Selection for alternative male reproductive tactics alters intralocus sexual conflict Agata Plesnar-Bielak1, Anna M. Skrzynecka1, Krzysztof Miler1, Jacek Radwan2
1
Institute of Environmental Sciences, Jagiellonian University, Gronostajowa 7, 30-387 Kraków, Poland
2
Institute of Environmental Biology, Faculty of Biology, Adam Mickiewicz University, Poznań, Poland
Corresponding author: Agata Plesnar-Bielak, e-mail: [email protected], tel.: (+48) 12 664 51 51, fax: (+48) 12 664 69 12
Full title: Selection for alternative male reproductive tactics alters intralocus sexual conflict. Running title: Intralocus sexual conflict and male morphs. Keywords: intralocus sexual conflict, sexually antagonistic selection, genetic correlation, alternative reproductive tactics, male morphs. Word count: 4057. The manuscript contains 1 table and 2 figures plus 1 supplementary figure. Data available from the Dryad Digital Repository: doi:10.5061/dryad.s4t7m
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/evo.12409. This article is protected by copyright. All rights reserved.
1
Abstract Intralocus sexual conflict (IASC) arises when fitness optima for a shared trait differ between the sexes; such conflict may help maintain genetic variation within populations. Sex-limited expression of sexually antagonistic traits may help resolve the conflict, but the extent of this resolution remains a subject of debate. In species with alternative male reproductive tactics, unresolved conflict should manifest more in a more sexually dimorphic male phenotype. We tested this prediction in the bulb mite (Rhizoglyphus robini), a species in which aggressive fighters coexist with benign scramblers. To do this, we established replicated lines in which we increased the proportion of each of the alternative male morphs using artificial selection. After approximately 40 generations, the proportion of fighters and scramblers stabilized at >0.9 in fighter- and scrambler-selected lines, respectively. We then measured several female fitness components. As predicted by IASC theory, female fecundity and longevity was lower in lines selected for fighters and higher in lines selected for scramblers. This finding indicates that sexually selected phenotypes are associated with an ontogenetic conflict that is not easily resolved. Furthermore, we suggest that IASC may be an important mechanism contributing to the maintenance of genetic variation in the expression of alternative reproductive tactics.
Keywords: intralocus sexual conflict, sexually antagonistic selection, genetic correlation, alternative reproductive tactics, male morphs.
This article is protected by copyright. All rights reserved.
2
Introduction Although the sexes demonstrate a high degree of overlap in their genomes and many traits are expressed in both males and females, fitness optima for a number of shared traits may be noticably sex-specific (Glucksman 1981, Rice 1984, Chippindale et al. 2001). In particular, traits that increase fitness when expressed in one sex may decrease fitness when expressed in the other sex. These differences may arise because the ecological niches of the sexes differ (e.g., Shine 1989, Forsman 1995, Merilä et al. 1997) or because some traits may be under strong sexual selection in males but not females (e.g., Chippindale et al. 2001, Björklund & Senar 2001, Price and Burley 1994, Robinson et al. 2006). This kind of antagonism, called intralocus sexual conflict (IASC), is associated with sex-specific differences in the direction of selection on allelic variants at a locus and leads to negative fitness correlations between the sexes for a given trait (Lande 1980). IASC has been observed in many species of different taxa (see Rice & Chippindale 2001, Bonduriansky & Chenoweth 2008, and van Doorn 2009 for review), although its generality and evolutionary importance are still debated (e.g., Cox & Calsbeek 2009, Bonduriansky & Chenoweth 2008). IASC may potentially have consequences for various evolutionary processes such as speciation (Parker & Partridge 1998, Rice & Chippindale 2002), the evolution of sex chromosomes (Bull 1983, Rice 1987, Charlesworth 1991), the evolution of sex determination (Rice 1986, Kraak & Pen 2002, van Doorn & Kirkpatrick 2007), the regulation of gene expression (Ellegren & Parsch 2007), aging (Vieira et al. 2000, Bonduriansky et al. 2008), sexual selection (Brommer et al. 2007, Pischedda & Chippindale 2006), sex allocation (Alonzo & Sinevo 2007), and the maintenance of genetic variation in natural populations (Chippindale & Rice 2001). Here, we hypothesize that it may also facilitate the maintenance of genetically determined alternative reproductive tactics (see below). However, the evolutionary significance of IASC depends on how easily it can be resolved and, hence, how prevalent it is.
