SEARCH

SEARCH BY CITATION

Keywords:

  • chromosomal rearrangements;
  • courtship;
  • inversions;
  • mate choice;
  • pleiotropy;
  • sexual selection;
  • speciation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The evolution of animal genitalia has gained renewed interest because of their potential roles during sexual selection and early stages of species formation. Although central to understanding the evolutionary process, knowledge of the genetic basis of natural variation in genital morphology is limited to a very few species. Using an outbred cross between phylogenetically distinct lines of Drosophila montana, we characterized quantitative trait loci (QTLs) affecting the size and shape of the distiphallus, a prominent part of the male intromittent organ. Our microsatellite-based linkage analysis shows that intra-specific variation in the distiphallus involves several QTLs of largely additive effect and that a highly significant QTL co-localizes with the same inversion where we have earlier localized a large QTL for a sexually selected courtship song trait. The latter indicates that inversions can play an important role in shaping the evolution of rapidly evolving traits with a potential influence on speciation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Understanding the genetic basis underlying morphological diversification of traits involved in adaptation, and early speciation has been of a longstanding interest in evolutionary biology. Animal genitalia might prove to be particular powerful traits for addressing this question, given their rapid evolutionary change and potential effects on reproductive isolation (e.g. Arnqvist, 1998; Sota & Tanabe, 2008, 2010). Accumulating data suggest that the diversification of genital structures may be primarily driven by sexual selection quite similar to elaborate male secondary sexual traits (Eberhard, 1985, 2010; Arnqvist, 1998; Hosken & Stockley, 2004; Birkhead, 2010), although alternative processes such as drift, pleiotropy (Mayr, 1963; Arnqvist, 1997; House & Simmons, 2005) or reinforcement, i.e. some sort of lock and key between the sexes to avoid inter-species hybridizations, may also play a role (Mikkola, 1992; Sota & Kubota, 1998; Sota & Tanabe, 2010). Contrary to an increasing amount of data addressing the selective forces acting on natural variation in genital morphology (Hosken & Stockley, 2004; Eberhard, 2010; Birkhead, 2010; and refs. therein), surprisingly little is known about the genetic architecture underlying this variation despite its critical role in understanding the evolutionary process. For most taxa, this challenge has become feasible only very recently in parallel with the improvements in molecular methodology, which facilitates the development of sufficient dense genetic linkage maps to localize quantitative trait loci (QTLs).

To our knowledge, detailed information on QTLs affecting variation in male genital morphology is only available for hybrid crosses between species belonging to the Drosophila simulans lineage of the Drosophila melanogaster clade (Liu et al., 1996; Laurie et al., 1997; True et al., 1997; Macdonald & Goldstein, 1999; Zeng et al., 2000) and for one cross between two species of carabid beetles (Sasabe et al., 2010). Hybrid analyses of crosses between D. simulans with D. mauritiana and D. sechellia revealed that species differences in the size/shape of the posterior lobe of the male genital arch are inherited in a largely additive manner involving multiple QTLs with relatively small effect. For example, in a high-resolution multiple interval mapping (MIM) study, Zeng et al. (2000) were able to detect at least 19 QTLs encoding for differences in the genital arch between D. simulans and D. mauritiana, none of which explained more than 11% of the parental species difference. Furthermore, morphological change likely occurred owing to episodes of strong directional selection during the species’ evolutionary history as indicated by the unidirectional signs of the QTL effects (Liu et al., 1996; Laurie et al., 1997; Macdonald & Goldstein, 1999). A polygenetic basis with unidirectional signs of additive QTL effects has also been implicated from a recent study on the size and width of the male intromittent organ in carabid beetles (Sasabe et al., 2010), which raises a question of whether this kind of genetic architecture is common in inter- and intra-specific variation in genital morphology.

