Copyright ? 2006 by the Genetics Society of America
Divergence With Gene Flow in Anopheles funestus From the Sudan Savanna
of Burkina Faso, West Africa
Andrew P. Michel,* Olga Grushko,* Wamdaogo M. Guelbeogo,†Neil F. Lobo,* N’Fale Sagnon,†
Carlo Costantini‡and Nora J. Besansky*,1
*Center for Tropical Disease Research and Training, Department of Biological Sciences, University of Notre Dame, Notre Dame, Indiana, 46556,
†Centre National de Recherche et de Formation sur le Paludisme, Ouagadougou 01, Burkina Faso and
‡Institut de Recherche pour le De ´veloppement, Ouagadougou 01, Burkina Faso
Manuscript received April 19, 2006
Accepted for publication April 26, 2006
Anopheles funestus is a major vector of malaria across Africa. Understanding its complex and non-
equilibrium population genetic structure is an important challenge that must be overcome before vector
populations can be successfully perturbed for malaria control. Here we examine the role of chromosomal
inversions in structuring genetic variation and facilitating divergence in Burkina Faso, West Africa, where
two incipient species (chromosomal forms) of A. funestus, defined principally by rearrangements of chro-
mosome 3R, have been hypothesized. Sampling across an ?300-km east–west transect largely contained
within the Sudan–Savanna ecoclimatic zone, we analyzed chromosomal inversions, 16 microsatellite loci
distributed genomewide, and 834 bp of the mtDNA ND5 gene. Both molecular markers revealed high
genetic diversity, nearly all of which was accounted for by within-population differences among individuals,
owing to recent population expansion. Across the study area there was no correlation between genetic
and geographic distance. Significant genetic differentiation found between chromosomal forms on the
basis of microsatellites was not genomewide but could be explained by chromosome 3R alone on the basis
of loci inside and near inversions. These data are not compatible with complete reproductive isolation but
are consistent with differential introgression and sympatric divergence between the chromosomal forms,
facilitated by chromosome 3R inversions.
specific cause of morbidity and mortality in children
from tropical Africa (World Health Organization
2003). Among its principal African vectors are two
highly anthropophilic mosquito species, Anopheles gam-
biae and A. funestus, whose population genetic structure
shows remarkable parallels (Lehmann et al. 2003;
Michel et al. 2005b). The tendency of both species to
specialize on humans not only maintains a high malaria
transmission intensity but also has shaped their popula-
tion history. This is evidenced byvery shallow population
structure across Africa and signs of recent population
expansion inresponse tohuman populationgrowthtrig-
gered by the rise and spread of agriculture. The shallow
and nonequilibrium population genetic structure of
these mosquitoes represents a major challenge to under-
standing contemporary rates and patterns of gene flow
but one that must be overcome if vector populations are
and Fontenille 2004).
ALARIA continues to claim over 1 million lives
annually and remains the leading pathogen-
Against the backdrop of weak population genetic
structure inferred from neutral molecular markers,
nonrandom patterns of chromosomal inversion poly-
morphism in West Africa suggest an ongoing speciation
process in both vectors, associated with environmental
changes, including irrigation schemes for rice cultiva-
tion (della Torreet al. 2002). Although this phenom-
enon has been most extensively studied in A. gambiae
(Coluzzi et al. 2002), karyotypic analysis of .5,000 A.
funestus specimens also has revealed significant and
stable departures from Hardy–Weinberg equilibrium
due to heterokaryotype deficit and linkage disequi-
librium between arrangements on independently as-
sorting chromosome arms (Costantini et al. 1999;
Guelbeogo et al. 2005). Equilibrium was restored under
the hypothesis of two sympatric and morphologically
indistinguishable but assortatively mating chromosomal
forms designated Folonzo and Kiribina. Corresponding
is associated with rice cultivation and is less likely than
Folonzo to rest indoors, feed on humans, or carry ma-
laria parasites (Costantini et al. 1999).
Owing to their role in suppressing recombination
between alternative arrangements, chromosomal inver-
sions may facilitate divergence in the face of gene flow
between incompletely isolated populations (Noor et al.
Sequence data from this article have been deposited with the EMBL/
GenBank Data Libraries under accession nos. DQ126687–127169.
1Corresponding author: 317 Life Sciences Bldg., P.O. Box 369, University
of Notre Dame, Notre Dame, IN, 46556. E-mail: firstname.lastname@example.org
Genetics 173: 1389–1395 ( July 2006)
2001; Rieseberg 2001; Ayala and Coluzzi 2005).