This article is protected by copyright. All rights reserved.
3
Several ways of resolving IASC have been proposed (reviewed in Stewart et al. 2010). In one scenario, a locus subject to antagonistic selection may be duplicated, and one of the duplicated genes might be expressed in males while the other is expressed in females (Partridge & Hurst 1998, Rice & Chippindale 2001, Wyman et al. 2012). IASC may also be resolved by sex-dependent regulation of gene expression (McIntyre et al. 2006) or genomic imprinting (Day & Bonduriansky 2004). As a consequence, antagonistic traits may evolve sexually dimorphic expression that demonstrates sex-specific optima, which is considered to be evidence of IASC resolution (see Cox & Calsbeek 2009; Stewart at al. 2010 and Pennell & Morrow 2013 for review). However, evidence is accumulating that IASC is not easily resolved, as manifested by positive genetic correlations between male and female traits that have opposite fitness effects in the sexes (reviewed by Poissant et al. 2010, see also Cox & Calsbeek 2009) or gene expression levels for such traits (Griffin et al. 2013). For example, sexually dimorphic locomotor activity in Drosophila melanogaster still showed significant inter-sexual genetic correlation despite being selected in different directions in males and females (Long & Rice 2007). In another study, Harano et al. (2010) recently showed that sexlimited trait expression does not fully resolve IASC; their study examined broad-horned flour beetles, a species in which males exhibit enlarged mandibles used in fights with rivals (a trait subject to sexual selection), a feature that is completely lacking in females. The authors showed that females from populations subject to selection for large mandibles (and thus higher fitness) in males had significantly lower fitness than females from control populations not subject to selection. In contrast, selection for small mandibles in males increased female fitness at the cost of male reproductive success. This finding indicates that, despite the sex-limited expression of horns, genes associated with horn elaboration decrease female fitness, which the authors attribute to genetic correlations between sexually antagonistic traits and other features that contribute to fitness. Alternative reproductive tactics (ARTs) are well suited to the study IASC because they typically comprise discontinuous male phenotypes which differ in the degree of sexual dimorphism.
This article is protected by copyright. All rights reserved.
4
In such cases, the more sexually dimorphic ART can be expected to be associated with stronger IASC. ARTs often manifest themselves as male polymorphism, where the morphs differ in reproductive strategies as well as the morphological structures used in male-male competition (reviewed in Brockman 2001, Oliveira et al. 2008). Typically, a morph possessing exaggerated fighting structures is more dominant and aggressive, whereas the other, less sexually dimorphic morph lacks weapons and is more subordinate. Some well-known examples of such dimorphisms include horns in scarab beetles (Emlen 1994, Kotiaho et al. 2003) or leg dimorphism in acarid mites (Woodring 1969, Radwan 1995, 2009); however, males may also differ in their expression of sexual ornaments associated with status and sexual attractiveness (e.g., color morphs in side-blotched lizards (Alonzo & Sinervo 2001) or dominant and subordinate morphs in turkeys (Buchholz 1995)). Higher degree of sexual dimorphism should be associated with higher IASC, but also with higher pressure for IASC resolution. Consistent with this expectation, dominant wild male turkeys (Pointer et al. 2012) as well as armored male beetles (Snell-Rood et al. 2010) and bulb mites (Stuglik et al. 2014) demonstrate a higher degree of sex-specific bias in gene expression. However, it is not known to what extent the observed bias resolves IASC. If the conflict is not fully resolved, genes involved in IASC in the more sexually dimorphic morph are predicted to lead to lower fitness for females, in which these genes are also expressed. Here, we test this prediction in the bulb mite, Rhizoglyphus robini. In this species, male morphology is heritable (Radwan 1995, 2001) and, as a consequence, the relative proportion of different male morphs in a population can be manipulated using artificial selection. This fact makes the species an ideal model for investigating the evolutionary and genetic aspects of ARTs. Although more aggressive, armored fighters achieve higher reproductive success than scramblers and may even monopolize access to females in small colonies (Radwan & Klimas 2001), scramblers remain present in most populations. As morph fitness does not appear to be frequency-dependent (Radwan & Klimas 2001), it remains unclear which mechanisms maintain genetically based male dimorphism
This article is protected by copyright. All rights reserved.
