Evolution of the Proportions of Two Sigma Viral Types in Experimental Populations of Drosophila melanogaster in the Absence of the Allele That Is Restrictive of Viral Multiplication

Corresponding author: Annie Fleuriet, Université Blaise Pascal, 63177 Aubière Cedex, France.

Communicating editor: A. A. HOFFMANN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A minority of flies in natural populations of Drosophila melanogaster are endemically infected by a rhabdovirus, sigma. The virus is vertically transmitted through male and female gametes. Two alleles of a fly locus, the ref(2)P locus, are present as a polymorphism in all populations: O permissive, and P restrictive for viral multiplication and transmission. Two viral types are known, Type I, which is very sensitive to the P allele, and Type II, which is more resistant. Previous observations have shown that, in presence of the P allele, viral Type II is selected for, in both natural and experimental populations. The aim of the present study was to determine whether, in the absence of P, Type I is selected for, or whether the two types are equivalent. For this purpose, experimental populations deprived of the P allele and differing in the initial proportions of the two viral types were established. After several generations, and despite a possible bias toward Type I, the frequencies of Type I and Type II clones differed in the various populations, depending on their initial values. These findings do not rule out selective advantage of viral Type I in the absence of P, but suggest that, if any, this advantage is in no way comparable to that displayed by viral Type II in the presence of P.

A rhabdovirus, sigma, is found in natural populations of Drosophila melanogaster and endemically infects a minority of individuals worldwide (for a review, see FLEURIET 1988 ). The virus is not contagious from fly to fly; it is vertically transmitted through male and female gametes. It never integrates into the fly chromosome but multiplies in the cytoplasm (BRUN and PLUS 1980 ; EMENY and LEWIS 1984 ). Infected flies can very easily be identified in this system because the virus confers a clear symptom of CO₂ sensitivity upon its host and the rules of transmission are very specific.

A few loci allow resistance to the sigma virus. Of these, the ref(2)P locus has been extensively analyzed (GAY 1978 ; DRU et al. 1993 ). Two alleles, ref(2)P^o and ref(2)P^p (O and P in this article) are respectively permissive and restrictive for viral multiplication and transmission. Wild populations of D. melanogaster are regularly polymorphic for O and P, with the latter being in the minority (FLEURIET 1988 ). Viral resistance to the P allele has developed. Two viral types are found in natural populations: Type I, which is very sensitive to the P allele, and Type II, which is more resistant (FLEURIET 1988 ). Replacement of viral Type I by Type II in the presence of P was unequivocally observed in at least two wild populations, in Southern France (FLEURIET and PERIQUET 1993 ) and in Germany (FLEURIET and SPERLICH 1992 ). All the collected data are in agreement with the working hypothesis that Type II, which is better adapted to the P allele, is replacing Type I in natural populations. The change might have started in France in the seventies (FLEURIET 1988 ) and the phenomenon is expected to spread progressively to other populations, the further from France the later.

These observations confirm that attempts to biologically control parasites by introducing alleles for resistance into host populations can only be a temporary solution. Parasites will eventually adapt to these alleles; the Drosophila sigma system shows that sensitive genotypes can be replaced by resistant ones within a few years (FLEURIET and PERIQUET 1993 ).

The question arises as to what would happen to the system if the P allele was absent. Would viral Type I be selected for or would both types be neutral? Host-parasite interactions are acknowledged as an important driving force in the evolution of populations. However, this side of the problem may be purely academic because it is questionable whether this particular situation ever occurs in the wild. In a classic "Red Queen" scenario, the presence of a parasite would increase the frequency of an allele for resistance H^R in the host population. This would in turn lead to the presence in significant proportions of a resistant P^R type in populations of the parasite. Would this P^R type be present with significant frequency if the H^R allele had never existed? Such a situation might be encountered in a small population in which the H^R allele had secondarily been lost by genetic drift but it would only be locally relevant. The disappearance of the H^R allele would not necessarily lead to the elimination of the P^R type from the population. Alleles for resistance are being artificially introduced into numerous plant species and this possible consequence of the neutrality of P^S and P^R types with respect to the H^S allele has to be known.

