Annotated Bibliography

Copyright © 1998 by the Genetics Society of America

Evidence for Genetic Hitchhiking Effect Associated With Insecticide Resistance in Aedes aegypti

Guiyun Yan,* Dave D. Chadee† and David W. Severson*,1

* Department of Animal Health and Biomedical Sciences, University of Wisconsin, Madison, Wisconsin 53706 and † Insect Vector Control Division, Ministry of Health, St. Joseph, Trinidad and Tobago, West Indies

Manuscript received April 3, 1997 Accepted for publication October 29, 1997

A B S T R A C T Information on genetic variation within and between populations is critical for understanding the evo-

lutionary history of mosquito populations and disease epidemiology. Previous studies with Drosophila sug- gest that genetic variation of selectively neutral loci in a large fraction of genome may be constrained by fixation of advantageous mutations associated with hitchhiking effect. This study examined restriction frag- ment length polymorphisms of four natural Aedes aegypti mosquito populations from Trinidad and Tobago, at 16 loci. These populations have been subjected to organophosphate (OP) insecticide treatments for more than two decades, while dichlor-diphenyltrichlor (DDT) was the insecticide of choice prior to this pe- riod. We predicted that genes closely linked to the OP target loci would exhibit reduced genetic variation as a result of the hitchhiking effect associated with intensive OP insecticide selection. We also predicted that genetic variability of the genes conferring resistance to DDT and loci near the target site would be sim- ilar to other unlinked loci. As predicted, reduced genetic variation was found for loci in the general chro- mosomal region of a putative OP target site, and these loci generally exhibited larger FST values than other random loci. In contrast, the gene conferring resistance to DDT and its linked loci show polymorphisms and genetic differentiation similar to other random loci. The reduced genetic variability and apparent gene deletion in some regions of chromosome 1 likely reflect the hitchhiking effect associated with OP in- secticide selection.

MOSQUITOES are important vectors for several human pathogens because of their close associa- tion with humans. Mosquito habitats often change rap- idly as a result of vector control efforts; therefore, successful adaptation to varying human habitats is es- sential for mosquito reproduction. Adaptation ability of an organism depends on its genetic variability. Infor- mation on genetic variation within and between popu- lations is critical for understanding the evolutionary history of mosquito populations and disease epidemiol- ogy (Tabachnick and Black 1996). Protein electro- phoresis and DNA sequence analyses have revealed remarkable variation in many genes in natural popula- tions of Drosophila and other species, but the genetic variability seems to differ substantially for genes in dif- ferent genome regions (Aquadro 1992). Distribution patterns of genetic variants in natural populations are the joint effects of various evolutionary forces and de- mographic factors, including random genetic drift, se- lection, recombination, mutation, gene flow, mating system and life history (e.g., colonization, range expan- sions or contractions; Slatkin 1985). Population life

Corresponding author: Guiyun Yan, Department of Biological Sci- ences, State University of New York, 109 Cooke Hall, Buffalo, NY 14260. E-mail: [email protected]

1Present address: Department of Biological Sciences, State Univer- sity of Notre Dame, Notre Dame, IN 46556.

history and mating structure influence all loci equally, but selection affects only the target loci (Kreitman and Akashi 1995). Variation of selectively neutral loci may also be constrained by the hitchhiking effect, particu- larly in genome regions with low recombination rates and under extensive selection (Maynard Smith and Haigh 1974). Recent studies with D. melanogaster sug- gest that the hitchhiking effect may have occurred over a large fraction of the insect genome (Begun and Aquadro 1992). In this study we analyzed restriction site variation of 16 loci in natural populations of the yellow fever mosquitoes, Aedes aegypti, and provide evi- dence that the hitchhiking effect may have reduced ge- netic variation in the genome regions around a puta- tive insecticide resistance locus.

A. aegypti is an important vector of yellow fever and dengue fever viruses in many tropical countries, includ- ing Trinidad and Tobago, West Indies. Control efforts for A. aegypti have focused primarily on habitat reduc- tion and chemical treatment, which is based on the de- struction of breeding sites and the use of insecticides, including dichlor-diphenyltrichlor (DDT) in the 1950s and several organophosphates (OP) since the 1960s. The wide use of insecticides has been a powerful selec- tion agent, and rapid development of resistance to DDT and OPs is well documented (Gilkes et al. 1956; Rawlins and Wan 1995). The genetic mechanisms of insect resistance to various insecticides have been well

Genetics 148: 793–800 (February, 1998)