5
in this species. We hypothesize that unresolved IASC may play a role because lower survival and reproductive success of scrambler males (Radwan and Klimas 2001) may be compensated by increased fitness of their daughters. A similar argument has been made for male homosexuality in humans. It has been found that mothers of homosexuals are more fecund than mothers of heterosexuals, which suggests that genes associated with homosexuality in men might be associated with high fecundity in women (Ciani et al. 2008). In our study, replicated groups of R. robini were subjected to artificial selection to increase proportions of either fighters (F lines) or scramblers (S lines). After the lines diverged, we measured the most important female fitness components, fecundity and longevity, in both. If intralocus sexual conflict is not fully resolved and thus potentially contributes to the maintenance of ARTs in this species, we predict that females from lines with low male fitness (S lines) should be more fit than females from lines with high male fitness (F lines).
Materials and methods General procedures and stock colonies: Stock colonies and large groups of mites were maintained in plastic containers (2 cm high and 2.5 cm in diameter), whereas individuals and pairs were kept in glass tubes (2 cm high and 0.8 cm in diameter), which contained plaster-of-Paris bases soaked in water. Mites were fed powdered yeast ad libitum and maintained at 24±1°C and >90% humidity. Selection lines were derived from two stock colonies that had been established using natural colonies of approximately 200 individuals collected from onions in a garden near Kraków, Poland in 1998 and 2008. Since then, the colonies have been allowed to grow (>1000 individuals) and have been maintained in the lab under the conditions described above.
Selection lines:
This article is protected by copyright. All rights reserved.
6
We used artificial directional selection to create two types of mite lines: the F lines contained an increased proportion of fighters and the S lines contained an increased proportion of scramblers. Four replicates of each line type were generated. Two replicates in each set were derived from the 1998 stock colony, and two replicates came from the 2008 stock colony. Since our aim was to obtain divergent selection lines and not to estimate heritability (which was done in earlier studies; see Radwan 1995, 2001), we applied a simplified selection protocol in which selection differential and the response to selection could not be estimated precisely and were thus not recorded. After the production of each generation, 50 males of the desired morph and 50 females were placed in a 2.5-cm-diameter dish. After being allowed to interact for 3 days and lay eggs, the adults were removed, and the eggs were allowed to develop. After 5-7 days, around 200 tritonymphs (the last juvenile stage of the bulb mite) from each line were transferred to a new container. On subsequent days, the containers were checked twice a day for newly emerged mites and males of the undesired morph were removed from the containers. After all of the mites had reached adulthood, 50 males of the desired morph and 50 females were transferred to a new container to initiate the next generation. This protocol, therefore, could not exclude the possibility of males of the undesired morph mating with females during the few hours between their emergence and removal from the container; however, this approach was very efficient logistically, and lastmale sperm precedence (Radwan 1997) ensured that most eggs were fertilized by males of the desired morph. Starting with generation 25, we regularly recorded the proportion of the different morphs in order to document the divergence of our selection lines.
Female fitness components: Female fitness components were estimated at generation 50 (fecundity-block one) and 55 (fecundity-block two and longevity); at generation 55, we also recorded female body size. To obtain equal-age virgin females for our fitness assays, tritonymphs were placed into individual tubes,
This article is protected by copyright. All rights reserved.
7
where they remained isolated until they reached maturity. Adult mites were sexed and assigned to one of the fitness component tests.
a) Body size Approximately 10 females (5-10 days old) from each line were photographed. Their body lengths (excluding mouthparts) were measured to the nearest 0.01mm using ImageJ software.
b) Fecundity and infertility For logistical reasons, two blocks of females were used for the fecundity assay; the first block consisted of females from generation 50, while the second consisted of females from generation 55. As our measure of fecundity, we determined the number of eggs laid over a seven-day period. Because females oviposit for the first two – three weeks at a nearly constant rate, this provides a good estimate of a female’s lifetime reproductive success (Konior et al. 2001, Tilszer et al. 2006). Each block consisted of approximately 15-20 females from each line (2-4 days old); within each block, each female was placed in an individual tube and allowed to mate with a male from the stock colony, whose morph had been previously recorded. After 24 hours, the male was replaced with another stock colony male of the same morph; a third replacement was performed after another 24 hours. Allowing the female to mate with three males ensured that sperm supply did not limit her fecundity. The female was left with the last male for five days, at which time the assay ended. The males and females were then eliminated, and the eggs laid by each female were counted. We conducted separate analyses of the data on female infertility (the proportion of females that did not lay eggs) and fecundity (the number of eggs laid by females, after excluding females that did not lay eggs).
c) Longevity
This article is protected by copyright. All rights reserved.
