Abstract

Seasonal environmental changes have the potential to influence the genetic structure of species with a short generation time, such as Drosophila. We previously found the seasonal change in linkage disequilibrium (LD) between the chemoreceptor (Cr) genes in a local Japanese population (Kyoto [KY]). This could be caused by fluctuation in the population size or selection in temporally heterogeneous environments or both. Here, we analyzed the scale of LD between 51 X-linked polymorphisms (10 Cr and 41 non-Cr gene markers) in the 2 seasonal samples from the KY population and an autumn sample from 106 localities in and around Japan (Ja03au). Many of the non-Cr genes have receptor function but fewer functional connections to each other. The magnitude of LD in Ja03au did not significantly differ from that in the KY autumn sample. The lack of local differentiation was confirmed in an autumn sample from another local Japanese population. On the other hand, the magnitude of LD was significantly larger in spring than in autumn in the 2 independent KY samples. This suggests that reduction in the population size during winter increased the magnitude of LD in spring in the mainland population in Japan. Long-distance LD could be a useful measure for assessing seasonal fluctuation in effective population size.

Seasonal environmental changes have the potential to influence the genetic structure of species with a short generation time, such as Drosophila. Seasonal changes in the frequency of chromosomal inversions have long been known in Drosophila species, and it was proposed that they were caused by varying selection (Dobzhansky 1948, 1970). Furthermore, the population size likely varies seasonally, which can also affect the genetic structure of the population. In spite of extensive surveys of the genetic structure of natural populations of several Drosophila species, relatively little is known about seasonal fluctuation of population size (Lumme and Lakovaara 1983). A better assessment of the impact of seasonality on the genetic structure can deepen our understanding of currently acting selection and the causes of within-species variation.

The impact of random genetic drift in natural populations has been inferred from changes in allele frequency over generations (e.g., Fisher and Ford 1947; Schaffer et al. 1977; Mueller et al. 1985). Such studies require multiple samples for a single estimation, and thus, a seasonal comparison of their magnitudes is difficult. A recent population bottleneck can be tested by distortion of allele frequency distribution from the neutrality (Cornuet and Luikart 1996; Luikart et al. 1998). However, this test is unlikely able to monitor short-term changes such as seasonal fluctuation because it would take a long time to recover the frequency distribution.

In this context, nonrandom association of polymorphisms at different sites, that is, linkage disequilibrium (LD), may serve as an alternative measure. LD can arise from physical linkage, finite population size, admixture and structure of populations, or epistatic natural selection, whereas it gradually decays by recombination. Therefore, long-distance LD is smaller in magnitude than short-distance LD and less dependent on the past history, the initial state of polymorphisms (ancestral haplotype), and their respective ages. Generally, in Drosophila, little LD is observed between pairs of sites separated by >2 kb (e.g., Miyashita and Langley 1988), except when polymorphic inversions are involved (e.g., Mukai et al. 1971; Charlesworth B and Charlesworth D 1973; Langley et al. 1974). This low level of background LD makes interlocus LD more sensitive to short-term demographic changes, and we may be able to find signatures of seasonal change in the genetic structure in terms of long-distance LD.

We previously found a greater amount of LD between polymorphisms in the 98 Drosophila chemoreceptor (Cr) genes in a spring sample compared with an autumn sample and a significant excess of associations between one frequent and one less common allele only for replacement polymorphisms in the spring sample (Takano-Shimizu et al. 2004). It seems unlikely that these were simply caused by seasonal bottlenecks associated with overwinter mortality. Because the Cr genes could be functionally associated with one another, we inferred that epistatic selection on these genes, in combination with bottleneck, was responsible for the seasonal changes in scale of LD. This can be tested by studying genes that are functionally independent to each other. Under this hypothesis, seasonal changes in LD for such genes are expected to be smaller than those for functionally connected genes.

Here, we report LDs between 51 polymorphisms at 50 X-linked genes. These genes are distributed throughout the X chromosome and mostly loosely linked. Most of them have no obvious functional connection to one another. In addition, no inversion polymorphism is known to exist on the X chromosome. Thus, we could ignore these potentially confounding effects. The main objective of this study is to test whether the seasonality is unique to the Cr genes or a common feature of genes on the X chromosome. Other objectives are to examine the differences in LD between polymorphisms at replacement and silent sites and to evaluate the genetic structure of extant populations. For the latter purpose, we used a structured sample (i.e., a collection of 1 or 2 males from each of 106 localities in and around Japan) in addition to samples from local populations with different population densities. Population structure is expected to increase LD, and then its effect may be detected by comparing the amount of LD between such samples. The present study revealed that the seasonal change of the interlocus LD was common to the X-linked genes and that the scale of LD was little affected by sampling strategy. Long-distance LD could be a useful index for detecting short-term changes in effective population size.