Whereas gene flow would be expected to homogenize
homosequential regions of thegenome, gene exchange
is restricted between rearranged regions, allowing dif-
ferences to accumulate and leading to the prediction of
greater divergence in such regions. To date, molecular
evidence supporting this prediction for A. funestus has
been contradictory between recent studies in Burkina
Faso and Cameroon (Cohuet et al. 2005; Michel et al.
2005a). Here, we examine this problem on a wider
geographic scale in Burkina Faso, where Folonzo and
Kiribina forms are sympatric. Using chromosomal in-
versions, 16 microsatellite markers, and mitochondrial
DNA sequence, we assessed the relative impact of inver-
sions and geographic distance on the partitioning of
MATERIALS AND METHODS
Mosquito sampling and processing: Villages, the spatial
sampling unit used in this study, are 1–3 km in radius
consisting of 20- to 50-m2compounds, each containing 2–13
small, closely spaced huts. Collections were made inside
multiple huts and compounds per village by pyrethrum spray
catches. Eight villages were sampled along an east–west
transect in Burkina Faso during November and December
2002 (Figure 1). The villages of Vi (11?459N, 3?89W), Siby
(11?519N, 2?589W), Ividie (11?499N, 2?389W), La (12?49N,
2?189W), Pehele (12?159N, 1?119W), and Sabtenga (11?489N,
0?309W) are all located within the Sudan–Savanna ecoclimatic
zone; Bagre (11?329N, 0?289W) and Dirze (11?249N, 0?369W)
lie on or just inside the border with the Guinean–Savanna
zone. The area sampled is characterized by annual rainfall
amounts of 550–900 mm and large flat expanses of tall grasses
or shrubs with scattered trees.
Mosquitoes were sorted morphologically to the Funestus
Group (Gillies and Coetzee 1987) in the field under a
dissecting scope. Ovaries from females at the appropriate
gonotrophic stage were dissected in the field into individually
labeled tubes containing modified Carnoy’s solution (etha-
nol:glacial acetic acid, 3:1) and held on ice until they could be
stored at ?20? for later polytene chromosome analysis. Mos-
quito carcasses were stored individually at room temperature
preparation followed Sharakhov et al. (2001). Assignment of
individual females to the Folonzo or Kiribina form was per-
formed on the basis of karyotype, according to Guelbeogo
et al. (2005). Only those specimens that could be unambigu-
ously assigned were analyzed further. After removal of abdo-
mens for later blood meal analysis (results will be presented in
a future article), genomic DNA was extracted from single
mosquito carcasses in 96-well plates (Wizard SV-96 genomic
DNA purification system, Promega) with the aid of a Biomek
FX workstation (Beckman Coulter). Prior to analysis by PCR,
genomic DNA was diluted 1:10 in H2O (?5 ng/ml). Identifica-
tionofA.funestus sensustricto apartfromcloselyrelated species
with similar or identical adult morphology was achieved by an
Michel et al. (2005a). Sporozoite infection was assayed by
nested PCR targeting the MspI gene (Ranford-Cartwright
et al. 1993; Michel et al. 2005a).
mtDNA sequencing and microsatellite genotyping: A seg-
ment of the mtDNA ND5 gene was amplified by PCR and
sequenced directly on an ABI-3700 (Applied Biosystems) as
described previously (Michel et al. 2005a). Sequences were
aligned, trimmed to a common 834-bp segment, and edited
using Lasergene software (DNASTAR). Sequences have been
deposited under GenBank accession nos. DQ126687–127169.
Sixteen physically mapped reference microsatellite loci
(Sharakhov et al. 2004) were amplified and products were
diluted, pool-plexed (two groups of eight loci each), and
genotyped on a Beckman Coulter CEQ8000 as described pre-
viously (Michel et al. 2005a).
Genetic data analysis: For each village and chromosomal
form, summary statistics of mtDNA and microsatellite poly-
and Fstat220.127.116.11 (Goudet 2001), respectively. Fstat18.104.22.168
was also used to assess deviations from Hardy–Weinberg equi-
ing the inbreeding coefficient Fis) and linkage disequilibrium
between pairs of loci within each village/chromosomal form.
Significance was tested using the randomization approach
implemented in Fstat, which applies Bonferroni corrections.
Within each village the frequency of suspected null alleles was
calculated using the Brookfield 2 estimate (Brookfield
1996). Allele and genotype frequencies were modified accord-
ingly in MICRO-CHECKER (van Oosterhout et al. 2004) to
create a null allele-adjusted data set that was compared to the
original data set to explore the impact of null alleles on esti-
mates of genetic differentiation.