8
Fourteen to sixteen one-day-old females from each line were put into a common container along with about the same number of randomly selected stock colony males (there were mostly scramblers in our stock colonies at that time). Each day, the number of dead females was recorded and lifespan was calculated for each female. Dead males were partly replaced with other males every 5-6 days to maintain a constant sex ratio.
Inbreeding assay To rule out the possibility that the differences in female fecundity between selection treatments could have resulted from differential inbreeding in our lines, we performed an additional experiment. We paired virgin females (2-4 days old) from each line with either a male from their own line or a male from another line of the same selection treatment and derived from the same stock colony (i.e., established in 1998 or 2008). This process yielded 8 “inbred” and 8 “outbred” replicates in total (i.e., four “inbred” and four “outbred” replicates in each selection treatment). After 24 hours, the males were eliminated and the females from each replicate were placed in a common container and allowed to lay eggs for two days. The females were then removed, and the eggs were allowed to develop. After their emergence, the tritonymphs were put into separate glass tubes to ensure that they remained virgins. After reaching adulthood, around 10 females from each replicate were mated to randomly chosen males from the stock colony. Because the morph of a female’s mate did not affect her fecundity (see results), we did not control for morph in this assay. In a given tube, after 24 hours, the male was replaced by another stock colony male; after another 24 hours, a third replacement was made. The third male stayed with the female for another three days until the completion of the assay. Afterwards, both the male and female were eliminated, and the eggs laid by the female were counted.
Statistics
This article is protected by copyright. All rights reserved.
9
The analyses were carried out using general linear models implemented in Statistica 10. One of the fitness components was used as the dependent variable, selection treatment (S and F) as a fixed factor, and line identity nested in selection treatment as a random factor. In the analysis of fecundity (number of eggs laid) and infertility (proportion of females that failed to lay eggs in a line within each block and mate morph treatment), the morph of a female’s mate (scrambler or fighter), and its interaction with selection treatment, was also included as a fixed factor, and block was included as a random factor. The lines originating from the different stock colonies (1998 vs. 2008) were analyzed jointly as there were no significant differences between the two for any of the fitness components we measured (see results). To determine if inbreeding depression differed between the selection treatments, we compared the fecundity of daughters whose parents originated either from the same line (“inbred” mites) or from different lines within the same selection treatment (“outbred” mites). We used a general linear model with selection treatment (S and F) and cross type (inbred and outbred) as fixed factors; the mother’s line identity nested in selection treatment was a random factor. To compare effect sizes between different analyses, we calculated partial eta-squared (ŋ2 = SS factor / (SS factor + SS error); Cohen, 1973) for the biologically relevant effects.
Results The selection treatments were clearly successful. After 42 generations, more than 90% of the males in a given line exhibited the desired ART, and that percentage remained stable across successive generations of selection (Fig. S1). Because there were no differences between the lines obtained from the different stock colonies for any of fitness components we measured (body size: F1,66=0.000, p=0.932; fecundity: F1,6=0.010, p=0.939; longevity: F1,6=0.100, p=0.764), we did not include stock colony identity as a factor in subsequent analyses.
This article is protected by copyright. All rights reserved.
10
Females from the S lines were more fecund than females from the F lines, irrespective of the morph of males with whom they mated (Table 1, Fig. 1). The effect of line identity was not significant, but there were significant differences in female fecundity between blocks (Table 1). The S line females lived longer than the F line females (F1,6=9.566, p=0.021, ŋ2=0.614); again, the effect of line identity was not significant (F6,110=0.969, p=0.449, ŋ2=0.502, Fig. 2). There were no significant effects of any of the tested factors on female infertility (selection treatment: F1,6=0.276, p=0.618, ŋ2=0.044; mate morph: F1,21=0.961, p=0.338, ŋ2=0.044; interaction between selection treatment and mate morph: F1,21=0.060, p=0.809, ŋ2=0.002; line ID: F6,21=0.286, p=0.946, ŋ2=0.071; block: F1,21=0.000, p=0.992, ŋ2=0.001). Also, females did not differ in body size either between selection treatments or lines (selection treatment: F1,6=0.879, p=0.381, ŋ2=0.115; line ID: F6,86=0.758, p=0.605, ŋ2=0.050). Female fecundity was not affected by inbreeding (mean ±SD: inbred S = 76.83± 6.67, outbred S = 77.95±4.39, inbred F = 62.37±5.36, outbred F = 61.06±4.90; F1,127=0.028, p=0.868, ŋ2
1
Institute of Environmental Sciences, Jagiellonian University, Gronostajowa 7, 30-387 Kraków, Poland
2
Institute of Environmental Biology, Faculty of Biology, Adam Mickiewicz University, Poznań, Poland
Corresponding author: Agata Plesnar-Bielak, e-mail: [email protected], tel.: (+48) 12 664 51 51, fax: (+48) 12 664 69 12
Full title: Selection for alternative male reproductive tactics alters intralocus sexual conflict. Running title: Intralocus sexual conflict and male morphs. Keywords: intralocus sexual conflict, sexually antagonistic selection, genetic correlation, alternative reproductive tactics, male morphs. Word count: 4057. The manuscript contains 1 table and 2 figures plus 1 supplementary figure. Data available from the Dryad Digital Repository: doi:10.5061/dryad.s4t7m
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/evo.12409. This article is protected by copyright. All rights reserved.
