SEARCH

SEARCH BY CITATION

Keywords:

  • Anopheles baimaii;
  • malaria vector;
  • phylogeography;
  • Pleistocene glaciations;
  • population expansion;
  • Southeast Asia;
  • tropical forest

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Anopheles dirus and Anopheles baimaii are closely related species which feed on primates, particularly humans, and transmit malaria in the tropical forests of mainland Southeast Asia. Here, we report an in-depth phylogeographic picture based on 269 individuals from 21 populations from mainland Southeast Asia. Analysis of 1537 bp of mtDNA sequence revealed that the population history of A. baimaii is far more complex than previously thought. An old expansion (pre-300 kyr BP) was inferred in northern India/Bangladesh with a wave of south-eastwards expansion arriving at the Thai border (ca 135–173 kyr BP) followed by leptokurtic dispersal very recently (ca 16 kyr BP) into peninsular Thailand. The long and complex population history of these anthropophilic species suggests their expansions are not in response to the relatively recent (ca 40 kyr BP) human expansions in mainland Southeast Asia but, rather, fit well with our understanding of Pleistocene climatic change there.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

The genetic diversity of a species may be shaped by both historical and contemporary factors. In the case of a species with a strong association with humans or human-altered environments, genetic diversity may have been influenced by historical changes in human populations (e.g. Musca domestica, Krafsur et al., 2005; Ceratitis capitata, Gasperi et al., 2002). Human influence has been claimed for the origins and spread of the major malaria vector in Africa, Anopheles gambiae, in response to the rise of agriculture ca 4000 years ago in the rainforest home of this mosquito in West Africa (Ayala & Coluzzi, 2005). In mainland Southeast Asia, the major vectors of malaria belong to the highly anthropophilic A. dirus species complex. Population expansions inferred previously using mitochondrial DNA (mtDNA) and microsatellites in two widespread species in the A. dirus complex (A. dirus and A. baimaii) were suggested to have been governed by human population expansions in the region (Walton et al., 2000).

An alternative explanation for the population expansions in A. dirus and A. baimaii may be their response to changes in forest extent during the Pleistocene, as these mosquitoes are only found in tropical forests (Scanlon & Sandhinand, 1965; Wilkinson et al., 1978). During the periodic glaciations of the Pleistocene, many tropical forests in mainland Southeast Asia were replaced by pine forests and savannah (Penny, 2001; Hope et al., 2004). Tropical forest was able to persist in some areas of mainland Southeast Asia which would have acted as refugia for forest-dependent species (Brandon-Jones, 1996; Gathorne-Hardy et al., 2002; White et al., 2004) and from which they would have expanded with the spread of forest cover during interglacial periods. There have been very few phylogeographic studies in Southeast Asia to aid the understanding of the effects of Pleistocene climate change on genetic diversity. The aim of this study was to infer the detailed population history of A. dirus and A. baimaii over a wide geographical area, including estimates of timing of events. By relating this history to our knowledge of the spread of human populations in Southeast Asia and to our knowledge of the effects of past climatic change on tropical forests in Southeast Asia we sought to determine which of these factors has most probably shaped genetic diversity in these mosquitoes.

There are at least seven (Sallum et al., 2005) and possibly eight (Walton et al., 1999a) species in the A. dirus complex. These essentially isomorphic species were initially detected from polytene chromosomal banding patterns and from the lack of viability and fertility in hybrids generated from forced matings (Baimai et al., 1987, 1988a). Species designations are further supported by each species having a distinct sequence for the second internal transcribed spacer (ITS2) of ribosomal DNA. Conclusive evidence of the species status of four members of the complex in Thailand (A. dirus, A. baimaii, A. scanloni and A. cracens) comes from allozyme data demonstrating that the chromosomal forms remain distinct even in sympatry (Green et al., 1992). Anopheles dirus and A. baimaii (named recently by Sallum et al., 2005 but known previously as A. dirus A and A. dirus D) are the most geographically widespread species of the A. dirus complex. Anopheles dirus extends across northern Thailand eastwards into Laos, Vietnam and Cambodia, and A. baimaii extends westwards from northern Thailand into Myanmar, Bangladesh and India and southwards into peninsular Thailand (Rosenberg & Maheswary, 1982; Tun-Lin et al., 1995; Baimai et al., 1988b).

