genetic population structure and introgression in anopheles dirus mosquitoes in south-east asia

Molecular Ecology (2001)

10

, 569–580

© 2001 Blackwell Science Ltd

Blackwell Science, Ltd

Genetic population structure and introgression in

Anopheles dirus

mosquitoes in South-east Asia

CATHERINE WALTON,* JANE M. HANDLEY,* FRANK H. COLLINS,† VISUT BAIMAI ,‡ RALPH E . HARBACH,§ VANIDA DEESIN¶ and ROGER K. BUTLIN**

School of Biology, University of Leeds, Leeds LS2 9JT, UK,

†

Department of Biological Sciences, University of Notre Dame, Notre Dame, USA,

‡

Department of Biology, Mahidol University, Bangkok, Thailand,

§

Department of Entomology, The Natural History Museum, London, UK,

¶

Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand

Abstract

Genetic structure and species relationships were studied in three closely related mosquitospecies,

Anopheles dirus

A, C and D in Thailand using 11 microsatellite loci and comparedwith previous mitochondrial DNA (mtDNA) data on the same populations. All threespecies were well differentiated from each other at the microsatellite loci. Given the almostcomplete absence of mtDNA differentiation between

An. dirus

A and D, this endorses theprevious suggestion of mtDNA introgression between these species. The high degree ofdifferentiation between the northern and southern population of

An. dirus

C (

R

ST

= 0.401),in agreement with mtDNA data, is suggestive of incipient species. The lack of geneticstructure indicated by microsatellites in four populations of

An. dirus

A across northernThailand also concurs with mtDNA data. However, in

An. dirus

D a limited but significantlevel of structure was detected by microsatellites over ~400 km in northern Thailand,whereas the mtDNA detected no population differentiation over a much larger area(>1200 km). There is prior evidence for population expansion in the mtDNA. If this is dueto a selective sweep originating in

An. dirus

D, the microsatellite data may indicate greaterbarriers to gene flow within

An. dirus

D than in species A. Alternatively, there may havebeen historical introgression of mtDNA and subsequent demographic expansion whichoccurred first in

An. dirus

D so enabling it to accumulate some population differentiation.In the latter case the lack of migration-drift equilibrium precludes the inference of absoluteor relative values of gene flow in

An. dirus

A and D.

Keywords

:

Anopheles dirus

, gene flow, introgression, malaria vectors, microsatellites, populationexpansion

Received 4 June 2000; revision received 12 September 2000; accepted 12 September 2000

Introduction

One of the problems faced in evolutionary studies is thedifficulty of distinguishing between the effects of long-termpopulation history and ongoing gene flow (Nichols &Beaumont 1996). This problem is of particular relevance tothose species of anopheline mosquito that are vectors ofhuman malaria because an understanding of contemporarygene flow would have potential applied benefits. Forexample, both rates of gene flow within species and the

potential for gene exchange between species are relevantto the proposal to release genetically modified mos-quitoes unable to transmit the malarial parasite (Collins1994), and to the world-wide spread of insecticide resistance(Hemingway & Ranson 2000).

Anopheline mosquitoes are also of considerable interestin studies of speciation, hybridization and introgressionbecause the genus

Anopheles

is very species rich and manyof the species occur as complexes consisting of several veryclosely related taxa. There is evidence for introgressionacross species boundaries in some well-studied complexes(Kamau

et al

. 1998). Furthermore, because many anophelinemosquitoes are vectors of human malaria, they have beenthe subject of intensive study and consequently there is a

Correspondence: Catherine Walton. Fax: 0113 233 2835; E-mail:[email protected]

MEC1201.fm Page 569 Monday, March 12, 2001 6:15 PM

570

C . WA LTO N

E T A L

.

© 2001 Blackwell Science Ltd,

Molecular Ecology

, 10, 569–580

large amount of information available on their distributionand natural history. This is particularly true of two of themajor African vectors,

Anopheles gambiae

and

An. arabiensis

,but also to a lesser extent of the

An. dirus

complex fromSouth-east Asia. Here we present a phylogeographic studyof

An. dirus

species in Thailand in which the aims were notonly to gain an understanding of the evolutionary relation-ships among species but also to discern contemporarygene flow patterns in these mosquitoes.

The

An. dirus

complex comprises at least seven species(Peyton 1989). They occur in tropical evergreen rainforeststhroughout South-east Asia and most species are highlycompetent vectors of malaria, reflected in the etymologyof their name, from ‘dire’ (Scanlon & Sandinand 1965;Rosenberg 1982; Rosenberg

et al

. 1990)

. An. takasagoensis

Morishita and

An. dirus E

are restricted to Taiwan and south-western India, respectively (Sawadipanich

et al

. 1990;Peyton & Harrison 1980). The other five recognized speciesall occur in Thailand (Baimai

et al

. 1987, 1988a,b

,

c). SpeciesB occurs only in southern peninsular Thailand and extendssouthwards into peninsular Malaysia.

An. dirus

F (nownamed formally as

An. nemophilous

; Peyton & Ramalingam1988) may be widespread in Thailand but little is knownabout it since it is not highly anthropophilic. This studyconcerns only three species,

An. dirus

A, C and D (Fig. 1).Species A and D both have widespread distributions but,whereas species A is found throughout northern Thailand,species D is only found along the border with Myanmarfrom where its range extends north-westwards intoMyanmar (Tun-Lin

et al.

1995).

An. dirus

C is intimatelyassociated with limestone geography (Poopittyasataporn& Baimai 1995). Consequently this species appears to havea patchy distribution dependent on the presence of rockylimestone outcrops found along the Thai-Myanmar borderand further south into peninsular Thailand.

Initial recognition of the five species in Thailand wasachieved primarily with crossing studies and polytenechromosome banding patterns (Baimai

et al

. 1987, 1988a,c)and supported subsequently by allozymes (Green

et al

.1992). In laboratory crosses between all pairwise combina-tions of

An. dirus

A, C and D,

F

1

hybrids were producedfrom the

An. dirus

A and C cross and the

An. dirus

A and Dcross (Baimai

et al

. 1987). In the former case the progenywere fertile and viable with the exception of sterile malesproduced by the cross (female A

×

male C), whereas in thelatter case the hybrids (only produced from femaleD

×

male A) had very low viability. In addition, an

An. dirus

C-D hybrid has been identified from field collections(Walton

et al

. 1999b) so the potential exists for introgressionbetween all species pairs, although the nature and exist-ence of any premating barriers is unknown. On the basis ofthese data and inferred relationships between polytenechromosome banding patterns, it has been suggested that

An. dirus

C has diverged recently from

An. dirus

A with

species A and D having a much more distant relationship(Poopittayasataporn & Baimai 1995).