Genetic differentiation between pairs of population sam-
ples within and between chromosomal forms was estimated
using FST. For mtDNA, FSTbased on pairwise nucleotide dif-
For microsatellites, FST was computed in Microsatellite
Analyzer (Dieringer and Schlotterer 2003). Global esti-
mates of FSTfor both markers were generated by analysis
of molecular variance (AMOVA) implemented in Arlequin
2.0. For all calculations, significance was assessed by 10,000
(Raymond and Rousset 1995) using a semi-matrix of FST
values based on mtDNA or microsatellite data sets. Signifi-
cance of the regression of genetic distance on geographic dis-
tance was tested using the Mantel procedure (Mantel 1967),
permuting columns of the semi-matrix 10,000 times. For
microsatellites we also used the program SPAIDA (Palsson
2004), which calculates two estimates of spatial autocorrela-
of repeats (Geary’s C) assuming a stepwise mutation model of
microsatellite evolution. Significance of the correlation for
each distance class was evaluated by 1000 permutations of in-
dividuals among locations.
Three approaches were followed to investigate whether
populations cluster by geographic proximity or chromosomal
form. In the first, we reconstructed relationships on the basis
of microsatellite or mtDNA genetic distances. For microsatel-
lites, Nei et al.’s (1983) DAdistance was used to construct a
neighbor-joining tree with Populations 1.2.28 software (http:/ /
strapping over individuals 100 times was used to assess robust-
ness of individual branches. For mtDNA, a haplotype network
was constructed using the statistical parsimony method
implemented in TCS 1.13 (Clement et al. 2000). The second
approach, applied only with microsatellites, used the model-
based clustering method implemented in STRUCTURE 2.0
(Pritchard et al. 2000), which estimates the most likely num-
ber of distinct populations on the basis of multilocus geno-
types with no a priori assumptions of population structure. Each
run length of 1,000,000 chains for each of K ¼ 1–8, replicated
1390A. P. Michel et al.
using microsatellite data were performed with the program
SPASSIGN (Palsson 2004) using a frequency-based method
developed byPaetkauetal.(1995) tocalculatetheprobability
of assignment to the Kiribina or Folonzo chromosomal forms.
Tests of non-neutral evolution were performed for micro-
1989), R2(Ramos-Onsins and Rozas 2002), and Fu’s Fs(Fu
1997) were computed using DnaSP. The first two statistics are
based on the distribution of mutation frequencies, whereas
where low-frequency mutations are in excess, as expected fol-
lowing a selective sweep or a population expansion. Statistical
significance was evaluated by coalescent simulations (10,000
b-imbalance index was computed using the b1estimator of
b(t) (King et al. 2000). This statistic is based on the imbalance
between allele size variance and heterozygosity at a locus, and
it assumes values ,1 following a population expansion. For
by bootstrapping over loci using an SAS program kindly pro-
vided by T. Lehmann (Donnelly et al. 2001).
During November and December 2002 we collected
indoor-resting A. funestus from eight villages spanning
an ?300-km east–west transect across Burkina Faso
(Figure 1). Although both Folonzo and Kiribina forms
were collected from each village, their relative propor-
tions were unbalanced such that in only two locales (La
and Pehele villages) were adequate sample sizes (i.e.,
.30) obtained for both forms in the same collection.
mtDNA and microsatellite polymorphism: Levels of
polymorphism for both marker classes were high
(supplemental Table S1 at http:/ /www.genetics.org/
supplemental/). In the set of 483 mtDNA ND5 sequen-
ces (302 Folonzo, 181 Kiribina), 180 of 834 sites were
tons, resulting in a large total number of haplotypes
(280) and a high haplotype diversity (0.98). Average
nucleotide diversity overall was only 0.006 per site, or
4.6 nucleotide differences between sequences. Thus,
the total mtDNA data set is composed of closely related
yet often unique sequences, as reflected in the greater
estimates of Nem based on u vs. p (supplemental Table
S1, http:/ /www.genetics.org/supplemental/).
