilis genotype with genotypes (PLu)F1 XViYLuAViALu, (PVi/Lu)Fb XViYViAViALu, (PVi/Lu)Fb XViYViAViAVi, (PVi/Lu)Fb XViYLuAViALu, and (PVi/Lu)Fb XViYLuAViAVi. If a significant epigenetic or genetic effects of the Y chromosome is lacking, the genotype (PVi)F1 XLuYViAViALu will cluster with genotypes (PLu)F1 XViYLuAViALu, (PVi/Lu)Fb XViYViAViALu, (PVi/Lu)Fb XViYViAViAVi, (PVi/Lu)Fb XViYLuAViALu, and (PVi/Lu)Fb XViYLuAViAVi. If the autosomes exert a dominant effect characteristic of D. virilis or D. lummei in this case, strains with the intermediate values of indicator variables will cluster together with the respective parental genotype. The incomplete design of crosses makes the results difficult to interpret. For example, male genotype (PVi)F1 XLuYViAViALu was the only genotype that had the intermediate value 1 for the indicator variables A3 (factor P→AUT(add) + P→Y), A6 (factor P→X + P→AUT(dom)), and A7 (factor X→AUT(dom.epist) + X + AUT(dom)). It is clear that the phenotypic features of this genotype are determined by combined effects of the three factors, and its clustering with other genotypes may therefore be distorted in the case of factor P→AUT(add) + P→Y.
Thus, the estimates confirm that epistatic interactions of the sex chromosomes and autosomes and epigenetic effects of the male parent origin from interspecific crosses influence the expression of D. virilis species-specific traits in the shape of the male copulatory system.