But whether or not selection occurs in favor of Type I, the consequences would nevertheless have an effect on the evolution of the system. Symmetrical advantages of P^S and P^R, the two types of the parasite, with respect to H^S and H^R, the two alleles of the host, create a balanced equilibrium; it would prevent the sensitive type P^S from disappearing in large populations of the host that are polymorphic for the two alleles.

In the Drosophila-sigma system, the almost total elimination of viral Type I observed in natural populations (FLEURIET and PERIQUET 1993 ) does not favor such a hypothesis. The problem cannot be analyzed in the wild because the P-O polymorphism is ubiquitous. One way of answering the question is to monitor experimental populations deprived of the P allele and differing in the initial proportions of the two viral types, then to measure the frequency of Type I and Type II viral clones in subsequent generations. If the frequency remains more or less similar to its initial value, both types may be considered as almost equivalent in the O/O genotype. If a comparable equilibrium is reached in the various populations, selective forces may be expected to act on the system. Such an experiment was started with two samples recently collected in the wild.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Procedures:
In the laboratory, flies were maintained on axenic food (DAVID 1959 ), at 20°, under natural light conditions. The CO₂ test used to measure the frequency of infected flies is described in PLUS 1954 .

Genotypes at the ref(2)P locus:
Each male was classified as O/O, O/P, or P/P at the ref(2)P locus, depending on whether, after crossing with a female of a reference strain, its adult progeny was entirely CO₂ sensitive, half CO₂ sensitive, or entirely CO₂ resistant (FLEURIET 1976 ).

Efficiency of viral transmission by males:
In the Drosophila-sigma system, two kinds of males are found. In the nonstabilized condition, males do not transmit the virus. In the stabilized condition, males transmit the virus to a proportion of their offspring. This proportion is called male valence and is the most important factor for the viral perpetuation in a population (FLEURIET 1988 ).

Isofemale lines were isolated from each sample; only CO₂-sensitive lines (i.e., infected lines) were kept. They are assumed to carry one viral clone only. For each line, the valence of 5–10 G1-stabilized males, i.e., the frequency of infected flies in their progeny, was determined by mating the males individually with O/O uninfected females from a reference strain. The frequency of CO₂-sensitive flies was then measured in their offspring. The average valence of males from the same line was obtained, as a characteristic of that line, by pooling the values observed for each of the males.

Determination of the viral type:
The method used to determine the type of a viral clone (with respect to the P allele) has been described in detail elsewhere (FLEURIET 1980 ). The valence of a male mated with an O/O female (see above) is taken as a reference. The valence of the same male with a P/P female is then measured. If the latter value is zero (or close to zero), the male carries a Type I viral clone; if it is only slightly lower than the reference value, it carries a Type II viral clone.

Transovarial transmission and number of adult progeny produced by females:
In the Drosophila-sigma system, two kinds of females are found. In the stabilized condition, females transmit the virus to all their progeny. In the nonstabilized condition, females transmit the virus to only a portion; this is a function of the rate of transovarial transmission (for a review see FLEURIET 1988 ). G1 males whose valence had been measured were in the stabilized condition (i.e., were able to transmit the virus to some offspring). Their infected daughters were nonstabilized (FLEURIET 1988 ) and were used to measure the rate of transovarial transmission as follows.

Virgin females were collected in the offspring of G1 males mated with O/O uninfected females (see above). They were distributed individually into vials with two uninfected O/O males. Every other day they were transferred to fresh vials. Before the females were discarded (after ~20 days), they were submitted to CO₂ treatment to determine which were infected. This method ensured that females had been handled in exactly the same way, whatever their infection status.

After emergence, adult offspring were submitted to CO₂ treatment, first to count them and second to determine the frequency of infected flies, i.e., the rate of transovarial transmission in infected females.

