Effect of Breeding Structure on Population Genetic Parameters in Drosophila

Corresponding author: Michel Veuille, 7 quai saint Bernard, 75252 Paris Cedex 05, France., mveuille{at}snv.jussieu.fr (E-mail)

Communicating editor: W. STEPHAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The breeding structure of populations has been neglected in studies of Drosophila, even though Wright and Dobzhansky's pioneering work on the genetics of natural populations was an attempt to tackle what they regarded as an essential factor in evolution. We compared the breeding structure of sympatric populations of D. melanogaster and D. simulans, two sibling species that are widely used in evolutionary studies. We recorded changes in population density and microsatellite variation patterns for 3 years in a temperate environment of southwestern France. Results were distinctively different in the two species. Maximum population levels in summer and in autumn were similar and fluctuated greatly over years, each species being in turn the most abundant. However, genetic data showed that D. melanogaster made up a continuous breeding population in time and space of practically infinite effective size. D. simulans was fragmented into isolates with a local effective size of between 50 and 350 individuals. A consequence of this was that, while a local sample provided a reliable estimate of regional genetic variability in D. melanogaster, a sample from the same area provided an underestimate of this parameter in D. simulans. In practical terms, this means that variations in breeding structure should be accounted for in sampling schemes and in designing evolutionary genetic models. More generally, this suggests the existence of differential reactions to local environments that might contribute to several genomic differences observed between these species.

INTEREST in the genetic structuring of natural populations arose in the early 1930s, after the population genetic syntheses published by FISHER 1930 and WRIGHT 1931 revealed their diverging opinions concerning the effect of population breeding structure on the dynamics of evolution through natural selection. The need to estimate population parameters led Dobzhansky to investigate the genetics of natural populations in Drosophila pseudoobscura (LEWONTIN et al. 1981 ). Using the allelism of lethals, DOBZHANSKY and WRIGHT 1941 and WRIGHT et al. 1942 were able to provide joint estimates for N_e (the effective population size) and m (the migration rate between populations). These studies led to growing interest in population structuring and encouraged Wright to develop F-statistics (WRIGHT 1951 ). Lethal studies were also conducted in D. melanogaster (IVES 1945 , IVES 1950 , IVES 1959 ). The availability of balancers in this species made it possible to simultaneously investigate the distribution of deleterious effects (see e.g., MUKAI and YAMAGUCHI 1974 ). However, lethal studies had limitations due to extensive line heterogeneity in mutation rates resulting from transposable elements (MUKAI et al. 1985 ; IVES and BAND 1986 ; KEIGHTLEY and EYRE-WALKER 1999 ). The introduction of allozyme techniques (HUBBY and LEWONTIN 1966 ) raised new interest in the breeding structure of Drosophila and led to observations, still to be confirmed, showing substantial local structuring (DANIELI and COSTA 1977 ; TAYLOR and POWELL 1977 ). Microsatellite markers now provide a powerful way of detecting microscale structuring in Drosophila populations (AGIS and SCHLOTTERER 2001 ). The microgeographic distribution of Drosophila populations is still poorly understood despite ecological studies in several species (e.g., BEGON 1977 ). In particular, more information in this domain is needed from D. melanogaster and D. simulans. These species have become major evolutionary genetics models in the last decade and show confusing differences in their patterns of molecular variation, which suggest demographic explanations (e.g., ANDOLFATTO 2001 ; BEGUN 2001 ; CHARLESWORTH 2001 ). Moreover, captive Drosophila populations have recently been used as experimental models of endangered species (FRANKHAM 1995 ), thus providing additional reasons to study their breeding structure in nature.

We report a microscale study of population structuring in D. melanogaster and D. simulans in the Bordeaux vineyard area of southwestern France. The "fruit fly" or "vinegar fly" (GREEN 2002 ; the "vintage fly" or "musset" of local vine growers) lives on fermenting fruit, forming dense populations on the rotting leftovers of vine agriculture. These two species are thought to originate in Africa and to have extended their range to Europe with the rise of agriculture in neolithic times (LACHAISE et al. 1988 ; BENASSI and VEUILLE 1995 ). Southwestern France was planted with vines in the Roman era and has remained an active wine-producing region ever since. It was already a wine-exporting province in the middle ages and we can suppose that the region has sustained a Drosophila population for centuries, a period spanning from 5000 to 10,000 Drosophila generations. American populations of these species are thought to have colonized the United States as recently as the 1870s (STURTEVANT 1920 ). The Bordeaux area is evenly and densely planted with vine, thus providing a continuous landscape that is ideal for studying microscale migration. The ecophysiology of fruit flies is poorly understood. In temperate countries natural populations are found only from late August to early December with fluctuations depending on the year. They are rarely found outside of this period. Low density in winter could be critical in determining their effective population size. In this study, we used microsatellite genetic variation to determine changes in effective population size over time and space. Given the paucity of reference literature on the subject, we designed an observational scheme that balanced sampling effort over time and over space, each factor being considered at two nested scales (from 3 months to 3 years and from 4 to 30 km, respectively) to be able to adjust the focus in further studies.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Study area:
The sampling design is summarized in Table 1. It involves four collecting sites that formed two 30-km-distant regions, each being divided in turn into two 4-km-distant sampling sites. The first region is in the Graves vineyard, immediately south of Bordeaux. The two sampling sites were at the "Grande Ferrade" château (hereafter "Bordeaux 1"), an agricultural research station owned by Institut National de la Recherche Agronomique, and at Couhins village (hereafter "Bordeaux 2"). The second region was in the Sauternes vineyard in Preignac, a village that lies on the left bank of the Garonne river, between Cadillac and Langon. Our sampling sites were "Preignac 1" and "Preignac 2." All sampling took place in vineyards and was repeated over 3 years according to the following scheme:

• Bordeaux 1: a test sample was made in late autumn 1996 (19 December). Several samples were collected in summer 1997 over an extended period (see below) and in early autumn 1998 (28 September).