The larval habitat of these mosquitoes is shaded temporary ground pools in mature (often primary) tropical forest (for example, elephant foot prints, animal wallows and puddles), with adults only leaving the forest briefly, and if necessary, in search of a blood meal (Scanlon et al., 1968; Prakash et al., 1997; C. Walton, personal observation). Anopheles baimaii has been found breeding in rock-lined wells shaded by mature trees in southern Myanmar (Tun-Lin et al., 1987; Oo et al., 2002) and both species have occasionally been found in mature plantations and orchards (Rosenberg et al., 1990; Obsomer et al., 2007). These are the limits of the use of man-altered environments by these species and both of these settings imitate their natural habitat. These species are highly anthropophilic with A. dirus (Trung et al., 2005) and A. baimaii (Dutta et al., 1996) feeding almost exclusively on humans rather than bovines when given the choice. This, along with their high susceptibility to the malaria parasite and exophilic behaviour, mean that A. dirus and A. baimaii are the most efficient malaria vectors in Asia (Obsomer et al., 2007 and references therein). Population genetic studies of these mosquitoes are therefore important from an applied perspective as an understanding of their population structure and gene flow can play an important role in the control of malaria (Collins et al., 2000).

Previous studies of the population structure and history of A. dirus and A. baimaii used the cytochrome oxidase I (COI) gene of mitochondrial DNA (mtDNA) in samples from Thailand, Myanmar and Bangladesh (Walton et al., 2000) and microsatellites in Thailand (Walton et al., 2001). In this paper, we have sampled mosquitoes from most of mainland Southeast Asia extending the previous sampling area of Bangladesh, Myanmar and northern Thailand to include Cambodia, Laos, north-east India and peninsular Thailand. Both the COI gene (used originally) and the cytochrome oxidase II (COII) gene of mtDNA have been used. This enabled us to generate long sequences giving considerable power to the reliable inference of population history, although this is necessarily that of a single locus. We have used a combination of summary statistics, genetic clustering (samova), model-based methods, and a reconstructed gene genealogy to infer population structure and history. For the genealogy we used a novel multidimensional vector space (MVS) method of haplotype network construction (Okabayashi et al., 2006) which allows for some alternative evolutionary pathways while providing a solution to the problem of tree uncertainty that can be caused by homoplasy (Kitazoe et al., 2005). The greater resolution of population history we have been able to infer reveals that it is much more complex than apparent previously, shedding new light on our understanding of the factors underlying genetic diversity in these species.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Mosquito collection and identification

In addition to the mosquitoes available from the previous study from Bangladesh, Myanmar and Thailand (Walton et al., 2000, 2001), collections were made in Cambodia, India, Laos and northern and southern Thailand from 2002 to 2003 to make a total of 21 sites (Fig. 1). Along the Thai-Myanmar border there were sympatric populations of A. dirus and A. baimaii at two sites (1 and 2). With the exception of Myanmar where mosquito larvae were collected, adult females were collected from all sites and assumed to be unrelated. Mosquitoes were identified as members of the A. dirus complex using morphological characters (Rattanarithikul & Panthusiri, 1994) and then identified to species using allele-specific PCR of the ITS2 rDNA (Walton et al., 1999a).

image

Figure 1.  Relief map of mainland Southeast Asia showing the mosquito collection sites and population groups defined by samova analysis for k = 8. Sites are colour coded according to species collected: yellow = Anopheles dirus, blue = Anopheles baimaii and white = A. dirus and A. baimaii. Seven samova groupings are indicated by black circles. The eighth grouping (A. dirus at site 2, Ratchaburi) is indicated by 2*.

Download figure to PowerPoint

DNA extraction and sequencing of mtDNA

Total DNA was extracted from individual mosquitoes using either a ‘salting-out’ protocol (Walton et al., 1999a) or standard phenol chloroform extraction (Sambrook, 1989). The COI sequences of 78 mosquitoes were available from Walton et al. (2000). New amplifications of this gene were performed using A. dirus specific primers D1664-86 and D2893-70 (Walton et al., 2000) to yield a product of 1230 bp. A 777-bp fragment of the COII gene was amplified from all mosquitoes using the LEU and LYS primers (Sharpe et al., 2000). The PCR reactions were performed in 50-μL volumes on a GeneAmp 9700 thermocycler (Applied Biosystems, Warrington, UK). Each reaction mix included 1–2 μL of DNA (equivalent to ∼1/500 of a mosquito), COI primers at 1 μm or COII primers at 0.24 μm, 200 μm dNTP, 1.5 mm MgCl2 and 2.5 units of Thermoprime Plus DNA polymerase in 1× PCR buffer (CLP, Northampton, UK). For COI, initial denaturation was for 5 min at 94 °C, followed by 35 cycles of amplification (20 s at 94 °C, 30 s at 45 °C and 1 min at 72 °C) and a final incubation of 3 min at 72 °C. For COII, initial denaturation was for 5 min at 94 °C, followed by 39 cycles of amplification (1 min at 94 °C, 1 min at 51 °C and 2 min at 72 °C) and a final incubation of 10 min at 72 °C. The products were purified on Montage columns (Millipore, Billerica, MA, USA). All products were sequenced in both directions (by MWG-Biotech, Ebersberg, Germany) using the amplification primers with the exception of the forward direction of COI which was sequenced with the primer D1908-30 (Walton et al., 2000). DNA sequences were assembled using the SeqEdTM multiple sequence editor program (Applied Biosystems) and checked manually. For all analyses, the COI (871 bp) and COII (666 bp) mtDNA sequences were concatenated with the reading frame maintained to give sequences of 1537 bp.