Neostabilization:
The measurement of neostabilization gives an indication of the invading abilities of a viral strain. Two kinds of infected flies may be found. In the stabilized condition, which is the more efficient for transmission of the virus, females transmit the virus to all their progeny and males to only some (valence). In the nonstabilized condition, females transmit the virus to only some offspring and males do not transmit it at all (FLEURIET 1988 ).

The infected G2 offspring of a stabilized G1 male are in the nonstabilized condition. A proportion of G3 flies from a G2 nonstabilized female will have recovered the stabilized condition. The frequency of neostabilization is the frequency of stabilized G3 flies in the infected offspring of these G2 females.

Collection of samples:
Large samples of flies were collected in September 1993 in Sainte Foy (SF) near Lyon, France, and Tübingen (TUB) in Germany. Isofemale lines were isolated from each sample: some were CO₂ resistant (uninfected) and some CO₂ sensitive (infected by the sigma virus). The latter were analyzed to determine which viral type they were carrying (I or II).

Isolation of O/O lines:
Three kinds of O/O lines were isolated from brother-sister matings:

• Uninfected: SF: three lines, nos. 69, 160, and 163. TUB: three lines, nos. 2, 47, and 50.

• Infected by a Type I viral clone: SF: two lines, nos. 14 and 183. TUB: one line, no. 61.

• Infected by a Type II viral clone: SF: one line, no. 60. TUB: two lines, nos. 9 and 44.

Experimental populations:
Each population was founded with 50% infected (⁺) O/O flies and 50% uninfected (^-) O/O flies. Three sorts of populations were set up, depending on the proportions of O/O flies respectively infected by a Type I or Type II viral clone (0.9, 0.1; 0.5, 0.5; and 0.1, 0.9).

Populations originating from the SF sample: See Table 1. Equal numbers of males and females were used (e.g., 400 flies = 200 virgin females + 200 males). Uninfected flies were a mixture of the three O/O lines (nos. 60, 160, and 163). Two lines infected by a Type I viral clone were used (no. 183 or 14). Three replicates were set up for each system (A1, A2, A3; B1, B2, B3, etc.) so that there were 18 populations in all.

Populations originating from the TUB sample: See Table 2. Uninfected flies were a mixture of the three O/O lines (nos. 2, 47, and 50). Two lines infected by a Type II viral clone were used (no. 44 or 9). Three replicates were set up for each system (H1, H2, H3, etc.) to give 18 populations in all.

The 36 populations were perpetuated with large numbers of flies (~1000). The frequency of infected flies was measured each generation. After at least 12 generations, isofemale lines were collected in each population. Infected lines were analyzed to determine whether they carried Type I or Type II clone and thus the respective frequencies of the two types in the corresponding population at the time (because it is taken that one isofemale line carries one clone).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Frequency of infected flies:
This was regularly recorded in all populations (Figure 1 Figure 2 Figure 3 Figure 4). In populations originating from the SF sample (A, B, C, D, E, and F), the frequency decreased quickly from its initial value 0.5 toward 0, except in A1, where it slowly started to increase after ~25 generations until almost all the flies were infected (from generation 60 onward, not shown in the figure). For populations originating from the TUB sample, the frequency increased from 0.5 to ~0.8 in J and M and remained steady. In H, I, K, and L, it increased sharply and then started to decrease slowly. Frequencies were very similar in the three replicates during the first 20 generations, but thereafter started to diverge; in some, the frequency continued to decrease slowly, in others more sharply (I3, K1, and L1), or, in contrast, began to increase until almost all the flies were infected (H3, K2, and L2).

View larger version (13K): In this window In a new window Download PPT slide   Figure 1. Frequency of infected flies in nine populations of SF origin: A, B, and C. x-axis, generations; y-axis, frequency of infected flies.

View larger version (9K): In this window In a new window Download PPT slide   Figure 2. Frequency of infected flies in nine populations of SF origin: D, E, and F. Definitions as in Figure 1.

View larger version (15K): In this window In a new window Download PPT slide   Figure 3. Frequency of infected flies in nine populations of TUB origin: H, I, and J. Definitions as in Figure 1.

