doi: 10.1046/j.1420-9101.2003.00676.x

What maintains noncytoplasmic incompatibility inducing Wolbachia in their hosts: a case study from a natural Drosophila yakuba population S. CHARLAT,* J. W. O. BALLARD  & H. MERC¸ OT* *Institut Jacques Monod, CNRS-Universite´s Paris 627, Laboratoire Dynamique du Ge´nome et Evolution, Jussieu, Paris Cedex, France  University of Iowa, Biological Sciences, Iowa City, IA, USA

Keywords:

Abstract

Cytoplasmic incompatibility; Drosophila yakuba; field work; Wolbachia.

Cytoplasmic incompatibility (CI) allows Wolbachia to invade hosts populations by specifically inducing sterility in crosses between infected males and uninfected females. In some species, non-CI inducing Wolbachia, that are thought to derive from CI-inducing ancestors, are common. In theory, the maintenance of such infections is not possible unless the bacterium is perfectly transmitted to offspring - and/or provides a fitness benefit to infected females. The present study aims to test this view by investigating a population of Drosophila yakuba from Gabon, West Africa. We did not find any evidence for CI using wild caught females. Infected females from the field transmitted the infection to 100% of their offspring. A positive effect on female fecundity was observed one generation after collecting, but this was not retrieved five generations later, using additional lines. Similarly, the presence of Wolbachia was found to affect mating behaviour, but the results of two experiments realized five generations apart were not consistent. Finally, Wolbachia was not found to affect sex ratio. Overall, our results would suggest that Wolbachia behaves like a neutral or nearly neutral trait in this species, and is maintained in the host by perfect maternal transmission.

Introduction The aim of this study is to test whether Wolbachia infection frequencies in a natural Drosophila yakuba population from Gabon (West Africa) were satisfactorily explained by current models. Wolbachia is a maternally inherited endocellular bacterium, widespread in arthropods and nematodes (O’Neill et al., 1997; Stouthamer et al., 1999). In its arthropod hosts, Wolbachia has evolved a number of ‘reproductive manipulations’ allowing invasion of uninfected populations. Based on current data, cytoplasmic incompatibility (CI) (Hoffmann & Turelli, 1997; Charlat et al., 2001) seems to be the most widespread phenomenon. In its simplest form, CI is expressed when males bearing the bacterium mate with uninfected females: such a cross results in embryo death. Correspondence (present address): Sylvain Charlat, University College London, Department of Biology, Wolfson House, 4 Stephenson Way, London, NW1 2HE, UK. Tel.: +689 56 52 87; e-mail: [email protected]

322

By contrast, crosses involving infected females are normally fertile, regardless of male infection status, so that infection frequency increases. Based on theoretical and empirical work in Drosophila simulans (Caspari & Watson, 1959; Fine, 1978; Turelli & Hoffmann, 1995; Hoffmann & Turelli, 1997), the spread of CI-inducing Wolbachia in uninfected populations, as well as its maintenance after invasion, are best understood by considering three factors: (i) the level of CI (the percentage of embryos that do not hatch because of CI in crosses between infected males and uninfected females), (ii) the maternal transmission efficiency and (iii) fitness effects. Basically, high levels of CI, efficient maternal transmission and low fitness costs will facilitate invasion of and maintenance in host populations. Understanding the long-term stability of Wolbachia-host associations requires the determination of the evolutionary trajectories of these different parameters. Turelli (1994) investigated this issue by delineating the selective pressures acting on Wolbachia and their hosts. He concluded that both host and symbiont factors are selected for increasing transmission rates and

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

What maintains Wolbachia in Drosophila yakuba

decreasing negative fitness effects. Predictions regarding CI levels are less straightforward. If infection is not fixed, host factors that can decrease CI can invade, because CI reduces hatching rates in crosses between infected males and uninfected females. On the contrary CI levels are selectively neutral to the bacterium within infected populations, as long as population structure is not too strong (Frank, 1998). This nonintuitive conclusion, also reached by Prout (1994) can be understood by noting that the bacterial factors determining CI levels are expressed only in males. Because Wolbachia is transmitted by females only, any variations affecting these determinants are neutral (unless one assumes they can have phenotypic effects on females through pleitropy). On the basis of such models, it has been hypothesized that long term Wolbachia-host co-evolution would lead to reduced CI levels (Turelli, 1994; Hurst & McVean, 1996). Consistently, non-CI inducing Wolbachia, that are thought to derive from CI-inducing ancestors, have been observed in several species (Bourtzis et al., 1994; Giordano et al., 1995; Rousset & Solignac, 1995; Bourtzis et al., 1996; Hoffmann et al., 1996; Bourtzis et al., 1998; Merc¸ot & Poinsot, 1998; Reynolds & Hoffmann, 2002; Charlat et al., 2003). Simple population models predict that such variants should be lost from their host unless transmission from mothers to offspring is perfect, or Wolbachia increases host fitness; a prediction that has previously been tested in two Drosophila species: D. melanogaster and D. simulans. In D. melanogaster, infected populations can be found throughout the whole species distribution (Solignac et al., 1994). CI does not appear to be expressed in the field, but it can be detected in the laboratory, especially if very young males are used; transmission rates are not perfect (95% confidence intervals ranging from 83 to 99.2%) and positive fitness effects are not apparent (Hoffmann et al., 1998; Olsen et al., 2001; Reynolds & Hoffmann, 2002; Weeks et al., 2002). Infection maintenance thus represents something of a paradox, the solution of which might lie in yet unidentified positive fitness effects. In D. simulans, some Australian populations, infected by the w Au variant, have been investigated (Hoffmann et al., 1996). Here, the infection was not found to cause CI. However, its maintenance in these populations is not confounding, because maternal transmission appears to be perfect. No positive or negative effects on host fitness have been detected, suggesting that the w Au infection is maintained as a neutral trait. Interestingly, a very closely related variant has been detected in the three species forming the Yakuba complex: D. yakuba, D. teissieri and D. santomea (Lachaise et al., 2000). Here we report on a field study realized on a D. yakuba population from Gabon (West Africa). In an attempt to test if the infection frequency in this species was satisfactorily explained by current models, we addressed the following questions: (i) what is the infection frequency, (ii) is there any evidence for CI, (iii) how efficient is maternal transmis-

