|

 

Evol_56_703.1331_1339.tp

Evolution, 56(7), 2002, pp. 1331–1339 WITHIN- AND BETWEEN-POPULATION VARIATION FOR WOLBACHIA-INDUCED REPRODUCTIVE INCOMPATIBILITY IN A HAPLODIPLOID MITE F. VALA,1,2 A. WEEKS,3 D. CLAESSEN,4 J. A. J. BREEUWER,5 AND M. W. SABELIS6 Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, P.O. Box 94084, 1090 GB Amsterdam, The Netherlands Wolbachia pipientis is a bacterium that induces cytoplasmic incompatibility (CI), the phenomenon in which infected males are reproductively incompatible with uninfected females. CI spreads in a population of hosts becauseit reduces the fitness of uninfected females relative to infected females. CI encompasses two steps: modification (mod)of sperm of infected males and rescuing (resc) of these chromosomes by Wolbachia in the egg. Infections associatedwith CI have mod1resc1 phenotypes. However, mod2resc1 phenotypes also exist; these do not result in CI. Assumingmod/resc phenotypes are properties of the symbiont, theory predicts that mod2resc1 infections can only spread in ahost population where a mod1resc1 infection already occurs. A mod2resc1 infection spreads if the cost it imposes onthe infected females is lower than the cost inflicted by the resident (mod1resc1) infection. Furthermore, introductionof a mod2 Wolbachia eventually drives infection to extinction. The uninfected population that results can be recolonizedby a CI-causing Wolbachia. Here, we investigated whether variability for induction of CI was present in two Tetranychusurticae populations. In one population all isofemale lines tested were mod2. In the other, mod1resc1 and mod2resc1isofemale lines coexisted. We found no evidence for a cost difference to females expressing either type (mod1/2).
Infections in the two populations could not be distinguished based on sequences of two Wolbachia genes. We considerthe possibility that mod2 is a host effect through a population dynamics model. A mod2 host allele leads to infectionextinction in the absence of fecundity differences. Furthermore, the uninfected population that results is immune toreestablishment of the (same) CI-causing Wolbachia.
Key words. Cytoplasmic incompatibility, modification and rescue, resistance, spider mite, theoretical model, Wol- Received August 13, 2001.
Accepted March 11, 2002.
Wolbachia pipientis is a cytoplasmically transmitted bac- so haplodized eggs develop as males. Depending on the de- terium that infects several arthropod and nematode hosts. In gree of aneuploidy, eggs may: develop as a male, if eggs the two-spotted spider mite, Tetranychus urticae Koch, a phy- revert to the haploid state; die, if haplodization is not com- tophagous haplodiploid arthropod, Wolbachia can induce plete but insufficient to develop as a female; or develop as both cytoplasmic incompatibility (CI; Breeuwer 1997; Vala an aneuploid female (see Breeuwer 1997; Vala et al. 2000).
et al. 2000) and hybrid breakdown (Vala et al. 2000).
Mortality of aneuploid embryos would explain why, in hap- CI is expressed in crosses between uninfected (U) females lodiploids, CI is expressed as a bias of F1 sex ratio toward and infected (W) males (reviewed by Hoffmann and Turelli males associated with an increase in mortality (cf. Breeuwer 1997; Stouthamer et al. 1999). CI is not induced if the same 1997; Vala et al. 2000; Vavre et al. 2000).
Wolbachia strain that was present in the male is also presentin the fertilized egg (i.e., /W 3 ?W). CI reduces the fitness The Modification and Rescuing Model of uninfected females (which are incompatible with W males)relative to infected females (which are compatible with both Although the molecular details of CI are not known, the W and U males). As a result, the frequency of infected fe- phenomenon is thought to involve a Wolbachia-produced tox- males (and the CI trait) increases in the host population.
in and antitoxin (Hurst 1991; Rousset and Raymond 1991) Cytological studies of CI in Nasonia wasps (Reed and Wer- or, similarly, Wolbachia-mediated modification and rescuing ren 1995) and Drosophila simulans (Callaini et al. 1997) steps (Hoffmann and Turelli 1997; Werren 1997). Chromo- showed that in /U 3 ?W crosses the paternal set of chro- somes from infected males are modified by Wolbachia and mosomes does not segregate properly in mitotic divisions become unable to respond properly to cell-cycle cues in un- early in embryonic development. This results in haploid or infected eggs. If the fertilized egg is infected with the same aneuploid embryos. In diploid species haploid and aneuploid bacterial strain that was present in the father, paternal chro- embryos abort, thus CI is expressed as increased F1 mortality.
mosomes are rescued, that is, they segregate properly during In haplodiploids, females are diploid and males are haploid, An infection in which Wolbachia induces CI is denoted 2 Present address: Department of Biology, University College mod1resc1 (Werren 1997). However, if CI is not induced, London, Wolfson House, 4 Stephenson Way, London NW1 2HE, the infection phenotype may be either mod2resc2, an infec- United Kingdom.
tion that can neither modify sperm nor rescue modified chro- 3 Present address: Department of Entomology, University of Cal- ifornia, Riverside, California 92521; E-mail: [email protected].
mosomes, or mod2resc1, an infection that cannot modify sperm but can rescue modified chromosomes. The test to q 2002 The Society for the Study of Evolution. All rights reserved.