Population structure and genetic diversity analysis

Alternative models of evolution were tested using modeltest 3.5 (Posada & Crandall, 1998). The data for both species best fitted the HKY (Hasegawa, Kishino and Yano) substitution model with a transition/transversion parameter of 9.642 for A. dirus and 9.025 for A. baimaii. The gamma distribution shape parameter was given as 0.926 for A. dirus and 0.854 for A. baimaii. This model was used in further analyses as appropriate but in arlequin (see below) the Tamura and Nei model of evolution used as the HKY model is not available.

Spatial analysis of molecular variance (samova: Dupanloup et al., 2002) was used to cluster populations into user-defined numbers of groups (k) by maximizing between group variation (i.e. maximizing FCT) and minimizing genetic differentiation among populations within groups (i.e. minimizing FSC). samovas were computed for k = 2–12, using 1000 simulated annealing steps from each of 100 sets of initial starting conditions. Hierarchical amova and pairwise FST values were subsequently estimated using the optimal population grouping in arlequin 2.0 (Schneider et al., 2000) and the significance tested using 1000 permutations. arlequin 2.0 was also used to perform Mantel tests, to test an isolation-by-distance model, and to make estimates of genetic diversity.

Haplotype networks

Haplotype networks were constructed for A. dirus and A. baimaii separately and together using the median joining algorithm (Bandelt et al., 1999) in the software network 4.1.0.7 (http://fluxus-engineering.com). Several sequences from A. scanloni (formerly known as A. dirus C), a closely related but distinct species of the A. dirus complex that is confined to karst habitat (Baimai et al., 1988a, b; O’loughlin et al., 2007) were included as outgroups. As these networks exhibited extensive reticulation which gives rise to uncertainty of node sequences, we used an alternative, novel procedure (called the ‘core set approach’) to reduce this uncertainty in intraspecific tree construction (Okabayashi et al., 2006).

The core set approach is based on the MVS representation method. Using a unified index to measure deviations from the additivity of evolutionary distances, MVS helps detect convergent evolution and modify the initial pairwise distances so as to remove this convergent effect (Kitazoe et al., 2005, 2007). MVS can also help exclude sequences which cause large deviations from additivity. Here, the core set approach selected the largest subset (first core set) of haplotypes for which maximum parsimony provided a single tree because of a small number of multiple substitutions. Applying the same criteria, a second core set was selected from the remaining sequences. Then obtaining all the intranode sequences of the two subtrees, a node in each tree was determined that enabled these trees to be connected by the maximum parsimony criterion. (When we could not determine the node positions uniquely, we memorized these positions to analyse all of them in the estimation of a final tree.) In the tree-merging process, the two core sets are redefined as one new core set, which was fixed for further tree building. Continuing this procedure gave rise to a final tree structure that represents the evolutionary pathway of node-to-node sequences. This resulting MVS network corresponded well to the branching pattern indicated by neighbour joining, but, by removing the influence of long distance attraction, it suppressed the number of branches proposed by the standard method, paup* (Swofford, 2002) to a large extent.

Inference of population history

Two tests of neutrality, Tajima’s D and Fu’s Fs, both of which are negative when there has been population expansion, were performed in arlequin 2.0. arlequin 2.0 was also used to estimate the observed frequency distributions of pairwise nucleotide differences among haplotypes (i.e. mismatch distributions) for species or population groups and compared with the expectations of the sudden expansion model (Rogers & Harpending, 1992). Demographic history was further explored for species or population groups using generalized skyline plots (Strimmer & Pybus, 2001). These are nonparametric graphical representations of effective population size, Ne, going back through time where Ne is estimated from the spacing of coalescent intervals. The skyline plots were constructed for each species and for the separate regions of A. baimaii using the software genie 3.0 (http://evolve.zoo.ox.ac.uk/software.html?id=genie). In accordance with the standard methodology, skyline plots were constructed using maximum likelihood (ML) genealogies from paup*4.0b10 that were generated by a heuristic search approach with a forced molecular clock and general transition rate (GTR) model. The high level of homoplasy (see above) means that there are many alternative genealogies with similar likelihood values. We expected these alternative genealogies to be similar in their spacing of coalescent events and this was verified as different randomly selected alternative genealogies did indeed yield SKYLINE plots with the same overall profile. However, we also used genealogies generated by the MVS algorithm to construct skyline plots, and where there is congruence between these two approaches we can have confidence in their findings. Coalescent times were estimated by an ML procedure. As the branch lengths of the MVS genealogy represent the realized number of substitutions, they were assumed to follow a Poisson distribution (r8s, Sanderson, 2003). A molecular clock was assumed with the expectation that the effect of rate variability will be small for a study of within population variation. genie was also used to test alternative parametric models of population history using an ML approach. Log-likelihood ratio tests were carried out for nested models. The demographic models (constant population size, exponential growth, expansion growth, logistic growth, piecewise expansion growth and piecewise logistic growth) are described in detail in Pybus & Rambaut (2002) and in the genie manual.