View larger version (15K): In this window In a new window Download PPT slide   Figure 4. Frequency of infected flies in nine populations of TUB origin: K, L, and M. Definitions as in Figure 1.

Several factors could explain why the frequency of infected flies increased or decreased in a population (Table 3).

1. The most important is the frequency of infected flies (or valence) in the progeny of stabilized males. This efficiency of transmission in absence of the P allele was significantly higher for males from populations in which the frequency of infected flies increased (0.323 vs. 0.636 and 0.488 vs. 0.606).

   
   

   View this table: In this window In a new window   Table 3. Differences in efficiency of viral transmission, in absence of the P allele, between populations in which the frequency of infected flies increased or decreased

2. In some isofemale stabilized lines, males did not transmit the virus at all. This was the case for 7% of the lines isolated from populations of SF origin, in which the frequency of infected flies decreased. None were found in the other populations: males transmitted the virus to at least some of their progeny in populations of SF origin, in which the frequency of infected flies increased, and in all the populations of TUB origin.

3. Significantly more infected females did not transmit the virus in populations in which the frequency of infected flies decreased (12.5%) than in populations in which the frequency increased (0.8%). This was true for flies of either origin (SF or TUB). The result for populations of SF origin is only given as an indication because the frequency of infected flies increased in one population only (A1) in which females of one line only did not transmit. Such a result does not allow reliable comparison.

4. Another important factor that explains the fast elimination of the virus from populations D, E, and F is the almost complete absence of neostabilization (see MATERIALS AND METHODS).

Proportions of Type I and Type II clones:
The proportions of the two types of clones were measured after at least 12 generations in each population (Table 4 and Table 5). Populations in which the frequency of infected flies decreased were examined first. In some of these (e.g., D2, E1, F2, and F3), infected flies were so few (1–2%) that it was difficult to obtain stabilized lines and only a few clones could be examined. In one population, E3, the virus was eliminated so rapidly that from generation 8 onward no infected flies could be found.

In most populations, one type was rapidly eliminated or remained infrequent (Table 4 and Table 5). In populations in which both types were still present in significant proportions, a second analysis was made. With the exception of I1, in which both types were equally represented, either one type remained or one was predominant. In D2, D3, K1, and L1 the virus was eliminated too fast for a second measurement. The loss of one type of clone in these populations may be expected but it is difficult to predict which one will be eliminated. The evolution observed in I1, from generation 18 to generation 53, shows that a trend can be reversed after some time. Population J3 shows that both types may even coexist for a long time. In a few cases, opposite trends were observed in replicate populations (e.g., B1 and B2 vs. B3 or I2 vs. I3).

Table 6 shows how populations can be distributed according to the initial frequencies of the two types and their evolution. As explained previously, populations classified in the second group (I = II), D2, D3, K1, and I1 would certainly have gone over to one of the other groups had enough time elapsed. Populations H1, H2, B3, and L1 were arbitrarily put in the third group (II in the majority) though, as noted above, it is not certain that the eliminated clone will be Type I.

When Type I clones were initially in the majority, Type II clones were systematically eliminated (Table 6). When both types were equally represented at the beginning, they sometimes coexisted over a long period; in most cases, Type II clones were eliminated, but in a few populations they predominated. When Type II clones were initially in the majority, they were eliminated in only three cases.

Comparison of Type I and Type II clone characteristics:
The efficiency of viral transmission by males, in the absence of the P allele, was systematically measured for all the clones tested, because the distinction between Type I and Type II clones is based on this parameter (see MATERIALS AND METHODS). The value given in Table 3 is the average value measured for all the clones in the population. The same value was also calculated separately for Type I and Type II clones (Table 7). Type I clones were more efficiently transmitted by males than Type II clones, whatever the origin of the clones. However, the efficiency of transmission was higher in populations of TUB origin than of SF origin.

The efficiency of transmission of the virus can also be measured in nonstabilized females, which, unlike stabilized females, transmit to only some progeny. This parameter was not monitored during the experiment but had been measured in the infected lines used to establish the populations. As expected, efficiency was great (FLEURIET 1988 ), with no differences between Type I and Type II clones (Table 7).