F. VALA ET AL.
distinguish between these alternatives is to mate females har- and PCR were as in Breeuwer (1997). The R and C strains boring a mod2resc? infection to males with a mod1resc1 Wol- are the same as those described in Vala et al. (2000).
bachia (Bourtzis et al. 1998; Merc¸ot and Poinsot 1998a). IfCI is observed, then the infection (in the female) is Isofemale Lines mod2resc2. Of course, this test is valid only if females andmales are reproductively compatible in the absence of Wol- Isofemale lines of the two strains were created by taking bachia. Note that this test does not indicate whether mod2 virgin females from the infected base populations and per- is a property of the symbiont or of the host.
forming mother 3 son matings for four consecutive gener- Mod2 is typically assumed to be a property of the sym- ations; for arrenotokous haplodiploid organisms this gives biont. Theory predicts that in a population where a mod1resc1 an expected inbreeding coefficient of 0.98 (Hartl and Clark Wolbachia is fixed, a mod2resc1 infection increases when 1997). From each inbred isofemale line an uninfected coun- rare, if it inflicts a cost to the infected female that is lower terpart was created either by tetracycline curing as described than the cost imposed by the resident type of infection (Prout by Breeuwer (1997) or by heat treatment as described by Van 1994; Turelli 1994; Hurst and McVean 1996). However, if Opijnen and Breeuwer (1999; whatever method worked first).
there are uninfected individuals in the population (e.g., be- For tetracycline treatment, 20–30 females were placed on cause not all progeny of an infected mother is infected), arenas and fed an antibiotic solution (described by Breeuwer mod2resc1 infections cannot spread or persist without the 1997). For heat treatment, 20–30 females were placed at 328C mod1resc1 type (Hurst and McVean 1996). Therefore, (Van Opijnen and Breeuwer 1999). Mites were reared as a mod2resc1 phenotype is expected to coexist with a mod1resc1 culture at this temperature for eight to nine generations to maximize the number of uninfected females at the end of The mod2resc1 Wolbachia type (initially) increases in fre- treatment. Isofemale lines cured by tetracycline are labeled quency within the infected subpopulation because it reduces TET, and isofemale lines cured by heat treatment are labeled host fecundity less than the resident mod1resc1 Wolbachia.
HT. For one isofemale line (R3), two uninfected sublines However, mod2resc1 Wolbachia rely on the sterilizing effect were established, one by curing with tetracycline the other of the mod1resc1 infection to reduce the fitness of uninfectedindividuals. As mod2resc1 Wolbachia increase within the in- by curing with heat treatment. The TET and HT R3 sublines fected subpopulation, the mod1resc1 phenotype decreases in were compared to control for treatment effects. To establish frequency and, consequently, the frequency of uninfected the uninfected sublines, 10–15 mated females per isofemale hosts increases. Thus, spread of a mod2 Wolbachia eventually line were placed alone on leaf discs to oviposit for three days leads to infection extinction (Hurst and McVean 1996). Both and subsequently collected for PCR. For each isofemale line, mod2 and mod1 infection types disappear, and the population offspring from females that did not give amplification prod- returns to the uninfected state. Because potentially any un- ucts on a PCR with Wolbachia-specific primers were kept.
infected population may be (re)colonized by a CI-causing Offspring from females positive in the PCR was discarded.
Wolbachia and again revert to the uninfected state, in a sense, This process was repeated twice. Finally, F3 females were reversible or cyclic evolution is possible (Hurst and McVean pooled to establish the uninfected sublines. PCR assays with Wolbachia-specific primers and DNA isolation were as in Mod2resc1 infection phenotypes have been described in Breeuwer (1997).
Drosophila (Bourtzis et al. 1998; Merc¸ot and Poinsot 1998a).
Because lines were inbred prior to curing and thus are However, it was not investigated whether these mod2 infec- expected to be nearly homozygous, differences between the tions coexist with mod1resc1 infections as predicted by the- infected and uninfected sublines of each isofemale line are ory. In the present study we investigate whether variability most likely due to presence or absence of Wolbachia. As- for mod and resc phenotypes is present between and/or within suming nuclear genetic variation in the base population, dif- two spider mite populations.
ferences between isofemale lines are likely to be due to ge-netic differences at the nuclear level (the genetic similarity MATERIALS AND METHODS of Wolbachia in the two populations is discussed below).