Dating of expansions

Estimates of the expansion start times were obtained from the skyline plots as the transition point between two time intervals which showed an increase in population size from one interval to the next. Estimates of the dates of expansions were made by the generalized nonlinear least squares approach using a model of sudden expansion (Schneider & Excoffier, 1999) implemented in arlequin 2.0. Times of expansion were also generated from the median joining haplotype networks using the rho (ρ) estimate of average mutational distance to ancestral haplotype with standard error σ as calculated in network 4.1.0.7 (Saillard et al., 2000). A divergence rate of 2.3% per Myr (Brower, 1994) and a generation time of 1/10th of a year (Walton et al., 2000) was assumed for all datings. The widely used Brower estimate of 2.2–2.4% per Myr is based mainly on dipteran species, mostly with a splitting date of less than 1 Myr and for different parts of the mitochondrial genome including the CO1 gene. The robustness of an insect mitochondrial molecular clock has been further confirmed by relating an insect molecular phylogeny to palaentological and biogeographical data (Gaunt & Miles, 2002). We therefore consider this divergence rate to be the best estimate available for use in an anopheline study at this time, but the accuracy of our estimated timings necessarily depend on how well this rate applies to these particular Anopheles species.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

MtDNA sequence variation

Several individuals from the Myanmar populations were collected as larvae from the same site at the same time increasing the chance that siblings were collected; only one of each haplotype per population was retained for analysis to avoid an excess of consanguinity. Although this may increase the estimates of diversity of these populations, this effect is expected to be slight as only three individuals were excluded per population. Taking this into account, a total of 137 A. dirus and 132 A. baimaii were sequenced for COI and COII. Haplotype diversity was very high, with 197 haplotypes found amongst the 269 individuals (sequences are deposited in the EMBL database with accession numbers AJ877309AJ877577). The distribution of haplotypes among the 21 populations and other summary statistics are shown in Table 1. Nucleotide substitutions were identified at 181 of the 1537 sites (11.8%), the majority of which were transitions (89%). At sites 2026, 2269, 2542, 2614, 2698 and 2719, there were three segregating bases and at site 2572 all four bases were segregating (site numbers relate to the published A. gambiae mtDNA sequence (Beard et al., 1993).

Table 1.   Summary data for haplotypes among the sites and species.
SpeciesCountrySiteNLat °N Long °EHaplotypesSθπ ± SDθs ± SD
  1. N is the sample size. Haplotypes in bold occur more than once. A number in brackets indicates the frequency of a haplotype at a particular site. S is the number of segregating sites. θπ and θs are estimates of genetic diversity (see Materials and methods).

  2. *A haplotype occurs in Anopheles dirus and Anopheles baimaii.