323

sion and (iv) can we identify other effects that could select for Wolbachia maintenance.

Material and methods Line collection Thirty three D. yakuba females were collected in Gabon in March 2002 on the border of the Ogoue River, about 250 km east from Libreville (GPS position: latitude: S 0006.073¢; longitude: E 01135.594¢), using banana and tomato baits. Females were kept in plastic vials with instant medium (formula 4–2, Carolina biological supply company, Burlington, NC 27215, USA) until arrival in the laboratory (6 days later). Fourteen females survived the trip back to Paris. Additionally, ethanol preserved samples were obtained from the Ogoue River (98 flies), Franceville (15 flies) and Libreville (three flies) in the Gabon, and from Nguti (33 flies) in the Cameroon (sample kindly provided by Peter Andolfatto, University of Edinburgh). These were used to gain a more robust estimate of the Wolbachia infection frequency. Wolbachia detection and identification DNA was extracted according to O’Neill et al. (1992). The presence or absence of Wolbachia was determined by polymerase chain reaction (PCR) amplification using 16S general primers (76F, 994R) (O’Neill et al., 1992). When infection was not detected, quality of DNA extracts was checked by amplifying mitochondrial DNA using primers Dick and Pat from Simon et al. (1994). If these mtDNA primers did not successfully amplify DNA the sample was excluded from further analysis. The identity of the Wolbachia variant in our sample was first checked by amplifying a fragment of the wsp gene with primers specifically designed for the Mel Wolbachia clade (Zhou et al., 1998; Riegler & Stauffer, 2002), to which w Au belongs. Sequencing was performed in seven lines (named in the Results section). PCR products were obtained with general wsp primers 81F and 691R (Zhou et al., 1998), purified using a GenElute PCR clean-up kit (Sigma, The Woodlands, TX, USA), and sent for sequencing with primers 81F and 691R. Both strands were sequenced for each line. Transmission rates Transmission rates were estimated in four lines from G0 to G1 and from G4 to G5, based on the proportion of infected adults in G1 and G5, respectively. Because there is no expression of CI (see Results section), this estimate is not biased: uninfected embryos do not have lower probability of survival than infected ones. If CI had been detected, it would have been necessary to cross-infected females with uninfected males in order to measure transmission efficiency.

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

324

S. CHARLAT ET AL.

Cytoplasmic incompatibility assays Upon arrival in the laboratory in Paris, 14 wild caught females (generation 0: G0) were left to oviposit for 48 h at 25 C on Petri dishes filled with axenic medium (David, 1962) coloured with neutral red (making egg counting easier) and thinly layered with yeast. However, among the 14 G0 females, only nine laid more than 10 eggs (three infected females and six uninfected). Eggs were left to hatch for 24 h and the number of hatched and unhatched eggs were counted. Embryonic mortality was determined as the percentage of unhatched eggs. The infection status of G0 females was then checked by PCR. G1 flies were obtained from vials where G0 had laid before arrival in the lab, allowing an experiment with larger sample size, and with more controlled conditions to be performed. Forty-eight virgin males and virgin females from the wild-caught lines (see lines names in the Results section) were collected from 2 to 6 days prior to the experiment, which was performed as follows. Mating was controlled and crosses where copulation lasted for less than 15 min were discarded, in order to ensure insemination. Inseminated females were left to lay and embryonic mortality was measured as described above. Finally, the infection status of all flies was checked by PCR. This protocol was repeated in G5 with 61 crosses. Fertility and fecundity assays Fertility and fecundity data were obtained during the CI experiment. Fecundity was estimated as the total number of eggs laid during 48 h (98 crosses in G1, 113 crosses in G5). Fertility was estimated as the hatching rates in crosses with uninfected males (48 crosses in G1, 58 crosses in G5). Mating behaviour Two aspects of mating behaviour were monitored: the time separating contact from copulation, and the duration of copulation. These measures were performed during CI assays, with one male and one female in each vial (80 crosses in G1, 125 crosses in G5). Sex-ratio assays In G5, females used in CI assays were left to lay in vials filled with axenic medium for 48 hours at 25 C. Sex ratio was estimated by counting males and females emerging from these vials (87 crosses). Statistical analysis Embryonic mortality was often close to 0%, so that data was not normally distributed. We therefore used