Materials and Methods

Fly Samples

Ja03au Sample

By sweeping over sites attractive for Drosophila (harvested grapes, refuse at fruits market, or fermenting fruit waste in orchards) or with banana bait traps, male flies were collected at 106 localities in and around Japan (Figure 1), 1 in Hokkaido, 70 in the mainland, 4 in Shikoku, 24 in Kyushu, 3 in Okinawa and isolated islands, and 4 in Korea. All sites were at least several kilometers apart. Collections were carried out in 2003 between September and December. A sample of 198 male flies (n = 198 X chromosomes) was obtained, composed of 1 or 2 flies from each location.

Figure 1

Fly collecting sites for the Ja03au samples.

Figure 1

Fly collecting sites for the Ja03au samples.

KA03au Sample

The Katsunuma area is one of the centers of wine production in Japan. The town of Katsunuma (35.5°N) is rather small (34 km2), but more than 30 wineries and many grape orchards are densely distributed in the western parts of the town. KA03au sample comprised of 191 males collected on a single day by sweeping over harvested grapes in a winery in September 2003. We regarded the KA03 sample as a high-density population sample.

Kyoto and Iriomote Samples

Four Kyoto (KY) samples, KY01au (2001 autumn, n = 177), KY02sp (2002 spring, n = 187), KY02au (2002 autumn, n = 186), and KY03sp (2003 spring, n =191), and 2 Iriomote (IR) samples, IR01au (2001 autumn, n = 192) and IR03sp (2003 spring, n = 191), were also used (Takano-Shimizu et al. 2004). At all locations, males were collected with banana bait traps. The climate and flora of IR (24.2°N), which is a subtropical island in the southernmost region of Japan, are considerably different from those of KY (35.0°N) and other localities in the mainland of Japan.

Field-caught males were crossed separately to an inbred attached X chromosome strain, TT-35 (C(1)RM, y wa/ y w). The F1 male progeny, which had an identical X chromosome to that of their father, was collected and used for typing.

Markers and Data Sets

In addition to the 10 markers in the 10 Cr genes (Takano-Shimizu et al. 2004), we newly identified 41 biallelic single nucleotide polymorphisms (SNPs) in 40 X chromosome genes including another Cr gene, Gr9a, (Supplementary Table 1). Many of these 40 genes have receptor functions but are thought to have fewer functional connections to one another. Of the 51 markers, 27 were replacement and 24 were silent polymorphisms. All are SNPs except Gr2a, which is a complex change of CAC/GGCA and categorized as a replacement polymorphism (Takano-Shimizu et al. 2004). The average distance between the 51 polymorphisms is 0.13 in terms of recombination frequency per generation.

In this study, we typed the above 51 markers for Ja03au and the 10 Cr gene markers for KA03au. In addition, the newly developed 41 markers were typed for KY01au and KY02sp. For the 10 Cr gene markers, we revisited the data of 6 samples in Takano-Shimizu et al. (2004): KY01au, KY02sp, KY02au, KY03sp, IR01au, and IR03sp. Consequently, we obtained the data of 3 samples (KY01au, KY02sp, and Ja03au) for the 51 markers and those of 8 samples (KY01au, KY02sp, KY02au, KY03sp, IR01au, IR03sp, KA03, and Ja03) for the 10 Cr gene markers.

DNA Extraction, Polymerase Chain Reaction, and Marker Typing

We extracted genomic DNA from F1 male progeny using the GenElute Mammalian Genomic DNA Miniprep Kit (Sigma, St Louis, MO). To amplify the variable sites, polymerase chain reaction (PCR) was carried out using the following protocol: 32 cycles of denaturing at 95 °C for 30 s, annealing at 60 °C for 30 s, and polymerizing at 72 °C for 1 min. Most genotyping was performed by allele-specific oligonucleotide hybridization (Saiki et al. 1986). The PCR primers and the allele-specific oligonucleotide probes are listed in the Supplementary Table 1. The polymorphisms in the Or2a, Gr5a, and Or7a were typed based on restriction fragment length polymorphisms according to Takano-Shimizu et al. (2004).