794 G. Yan et al.

characterized. For example, a point mutation in the para sodium channel gene confers one form of resis- tance to DDT (Williamson et al. 1996), and esterase (EST ) gene amplification is associated with resistance to OPs in Culex mosquitoes (Mouchès et al. 1990). A genetic linkage map, based largely on random cDNA sequences, has been constructed for A. aegypti (Sever- son et al. 1993), and several insecticide resistance genes have been mapped (Severson et al. 1997). In this study, we used cDNA markers distributed across the mosquito genome to examine DNA polymorphism and popula- tion genetic differentiation, and to examine mosquito genome structural changes associated with strong selec- tion imposed by insecticides. Several studies have investigated the genetic variation of various genera of mosquito populations with isozyme, RAPD-DNA, mic- rosatellite and mitochondrial DNA markers (Powell et al. 1980; Tabachnick and Wallis 1985; Conn et al. 1993; Chevillon et al. 1995; Apostol et al. 1996). Re- striction fragment length polymorphism (RFLP) mark- ers are particularly suitable for population genetic stud- ies, because they are presumably neutral, highly polymorphic, segregate as codominant markers, and can be used for studies of other mosquito species (Sev- erson et al. 1994a). We chose Trinidad and Tobago populations because the population history is known and surveillance programs have been well established there. Population historical information is important for the interpretation of genetic data. Because the mos- quito populations have been under selection of OP in- secticides, genetic variation of loci closely linked to an esterase locus conferring resistance would be reduced if the hitchhiking effect has occurred. The hitchhiking effect would be prominent in genome regions with strong linkage disequilibrium and intense selection (e.g., insecticides). In contrast, gene diversity at the para locus and other neighboring loci is expected to be similar to unlinked loci in the genome, because selec- tion pressure has been removed at the para locus since DDT was abandoned more than two decades ago.

M AT E R I A L S A N D M E T H O D S

Natural history of A. aegypti in Trinidad and Tobago: It is gen- erally believed that domestic A. aegypti originated from an African sylvan ancestor, and was introduced to the New World from West Africa via transoceanic trade during the fifteenth to seventeenth centuries (Tabachnick 1991). Caribbean pop- ulations probably represent the initial introduction of the mosquito species into the New World in the course of New World colonization. The first outbreak of yellow fever in Trin- idad was recorded in 1796, and in 1820 in Tobago.

In the 1950s, intensive vector control programs aimed to- ward mosquito eradication were adopted, primarily by the widespread usage of DDT. In the early 1960s, Trinidad was considered free of A. aegypti, but was reinfested in 1962. Over- all, mosquito populations in Tobago have been exposed to fewer insecticides than Trinidad populations. A. aegypti is now widely distributed in Trinidad and Tobago, despite continued

intensive vector control efforts through the use of OP insecti- cides. Insecticide applications not only impose strong selec- tion on the target loci, but also lead to recurrent reductions of population sizes.

Collection of samples: In conjunction with the A. aegypti surveillance program, in April, 1995, we collected three geo- graphically-distinct samples from Trinidad and one sample from Tobago (Figure 1). These four villages share similar cli- mates, including temperature and the annual amount of rain- fall. For each village, 100 ovitraps were distributed, and about half of the village’s residential area was covered (approxi- mately two traps every five houses). Each ovitrap consisted of a black plastic container roughly half-filled with water into which a rectangular masonite strip was placed in an upright manner. Female A. aegypti mosquitoes will readily oviposit on the masonite strip, near the water interface. After 2–3 days, the masonite strips were removed, transported to the labora- tory, where attached eggs were allowed to hatch, and reared into adults. All adults were identified as A. aegypti by micro- scopic examination, and were frozen for subsequent DNA analysis. Previous studies with A. aegypti in Puerto Rico suggest that the mean number of families represented per ovitrap was 4.7 (e.g., several female mosquitoes frequently oviposit in the same container; Apostol et al. 1994). Therefore, it is unlikely that siblings within a subpopulation would be sampled.

RFLP and probe selection: We genotyped a total of 870 mosquitoes for four populations (n 5 150 for Curepe, 262 for Couva, 258 for San Fernando, and 200 for Tobago). DNA ex- traction from individual mosquitoes, digestion with EcoRI, Southern blotting and hybridization were as previously de- scribed (Severson et al. 1993). Fifteen mapped RFLP markers were selected to provide broad coverage of the A. aegypti ge- nome with an average resolution of 10.6 cM (Figure 2). All clones used were random cDNA clones with the exception of the para and Mal I clones. Mal I is a gene specifically expressed in the salivary glands, and its putative function is related to

Figure 1.—Map of Trinidad and Tobago. Three samples were collected from Trinidad: Curepe (10838.62'N, 61824.23'W), Couva (10826.12'N, 61828.19'W), and San Fernando (10818.11'N, 61828.21'W). One sample was col- lected from Tobago (11811.23'N, 60844.21'W).

795 Hitchhiking Effect in Mosquitoes

Figure 2.—Relative map positions of the 16 Aedes aegypti (2N 5 6) RFLP loci used in the study. Chromosome numbers are in italics. Map distances are in Kosambi centimorgans. Underlined loci were not used in the study. Esterase gene amplification is involved with resistance to organophosphate insecticides in mosquitoes (Mouchès et al. 1990). Three esterase loci were mapped to the general chromosomal loca- tions as shown in the figure (Munstermann 1990). LF250 represents duplicated loci. Markers in italics are genes with known functions; other markers are random cDNA.

sugar metabolism ( James et al. 1989). The mosquito genome consists of single or low-copy DNA sequences and repetitive DNA with short-period interspersion (Black and Rai 1988). In this study, we focused on allelic variations of single- or low- copy cDNA sequences.