nonparametrical tests to analyze fertility and CI data (Wilcoxon tests were used when two data sets were to be compared, Kruskall–Wallis tests were used for comparing more than two data sets). Fecundity and behavioural data were analyzed as a nested A N O V A . Sex-ratio results were analyzed as a nested A N O V A after arcsine transformation. A N O V A s were performed using the JMP software package (vers. 3.2.2 for the MacIntosh; SAS-Institute, 1995). Lines denomination Throughout this article, the following nomenclature will be used to name the lines collected: ‘GN’ (standing for Gabon), followed by the isofemale line number, followed by ‘-W’ (standing for ‘Wolbachia’) for infected lines or ‘-U’ (standing for ‘uninfected’) for uninfected lines. For instance, GN42-W is infected by Wolbachia and was the 42nd isofemale line collected in Gabon.

Results Infection frequencies Fourteen wild-caught lines were established from flies collected in Gabon. Among these, five were infected and nine uninfected; infection status was confirmed each generation until G5. Infection frequency was also estimated using materials stored in ethanol. Ninety-eight individuals were analyzed and 32 were found infected using 16S general primers. Using a pool from isofemale lines and samples stored in ethanol, the estimated infection frequency was thus 33.9% (n ¼ 112, 95% confidence interval: 25.0–42.8%). Fifteen individuals stored in ethanol from another site (Franceville, about 250 km east) were also analyzed. Four were found infected and one negative sample did not amplify mitochondrial DNA, making the estimated infection frequency 28.6% (n ¼ 14, 4.9–52.3%). Finally, three ethanol stored flies from the Atlantic Ocean coast (Libreville) were analyzed, two of which were positive and one a true negative. In addition to these samples from Gabon, we were able to estimate infection frequency in a population from Nguti (Cameroon, Banyang-Mbo Research Station, Nguti, Southwest Province). Following collection, individuals of generation 2 or 3 stored in ethanol were analyzed. Nine lines of 33 were found infected, and 24 were true negative, making the estimated infection frequency 27.3% (n ¼ 33, 12.0–42.5%). Although we could not determine infection frequency in G0, we are confident that this infection frequency represents the wild type situation. Indeed, three to six individuals were tested separately for infection in every line, and either all or none were found infected. Furthermore, as showed below, transmission rates appear to be 100% in D. yakuba.

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

What maintains Wolbachia in Drosophila yakuba

Infection identity A 588 bp fragment of the wsp gene was sequenced in five infected lines from Gabon (GN40-W, GN42-W, GN43-W, and GN45-W, GN54-W) and in two infected lines from Cameroon. The seven sequences obtained were identical to each other. Sequences from Gabon and Cameroon were deposited in GenBank under references AY291346 and AY291348, respectively. Based on earlier work (Lachaise et al., 2000), we expected this sequence to be identical to the wsp sequence from w Au, naturally infecting D. simulans (AF020067 or AF290890). However, we found a 1 bp difference between w AuSim (Zhou et al., 1998; Charlat et al., 2003) and this w Au-like Wolbachia. In fact, the sequence we found is identical to the one of w Cer2 (AF418557), a Wolbachia naturally infecting the cherry fruit fly Rhagoletis cerasi (Riegler & Stauffer, 2002). Notably, we obtained an identical sequence from two D. teissieri lines from Gabon (GenBank AY291347) (not shown). Transmission efficiency To measure the ability of females from the field to transmit infection to their offspring, a total of 208 G1 flies (110 males and 98 females) from four different infected G0 females were tested for the presence of Wolbachia. Perfect transmission was found in the four lines (no variation between lines), so that data can be pooled, making the estimated transmission efficiency 100% (n ¼ 208, lower limit of the 95% confidence interval: 0.98). On the contrary, the 64 G1 flies from lines that had been found uninfected in G0 were all uninfected, as expected. The efficiency of transmission was measured again from G4 to G5 using four lines derived from the same four infected G0 females. A total of 111 flies from infected mothers (57 males and 54 females) were tested for infection. All of them were found positive.