Data Analysis

Average heterozygosity (H), minimum genetic distance (Dm), and their standard errors were estimated as described in Nei and Roychoudhury (1974). We estimated fixation index (Wright 1951), Fst, without sampling correction using an expected H in Ja03au as the expected total H.

LD between polymorphisms was statistically tested by a 2-tailed Fisher's Exact test. Because we performed 1275 tests in the analysis of the 51 X-linked polymorphisms, the critical value for Bonferroni multiple test correction was 5%/1275 = 0.0039%. Squared correlation coefficient (r2) was calculated as a measure of LD, excluding 7 closely located polymorphism pairs whose physical distances were less than 10 kb.

For the sign test (Lewontin 1995), alleles at all marker sites were assigned so that the LD sign was negative when there was an excess of chromosomes with one frequent and one less common allele. Polymorphisms were excluded when the frequency of the less common allele was 0.4 or greater. A singleton was also excluded from the sign test.

To statistically test a difference in the average r2 values between 2 samples, permutation tests were performed. Chromosomes from the 2 samples of sizes n1 and n2 were randomly divided into 2 samples of n1 and n2 chromosomes, and the average r2 value was calculated for each sample. By repeating this process 5000 times, we obtained the probability that the between-sample difference in r2 values was equal to, or larger than, the observed difference. The tests were 2 sided. We also used permutations to test difference in the increase in the average r2 value from autumn to spring between the Cr and non-Cr genes. In this case, we calculated the average r2 values in KY01au and KY02sp samples for randomly selected 11 genes, and the distribution of their differences was used to establish significance thresholds.

Recombination Frequency

Recombination frequency between polymorphisms was calculated using the Kosambi formula (Kosambi 1944) and the standard genetic map of Drosophila melanogaster (Lindsley and Zimm 1992; Drysdale et al. 2005).

Results

Fifty-One Polymorphisms in 3 Samples

We typed 51 diallelic X chromosome polymorphisms for 3 sets of samples: Ja03au (198 chromosomes), KY01au (177 chromosomes), and KY02sp (187 chromosomes) (Supplementary Tables 2, 3, and 4). The average H was estimated to be 0.32 for all the 3 samples, meaning that there was no excess of homozygosity in the local KY samples compared with the combined fly sample (Ja03au). All 3 samples shared the same rare allele in 48 of 51 polymorphisms; the exceptions were CG6986, mys, and CG32704. The estimates of Dm were small (0.002 ± 0.001 between KY01au and KY02sp, 0.001 ± 0.001 between KY01au and Ja03au, and 0.003 ± 0.001 between KY02sp and Ja03au).

The results of LD analysis are given in Figure 2 and Table 1. Of the 1275 polymorphism pairs in Ja03au, KY01au, and KY02sp, 16, 21, and 53 pairs showed significant LDs at the 1% significance level and 5, 0, and 1 pairs did so after the Bonferroni correction, respectively (Figure 2). Excluding 7 closely located polymorphism pairs whose physical distances were less than 10 kb, the average r2 values were 0.0062 in Ja03au, 0.0073 in KY01au, and 0.0095 in KY02sp (Table 1). Although the difference in average r2 values between Ja03au and KY01au was not significant (P > 0.2), the difference between KY01au and KY02sp was highly significant (P = 0/5000). The differences between KY01au and KY02sp were all statistically significant even after dichotomizing the 51 markers into Cr and non-Cr genes and into replacement and silent sites. This finding indicates that the amount of LD increased from autumn to spring at the chromosome-wide level in the KY population. The increase in the amount of LD in the spring was most pronounced in the Cr genes (0.0074 in KY01au to 0.0135 in KY02sp in terms of average r2; see Table 1), although the observed difference in the Cr genes was not significantly larger than that in the non-Cr genes (P = 0.06 by a permutation test).