  
  

  View this table: In this window In a new window   Table 1. Sampling scheme

• Bordeaux 2: samples were collected in summer 1997 (22 August).

• Preignac 1: samples were collected in summer 1997 (23 August) and in autumn 1998 (10 September–15 October).

• Preignac 2: samples were collected in summer 1997 (23–27 August) for the genetic analysis and also in summer 1998 for the demographic study (see below).

Long-term study:
In 1997 and 1998, Bordeaux 1, Preignac 1, and Preignac 2 samples were collected twice a week from the first week of August to the first week of November for an ecological survey that will be published elsewhere. In this article, pooled data from these three locations were used for estimating sex-ratio values and species abundance.

Short-term study in Bordeaux 1:
In Bordeaux 1 samples were taken at several times for the same population in 1997. They were collected between 9 and 13 August (about the time of year when fruit flies are first observed), on 25 August (about the time of peak abundance), and on 8 October (about the time of declining populations).

Trapping device:
A technical difficulty in the ecological genetics of Drosophila is that, to our knowledge, previous studies trapped flies using attractive baits made of fermenting fruit. Although very efficient, this technique can introduce sampling biases (e.g., Wahlund effects) by attracting flies from a wide area. We therefore used a nonattractive device adapted from traps previously used for collecting aphids (LABONNE et al. 1983 ). Each trap consisted of a wooden frame inside which a 25 x 25-cm screen was made using parallel nylon threads spaced at 5-mm intervals. The threads were covered with glue. These traps were hung in the vineyard for 3–4 days. We consider that this method was nonattractive and collected a constant proportion of the flies that went through the screens. There was no indication of saturation or of variation in efficiency over time, although we did not monitor the screens for this. The trapped flies were recovered by covering the traps with turpentine for 30–120 min. The flies were then stored in 70% ethanol/30% TE until DNA extraction. This protocol provided suitable material for PCR amplification. Variations in the output of amplification on a sample-by-sample basis, however, point to unidentified drawbacks of the collecting method, possibly due to decay of trapped material between field visits. There were 5–15 trapping screens at 1.5-m intervals/sampling site. For each of them, an individual sample of 20 males was used for analysis. Screens from the same area in the same vineyard were pooled when necessary; however, this produced no artifactual structuring (see RESULTS). The number of chromosomes observed is indicated as 2n in Table 1.

Recording genetical data:
DNA extraction, PCR amplification, and examination of polymorphism at 10 microsatellite loci from chromosome II [bib, cad, dl, Elf-1, mam, odd, slobo, Su(H), Su(z)2, and twist] were carried out as described by MICHALAKIS and VEUILLE 1996 . These loci are taken from long coding trinucleotide repeats that vary widely in the number of codons within species, but are relatively constant in the average length across species (MICHALAKIS and VEUILLE 1996 ; COBB et al. 2000 ; VEUILLE et al. 2004 ). These repeated regions are frequent in Drosophila regulatory proteins associated with development and neurogenesis, and normalizing selection seems negligible in terms of deviation from a neutral distribution at the population scale (MICHALAKIS and VEUILLE 1996 ). When used as genetic markers in different species, these loci do not appear to be liable to ascertainment bias, contrary to noncoding microsatellites (MOUSSET and DEROME 2004 ; VEUILLE et al. 2004 ). The 10 loci used are evenly spaced over chromosome II, which makes up 40% of the Drosophila genome.

We used only males, since the females of these two species are morphologically similar. Standard statistical analyses were carried out using Genepop version 3.3 of March 2001 (RAYMOND and ROUSSET 1995 ). This program carries out a Hardy-Weinberg exact test using a Markov chain method from GUO and THOMPSON 1992 . It also tests deviation at several loci simultaneously using ROUSSET and RAYMOND's (1995) method. We could use this option since the markers are evenly spaced on the Drosophila genetic map and can be assumed to be independent, except for those linked to the breakpoints of the In(2L)t inversion in D. melanogaster (MICHALAKIS and VEUILLE 1996 ). Population differentiation was tested using an exact test (RAYMOND and ROUSSET 1995 ). Wright's F_ST was calculated after WEIR and COCKERHAM 1984 . Population heterozygosity was estimated as

where p_i is the frequency of the ith allele in a sample of 2n chromosomes. The estimate of effective population size and of its confidence interval (C.I.) was calculated using WAPLES's (1989) "F" genetic correlation coefficient between samples taken from the same population at different generations. It is estimated as