The measurement of this parameter also made it possible to determine the average daily number of adult flies in the progeny of infected or uninfected young females (see MATERIALS AND METHODS). For infected females it was significantly lower in lines of SF origin (11.66 ± 1.10 vs. 20.00 ± 1.71; t = 6.95, P < 0.001) and in lines of TUB origin (12.85 ± 1.59 vs. 19.82 ± 2.65; t = 4.68, P < 0.001). However, in lines of SF origin, the average daily number of adult flies in the progeny of infected females was lower when they were infected by Type I clones than by Type II clones (9.62 ± 1.34 vs. 13.43 ± 1.49; t = 3.81, P < 0.001). The opposite was true in lines of TUB origin (22.0 ± 6.10 vs. 11.86 ± 1.40; t = 4.57, P < 0.001).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Previous observations in experimental and natural populations of D. melanogaster have shown that, in the presence of the restrictive P allele, Type II clones of sigma are selected for (FLEURIET 1988 ; FLEURIET and SPERLICH 1992 ; FLEURIET and PERIQUET 1993 ). The purpose of the present work was to determine whether, in populations in which P is absent, Type I clones are selected for or whether the two types are equivalent. The experimental design was to establish populations of O/O genotype with different proportions of Type I and Type II clones and to follow the evolution of their frequencies.

The easiest way to initiate these populations would have been, first, to isolate O/O flies by classical crosses with a reference strain carrying a balanced system on chromosome II (where the ref(2)P locus is located); and then to obtain infected isofemale lines by injection with Type I or Type II clones. This method was discarded for two reasons. To stay as close as possible to the original "natural" genetic and cytoplasmic background, it was not possible to cross with a laboratory strain. Previous studies showed that the virus is sometimes better controlled in the original genotype of the population from which it came than in another one (FLEURIET 1991 , FLEURIET 1996 ). It is also known that different viral genotypes are selected through injection or hereditary transmission (FLEURIET 1988 ). The method used, brother-sister mating, is more time-consuming and all the more difficult because, first, genotypes at the ref(2)P locus in females cannot be determined, and in males can only be established by analysis of their progeny after crossing with females of a reference strain (see MATERIALS AND METHODS). Second, individual crosses with reference strains are also required to determine the viral type carried by an infected line. Third, it is preferable to use viral clones (I and II) and O alleles originating from the same natural population; and viral Type I is now rather rare in European populations (FLEURIET 1988 ). This is why it took 8 months of intensive work to establish populations after collection of flies in the wild. Yet, given the problems stated above, it was the only possible way to get the three types of lines necessary for the experiment.

The first parameter followed in the experimental populations was the frequency of infected flies, from its initial value of 0.5. In populations of TUB origin, the frequency usually remained higher than in SF populations, in which the virus was eliminated very quickly. Comparable results were observed in the two large populations founded with the total samples of flies collected at SF and TUB and perpetuated ever since. The virus was eliminated from the SF population within 20 generations and was still present after 90 generations in ~20% of the flies in the TUB population.

There was a clear correlation between the increase or decrease in the frequency of infected flies in populations and the value of certain parameters, in particular the efficiency of viral transmission. The efficiency of transmission by males, which is a cornerstone for the maintenance of the virus in populations (L'HERITIER 1970 ; FLEURIET 1988 ), was lower in populations in which the frequency of infected flies decreased and also lower in populations of SF origin than in those of TUB origin. There were even some lines in SF populations in which males did not transmit at all. Furthermore, in some populations (D, E, and F) there was almost no neostabilization, which means that once the virus is in a male, it is lost for the following generations (FLEURIET 1988 ). This had never previously been observed in flies of recent wild origin in which neostabilization is usually very high (almost 100%; FLEURIET 1988 ). These data explain why the virus was eliminated so fast in some populations, mainly of SF origin.