Table 1

Average r2 values for 51 X-linked genes

Sample Polymorphisms No. of pairs Average distancea Average frequencyb Average r2 
Ja03au All 1218 0.13 (0.045) 0.24 0.0062 
Crc 52 0.11 (0.038) 0.29 0.0075 
Non-Crd 739 0.13 (0.041) 0.23 0.0056 
Re 324 0.13 (0.028) 0.21 0.0075 
Sf 275 0.13 (0.044) 0.28 0.0048 
KY01au All 1268 0.13 (0.043) 0.24 0.0073 
Crc 52 0.11 (0.038) 0.29 0.0074 
Non-Crd 778 0.13 (0.039) 0.22 0.0074 
Re 350 0.13 (0.026) 0.19 0.0069 
Sf 275 0.13 (0.044) 0.29 0.0067 
KY02sp All 1268 0.13 (0.043) 0.24 0.0095 
Crc 52 0.11 (0.038) 0.29 0.0135 
Non-Crd 778 0.13 (0.039) 0.23 0.0089 
Re 350 0.13 (0.026) 0.21 0.0096 
Sf 275 0.13 (0.044) 0.27 0.0097 
Sample Polymorphisms No. of pairs Average distancea Average frequencyb Average r2 
Ja03au All 1218 0.13 (0.045) 0.24 0.0062 
Crc 52 0.11 (0.038) 0.29 0.0075 
Non-Crd 739 0.13 (0.041) 0.23 0.0056 
Re 324 0.13 (0.028) 0.21 0.0075 
Sf 275 0.13 (0.044) 0.28 0.0048 
KY01au All 1268 0.13 (0.043) 0.24 0.0073 
Crc 52 0.11 (0.038) 0.29 0.0074 
Non-Crd 778 0.13 (0.039) 0.22 0.0074 
Re 350 0.13 (0.026) 0.19 0.0069 
Sf 275 0.13 (0.044) 0.29 0.0067 
KY02sp All 1268 0.13 (0.043) 0.24 0.0095 
Crc 52 0.11 (0.038) 0.29 0.0135 
Non-Crd 778 0.13 (0.039) 0.23 0.0089 
Re 350 0.13 (0.026) 0.21 0.0096 
Sf 275 0.13 (0.044) 0.27 0.0097 

The number of pairs compared differed among the samples because there was a singleton site in Ja03au.

a

Average recombination frequency per generation (fraction of pairs of 0.01 or less recombination frequency).

b

Average rare allele frequency.

c

Between Cr and Cr polymorphisms.

d

Between non-Cr and non-Cr polymorphisms.

e

Between replacement polymorphisms.

f

Between silent polymorphisms.

Figure 2

LD (Fisher's Exact test) between 51 polymorphisms. (A) Ja03au, (B) KY01au, and (C) KY02sp. The genes are arranged from top left to bottom right according to the order in Supplementary Table 1.

Figure 2

LD (Fisher's Exact test) between 51 polymorphisms. (A) Ja03au, (B) KY01au, and (C) KY02sp. The genes are arranged from top left to bottom right according to the order in Supplementary Table 1.

The direction of LD was tested by the sign test (Lewontin 1995), and no significant deviation from the expectations was obtained regardless of the class of polymorphisms (replacement, silent, Cr, non-Cr, and all polymorphisms) and samples (Table 2).

Table 2

Numbers of positive- and negative-phase LDs

Sample Polymorphisms No. of observed/expected pairs
 
Positive or zero Negative 
Ja03au All 332/329 371/374 
Cra 12/14 16/14 
Non-Crb 217/203 218/232 
Rc 100/108 131/123 
Sd 56/58 64/62 
KY01au All 394/382 428/440 
Cra 17/14 12/15 
Non-Crb 251/244 277/284 
Rc 150/152 175/173 
Sd 46/48 59/57 
KY02sp All 374/379 450/445 
Cra 23/22 22/23 
Non-Crb 213/209 254/258 
Rc 118/132 161/147 
Sd 57/61 79/75 
Sample Polymorphisms No. of observed/expected pairs
 
Positive or zero Negative 
Ja03au All 332/329 371/374 
Cra 12/14 16/14 
Non-Crb 217/203 218/232 
Rc 100/108 131/123 
Sd 56/58 64/62 
KY01au All 394/382 428/440 
Cra 17/14 12/15 
Non-Crb 251/244 277/284 
Rc 150/152 175/173 
Sd 46/48 59/57 
KY02sp All 374/379 450/445 
Cra 23/22 22/23 
Non-Crb 213/209 254/258 
Rc 118/132 161/147 
Sd 57/61 79/75 
a

Between Cr and Cr polymorphisms.

b

Between non-Cr and non-Cr polymorphisms.

c

Between replacement polymorphisms.

d

Between silent polymorphisms.