where x_i and x_j are allele frequencies in successive population samples, the sum being calculated over k alleles. According to a classical relation, genetic variation decreases as H_t = H₀ exp(-t/2N_e) between generations 0 and t (WRIGHT 1931 ; MALECOT 1946 , MALECOT 1948 ). From this, WAPLES 1989 showed that F depends on the effective population size and on sample sizes (2n₁ and 2n₂) according to E(F) (2n₁ + 2n₂)/(8n₁n₂) + t/(2N_e), from which the effective population size can be obtained. Confidence intervals ( = 0.05, 1 - = 0.95) were computed according to WAPLES 1989 as n/(²_()/2n), n/(²_(1-)/2n). Significance levels in multiple tests of genetic differentiation between samples were determined according to the sequential Bonferroni method, using the total number of tests as a reference. It was therefore slightly conservative in pairwise comparisons, since the tests were not independent.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Population density levels:
Table 2 shows the total number of male flies collected in Bordeaux 1, Preignac 1, and Preignac 2 over 3 months (August–October) in 1997 and 1998. There were large fluctuations, mostly due to differences in maximal values over years and over locations (E. GRAVOT, unpublished results). The dominant species was D. melanogaster in all Preignac samples. In Bordeaux 1, the dominant species changed from 82.0% D. melanogaster in 1997 to 78.9% D. simulans in 1998. We do not know whether the distribution was the same in females. In pooled data for the two species, the sex ratio fluctuated over time, a balanced sex ratio being found only once in six observations. In these three populations, trapped males were significantly less abundant than trapped females in August and September and reached a 50% proportion in October (data not shown). The cause of these variations is unknown. They may thus result from differences in the primary sex ratio, from differential survival, from differential migration, or from differential activity, in which case they would depend on the trapping device.

Genetic variation levels:
Heterozygosity levels are shown in Table 3 (D. melanogaster) and Table 4 (D. simulans). Due to technical difficulties and to a small sample size, genetic variation for Bordeaux 1-96 in D. melanogaster was calculated from only 5 loci: bib, cad, slobo, Su(z)2, and twist. Data for all other samples were calculated using 10 loci and were very similar. All loci were polymorphic in D. melanogaster. Only 8 loci were polymorphic in D. simulans. However, the 2 monomorphic loci in the latter species (odd and twi) also showed low variation levels in D. melanogaster. Overall, D. melanogaster was less variable than D. simulans. The average heterozygosity was 0.375 in the first species (range 0.338–0.397 over 10 loci) and 0.410 in the second species (range 0.363–0.482 over 10 loci). Three D. simulans samples showed especially low values (Bordeaux 1-98, Preignac 1-97, and Preignac 2-97). Two of these samples correspond to sites that were sampled for two successive years. The Bordeaux 1-98 sample showed a marked decrease in heterozygosity relative to the preceding sample (Bordeaux 1-97) since the average value between samples dropped from 0.454 to 0.363. Preignac 1-97 showed a much lower value than did the following sample, Preignac 1-98, with an increase from 0.374 to 0.433. The Preignac 2 sample was studied only once. Interestingly, two of these three low-variation samples (Preignac 1 and Preignac 2) were taken the same year (1997) from neighboring sites. An inspection of genetic diversity at individual loci reveals similar tendencies. In the two Preignac samples, mam, slobo, and Su(H) showed a marked decrease in variability, leading to their lowest values for the whole survey. The values at the other loci were also very similar in the two samples. The third low-variation sample (Bordeaux 1-98) showed a decrease in heterozygosity for a different set of loci: bib, dl, slobo, and Su(H). There is thus no indication that selection at one locus is responsible for the dramatic decrease in frequency in three samples, these observations being rather compatible with drift. It also appears that the three outlying samples in D. simulans reduce to two variation reduction events, one of these events extending over the 4 km between the two Preignac sampling sites. The general picture is that for most of the time D. simulans showed high levels of heterozygosity (range 0.410–0.482) but in three instances dropped to values (0.363–0.378) close to those observed in D. melanogaster (0.338–0.397).

The proportion of heterozygotes was not significantly different from values predicted from heterozygosity (i.e., parametric gene diversity) at the population scale, except in two instances, one in each species. This indicates that in general there was no microgeographic structuring within sampling sites. This means that the nonattractive traps used for collecting the flies produced no Wahlund effect or that they recruited flies from a panmictic population that was large enough for randomizing genetic correlation across kin groups. This also indicates that the flies collected at the beginning of the annual demographic expansion showed no consanguinity, contrary to some previous studies (see DISCUSSION).

Genetic differentiation between samples:
Genetic differentiation between samples involved 36 comparisons. Results are summarized in Table 5 (D. melanogaster) and Table 6 (D. simulans). None of them was significant in D. melanogaster. Sixteen of them were significant in D. simulans using the sequential Bonferroni procedure. This is unlikely to result from a higher power of the tests due to the higher heterozygosity in D. simulans. Roughly one-half of F_ST's in D. melanogaster were "negative" (20 vs. 16) as expected from sampling fluctuations around a null expectation, compared to very few of them (5 vs. 31) in D. simulans. These tests are not independent from each other; however, there is no indication that a difference of power of the test in one species is responsible for the contrast found between them. On the contrary, and interestingly, almost all significant tests in D. simulans implicate the three samples in which heterozygosity dropped to low values.

No significant differences were observed in either species between the three samples from the time series for Bordeaux 1 in 1997. This indicates that no change in allele frequency occurred during the annual demographic expansion. These samples were therefore pooled. Bordeaux 1-97 then remained significantly different from Bordeaux 1-98 and from Preignac 1-97 in D. simulans (sequential Bonferroni method over 21 comparisons). In Table 6, structuring can be examined between samples from different populations of this species collected at the same time or between samples from the same population collected at different times.