325

what is often seen in laboratory lines, where inbreeding depression usually causes at least 10% embryonic mortality in the absence of Wolbachia (see below for an example). Clearly, uninfected females did not have a higher embryonic mortality than infected ones, suggesting that there was no CI in the field. To corroborate the absence of incompatibility, we repeated the CI tests in G1. Virgin G1 flies were collected in two infected lines (GN40-W and GN43-W) and two uninfected lines (GN50-U and GN65-U). In order to test if CI was expressed, infected males and uninfected males were crossed with uninfected females. Results are presented in Table 1. Here again, embryonic mortality was very low (36 unhatched eggs over 3456). No significant difference was found between the four data sets (Kruskal–Wallis, H ¼ 3.13, d.f. ¼ 3, n.s.), and no effect of male infection status was detected (Wilcoxon, W ¼ 1.04, d.f. ¼ 1, n.s.). These data support the results obtained from field collected flies. The experiment was repeated in G5, using the same lines as in G1, together with two additional infected lines (GN42-W and GN45-W) and two additional uninfected lines (GN52-U and GN67-U). Results are presented in Table 1. In general, embryonic mortality was much higher than in G1, presumably because of inbreeding depression (these are isofemale lines). No significant difference was found between the eight data sets (Kruskal–Wallis, H ¼ 10.54, d.f. ¼ 7, n.s.), and no effect of male infection status was detected (Wilcoxon, W ¼ 0.54, d.f. ¼ 1, n.s.). Thus, there was no evidence for CI, or for variations between lines. Inbreeding depression weakens the power of this experiment, and it is arguable that crosses among lines would have circumvented this problem. However, at the time of this experiment, fertility had been previously tested only in G0 with (obviously) low inbreeding depression. We were Table 1 Cytoplasmic incompatibility assay in G1 and G5: descriptive statistics. Male

No. cross

Mean EM (%)

SE (%)

No. eggs

G1 experiment GN50-U GN65-U GN40-W GN43-W

10 13 10 15

2.2 0.9 0.9 0.8

0.8 0.3 0.3 0.2

600 993 706 1157

G5 experiment GN49-U GN50-U GN65-U GN67-U GN40-W GN42-W GN43-W GN45-W

8 8 8 8 8 6 8 7

23.9 9.5 15.7 11.9 50.6 4.3 2.7 17.3

14.8 6.3 12.7 8.7 18.3 1.7 1.7 15.0

577 744 711 711 583 458 686 587

Cytoplasmic incompatibility assays The experiment using G0 flies was carried out totally blindly, as the infection status of the females from the field were unknown. Based on previous estimates of infection frequency (Lachaise et al., 2000), we were however expecting a polytypic situation (with both infected and uninfected individuals) that would allow to compare embryonic mortality in offspring from infected vs. uninfected mothers. This was the case. However, among the 14 G0 females, only nine laid more than 10 eggs (three infected females and six uninfected). The low number of eggs laid might have been caused by female age, hard conditions in the field, and a long trip from Africa to Europe. Embryonic mortality was very low in all crosses: only one egg over 286 did not hatch (from one uninfected female). This is clearly different from

All males were crossed with uninfected females. EM: embryonic mortality.

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

326

S. CHARLAT ET AL.

Table 2 Fertility assay in G1 and G5: descriptive statistics. SE (%)

Table 3 Fecundity in G1 and G5: descriptive statistics.

Female

No. cross

Mean HR (%)

No. eggs

G1 experiment GN50-U GN65-U GN40-W GN43-W

11 12 13 12

97.7 99.2 99.3 97.6

0.7 0.4 0.3 1.5

725 868 1121 1164

G5 experiment GN49-U GN50-U GN65-U GN67-U GN40-W GN42-W GN43-W GN45-W

8 8 8 8 6 8 5 7

59.4 95.6 98.5 85.4 98.4 79.2 87.7 70.2

14.3 1.5 1.0 12.3 1.2 5.2 9.3 19.2

815 632 714 582 500 710 500 649

All females were crossed with uninfected males. HR: hatching rates.

therefore unaware that inbreeding depression would be so strong. Fertility and fecundity In G1, we tested whether Wolbachia could affect female fertility by crossing uninfected males with females from two infected and two uninfected lines (GN40-W, GN43W, GN50-U and GN65-U). The results are presented in Table 2. Fertility was very high in all crosses. No significant difference was found between the four data sets (Kruskal–Wallis, H ¼ 6.07, d.f. ¼ 3, n.s.), and no effect of female infection status was detected (Wilcoxon, W ¼ 0.66, d.f. ¼ 1, n.s.). The experiment was repeated in G5 using the same four lines as in G1, with two infected and two uninfected lines added (GN42-W, GN45-W, GN52-U and GN67-U). As showed in Table 2, fertility was much lower and more variable. Significant heterogeneity was found among lines (Kruskal–Wallis, H ¼ 19.9, d.f. ¼ 7, P < 0.01), but there was no effect of female infection status (Wilcoxon, W ¼ 0.86, d.f. ¼ 1, n.s.). Again, it appears that some lines were suffering from strong inbreeding depression. Too few eggs were laid in G0 to make meaningful fecundity estimates. In G1, the effects of infection on female fecundity were tested by comparing the number of eggs laid by infected and uninfected females (lines GN40-W, GN43-W, GN50-U and GN65-U), in crosses involving infected and uninfected males. The results, shown in Table 3, were analyzed by A N O V A (Table 4). A significant effect of female infection status was found, with infected females laying more eggs in average than uninfected ones (88.3 vs. 72). However, a possible bias in the fecundity data must be noted: because CI assays require many G1 males and females, the infected and uninfected lines used (GN40-W, GN43-W, GN50-U and GN65-U) derived from the most fecund G0 females in