It is only recently that DNA markers have been appliedto the

An. dirus

complex but they offer the opportunity tomore fully understand evolutionary history and gene flowin these mosquitoes. There may even be unrecognizedspecies in the complex: sequence data from the secondinternal transcribed spacer (ITS2) of the ribosomal DNAsuggest that the species recognized as

An. dirus

D on thebasis of chromosomal morphology in China is, in fact,another species altogether (Xu & Qu 1997; Walton

et al

.1999a). Mitochondrial markers have been applied to

An. dirus

A, C and D populations from South-east Asia (Walton

et al

. 2000b). Significant genetic structure was observedwithin species C (possibly related to its patchy distribu-tion) but, despite substantial variation, there was no popu-lation differentiation within

An. dirus

A or D. The commongenealogy of the mitochondrial DNA (mtDNA) haplo-types of species A and D (Fig. 2) indicated that there mayhave been introgression of mtDNA between species A and

Fig. 1 Approximate geographical distributions of Anopheles dirusA, C and D in South-east Asia. The asterisks denoting species Ccorrespond to populations sampled previously (Baimai et al.1988b; Rattanarithikul et al. 1996) but is not meant to imply thatthese are the only populations of this species.


A N O P H E L E S D I R U S

M O S Q U I TO P O P U L AT I O N S T R U C T U R E

571


Molecular Ecology

, 10, 569–580

D. The excess of rare mutations and smooth unimodal mis-match distributions that were observed (underlying thestar-like genealogy in Fig. 2) were thought to indicate ahistory of population expansion, either a mitochondrialselective sweep or a demographic expansion. This couldexplain the observed intraspecific homogeneity in

An. dirus

A and D without any need to invoke ongoing gene flow.However, interpretation of the mtDNA data was compro-mised to some extent by the extreme level of homoplasyobserved (Fig. 2) as this can give some of the same patternsas population expansion. Although usually attributedto site hypervariability, it has been suggested that someof this homoplasy could also be due to mitochondrialrecombination (Walton

et al

. 2000b). An alternativeexplanation, that the pattern of polymorphism in

An. dirus

A and D is due to background selection (Charlesworth

et al

. 1993), cannot be rejected.In this study, microsatellite analysis has been applied to

populations of

An. dirus

A, C and D from Thailand toaddress the issues of gene flow within each species andmtDNA introgression between species A and D. This hasenabled a comparison between the mitochondrial and

nuclear markers. The comparison may allow a distinc-tion to be made between demographic and mtDNAexpansions and the higher mutation rate of microsatellitesis expected to resolve more recent population historythan mtDNA.

Materials and methods

Sample collection and identification

The samples were collected in 1995 and 1996 from sevensites distributed around Thailand (Fig. 1). Two of the siteshad sympatric populations (species A and D at site 4 andspecies C and D at site 5) making a total of nine sampledpopulations (four of

Anopheles dirus

species A, three ofspecies D and two of species C). To enable easy comparisonwith a previous study of the same populations usingmtDNA (Walton

et al

. 2000b), the same site numbers arealso used here. Sites 1, 2, and 3 in Bangladesh andMyanmar and site 11 from Thailand could not be includedin the microsatellite analysis due to limited sample sizes.Our objective was to obtain 30 individuals from each

Fig. 2 Haplotype network of Anophelesdirus A (white circles) and D (black circles)showing individually numbered haplotypesconnected by lines representing mutations.Grey circles indicate haplotypes sharedbetween species and 0 indicates inferredmissing haplotypes. The excessive homoplasyis illustrated by the numbers 1–5 along thelines indicating homoplasic substitutions forfive particular mutational events. Reprintedwith permission from Molecular Biologyand Evolution.


572

C . WA LTO N

E T A L

.


Molecular Ecology

, 10, 569–580

population but this was not achieved for two populationsof species A at site 4 (sample size = 21) and at site 8 (samplesize = 26) or for species D at site 6 (sample size = 10).Because all the sampled individuals were collected as adultbiting females it is reasonable to assume that they areunrelated. Individual mosquitoes were preserved bydesiccation over silica gel and their DNA extracted using a‘salting-out’ protocol (Sunnucks & Hales 1996; Walton

et al

.1999a). These three species cannot be separated reliably bymorphology and so were identified using allele-specificpolymerase chain reaction (PCR) of the ITS2 region(Walton

et al

. 1999a). The strong correlation between ITS2sequence and the polytene chromosome banding patternsthat were first used to define the species, make this areliable diagnostic tool to use.

Microsatellite genotyping

The 11 microsatellite markers used here were developedin

An. dirus

A and the primers and amplification conditionswere as described in Walton

et al

. (2000a). Some of the lociwere isolated from the same lambda phage clone and aretherefore physically linked within 20 kb of each other.There is a mixture of CAG and CA-repeat microsatellites.Having the same clone number at the start of the micro-satellite name indicates linked microsatellites. Loci wereisolated in this way to investigate the possibility thatlinkage disequilibrium between linked loci might provideadditional information in studies of population structure.Microsatellite analysis was performed on an ABI 373automated sequencer (Applied Biosystems). Typically, fiveor six fluorescently labelled microsatellite products wererun per lane together with internal size standards. In addi-tion, a panel of microsatellite products of differing sizes wasrun on three lanes distributed across each gel to ensure thatsize calling was consistent both within and between gels.This also served as a check on correct lane tracking. Theindividuals were genotyped using

genotyper

software(A.B.I. 1998).

Statistical analysis

The

arlequin

1.1 software package (Schneider

et al

. 1997)was used for the following analyses. Expected hetero-zygosity (

H

E

) was calculated according to Nei (1987)and an exact test of Hardy–Weinberg (HW) equilibriumwas performed using the Markov-chain algorithm of Guo& Thompson (1992). Tests of linkage disequilibrium wereperformed between all pairs of loci within each popula-tion using a likelihood ratio test in which the likelihood ofthe data assuming linkage equilibrium is compared to thelikelihood of the data when that assumption is relaxed. Thelatter quantity is computed from estimated haplotypefrequencies because the gametic phase is unknown.

F

ST

was estimated from allele frequencies using a conven-tional analysis of variance framework (Weir & Cockerham1984) in

arlequin

. An analogue of

F

ST

, Slatkin’s (1995)

R

ST

,assumes a stepwise mutation model which may be moreappropriate for microsatellites, and therefore takes intoaccount length differences between alleles as well as allelefrequency differences. An unbiased estimate of

R

ST

,

ρ

, wascalculated using

R

ST

-CALC (Goodman 1997) in which theoverall

R

ST

estimates were based on averaging variance com-ponents across loci, rather than averaging across individuallocus estimates of

R

ST

. This method also accommodatesdifferent sample sizes. Ohta’s (1982)

D

IT2

was calculatedin

popgene

(Yeh & Boyle 1997). This is the total variance oflinkage disequilibrium in a structured population. It isexpected to decline as gene flow between populationsincreases. It was calculated for populations of each speciesand also for the whole sample, treating species as populations.