Spatial structuring involved six independent comparisons between the four populations in 1997 (Bordeaux 1, Bordeaux 2, Preignac 1, and Preignac 2) and a unique comparison between two populations in 1998 (Bordeaux 1 and Preignac 1), making a total of seven comparisons. Three of them were significant, all involving 30-km-distant sites in 1997: Bordeaux 1-Preignac 1, Bordeaux 2-Preignac 1, and Bordeaux 2-Preignac 2.

Time structuring involved three pairwise comparisons: the two pairwise comparisons between the three successive Bordeaux 1 samples (1996–1997 and 1997–1998) and those between the two successive Preignac 1 samples (1997 and 1998). Two of them were significant (Table 6): Bordeaux 1 and Preignac 1, when compared between 1997 and 1998.

These results are compatible with the hypothesis that structuring in D. simulans results from the low variation observed in Bordeaux in 1998 and in Preignac in 1997. It would thus be a temporary phenomenon.

Since neither the different samples from the Bordeaux area nor those from the Preignac area differed in each species in 1997, samples were pooled within each area to assess the level of genetic differentiation from a larger data set. Genetic differentiation between Bordeaux and Preignac then remained low in D. melanogaster (Weir and Cockerham's F_ST = 0.0005, P value = 0.156) and was somewhat higher in D. simulans (F_ST = 0.0302, P value < 10^-5).

Genetic drift in D. simulans populations:
Overall these data suggest that D. simulans samples were collected in a neighborhood of a small effective size. Genetic differentiation within species can result from either genetic drift alone or a balance between migration and genetic drift. Our sampling design was too simple for us to test detailed models of breeding structure. One way to interpret our data is to consider that the changes are temporal and mainly involve genetic drift. Estimates for N_e/t for a year using WAPLES's (1989) model under this assumption are: N_e/t = 352.69 for Bordeaux 1996–1997 (C.I._0.05 = 226.51–633.81); N_e/t = 48.00 for Bordeaux 1 1997–1998 (C.I._0.05 = 39.61–57.77); and N_e/t = 99.53 for Preignac 1997–1998 (C.I._0.05 = 76.77–130.02). For Preignac, we assume that the higher heterozygosity in the second year (1998) results from a return to a normal value after a population drop in 1997. There are probably many generations per year. However, in summer and autumn, there are huge fruit-fly populations: the sampling effect of successive generations would not be apparent in a survey of 10 loci on 40 chromosomes. The very low population level in winter suggests that our estimate is close to that of an overwintering generation.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Few studies on the population structure of species from the melanogaster subgroup have been carried out. Our observations were conducted at a relatively high latitude (45° N 35'–45° N 45') and a mild climate in a continuous agrosystem that seems to sustain dense populations of these species. We observed facts that apparently pertain more to demography than to spatial differentiation. While spatial structuring is easy to account for in population genetic studies, demographic perturbations are relatively unpredictible and may differ from place to place, thus confounding genetic analysis. Below, we discuss these facts, discuss their biological meaning, and then consider their consequences for genetical research.

Microscale contrast between D. melanogaster and D. simulans:
We found no substantial deviation from Hardy-Weinberg proportions in either species over the period during which these flies are abundant enough to be observed (August–December). This contradicts previous results by DANIELI and COSTA 1977 . These authors recorded EST-6 allozyme variation in six different places from Venetia (Italy) over 2 years, 1971 and 1974. They consistently found an excess of homozygotes at the beginning of the sampling season. Inbreeding decreased in the following generations. They interpreted this as meaning that in late August D. melanogaster is composed of micropopulations that have been isolated in winter and have undergone inbreeding. According to their Figure 1, DANIELI and COSTA 1977 found an inbreeding coefficient F = 0.6 in early September, and this value decreased to zero in late October. This is a very high value for an inbreeding coefficient. An average F = 0.59 is expected only after four generations of brother-sister mating. No D. melanogaster populations are observed for 7 months, from January to August. D. melanogaster's developmental time in winter must be very long, since it depends on environmental temperature. Development lasts 2 weeks at 25°, but drops to 4 weeks at 17°, a temperature below which males are unable to reproduce. Four generations thus appear a probable maximum for winter populations. The high inbreeding estimate reported by DANIELI and COSTA 1977 would mean that, for a considerable period of the year, fruit flies are virtually reduced to populations of a couple of breeding individuals. However, our data overlapped the same sampling season, and no evidence of consanguinity was found. The cause of the differences between the two studies is unknown. They differ in a number of aspects: the sampling device, the sampling area, the kind of markers used, and the number of loci. Only comparative studies using the same methods in the two areas could solve this point.

In D. melanogaster from southwestern France, no population structuring was found over space or over time. This species thus appears to form a relatively evenly distributed panmictic population at this geographical scale. For similar latitudes, high population levels (>10,000) were obtained in Japan (MUKAI and YAMAGUCHI 1974 ) and in the population of Raleigh, North Carolina (MUKAI et al. 1971 ; MUKAI and VOELKER 1997 ) using the allelism of lethals, with no indication of population substructuring. An allozyme study by SMITH et al. 1978 found no evidence of microscale structuring in North American D. melanogaster populations.