Data analysis: DNA polymorphism and Hardy-Weinberg equilib- rium (HWE) tests: Molecular weights of fragments detected by each clone were estimated by comparing them to lambda- HindIII digest standards included on each gel, using the Eagle Sight image capture and analysis software (Stratagene, La Jolla, CA). DNA polymorphisms may be measured by the proportion of polymorphic loci, number of alleles, and het- erozygosity. Conformance with HWE was tested using the probability test for each locus and each population, using the GENEPOP computer program (Raymond and Rousset 1995). Because this test is robust to allele frequencies, rare al- leles were not pooled. We further tested whether distortion from HWE resulted from deficient or excessive heterozygos- ity, using the FIS statistics (Weir 1990; Rousset and Raymond 1995). FIS is defined as [1 2 (observed heterozygosity/ex- pected heterozygosity from HWE)]. Because FIS estimates at individual loci may be unduly influenced by rare alleles, we tested the significance of the average FIS over all loci using the method of Robertson and Hill (1984). Variations in het- erozygosity among the populations were analyzed following the method of Weir (1990), using the analysis of variance (ANOVA) with subpopulations, individuals, loci and inter- actions of loci and individuals as factors. All factors were treated as random effects except loci.

Population genetic structure, gene flow and genetic distance: Population genetic structure was examined with Wright’s F-statistics, based on the procedure of Weir and Cockerham (1984) and using the FSTAT computer program (Goudet 1995). Standard deviations (SD) of F-statistics were obtained

for each locus by a jackknife procedure over the alleles, and were used to test the significance of the F statistics. We first tested whether the three populations from Trinidad were sig- nificantly substructured, then included the Tobago popula- tion data in the analysis.

Gene flow (Nm) was estimated from the standardized- among-population genetic variance (FST) estimate of each lo- cus using the relationship Nm 5 (1/FST 2 1)/4, where N is the effective population size of a deme, and m is the rate of gene flow (Wright 1943). This equation assumes the infinite-is- land model of population structure and gene flow. Few popu- lations probably conform to this assumption, but it provides a useful approximation of the relative magnitude of gene flow. Gene flow was also estimated using the private-alleles method for the appropriate loci (Slatkin and Barton 1989). Private alleles are the alleles unique to a given deme. Nei’s unbaised genetic distance for all pairs of populations was calculated based on population allele frequency for all loci (Nei 1987).

R E S U LT S

DNA polymorphisms and HWE tests: Fifteen cDNA markers examined in this study were all polymorphic. The RFLP patterns of one marker (LF250) indicate that this marker represents a gene duplication (data not shown), and therefore, the 15 markers represented a total of 16 loci. A total of 91 unique alleles were iden- tified, 68 alleles (74.7%) were common to all four pop- ulations. The average number of alleles was about five per locus (Table 1). Six loci (LF198, ARC1, LF250a, para, LF168 and Mal I) exhibited private alleles, and five private alleles were in the Tobago population. An excess of rare alleles was found: 16 alleles (18%) had a frequency less than 0.05. Under the infinite alleles model (equation 8.24; Kimura 1983), we expected to find only four alleles in this frequency class with our sample size (n 5 870). The mean sizes of restriction fragments detected by the cDNA clones weighted by their frequencies ranged from 0.73 kb at locus LF250b to 12.90 kb at locus LF352, and exhibited an overall mean of 5.13 kb (95% confidence interval: 3.61–6.64 kb).

In general, high heterozygosity was observed in all four mosquito populations, except at the LF90 locus. The LF90 locus showed significantly lower heterozygosity than the other 15 loci examined (Table 1; ANOVA, t 5 8.37, d.f. 5 1, P , 0.0001). The most heterozygous loci were LF178 on chromosome 1 and LF282 on chromo- some 2. The high heterozygosity at the LF178 locus does not seem to be a result of sex linkage (see Figure 2), be- cause males and females showed similar heterozygosity (data not shown). Population average heterozygosity over all 16 loci varied little among populations (ranged from 0.582 for Couva to 0.627 for Tobago), and such variations were not statistically significant (Table 1; ANOVA, F 5 1.08, d.f. 5 3, 49, P . 0.05). Heterozygosity is not corre- lated with the mean size of restriction fragments weighted by frequencies at a locus (r 5 0.22, d.f. 5 15, P . 0.05), but seems to correlate with the number of ob- served alleles (r 5 0.49, d.f. 515, P 5 0.052).

The genotype frequencies at several loci did not con-

796 G. Yan et al.

TABLE 1

RFLP polymorphisms of four Aedes aegypti populations from Trinidad and Tobago, measured by observed heterozygosity and the number of alleles

Curepe Couva San Fernando Tobago

Chromosome Locus n Hobs aFIS n Hobs FIS n Hobs FIS n Hobs FIS

1 LF90 5 0.122 20.036 3 0.159 20.076 5 0.256 0.160* 3 0.391 0.244*** LF230b 3 0.250 0.600*** 3 0.069 0.880*** 3 0.129 0.774*** 3 0.296 0.535*** LF198 6 0.655 0.096 6 0.504 0.100** 6 0.775 0.029 7 0.738 20.120** LF178 6 0.761 0.012 6 0.849 20.075* 6 0.851 20.100*** 6 0.828 20.120** TY7 5 0.503 0.160* 5 0.546 0.151* 5 0.543 20.001 5 0.656 0.060*