The following tests were made at the web page of J.Brzustowski (http://www.biology.ualberta.ca/jbrzusto/).Assignment tests (Paetkau

et al

. 1995) were performed inwhich an individual is assigned to a population fromwhich its 11-locus genotype is most likely to have beensampled based on the allele frequency distributions of eachlocus in each population. To avoid bias, the allele frequen-cies of the population that an individual was sampled fromwere calculated excluding that individual and wherenecessary absent alleles were given a frequency of 0.01to avoid zero values. These likelihood values were usedto compute the genotype likelihood ratio distance (

D

LR

;Paetkau

et al

. 1997) for all pairs of populations. This dis-tance measure was used because it has been shown to havegood linearity and relatively low variance when appliedto fine scale population structure. The distance matrixproduced was used to construct a neighbour-joining tree(Saitou & Nei 1987) which was subsequently visualizedusing

treeview

1.40 (Page 1996).A multilocus test for population expansion was carried

out using the imbalance index,

β

, of Kimmel

et al

. (1998)which assumes a stepwise mutation model. The index isthe ratio of two estimators of

θ

, one based on the varianceof allele sizes averaged across loci and the other based onthe heterozygosity averaged across loci. The populationparameter

θ

= 4

N

µ

, where

N

is the effective population sizeand

µ

is the mutation rate. A value of

β

> 1 indicates popu-lation expansion preceded by a period of reduced popu-lation size.

Results

Within population genetic diversity

In total, 237 individuals were genotyped for 11 micro-satellite loci. There were 16 nonamplifications in total, fiveof which are thought to be due to poor quality samples of


A N O P H E L E S D I R U S

M O S Q U I TO P O P U L AT I O N S T R U C T U R E

573


Molecular Ecology

, 10, 569–580

species A and D from site 4. Although the microsatelliteswere isolated in

Anopheles dirus

A they appear to besimilarly polymorphic in species C and D (Table 1) withaverage expected heterozygosities of 0.518, 0.486 and

0.536 in species A, C and D, respectively. However, not allthe microsatellites are polymorphic across all species orin some cases across all populations within a species.Genotypes are available on request.

Table 1 Within-population diversity for all populations at all loci

Microsatellite locus

Population 6A10 6A11L 6A11S 7R11L 7R5 4A11L 4A10 11R11 13H11 3R5 9R11 Mean

A4NA 1 9 8 9 4 16 3 11 14 1 4 7.27HE — 0.756 0.793 0.818 0.253 0.921 0.138 0.662 0.805 — 0.619 0.524HO — 0.571* 0.524* 0.750 0.264 0.667* 0.143 0.619 0.600* — 0.429* 0.415PHW — 0.035 0.011 0.373 1.000 0.001 1.000 0.756 0.002 — 0.010A8NA 5 5 6 11 4 23 4 16 6 1 4 7.73HE 0.149 0.775 0.722 0.851 0.529 0.953 0.113 0.761 0.716 — 0.465 0.549HO 0.154 0.769 0.692 0.769* 0.538 0.846 0.115 0.692 0.500 — 0.346 0.493PHW 1.000 0.320 0.804 0.019 1.000 0.095 1.000 0.623 0.150 — 0.162A9NA 3 6 5 11 4 22 3 14 5 1 5 7.18HE 0.066 0.776 0.712 0.841 0.397 0.947 0.066 0.611 0.603 — 0.245 0.479HO 0.067 0.767 0.467* 0.733 0.367 0.767* 0.067 0.483* 0.433* — 0.200 0.396PHW 1.000 0.918 0.023 0.325 0.306 0.001 1.000 0.017 0.001 — 0.427A10NA 1 5 7 13 4 20 4 19 6 1 5 7.73HE — 0.723 0.749 0.827 0.319 0.951 0.159 0.726 0.672 — 0.571 0.518HO — 0.700 0.833 0.800 0.367 0.867 0.100 0.667 0.500* — 0.600 0.494PHW — 0.779 0.468 0.058 1.000 0.060 0.148 0.614 0.047 — 0.551C5NA 1 10 5 15 3 17 1 12 4 2 2 6.55HE — 0.850 0.219 0.9706 0.505 0.819 — 0.854 0.531 0.427 0.033 0.474HO — 0.833 0.200 0.800 0.567 0.633* — 0.433* 0.333* 0.533 0.033 0.397PHW — 0.387 0.260 0.073 0.605 0.045 — 0.000 0.003 0.216 1.000C7NA 1 7 11 14 3 16 2 17 6 1 1 7.182HE — 0.636 0.867 0.915 0.474 0.915 0.155 0.850 0.659 — — 0.497HO — 0.633 0.933 0.900 0.300* 0.700* 0.167 0.586* 0.533 — — 0.432PHW — 0.208 0.256 0.735 0.033 0.000 1.000 0.001 0.115 — —D4NA 3 4 7 3 6 16 5 4 16 3 3 6.364HE 0.671 0.371 0.818 0.098 0.161 0.915 0.299 0.461 0.871 0.239 0.657 0.506HO 0.667 0.333 0.821 0.100 0.167 0.862 0.133* 0.433* 0.667* 0.267 0.600 0.459PHW 0.394 0.461 0.709 1.000 1.000 0.372 0.001 0.019 0.000 1.000 0.889D5NA 3 6 8 2 6 14 2 5 18 1 6 6.45HE 0.671 0.610 0.829 0.097 0.277 0.903 0.259 0.465 0.935 — 0.644 0.517HO 0.433* 0.467 0.767 0.100 0.267 0.800 0.300 0.500 0.458* — 0.800 0.445PHW 0.039 0.341 0.137 1.000 0.530 0.204 1.000 0.279 0.000 — 0.389D6NA 3 4 8 1 2 11 2 4 7 2 4 4.364HE 0.668 0.668 0.879 — 0.479 0.921 0.190 0.605 0.879 0.395 0.753 0.585HO 0.300* 0.500 1.000 — 0.100* 0.800* 0.000 0.400 0.200* 0.500 0.400* 0.382PHW 0.032 0.405 0.152 — 0.022 0.009 0.052 0.226 0.000 1.000 0.007

Number of alleles detected (NA), expected and observed heterozygosity (HE, HO), exact probability of Hardy—Weinberg equilibrium (PHW). An asterisk denotes HO values where PHW is <0.05.