1
Abstract Intralocus sexual conflict (IASC) arises when fitness optima for a shared trait differ between the sexes; such conflict may help maintain genetic variation within populations. Sex-limited expression of sexually antagonistic traits may help resolve the conflict, but the extent of this resolution remains a subject of debate. In species with alternative male reproductive tactics, unresolved conflict should manifest more in a more sexually dimorphic male phenotype. We tested this prediction in the bulb mite (Rhizoglyphus robini), a species in which aggressive fighters coexist with benign scramblers. To do this, we established replicated lines in which we increased the proportion of each of the alternative male morphs using artificial selection. After approximately 40 generations, the proportion of fighters and scramblers stabilized at >0.9 in fighter- and scrambler-selected lines, respectively. We then measured several female fitness components. As predicted by IASC theory, female fecundity and longevity was lower in lines selected for fighters and higher in lines selected for scramblers. This finding indicates that sexually selected phenotypes are associated with an ontogenetic conflict that is not easily resolved. Furthermore, we suggest that IASC may be an important mechanism contributing to the maintenance of genetic variation in the expression of alternative reproductive tactics.
Keywords: intralocus sexual conflict, sexually antagonistic selection, genetic correlation, alternative reproductive tactics, male morphs.
This article is protected by copyright. All rights reserved.
2
Introduction Although the sexes demonstrate a high degree of overlap in their genomes and many traits are expressed in both males and females, fitness optima for a number of shared traits may be noticably sex-specific (Glucksman 1981, Rice 1984, Chippindale et al. 2001). In particular, traits that increase fitness when expressed in one sex may decrease fitness when expressed in the other sex. These differences may arise because the ecological niches of the sexes differ (e.g., Shine 1989, Forsman 1995, Merilä et al. 1997) or because some traits may be under strong sexual selection in males but not females (e.g., Chippindale et al. 2001, Björklund & Senar 2001, Price and Burley 1994, Robinson et al. 2006). This kind of antagonism, called intralocus sexual conflict (IASC), is associated with sex-specific differences in the direction of selection on allelic variants at a locus and leads to negative fitness correlations between the sexes for a given trait (Lande 1980). IASC has been observed in many species of different taxa (see Rice & Chippindale 2001, Bonduriansky & Chenoweth 2008, and van Doorn 2009 for review), although its generality and evolutionary importance are still debated (e.g., Cox & Calsbeek 2009, Bonduriansky & Chenoweth 2008). IASC may potentially have consequences for various evolutionary processes such as speciation (Parker & Partridge 1998, Rice & Chippindale 2002), the evolution of sex chromosomes (Bull 1983, Rice 1987, Charlesworth 1991), the evolution of sex determination (Rice 1986, Kraak & Pen 2002, van Doorn & Kirkpatrick 2007), the regulation of gene expression (Ellegren & Parsch 2007), aging (Vieira et al. 2000, Bonduriansky et al. 2008), sexual selection (Brommer et al. 2007, Pischedda & Chippindale 2006), sex allocation (Alonzo & Sinevo 2007), and the maintenance of genetic variation in natural populations (Chippindale & Rice 2001). Here, we hypothesize that it may also facilitate the maintenance of genetically determined alternative reproductive tactics (see below). However, the evolutionary significance of IASC depends on how easily it can be resolved and, hence, how prevalent it is.
This article is protected by copyright. All rights reserved.