Ten Cr Polymorphisms for 8 Samples

To compare the KY fly samples with other local population samples, we typed 10 of the above 11 Cr polymorphisms on 191 X chromosomes of the Katsunuma sample (KA03au) (Supplementary Table 5). These 10 polymorphisms were previously typed for 2 other KY samples, KY02au and KY03sp, and 2 IR samples, IR01au and IR03sp (Takano-Shimizu et al. 2004). Table 3 gives the estimates of average H and Dm for the 8 samples including Ja03au, KY01au, and KY02sp. As in the samples harboring 51 polymorphisms, the 4 KY samples did not show a reduction in H compared with Ja03au. The genetic distances between the populations were concordant with their geographical locations; the distance between Katsunuma and IR was the largest. The Fst estimate was 0.08 based on the average H in the three 2003 local samples (KY03sp, KA03au, and IR03sp) and that in Ja03au, implying little differentiation among the localities.

Table 3

Average H and Dm with standard error for 10 Cr polymorphisms in 8 samples

Sample H Dm
 
KY01au KY02sp KY02au KY03sp KA03au IR01au IR03sp 
Ja03au 0.412 ± 0.031 0.000 ± 0.001 0.004 ± 0.001 0.001 ± 0.001 0.003 ± 0.002 0.009 ± 0.003 0.021 ± 0.010 0.019 ± 0.008 
KY01au 0.414 ± 0.031  0.007 ± 0.004 0.003 ± 0.001 0.005 ± 0.002 0.010 ± 0.005 0.025 ± 0.012 0.021 ± 0.008 
KY02sp 0.414 ± 0.026   0.001 ± 0.001 0.004 ± 0.003 0.016 ± 0.006 0.029 ± 0.011 0.026 ± 0.011 
KY02au 0.402 ± 0.032    0.000 ± 0.001 0.017 ± 0.005 0.021 ± 0.010 0.020 ± 0.011 
KY03sp 0.396 ± 0.021     0.015 ± 0.004 0.024 ± 0.013 0.021 ± 0.014 
KA03au 0.360 ± 0.036      0.042 ± 0.021 0.032 ± 0.014 
IR01au 0.385 ± 0.026       0.001 ± 0.001 
IR03sp 0.384 ± 0.022        
Sample H Dm
 
KY01au KY02sp KY02au KY03sp KA03au IR01au IR03sp 
Ja03au 0.412 ± 0.031 0.000 ± 0.001 0.004 ± 0.001 0.001 ± 0.001 0.003 ± 0.002 0.009 ± 0.003 0.021 ± 0.010 0.019 ± 0.008 
KY01au 0.414 ± 0.031  0.007 ± 0.004 0.003 ± 0.001 0.005 ± 0.002 0.010 ± 0.005 0.025 ± 0.012 0.021 ± 0.008 
KY02sp 0.414 ± 0.026   0.001 ± 0.001 0.004 ± 0.003 0.016 ± 0.006 0.029 ± 0.011 0.026 ± 0.011 
KY02au 0.402 ± 0.032    0.000 ± 0.001 0.017 ± 0.005 0.021 ± 0.010 0.020 ± 0.011 
KY03sp 0.396 ± 0.021     0.015 ± 0.004 0.024 ± 0.013 0.021 ± 0.014 
KA03au 0.360 ± 0.036      0.042 ± 0.021 0.032 ± 0.014 
IR01au 0.385 ± 0.026       0.001 ± 0.001 
IR03sp 0.384 ± 0.022        

Figure 3 illustrates the average r2 values between the 10 Cr polymorphisms in the 8 samples. Despite possible differences in demographic characteristics, the values in Ja03au, KA03au, IR01au, and IR03sp were very similar to those in the 2 KY autumn samples. In contrast, the amounts of LD in the 2 KY spring samples were consistently larger than those in the other samples. The permutation probability that between-sample differences in average r2 value are equal to or larger than the observed difference was significant in both the KY01au–KY02sp (P= 0.03) and KY02au–KY03sp comparisons (P= 0/5000). These findings implied that the amounts of LD between the Cr polymorphisms are significantly larger in spring than in autumn in the 2 consecutive years. In addition, the similar amounts of r2 values in the KYau and KAau samples suggested that population density did not greatly affect the amount of LD at the present resolution level.

Figure 3

Average r2 values between 10 Cr polymorphisms.

Figure 3

Average r2 values between 10 Cr polymorphisms.