MEC1201.fm 73 Monday, March 12, 2001 6:15 PM

574 C . WA LTO N E T A L .

© 2001 Blackwell Science Ltd, Molecular Ecology, 10, 569–580

Tests of disequilibrium

Exact tests of HW equilibrium were performed for the 86polymorphic population/locus combinations (Table 1).For this number of tests only four or five rejections of HWequilibrium would be expected due to Type I error (withP < 0.05) yet there were 27 population/locus combinationswith an excess of homozygotes (sometimes with probabilitylevels much lower than 0.05). The distribution across loci isunequal with the greatest deficiencies of heterozygotesoccurring in loci 4A11L, 11R11 and 13H11. Although thismay sometimes be due to the nondetection of a muchlarger second allele in a pair (Walton et al. 2000a) it alsoimplies the presence of null alleles. Null alleles are alsoindicated by the 11 nonamplifications that are not due topoor sample quality. All of these cases were associatedwith strong deviations from HW equilibrium: two non-amplifications at locus 11R11 (populations A9 and C7); andnine nonamplifications at locus 13H11 (6 in population D5and 3 in D6).

The levels of disequilibrium amongst loci from clones 4,6 and 7, within populations, are shown in Table 2. Only theloci from clone 6 exhibit significant linkage disequilibrium.One of the loci from clone 4, 4A10, is not very variable(Table 1) which will reduce the ability to detect linkagedisequilibrium. However, why no significant linkage dis-equilibrium was detected amongst the loci from clone 7is not known. Conversely, amongst the loci isolated from

independent clones 25 out of 336 tests were positive (7.44%relative to the chance expectation of 5%). In 15 of the 25positive tests there was evidence of a significant deficiencyof heterozygotes in at least one of the pair of loci beingcompared in that population. Some of the positive resultscould, therefore, have been due to departure from HWequilibrium, which the test assumes (Excoffier & Slatkin1998).

Within and between species population differentiation

Pairwise comparisons of population differentiation (RSTand FST) are shown in Table 3. In a regression of RST valueson FST values the adjusted r2 = 0.81. The regressioncoefficient of 1.47 indicates that RST gets proportionatelylarger for increased values of FST. The C5 population standsout as being very highly differentiated from all of the otherpopulations even including the comparison with theother population of An. dirus C, population C7. All of theinterspecies comparisons have high RST and FST valuesand within species comparisons are always lower. This isalso true for An. dirus C even though the level of geneticdifferentiation between the two C populations is notmuch lower than the levels of interspecific differentiationobserved. Within species A, there is very little if anydifferentiation amongst populations. By contrast, withinspecies D all the pairwise comparisons indicate that thereis significant genetic structure, albeit at a relatively low

Pairwise comparisons of loci individually significant at P < 0.05 (maximum number of possible tests per population)

Within clone 4 Within clone 6 Within clone 7 All othersPopulations (1) (3) (exact P*) (1) (50)

A4 0 2 (0.002) 0 2 (0.005)

A8 0 1 (0.004) 0 2A9 0 0 0 3A10 0 0 0 3C5 0 1 (0.007) 0 1C7 0 1 (0.007) 0 3D4 0 2 (0.032) 0 6

(0.006)D5 0 3 (<0.001) 0 2

(0.005) (0.030)

D6 0 1 (<0.001) 0 3

Total significant (from total tests)

0 (from 8†) 11 (from 19†) 0 (from 8†) 25 (from 336†)

*Exact P for each significant pairwise comparison.†Indicates the total number of tests excluding those involving monomorphic loci.

Table 2 Comparisons of linkage disequilibriabetween loci, within populations


A N O P H E L E S D I R U S M O S Q U I TO P O P U L AT I O N S T R U C T U R E 575


level of differentiation. Both RST and FST values indicatethat population D6 is the most divergent of the threepopulations.

In the assignment tests (Table 4), all individuals werecorrectly assigned to species. This reconfirms the use ofITS2 as a species diagnostic marker, although the possibilityof introgression of ITS2 is one that must always be borne inmind. Within species, all individuals of species C and 73%of species D were correctly assigned to population. InAn. dirus A, 31% were correctly assigned, only slightlyabove the ~25% level of random assignment. It should benoted that the assignment method assumes HW and linkageequilibrium within each population since probabilities ofeach genotype are derived from allele frequencies and

multiplied across loci. Although the assignment probabilitiesfor an individual will be influenced by the lack of independ-ence between loci, this will apply similarly to the estimatedprobabilities in the population an individual was sampledfrom and in the population that is being compared. Ineffect, the assignment test is ignoring the informationpresent in the associations between alleles at different lociand will, if anything, underestimate genetic distance. Fur-ther, only clone 6 showed significant linkage disequilib-rium, so any overall effect resulting from the equilibriumassumption will not be large. The unrooted neighbour-joining DLR distance tree of all the populations is shown inFig. 3.

Examining the variance in linkage disequilibrium is oneway to reveal the information about genetic structurewithin and between species revealed by differences in thedistributions of multilocus genotypes. DIT

2 values (Table 5)show a large degree of differentiation between the twopopulations of An. dirus C and more structure in An. dirusD than A, consistent with the pattern of population struc-ture revealed by FST and RST. However, there is apparentlyno additional information being gained from the linkedloci because the same signature in DIT

2 is found with bothlinked and unlinked loci. This is, in fact, expected as DIT

2

has been shown to be sensitive to the numbers of migrantsbut not sensitive to the recombination rate (Ohta 1982).

The mean FST values estimated from the microsatel-lite loci for within and among group comparisons areshown in Fig. 4 relative to similar estimates of FST madefrom mtDNA sequence data (Walton et al. 2000b). The

Table 3 Pairwise comparison of population differentiation

A4 A8 A9 A10 C5 C7 D4 D5 D6

A4 0.011 0.020* 0.001 0.284** 0.207** 0.229** 0.219** 0.212**A8 –0.0035 0.002 –0.003 0.284** 0.193** 0.224** 0.218** 0.187**

–0.007–0.061A9 0.0008 –0.006 0.005 0.326** 0.246** 0.258** 0.258** 0.245**

–0.001–0.060 –0.008–0.048A10 0.003 – 0.009 0.011 0.302** 0.209** 0.230** 0.225** 0.210**

– 0.002–0.068 – 0.009–0.039 0.000–0.065C5 0.510 0.506 0.525 0.544 0.188** 0.382** 0.383** 0.375**

0.444–0.587 0.447–0.583 0.467– 0.595 0.481–0.612C7 0.269 0.252 0.311 0.279 0.401 0.322** 0.310** 0.287**

0.220–0.356 0.212–0.320 0.273–0.375 0.229–0.353 0.332–0.495D4 0.191 0.215 0.230 0.242 0.534 0.384 0.009 0.075**

0.157–0.272 0.181–0.285 0.193–0.305 0.198–0.323 0.477–0.607 0.334–0.464D5 0.227 0.270 0.280 0.297 0.559 0.433 0.070 0.072**

0.174–0.330 0.215–0.359 0.232–0.363 0.243–0.384 0.508–0.624 0.388–0.502 0.025–0.172D6 0.365 0.406 0.430 0.423 0.660 0.552 0.220 0.120

0.288–0.508 0.346–0.531 0.373–0.550 0.358–0.551 0.616–0.739 0.499–0.658 0.138–0.379 0.076–0.262

RST estimates are shown below the diagonal with 95% confidence intervals and FST estimates above the diagonal with significance levels indicated by *P < 0.05, **P < < 0.001.