3
Several ways of resolving IASC have been proposed (reviewed in Stewart et al. 2010). In one scenario, a locus subject to antagonistic selection may be duplicated, and one of the duplicated genes might be expressed in males while the other is expressed in females (Partridge & Hurst 1998, Rice & Chippindale 2001, Wyman et al. 2012). IASC may also be resolved by sex-dependent regulation of gene expression (McIntyre et al. 2006) or genomic imprinting (Day & Bonduriansky 2004). As a consequence, antagonistic traits may evolve sexually dimorphic expression that demonstrates sex-specific optima, which is considered to be evidence of IASC resolution (see Cox & Calsbeek 2009; Stewart at al. 2010 and Pennell & Morrow 2013 for review). However, evidence is accumulating that IASC is not easily resolved, as manifested by positive genetic correlations between male and female traits that have opposite fitness effects in the sexes (reviewed by Poissant et al. 2010, see also Cox & Calsbeek 2009) or gene expression levels for such traits (Griffin et al. 2013). For example, sexually dimorphic locomotor activity in Drosophila melanogaster still showed significant inter-sexual genetic correlation despite being selected in different directions in males and females (Long & Rice 2007). In another study, Harano et al. (2010) recently showed that sexlimited trait expression does not fully resolve IASC; their study examined broad-horned flour beetles, a species in which males exhibit enlarged mandibles used in fights with rivals (a trait subject to sexual selection), a feature that is completely lacking in females. The authors showed that females from populations subject to selection for large mandibles (and thus higher fitness) in males had significantly lower fitness than females from control populations not subject to selection. In contrast, selection for small mandibles in males increased female fitness at the cost of male reproductive success. This finding indicates that, despite the sex-limited expression of horns, genes associated with horn elaboration decrease female fitness, which the authors attribute to genetic correlations between sexually antagonistic traits and other features that contribute to fitness. Alternative reproductive tactics (ARTs) are well suited to the study IASC because they typically comprise discontinuous male phenotypes which differ in the degree of sexual dimorphism.
This article is protected by copyright. All rights reserved.
4
In such cases, the more sexually dimorphic ART can be expected to be associated with stronger IASC. ARTs often manifest themselves as male polymorphism, where the morphs differ in reproductive strategies as well as the morphological structures used in male-male competition (reviewed in Brockman 2001, Oliveira et al. 2008). Typically, a morph possessing exaggerated fighting structures is more dominant and aggressive, whereas the other, less sexually dimorphic morph lacks weapons and is more subordinate. Some well-known examples of such dimorphisms include horns in scarab beetles (Emlen 1994, Kotiaho et al. 2003) or leg dimorphism in acarid mites (Woodring 1969, Radwan 1995, 2009); however, males may also differ in their expression of sexual ornaments associated with status and sexual attractiveness (e.g., color morphs in side-blotched lizards (Alonzo & Sinervo 2001) or dominant and subordinate morphs in turkeys (Buchholz 1995)). Higher degree of sexual dimorphism should be associated with higher IASC, but also with higher pressure for IASC resolution. Consistent with this expectation, dominant wild male turkeys (Pointer et al. 2012) as well as armored male beetles (Snell-Rood et al. 2010) and bulb mites (Stuglik et al. 2014) demonstrate a higher degree of sex-specific bias in gene expression. However, it is not known to what extent the observed bias resolves IASC. If the conflict is not fully resolved, genes involved in IASC in the more sexually dimorphic morph are predicted to lead to lower fitness for females, in which these genes are also expressed. Here, we test this prediction in the bulb mite, Rhizoglyphus robini. In this species, male morphology is heritable (Radwan 1995, 2001) and, as a consequence, the relative proportion of different male morphs in a population can be manipulated using artificial selection. This fact makes the species an ideal model for investigating the evolutionary and genetic aspects of ARTs. Although more aggressive, armored fighters achieve higher reproductive success than scramblers and may even monopolize access to females in small colonies (Radwan & Klimas 2001), scramblers remain present in most populations. As morph fitness does not appear to be frequency-dependent (Radwan & Klimas 2001), it remains unclear which mechanisms maintain genetically based male dimorphism
This article is protected by copyright. All rights reserved.