Species belonging to the Drosophila virilis group have diverged from the D. melanogaster group about 40 million years ago (Russo et al., 1995). Comparative morphometric analyses documented that the shape of the distiphallus (the most prominent sclerotized structure of the male intromittent organ) has evolved in a highly species-diagnostic manner and is also variable within species (Kulikov et al., 2004). However, the selective process behind this diversification has not yet been studied in detail in any species of the D. virilis group. Functional studies of the mechanics of copulation in the D. melanogaster group provided little support for a lock and key mechanism of isolation, but rather indicate post-copulatory sexual selection (involving traumatic insemination) as main driving force in shaping male intromittent genitalia (Kamimura, 2007, 2010; Polak & Rashed, 2010). Drosophila montana, one species of the D. virilis group, offers a good opportunity to study the genetic architecture underlying naturally occurring variation in the distiphallus at early stages of speciation. The species has been characterized by substantial population genetic variation in inversions, mitochondrial and nuclear DNA sequences with populations from Japan, Finland and Northern America forming clearly distinct genetic entities (Throckmorton, 1982; Mirol et al., 2007). The large phylogenetic distinctness is accompanied by some degree of prezygotic and post-zygotic isolation (Jennings et al., 2011) and by significant inter-population differentiation in several traits linked to reproductive behaviour (Routtu et al., 2007). This in particular holds true for differences in male courtship song and corresponding female preferences (Klappert et al., 2007), but it may also apply to the shape variation in the distiphallus where patterns of divergence were found to be inconsistent with those expected under neutrality most likely owing to episodes of directional or diversifying selection (Routtu et al., 2007).

In this study, we have localized QTLs affecting genitalic variation in the distiphallus by analysing a reciprocal outbred cross between two phylogenetically isolated populations of D. montana. The same cross has been previously used to develop a microsatellite-based linkage map and to successfully localize QTLs encoding for a sexually selected male courtship song component (Schäfer et al., 2010). The present data indicate that intra-specific variation in the male distiphallus has a polygenic basis with largely additive effects and that a major QTL affecting male genital variation co-localizes with a QTL for the song difference in a genomic region containing a polymorphic inversion.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Fly strains and crossing scheme

Because the crossing scheme has been already described in detail (Schäfer et al., 2010), we only provide a brief summary here. Our mapping population was derived from reciprocal crosses between two phylogenetically distinct lines of D. montana (O66, collected during 2003 from Oulanka in Finland and C13, collected during 2003 in the surroundings of Crested Butte in Colorado, USA). The resulting F1 generation was then intercrossed in all four possible combinations to obtain the recombining F2 generation, which also allowed testing for Y-chromosomal effects. Flies were raised in vials containing malt medium in continuous light in a culture room (19 ± 1 °C). After emergence, males and females were kept separately for at least 3 weeks and then stored in 70% ethanol for the genital preparations. The study lines maintained relatively high allelic diversity with a mean number of about 2.4 alleles per locus in both strains. In the linkage analyses, different alleles were pooled within lines, but all microsatellites segregated for alternative alleles between the parental lines.

Genital preparation and morphometrics

The genital preparations followed the protocol given in Routtu et al. (2007). In brief, the distal part of the male abdomen was transferred to Eppendorf tubes containing 0.1 m NaOH. After heating the probes for about 8 min at 95 °C, the NaOH was replaced by water. Then, the distiphallus was manually removed from the genital apparatus, placed on a glass slide and covered with EUKITT (Kindler, Freiburg, Germany). Slides were photographed with a SPOT Insight colour digital camera installed on a Zeiss Axioscop 40 microscope and stored as bitmap files for the morphometric analyses. A photograph of a genital preparation is shown in Fig. 1. Further information on the positioning of the distiphallus relative to other internal as well as to external male genital structures can be found in Bächli et al. (2004).

image

Figure 1.  Side view of the male distiphallus in Drosophila montana. The posterior end of the distiphallus has two hook-like structures (H) where the sperm extrudes during copulation. The scale bar corresponds to 0.1 mm.