Microsatellite polymorphism levels were high at all
16 loci genotyped from 502 A. funestus (315 Folonzo,
187 Kiribina). Overall allelic richness averaged 9.0 per
locus with a range of 3–20. Values of observed hetero-
Weinberg equilibrium within some samples of Kiribina
and Folonzo (19 of 160 tests) indicated heterozygote
deficit (supplemental Table S1, http:/ /www.genetics.
org/supplemental/), most likely due to null alleles as
of null alleles in microsatellite studies of this and other
species (Lehmann et al. 1996; Dakin and Avise 2004;
Michel et al. 2005a; Stump et al. 2005). Notably, all pop-
ulation samples showed heterozygote deficits at locus
FunD, and three more showed significant deficits at
AFND12 and AFUB12, together accounting for 16 of 19
significant tests. Under the assumption of random mat-
ing, we estimated null allele frequencies for each locus
and population and explored their impact by repeating
ments: (1) with a data set adjusted for estimated null
allele frequencies, or (2) by excluding FunD, AFND12,
and AFUB12. The effect of either treatment was mini-
mal and not statistically significant.
Figure 1.—Map of Burkina Faso showing the
capital Ouagadougou, the villages of Koubri
and Kuiti sampled in a previous study (Michel
et al. 2005a), and the villages where mosquito col-
lections were made for the present study. Three
climatic regions indicated by different shading
(from north to south: Sahel, Sudan–Savanna,
and Sudan–Guinea) are delimited by the 600-
and 900-mm isohyets.
Genetic Structure of A. funestus in Burkina Faso1391
Genetic clustering: Neighbor-joining trees were re-
constructed to explore whether A. funestus would cluster
on the basis of physical proximity of sampling sites or
ces were uninformative, resulting in a star-like phylogeny
(data not shown). However, microsatellite DAdistances
ples of both chromosomal forms collected in sympatry
from the villages of La and Pehele were more closely re-
lated to samples of the same form separated by ?120 km
than to sympatric samples of the other form.
The same pattern was revealed by unsupervised
2). Two clusters were inferred as most likely correspond-
ing to the chromosomal forms. Predefined Kiribina and
Folonzo samples were consistently different in their
proportion of membership to cluster 1 (mean, 0.67 6
0.05 and 0.40 6 0.02 for Kiribina and Folonzo, respec-
tively) or cluster 2 (0.33 6 0.05 and 0.60 6 0.02 for
of Kiribina specimens and 74% of Folonzo specimens.
mosomal form was significantly better than random
(X2¼ 143.7, P , 0.001). By contrast, the proportion of
Kiribina or Folonzo specimens collected from La and
Pehele villages that were correctly assigned to those
villages was not significantly different from random
(X2¼1.48,P .0.1 and X2¼2.64,P .0.1,respectively).
Population structure: The genetic structure of these
A. funestus populations was investigated using a hierar-
chical AMOVA approach based on mtDNA and micro-
satellites. We defined three hierarchical levels: within
populations, among populations within chromosomal
forms, and between forms. Results for both types of
markers were virtually identical and in accord with the
clustering analyses. Almost all genetic diversity (99%)
was partitioned within populations, indicating high
genetic similarity overall. Global FSTestimates revealed
slight but significant overall genetic structure (mtDNA:
ever, most AMOVA tests of population structure within
forms (i.e., two hierarchical levels: within populations
and among populations within forms) were not signif-
icant (Kiribina mtDNA FST¼ 0.002, P ¼ 0.32 and micro-
satellite FST¼ 0.003, P ¼ 0.06; Folonzo mtDNA FST¼
0.003, P ¼ 0.11 and microsatellite FST¼ 0.004, P ¼
Within chromosomal forms, pairwise FSTcomparisons
Table S2 at http:/ /www.genetics.org/supplemental/).
Of 40 comparisons in total, only 7 revealed significant
values after correction for multiple tests. Two of the
largest such values, corresponding to mtDNA contrasts
between Siby and either La or Ividie, involved a trio of
nearby villages, whereas comparisons across much larger
distances (e.g., Ividie vs. Sabtenga) were not statistically
different from zero. These results are inconsistent with
positively correlated geographic and genetic distances
and instead may reflect heterogeneous sampling of rare
alleles, which collectively compose a substantial fraction
Figure 2.—Clustering of A. funestus study pop-
ulations from Burkina Faso, inferred from micro-
satellites. Top: Neighbor-joining tree based on
Nei et al.’s (1983) DAdistance. Bootstrap values
.50% are shown. Branches are labeled by sam-
pling location (village) and chromosomal form
assignment (solid circles, Folonzo; open circles,
Kiribina). Bottom: Plot from an unsupervised
STRUCTURE run of highest estimated probabil-
ity at k ¼ 2 (two clusters). Each individual is rep-
resented by a vertical line in which shading (solid
or open) indicates membership coefficients (scale
bar at left) for alternative clusters corresponding
to Folonzo and Kiribina. Plot is partitioned into
10 segments representing samples of Kiribina
or Folonzo from different villages. Village names
are indicated below the plot, and chromosomal
form membership (determined independently
by karyotyping) is indicated atthe bottom by a bar.