WM

Female

No. cross

No. eggs

SE

G1 experiment U GN40-W U GN43-W U GN50-U U GN65-U W GN40-W W GN43-W W GN50-U W GN65-U

13 12 11 12 13 12 12 13

86.2 97.0 65.9 72.3 89.8 80.3 76.2 72.9

4.8 3.7 5.1 6.7 3.7 7.9 4.2 4.5

G5 experiment U GN40-W U GN42-W U GN43-W U GN45-W U GN49-U U GN50-U U GN65-U U GN67-U W GN40-W W GN42-W W GN43-W W GN45-W W GN49-U W GN50-U W GN65-U W GN67-U

6 8 5 7 8 8 8 8 5 7 8 8 7 8 7 7

83.3 88.7 100.0 92.7 101.9 79.0 89.2 72.7 84.0 91.3 92.5 87.5 93.0 79.9 85.4 60.9

13.4 6.9 5.4 10.0 10.7 7.1 6.0 11.0 9.9 6.9 9.9 8.6 15.6 10.0 13.0 15.2

For clarity, male lines with identical infection status were pooled in this table, after checking that male lines did not differ significantly (see Table 4). WM: male infection status (U: uninfected, W: infected). No. eggs: Mean number of eggs.

Table 4 Fecundity in G1 and G5:

A N O V A s.

Source

d.f.

Mean square

F

P

G1 experiment WM WF LM (WM) LF (WF) WM · WF LM (WM) · LF (WF) Error

1 1 2 2 1 4 87

15.45 6678.04 667.24 4.32 819.83 99.97 315.15

0.05 20.95 2.09 0.01 2.57 0.31

0.83 <0.01 0.13 0.99 0.11 0.87

G5 experiment WM WF LM (WM) LF (WF) WM · WF LM (WM) · LF (WF) Error

1 1 6 6 1 36 63

103.38 1661.60 455.20 1534.82 296.10 677.57 627.07

0.16 2.65 0.73 2.45 0.47 1.08

0.69 0.11 0.63 0.03 0.49 0.39

WM, male infection status (infected/uninfected); WF, female infection status (infected/uninfected); LM, line male; LF, line female. LM and LF are nested within WM and WF, respectively.

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

What maintains Wolbachia in Drosophila yakuba

each category, having laid respectively 39, 71, 23 and 40 eggs. The effect observed might result from this nonrandom sampling, or also from small sample size giving lines specific effects. The G5 experiment allowed us to further investigate the fecundity of infected flies. The two additional infected and uninfected lines (GN42-W, GN45-W, GN52-U and GN67-U) were randomly chosen. The results, shown in Table 3, were analyzed by A N O V A (Table 4). Here no effect of infection status was detected, but the line female (LF) effect was found significant at the 5% threshold, confirming that the effect observed in G1 might result from an experimental bias. Mating behaviour We investigated potential effects of infection status on sexual behaviour by measuring the time between contact and copulation and the duration of copulation in single pair crosses. This experiment was performed in G1 using females from two infected and two uninfected lines (GN40-W, GN43-W, GN50-U and GN65-U), mated with infected and uninfected males. The results were analyzed by A N O V A (data not shown). No factor was found to affect time before copulation significantly, but copulations were significantly longer with infected than with uninfected females (40.4 vs. 36.1 min). The experiment was repeated in G5 using the same four lines as in G1, with two infected and two uninfected lines added (GN42-W, GN45-W, GN52-U and GN67-U). The results were analyzed by A N O V A (data not shown). Here female infection status was found to affect time before copulation significantly, with infected females tending to mate earlier than uninfected ones (49.8 vs. 71.4 min). No factors affected copulation duration.

327

Table 5 Sex ratio in G5: descriptive statistics. WM

WF

N

SR (%)

SE (%)

U U W W

U W U W

25 17 23 22

53.0 48.3 48.4 50.9

1.3 1.7 1.4 1.1

WM, male infection status (U: uninfected, W: infected); WF, female infection status (U: uninfected, W: infected); SR, sex ratio (male/ total number of adults).

Table 6 Sex ratio in G5:

ANOVA.

Source

d.f.

Mean square

F

P

WM WF LM (WM) LF (WF) WM · WF LM (WM) · LF (WF) Error

1 1 6 6 1 36 35

0.0017 0.0989 0.0807 0.0448 0.0160 0.0042 0.0033

0.52 0.01 0.93 1.33 4.83 1.423

0.48 0.91 0.48 0.27 0.03 0.15

WM, male infection status (infected/uninfected); WF, female infection status (infected/uninfected); LM, line male; LF, line female LM and LF are nested within WM and WF, respectively.

Discussion In this study D. yakuba flies were collected in West Africa in order to test whether Wolbachia infection frequencies can be explained by current models. Every effort was made to collect and maintain large numbers of flies from Africa to Europe, but the survival rate was low: only 14 females survived back to the laboratory, five of which were infected. These data nevertheless provide a unique window into a naturally occurring population.