Results from other parameters support this finding. The proportion of infected females, which, as individuals, did not transmit the virus, was lower in populations in which the frequency of infected flies increased. Infected females produced fewer adults than uninfected ones and the difference was greater in females of SF origin than in those of TUB origin.

These findings can therefore explain why the frequency of infected flies tended to increase or decrease in a population. It has been shown (FLEURIET 1994 , FLEURIET 1996 ) that different systems of mutually coadapted sets of parameters can be found in populations and that they offer an explanation of the respective evolution observed in the populations. No clear evidence has ever been given of the factor(s) responsible for the "choice" of the particular set of parameters in a population that decides its later evolution.

There seems to be no correlation between the viral type fixed in a population and the long-term increase or decrease in the frequency of infected flies. Opposite results can be found, depending on the populations. However, in populations in which Type I was initially predominant, infection frequency tended to increase, or to decrease more slowly (e.g., C, J, and M).

There was no systematic evolution in the proportions of Type I and Type II clones in the populations. In contrast, all previous observations have shown that, in the presence of the P allele, there is an increase in the frequency of Type II clones, both in wild and in experimental populations (FLEURIET 1988 ; FLEURIET and SPERLICH 1992 ; FLEURIET and PERIQUET 1993 ). In the absence of P, one type or the other predominated, Type I in 22 cases, Type II in 9 cases. In four populations, both types still coexisted in significant proportions when the virus was eliminated. There was some evidence of a bias in favor of Type I, all the more so when its initial frequency in the population was high. It was never eliminated from populations in which its initial frequency was 0.9; it was less often eliminated than Type II in populations in which both types were equally represented at the start. It is only in populations in which Type II was initially predominant that Type I was eliminated more often.

The most probable explanation for the advantage of Type I is its better transmission by males in the absence of the P allele. It would of course have been preferable to do the experiment with Type I and Type II clones equally transmitted by males. It is almost impossible now to find clones with such characteristics in Europe (which, in 1993, seemed to be the only place where both types coexisted). The efficiency of transmission of Type II clones by males in the absence of P decreased at the same time as the frequency of these clones increased in populations (FLEURIET and PERIQUET 1993 ). A possible interpretation is that the decrease prevented them from being too "efficient." The consequence is that, when clones are now collected in the same wild population, Type II will be less efficiently transmitted than Type I. One solution might have been to use Type I clones from African populations, in which their efficiency of transmission is lower than in Europe (and in which Type II clones seem to be absent or at least extremely rare; FLEURIET 1988 ). The result would have been a mixture of clones of different origins, adapted to different backgrounds, and the observations would not necessarily have been easier to interpret. It was thus deliberately decided to conduct the experiment with clones, known to differ in fitness by one parameter, but not the parameter tested in the experiment. The objection might be raised that "the" hypothetical selective advantage that was looked for in these experiments lay in this better transmission of Type I clones in the presence of the O allele alone. However, this better transmission is not an intrinsic characteristic of viral Type I but is a consequence of the decreased transmission of Type II clones observed in populations since 1982 (FLEURIET 1990 ; FLEURIET et al. 1990 ; FLEURIET and PERIQUET 1993 ).

The problem is whether these data can be used to determine if the two types are otherwise equivalent with respect to the O allele or whether Type I, because of its function, is selected for in presence of O. Table 6 shows that Type II is able to win in a population when it is not the rarest, despite the fact that its lower transmission by males puts it at a great disadvantage. This suggests that if the simple presence of the O allele made the fitness of Type II even lower than that of Type I, then it would never be able to predominate. A possible conclusion is that, in the presence of the O allele alone, the two types are almost equivalent. If they are not, and if, nevertheless, Type I is selected for in the presence of O, its advantage cannot be very great and in no way comparable to that enjoyed by Type II in the presence of P. Comparable results were obtained with the two different samples of virus used in these experiments. The problem cannot be analyzed in the wild, because the polymorphism for O and P at the ref(2)P locus seems to be ubiquitous, at varying frequencies, in natural populations of the fly. Knowledge of the molecular differences between Type I and Type II, and of the effect of the two alleles upon the virus, might provide a fuller understanding of the problem. If the P allele were reintroduced into some of the O/O populations in which both types are still present, the frequency of Type II clones would be expected to rapidly increase. Preliminary results show that such is the case.