Discussion

The present study was aimed at quantitative assessment of seasonal change and population structure in terms of LD. For these purposes, we studied 3 fly samples, Ja03au, KY01au, and KY02sp, for the 51 polymorphisms, and the 5 other local population samples, KA03au, KY02au, KY03sp, IR01au, and IR03sp, for the 10 Cr polymorphisms. The average r2 for the 51 X-linked polymorphisms in the spring sample (KY02sp) was significantly larger than that in the autumn sample (KY01au), irrespective of the types of genes and polymorphisms. We thus concluded that the magnitude of LD increased at the chromosome-wide level from autumn to spring every year in the mainland population of Japan. On the other hand, there was no obvious geographic differentiation in the scale of LD.

It is very likely that the fly population is severely reduced over the winter season (Watanabe et al. 1984). Flies seem to pass the winter as adults, and there are only a few new generations over this period. Indeed, adults are more frost resistant compared with eggs, larvae, and pupae (Tucic 1979). If this is the case, the genetic structure of a population will change little during winter. To quantify the effect of bottleneck on LD, we performed computer simulations (Figure 4). In the simulation analysis, we assumed that population size alternates cyclically between N1 and N2, with the duration time of 10 and 2 generations, respectively. Assuming N1 = 5000, we studied changes in the average r2 values with 3 different bottleneck sizes (N2 = 200, 500, or 1000). We carried out the gamete-based stochastic simulation (Lynch and Force 2000), where 2 biallelic loci are located with recombination rate of 0.1. Recombination is treated as deterministic process, whereas mutation and selection are neglected. At generation 0, the allele frequencies are assumed to be 0.5 at both loci, and they are in linkage equilibrium. The r2 value was calculated from 200 gametes randomly sampled in each run and time point, and the average values were obtained from 104 independent runs. Our simulation results (Figure 4) indicate the change in amount of LD and suggest that if the observed change in the magnitude of LD during winter was caused by bottlenecks of short duration, severe reduction in population size would be required. Therefore, long-distance LD could be useful for assessing the short-term fluctuation in population size.

Figure 4

Changes in average r2 values when population size fluctuates through bottlenecks. Population size alternates cyclically between N1 and N2, with the duration time of 10 and 2 generations, respectively (the lower panel). Results for 3 different N2 values (square: 200; cross: 500; and triangle: 1000) are illustrated; N1is assumed to be 5000 throughout. Essentially, the same results were obtained with N1 = 10 000 (data not shown). The expected r2 value may be obtained by forumla, where c is the recombination rate, Ne is the effective population size, and n is the sample size (Weir and Hill 1980). In the present case, this becomes 0.0054 for Ne = 5000, 0.0072 for Ne = 1000, 0.0093 for Ne = 500, and 0.0158 for Ne = 200.

Figure 4

Changes in average r2 values when population size fluctuates through bottlenecks. Population size alternates cyclically between N1 and N2, with the duration time of 10 and 2 generations, respectively (the lower panel). Results for 3 different N2 values (square: 200; cross: 500; and triangle: 1000) are illustrated; N1is assumed to be 5000 throughout. Essentially, the same results were obtained with N1 = 10 000 (data not shown). The expected r2 value may be obtained by forumla, where c is the recombination rate, Ne is the effective population size, and n is the sample size (Weir and Hill 1980). In the present case, this becomes 0.0054 for Ne = 5000, 0.0072 for Ne = 1000, 0.0093 for Ne = 500, and 0.0158 for Ne = 200.

Although the difference in the average r2 values between KY03sp and KY02sp was not statistically significant (P > 0.2), the increase in the magnitude of LD was larger in the KY02au–KY03sp than in the KY01au–KY02sp (Figure 3). This could be caused by milder winter conditions in 2001–2002. Indeed, the winter of 2001–2002 was the record mildest winter in KY. The monthly mean air temperatures of January, March, and April in 2002 were the ninth, first, and fourth highest since 1881, respectively, whereas those in 2002–2003 were ordinary (the climatic statistics of Japan Meteorological Agency: http://www.jma.go.jp/jma/index.html).

We previously studied LD between 98 Cr genes in the entire Drosophila genome and found a large difference not only in the magnitude of LDs but also in their direction between KY01au and KY02sp (Takano-Shimizu et al. 2004). There is a significant shortage of coupling disequilibrium with associations of less common alleles but only for replacement polymorphism pairs in KY02sp. Such a bias in LD direction was not found in all the samples in this study (Table 2). Most of the genes studied here have no obvious functional connection to one another. Although further study is needed, the significant bias in the Cr genes might be due to their functional associations.