5
in this species. We hypothesize that unresolved IASC may play a role because lower survival and reproductive success of scrambler males (Radwan and Klimas 2001) may be compensated by increased fitness of their daughters. A similar argument has been made for male homosexuality in humans. It has been found that mothers of homosexuals are more fecund than mothers of heterosexuals, which suggests that genes associated with homosexuality in men might be associated with high fecundity in women (Ciani et al. 2008). In our study, replicated groups of R. robini were subjected to artificial selection to increase proportions of either fighters (F lines) or scramblers (S lines). After the lines diverged, we measured the most important female fitness components, fecundity and longevity, in both. If intralocus sexual conflict is not fully resolved and thus potentially contributes to the maintenance of ARTs in this species, we predict that females from lines with low male fitness (S lines) should be more fit than females from lines with high male fitness (F lines).
Materials and methods General procedures and stock colonies: Stock colonies and large groups of mites were maintained in plastic containers (2 cm high and 2.5 cm in diameter), whereas individuals and pairs were kept in glass tubes (2 cm high and 0.8 cm in diameter), which contained plaster-of-Paris bases soaked in water. Mites were fed powdered yeast ad libitum and maintained at 24±1°C and >90% humidity. Selection lines were derived from two stock colonies that had been established using natural colonies of approximately 200 individuals collected from onions in a garden near Kraków, Poland in 1998 and 2008. Since then, the colonies have been allowed to grow (>1000 individuals) and have been maintained in the lab under the conditions described above.
Selection lines:
This article is protected by copyright. All rights reserved.
6
We used artificial directional selection to create two types of mite lines: the F lines contained an increased proportion of fighters and the S lines contained an increased proportion of scramblers. Four replicates of each line type were generated. Two replicates in each set were derived from the 1998 stock colony, and two replicates came from the 2008 stock colony. Since our aim was to obtain divergent selection lines and not to estimate heritability (which was done in earlier studies; see Radwan 1995, 2001), we applied a simplified selection protocol in which selection differential and the response to selection could not be estimated precisely and were thus not recorded. After the production of each generation, 50 males of the desired morph and 50 females were placed in a 2.5-cm-diameter dish. After being allowed to interact for 3 days and lay eggs, the adults were removed, and the eggs were allowed to develop. After 5-7 days, around 200 tritonymphs (the last juvenile stage of the bulb mite) from each line were transferred to a new container. On subsequent days, the containers were checked twice a day for newly emerged mites and males of the undesired morph were removed from the containers. After all of the mites had reached adulthood, 50 males of the desired morph and 50 females were transferred to a new container to initiate the next generation. This protocol, therefore, could not exclude the possibility of males of the undesired morph mating with females during the few hours between their emergence and removal from the container; however, this approach was very efficient logistically, and lastmale sperm precedence (Radwan 1997) ensured that most eggs were fertilized by males of the desired morph. Starting with generation 25, we regularly recorded the proportion of the different morphs in order to document the divergence of our selection lines.
Female fitness components: Female fitness components were estimated at generation 50 (fecundity-block one) and 55 (fecundity-block two and longevity); at generation 55, we also recorded female body size. To obtain equal-age virgin females for our fitness assays, tritonymphs were placed into individual tubes,
This article is protected by copyright. All rights reserved.
7
where they remained isolated until they reached maturity. Adult mites were sexed and assigned to one of the fitness component tests.
a) Body size Approximately 10 females (5-10 days old) from each line were photographed. Their body lengths (excluding mouthparts) were measured to the nearest 0.01mm using ImageJ software.
b) Fecundity and infertility For logistical reasons, two blocks of females were used for the fecundity assay; the first block consisted of females from generation 50, while the second consisted of females from generation 55. As our measure of fecundity, we determined the number of eggs laid over a seven-day period. Because females oviposit for the first two – three weeks at a nearly constant rate, this provides a good estimate of a female’s lifetime reproductive success (Konior et al. 2001, Tilszer et al. 2006). Each block consisted of approximately 15-20 females from each line (2-4 days old); within each block, each female was placed in an individual tube and allowed to mate with a male from the stock colony, whose morph had been previously recorded. After 24 hours, the male was replaced with another stock colony male of the same morph; a third replacement was performed after another 24 hours. Allowing the female to mate with three males ensured that sperm supply did not limit her fecundity. The female was left with the last male for five days, at which time the assay ended. The males and females were then eliminated, and the eggs laid by each female were counted. We conducted separate analyses of the data on female infertility (the proportion of females that did not lay eggs) and fecundity (the number of eggs laid by females, after excluding females that did not lay eggs).
c) Longevity
This article is protected by copyright. All rights reserved.