	ACKNOWLEDGMENTS

The author thanks R. Allemand and D. Sperlich for collecting flies in Sainte Foy and Tübingen; D. Rodriguez and A. Guermite for efficient technical assistance; and two anonymous reviewers. This work was supported by the Centre National de la Recherche Scientifique (UMR 6547) and the Université Blaise Pascal.

Manuscript received December 15, 1998; Accepted for publication August 30, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BRUN, G., and N. PLUS, 1980 The viruses of Drosophila, pp. 625–702 in The Genetics and Biology of Drosophila, edited by M. ASHBURNER and T. R. F. WRIGHT. Academic Press, New York.

DAVID, J., 1959  Etude quantitative du développement de la Drosophile élevée en milieu axénique. Bull. Soc. Biol. Fr. Belg. 93:472-505.

DRU, P., F. BRAS, S. DEZELEE, P. GAY, and A. M. PETITJEAN et al., 1993  Unusual variability of the D. melanogaster ref(2)P protein which controls the multiplication of sigma Rhabdovirus. Genetics 133:943-954[Abstract].

EMENY, J. M., and M. J. LEWIS, 1984 Sigma virus of Drosophila as a vector model, pp. 93–112 in Vectors in Virus Biology, edited by M. A. MAYO and K. A. HARRAPS. Academic Press, New York.

FLEURIET, A., 1976  Presence of the hereditary Rhabdovirus sigma and polymorphism for a gene for resistance to this virus in natural populations of D. melanogaster.. Evolution 30:735-739.

FLEURIET, A., 1980  Polymorphism of the hereditary Rhabdovirus sigma in natural populations of D. melanogaster.. Genetics 95:459-465[Abstract/Free Full Text].

FLEURIET, A., 1988 Maintenance of a hereditary virus, the sigma virus, in populations of its host, D. melanogaster, pp. 1–30 in Evolutionary Biology, edited by M. K. HECHT and B. WALLACE. Plenum, New York.

FLEURIET, A., 1990  Evolution of natural populations in the D. melanogaster-sigma virus system. II. Northern and central France. Genetica 81:33-41[Medline].

FLEURIET, A., 1991  Polymorphism of the hereditary Rhabdovirus sigma in wild populations of its host, D. melanogaster.. Genet. Sel. Evol. 23:25-37.

FLEURIET, A., 1994  Female characteristics in the D. melanogaster-sigma virus system in natural populations from Languedoc. Arch. Virol. 135:29-42[Medline].

FLEURIET, A., 1996  Polymorphism of the D. melanogaster-sigma virus system. J. Evol. Biol. 9:471-484.

FLEURIET, A. and G. PERIQUET, 1993  Evolution of the D. melanogaster-sigma virus system in natural populations from Languedoc. Arch. Virol. 129:131-143[Medline].

FLEURIET, A. and D. SPERLICH, 1992  Evolution of the D. melanogaster sigma virus system in a natural population from Tübingen. Theor. Appl. Genet. 85:186-189.

FLEURIET, A., G. PERIQUET, and D. ANXOLABEHERE, 1990  Evolution of natural populations in the D. melanogaster-sigma virus system. I. Languedoc. Genetica 81:21-31.

GAY, P., 1978  Les gènes de la Drosophile qui interviennent dans la multiplication du virus sigma. Mol. Gen. Genet. 159:269-283[Medline].

L'HERITIER, PH., 1970 Drosophila viruses and their role as evolutionary factors, pp. 185–189 in Evolutionary Biology, edited by M. K. HECHT and B. WALLACE. Plenum, New York.

PLUS, N., 1954  Etude de la multiplication du virus de la sensibilité au gaz carbonique chez la Drosophile. Bull. Soc. Biol. Fr. Belg. 88:1-46.