Base Populations To detect variation in Wolbachia-induced reproductive in- compatibility, several isofemale lines from each population Two populations of T. urticae spider mites were established were tested for CI. The test for aneuploidy of F in the laboratory, one from mites collected from rose plants hybrid breakdown test, is presented in a separate paper. To (the R strain, hereafter) in a greenhouse at Aalsmeer, The test for CI, all possible crosses between infected (W) and Netherlands and another from mites collected from cucumberplants (the C strain, hereafter) obtained from the Institute for uninfected (U) individuals were performed (/ 3 ?: W 3 Horticultural Plant Breeding in Wageningen, The Nether- W, W 3 U, U 3 U, U 3 W). In haplodiploids, CI may result lands. Since collection, spider mites have been reared on in a male bias of F1 sex ratio associated with an increase in detached leaves of Phaseolus vulgaris ‘Arena'. Cultures were mortality in /U 3 ?W crosses compared to U 3 U crosses.
maintained, and experiments were performed in one climate If Wolbachia is present in the female and/or absent in the room, at 238C, 60–80% relative humidity, and 16L:8D pho- male (thus, / 3 ?: W 3 W and W 3 U), crosses should be toperiod. Both strains were infected with Wolbachia based on a polymerase chain reaction (PCR) assay with Wolbachia- mod2resc1, crosses were performed between mod2resc? fe- specific primers (Breeuwer and Jacobs 1996). DNA isolation males and mod1resc1 males of another isofemale line.
Procedures for All Experiments ftsZ1262R, which amplify 730 base pairs (bp) of the cell-division gene (Holden et al. 1993), and the primer pairs Twenty-five to 30 females of each line laid eggs on de- wsp81F and wsp619R (Zhou et al. 1998), which amplify 590– tached bean leaves (P. vulgaris) placed on water-soaked cot-ton balls. These females were transferred at three-day inter- 632 bp of the wsp gene, were used in separate PCR ampli- vals to produce age cohorts. Offspring from these cohorts fications. PCR reaction mixes and amplification conditions were used in the experiments to ensure that all mites tested were the same as described in Weeks and Breeuwer (2001).
were of the same age. All experiments were performed on PCR products were then cleaned using Genecleant (BIO 101, bean leaf discs (1.5 cm in diameter). Leaf discs were placed Inc., Bingham, Nottingham, U.K.) and cloned into a pGEMt- on water-soaked cotton sheets stretched on sponges. Sponges T vector (Promega, Madison, WI). We extracted five vectors were placed on plastic trays, and water was added regularly from recombinant colonies for each gene from each strain to prevent the leaf discs from drying. In all experiments, using the alkaline-lysis method (Sambrook et al. 1989). After crosses and spider mite lines were randomized across spong- extraction, 1 mg of vector DNA was used as template for a cycle sequencing reaction (Thermosequenase kit, Amersham/ For F1 analysis (test for CI), experimental females were Pharmacia, Piscataway, NJ) using fluorescent-labeled primer collected at the last molting stage from age cohorts (to ensure (IRD 700/800, Biolegio, Malden, The Netherlands) and sub- they were virgin) and placed in groups of five females and sequently run on an NEN Global IR2 DNA analyzer (LI- three males for 48 h to mate. Subsequently, females were COR, Lincoln, NE).
transferred individually to fresh leaf discs for oviposition. Intotal six days of oviposition were scored, and females were transferred to a fresh leaf disc after three days. Offspring (F1female, F1 male, unhatched eggs, and dead individuals) were Eight isofemale lines of two spotted spider mites were counted 10–12 days later and used to compute clutch size established through mother 3 son mating. One reason why (CS 5 number unhatched eggs 1 number dead 1 number it may be easy to establish these lines is haplodiplody: Re- F1 females 1 number F1 males), F1 sex ratio (SR 5 number cessive deleterious mutations are mostly eliminated through F1 males/[number F1 females 1 number F1 males]), and F1 male mortality (Crozier 1985). Thus, in a sense, mothers mortality (mortality 5 [number unhatched eggs 1 number usually mated to good sons.
of dead]/CS).
Lines established from PCR-negative females in an assay with Wolbachia-specific primers remained negative without further treatment. For isofemale line 3, two cured sublines Effect of factors was analyzed by MANOVAs on derived were established by curing with tetracycline (TET) and by variables (i.e., variables computed from what was actually curing with heat treatment (HT). No differences were found measured in the experiments: clutch size, sex ratio, and mor- in crosses with mites cured by one or the other method (Table tality) because these variables will generally not be inde- 1, R3). Therefore, for assessment of reproductive compati- pendent. We report the MANOVA Wilk's l test statistic.
bility, we conclude that method of curing had no effect other Normality of data was tested graphically and significance was than removal of the symbiont.
examined using the Shapiro-Wilk test. Homocedasticity(equality of group variances) was tested using Levine's test.