Average over chromosome 1 5.5 0.510 0.059 5.0 0.515 0.025 5.5 0.606 0.022 5.3 0.654 0.016

2 ARC1 5 0.750 20.046 5 0.724 0.015 5 0.702 20.038 6 0.725 0.010 LF138 4 0.729 20.083 4 0.714 20.201** 4 0.492 20.075 4 0.487 0.145** LF282 9 0.838 20.009 7 0.795 0.025 8 0.864 20.060* 7 0.793 20.102* LF98 5 0.642 0.092 5 0.654 20.035 6 0.682 20.073 6 0.878 20.068* LF250a 4 0.615 0.060 3 0.596 20.032 4 0.800 20.154* 3 0.562 20.183*** LF250b 3 0.644 20.009 3 0.687 20.038 3 0.662 20.238*** 3 0.557 20.160* LF115 5 0.514 20.031 5 0.366 20.016 5 0.566 20.101 4 0.562 20.077

Average over chromosome 2 5.0 0.676 20.003 4.6 0.648 20.040 5.0 0.681 20.105 4.7 0.652 20.062

3 LF352 6 0.548 0.268** 6 0.727 0.113** 6 0.441 0.292*** 6 0.550 0.263*** LF261 4 0.483 0.035 4 0.279 20.061 4 0.476 20.058 3 0.592 0.008 para 4 0.469 0.152 4 0.563 0.024 5 0.598 0.081 6 0.555 0.103 LF168 6 0.667 0.042** 7 0.516 0.145** 7 0.667 0.109 6 0.582 20.043 MalI 4 0.627 0.103** 4 0.637 0.009 3 0.640 0.034 4 0.577 0.137*

Average over chromosome 3 4.8 0.559 0.119 5.0 0.544 0.046 5.0 0.564 0.092 5.0 0.571 0.093

Average over all loci 5.1 0.598 0.050 4.8 0.582 0.003 5.1 0.626 20.012 4.9 0.627 0.006

*P , 0.05, ** P , 0.01, ***P , 0.001. Hobs, observed heterozygosity; n, number of alleles. a Significant FIS also indicates distortion from HWE. Positive FIS indicates heterozygosity deficit from HWE expectation; negative

FIS indicates excess of heterozygosity. b A large proportion of individuals show an apparent gene deletion at the LF230 locus, therefore, this locus was not used for

chromosomal average heterozygosity calculation. The heterozygosity and FIS were based on the individuals without deletions. The percentage of individuals showing the gene deletion at the LF230 locus was 41.4 for the Curepe population, 58.6 for Couva, 53.7 for San Fernando, and 46.0 for Tobago.

form to HWE. Loci on chromosome 2 generally exhib- ited a heterozygote excess, but loci on chromosome 3 that showed HWE distortion exhibited a heterozygote deficit (Table 1). The FIS values varied greatly among the loci, suggesting no systematic inbreeding occurred in these populations. The average FIS over all loci was not significantly different from 0 for each population (Table 1). Departure from HWE probably reflects either the effect of insecticide selection on some loci linked to the resistance loci, or simply sampling error.

Population genetic structure, gene flow and genetic distance: Analysis of F statistics for the three Trinidad populations found small, but statistically significant FST estimates for all loci (Table 2), suggesting that these populations are genetically differentiated. FST estimates showed a six-fold difference among loci, with an aver- age FST over all loci of 0.043. When the Tobago popula- tion was included in the analysis, the basic pattern of

estimation among the loci was not altered (Table FST

2). As expected, slightly larger FST values were obtained for most loci, and the average FST over all loci was 0.056. The para locus exhibited similar polymorphism and genetic differentiation as other random loci.

Assuming that the populations are at an equilibrium between migration and random drift, the average num- ber of migrants exchanged per generation can be cal- culated. Average gene flow (Nm) among the four popu- lations, based on the FST method, was 4.2 migrants per generation (95% confidence interval: 3.2–5.7). This es- timate was similar to the estimate based on the average frequency of six private alleles present in the popula- tions (Nm 5 4.5). Table 3 shows genetic distances and gene flow between each pair of populations calculated from the pair-wise average FST. A large gene flow be- tween the Tobago and Trinidad populations was de- tected. There was no significant correlation between ge- netic distance and geographic distance (r2 5 0.5, d.f. 5 5, P . 0.05).

797 Hitchhiking Effect in Mosquitoes

TABLE 2

FST statistics and Nm estimates of four Aedes aegypti populations of Trinidad and Tobago

FST Nm estimates of all populations a

Chromosome Locus Trinidad

populationsb Trinidad and Tobago

populationsa Based on

FST Based on the

private-alleles method

1

2

3

LF90 LF198 LF178 TY7 ARC1 LF138 LF282 LF98 LF250a LF250b LF115 LF352 LF261 para LF168 MalI

0.053 6 0.022 0.121 6 0.062 0.011 6 0.006 0.033 6 0.022 0.064 6 0.046 0.044 6 0.032 0.019 6 0.011 0.042 6 0.012 0.081 6 0.053 0.106 6 0.067 0.021 6 0.012 0.119 6 0.059 0.038 6 0.034 0.039 6 0.025 0.020 6 0.013 0.034 6 0.019