Sex ratio In G5, we tested the effect of female infection status on sex ratio using females from four infected and four uninfected lines (GN40-W, GN42-W, GN43-W, GN45-W, GN50-U, GN52-U, GN65-U and GN67-U), mated with infected and uninfected males. The results, presented in Table 5, were analyzed by A N O V A after arcsine transformation (Table 6). Most importantly, the Wolbachia female (WF) factor was not found significant, suggesting that infected females do not produce female biased progeny. A significant interaction between male and female infection status was detected. Infected females tended to produce more males when mated with infected males rather than uninfected males (male proportion 50.9% vs. 48.3%), while the opposite was observed for uninfected females, who tended to produce fewer males when mated with infected males rather than uninfected males (male proportion 48.4% vs. 53.0%). Overall, these effects were quantitatively very small.

Parameter estimates In the population under study, the observed infection frequency was 33.9% (n ¼ 112, 25.0–42.8%), which is higher than a previous estimate performed at this same site (9.26%, n ¼ 54, 1.53–16.99%) (Lachaise et al., 2000). Based on smaller samples, infection frequencies were estimated in two other sites as 28.6% (n ¼ 14, 95% confidence interval 4.9–52.3%) and 27.3% (n ¼ 33, 95% confidence interval 15.5–39.0%). Sequencing revealed that the infection in D. yakuba can be distinguished from the w Au Wolbachia from D. simulans, based on a single substitution within the wsp gene. In fact, the infection shows the same wsp sequence as wCer2, a Wolbachia naturally infecting the cherry fruit fly R. cerasi (Riegler & Stauffer, 2002). We investigated some of the parameters that are known to determine infection frequency at equilibrium: transmission efficiency, CI levels, effects on host fitness

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

328

S. CHARLAT ET AL.

and sex ratio. Transmission from wild caught infected females to their offspring was found perfect: no uninfected individual was detected among 208 tested. The same was true five generations later, where no uninfected individual was found among 111 tested. In previous experiments based on lab-maintained lines, the Wolbachia variant from D. yakuba was not thought to induce CI (S. Zabalou, A. Nirgianaki, SC, HM and K. Bourtzis, unpublished data). The absence of CI in the field was thus expected, as CI expression is usually found higher in the lab than in nature (Hoffmann et al., 1998; Turelli & Hoffmann, 1995). Infected G0 females did not show higher hatching rates than uninfected ones, but the sample was very small. In a bigger experiment with two infected and two uninfected lines, no CI was detected in G1. The same conclusion was derived from an experiment conducted five generations after collecting using four infected and four uninfected lines, although embryonic mortality was in average much higher, presumably because of five generations of inbreeding. No effect of Wolbachia infection on female fertility was found: in crosses with uninfected males, infected females did not show higher hatching rates than uninfected ones. An effect of Wolbachia infection on fecundity was observed in G1, but this possibly results from the sample size. Consistent with this interpretation, the fecundity benefit was not detected in the G5 experiment, where additional lines had been included. Fecundity benefits have been observed in other dipteran species (Dobson et al., 2002). Additional experiments would be necessary to further investigate this issue in D. yakuba. Although these traits are probably not important components of female fitness, our CI-assay protocol also allowed us to investigate potential effects of female infection status on time before copulation and copulation duration. Although some effects were observed sporadically, no clear pattern emerged. In G1, infected females were found to mate longer but not earlier, while in G5 they mated earlier but not longer. If time before copulation and copulation duration are indicators of male choice, these results can be interpreted as males preferring infected rather than uninfected females. However, the discrepancy between generations is confounding. We examined sex-ratio effects in G5, and found no female biased progeny in offspring from infected females. Thus, sex-ratio effects do not seem to contribute here to Wolbachia maintenance. A significant (although quantitatively small), interaction between male and female infection status was observed, which we fail to interpret in adaptive terms. Wolbachia maintenance (and its origin) Infection dynamics models predict that in the absence of CI expression or sex ratio distortion, Wolbachia infections should be lost from natural populations unless beneficial to the host or perfectly transmitted from infected mothers