The Effect of Wolbachia on Cytoplasmic Incompatibility In MANOVAs, equality of covariance matrices was testedusing the Box's test. Nonparametric tests (Kruskal-Wallis) Variability for Wolbachia-induced reproductive incompat- were used when assumptions of normality and homocedas- ibility was found among isofemale lines of the rose strain.
ticity were violated (provided they could not be solved by Presence of Wolbachia in males of isofemale line R1 and R2 transformation). When (M)ANOVA were performed, sex ra- resulted in induction of CI when mated with R1 and R2 tio and mortality were arcsine-square-root transformed. Sta- uninfected females, respectively. CI was expressed as in- tistic analysis was performed using SPSS (Chicago, IL).
creased F1 mortality and a sex-ratio bias toward males (Table MANOVAs were followed by a series of univariate ANO- 1). However, in crosses of infected R3 males with uninfected VAs. The significance a-level of these ANOVAs (P 5 0.05) R3 females CI was not observed (Table 1). PCR with Wol- was adjusted following the Bonferroni procedure to correct bachia primers confirmed that this result was not due to a for multiple analysis (Field 2000). Pairwise comparisons change in infection status of the uninfected sublines (or of were performed using Tukey post hoc tests.
the infected one). Finally, absence of CI was stable over time;the same result was obtained in later experiments (cf. Table Cloning and Sequencing of Wolbachia ftsZ and wsp Genes Two Wolbachia genes (wsp and ftsZ) from both the cu- Although in isofemale line C3 higher F1 mortality was cumber and rose populations of T. urticae were cloned and observed in U 3 W crosses (Table 2), this effects was not sequenced to determine the relatedness of their Wolbachia statistically significant. Thus, the cucumber isofemale lines strains. Ten individual female mites from each mass-bred tested did not show CI associated with presence of Wolbachia population of cucumber and rose were pooled separately.
in males—all infections in C lines were associated with mod2 DNA was extracted using the CTAB method adapted for phenotypes. In C5 a significant effect in sex ratio was found: mites from Breeuwer (1997). The primers ftsZ491F and Infected females produced more female-biased sex ratios. A F. VALA ET AL.
TABLE 1. Test for induction of cytoplasmic incompatibility: crosses within isofemale lines from the Rose population. W, Wolbachia infected;U, uninfected (cured); TET, cured by tetracycline; HT, cured by heat treatment; variables significant (after Bonferroni correction) according tounivariate ANOVAs are marked with an asterisk; identical superscripts (a, b, c, d) within columns indicate nonsignificant differences betweencrosses at the 5% level (Tukey test).
5 29.09, Wilk's l 5 0.21, P , 0.001 5 3.70, P , 0.001 5 4.13, P , 0.01 5 37.06, Wilk's l 5 0.22, P , 0.001 5 3.70, P , 0.001 5 4.13, P , 0.001 Rose 3 (HT and TET ) 5 3.96, Wilk's l 5 0.81, P , 0.001 5 5.64, P , 0.001 5 3.70, P 5 0.001 5 2.54, P 5 0.015 sex-ratio shift toward females provides a spreading mecha- Wsp and ftsZ Sequence Variation for the Two Strains of nism for Wolbachia (Egas et al. 2002) and is consistent with results obtained previously in the base population (Vala et No differences were found in either of the ftsZ or wsp al. 2000, 2002).
sequences within or between the cucumber and rose mass-bred populations of T. urticae. All 10 inserts sequenced (five from cucumber and five from the rose populations) for both To test whether infected R3 females retained the property ftsZ and wsp were identical (Genbank accession numbers are of rescuing modified sperm, despite the fact that sperm from ftsZ-AF404763 and wsp-AF404765 for cucumber and ftsZ- infected R3 males is not modified, R3 and R1 mites were AF404764 and wsp-AF404766 for rose).
crossed. Results are presented in Table 3. First, as for pre-vious experiments (Tables 1, 2), infected R1 males induced CI in uninfected R1 females, but infected R3 males did not This is the first report of within-population variation for induce CI in uninfected R3 females. Second, infected R1 Wolbachia-induced phenotypes. The variation reported here males induced CI in uninfected R3 females, whereas infected is likely to reflect genetic differences because environmental R3 males did not induce CI in uninfected R1 females. Third, conditions were constant throughout experiments. The ques- presence of Wolbachia in R3 females eliminated incompat- tion is whether these differences are due to genetic variation ibility in crosses with infected R1 males. In other words, at the host, at the symbiont level, or both.
mortality and sex ratio of /R3W 3 ?R1W was not signif-icantly different from mortality and sex ratio of /R3U 3 Is mod2 a Property of Wolbachia or of the Host? R1U crosses, whereas both differed from /R3U 3 ?R1W crosses. Thus, we conclude that the phenotype of the Wol- Three genes are commonly used to infer Wolbachia phy- bachia-host association in isofemale line R3 is of type logeny, 16S rDNA, wsp, and ftsZ. The latter two evolve faster and wsp is the most variable and informative (Zhou et al.
TABLE 2. Test for induction of cytoplasmic incompatibility: crosses within isofemale lines from the cucumber population (for table legend,see Table 1).