The picture is different in D. simulans. Statistically significant differences in allele frequencies were consistently found between locations and between successive years. It is unlikely that significance was due to local heterogeneity within vineyards, since this would also have induced differentiation between sampling sites 4 km apart. Spatial heterogeneity was associated with a reduction in heterozygosity between successive years. In the only case in which this could be observed, the drop in variation was simultaneous 4 km apart, thus giving the spatial extent of a neigborhood. A 4-km patch is substantial, and the effective size relatively small (down to 50 individuals) for an organism occurring in huge populations in summer.

Evidence of demographic instability in D. simulans:
Overall, the genetic variation in D. simulans is much higher than that in D. melanogaster, showing that long-term population size is substantial. The significant fixation indices between samples indicate fluctuations in population size in this species. For this reason, we do not interpret the significant F_ST values in D. simulans as reflecting stable geographic differentiation at a 30-km scale, but as temporary, probably annual, local changes in effective population size. No strong contrast was apparent in population abundance between D. melanogaster and D. simulans. Depending on the year, either of the two species was the most frequent. This suggests that the difference in effective population size was not due to demographical differences in summer, but probably to differences in winter. It is reasonable to assume that winter populations of either species are fragmented into small overwintering isolates that expand locally in summer, coalesce, and finally restore a dense and continuous population. The most likely population regime would involve two steps, with random genetic drift occurring in winter and gene flow in summer. We can thus imagine the D. simulans population as a field of neighborhoods. Fruit flies from a given area would originate from a limited number of surviving individuals, resulting in a temporary level of inbreeding. However, the resulting structuring would not last long, since populations exchange individuals. Our lowest estimates of effective population size in D. simulans in Bordeaux is N_e/t = 48 for a 1-year cycle. Since the effective population size of a population over some period of time is the geometric mean of the elementary population sizes of each generation, our estimates are likely to be close to the size of the winter breeding population. For the rest of the year, the fruit flies from a given area would retain a gene identity of f = 1/(2N_e + 1) 1/100, despite their high population level. It would be tempting to estimate a migration rate between sampling sites. Unfortunately, its estimation would depend on a stable population model, which is confounded by our results. The observation that the decreased heterozygosity in Preignac in 1997 was completely restored in 1998 (and the F_ST nonsignificant) suggests that extensive gene flow occurs.

For D. melanogaster, no genetic differentiation was detected in this study, but we cannot exclude that fluctuations also exist. We can note only that no bottleneck event was detected in D. melanogaster while two strong events were found in D. simulans in the same sampling sites. Thus there is a difference, but maybe only one of intensity. Even though the difference between the two species may lie in trivial quantitative changes within a single ecological framework, there are conspicuous differences in the genetic distribution patterns.

We do not assume that the difference observed in southwestern France will be found throughout the overlap zone of these two species, but only that they are able to react differently to their environments in such a location. Microhabitat studies show that D. melanogaster is more abundant than D. simulans in villages and in houses. This has been observed in a number of areas, including tropical Africa (DAVID 1979 ) and Tunisia (ROUAULT and DAVID 1982 ). This ecological difference is likely to restrict D. simulans to unsheltered microhabitats. Moreover, ecophysiological studies carried out by BOULETREAU-MERLE 1992 on reproduction in Drosophila from southeastern France showed that D. simulans is less adapted to temperate areas than is D. melanogaster. Temperate D. melanogaster females differ from tropical populations in the control of fecundity, thus allowing them to cope with annual environmental changes through individual adaptation. This is not observed in D. simulans, where tropical and temperate females behave similarly in experimental designs simulating the two environments (BOULETREAU-MERLE 1992 ). These ecological differences would contribute to limit population level in D. simulans. A study of 15 D. melanogaster populations over a 700-km north-south gradient in southeastern France by GIRARD and PALABOST 1976 showed no allozyme frequency differences. However, groups of populations differed significantly in ovariole number, suggesting ecological adaptations. Unfortunately, no corresponding study is available for D. simulans.

Practical consequences for population genetic studies:
Of the two species used in this study, D. melanogaster appears to form a large and stable population, whereas D. simulans appears to be fragmented into small drifting demes. These population profiles are reminiscent of the conceptions of population put forward by Fisher and Wright, respectively. The boundaries of our observation design set a limit to the generality of this conclusion, which should be confirmed by independent evidence. It is, however, important to consider its potential implications, since these two species are widely used as models in evolutionary biology. The F_ST can be interpreted in terms of a decomposition of genetic diversity. Of the total variation that is present within a 30-km area, a local population of D. simulans represents only 1 - F_ST 97.0% (using the average F_ST between sampled sites in this study). The balance is the amount of variation that would be locally and temporarily lost in the overwintering sampling process.

For instance, VEUILLE et al. 2004 showed that, for the 10 loci used in this study, heterozygosity in a D. simulans population from Zimbabwe (H = 0.505) was larger than that in populations from France (H = 0.437). This is in agreement with earlier results from HAMBLIN and VEUILLE 1999 showing that non-African populations from this species are less variable than African ones and probably originated through a population bottleneck. If we assume, however, that this species is structured as a metapopulation in France and not in Zimbabwe, then using a single local sample introduces a bias. The value of H is underestimated for Europe and should be corrected as H/(1 - F_ST) 0.450. This should also occur if the two sexes migrate differently between demes in the metapopulation. For instance, if females migrate less than males, then genetic structuring for mitochondria and for X chromosomes will be increased. This may inflate contrasts when comparing populations from different continents for X-linked genes (HAMBLIN and VEUILLE 1999 ).