Table 4 Pairwise population assignments showing the number ofindividuals from each column population that has been assignedto each row population

Population A4 A8 A9 A10 C5 C7 D4 D5 D6

A4 8 1 5 7A8 1 6 8 11A9 1 10 12 7A10 8 3 12 7C5 30 0C7 0 30D4 24 5 1D5 9 20 1D6 3 0 7




mtDNA-based FST estimates are generally higher that thoseestimated from microsatellite data. Notable exceptionsto this are the comparisons within species D and betweenspecies A and D.

Population expansion in An. dirus A and D

The imbalance index, β, is greater than one for both An. dirusA (β = 3.8) and An. dirus D (β = 3.4). The majority of loci arecontributing to these values because in individual locustests all 11 loci in species D and eight of the 10 variable lociin An. dirus A had values greater than one. Although this isconsistent with a population expansion following a periodof reduced genetic diversity there could be other causes forthe imbalance between variance and heterozygosity.

Discussion

Microsatellites

As is generally found, the microsatellites used in thisstudy are highly polymorphic (Jarne & Lagoda 1996) andtherefore potentially useful in addressing the issue ofpopulation genetic structure. The relatively high incidenceof null alleles found in this study is similar to that reportedfrom several microsatellite studies in the African vectors

Fig. 3 The unrooted neighbour-joining DLR distance tree of allpopulations of Anopheles dirus A, C and D. A DLR distance of twoindicates that, on average, the genotypes of individuals from the twopopulations being compared are two orders of magnitude morelikely to occur in their own population than in the other population.

Fig. 4 Comparison of within and amonggroup FST estimates from microsatellites(averaged across all loci) with FST estimatesfrom mtDNA cytochrome oxidase I sequencedata (from Walton et al. 2000b). For bothsets of estimates in Anopheles dirus A and D,a group corresponds to the species whereasin species C each population is classified asa group. Note that, in the mtDNA study,there were two additional populations ofspecies A and three of species D as indicatedin Fig. 1. All values are significant to at leastthe 0.01 level except within An. dirus A(both markers) and within An. dirus D(mtDNA only).

Table 5 Within and between species comparisons of DIT2

Loci

Clone 4 Clone 6 Clone 7 Others

Within A 0.0218 0.0146 0.0131 0.0181Within C 0.0546 0.1008 0.0172 0.0771Within D 0.0192 0.0269 0.1718 0.0491Between species 0.0218 0.0909 0.1195 0.1479




Anopheles arabiensis and An. gambiae (Kamau et al. 1999 andreferences therein; Donnelly et al. 1999). It is possible thatorganisms such as mosquitoes that have large populationsizes with correspondingly high levels of polymorphismmay be more prone to harbour null alleles than organismswith smaller population sizes. Microsatellites have beenisolated in at least two other Anopheles species, An. maculatus(Rongnoparut et al. 1996) and An. funestus (Sinkinset al. 2000). Evidence for the presence of null alleles inAn. maculatus is mixed (Rongnoparut et al. 1996, 1999) andin An. funestus the relevant data have not yet been reported.

There is much debate on the most appropriate statistic todetect population structure for microsatellites ( Jarne &Lagoda 1996). Essentially, should measures of differentia-tion be based on the variance in allele frequencies (assum-ing an infinite alleles model) or the variance in allele repeatnumber (assuming a stepwise mutation model)? If a step-wise mutation model does apply to microsatellites then RSTvalues are predicted to be larger than FST values (Slatkin1995). This was found to be the case here with the regres-sion coefficient of RST regressed onto FST being greater thanone (b = 1.47). The high correlation coefficient (adjustedr2 = 0.81) indicates that the overall pattern of genetic differ-entiation revealed by both statistics (estimated over all 11loci) is the same. However, there are differences in whichindividual loci contribute the most to the overall FST andRST estimates (data not shown). This was also found byDonnelly et al. (1999) within An. arabiensis from Africa. Thismay reflect between-locus differences in the relative effectsof mutation and drift. However, the overall agreementbetween FST and RST values noted above indicates thatthere are sufficient loci to average such differences out.Therefore, only the global FST and RST values are discussedbelow.

Between species relationships and introgression

In the between-species comparisons (Table 3 and Fig. 4)both RST and FST values show marked differentiationbetween all three species of An. dirus. The neighbour-joining tree based on DLR distances also showed the threespecies being well differentiated from each other. This is inmarked contrast to previous findings with mtDNA wherethere was almost no differentiation between An. dirus Aand D (Fig. 4 and Walton et al. 2000b). Although themtDNA pattern could be due to the retention of ancestralpolymorphism in large populations, this explanationseems unlikely given that the haplotype network of An.dirus C mtDNA was separated from that of the combinedspecies A-D network by seven fixed differences. The cleardifferentiation of species A, C and D at microsatellite loci,therefore, reinforces the previous suggestion (Walton et al.2000b) that mtDNA has introgressed between An. dirus Aand D. Because species A and D have only a limited range

of overlap and their hybrids are very unfit (Baimai et al.1987), contemporary introgression on a large geographicalscale is unlikely. Analysis of the cytochrome oxidase I(COI) mtDNA sequence data suggested populationexpansion (selective sweep or demographic expansion) ofAn. dirus D followed subsequently by expansion of An.dirus A (Walton et al. 2000b). The observed pattern ofmtDNA variation is, therefore, more likely to be due toeither: (i) historical introgression predating demographicexpansion when both species shared a limited distribution;or (ii) a selective sweep of mtDNA that began in species Dand subsequently crossed into species A. On the basis ofthe imbalance index, the microsatellite data appear tosupport the possibility of demographic expansion. Aselective sweep had previously been thought to be unlikelybecause the expansion appears not to come from a singlemitochondrial haplotype (Fig. 2). However, the high levelof homoplasy in the mtDNA data could remove thisobjection to a selective sweep argument.