Cucumber 1 (TET ) 5 1.85, Wilk's l 5 0.89, ns Cucumber 2 (TET ) 5 2.61, Wilk's l 5 0.81, P 5 0.007 5 3.70, P 5 0.014 5 4.13, P 5 0.008 5 6.49, Wilk's l 5 0.68, P , 0.001 5 16.75, P , 0.001 5 5.75, P 5 0.001 Cucumber 4 (TET ) 5 3.79, Wilk's l 5 0.77, P , 0.001 5 4.54, P 5 0.005 5 4.54, P , 0.004 5 4.74, Wilk's l 5 0.78, P , 0.001 5 4.17, P 5 0.007 5 7.70, P , 0.001 1998; Jiggins et al. 2001). In our study, wsp and ftsZ se- genes through recombination (Jiggins et al. 2001; Werren quences did not correlate with infection phenotype. Sequenc- and Bartos 2001). Alternatively, the phenotypes associated es were identical in the rose and cucumber strains, and re- with these infections may be influenced by host effects.
productive incompatibility was induced in isofemale lines of Crosses within isofemale lines of the cucumber strain re- the first, but not of the second strain. Fialho and Stevens vealed that infection phenotype was mod2resc? in all C iso- (2000) report a similar result. In Tribolium madens Wolbachia female lines (Tables 2). The same result is observed in crosses is associated with male killing and in T. confusum the infec- within the cucumber base population (Vala et al. 2002). Pre- tion results in CI. However, based on ftsZ and wsp sequences vious results suggest that C infections can, to some extent, the two bacteria cannot be distinguished (Fialho and Stevens rescue R-modified sperm (cf. Vala et al. 2000). However, the 2000). Of course, it is possible that the Wolbachia present two strains are reproductively isolated even in the absence in these pairs of strains and species differs in genes other of Wolbachia (Vala et al. 2000). Thus, it is difficult to con- than the two we have sequenced. The two strains of Wol- clusively determine whether infected C females can rescue bachia may have acquired similar copies of wsp and ftsZ sperm from infected R males. In any case, how does an in- F. VALA ET AL.
TABLE 3. Crosses between R1 and R3 isofemale lines for resc2 or in R3. If we repeat these calculations for sex ratio, then R1 resc1 status (for table legend, see Table 1).
has the least female-biased sex ratio (0.48), whereas R2 andR3 have similar sex ratios (0.27 and 0.26). Thus, infections in R1 or R2 do not appear more costly to females than the infection in R3. Absence of a difference in cost to infected Does R1W induce CI in R1U? Yes.
R1, R2, and R3 females is expected if the Wolbachia infecting these lines is the same. This hypothesis cannot be refuted based on our sequence data. The phenotypic differences ob- Does R1W induce CI in R3U? Yes.
served may be due to genetic differences between hosts.
Like a symbiont mod2 allele, a host mod2 allele cannot invade a population unless there is a resident mod1. CI pro- Can R3W rescue R1W? Yes.
vides a spreading mechanism to the symbiont, but for nuclear host genes CI means that not all crosses between infected Does R3W induce CI in R3U? No.
and uninfected individuals will produce viable offspring. In a population with infected and uninfected hosts, males that possess a nuclear allele conferring resistance to sperm mod- Does R3W induce CI in R1U? No.
ification are compatible with all females in the population.
Therefore, such a host allele invades (Turelli 1994), even if the mod2 trait is not associated with a lower cost to the 5 18.74, Wilk's l 5 0.48, P , 0.001 infected (mod2) host. The allele spreads in the absence of 5 23.24, P , 0.001 F 5 21.11, P , 0.001 other fitness differences because, being a nuclear allele, it istransmitted by both sexes. If uninfected individuals are notpresent in the population, like in our laboratory population, fection that does not induce CI maintain itself in the host a mod2 host allele is, like a symbiont mod2 allele, a neutral population? A separate study suggests that C-Wolbachia in- duces a sex-ratio bias toward females (Vala et al. 2002).
To summarize, in a field situation, where uninfected in- Mathematical analysis shows that such an effect provides a dividuals are likely to be present, a mod2 host allele invades spreading mechanism for the bacteria (Egas et al. 2002).
if there is a mod1 allele present. Under the same circum- Mod1resc1 and mod2resc1 infection phenotypes were iso- stances a symbiont allele can only invade if it entails a lower lated from the rose strain. Crosses within isofemale line R3 cost to infected females than the resident mod1. Our results did not show induction of CI (Table 1), whereas CI was do not support the hypothesis that mod2 is less costly to induced in crosses within isofemale lines R1 and R2. Ad- infected females than mod1. Thus, if mod2 were a symbiont ditional crossing experiments demonstrated that the infection trait, it would be rare and essentially neutral in the field.
in R3 exhibits a mod2resc1 phenotype (Table 3). Thus, Consequently, in the absence of a cost effect to infected fe- mod1resc1 and mod2resc1 phenotypes co-exist within a sin- males, the probability of picking up a host mod2 in the field gle host population.
seems higher. Once in the laboratory, where uninfected in- Theory predicts that if a resident mod1resc1 Wolbachia is dividuals are not present, a mod2 host allele is a neutral trait.
present and the mod2resc1 type is more deleterious than the The probability of it drifting to fixation or to extinction de- resident mod1, then mod2 will be excluded by selection. If pends on its original frequencies. If introduced at interme- the mod2 and mod1 infections are equally harmful, then mod2 diate frequencies it could be maintained.