A. dirus Thailand1 Maesod11N16.59 E98.681, 2, 3, 4, 5, 6, 7 (2), 8, 9*, 10180.0029 ± 0.002 0.004 ± 0.002
2 Ratchaburi10N13.53 E99.3611, 12, 13, 14, 15 (2), 16, 17, 18, 19170.0035 ± 0.0020.0039 ± 0.002
3 Loei12N17.46 E101.419, 20, 21, 22, 23*, 24, 25, 26, 27, 28, 29, 30300.0043 ± 0.0030.0065 ± 0.003
4 Sukhon Nakhon13N17.09 E104.331, 32, 33, 34, 35, 36*, 37, 38, 39, 40, 41, 42, 43280.0035 ± 0.0020.0059 ± 0.002
5 Ubon Ratchathani14N14.72 E105.094 (2), 7, 104, 105*, 106, 107, 108, 109, 110*, 111, 112, 113, 114240.0034 ± 0.0020.0049 ± 0.002
6 Chantaburi5N12.67 E102.147, 44, 45, 46, 47 80.0024 ± 0.0020.0025 ± 0.002
7 Lampang9N18.45 E99.7948, 49, 50, 51, 52, 53*, 54, 55, 56210.0038 ± 0.0020.0050 ± 0.002
Cambodia8 Ratanakiri9N13.75 E106.953 (2), 91, 92, 93* (2), 94, 95, 96130.0027 ± 0.0020.0031 ± 0.002
9 Pursat9N12.26 E102.9215, 20, 84, 85, 86, 87, 88, 89, 90230.0042 ± 0.0030.0055 ± 0.003
10 Mondulkiri9N12.11 E106.8636*, 60*, 97, 98, 99, 100, 101, 102, 103230.0040 ± 0.0020.0055 ± 0.003
Laos11 Sekong13N15.37 E106.677, 41, 57, 58, 59, 60*, 61, 62, 63, 64, 65, 66, 67290.004 ± 0.00230.0061 ± 0.003
12 Savannakhet12N16.57 E106.4436*, 67, 68, 69 (2), 70, 71, 72, 73, 74, 75, 76270.0037 ± 0.0020.0058 ± 0.003
13 Vientiane11N18.35 E102.2613 (2), 23*, 41, 77, 78, 79, 80, 81, 82, 83210.0039 ± 0.0020.0047 ± 0.002
A. baimaiiThailand1 Maesod12N16.59 E98.6893*, 115, 116, 117, 118 (2), 119, 120, 121, 122, 123, 124230.0036 ± 0.0020.005 ± 0.002
2 Ratchaburi11N13.53 E99.36135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145350.0051 ± 0.0030.0078 ± 0.003
14 Kanchanaburi14N14.33 E98.9923*, 105*, 125, 126 (2), 127, 128, 129, 130 (2), 131, 132, 133, 134280.0041 ± 0.0020.0057 ± 0.002
15 Ranong11N10.04 E98.649*, 53*, 140, 156, 157, 158 (3), 159, 160, 161240.0038 ± 0.0020.0053 ± 0.002
16 Phang Nga50N8.60 E98.579* (2), 36* (6), 60* (3), 110* (3), 162, 163, 164 (5), 165 (7), 166 (2), 167, 168 (2), 169, 170, 171, 172 (2), 173, 174, 175, 176, 177, 178 (2), 179, 180, 181430.0045 ± 0.0020.0062 ± 0.002
17 Lan Sa Ka4N8.35 E99.789* (3), 182 50.0016 ± 0.0010.0018 ± 0.001
Myanmar18 Myaing7N19.19 E97.14110*, 146 (2), 147, 148, 149, 150 200.0049 ± 0.0030.0053 ± 0.003
19 Kyauk7N16.61 E98.24146, 148, 151, 152, 153, 154, 155210.0049 ± 0.0030.0056 ± 0.003
Bangladesh20 Chaklapunjee6N21.92 E92.1793*, 193, 194, 195, 196, 197210.0051 ± 0.0030.006 ± 0.003
Assam21 Sonitpur10N26.94 E92.99183, 184, 185, 186, 187, 188, 189, 190, 191, 192470.0084 ± 0.0050.0108 ± 0.003

Population genetic structure and geography

The substantial increase in sampling area, sample size and sequence length in this study uncovered differences in the population structure of A. dirus and A. baimaii that were not evident from the previous study of mtDNA. The samova revealed no obvious geographical boundaries as the variation among groups, and among populations within groups, declines fairly continuously as the number of groups (k) decreases. There were also no obvious species boundaries; k = 2 did not separate A. dirus and A. baimaii. However, far greater structuring was revealed in A. baimaii than in A. dirus. As the number of groupings increased from k = 2 to k = 10, the majority of new groupings formed consisted of individual A. baimaii populations, with populations from Assam, Bangladesh and Lan Sa Ka (sites 21, 20 and 17) being the most distinctive (i.e. they were pulled out successively for k = 2–5). The exceptions to this trend were the placement of A. dirus populations from Ratchaburi (site 2) and Ratanakiri (site 8) into separate groups at k = 7 and 9 respectively. The grouping structure (k = 8) that maximizes the pooling of geographically adjacent populations is shown in Fig. 1. In this, most A. baimaii populations form a group themselves except for the two populations from Myanmar, which form a single group, and the three A. baimaii border populations which form a group with all the A. dirus populations except the A. dirus population from site 2. The differentiation of the geographically close populations of A. baimaii from Maesod (site 1) and Kyauk (site 19) may be related to the Tenasserim Mountain range that runs north–south between them. The distinctiveness of A. dirus from site 2 can be seen as a cluster of closely related haplotypes (11, 13, 15 and 19) in the MVS haplotype network (Fig. 2). This population is genetically isolated with the nearest known A. dirus population at Maesod, ∼300 km away. Excluding this population, this grouping structure indicates that in contrast to A. baimaii there is no population structure in A. dirus despite the large increase in sampling area.

image

Figure 2.  Multidimensional vector space haplotype network for Anopheles dirus and Anopheles baimaii. Numbered circles represent haplotypes and their size is proportional to frequency. Intermediate unsampled haplotypes are represented by small open circles. There is one mutational step between haplotypes. Dotted branches show alternative positions of haplotypes where there is uncertainty. The long lineage to haplotype 188 has been condensed by six mutational steps. For simplicity, one uncertain haplotype is not shown on the network: haplotype 176 which is three mutational steps from either 126 or 9.