A fragmented overwintering population may also introduce changes in the genomic makeup of a species. For instance, homozygotes for deleterious mutations are more likely to appear during strong bottlenecks. This may purge the genome of many deleterious variants, especially lethal genes, and decrease the effect of background selection (CHARLESWORTH et al. 1993 ). A theoretical study by WRIGHT 1937 showed that the steady-state frequency of recessive lethals is smaller in small populations than in large ones (see WRIGHT 1969 , pp. 363–366). This model considered strictly recessive lethals. However, many lethal and mildly detrimental alleles in Drosophila have a dominant effect and are thus eliminated as heterozygotes in large populations (GREENBERG and CROW 1960 ). Models taking this into account show that the size of the population still contributes to the load through random fluctuations in allele frequencies (KIMURA et al. 1963 ; NEI 1968 ). However, the mating system could also contribute to the load adjustment by eliminating lethals as homozygotes. Temporal inbreeding in a natural population could thus be a key factor in determining variation patterns.

A model put forward by KIMURA and CROW (1970, equation 9.4.12) takes this factor into account. Lethal genes are assumed to follow a gamma distribution (V,S), where V = 4N_ev and S = 4N_e(h + f), given that v is the lethal mutation rate, f is the inbreeding coefficient, and h is the dominance disadvantage of lethal heterozygotes. It is assumed that the selection coefficient of lethal homozygotes is s = 1. The expected frequency of lethals E(q) = v/(h + f) is independent of the effective population size, but depends on the inbreeding coefficient.

In our results, a striking difference between species lies in f, the inbreeding coefficient of populations. Its value is negligible in this study in D. melanogaster, but significant in D. simulans. A consequence of this would be a larger load of lethal mutations in D. melanogaster than in D. simulans. We can use the above model to evaluate the magnitude of this effect using realistic values. The average heterozygous disadvantage of a lethal mutation in Drosophila is thought to be h = 1/40. Let f = 0 in D. melanogaster and f = 1/100 in D. simulans. Then, the average equilibrium frequency of lethals in D. simulans would be only 0.715 times its value in D. melanogaster. In other words, temporal fluctuations in effective population size would result in an almost 30% decrease of lethals. The breeding structure could also affect the mildly detrimental load. These factors would decrease the effect of background selection. The mutation load is thought to decrease the amount of segregating neutral variation by a factor f₀ = exp[-v/(hs + r)] (HUDSON and KAPLAN 1995 ; NORDBORG et al. 1996 ), where r is the local recombination rate. In this expression, the mean selective disadvantage hs of mutant heterozygotes depends only on dominance, whereas our results suggest that it could also be affected by the breeding structure of populations. Provided that the interspecific difference in breeding structure extends over a substantial part of their range, this might be an additional factor involved in differences in genetic variation between the two fruit-fly species. The D. simulans genome not only is more variable than that of D. melanogaster at the nucleotide level (AQUADRO et al. 1988 ; MORIYAMA and POWELL 1996 ), but also is less variable for natural inversion polymorphisms (LEMEUNIER and AULARD 1992 ). In this and in the differential chromosome patterns of molecular variation in its African range (ANDOLFATTO 2001 ), it resembles D. melanogaster X variation more than autosomal variation. A possibility is that the purging effect of temporal variations in population size contributes to all of these effects.

It thus appears that, even though D. melanogaster and D. simulans are very closely related, a correct interpretation of the differences in their population genetic parameters may require an extensive knowledge of the demographic regime of the populations from which the samples are collected, knowledge which is as yet lacking.

	ACKNOWLEDGMENTS

We thank Robert Barbault for encouragement and Richard Lewontin for discussions on the genetics of natural populations. Matthew Cobb and Christian Schlötterer made fruitful comments on the first draft of this article. We thank two anonymous referees for their substantial and useful criticisms. M.V. expresses special gratitude to Matthew Cobb for 18 years of friendship and comradeship. E.G. was financially supported by a grant of the Conseil Interprofessionnel du Vin de Bordeaux. Her laboratory work was financially supported by Centre National de la Recherche Scientifique (CNRS) Institut d'Ecologie Fondamentale et Appliquée FR 101. M.V. is financially supported by the Groupe de Recherche CNRS 1928 Génomique des Populations and the Plan Pluri-Formation network Populations Fractionnées et Insulaires of the Ecole Pratique des Hautes Etudes.

Manuscript received April 29, 2003; Accepted for publication October 27, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AGIS, M. and C. SCHLÖTTERER, 2001  Microsatellite variation in natural Drosophila melanogaster populations from New South Wales (Australia) and Tasmania. Mol. Ecol. 19:1197-1205.

ANDOLFATTO, P., 2001  Contrasting patterns of X-linked and autosomal nucleotide variation in Drosophila melanogaster and Drosophila simulans.. Mol. Biol. Evol. 18:279-290.[Abstract/Free Full Text]

AQUADRO, C. F., K. M. LADO, and W. A. NOON, 1988  The rosy region of Drosophila melanogaster and Drosophila simulans. I. Contrasting levels of naturally occurring DNA restriction map variation and divergence. Genetics 119:875-888.[Abstract/Free Full Text]

BEGON, M., 1977  The effective size of a natural Drosophila subobscura population. Heredity 38:13-18.[Medline]

BEGUN, D., 2001  The frequency distribution of nucleotide variation in Drosophila simulans.. Mol. Biol. Evol. 18:1343-1352.[Abstract/Free Full Text]

BÉNASSI, V. and M. VEUILLE, 1995  Comparative population structuring of molecular and allozyme variation of Drosophila melanogaster Adh between Europe, West Africa and East Africa. Gen. Res. 65:95-103.