0.107 6 0.102 0.109 6 0.046 0.025 6 0.013 0.034 6 0.011 0.049 6 0.034 0.030 6 0.022 0.041 6 0.027 0.061 6 0.021 0.099 6 0.045 0.110 6 0.041 0.055 6 0.039 0.079 6 0.037 0.102 6 0.060 0.035 6 0.017 0.040 6 0.021 0.020 6 0.015

2.1 2.0 9.8 7.1 4.9 8.1 5.9 3.9 2.3 2.0 4.3 2.9 2.2 6.9 6.0

12.3

—c 3.5

— —

177.6 — — — 0.8

— — — — 70.9 1.7 9.2

Summary over 16 loci 0.043 6 0.007 0.056 6 0.008 4.2 4.5

Values are 6 SD. All FST values were significantly larger than 0 at P , 0.001. The test was performed using a jackknifing proce-

dure over samples. a n 5 4. b n 5 3. c The estimate was not available because no private alleles existed for the locus.

Hitchhiking effect on DNA polymorphisms: Gene- tic hitchhiking occurs when a (neutral) mutation changes frequency through genetic linkage to a muta- tion that is selected, resulting in reduced genetic varia- tion surrounding the target site of selection. Low DNA polymorphism at the LF90 locus suggests that hitchhik- ing has probably occurred in the genome region of the EST-4 locus. We collected additional evidence to test for this hypothesis by examining genetic polymorphism at the LF230 locus, which also is in the general genomic region of EST-4 (Figure 2). An apparent chromosomal deletion event occurred around this locus in 42–59% of the individuals (Figure 3), and low heterozygosity (0.07–0.29) was observed among those individuals with- out the apparent deletion (Table 1). However, substan- tial reduction of heterozygosity for the para locus and other loci in the vicinity of para was not observed (Ta- ble 1).

D I S C U S S I O N

DNA polymorphisms of four A. aegypti mosquito populations were examined using RFLP markers. The Trinidad populations have been exposed to OPs every 3–4 months for about two decades. These populations have therefore experienced intense selection by insecti- cides, that probably resulted in periodic population bottlenecks. A population bottleneck maintains a long-

term effect on population heterozygosity, even for spe- cies with a large intrinsic rate of growth such as A. aegypti (Nei et al. 1975). Thus, genetic polymorphisms are ex- pected to decline rapidly during insecticide use for any locus in the mosquito genome. Loci conferring OP re- sistance are expected to maintain lower genetic vari- ability than other random loci in the genome, and ge- netic variation of other closely-linked neutral loci may be reduced through genetic linkage.

If recurrent population bottlenecks have occurred in the mosquito populations, low polymorphism for all loci in the genome would be expected. In contrast to the expectation, we observed high polymorphisms for most loci. For the same loci, average heterozygosities of

TABLE 3

Nm matrix based on pairwise FST estimates and Nei’s unbiased genetic distance matrix

Curepe Couva San Fernando Tobago

Curepe Couva San Fernando Tobago

11.4 8.6 4.3

0.041

3.7 2.6

0.057 0.112

3.7

0.113 0.166 0.124

Numbers below diagonal line are FST estimates, above the diagonal line are Nei’s unbiased genetic distance.

798 G. Yan et al.

Figure 3.—Southern blot analysis of natural Aedes aegypti populations probed with cDNA clones LF230 (top) and LF90 (bottom). The mosquito genomic DNA was digested with EcoRI. Each lane is for a single mosquito. (Top) Apparent gene deletion around the LF230 locus in 55% of individuals (11 out of 20). (Bottom) Probe LF90 was used as a control to demonstrate that absence of hybridization of mosquito ge- nomic DNA to LF230 was not due to incomplete DNA diges- tion, or to poor probe conditions. DNA hybridization was ob- served for the same individuals with all other markers tested. See Table 1 for heterozygosity and percentage of gene dele- tion at the LF230 locus.

the populations studied here are substantially higher than a laboratory population, which has not been ex- posed to insecticides for more than 20 years and has not experienced population bottleneck (Yan et al. 1997). The highest heterozygosity was observed for loci at two chromosomal regions (LF198-LF178 on chromo- some 1, and LF282-LF98 on chromosome 2). Coinci- dentally, these two chromosomal regions in A. aegypti harbor genes determining vector competence for filar- ial worms and malaria parasites (Severson et al. 1994b, 1995). The high levels of heterozygosity observed may be explained by two mechanisms. The first is that the effective size of population bottlenecks has never been small, because heterogeneous habitats may provide effective shelters for the field populations. The sec- ond is that genetic polymorphisms are introduced and maintained by large gene flow among populations. The gene flow estimates seem to support the second hy- pothesis.