is essentially neutral (Hurst and McVean 1996). Its frequencymay drift around the level at which it was introduced and, Population Dynamics Consequences of a Host mod2 Allele assuming the mutation that causes the shift in mod is notvery common, disappear. Thus, the probability of finding a Hosts may mutate or otherwise protect the target sites of neutral mod2 symbiont seems low. A mod2resc1 Wolbachia modification by Wolbachia, or bacterial growth in males may increases in frequency when rare if it entails a lower cost to be prevented (Bressac and Rousset 1993; Poinsot et al. 1998; infected females than the resident mod1 (Prout 1994; Turelli McGraw et al. 2001). Two examples suggest mod2 host al- 1994; Hurst and McVean 1996). If uninfected individuals are leles may occur in Drosophila hosts. First, a Wolbachia strain present in the population, spread of the mod2resc1 type leads that does not induce CI in D. melanogaster, because it is both infections to extinction (Hurst and McVean 1996). If excluded from sperm cysts, induces CI when present D. si- uninfected individuals are not present, like in our infected mulans males (McGraw et al. 2001). Second, the observation laboratory populations, a less costly mod2 symbiont type that wAu, a symbiont type occurring in D. simulans, fails to would spread to fixation.
induce CI in flies from Australia (Hoffmann et al. 1996; Our results do not indicate any correlated differences be- Merc¸ot and Poinsot 1998b), but induces CI in some isofemale tween fecundity costs to infected females and CI induction.
lines of flies collected in Florida (Ballard et al. 1996).
Infected R1 and R3 females generally produce similar clutch It is therefore interesting to ask what happens after invasion sizes (Table 1). Average F1 mortality in W 3 W and W 3 by a mod2 host allele. It is conceivable that spread of a U crosses is 0.25 for R1, 0.16 for R2, and 0.21 for R3.
nuclear mod2 would result in conditions for reestablishment Therefore, the mod1resc1 infection in R2 incurs lower mor- of uninfected individuals, analogous to spread of a Wolbachia tality in broods of W females than the mod2resc1 infection mod2 (Hurst and McVean 1996). Simulations presented in infected with wAu in Australia constitute an example of hostallele–mediated dynamics such as those simulated here.
To conclude, we found within-population variation for mod. The possibility that the differences we observed are dueto host effects cannot be excluded. Simple mathematical anal-ysis of the dynamics of a host mod2 allele suggests interestingevolutionary implications and novel interpretations of exist-ing data. Clearly, more effort should be made to distinguishbetween Wolbachia and host mod2 alleles.
F. Vala was supported by Fundac¸a˜o para a Ciencia e Tec- nologia (scholarship reference: Praxis XXI/BD/9678/96).
Ballard, J. W. O., J. Hatzidakis, T. L. Karr, and M. Kreitman. 1996.
Reduced variation in Drosophila simulans mitochondrial DNA.
Genetics 144:1515–1528.
Bourtzis, K., S. L. Dobson, H. R. Braig, and S. L. O'Neill. 1998.
Rescuing Wolbachia have been overlooked. Nature 391:852–853.
Breeuwer, J. A. J. 1997. Wolbachia and cytoplasmic incompatibility in the spider mite Tetranychus urticae and T. turkestani. Heredity79:41–47.
Breeuwer, J. A. J., and G. Jacobs. 1996. Wolbachia: intracellular The effect of a host allele resistant to modification by manipulators of mite reproduction. Exp. Appl. Acarol. 20: Wolbachia on the population dynamics of the infection (see Ap- pendix and Table A1 for details). The dynamics depicted here are Bressac, C., and F. Rousset. 1993. The reproductive incompatibility based on m 5 0.9, F 5 0.9, H 5 0.1. With these parameters and system in Drosophila simulans: DAPI staining analysis of the in the absence of mod2 (p 5 0), there is an unstable equilib- Wolbachia symbionts in sperm cysts. J. Invertebr. Pathol. 61: rium at p 5 0.765. With initial conditions ps 0, the dynamics quickly converge to the Callaini, G., R. Dallai, and M. G. Riparbelli. 1997. Wolbachia- equilibrium p 5 0.014. After one generation, a mod2 induced delay of paternal chromatin condensation does not pre- allele was introduced (p 5 1025), thus mimicking a mutation in a vent maternal chromosomes from entering anaphase in incom- single infected host. (a) Frequency of types susceptible to modi- patible crosses of Drosophila simulans. J. Cell Sci. 110:271–280.
fication by Wolbachia (mod1): ps (dashed) and qs (solid). (b) Fre- Crozier, R. H. 1985. Adaptive consequences of male-haploidy. Pp.
quency of types resistant to modification by Wolbachia (mod2): pr 201–222 in W. Helle and M. W. Sabelis, ed. Spider mites: their (dashed) and qr (solid).
biology, natural enemies and control. Elsevier, Amsterdam.