Download figure to PowerPoint

With these eight groups the hierarchical amova showed that within-group variation contributes the majority of the overall variance (93%), with the between-group component accounting for 7.5% of the variation. Consequently, the pairwise FST values for the eight population groups (Table 2) are fairly low. After a Bonferroni correction for multiple comparisons (with P ≤ 0.0019) six of the comparisons retained significance. The highest level of differentiation (P ≤ 0.001) was observed in between-species comparisons with two of the four significant comparisons involving the most northern A. baimaii population (site 21, Assam). The remaining two comparisons were A. dirus with A. baimaii populations from Myanmar (sites 18 and 19) and southern Thailand (site 16, Phang Nga). Some high FST values are not significant, notably involving comparisons with A. dirus from site 2 and A. baimaii from site 17; this is most likely due to the small sample sizes of these populations, although they are probably genetically distinctive as they are pulled out by the samova. Interestingly, there is also considerable differentiation between northern and southern A. baimaii populations with four comparisons having FST values > 0.12, two of which are significant. Although quite high, this level of differentiation does not indicate the presence of cryptic species within A. baimaii being much lower than, for example, the divergence estimated between A. dirus and A. scanloni (Walton et al., 2000), and between populations within A. scanloni (O’loughlin et al., 2007), using mtDNA. In addition the sequence of the ITS2 locus, which is a good marker for discriminating even very closely related species in Anopheles (Walton et al., 1999b), does not vary in A. baimaii across its geographic range.

Table 2. FST estimates for the population groups defined by samova.
 All Anopheles dirus (except site 2) and Anopheles baimaii sites 1, 2, 14A. dirus site 2A. baimaii site 21A. baimaii site 20A. baimaii sites 18 and 19A. baimaii site 15A. baimaii site 16
  1. Significance levels: *P < 0.0019, **P < 0.001.

A. dirus site 20.053      
A. baimaii site 210.130**0.099**     
A. baimaii site 200.1020.1030.044    
A. baimaii sites 18  and 190.056**0.0860.0760.057   
A. baimaii site 150.0440.1100.0800.127*0.094  
A. baimaii site 160.061**0.1030.136*0.0570.0700.071 
A. baimaii site 170.0730.2330.0680.1980.1430.0810.094

Given this pattern of differentiation, and the expectation that population history will be similar for geographically connected populations, the A. baimaii populations were grouped into three geographical regions for further population history analysis: northern (sites 20 and 21, Assam and Bangladesh), central (sites 1, 2, 14, 18 and 19, Myanmar and the Thai border) and southern (sites 15, 16 and 17, peninsular Thailand). The A. dirus populations were not subdivided as the samova analysis found almost no population structure in A. dirus, and as the isolated A. dirus site 2 population was the least differentiated from the main A. dirus population group (Table 2). The Mantel test was not significant for A. dirus populations, or for the combined central and southern A. baimaii. It was significant for the combined central and northern A. baimaii populations.

Genealogical distribution of Anopheles dirus and Anopheles baimaii haplotypes

The median joining network (see Supporting Information Fig. S1) has a high degree of uncertainty with large numbers of alternative connections particularly around the core. By comparison, the MVS haplotype network for A. dirus and A. baimaii (Fig. 2) is well resolved with 163 of the haplotypes placed uniquely. There are only 34 haplotypes with alternative connections each with only two possible positions in the genealogy. In the MVS network the repeated mutations were inferred to occur at 12 sites with more than four inferred mutations each (Table 2 in Okabayashi et al., 2006) indicating the presence of a relatively small number of hypervariable sites (see Okabayashi et al., 2006 for a further comparison of the MVS network with neighbour-joining and statistical parsimony methods). The A. scanloni outgroup sequences were separated from the dirus–baimaii MVS network by 12 fixed differences and join it at the central, highest frequency (n = 9) haplotype (36) suggesting that this is the oldest haplotype sampled. As in Walton et al. (2000), the haplotypes of A. dirus and A. baimaii are intermingled with seven haplotypes shared between the two species and there is an overall star-like shape characteristic of population expansion.