1392A. P. Michel et al.
of allelic variation. An excess of rare mtDNA alleles was
singletons alone accounted for 44% of mtDNA variable
sites. For microsatellites, rare alleles (defined here as
having frequencies ,2.5%) accounted for 43% of alleles
at a locus in the Kiribina form (range across 16 loci, 20–
62%) and 52% of alleles at a locus in Folonzo (33–65%).
study from Koubri/Kuiti (Michel et al. 2005a) was
similarly high in both forms (54% of alleles from ?380
Kiribina individuals, 41% of alleles from ?180 Folonzo
Formal tests of isolation by distance were conducted
separately for each chromosomal form and marker
class. For Folonzo, tests included the villages of Vi, Siby,
Ividie, La, Pehele, and Sabtenga, spanning pairwise
distances of 20–280 km. For Kiribina, the tests included
La, Pehele, Dirze, and Bagre (19–200 km; see Figure 1).
In no case did the data fit the isolation-by-distance
model based on Mantel tests (mtDNA: Kiribina P .
0.08, Folonzo P . 0.95; microsatellites: Kiribina P .
0.08; Folonzo P . 0.54). In addition, using the SPAIDA
program to calculate two estimates of spatial autocorre-
lation extended for microsatellites (Palsson 2004), we
found no significant correlations with either Moran’s I
under the infinite allele model or Geary’s C under the
stepwise mutation model. Therefore, for distances ,300
km the data do not support an isolation-by-distance
tests of isolation by distance using microsatellite data
from each village without regard to chromosomal form
membership of the samples produced a potentially mis-
leading, marginally significant correlation (P , 0.053).
This emphasizes the need for karyotype information on
Folonzo and Kiribina forms of A. funestus.
For comparisons between chromosomal forms, the
analysis was limited to samples in which sufficient num-
bers of both forms were captured simultaneously from
thesame locale (La and Pehelevillages). The results are
given in Table 1, which also provides data from a pre-
vious study in Koubri/Kuiti (see Figure 1). The micro-
satellite data were consistent across space (among
villages) and time (among years) and gave evidence of
significant differentiation between forms, although the
magnitude of differentiation was slight. Notably, micro-
satellite differentiation was not genomewide, as shown
3Rb, whose arrangement is crucial in chromosomal
forms (and none within forms) could be explained by
the five 3R loci alone: three outside of inversions, two
within inversion 3Rb. The lack of microsatellite differ-
entiation outside of 3R is consistent with slight or
nonexistent mtDNA differentiation.
Demographic inference: Typical of villages across the
transect, an excess of low-frequency mtDNA alleles in
Folonzo and Kiribina samples from La and Pehele
villages was indicated by each of three tests on the basis
of the frequencies of segregating sites or the haplotype
distribution (Table 2). These results are consistent with
alternative scenarios—a recent selective sweep or pop-
ulation expansion—both of which are characterized by
an elevated proportion of recent mutations occurring
on the external branches of the genealogy. Supporting
evidence from 16 microsatellite loci, on the basis of a b-
imbalance index significantly ,1 in both Kiribina and
Folonzo from La and Pehele, favors population expan-
sion as the more likely explanation (Kimmel et al. 1998;
King et al. 2000) because the effect is genomewide. This
index was significantly ,1 in virtually all individual
Microsatellite and mtDNA differentiation (FST) between
A. funestus chromosomal forms in Burkina Faso
Data set La PeheleKoubri/Kuitia
All loci (16)
3R loci (5)
Non-3R loci (11)
NS, P . 0.05; *P , 0.05; **P , 0.01; ***P , 0.001.
aFrom Michel et al. (2005a).
Summary of neutrality tests against population growth
Mitochondrial DNA Microsatellites
*P , 0.05; **P , 0.01; ***P , 0.001.
a95% confidence interval based on bootstrapped b1values.
Genetic Structure of A. funestus in Burkina Faso 1393
samples (only Folonzo from Pehele was marginally sig-
nificant; data not shown).
As in Burkina Faso, chromosomal polymorphism in
Cameroon appears consistent with the presence of
Kiribina and Folonzo forms, given heterokaryotype
deficits and linkage disequilibria between inversions in
central localities of that country (Cohuet et al. 2005).