De Boer, R. 1985. Reproductive barriers. Pp. 193–200 in W. Helle and M. W. Sabelis, ed. Spider mites: their biology, natural en- Figure 1 show that this is indeed the case (see the Appendix emies and control. Elsevier, Amsterdam.
for details). However, invasion by a host allele conferring Egas, M., F. Vala, and J. A. J. Breeuwer. 2002. On the evolution resistance to sperm modification by Wolbachia leads to es- of cytoplasmic incompatibility in haplodiploids. Evolution 56:1101–1109.
tablishment of resistant uninfecteds. Spread of a host mod2 Fialho, R. F., and L. Stevens. 2000. Male-killing Wolbachia in a allele differs from spread of a mod2 Wolbachia because the flour beetle. Proc. R. Soc. Lond. B 267:1469–1474.
uninfected population that results can only be recolonized by Field, A. 2000. Complex ANOVA (GLM2). Pp. 294–322 in D. B.
a Wolbachia using a novel modification site (i.e., a new in- Wright, ed. Discovering statistics using SPSS for Windows.
compatibility type). This could provide selective pressure for SAGE Publications, London.
Hartl, D. L., and A. G. Clark, 1997. Population substructure. Pp.
new incompatibility types. For example, in D. simulans five 117–137 in D. L. Hartl and A. G. Clark, eds. Principles of pop- Wolbachia strains have been identified (wRi, wHa, wAu, wKi, ulation genetics. Sinauer Associates, Sunderland, MA.
wMa, and wNo), a classification that is consistent with se- Hoffmann, A. A., and M. Turelli. 1997. Cytoplasmic incompatibility quence data from wsp and 16S rDNA (cf. James and Ballard in insects. Pp. 42–80 in S. L. O'Neill, A. A. Hoffmann, and J.
H. Werren, eds. Influential passengers. Oxford Univ. Press, Ox- 2000). These symbiont types correspond to four incompati- bility groups. It would be interesting to test if D. simulans Hoffmann, A. A., D. Clancy, and J. Duncan. 1996. Naturally oc- hosts resist modification by other simulans Wolbachia types.
curring Wolbachia infection in Drosophila simulans that does The theoretical result that a host mod2 allele may lead to not cause cytoplasmic incompatibility. Heredity 76:1–8.
symbiont exclusion is also interesting. For example, infec- Holden, P. R., J. F. Y. Brookfield, and P. Jones. 1993. Cloning and characterization of an fstZ homologue from a bacterial symbiont tions by wAu in Australia occur at zero to low frequencies of Drosophila melanogaster. Mol. Gen. Genet. 240:213–220.
in different populations (Hoffmann et al. 1996). The effect Hurst, L. D. 1991. The evolution of cytoplasmic incompatibility or of these infections on reproductive incompatibility was tested when spite can be successful. J. theor. Biol. 148:269–277.
on pooled samples from an Australian population and shown Hurst, L. D., and G. T. McVean. 1996. Clade selection, reversible evolution and the persistence of selfish elements: the evolution- not to cause CI. However, when isofemale lines were created ary dynamics of cytoplasmic incompatibility. Proc. R. Soc.
from Florida populations, CI was detected in some isofemale Lond. B. 263:97–104.
lines (Ballard et al. 1996). It is possible that populations James, A. C., and J. W. O. Ballard. 2000. Expression of cytoplasmic F. VALA ET AL.
incompatibility in Drosophila simulans and its impact on infec- 2000. Evidence for female mortality in Wolbachia-mediated cy- tion frequencies and distribution of Wolbachia pipientis. Evo- toplasmic incompatibility in haplodiploid insects: epidemiologic and evolutionary consequences. Evolution 54:191–200.
Jiggins, F. M., J. H. G. von der Schulenburg, G. D. D. Hurst, and Werren, J. H. 1997. Biology of Wolbachia. Annu. Rev. Entomol.
M. E. N. Majerus. 2001. Recombination confounds interpreta- tions of Wolbachia evolution. Proc. R. Soc. Lond. B 268: Werren, J. H., and J. D. Bartos. 2001. Recombination in Wolbachia.
Curr. Biol. 11:431–435.
McGraw, E. A., D. J. Merrit, J. N. Droller, and S. L. O'Neill. 2001.
Zhou, W., F. Rousset, and S. O'Neill. 1998. Phylogeny and PCR- Wolbachia mediated sperm modification is dependent on host based classification of Wolbachia strains using wsp gene se- genotype in Drosophila. Proc. R. Soc. Lond. B 268:2565–2570.
quences. Proc. R. Soc. Lond. B 265:509–515.
Merc¸ot, H., and D. Poinsot. 1998a. . and discovered on Mount Kilimanjaro. Nature 391:853.
Corresponding Editor: M. Riley ———. 1998b. Wolbachia transmission in a naturally bi-infected Drosophila simulans strain from New Caledonia. Entomol. Exp.
Appl. 86:97–103.
Poinsot, D., K. Bourtzis, G. Markakis, C. Savakis, and H. Merc¸ot.