A very close relationship between An. dirus A and C hasbeen suggested based primarily on cross-mating andcytogenetic studies (see Introduction). However, this wasnot observed with the microsatellite loci (Fig. 3) where,although species C and D are the most distant from eachother, they are both at a considerable distance from speciesA. On the other hand, sequence data from another nuclearlocus, ITS2 (Walton et al. 1999a), do indicate a close speciesA-C relationship, as expected. Differential introgressionamong sections of the nuclear genome is a possibility thatshould be considered in these species. Between An. dirus Aand C in particular, nuclear introgression might be facili-tated by their relatively fertile hybrids which exhibit a highdegree of chromosome synapsis (>90%) and the presenceof chiasmata (Baimai et al. 1987). In the comparable Africanmalaria vectors, An. gambiae and An. arabiensis, there is nowconsiderable evidence to suggest the exchange of geneticinformation between these species (Besansky et al. 1994;García et al. 1996; della Torre et al. 1997; Kamau et al. 1998).However, when dealing with such closely related species itis very difficult to differentiate introgression from theeffects of stochastic lineage sorting (Avise 1994). Becausemicrosatellites are putatively neutral and distributedthroughout the genome (Jarne & Lagoda 1996), they mightbe expected to provide a better hypothesis of the evolution-ary relationships between the species. However, there isconcern that microsatellite-based distance measures arenot useful even at this close level of species relationship.For example, Paetkau et al. (1997) found no microsatellitedistance measures (including DLR used here) that werecapable of resolving the sister species relationship betweenbrown and polar bears in relation to the much more dis-tantly related black bear. This is most probably due to con-straints on the allele size of microsatellites which causesgenetic distance measures to plateau even after relatively




low divergence times (Nauta & Weissing 1996; Feldmanet al. 1997). Clearly, information from additional nonmicro-satellite nuclear loci is required to help address this issue.

Within species genetic structure

The population that stands out as being the most differentfrom all of the others (for both RST and FST) is C5, thepopulation of An. dirus C from northern Thailand. Ininterspecific comparisons the level of differentiationmeasured by RST is always greater than 0.5. The two Cpopulations are also very different from each other(RST = 0.401, FST = 0.188) in agreement with previousmtDNA data (Fig. 3). In addition, in a genealogy of themtDNA sequences C5 was paraphyletic with respect to C7and no contemporary gene flow was indicated betweenthem (Walton et al. 2000b). In another mosquito species,An. maculatus, several populations from a north–southtransect in Thailand (from approximately site 4 in north-west Thailand to beyond site 7 in Fig. 1) have beencompared using microsatellites (Rongnoparut et al. 1999).Northern and southern populations were found to formtwo genotypic clusters and it was suggested that mountainranges could present a geographical barrier to gene flow.Despite the similarity between the two species, the overalllevel of genetic structure in An. maculatus (FST = 0.0406)was considerably less extreme than that observed inAn. dirus C. The low level of diversity of mtDNA sequencedata observed in population C5 relative to that in popula-tions C7 and all populations of species A and D, indicatesa period of low effective population size for C5 (Waltonet al. 2000b). The same pattern is not seen in the levelof polymorphism at the microsatellite loci with allpopulations of all species having comparable levels ofexpected heterozygosity (Table 1). However, the relativelylong branch length leading to C5 in Fig. 2 is consistent withthis population having been isolated for some time andexperiencing a period of independent evolution.

In the cross-mating studies that helped to establish spe-cies status (Baimai et al. 1987), intraspecific crosses betweennorthern and southern C populations were not performedand the allozyme studies (Green et al. 1992) only includedpopulations from southern Thailand. The status of An.dirus C as a single, distinct species therefore rests primarilyon polytene chromosome banding patterns (Baimai et al.1988c). The genetic data from both mtDNA and micro-satellites indicate that what is currently known as An. dirusC may actually be more than one species. Genetic data onthe intervening populations known to exist (Fig. 1) wouldbe very helpful in clarifying this situation.

The results in An. dirus C clearly indicate the ability ofthe microsatellites to detect genetic population structure ifit exists. However, the pairwise comparisons of RST and FST(Table 3) and the overall FST value (Fig. 4) indicate that

within species A from Thailand there is essentially nodetectable population structure. This is in complete agree-ment with the results of mtDNA sequence analysis of thesame and other populations from Thailand (Fig. 4). In An.dirus D, the FST value from previous analysis of mtDNAsequence data (Fig. 4), which included populations fromBangladesh and Myanmar in addition to those used here,also indicated no population structure. In contrast to this,the microsatellite FST value indicates a significant, albeitrelatively low, level of population structure even thoughthis has been detected over a much smaller geographicalarea confined to Thailand (Fig. 1). Population D6 appearsto be the most divergent (Table 3). If this was due in part toits small sample size (10 individuals), then this could biasthe overall FST within species D (Fig. 3). This is unlikely tobe a serious problem since the sample size of loci is highand at least three of the 11 loci contribute to the signal ofgenetic structure. Even if population D6 is excluded, thepopulation pairwise comparison of D4 and D5 still showsa significant level of differentiation from RST (Table 3)(although FST is not significant, P = 0.09). Furthermore, inthe assignment test (Table 4) the majority of individualsfrom populations D4 and D5, as well as those from D6, arecorrectly assigned.

In both species A and D the lack of genetic structuredetected by mtDNA was thought most probably to be thesignature of a population expansion (see Introduction;Walton et al. 2000b). Whilst this does not preclude thepossibility of ongoing gene flow, it does mean that the levelof gene flow cannot be inferred reliably from the F-statisticsbecause the populations are unlikely to be at migration-drift equilibrium (particularly given the large effectivepopulation sizes). The significantly lower levels of mtDNAnucleotide diversity observed in populations of An. dirus Arelative to populations of species D indicate that expansionoccurred first in An. dirus D and subsequently in species A.It is difficult from these data alone to distinguish a demo-graphic expansion from an expansion of a particularfavoured mtDNA haplotype, that is, a selective sweep. Thefact that the imbalance index, β, was greater than one forboth species, is more consistent with demographic expan-sion. However, it is not clear how robust this conclusion is,given the sample size, or whether the statistic adequatelydiscriminates between population expansion and otherfactors such as constraints on allele size or populationstructure within species. The levels of diversity at nuclearloci relative to mtDNA diversity offer another means oftesting these alternative hypotheses. The similar, highlevels of microsatellite diversity observed in An. dirus A andD (Table 1) are more consistent with a selective sweephypothesis because demographic expansion should resultin lower nuclear as well as mtDNA diversity in species A.However, this is not a powerful test because the highermutation rate of microsatellites than of mtDNA may allow




them to reach equilibrium levels of diversity more quickly.In other words, they may simply be revealing more recentpopulation history. The microsatellite mutation rate in An.dirus is unknown but it is likely to be similar to that measuredin Drosophila (6.3 × 10–6 per generation Schug et al. 1997).This is in contrast to the much higher estimates of micro-satellite mutation rate that have been made in other organisms(for example in humans, ~10–3 per locus per generation,Weber & Wong 1993). However, the observation that popula-tion C5 has high microsatellite diversity yet low mtDNAdiversity indicates that there may well have been sufficienttime to recover microsatellite diversity in An. dirus A.