to their offspring (reviewed in Hoffmann & Turelli, 1997). Our observations fit with this prediction: here transmission was found to be perfect, and a possible positive effect on host fecundity was observed. If, in the lack of certainty, one neglects the fecundity effect, the picture is very similar to that obtained earlier for w Au in Australian D. simulans (Hoffmann et al., 1996). In these two cases, Wolbachia seems to be maintained like a neutral trait. It is notable that Wolbachia infects D. yakuba at low infection frequencies. This observation prompts the question of the origin of uninfected flies, for which we see two hypotheses: these must derive either from cytoplasmic lineages that have never been infected or, if transmission efficiency is in fact less than perfect, from originally infected lineages (these two possibilities will be respectively referred to as the ‘never infected’ and ‘once infected’ hypotheses). Under the ‘never infected’ hypothesis two evolutionary scenarios can be proposed. First, Wolbachia might derive from recent horizontal transmission(s) of a non-CI inducing Wolbachia behaving like a neutral trait. The observed pattern would then represent a transitory equilibrium between horizontal transfer and drift. This view would imply that horizontal transfers are sufficiently rare and/or recent for the only possible long-term equilibrium (that is, fixation of the infection) not to have been reached. A second possibility is that Wolbachia has been once fixed in some, but not all, D. yakuba populations and that admixture followed. Such fixation could have occurred through recurrent horizontal transmission and drift, without CI expression, or much faster with the help of CI, which would then have been secondarily lost. The ‘once infected’ hypothesis implies that maternal transmission is in fact less than perfect, a possibility that our results do not totally rule out because of the small number of infected lines tested. The current situation might then be transitory, if infection leakage takes place at a higher rate than horizontal transmission, or stable, if the two processes occur at similar rates. As was the case under the ‘never infected’ hypothesis, the possibility of and ancestral expression of CI followed by its secondary loss is not ruled out. The ‘never infected’ and ‘once infected’ hypotheses might be tested with the help of mitochondrial sequence data as both mitochondria and Wolbachia are maternally inherited. Ballard & Kreitman (1994) showed that D. yakuba has relatively high amounts of mitochondrial DNA variation. Although the infection status of the lines used in this study was unknown, it suggests that Wolbachia has not caused CI in the recent past. The present work confirms that non-CI inducing Wolbachia can be maintained in natural populations. Now that Wolbachia screenings are based on PCR rather than phenotypic effect, similar cases of infections without any apparent consequences will probably prove to be common; probably not as common as they really are, as

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

What maintains Wolbachia in Drosophila yakuba

such infections, as sex ratio distorters (Jiggins et al., 2001), can persist at low frequencies and thus remain out of sight.

Acknowledgments We wish to thank the Gabonese ‘Ministe`re des Eaux et Foreˆts et du Reboisement’ and ‘Ministe`re de l’enseignement supe´rieur, de la recherche et de l’innovation technologique’ for allowing us to perform sampling and the ‘Centre International de Recherche de Franceville’ and giving access to a fantastic sampling site. We thank Peter Andolfatto for sending D. yakuba lines from Banyang-Mbo Research Station, Nguti, Southwest Province, Cameroon, and to the Wildlife Conservation Society and the Government of Cameroon, for permitting the collection of these lines and to Avis James and Kostas Bourtzis for commenting a previous version of this article. We are also grateful to Daniel Lachaise for invaluable help with species identification, and to Guillaume Charlat for contributing to the experiments, and to Vale´rie Delmarre and Chantal Labellie for technical assistance. Funding was provided by the National Science Foundation Grant No. DEB-9702824.

References Ballard, J.W.O. & Kreitman, M. 1994. Unraveling selection in the mitochondrial genome of Drosophila. Genetics 138: 757– 772. Bourtzis, K., Nirgianaki, A., Onyango, P. & Savakis, C. 1994. A prokaryotic dnaA sequence in Drosophila melanogaster: Wolbachia infection and cytoplasmic incompatibility among laboratory strains. Insect Mol. Biol. 3: 131–142. Bourtzis, K., Nirgianaki, A., Markakis, G. & Savakis, C. 1996. Wolbachia infection and cytoplasmic incompatibility in Drosophila species. Genetics 144: 1063–1073. Bourtzis, K., Dobson, S.L., Braig, H.R. & O’Neill, S.L. 1998. Rescuing Wolbachia have been overlooked. Nature 391: 852– 853. Caspari, E. & Watson, G.S. 1959. On the evolutionary importance of cytoplasmic sterility in mosquitoes. Evolution 13: 568–570. Charlat, S., Bourtzis, K. & Merc¸ot, H. 2001. Wolbachia-induced cytoplasmic incompatibility. In Symbiosis: mechanisms and model systems (J. Seckbach, ed.), pp. 621–644, Kluwer Academic Publisher, Dordrecht. Charlat, S., Le Chat, L. & Mercot, H. 2003. Characterization of non-cytoplasmic incompatibility inducing Wolbachia in two continental African populations of Drosophila simulans. Heredity 90: 49–55. David, J. 1962. A new medium for rearing Drosophila in axenic conditions. Dros. Inf. Serv. 93: 28. Dobson, S.L., Marsland, E.J. & Rattanadechakul, W. 2002. Mutualistic Wolbachia infection in Aedes albopictus. Accelerating cytoplasmic drive. Genetics 160: 1087–1094. Fine, P.E.M. 1978. On the dynamics of symbiont-dependent cytoplasmic incompatibility in Culicine mosquitoes. J. Invertebr. Pathol. 30: 10–18.