The LF90 locus consistently exhibited lower het- erozygosity than other loci in the genome for the four populations used in this study. The heterozygosity of a RFLP locus may be influenced by several factors, in- cluding the size of the probes, the size of the regions being probed by the probes, reduced mutation or re- combination rates in these genome regions, natural or artificial selection on a particular locus, and hitchhik- ing (selective sweep) of a selectively neutral locus by se- lectively favored substitutions at linked loci. We argue

that the polymorphism pattern of the LF90 locus likely reflects the result of a hitchhiking effect. First, the puta- tive function of the LF90 clone is coding for ribosomal protein S14 (Severson and Zhang 1996), and thus the RFLP fragments of LF90 themselves are presumably neutral. Second, LF90 is located in the general chro- mosomal region of EST-4, and gene amplification at an esterase locus is a common mechanism of OP resis- tance. Our populations have been under selective pres- sure by OP insecticides for decades. Third, low het- erozygosity of the LF90 locus is likely not related to the size of the genome region being probed by the LF90 marker, because we found no significant correlation be- tween heterozygosity and RFLP fragment sizes. Fourth, the observed heterozygosity of the LF90 locus in labora- tory colonies of A. aegypti that have not been exposed to insecticide selection was similar to other random loci across the genome (Yan et al. 1997).

Our argument for a hitchhiking effect is strength- ened by the RFLP data of the LF230 locus, which is linked to LF90 and also is in the general genomic re- gion of EST-4. We found very low DNA polymorphism and an apparent gene deletion for many individuals at this locus, a phenomenon which has not been observed in other laboratory colonies of A. aegypti (Yan et al. 1997). Gene deletions may be the result of unequal re- combination in this chromosomal region, associated with the esterase gene amplification. For example, OP resistant Culex mosquitoes possess 250–500 copies of a 30-kb esterase B1 gene, compared to a single copy in susceptible individuals (Mouchès et al. 1990). Given the genome size of A. aegypti of about 320 Mb (Zaitlin and Severson, unpublished results), if the magnitude of esterase gene amplification in the Aedes mosquitoes is similar to Culex mosquitoes, then the hitchhiking ef- fect associated with OP insecticide selection could af- fect meiotic pairing across a large genome region of chromosome 1, and could lead to chromosomal abnor- malities (i.e., deletions or duplications) within this re- gion. In contrast, a hitchhiking effect associated with DDT usage is not evident for the para locus, as indi- cated by the fact that genetic heterozygosity at the para locus and loci closely linked to para was similar to other random loci in the genome. This result is consistent with the hypothesis that in the years since DDT was aban- doned, the populations have had time to re-equilibrate.

Ideally, the hitchhiking effect should be demon- strated at the nucleotide diversity level (Kaplan et al. 1989). Unfortunately, nucleotide diversity cannot be appropriately estimated for the present data, because our RFLP data is based on one restriction enzyme. To statistically rule out the possibility of low heterozygosity caused by reduced mutation rates and/or increased functional constraints in the LF90 gene region, one needs to examine intraspecific variation and interspe- cific divergence over several gene regions for closely- related species. This method has been elegantly applied

799 Hitchhiking Effect in Mosquitoes

to Drosophila studies (Begun and Aquadro 1991, 1992). The rationale is that, if reduced mutation rate in a gene region leads to low intraspecific variation, then interspecific divergence is expected to be smaller than in other gene regions. However, hitchhiking effect only reduces intraspecific polymorphism, but will not affect interspecific divergence (Begun and Aquadro 1991). The magnitude of the hitchhiking effect should be in- versely proportional to the recombination distance. In addition, laboratory experiments should be conducted to investigate the hypothesis that gene deletions in the genome region containing LF230 result from unequal recombination between susceptible individuals and re- sistant individuals with an amplified esterase gene.

Gene flow estimates among the four mosquito popu- lations were very high compared to other animal spe- cies (Slatkin 1985). Our gene flow estimates are, how- ever, consistent with other studies of natural A. aegypti (Apostol et al. 1996) and C. pipiens (Chevillon et al. 1995) populations. Extinction and recolonization may constitute an important and powerful form of gene flow for the mosquito populations in small geographic areas. Suitable niches may often be vacated by insecti- cide applications, and then subsequently recolonized by mosquitoes. Direct estimates of A. aegypti’s natural dispersal ability, however, found limited flight ranges in urban areas, usually within 1 km when there are no geographic barriers (Reiter et al. 1995). Therefore, gene flow between the Trinidad and Tobago popula- tions must be assisted by human activities, because a 35-km strait is far beyond A. aegypti’s flight ability. A. ae- gypti-infested water containers transported from Trin- idad to Tobago have occasionally been detected since 1983 (Chadee 1990).

Slatkin and Barton (1989) showed that, in a subdi- vided population that is at a demographic equilibrium, both FST and private-alleles methods can provide rea- sonably accurate estimates of Nm under a variety of conditions. Our results indicate that Nm estimates based on FST were more consistent among loci and are probably more reliable than the private-alleles method. However, two factors may lead to biased estimates of gene flow for both methods. First, strong selection by insecticides may have significant effects on population allele frequencies, and produce local differentiation. Thus, gene flow based on FST would be underesti- mated. Second, the assumption that the populations are in genetic equilibrium may not be true for mosqui- toes. Assuming that mutation is small relative to migra- tion, the half time to equilibrium (t1/2) between gene flow and genetic drift is calculated as t1/2 5 Ln2/(2m 1 1/N ) (Crow and Aoki 1984). If the generation time of mosquitoes is 1 month, and Nm 5 4.2 (from the present study), t1/2 is about 3 years for a population with N 5 500. The interval between insecticide spray- ings is typically only 3–4 months, far less than the time required to reach genetic equilibrium.