However, at odds with the present study, molecular data
from 10 microsatellite loci—of which 7 were shared
discontinuities (Cohuet et al. 2005). Instead, the pres-
ence of significant population structure among localities
was attributednot togenetic isolation between the forms
but to an isolation-by-distance effect that encompassed
all loci both inside and outside of inversions. The dis-
crepancy between studies may reflect a true difference
between West and Central African populations of A.
funestus. However, as only 25% of the ?600 Cameroon
specimens analyzed molecularly also were karyotyped,
the data were not partitioned by chromosomal form
prior to analysis. In light of the nonrandom distribution
of chromosomal inversions along a north–south cline of
decreasing aridity in Cameroon, the significant correla-
tion of genetic and geographic distance may have been
driven by hidden population structure due to displace-
ment of the chromosomal forms along the cline. More-
over, the seeming contradiction regarding the influence
of chromosomal inversions on estimates of genetic dif-
ferentiation also may be a function of different method-
ology rather than biology. Whereas Cohuet et al. (2005)
partitioned their loci strictly according to whether they
fell inside or outside inversions, we found that loci on
chromosome 3R (whether within or near inversions)
accounted for all significant differentiation. Viewed in
this way, the data from Cameroon are more congruent
with the present study; their mean FSTestimate for all 4
loci from chromosome 3R (FunG, FunD, AFND19, and
AFND20) was 0.016, twofold higher than the estimate
across the 4 remaining loci whose genome position has
been mapped (0.008).
Under a model of allopatric speciation (or complete
reproductive isolation), it is expected that genetic di-
vergence should be found genomewide and in propor-
tion to time since isolation. The present data are not
compatible with this model. Yet the absence of genome-
of lineage splitting. Divergence with gene flow models
predict a mosaic pattern of undifferentiated genome
segments homogenized by recombination interspersed
with islands of differentiation due to strong selection
against introgression (Machado et al. 2002; Wu and
Ting 2004; Feder et al. 2005). These islands are likely to
coincide with low-recombination regions of the ge-
nome, including polymorphic chromosomal inversions
and pericentromeric areas, and are likely to contain
many of the genes underlying differential ecological
adaptations and premating isolation (Noor et al. 2001;
Ortiz-Barrientos et al. 2002). The African malaria
vector A. gambiae appears to represent this situation very
well, with nascent species M and S in sympatry across
West and CentralAfrica (Stump et al. 2005; Turneret al.
2005). In the case of Folonzo and Kiribina forms of A.
funestus, alternative arrangements on chromosome 3R
are core components of a deterministic algorithm that
makes taxonomic assignments on the basis of karyotype
(Guelbeogo et al. 2005); no independent molecular
markers have been discovered. Such an algorithm has
limitations. It should be emphasized that no inversions
or inversion combinations are strictly diagnostic for
different forms, neither for A. funestus nor for A. gam-
biae. Inversions are known to be shared across forms,
albeit at significantly different frequencies. Neverthe-
less, their importance in partitioning genetic variation
in A. funestus is underscored by the finding that all
significant nuclear differentiation derives from loci
within or near inversions on chromosome 3R. Neither
candidate genes nor candidate phenotypes exist that
can be mapped to this chromosome arm at present,
although tantalizing evidence suggests that the inver-
sions may be linked to alternative utilization of anthro-
pogenic (ricefield) vs. natural (e.g., marsh) habitats
(Costantini et al. 1999), to which they may be dif-
ferentially adapted. The present data do not support a
tion, as sporozoite rates were uniformly high in both
forms (8.8% and 9.1% for Kiribina and Folonzo,
As is thecasefor A. gambiae and our own species, most
genetic diversity in A. funestus (99%) is accounted for
ing to recent population expansion. The implications
librium so common to analyses of population genetic
structure and especially to indirect estimates of con-
temporarygeneflow isnot valid for either vector.Migra-
tion rates among demes within a chromosomal form, or
gene flow between chromosomal forms, might be much
smaller than what a simple transformation of FSTto Nm
would suggest (Whitlock and McCauley 1999). Sec-
ond, it suggests a shared history that follows from the
interplay between human population expansion, asso-
ciated anthropogenic modifications to the environ-
ment, and the correlated rise of pest species with the
infectious diseases they transmit. The parallel instances
of incipient speciation in West Africa by these major
malaria vectors, both associated with rice cultivation,
may offer mechanistic insights into the rapid ecological
adaptation and lineage splitting that seem to occur in
response to common environmental cues.
1394A. P. Michel et al.
We thank the entomological technicians at the Centre National de Download full-text
Recherche et de Formation sur le Paludisme for assistance with field
collections. This project was funded by National Institute of Health
grant R01-AI48842 to N.J.B. A.P.M was supported by an Arthur J.