Download figure to PowerPoint

We used elliptic Fourier analysis of an enclosed contour to describe shape variation in the distiphallus between the parental and the recombining F2 generation owing to lack of reliable landmarks (Fig. 1). The morphometric analysis was performed with the program SHAPE 1.2 (Iwata & Ukai, 2002) and is based on principal component analysis (PCA) performed on the elliptical Fourier descriptors (EFD, Kuhl & Giardina, 1982). The resulting normalized principal component (PC) scores can be used as trait values that describe the shape variation only. The size component (area) was measured separately with the program. Pictures were edited to black and white to enable automated measuring with the program. Repeatability was high and varied from almost 100 to 94% with smaller PCs showing the lower rep-eatability, respectively. To test for phenotypic differences between the parental strains, we measured the genitalia of 35 males from the O66 strain and of 26 males from the C13 strain, respectively. The QTL analysis was performed on genital measurements of 403 F2 offspring evenly distributed among the four reciprocal QTL crosses (range from N = 96 to N = 114).

Molecular map and QTL analysis

Our final linkage map consisted of 35 microsatellites distributed along the X-chromosome and the four autosomes. To achieve an even physical spacing between markers, we designed microsatellites at specific cytological positions predicted from published correspondence between the D. montana cytological map and the map of the distantly related D. virilis (see Schäfer et al., 2010 and references therein). Markers were designed for the region of interest via cross-amplification using D. virilis scaffolds. Nevertheless, the resulting genetic map indicated large heterogeneity in recombination rates. Whereas the recombination map of the X-chromosome was quite long resulting in two independent linkage groups separated by more than 50 cm, other parts of the genome showed large areas of suppressed recombination. Suppression of recombination on chromosome 2 could be to a large extent attributed to a polymorphic inversion 2Y (Schäfer et al., 2010), which most likely segregates in the Finnish population (Vieira et al., 1997; Morales-Hojas et al., 2007).

Quantitative trait loci were localized using composite interval mapping (CIM), which tests for QTL effects in a given interval while statistically controlling for other QTLs outside that interval (Jansen & Stam, 1994; Zeng, 1994). CIM was performed with model 6 as implemented in the Zmapqtl program of the QTL Cartographer software (Basten et al., 1997). A forward–backward regression procedure was applied to select significant markers as genetic background. We examined the robustness of CIM by using different backgrounds and window sizes (ranging from 2 to 20 cm), which appeared not to have a strong influence on the resolution of the QTLs. We further included a hypothetical, Y-linked marker in the analyses to test for Y-chromosomal effects. As dependent variables, we used the first four major PC axis, which explained the majority (72.51%) of the total phenotypic variation (Table 1) as well as genital area as an independent size measure. Genome-wide significance thresholds were adjusted by 1000 permutations of the trait data against the marker data prior to QTL analysis (Churchill & Doerge, 1994). We chose a constrained additive model to test for statistical significance of X-linked QTLs including all markers and a free model for autosomal QTLs to examine the dominance effects as well. In the free model, the sex-linked markers were excluded prior to the permutation procedure. All QTL analyses were performed using QTL cartographer version 2.5 (Basten et al., 1997) and Map Manager QTX b20 (Manly et al., 2001).

Table 1.   Principle components (PC) extracted from a variance–covariance matrix of elliptic Fourier coefficients describing genital shape variation in the study mapping population of Drosophila montana. Genital shape and size (area) differences between the parental strains were tested using a t-test. The corresponding statistics are given in the last two columns.
ComponentEigenvalueCumulative (individual) percentage of variance explainedt1,58P
  1. *t-test for unequal variances.

PC18.15−0424.63 (24.63)−0.9430.350
PC26.83−0445.29 (20.66)−1.426*0.159
PC36.05−0463.57 (18.28)−2.907*0.005
PC42.95−0472.51 (8.93)6.929*< 0.001
PC51.32−0476.49 (3.99)0.5540.582
PC61.14−0479.93 (3.45)1.8640.067
PC79.56−0582.82 (2.89)−0.3240.747
PC86.84−0584.89 (2.07)−12.638< 0.001
PC95.86−0586.67 (1.77)2.2240.030
PC105.51−0588.33 (1.67)−2.2930.026
PC114.14−0589.58 (1.25)−9.166< 0.001
PC123.92−0590.77 (1.18)−1.0130.315
PC132.97−0591.67 (0.90)2.8760.006
Area−1.8210.074