If there was a selective sweep of mtDNA, then the micro-satellite data imply a real difference in the relative rates ofongoing gene flow occurring in An. dirus A and D. Altern-atively, if there has been demographic expansion, there isunlikely to have been time for migration-drift equilibriumto be established at microsatellite loci due to the largeeffective population sizes limiting drift. In this scenario,the significant population structure detected by micro-satellites in An. dirus D would be consistent with geneticdifferentiation having had more time to accumulatepostexpansion in An. dirus D than in species A. Althoughthis indicates that barriers to gene flow exist in species D,their strength cannot be determined and similar barriersmight also exist in species A.

Conclusions

The inconsistencies observed among mitochondrial,microsatellite and other nuclear markers make an under-standing of the evolutionary history of these mosquitoeselusive. This is of evolutionary interest in understandingthe nature of species boundaries and genetic exchange. Inaddition, inferring population history is also of practicalrelevance to the control of malaria because any reasonableattempt to address the question of contemporary gene flowin these mosquitoes will first require a clear understandingof their evolutionary history to enable confounding effectsto be distinguished.

Acknowledgements

This work was funded by a Wellcome Trust fellowship to C. Waltonwith the sponsorship of R.K. Butlin. We would also like to thankMr Sawadwangporn, Mr Somsak, Mr Chow and the regionalMalaria Offices around Thailand for providing information andassistance with mosquito collections.

References

A.B.I. (1998) GENOTYPER 2.0. ABI, Foster City, USA.Avise JC (1994) Molecular markers, natural history and evolution,

pp. 126–138. Chapman & Hall, New York.Baimai V, Andre RG, Harrison BA, Kijchalao U, Panthusiri L

(1987) Crossing and chromosomal evidence for two additional

sibling species within the taxon Anopheles dirus Peyton andHarrison (Diptera: Culicidae) in Thailand. Proceedings of theEntomological Society of Washington, 89, 157–166.

Baimai V, Harbach RE, Kijchalao U (1988a) Cytogenetic evidenceof a fifth species within the taxon Anopheles dirus (Diptera:Culicidae) in Thailand. Journal of the American Mosquito ControlAssociation, 4, 333–338.

Baimai V, Kijchalao U, Sawadwongporn P, Green CA (1988b)Geographic distribution and biting behaviour of four species ofthe Anoheles dirus complex (Diptera: Culicidae) in Thailand.Southeast Asian Journal of Tropical Medicine and Public Health, 19,151–161.

Baimai V, Poopittayasataporn A, Kijchalao U (1988c) Cytologicaldifferences and chromosomal rearrangements in four membersof the Anopheles dirus complex (Diptera: Culicidae) in Thailand.Genome, 30, 372–379.

Besansky NJ, Powell JR, Caccone A, Hamm DM, Scott JA, Collins FH(1994) Molecular phylogeny of the Anopheles gambiae complexsuggests genetic introgression between principal malaria vectors.Proceedings of the National Academy of Sciences of the USA, 91,6885–6888.

Charlesworth B, Morgan MT, Charlesworth D (1993) The effect ofdeleterious mutations on neutral molecular variation. Genetics,134, 1289–1303.

Collins FH (1994) Prospects for malaria control through the geneticmanipulation of its vectors. Parasitology Today, 10, 370–371.

Donnelly MJ, Cuamba N, Charlwood JD, Collins FH, Townson H(1999) Population structure in the malaria vector, Anopheles ara-biensis Patton, in East Africa. Heredity, 83, 408–417.

Excoffier L, Slatkin M (1998) Incorporating genotypes of relativesinto a test of linkage disequilibrium. American Journal of HumanGenetics, 62, 171–180.

Feldman MW, Bergman A, Pollock DD, Goldstein DB (1997)Microsatellite genetic distances with range constraints: Analyticdescription and problems of estimation. Genetics, 145, 207–216.

García BA, Caccone A, Mathiopoulos KD, Powell JR (1996) Inver-sion monophyly in African Anopheline malaria vectors. Genetics,143, 1313–1320.

Goodman SJ (1997) RST-CALC: a collection of computer programsfor calculating estimates of genetics differentiation frommicrosatellite data and determining their significance. MolecularEcology, 6, 881–885.

Green CA, Munstermann LE, Tan SG, Panyim S, Baimai V (1992)Population genetic evidence for species A, B, C and D of theAnopheles dirus complex in Thailand and enzyme electromorphsfor their identification. Medical and Veterinary Entomology, 6,29–36.

Guo SW, Thompson EA (1992) Performing the exact test of Hardy–Weinberg proportion for multiple alleles. Biometrics, 48, 361–372.

Hemingway J, Ranson H (2000) Insecticide resistance in insectvectors of human disease. Annual Review of Entomology, 45, 371–391.

Jarne P, Lagoda PJL (1996) Microsatellites, from molecules topopulations and back. Trends in Ecology and Evolution, 11, 424–429.

Kamau L, Hawley WA, Lehmann T et al. (1998) Use of short tandemrepeats for the analysis of genetic variability in sympatricpopulations of Anopheles gambiae and Anopheles arabiensis. Heredity,80, 675–682.

Kamau L, Mukabana WR, Hawley WA et al. (1999) Analysis ofgenetic variability in Anopheles arabiensis and Anopheles gambiaeusing microsatellite loci. Insect Molecular Biology, 8, 287–297.




Kimmel M, Chakraborty R, King JP, Bamshad M, Watkins WS,Jorde LB (1998) Signatures of population expansion in micro-satellite repeat data. Genetics, 148, 1921–1930.

Nauta MJ, Weissing FJ (1996) Constraints on allele size at micro-satellite loci: implications for genetic differentiation. Genetics,143, 1021–1032.

Nei M (1987) Molecular Evolutionary Genetics. Columbia UniversityPress, New York.

Nichols RA, Beaumont MA (1996) Is it ancient or modern historythat we can read in our genes?. In: Aspects of the Genesis andMaintenance of Biological Diversity (eds Hochberg ME, BarbaultR), pp. 69–87. Oxford UP, Oxford.