Schmidt Graduate Fellowship from the University of Notre Dame.
Ayala, F. J., and M. Coluzzi, 2005
mans, Drosophila, and mosquitoes. Proc. Natl. Acad. Sci. USA
102(Suppl. 1): 6535–6542.
Brookfield, J. F., 1996 A simple new method for estimating null al-
Clement, M., D. Posada and K. A. Crandall, 2000
program to estimate gene genealogies. Mol. Ecol. 9: 1657–1659.
Coetzee, M., and D. Fontenille, 2004
Anopheles funestus, a major vector of malaria in Africa. Insect
Biochem. Mol. Biol. 34: 599–605.
Cohuet, A., I. Dia, F. Simard, M. Raymond, F. Rousset et al.,
2005 Gene flow between chromosomal forms of the malaria
vector Anopheles funestus in Cameroon, Central Africa and its rel-
evance in malaria fighting. Genetics 169: 301–311.
Coluzzi, M., A. Sabatini, A. Della Torre, M. A. Di Deco and
V. Petrarca, 2002 A polytene chromosome analysis of the
Anopheles gambiae species complex. Science 298: 1415–1418.
Costantini, C., N. F. Sagnon, E. Ilboudo-Sanogo, M. Coluzzi and
D. Boccolini, 1999 Chromosomal and bionomic heterogene-
Faso. Parassitologia 41: 595–611.
Dakin, E. E., and J. C. Avise, 2004
entage analysis. Heredity 95: 504–509.
della Torre, A., C. Costantini, N. J. Besansky, A. Caccone,
V. Petrarca et al., 2002 Speciation within Anopheles gambiae—the
glass is half full. Science 298: 115–117.
Dieringer, D., and C. Schlotterer, 2003
(MSA): a platform independent analysis tool for large microsatel-
lite data sets. Mol. Ecol. Notes 3: 167–169.
Donnelly, M. J., M. C. Licht and T. Lehmann, 2001
recent population expansion in the evolutionary history of the
malaria vectors Anopheles arabiensis and Anopheles gambiae. Mol.
Biol. Evol. 18: 1353–1364.
Feder, J. L., X. Xie, J. Rull, S. Velez, A. Forbes et al., 2005
Dobzhansky, and Bush and the complexities of sympatric speci-
ation in Rhagoletis. Proc. Natl. Acad. Sci. USA 102(Suppl. 1):
Fu, Y. X., 1997Statistical tests of neutrality of mutations against pop-
ulation growth, hitchhiking and background selection. Genetics
Gillies, M. T., and M. Coetzee, 1987
of Africa South of the Sahara. The South African Institute for Med-
ical Research, Johannesburg.
Goudet, J., 2001
FSTAT: A Program To Estimate and Test Gene Diversi-
ties and Fixation Indices (Version 2.9.3) (http:/ /www.unil.ch/izea/
Guelbeogo, W. M., O. Grushko, D. Boccolini, P. A. Ouedraogo,
N. J. Besansky et al., 2005Chromosomal evidence of incipient
speciation in the Afrotropical malaria mosquito Anopheles funes-
tus. Med. Vet. Entomol. 19: 458–469.
Kimmel, M., R. Chakraborty, J. P. King, M. Bamshad, W. S.
Watkins et al., 1998Signatures of population expansion in
microsatellite repeat data. Genetics 148: 1921–1930.
King, J. P., M. Kimmel and R. Chakraborty, 2000
of microsatellite-based statistics for inferring past population
growth. Mol. Biol. Evol. 17: 1859–1868.
Koekemoer, L., M. M. Weeto, L. Kamau, R. H. Hunt and
M. Coetzee, 2002 A cocktail polymerase chain reaction
(PCR)assay toidentify members of theAnophelesfunestus(Diptera:
Culicidae) group. Am. J. Trop. Med. Hyg. 66: 804–811.
Lehmann, T., W. A. Hawley and F. H. Collins, 1996
of evolutionary constraints on microsatellite loci using null al-
leles. Genetics 144: 1155–1163.
Lehmann, T., M. Licht, N. Elissa, B. T. Maega, J. M. Chimumbwa
et al., 2003Population structure of Anopheles gambiae in Africa.
J. Hered. 94: 133–147.
Chromosome speciation: hu-
TCS: a computer
Advances in the study of
Microsatellite null alleles in par-
A Supplement to the Anophelinae
A power analysis
Machado, C. A., R. M. Kliman, J. A. Markert and J. Hey, 2002
ferring the history of speciation from multilocus DNA sequence
data: the case of Drosophila pseudoobscura and close relatives. Mol.