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Genital shape and size

Similar to previous results (Routtu et al., 2007), the shape analysis of the male distiphallus in D. montana yielded a complicated pattern of correlations between EFDs, resulting in 13 significant PCs. PC 1 and PC 4 showed the largest variance at the terminal ends of distiphallus, PC 2 at the leverage point and PC 3 at the terminal end of the distiphallus hook (Fig. 2). These four PCs cumulatively explained more than 70% of the total variation in genitalia shape, but only two of them, PC3 and PC4, differed significantly between the parental lines (Table 1). Parental lines also differed in PC 8, 9, 10, 11 and 13, but these PCs only explained a minor fraction of the total phenotypic variation and were not used in the QTL analyses. In addition to PCs affecting the shape of male genitalia, we also used male genital area as an independent trait in the QTL analyses.

image

Figure 2.  Outlines of the four major principle components describing shape variation in the male distiphallus. Shapes correspond to the mean and two standard deviations in both directions and are normalized for size differences.

Download figure to PowerPoint

QTLs affecting genital morphology

Composite interval mapping analysis of the first four PCs identified a minimum number of five QTLs affecting shape variation in the distiphallus, which exceeded the genome-wide significance level at P < 0.01 (Fig. 3, Table 2). The five QTLs influencing the shape differences described by PC axes 2 and 4 mapped to chromosomes 2, 3 and 4, respectively (Table 2, Fig. 3), and the hypothetical Y-marker associated with differences in genital size (Table 2, Fig. 3). In addition, there were a number of marginally significant QTLs associated with PC 1, PC 3 and genital area (0.01 < P < 0.05), which mapped to the two linkage groups of the X-chromosome and the fourth chromosome, respectively (Fig. 3). Because the log-likelihood profiles of CIM indicated two closely situated peaks for PC2 on chromosome 3 (Fig. 3), we additionally performed MIM to distinguish between alternative QTL models. MIM has advantages over CIM in resolving QTLs when multiple loci cluster together on the same chromosome (e.g. Mayer, 2005). A model with a single QTL next to marker vir86ms explained as much of the phenotypic variance as a model in which we added a second QTL next to marker vir93ms (see Fig. 3), as further indicated by the Akaike information criterion (ΔAIC = −9.56). Otherwise, MIM and CIM yielded nearly identical results in the position and effect sizes of the putative QTLs as expected by the segregation pattern onto different chromosomes (e.g. Mayer, 2005).

image

Figure 3.  Composite interval mapping of male genital size and shape against marker positions of the two linkage groups of the X-chromosome and the four autosomes. All intervals exceeding the genome-wide significance threshold of P < 0.05 are shown up to the minimum log-likelihood ratio of the flanking intervals, and quantitative trait locus (QTLs) with P < 0.01 are considered significant. Significance thresholds for sex-linked QTL ranged from 9.9 to 10.6 (P < 0.05) and from 15.7 to 16.9 (P < 0.01), whereas the thresholds for the autosomes ranged from 11.9 to 12.7 (P < 0.05) and from 18.0 to 19.9 (P < 0.01), respectively.

Download figure to PowerPoint

Table 2.   Quantitative trait loci (QTL) positions (in cm) and the magnitude of their effects on the size and shape variation in male genital morphology in Drosophila montana. Only QTLs with P < 0.01 are listed.
TraitChromosomePeak positionAdditive effectDominance effectLikelihood ratio% variance explained
Areahyp Y978.7918.205.4
PC2249.80.172−0.00286.5020.91
PC2354.8−0.010−0.00128.487.31
PC4256.90.0070.00138.259.14
PC4367.20.0070.00028.478.45
PC4426.50.0060.00121.363.68