In contrast to the mainland population, a southern island population shows little seasonal difference in the magnitude of LD (Takano-Shimizu et al. 2004). This is not surprising, given the smaller seasonal climatic change in the southernmost region of Japan. The southern island populations are known to carry much larger additive genetic variance in viability than a northern mainland population of Japan, and diversifying selection is suggested to be the cause of the excessive variation (Tachida et al. 1983; Kusakabe and Mukai 1984; Takano et al. 1987). In light of the present findings, there may be other factors that contribute to the difference in genetic variance between the northern and southern populations, that is, heterogeneity in seasonal change. Reduction of population size can purge highly recessive deleterious mutations (Kirkpatrick and Jarne 2000; Glémin 2003). However, the purging effect alone cannot explain the reduced variation in the mainland population because mildly deleterious mutations that are largely responsible for the excessive genetic variance in the southern populations are only mildly recessive (the degree of dominance = 0.2–0.5; Mukai and Nagano 1983; Tachida et al. 1983). It is also conceivable that selection is more severe during winter. Increased selection intensity, in combination with population bottleneck, during winter may contribute to the reduction of the genetic load.

Repeated bottlenecks frequently occur in natural populations. Long-distance LD could be a powerful measure for assessing their impact on the genetic structure.

Supplementary Material

Supplementary material can be found at http://www.jhered.oxfordjournals.org/.

Funding

Cooperative Research Program, National Institute of Genetics (2002-B4, 2003-B4, 2004-B5, 2006-A17, and 2007-A20 to N.I., M.I., R.K., and T.T.-S.); Ministry of Education, Culture, Sports, Science and Technology of Japan (Grants-in-Aid for Scientific Research [C] 17570087 to T.T.-S. and N.I. and 20570100 to T.T.-S.); Yamada Science Foundation (to T.T.-S. and N.I.); Mitsubishi Foundation (2003-#18 to T.T.-S.).

We thank A. Takahashi, Y. Ishii, K. Suzuki, and K. Takeshita for their technical assistance, and A. E. Szmidt for correcting the English of the manuscript. We also thank the many farmers and shopkeepers who generously permitted our fly collection on their lands or possessions, especially the chief priest of the Ichijoji-Akanomiya Shrine in KY, and the manager of the Marufuji Winery in Katsunuma. Thank you to the many people who kindly collected flies for us.