BOULÉTREAU-MERLE, J., 1992 Two paths for geographical expansion, pp. 43–49 in Advances in Regulation of Insect Reproduction, edited by B. BENNETTOVÀ, I. GELBIC and T. SOLDÀN. Czech Academy of Sciences, Prague.

CHARLESWORTH, B., 2001  The effect of life-history and mode of inheritance on neutral genetic variability. Genet. Res. 77:153-166.[CrossRef][Medline]

CHARLESWORTH, B., M. T. MORGAN, and D. CHARLESWORTH, 1993  The effect of deleterious mutations on neutral molecular variation. Genetics 134:1289-1303.[Abstract]

COBB, M., M. HUET, D. LACHAISE, and M. VEUILLE, 2000  Fragmented forests, evolving flies: molecular variation in African populations of Drosophila teissieri.. Mol. Ecol. 9:1591-1597.[CrossRef][Medline]

CROW, J. F., and M. KIMURA, 1970 An Introduction to Population Genetics Theory. Harper & Row, New York.

DANIELI, G. A. and R. COSTA, 1977  Transient equilibrium at the est-6 locus in wild populations of Drosophila melanogaster.. Genetica 47:37-41.

DAVID, J., 1979  Attractive behavior toward human constructions helps to explain the domestic and cosmopolitan status of some drosophilids. Experientia 35:1436-1437.[CrossRef]

DOBZHANSKY, TH. and S. WRIGHT, 1941  Relations between mutation rate and accumulation of lethals in populations of Drosophila pseudoobscura.. Genetics 26:23-51.[Free Full Text]

FISHER, R. A., 1930 The Genetical Theory of Natural Selection. Clarendon Press, Oxford.

FRANKHAM, R., 1995  Conservation genetics. Annu. Rev. Genet. 29:305-327.[CrossRef][Medline]

GIRARD, P and L. PALABOST, 1976  Etude du polymorphisme enzymatique de 15 populations naturelles de Drosophila melanogaster.. Arch. Zool. Exp. Genet. 117:41-55.

GREEN, M. M., 2002  It really is not a fruit fly. Genetics 162:1-3.[Free Full Text]

GREENBERG, R. and J. F. CROW, 1960  A comparison of the effect of lethal and detrimental chromosomes from Drosophila populations. Genetics 45:1153-1168.[Free Full Text]

GUO, S. W. and E. A. THOMPSON, 1992  Performing the exact test of Hardy-Weinberg proportions for multiple alleles. Biometrics 48:361-372.[CrossRef][Medline]

HAMBLIN, M. T. and M. VEUILLE, 1999  Population structure among African and derived populations of D. simulans: evidence for ancient subdivision and recent admixture. Genetics 153:305-317.[Abstract/Free Full Text]

HUBBY, J. L. and R. C. LEWONTIN, 1966  A molecular approach to the study of genic heterozygosity in natural populations. I. The number of alleles at different loci in Drosophila pseudoobdcura. Genetics 54:577-594.[Free Full Text]

HUDSON, R. R. and N. L. KAPLAN, 1995  Deleterious background mutations with recombination. Genetics 141:1605-1617.[Abstract]

IVES, P. T., 1945  The genetic structure of American populations of Drosophila melanogaster.. Genetics 30:167-196.[Free Full Text]

IVES, P. T., 1950  The importance of mutation rate genes in evolution. Evolution 4:236-252.[CrossRef]

IVES, P. T., 1959  Chromosomal distribution of mutator- and radiation-induced mutations in D. melanogaster.. Evolution 13:526-531.[CrossRef]

IVES, P. T. and H. T. BAND, 1986  Continuing studies on the south Amherst Drosophila melanogaster natural population during the 1970's and 1980's. Evolution 40:1289-1302.[CrossRef]

KEIGHTLEY, P. D. and A. EYRE-WALKER, 1999  Terumi Mukai and the riddle of deleterious mutation rates. Genetics 153:515-523.[Abstract/Free Full Text]

KIMURA, M., T. MARUYAMA, and J. F. CROW, 1963  The mutation load in small populations. Genetics 48:1303-1312.[Free Full Text]

LABONNE, G., G. FAUVEL, T. LECLANT, and J. B. QUIOT, 1983  Intérêt des pièges à fils dans l'étude des populations de pucerons ailés. Agronomie 3:315-326.

LACHAISE, D., M.-L. CARIOU, J. R. DAVID, F. LEMEUNIER, and L. TSACAS, 1988  Historical biogeography of the Drosophila melanogaster species subgroup. Evol. Biol. 22:159-225.

LEMEUNIER, F., and S. AULARD, 1992 Inversion polymorphism in Drosophila melanogaster, pp. 339–405 in Drosophila Inversion Polymorphism, edited by C. B. KRIMBAS and J. R. POWELL. CRC Press, Boca Raton, FL.