Biol. Evol. 19: 472–488.
Mantel, N., 1967 The detection of disease clustering and a gener-
alized regression approach. Cancer Res. 27: 209–220.
Michel, A. P., W. M. Guelbeogo, O. Grushko, B. J. Schemerhorn,
M. Kern et al., 2005aMolecular differentiation between chro-
mosomally defined incipient species of Anopheles funestus. Insect
Mol. Biol. 14: 375–387.
Michel, A. P.,M. J. Ingrasci, B. J. Schemerhorn, M. Kern, G.Le Goff
et al., 2005bRangewide population genetic structure of the Afri-
can malaria vector Anopheles funestus. Mol. Ecol. 14: 4235–4248.
Nei, M., F. Tajima and Y. Tateno, 1983
logenetic trees from molecular data. II. Gene frequency data.
J. Mol. Evol. 19: 153–170.
mosomal inversions and the reproductive isolation of species.
Proc. Natl. Acad. Sci. USA 98: 12084–12088.
Ortiz-Barrientos, D., J. Reiland, J. Heyand M. A. Noor, 2002
combination and the divergence of hybridizing species. Genetica
Paetkau, D., W. Calvert, I. Stirling and C. Strobeck, 1995
satellite analysis of population structure in Canadian polar bears.
Mol. Ecol. 4: 347–354.
Palsson, S., 2004Isolation by distance, based on microsatellite data,
tested with spatial autocorrelation (SPAIDA) and assignment test
(SPASSIGN). Mol. Ecol. Notes 4: 143–145.
Pritchard, J. K., M. Stephens and P. Donnelly, 2000
population structure using multilocus genotype data. Genetics
Ramos-Onsins, S. E., and J. Rozas, 2002
new neutrality tests against population growth. Mol. Biol. Evol.
Ranford-Cartwright, L. C., P. Balfe, R. Carter and D. Walliker,
1993 Frequency of cross-fertilization in the human malaria par-
asite Plasmodium falciparum. Parasitology 107(Pt 1): 11–18.
Raymond, M., and F. Rousset, 1995
population genetics software for exact tests and ecumenicism.
J. Hered. 86: 248–249.
Rieseberg, L. H., 2001 Chromosomal rearrangements and specia-
tion. Trends Ecol. Evol. 16: 351–358.
Rozas, J., J. C. Sanchez-DelBarrio, X. Messeguer and R. Rozas,
2003 DnaSP, DNA polymorphism analyses by the coalescent
and other methods. Bioinformatics 19: 2496–2497.
Schneider, S., D. Roessli and L. Excoffier, 2000
ware for population genetics data analysis, Ver. 2.000. University
of Geneva, Geneva.
Sharakhov, I., O. Braginets, O. Grushko, A. Cohuet, W. M.
Guelbeogo et al., 2004 A microsatellite map of the African hu-
man malaria vector Anopheles funestus. J. Hered. 95: 29–34.
Sharakhov, I. V., M. V. Sharakhova, C. M. Mbogo, L. L. Koekemoer
and G. Yan, 2001Linear and spatial organization of polytene
chromosomes of the African malaria mosquito Anopheles funestus.
Genetics 159: 211–218.
Stump, A. D., J. A. Shoener, C. Costantini, N. Sagnon and N. J.
Besansky, 2005Sex-linked differentiation between incipient
species of Anopheles gambiae. Genetics 169: 1509–1519.
Tajima, F., 1989Statistical method for testing the neutral mutation
hypothesis by DNA polymorphism. Genetics 123: 585–595.
Turner, T. L., M. W. Hahn and S. V. Nuzhdin, 2005
lands of speciation in Anopheles gambiae. PLoS Biol. 3: e285.
van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills and P. F.
Shipley, 2004 Micro-Checker: software for identifying and cor-
recting genotyping errors in microsatellite data. Mol. Ecol. Notes
Whitlock, M. C., and D. E. McCauley, 1999
gene flow andmigration:FSTnot equalto1/(4Nm1 1).Heredity
82(Pt 2): 117–125.
World Health Organization, 2003
210 World Health Organization, Geneva.
Wu, C.-I, and C. T. Ting, 2004Genes and speciation. Nat Rev Genet
Accuracy of estimated phy-
Statistical properties of
GENEPOP Version 1.2. A
Arlequin: a soft-
Indirect measures of
World Health Report 2003.
Communicating editor: L. Harshman
Genetic Structure of A. funestus in Burkina Faso1395