Overall, the QTLs detected showed a largely additive mode of inheritance (Table 2, Fig. 4a,b). Individual loci had only relatively small effects on genital variation (R2 from 5 to 9%), except for one highly significant QTL, which explained 20.91% of the total phenotypic variation described by PC 2 (Table 2) and segregated with a region of suppressed recombination on chromosome 2 so could encompass a number of smaller effect QTLs. Suppression of recombination in the QTL region can be largely explained by the presence of an inversion (2Y), which has earlier been detected to be polymorphic within Finnish, but not within American populations (Vieira et al., 1997; Morales-Hojas et al., 2007). As the same genomic region was previously identified segregating with genetic differences in the carrier frequency of male courtship song (Schäfer et al., 2010), we tested for a correlation between these two traits within the recombining F2 mapping population. Albeit quite weak, the correlation was statistically highly significant (r = −0.215, N = 403, P < 0.001).

image

Figure 4.  Effects of representative microsatellite alleles of significant quantitative trait loci affecting genital shape variation described by PC2 (a) and PC4 (b). Effects are shown for both parental lines (O66; Finland) (C13; Colorado) and for each of the four reciprocal F2 crosses. F2 AA: F1(C13×O66) × F1(C13×O66); F2 AB: F1(C13×O66) × F1(O66×C13); F2 BA: F1(O66×C13) × F1(C13×O66); F2 BB: F1(O66×C13) × F1(O66×C13). The female origin in the F1 crosses and the four F2 crosses is always listed as first, and male origin is listed as second.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The present linkage mapping study, in which we characterized QTLs associated with intra-specific variation in male genital morphology between two geographically isolated strains of Drosophila montana, leads to two main conclusions. First, intra-specific variation in the male distiphallus involves multiple QTLs, which are inherited in a largely additive manner, and second, a highly significant QTL affecting genital shape maps to a polymorphic inversion, which has been previously identified to segregate with genetic differences in a sexually selected courtship song trait between the study lines. In the following, we first relate our findings to other quantitative genetic studies on genitalic and nongenitalic morphological traits and then discuss the implications for different evolutionary scenarios, which can lead to divergence in genital morphology.

Our finding that intra-specific morphological variation in the male distiphallus in D. montana is affected by multiple QTLs of largely additive, but of mostly relatively small effect is qualitatively similar to the results obtained from similar studies on the posterior lobe of the male genital arch within the D. simulans clade (Liu et al., 1996; Laurie et al., 1997; Macdonald & Goldstein, 1999; Zeng et al., 2000). This similarity might not have been necessarily expected a priori, given that we studied a nonhomologous internal genital trait involved in insemination (as opposed to the derived external genital arch in the D. melanogaster group which functions in mounting and coupling with the female, Jagadeeshan & Singh, 2006), and that phenotypic differences between our parental lines were much smaller. Although significant maternal and dominance effects on genitalic traits have been reported (House & Simmons, 2005; Sasabe et al., 2007, 2010), the available data so far clearly indicate a polygenetic basis with sufficient additive variation for morphological change if exposed to selection (Arnqvist & Thornhill, 1998; Simmons et al., 2009).

Although the arthropod genitalia appear to have diverged rapidly in their shape and complexity, they are usually highly genetically canalized in size as reflected by negative (hypoallometric) scaling relationships with other morphological traits (Eberhard et al., 1998; Eberhard, 2009). We found the shape differences in the male distiphallus to be influenced by at least five highly significant QTLs located on the autosomes, whereas the genital size only segregated with the hypothetical Y-marker and a marginally significant X-linked locus (Fig. 3). The association of genital area with the hypothetical Y-marker is interesting in light of ongoing debates about the potential importance of Y-linkage for the establishment and maintenance of sexually antagonistic genetic variation (e.g. Rice, 1987; Lindholm et al., 2004; Fry, 2010) but should ideally be re-tested in an isogenic background to establish its significance. Irrespective of the detailed genetic nature, our findings clearly show that shape and size components are regulated (at least to some degree) independently and that the genetic basis of the shape is more ‘complex’ than that of the size. When studying the genetic basis of wing vein positioning in D. melanogaster,Zimmerman et al. (2000) found that many more QTLs affect wing shape than wing size and that the two descriptors are affected by independent loci. The authors concluded that wing shape is regulated more locally through the regulation of the length and positioning of wing veins, whereas wing size may involve regulation in the activity of fewer general growth factors (see also Mezey et al., 2005). Studies on the mandible and molars of mice yielded a similar picture (Cheverud et al., 1997; Klingenberg et al., 2001; Workman et al., 2002). For example, Workman et al. (2002) identified 16 distinct QTLs encoding for molar shape but only three for centroid size. They argued that because overall size is likely to be controlled by the endocrine system, the genetic basis of molar size is ‘simpler’ than that of shape, which again might be regulated more locally. In light of this argument it seems plausible that the predominance of QTLs with effects on shape of the distiphallus in D. montana has a similar explanation. Alternatively, shape components might simply unravel a higher number of QTLs because multivariate shape measures potentially capture more complex and precise aspects of morphology than simple linear size measures do (Rohlf & Marcus, 1993).