LEWONTIN, R. C., J. A. MOORE, W. B. PROVINE and B. WALLACE, 1981 Dobzhansky's Genetics of Natural Populations, I–XLIII. Columbia University Press, New York.

MALÉCOT, G., 1946  La consanguinité dans une population limitée. C. R. Acad. Sci. Paris 222:841-843.

MALÉCOT, G., 1948 Les Mathématiques de l'Hérédité. Masson & Cie, Paris.

MICHALAKIS, Y. and M. VEUILLE, 1996  Length variation of CAG/CAA trinucleotide repeats in natural populations of Drosophila melanogaster and its relation to the recombination rate. Genetics 143:1713-1725.[Abstract]

MORIYAMA, E. N. and J. R. POWELL, 1996  Intraspecific nuclear DNA variation in Drosophila. Mol. Biol. Evol. 13:261-277.[Abstract]

MOUSSET, S. and N. DEROME, 2004  Molecular polymorphism in Drosophila melanogaster and D. simulans: What did we learn from recent studies? Genetica 120:79-86.[CrossRef][Medline]

MUKAI, T. and R. A. VOELKER, 1997  The genetic structure of natural populations of Drosophila melanogaster XIII: further studies on linkage disequilibrium. Genetics 86:175-185.

MUKAI, T. and O. YAMAGUCHI, 1974  The genetic structure of natural populations of Drosophila melanogaster. XI. Genetic variability in a local population. Genetics 76:339-366.[Abstract/Free Full Text]

MUKAI, T., L. E. METTLER, and S. I. CHIGUSA, 1971  Linkage disequilibrium in a local population of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 68:1065-1069.[Abstract/Free Full Text]

MUKAI, T., M. BABA, M. AKIYAMA, N. UOWAKI, and S. KUSAKABE et al., 1985  Rapid change in mutation rate in a local population of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 82:7671-7675.[Abstract/Free Full Text]

NEI, M., 1968  The frequency distribution of lethal chromosomes in finite populations. Proc. Natl. Acad. Sci. USA 60:517-524.[Free Full Text]

NORDBORG, M., B. CHARLESWORTH, and D. CHARLESWORTH, 1996  The effect of recombination on background selection. Genet. Res. 67:159-174.[Medline]

RAYMOND, M. and F. ROUSSET, 1995  Genepop (version 1.2): population genetics software for exact tests and ecumenism. J. Hered. 86:248-249.[Free Full Text]

ROUAULT, J. and J. DAVID, 1982  Evolutionary biology of Drosophila melanogaster and D. simulans: a behavioral divergence in microhabitat selection. Acta Oecol. 3:331-338.

ROUSSET, F. and M. RAYMOND, 1995  Testing heterozygosity excess and deficiency. Genetics 140:1413-1419.[Abstract]

SMITH, D. B., C. H. LANGLEY, and F. M. JOHNSON, 1978  Variance component analysis of allozyme frequency data from eastern populations of Drosophila melanogaster.. Genetics 88:121-137.[Abstract/Free Full Text]

STURTEVANT, A. H., 1920  Genetics studies on Drosophila simulans. I. Introduction. Hybrids with Drosophila melanogaster. Genetics 5:488-500.[Free Full Text]

TAYLOR, C. E. and J. R. POWELL, 1977  Microgeographic differentiation of chromosomal and enzyme polymorphisms in Drosophila persimilis.. Genetics 85:681-695.[Abstract/Free Full Text]

VEUILLE, M., M. COBB, N. DEROME, and E. GRAVOT, 2004  Historicity and population genetics parameters in D. melanogaster and D. simulans.. Genetica 120:61-70.

WAPLES, R. S., 1989  A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics 121:379-392.[Abstract/Free Full Text]

WEIR, B. S. and C. C. COCKERHAM, 1984  Estimating F-statistics for the analysis of population structure. Evolution 38:1358-1370.[CrossRef]

WRIGHT, S., 1931  Evolution in Mendelian populations. Genetics 16:97-159.[Free Full Text]

WRIGHT, S., 1937  The distribution of gene frequencies in populations. Proc. Natl. Acad. Sci. USA 23:307-320.[Free Full Text]

WRIGHT, S., 1951  The genetical structure of populations. Ann. Eugen. 15:323-354.

WRIGHT, S., 1969 Evolution and the Genetics of Populations: Vol. 2. The Theory of Gene Frequencies. University of Chicago Press, Chicago.

WRIGHT, S., TH. DOBZHANSKY, and W. HOVANITZ, 1942  The allelism of lethals in the third chromosome of Drosophila pseudoobscura. Genetics 27:363-394.[Free Full Text]

This article has been cited by other articles:

				A. Sanchez-Gracia and J. Rozas Unusual Pattern of Nucleotide Sequence Variation at the OS-E and OS-F Genomic Regions of Drosophila simulans Genetics, April 1, 2007; 175(4): 1923 - 1935. [Abstract] [Full Text] [PDF]

				E. Baudry, N. Derome, M. Huet, and M. Veuille Contrasted Polymorphism Patterns in a Large Sample of Populations From the Evolutionary Genetics Model Drosophila simulans Genetics, June 1, 2006; 173(2): 759 - 767. [Abstract] [Full Text] [PDF]

				J. J. Welch Estimating the Genomewide Rate of Adaptive Protein Evolution in Drosophila Genetics, June 1, 2006; 173(2): 821 - 837. [Abstract] [Full Text] [PDF]