For A. baimaii, the different geographical regions show different patterns of distributions in the network indicative of differing population histories in these regions. The haplotypes from the central region (Thai-Myanmar border and Myanmar) are distributed mostly as singletons throughout the network. The Assam and Bangladesh haplotypes are mostly found at the tips of long lineages and many are highly divergent from each other, whereas others show a tendency to clustering (haplotypes 183, 185, 189 and 196). Several highly divergent haplotypes (175, 186, 187, 188 and 189) were confirmed as those of A. baimaii by sequencing ITS2 ribosomal DNA (Walton et al., 1999a). The southern A. baimaii haplotypes are distributed throughout the network but many occur at high frequencies at the cores of expansion clusters and at several tips.

Demographic history

Anopheles baimaii has higher levels of genetic diversity than A. dirus with θ values being the highest in the northern A. baimaii populations (Tables 1 and 3). Both species had significantly negative values for neutrality tests (Table 3) and the mismatch distributions were smooth and unimodal and fitted the expected distributions under the sudden expansion model (Fig. 3). This is in agreement with earlier findings (Walton et al., 2000) and is indicative of population expansion with the offset to the right of the A. baimaii distribution leading to our previous inference of earlier expansion in A. baimaii than A. dirus. However, a major new finding of this study is that it is now apparent from dividing A. baimaii into geographical groupings that there are differences between regions in the demographic history of A. baimaii with each region having a distinctive mismatch distribution. For example, although the small number of comparisons requires caution to be taken in interpretation, the mismatch distribution for the northern region is notably ragged and has a higher mean number of pairwise differences (11). Some values range from 18 to 25, partially collapsing the wavelike signal and indicating longer term stability in this region. The only region to depart significantly from the sudden expansion model is central A. baimaii (P = 0.05). Here, the highly unimodal distribution is indicative of population expansion (Rogers & Harpending, 1992), but the lack of fit to the expansion model is probably due to lower than expected variance due to homoplasy caused by hypervariable sites (Walton et al., 2000). The regional groups of A. baimaii all had significantly negative neutrality test values, except for Tajima’s D for the southern populations. The neutrality tests for the central populations were the most negative, indicating that the signal of population expansion is the strongest in this region.

Table 3.   Summary data for genetic diversity and neutrality tests for all Anopheles dirus and Anopheles baimaii and for the A. baimaii geographical areas.
 NSθπ ± SEθs ± SETajima’s D (P) Fu’s Fs (P)
All A. dirus1371040.0036 ± 0.0020.0123 ± 0.003−2.3128 (0.002)−25.1896 (0.000)
All A. baimaii1321330.0049 ± 0.0030.0159 ± 0.004−2.2538 (0.003)−24.7469 (0.000)
A. baimaii north16590.0073 ± 0.0040.0116 ± 0.004−1.5982 (0.044)−7.7336 (0.000)
A. baimaii central51750.0044 ± 0.0020.0108 ± 0.003−2.0925 (0.002)−25.1113 (0.000)
A. baimaii south65530.0044 ± 0.0020.0073 ± 0.002−1.3396 (0.073)−13.9484 (0.000)
image

Figure 3.  Observed mismatch distributions among haplotypes in Anopheles dirus, Anopheles baimaii and different geographical regions of A. baimaii, indicating their fit to a sudden expansion model. Mean distance was greater in A. baimaii (7.4) than in A. dirus (5.4). The P-values are from the sum of squared deviation goodness-of-fit test for the sudden expansion model (Rogers & Harpending, 1992).

Download figure to PowerPoint

The generalized skyline plots and the best-fitting demographic models for A. dirus and A. baimaii, and the three geographical regions of A. baimaii, are shown in Fig. 4. The ML, Akaike information criterion (AIC) and parameter values for the demographic models are given in Supporting information Table S1. The plots for the ML and MVS trees are broadly similar. Both species, and all three geographical groupings of A. baimaii, show population expansion. Log-likelihood ratio tests (which can only be used for nested models) were used to compare the constant model with all other models, and the exponential model with all other models (results given in Supporting information Table S2). In most cases, a model of constant population size was rejected in comparison with a model with expansion, the only exception being the Southern A. baimaii populations using the ML genealogy and exponential, expansion and piecewise logistic models, and the Northern A. baimaii populations using the MVS genealogy and the piecewise logistic model. However, in all cases where the best-fitting model was compared with the constant or exponential model, there was a significant outcome meaning that the simpler model could be rejected. This indicated that a more complex model with additional parameters always improved the fit of the parametric model to the genealogy. It can also be seen that in each case the skyline plot deviates very strongly from the best model (Fig. 4) indicating that the parametric models are too simplistic. The most reliable estimates of expansion date are therefore probably given by the nonparametric skyline plots which do not rely on a specific simple demographic model. Further, the expansion dates estimated from the MVS skyline plots are expected to be the most accurate as these also have reduced effects of homoplasy. The expansion start dates estimated from the plots (and by other methods) are given in Table 4. Interestingly, both species have relatively high pre-expansion effective population sizes (of the order of 100 000).