329

Frank, S.A. 1998. Cytoplasmic incompatibility and population structure. J. Theor. Biol. 192: 213–218. Giordano, R., O’Neill, S.L. & Robertson, H.M. 1995. Wolbachia infections and the expression of cytoplasmic incompatibility in Drosophila sechellia and D. mauritiana. Genetics 140: 1307–1317. Hoffmann, A.A. & Turelli, M. 1997. Cytoplasmic incompatibility in insects. In: Influential Passengers: Inherited Microorganisms and Arthropod Reproduction (S.L. O’Neill, A.A. Hoffmann & J.H. Werren, eds), pp. 42–80, Oxford University Press, Oxford. Hoffmann, A.A., Clancy, D. & Duncan, J. 1996. Naturallyoccurring Wolbachia infection in Drosophila simulans that does not cause cytoplasmic incompatibility. Heredity 76: 1–8. Hoffmann, A.A., Hercus, M. & Dagher, H. 1998. Population dynamics of the Wolbachia infection causing cytoplasmic incompatability in Drosophilia melanogaster. Genetics 148: 221–231. Hurst, L.D. & McVean, G. 1996. Clade selection, reversible evolution and the persistence of selfish elements: the evolutionary dynamics of cytoplasmic incompatibility. P. Roy. Soc. Lond. B 263: 97–104. Jiggins, F.M., Bentley, J.K., Majerus, M.E. & Hurst, G.D. 2001. How many species are infected with Wolbachia ? Cryptic sex ratio distorters revealed to be common by intensive sampling. P. Roy. Soc. Lond. B 268: 1123–1126. Lachaise, D., Harry, M., Solignac, M., Lemeunier, F., Benassi, V. & Cariou, M.L. 2000. Evolutionary novelties in islands: Drosophila santomea, a new melanogaster sister species from Sao Tome. P. Roy. Soc. Lond. B 267: 1487–1495. Merc¸ot, H. & Poinsot, D. 1998. Rescuing Wolbachia have been overlooked and discovered on Mount Kilimanjaro. Nature 391: 853. O’Neill, S.L., Giordano, R., Colbert, A.M., Karr, T.L. & Robertson, H.M. 1992. 16S rRNA phylogenetic analysis of the bacterial endosymbionts associated with cytoplasmic incompatibility in insects. Proc. Natl. Acad. Sci. USA 89: 2699–2702. O’Neill, S.L., Hoffmann, A.A. & Werren, J.H. (eds) 1997. Influential Passengers: Inherited Microorganisms and Arthropod Reproduction. Oxford University Press, Oxford. Olsen, K., Reynolds, K.T. & Hoffmann, A.A. 2001. A field cage test of the effects of the endosymbiont Wolbachia on Drosophila melanogaster. Heredity. 86: 731–737. Prout, T. 1994. Some evolutionary possibilities for a microbe that causes incompatibility in its host. Evolution 48: 909–911. Reynolds, K.T. & Hoffmann, A.A. 2002. Male age, host effects and the weak expression or non-expression of cytoplasmic incompatibility in Drosophila strains infected by maternally transmitted Wolbachia. Genet. Res. 80: 79–87. Riegler, M. & Stauffer, C. 2002. Wolbachia infections and superinfections in cytoplasmically incompatible populations of the European cherry fruit fly Rhagoletis cerasi (Diptera, Tephritidae). Mol. Ecol. 11: 2425–2434. Rousset, F. & Solignac, M. 1995. Evolution of single and double Wolbachia symbioses during speciation in the Drosophila simulans complex. P. Natl. Acad. Sci. USA 92: 6389–6393. SAS-Institute. 1995. JMP Statistics and Graphics Guide. Ver. 3.2.2. SAS Institute, Inc., Cary, NC. Simon, C., Frati, F., Beckenbach, A., Crespi, B., Liu, H. & Flook, P. 1994. Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved PCR primers.. Ann. Entomol. Soc. Am. 87: 651–701. Solignac, M., Vautrin, D. & Rousset, F. 1994. Widespread occurence of the proteobacteria Wolbachia and partial incom-

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD

330

S. CHARLAT ET AL.

patibility in Drosophila melanogaster. C. R. Acad. Sci. III-Vie 317: 461–470. Stouthamer, R., Breeuwer, J.A. & Hurst, G.D. 1999. Wolbachia pipientis: microbial manipulator of arthropod reproduction. Ann. Rev. Microbiol. 53: 71–102. Turelli, M. 1994. Evolution of incompatibility-inducing microbes and their hosts. Evolution 48: 1500–1513. Turelli, M. & Hoffmann, A.A. 1995. Cytoplasmic incompatibility in Drosophila simulans: dynamics and parameter estimates from natural populations. Genetics 140: 1319–1338.

Weeks, A.R., Reynolds, K.T. & Hoffmann, A.A. 2002. Wolbachia dynamics and host effects: what has (and has not) been demonstrated? Trends. Ecol. Evol. 17: 257–262. Zhou, W., Rousset, F. & O’Neil, S. 1998. Phylogeny and PCRbased classification of Wolbachia strains using wsp gene sequences. P. Roy. Soc. Lond. B 265: 509–515. Received 29 May 2003; revised 9 October 2003; accepted 27 October 2003

J. EVOL. BIOL. 17 (2004) 322–330 ª 2004 BLACKWELL PUBLISHING LTD