In this study, we applied classical population genetic theory and molecular techniques to study the evolu- tionary consequences of insecticide utilization in medi- cally-important field pest populations. We made spe- cific predictions concerning gene polymorphisms and spatial variations based on the history of insecticide ap- plication. These predictions were then tested by RFLP analysis of loci representative of the mosquito genome. Our data were generally consistent with the predic- tions. We observed evidence for a hitchhiking effect in the general chromosomal region containing genes pre- sumed to confer resistance to OPs. The hitchhiking ef- fect was reflected by low DNA polymorphisms and gene deletions for loci surrounding the EST-4 locus gene re- gion. Gene deletions and reduced genetic variability in genome regions of chromosome 1 may be the result of the hitchhiking effect associated with the spread of the amplified EST-4 gene, which increases the fitness of the mosquitoes in the OP environment (Wood and Bishop 1981). Large gene flow among the mosquito popula- tions likely resulted from human-assisted migration, and may explain the rapid spread of insecticide resis- tant genes (Raymond et al. 1991).

We thank M. Fero, M. Kassner, V. Kassner, L. Smith and J. Walerak for technical assistance. M. Raymond, W. J. Tabachnick and D. Zaitlin provided valuable discussions. We are grateful to J. F. Crow, C. Denniston, K. F. Goodnight and two anonymous review- ers for critical review. This work was funded by a National Institutes of Health (NIH) National Research Service Award No. T32 (NIH grant AI-07414) to G.Y., and NIH grant AI-33127 to D.W.S.

L I T E R AT U R E C I T E D

Apostol, B. L., W. C. Black, P. Reiter and B. R. Miller, 1994 Use of randomly amplified polymorphic DNA amplified by poly- merase chain reaction markers to estimate the number of Aedes aegypti families at oviposition sites in San Juan, Puerto Rico. Am. J. Trop. Med. Hyg. 51: 89–97.

Apostol, B. L., W. C. Black, P. Reiter and B. R. Miller, 1996 Population genetics with RAPD-PCR markers: the breeding structure of Aedes aegypti in Puerto Rico. Heredity 76: 325–334.

Aquadro, C. F., 1992 Why is the genome variable? Insights from Drosophila. Trends Genet. 8: 355–361.

Begun, D. J., and C. F. Aquadro, 1991 Molecular population genet- ics of the distal portion of the X chromosome in Drosophila: evi- dence for genetic hitchhiking of the yellow-achaete region. Genet- ics 129: 1147–1158.

Begun, D. J., and C. F. Aquadro, 1992 Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. mel- anogaster. Nature 356: 519–520.

Black, W. C., and K. S. Rai, 1988 Genome evolution in mosquitoes: intraspecific and interspecific variation in repetitive DNA amounts and organization. Genet. Res. 51: 185–196.

Chadee, D. D., 1990 Aedes aegypti surveillance in Tobago, West In- dies (1983-88). J. Am. Mosq. Control Assoc. 6: 148–150.

Chevillon, C., N. Pasteur, M. Marquine, D. Heyse and M. Ray- mond, 1995 Population structure and dynamics of selected genes in the mosquito Culex pipiens. Evolution 49: 997–1007.

Conn, J., A. F. Cockburn and S. E. Mitchell, 1993 Population dif- ferentiation of the malaria vector Anopheles aquasalis using mito- chondrial DNA. J. Hered. 84: 248–253.

Crow, J. F., and K. Aoki, 1984 Group selection for a polygenic be- havioral trait: estimating the degree of population subdivision. Proc. Natl. Acad. Sci. USA 81: 6073–6077.

Gilkes, C. D., F. R. S. Kellett and H. P. S. Gillette, 1956 Yellow

800 G. Yan et al.

fever in Trinidad and the development of resistance in Aedes ae- gypti Linn, to D.D.T. formulations. W. I. Med. J. 5: 73–89.

Goudet, J., 1995 FSTAT version 1.2: a computer program to calcu- late F-statistics. J. Hered. 86: 485–486.

James, A. A., K. Blackmer and J. V. Racioppi, 1989 A salivary gland- specific maltase-like gene of the vector mosquito, Aedes aegypti. Gene 75: 73–83.

Kaplan, N. L., R. R. Hudson and C. H. Langley, 1989 The “hitch- hiking effect” revisited. Genetics 123: 887–899.

Kimura, M., 1983 The Neutral Theory of Molecular Evolution. Cam- bridge University Press, Cambridge.

Kreitman, M., and H. Akashi, 1995 Molecular evidence for natural selection. Ann. Rev. Ecol. Syst. 26: 403–422.

Maynard Smith, J., and J. Haigh, 1974 The hitch-hiking effect of a favorable gene. Genet. Res. 23: 23–35.

Mouchès, C., Y. Pauplin, M. Agarwal, L. Lemieux, M. Herzog et al., 1990 Characterization of amplification core and esterase B1 gene responsible for insecticide resistance in Culex. Proc. Natl. Acad. Sci. USA 87: 2574–2578.

Munstermann, L. E., 1990 Linkage map of the yellow fever mos- quito, Aedes aegypti, pp. 179–183 in Genetic Maps, Vol. 3, edited by S. J. O’Brien. Cold Spring Harbor Press, Cold Spring Harbor, NY.