8
Fourteen to sixteen one-day-old females from each line were put into a common container along with about the same number of randomly selected stock colony males (there were mostly scramblers in our stock colonies at that time). Each day, the number of dead females was recorded and lifespan was calculated for each female. Dead males were partly replaced with other males every 5-6 days to maintain a constant sex ratio.
Inbreeding assay To rule out the possibility that the differences in female fecundity between selection treatments could have resulted from differential inbreeding in our lines, we performed an additional experiment. We paired virgin females (2-4 days old) from each line with either a male from their own line or a male from another line of the same selection treatment and derived from the same stock colony (i.e., established in 1998 or 2008). This process yielded 8 “inbred” and 8 “outbred” replicates in total (i.e., four “inbred” and four “outbred” replicates in each selection treatment). After 24 hours, the males were eliminated and the females from each replicate were placed in a common container and allowed to lay eggs for two days. The females were then removed, and the eggs were allowed to develop. After their emergence, the tritonymphs were put into separate glass tubes to ensure that they remained virgins. After reaching adulthood, around 10 females from each replicate were mated to randomly chosen males from the stock colony. Because the morph of a female’s mate did not affect her fecundity (see results), we did not control for morph in this assay. In a given tube, after 24 hours, the male was replaced by another stock colony male; after another 24 hours, a third replacement was made. The third male stayed with the female for another three days until the completion of the assay. Afterwards, both the male and female were eliminated, and the eggs laid by the female were counted.
Statistics
This article is protected by copyright. All rights reserved.
9
The analyses were carried out using general linear models implemented in Statistica 10. One of the fitness components was used as the dependent variable, selection treatment (S and F) as a fixed factor, and line identity nested in selection treatment as a random factor. In the analysis of fecundity (number of eggs laid) and infertility (proportion of females that failed to lay eggs in a line within each block and mate morph treatment), the morph of a female’s mate (scrambler or fighter), and its interaction with selection treatment, was also included as a fixed factor, and block was included as a random factor. The lines originating from the different stock colonies (1998 vs. 2008) were analyzed jointly as there were no significant differences between the two for any of the fitness components we measured (see results). To determine if inbreeding depression differed between the selection treatments, we compared the fecundity of daughters whose parents originated either from the same line (“inbred” mites) or from different lines within the same selection treatment (“outbred” mites). We used a general linear model with selection treatment (S and F) and cross type (inbred and outbred) as fixed factors; the mother’s line identity nested in selection treatment was a random factor. To compare effect sizes between different analyses, we calculated partial eta-squared (ŋ2 = SS factor / (SS factor + SS error); Cohen, 1973) for the biologically relevant effects.
Results The selection treatments were clearly successful. After 42 generations, more than 90% of the males in a given line exhibited the desired ART, and that percentage remained stable across successive generations of selection (Fig. S1). Because there were no differences between the lines obtained from the different stock colonies for any of fitness components we measured (body size: F1,66=0.000, p=0.932; fecundity: F1,6=0.010, p=0.939; longevity: F1,6=0.100, p=0.764), we did not include stock colony identity as a factor in subsequent analyses.
This article is protected by copyright. All rights reserved.
10
Females from the S lines were more fecund than females from the F lines, irrespective of the morph of males with whom they mated (Table 1, Fig. 1). The effect of line identity was not significant, but there were significant differences in female fecundity between blocks (Table 1). The S line females lived longer than the F line females (F1,6=9.566, p=0.021, ŋ2=0.614); again, the effect of line identity was not significant (F6,110=0.969, p=0.449, ŋ2=0.502, Fig. 2). There were no significant effects of any of the tested factors on female infertility (selection treatment: F1,6=0.276, p=0.618, ŋ2=0.044; mate morph: F1,21=0.961, p=0.338, ŋ2=0.044; interaction between selection treatment and mate morph: F1,21=0.060, p=0.809, ŋ2=0.002; line ID: F6,21=0.286, p=0.946, ŋ2=0.071; block: F1,21=0.000, p=0.992, ŋ2=0.001). Also, females did not differ in body size either between selection treatments or lines (selection treatment: F1,6=0.879, p=0.381, ŋ2=0.115; line ID: F6,86=0.758, p=0.605, ŋ2=0.050). Female fecundity was not affected by inbreeding (mean ±SD: inbred S = 76.83± 6.67, outbred S = 77.95±4.39, inbred F = 62.37±5.36, outbred F = 61.06±4.90; F1,127=0.028, p=0.868, ŋ2