The genetic architectures of phenotypic differences between species and subspecies are likely to reflect the selective processes that have acted on the trait during the past. For instance, episodes of strong unidirectional selection are predicted to produce unidirectional signs of QTL effects conditioned on the phenotypic difference between the parental strains (Orr, 1998, 2001). Although comparative work provided strong evidence that directional (sexual) selection caused the genitalic diversification between D. simulans group species (Laurie et al., 1997; Macdonald & Goldstein, 1999; Zeng et al., 2000) and likely also between species of carabid beetles (Sasabe et al., 2010), our data remain less conclusive in this respect. The parental strains only differed statistically in PC 4, but not in PC 2 for which two highly significant QTLs were identified and one of them was rather transgressive in sign (Table 2, Fig. 3). Although we cannot reject neutrality based on the present data, this of course does not rule out any role of directional selection on some QTLs, nor does it contradict many other forms of selection such as stabilizing selection or cyclic changes in direction of selection as predicted for some models of male–female co-evolution (Iwasa & Pomiankowski, 1995; Orr, 2001). For example, sexual conflict has influenced the evolution of variety of genitalic traits in the arthropods (e.g. Arnqvist & Rowe, 2002; Rönn et al., 2007; Kuntner et al., 2009) including Drosophila species (Kamimura, 2007, 2010; Polak & Rashed, 2010). Behavioural data indicate intense sexual conflict over copulation in D. montana (Mazzi et al., 2009), which might have contributed to the morphological differentiation of the distiphallus as seen between allopatric populations of D. montana (Routtu et al., 2007).

Our finding that the QTL with the strongest effect on genital shape co-localizes with a genomic area containing an inversion polymorphism, which is also associated with a sexually selected courtship song trait within the mapping population, may reflect the fact that the chances of finding QTLs are increased in low recombining regions. QTL studies will integrate over loci of individually small, but cumulatively large effect within inversions (Noor et al., 2001). There are several reasons why traits with prospective roles in adaptation and speciation often co-localize with inversions (reviewed in Rieseberg, 2001; Hoffmann et al., 2004; Kirkpatrick & Barton, 2006; Hoffmann & Rieseberg, 2008). One implication is that genes affecting genital shape cannot evolve completely independent from other genes within the inversion, potentially encoding for a broad range of traits. Selection on the inversion itself or associated traits (e.g. courtship song) may lead as a by-product to correlated changes in genital morphology in analogy to the early pleiotropy hypothesis of genitalic evolution (Mayr, 1963). Whatever mechanism involved, the co-localization of QTLs for two sexual traits in a genomic region containing an inversion polymorphism points the potential importance of areas of low recombination in the diversification of rapidly evolving traits that can influence speciation.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This work was funded by the EU Research Training Network ‘Co-evolved traits’ (HPRN-CT-2002–00266). M.S. was supported by a scholarship (SCHA 1502/2-1) of the Deutsche Forschungsgemeinschaft (DFG) during the preparation of the manuscript. We further thank members of the EU RTN for helpful comments on the project and Erik Postma for fruitful discussions on the manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Data deposited at Dryad: doi: 10.5061/dryad.c10q7