Nei, M., 1987 Molecular Evolutionary Genetics. Columbia University Press, New York.

Nei, M., T. Maruyama and R. Chakraborty, 1975 The bottleneck effect and genetic variability in populations. Evolution 29: 1–10.

Powell, J. R., W. J. Tabachnick and J. Arnold, 1980 Genetics and the origin of a vector population: Aedes aegypti, a case study. Sci- ence 208: 1385–1387.

Rawlins, S. C., and J. H. Wan, 1995 Resistance in some Caribbean populations of Aedes aegypti to several insecticides. J. Am. Mosq. Control Assoc. 11: 59–65.

Raymond, M., and F. Rousett, 1995 GENEPOP (version 1.2): a population genetics software for exact tests and ecumenicism. J. Hered. 86: 248–249.

Raymond, M., A. Callaghan, P. Fort and N. Pasteur, 1991 Worldwide migration of amplified insecticide resistance genes in mosquitoes. Nature 350: 151–153.

Reiter, P., M. A. Amador, R. A. Anderson and G. G. Clark, 1995 Dispersal of Aedes aegypti in an urban area after blood feeding as demonstrated by rubidium-marked eggs. Am. J. Trop. Med. Hyg. 52: 177–179.

Robertson, A., and W. G. Hill, 1984 Deviations from Hardy-Wein- berg proportions: sampling variances and use in estimation of inbreeding coefficients. Genetics 107: 703–718.

Rousett, F., and M. Raymond, 1995 Testing heterozygote excess and deficiency. Genetics 140: 1413–1419.

Severson, D. W., and Y. Zhang, 1996 Generation of expressed se- quence tags and sequence-tagged sites as physical landmarks in the mosquito, Aedes aegypti, genome. Genome 39: 224–229.

Severson, D. W., A. Mori, Y. Zhang and B. M. Christensen, 1993 Linkage map for Aedes aegypti using restriction fragment length polymorphisms. J. Hered. 84: 241–247.

Severson, D. W., A. Mori, Y. Zhang and B. M. Christensen, 1994a The suitability of RFLP markers for evaluating genetic diversity among and synteny between mosquito species. Am. J. Trop. Med. Hyg. 50: 425–432.

Severson, D. W., A. Mori, Y. Zhang and B. M. Christensen, 1994b Chromosomal mapping of two loci affecting filarial worm susceptibility in Aedes aegypti. Insect Biochem. Mol. Biol. 3: 67–72.

Severson, D. W., V. Thathy, A. Mori, Y. Zhang and B. M. Chris- tensen, 1995 Restriction fragment length polymorphism map- ping of quantitative trait loci for malaria parasite susceptibility in the mosquito Aedes aegypti. Genetics 139: 1711–1717.

Severson, D. W., N. M. Anthony, O. Andreev and R. H. Ffrench- Constant, 1997 Molecular mapping of insecticide resistance genes in the yellow fever mosquito Aedes aegypti. J. Hered. 88: 520–524.

Slatkin, M., 1985 Gene flow in natural populations. Ann. Rev. Ecol. Syst. 16: 393–430.

Slatkin, M., and N. H. Barton, 1989 A comparison of three indi- rect methods for estimating average levels of gene flow. Evolu- tion 43: 1349–1368.

Tabachnick, W. J., 1991 Evolutionary genetics and arthropod- borne disease: the yellow fever mosquito. Am. Entomol. 37: 14– 24.

Tabachnick, W. J., and W. C. Black, 1996 Population genetics in vector biology, pp. 417–437 in The Biology of Disease Vectors, edited by B. J. Beaty and W. C. Marquardt. University Press of Colo- rado, Niwot, CO.

Tabachnick, W. J., and G. P. Wallis, 1985 Population genetic structure of the yellow fever mosquito Aedes aegypti in the Carib- bean: ecological considerations, pp. 371–382 in Ecology of Mosqui- toes: Proceedings of a Workshop, edited by L. P. Lounibos, J. R. Rey and J. H. Frank. Florida Medical Entomology Laboratory, Vero Beach, FL.

Weir, B. S., 1990 Genetic Data Analysis. Sinauer Associates, Sunder- land, MA.

Weir, B. S., and C. C. Cockerham, 1984 Estimating F-statistics for the analysis of population structure. Evolution 38: 1358–1370.

Williamson, M. S., D. Martinez-Torres, C. A. Hick and A. L. De- vonshire, 1996 Identification of mutations in the housefly para-type sodium channel gene associated with knockdown resis- tance (kdr) to pyrethroid insecticides. Mol. Gen. Genet. 251: 51– 60.

Wood, R. A., and J. A. Bishop, 1981 Insecticide resistance: genes and mechanisms, pp. 97–127 in Genetic Consequences of Man-Made Change, edited by J. A. Bishop and L. M. Cook. Academic Press, London.

Wright, S., 1943 Isolation by distance. Genetics 28: 114–138. Yan, G., B. M. Christensen and D. W. Severson, 1997 Com-

parisons of genetic variability and genome structure among mos- quito strains selected for refractoriness to a malaria parasite. J. Hered. 88: 187–194.

Communicating editor: A. G. Clark