References

Charlesworth
B
Charlesworth
D
A study of linkage disequilibrium in populations of Drosophila melanogaster
Genetics
 , 
1973
, vol. 
73
 (pg. 
351
-
359
)
Cornuet
JM
Luikart
G
Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data
Genetics
 , 
1996
, vol. 
144
 (pg. 
2001
-
2014
)
Dobzhansky
T
Genetics of natural populations. XVI. Altitudinal and seasonal changes produced by natural selection in certain populations of Drosophila pseudoobscura and Drosophila persimilis
Genetics
 , 
1948
, vol. 
33
 (pg. 
158
-
176
)
Dobzhansky
T
Genetics of the evolutionry process
 , 
1970
New York
Columbia University press
Drysdale
RA
Crosby
MA
The FlyBase Consortium
FlyBase: genes and gene models
Nucleic Acids Res
 , 
2005
, vol. 
33
 (pg. 
D390
-
D395
Available from: http://flybase.org/
Fisher
RA
Ford
EB
The spread of a gene in natural conditions in a colony of the moth Panaxia dominula L
Heredity
 , 
1947
, vol. 
1
 (pg. 
143
-
174
)
Glémin
S
How are deleterious mutations purged? Drift versus nonrandom mating
Evolution
 , 
2003
, vol. 
57
 (pg. 
2678
-
2687
)
Kirkpatrick
M
Jarne
P
The effects of a bottleneck on inbreeding depression and the genetic load
Am Nat
 , 
2000
, vol. 
155
 (pg. 
154
-
167
)
Kosambi
DD
The estimation of map distances from recombination values
Ann Eugen
 , 
1944
, vol. 
12
 (pg. 
172
-
175
)
Kusakabe
S
Mukai
T
The genetic structure of natural populations of Drosophila melanogaster. XVII. A population carrying genetic variability explicable by the classic hypothesis
Genetics
 , 
1984
, vol. 
108
 (pg. 
393
-
408
)
Langley
CH
Tobari
YN
Kojima
K
Linkage disequilibrium in natural populations of Drosophila melanogaster
Genetics
 , 
1974
, vol. 
78
 (pg. 
921
-
936
)
Lewontin
RC
The detection of linkage disequilibrium in molecular sequence data
Genetics
 , 
1995
, vol. 
140
 (pg. 
377
-
388
)
Lindsley
DL
Zimm
GG
The genome of Drosophila melanogaster
 , 
1992
San Diego (CA)
Academic Press
Luikart
G
Allendorf
FW
Cornuet
J-M
Sherwin
WB
Distortion of allele frequency distributions provides a test for recent population bottlenecks
J Hered
 , 
1998
, vol. 
89
 (pg. 
238
-
247
)
Lumme
J
Lakovaara
S
Ashburner
M
Carson
HL
Thompson
JN
Jr
Seasonality and diapause in Drosophilids
The genetics and biology of Drosophila. Vol. 3d
 , 
1983
New York
Academic Press
(pg. 
171
-
220
)
Lynch
M
Force
A
The probability of duplicate gene preservation by subfunctionalization
Genetics
 , 
2000
, vol. 
154
 (pg. 
459
-
473
)
Miyashita
N
Langley
CH
Molecular and phenotypic variation of the white locus region in Drosophila melanogaster
Genetics
 , 
1988
, vol. 
120
 (pg. 
199
-
212
)
Mueller
LD
Wilcox
BA
Ehrlich
PR
Heckel
DG
Murphy
DD
A direct assessment of the role of genetic drift in determining allele frequency variation in populations of Euphydryas editha
Genetics
 , 
1985
, vol. 
110
 (pg. 
495
-
511
)
Mukai
T
Mettler
LE
Chigusa
SI
Linkage disequilibrium in a local population of Drosophila melanogaster
Proc Natl Acad Sci USA
 , 
1971
, vol. 
68
 (pg. 
1065
-
1069
)
Mukai
T
Nagano
S
The genetic structure of natural populations of Drosophila melanogaster. XVI. Excess of additive genetic variance of viability
Genetics
 , 
1983
, vol. 
105
 (pg. 
115
-
134
)
Nei
M
Roychoudhury
AK
Sampling variances of heterozygosity and genetic distance
Genetics
 , 
1974
, vol. 
76
 (pg. 
379
-
390
)
Saiki
RK
Bugawan
TL
Horn
GT
Mullis
KB
Erlich
HA
Analysis of enzymatically amplified β-globin and HLA-DQα DNA with allele-specific oligonucleotide probes
Nature
 , 
1986
, vol. 
324
 (pg. 
163
-
166
)
Schaffer
HE
Yardley
D
Anderson
WW
Drift of selection: a statistical test of gene frequency variation over generations
Genetics
 , 
1977
, vol. 
87
 (pg. 
371
-
379
)
Tachida
H
Matsuda
M
Kusakabe
S-I
Mukai
T
Variance component analysis for viability in an isolated population of Drosophila melanogaster
Genet Res
 , 
1983
, vol. 
42
 (pg. 
207
-
217
)
Takano
T
Kusakabe
S
Mukai
T
The genetic structure of natural populations of Drosophila melanogaster. XX. Comparison of genotype-environment interaction in viability between a northern and a southern population
Genetics
 , 
1987
, vol. 
117
 (pg. 
245
-
254
)
Takano-Shimizu
T
Kawabe
A
Inomata
N
Nanba
N
Kondo
R
Inoue
Y
Itoh
M
Interlocus nonrandom association of polymorphisms in Drosophila chemoreceptor genes
Proc Natl Acad Sci USA
 , 
2004
, vol. 
101
 (pg. 
14156
-
14161
)
Tucic
N
Genetic capacity for adaptation to cold resistance at different developmental stages of Drosophila melanogaster
Evolution
 , 
1979
, vol. 
33
 (pg. 
350
-
358
)
Watanabe
TK
Inoue
Y
Watada
M
Adaptation of Drosophila simulans in Japan
Jpn J Genet
 , 
1984
, vol. 
59
 (pg. 
225
-
235
)
Weir
BS
Hill
WG
Effect of mating structure in linkage disequilibrium
Genetics
 , 
1980
, vol. 
95
 (pg. 
477
-
488
)
Wright
S
The genetical structure of populations
Ann Eugen
 , 
1951
, vol. 
15
 (pg. 
323
-
354
)

Author notes

Corresponding Editor: James Thompson