We study the dynamics of allele frequencies after invasion of the 1998. Wolbachia transfer from Drosophila melanogaster into D. mod2 host allele, with a population genetic model that follows simulans: host effects and cytoplasmic incompatibility relation- Turelli (1994). The variables in our model are the frequency of ships. Genetics 150:227–237.
infected hosts susceptible to modification by Wolbachia (ps), the Prout, T. 1994. Some evolutionary possibilities for a microbe that frequency of infected hosts resistant to modification by Wolbachia causes incompatibility in its host. Evolution 48:909–911.
(pr), the frequency of uninfected hosts possessing the allele for Reed, K. M., and J. H. Werren, 1995. Induction of paternal genome susceptibility to modification by Wolbachia (qs), and the frequency loss by the paternal sex-ratio chromosome and cytoplasmic in- of uninfected hosts possessing the resistant allele to modification compatibility bacteria (Wolbachia): a comparative study of early by Wolbachia (qr). Other notation follows Turelli (1994); m is the embryonic events. Mol. Reprod. Dev. 40:408–418.
proportion infected offspring produced by an infected mother, F is Rousset, F., and M. Raymond. 1991. Cytoplasmic incompatibility the fecundity of infected females relative to uninfected females, in insects: Why sterilize females? Trends Ecol. Evol. 6:54–57.
and H is the hatchability in incompatible crosses (H 5 0 implies Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular clon- that CI results in 100% F1 mortality). Note that m, F, and H cor- ing a laboratory manual. Cold Spring Harbor Laboratory Press, respond to a, 1 2 U, and 1 2 k, respectively, in Hurst and McVean Cold Spring Harbor, NY.
Stouthamer, R., J. A. J. Breeuwer, and G. D. D. Hurst. 1999. Wol- Table A1 lists all possible matings, the frequency at which they bachia pipientis: microbial manipulator of arthropod reproduc- occur assuming random mating, and the expected distribution of tion. Annu. Rev. Microbiol. 53:71–102.
offspring over the four different host categories. Assuming non- Turelli, M. 1994. Evolution of incompatibility-inducing microbes overlapping generations and haploid genetics for reasons of model and their hosts. Evolution 48:1500–1513.
tractability, the dynamics of the population is described by four Vala, F., J. A. J. Breeuwer, and M. W. Sabelis. 2000. Wolbachia- difference equations. For example, the frequency of infected, sus- induced ‘hybrid breakdown' in the two-spotted spider mite Te- ceptible hosts in the next generation (p 9 s ) is obtained by summing tranychus urticae Koch. Proc. R. Soc. Lond. B 267:1931–1937.
all separate contributions to Ws individuals in Table A1, and di- Vala, F., T. Van Opijnen, J. A. J. Breeuwer, and M. W. Sabelis.
viding by the sum of all contributions: 2002. Genetic conflicts over sex ratio: mite-endosymbiont in- teractions. Am. Nat. In press.
p9 5 (p mF 1 ½p p mF 1 . . 1 ½p q mF) Van Opijnen, T., and J. A. J. Breeuwer. 1999. High temperatures 4 (psmF 1 . . 1 q ) eliminate Wolbachia, a cytoplasmic incompatibility inducing en- dosymbiont, from the two-spotted spider mite. Exp. Appl. Acar- Note that because p 1 1, one variable can be ol. 23:871–881.
eliminated, leaving three difference equations.
Vavre, F., F. Fleury, J. Varaldi, P. Fouillet, and M. Bouletreau.
The objective of this exercise is to investigate whether Hurst and TABLE A1. List of possible crosses between host types, the frequency by which they occur, and the distribution of offspring over the fourtypes in a population that segregates a host allele resistant to modification by Wolbachia. Host types are classified by infection status (W,infected; U, uninfected) and susceptibility to modification (s, susceptible or mod1; r, resistant or mod2). The columns Ws, Wr, Us, and Ur areexpressed in units of the clutch size of an uninfected female.
(1 2 m)F H ½ (1 2 m)F ½(1 2 m)F ½(1 2 m)F ½(1 2 m)F ½(1 2 m)F H ½(1 2 m)F H ½(1 2 m)F ½(1 2 m)F McVean's (1996) predictions hold assuming mod2 is a property of that m , 1 and that the two equilibria (ps, qs, 0, 0) exist (i.e., the the host. In other words, do uninfected individuals reestablish fol- unstable, or threshold, equilibrium and the high prevalence, or poly- lowing invasion by mod2? Using the same parameter values as Hurst morphic, internal equilibrium).
and McVean (1996, fig. 2), we find: (1) invasion of the mod2 host Note that if there were a cost to mod2, the uninfected population allele, followed by extinction of Wolbachia (Fig. 1); (2) the resulting would slowly return to the susceptible state. However, population uninfected host population is immune to any Wolbachia that uses immunity is provided by any frequency q .
0. Therefore, we argue the same modification site; (3) results (1) and (2) hold provided that reinvasion of Wolbachia requires a new modification site.

Source: http://www.environnement.ens.fr/IMG/file/DavidPDF/evolution2002.pdf