Ohta T (1982) Linkage disequilibrium with the island model.Genetics, 101, 139–155.

Paetkau D, Calvert W, Stirling I, Strobeck C (1995) Microsatelliteanalysis of population structure in Canadian polar bears. Molec-ular Ecology, 4, 347–354.

Paetkau D, Waits LP, Clarkson PL, Craighead L, Strobeck C (1997)An empirical evaluation of genetic distance statistics usingmicrosatellite data from bear (Ursidae) populations. Genetics,147, 1943–1957.

Page RDM (1996) treeview: an application to display phylogenetictrees on personal computers. Computer Applications in the Bio-sciences, 12, 357–358.

Peyton EL (1989) A new classification for the Leucosphyrus Groupof. Anopheles (Cellia). Mosquito Systematics, 21, 197–205.

Peyton EL, Harrison BA (1980) Anopheles (Cellia) takasagoensisMorishita 1946, an additional species in the BalabacensisComplex of Southeast Asia (Diptera: Culicidae). MosquitoSystematics, 12, 335–347.

Peyton EL, Ramalingam S (1988) Anopheles (Cellia) nemophilous,a new species of the Leucosphyrus Group from peninsularMalaysia and Thailand (Diptera: Culicidae). Mosquito Systematics,20, 272–299.

Poopittayasataporn A, Baimai V (1995) Polytene chromosomerelationships of five species of the Anopheles dirus complex inThailand. Genome, 38, 426–434.

Rattanarithikul R, Linthicum KJ, Konishi E (1996) Seasonalabundance and parity rates of Anopheles species in southernThailand. Journal of the American Mosquito Control Association, 12,75–83.

Rongnoparut P, Sirichotpakorn N, Rattanarithikul R, Yaicharoen S,Linthicum KJ (1999) Estimates of gene flow among Anophelesmaculatus populations in Thailand using microsatellite analysis.American Journal of Tropical Medicine and Hygiene, 60, 508–515.

Rongnoparut P, Yaicharoen S, Sirichotpakorn N, Rattanarithikul R,Lanzaro GC, Linthicum KJ (1996) Microsatellite polymorphismin Anopheles maculatus, a malaria vector in Thailand. AmericanJournal of Tropical Medicine and Hygiene, 55, 589–594.

Rosenberg R (1982) Forest malaria in Bangladesh. III. Breedinghabitats of Anopheles dirus. American Journal of Tropical Medicineand Hygiene, 31, 192–201.

Rosenberg R, Andre RG, Somchit L (1990) Highly efficient dry sea-son transmission of malaria in Thailand. Transactions of the RoyalSociety of Tropical Medicine and Hygiene, 89, 22–28.

Saitou N, Nei M (1987) The neighbour joining method: a newmethod for constructing phylogenetic trees. Molecular Biologyand Evolution, 4, 406–425.

Sawadipanich Y, Baimai V, Harrison BA (1990) Anopheles dirusspecies E: chromosomal and crossing evidence for another

member of the Dirus Complex. Journal of the American MosquitoControl Association, 6, 477–481.

Scanlon JE, Sandinand U (1965) The distribution and biology ofAnopheles balabacensis in Thailand (Diptera: Culicidae). Journal ofMedical Entomology, 2, 61–69.

Schneider S, Kueffer J-M, Roessli D, Excoffier L (1997) ARLEQUIN 1.1.A software for population genetic data analysis. URL: http://anthropologie.unige.ch/arlequin.

Schug MD, Mackay TFC, Aquadro CF (1997) Low mutation rates ofmicrosatellite loci in Drosophila melanogaster. Nature Genetics, 15,99–102.

Sinkins SP, Hackett BJ, Costantini C et al. (2000) Isolation of poly-morphic microsatellite loci from the malaria vector Anophelesfunestus. Molecular Ecology, 9, 490–492.

Slatkin M (1995) A measure of population sub-division based onmicrosatellite allele frequencies. Genetics, 139, 457–462.

Sunnucks P, Hales DF (1996) Numerous transposed sequences ofmitochondrial cytochrome oxidase I-II in aphids of the genusSitobian (Hemiptera: Aphididae). Molecular Biology and Evoution,13, 510–524.

della Torre A, Merzagora L, Powell JR, Coluzzi M (1997) Selectiveintrogression of paracentric inversions between two sibling speciesof the Anopheles gambiae complex. Genetics, 146, 239–244.

Tun-Lin W, Thu MM, Than S, Mya MM (1995) Hyperendemicmalaria in a forested, hilly Myanmar village. Journal of theAmerican Mosquito Control Association, 11, 401–407.

Walton C, Chang MS, Handley JM et al. (2000a) The isolation andcharacterisation of microsatellites from Anopheles dirus mosqui-toes. Molecular Ecology, 9, 1665–1667.

Walton C, Handley JM, Kuvangkadilok C et al. (1999a) Identifica-tion of five species of the Anopheles dirus complex from Thailand,using allele-specific polymerase chain reaction. Medical andVeterinary Entomology, 13, 24–32.

Walton C, Handley JM, Tun-Lin W et al. (2000b) Populationstructure and population history in Anopheles dirus mosquitoes inSoutheast Asia. Molecular Biology and Evolution, 17, 962–974.

Walton C, Sharpe RG, Pritchard SJ, Thelwell NJ, Butlin RK (1999b)Molecular identification of mosquito species. Biological Journal ofthe Linnean Society, 68, 241–256.

Weber JL, Wong C (1993) Mutation of human short tandemrepeats. Human Molecular Genetics, 2, 1123–1128.

Weir BS, Cockerham CC (1984) Estimating F-statistics for theanalysis of population structure. Evolution, 38, 1358–1370.

Xu J-N, Qu F-Y (1997) Ribosomal DNA difference betweenspecies A and D of the Anopheles dirus complex of mosquitoesfrom China. Medical and Veterinary Entomology, 11, 134–138.

Yeh FC, Boyle TJB (1997) Population genetic analysis of co-dominant and dominant markers and quantitative traits.Belgian Journal of Botany, 129, 157.

This work presented here was performed as part of CW’s WellcomeTrust Fellowship in Biodiversity with molecular technical assistanceprovided by JH. The aim of the fellowship project was to use DNAsequence data and microsatellites to study population structure,gene flow, introgression and molecular systematics in Anophelesdirus mosquitoes. The fellowship involved the sponsorship of RKBand collaborations with VB, FHC, REH and VD. CW continues towork on mosquito genetic diversity in South-east Asia.


genetic population structure and introgression in anopheles dirus mosquitoes in south-east asia

Documents