image

Figure 4.  Skyline plots from genie analysis. Each plot shows the estimated effective population sizes (Ne) from the Skyline analysis and includes the curves of the parametric models with the best log-likelihood and Akaike information criterion values (AIC) values, for both the ML and MVS genealogies. The y-axis shows effective population size on a log scale; the x-axis shows years from present. The final plot is included to give examples of all parametric models using the maximum likelihood results for the Anopheles baimaii data. For the skyline plots, coalescent intervals were grouped to minimize the AIC values; for the maximum likelihood trees the ε values were set at 1.59e−6, 2.63e−4, 1.27e−3, 4.94e−4 and 2.83e−4 and for the MVS trees the ε values were set at 4.80e−4, 4.52e−4, 1.41e−3, 9.29e−4 and 2.64e−4 for Anopheles dirus, A. baimaii, A. baimaii north, A. baimaii central and A. baimaii south respectively.

Download figure to PowerPoint

Table 4.   Dates of expansions in years BP for Anopheles dirus and Anopheles baimaii estimated from the sudden expansion model, ρ estimate and x intercept of the SKYLINE plots assuming a divergence rate of 2.3% per Myr (Brower, 1994).
 Sudden expansion model (95%CI)ρ ± σ from median joining networkML skylineMVS skyline
A. dirus (all)153 517 (115 187–168 170)172 202 ± 22 130171 000135 000
A. baimaii (all)205 397 (142 032–253 203)234 443 ± 33 441255 000225 000
A. baimaii (north)208 310 (111 736–460 297)290 000327 000
A. baimaii (central)197 560 (126 446–235 806)214 000173 000
A. baimaii (south)220 163 (146 020–301 093)189 000209 000

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Population history

Overall, it can be seen that the population history of A. baimaii is far more complex than previously thought, with evidence of different times of expansion in different regions (Figs 3 and 4). Of course, the accuracy of the estimated absolute dates depends on the assumption of a molecular clock for which there is no calibrated rate in Anopheles. However, between closely related species, and between the same species at different geographical locations, the relative dates of expansions are expected to be reliable. The greater population structure detected in A. baimaii than in A. dirus using mtDNA mirrors findings using chromosomal inversions (Baimai et al., 1988c) and microsatellites (Walton et al., 2001). This, together with the concordance between the expansion signals for microsatellites and mtDNA (Walton et al., 2001) suggests that the variation in the mtDNA data here is due largely to demography.

The oldest and most stable A. baimaii populations are found in the north-west in the foothills of the Himalayas. The gradient of decreasing diversity from the north-west to the south (Table 3) suggests that the north-western A. baimaii populations are ancestral (Templeton et al., 1995) and that they have spread eastwards through Myanmar to the border with Thailand and then southwards. It has been proposed that A. baimaii and A. dirus speciated allopatrically (Baimai et al., 1988c). If so, they have probably made secondary contact along the Thai-Myanmar border. We proposed previously that the similarity of mtDNA of A. dirus and A. baimaii is most likely due to mitochondrial introgression from A. baimaii into A. dirus, rather than ancestral lineage sorting, based on the results of inter-specific crossing studies, and the inconsistency of A. dirus being extremely similar to its sister species A. scanloni at nuclear loci yet indistinguishable from A. baimaii at mtDNA (Walton et al., 2000, 2001 and references therein; unpublished data on the white gene intron). The placing of the border A. baimaii populations within the A. dirus group by samova further supports introgression, possibly when they made secondary contact.

For A. dirus, the mismatch distribution and skyline plot all support a single, simple population expansion that appears to be more recent than that of the central A. baimaii populations (ca 135 vs. ca 173 kyr BP from the MVS skyline plot). If the expansions in both species are in response to a common factor, for instance increase in human hosts or forest cover in a particular geographic region, they would be expected to occur at similar times. The estimated expansion date of A. baimaii in the central region may have been confounded by subsequent gene flow from older populations in the north-west.

In the south of A. baimaii’s range (peninsular Thailand) the lack of a good fit of the parametric models to the ML and MVS skyline plots is probably largely due to the many high-frequency haplotypes from this region, which are unusual in a network made up mainly of singleton haplotypes (Fig. 2). The high-frequency haplotypes generate many closely spaced coalescent events towards the tips of the genealogy resulting in an apparent recent decline in population size in the MVS and ML skyline plots. This is misleading as the presence of the high-frequency unrelated haplotypes actually indicates leptokurtic dispersal where early pioneer individuals (in this case predominantly a subset of the central haplotypes) are able to rapidly multiply in new habitat producing patches of homozygosity, with later migrants seen at low frequencies (Ibrahim et al., 1996; Hewitt, 1999). The narrowing of peninsular Thailand may favour this type of dispersal. Based on this scenario, we estimated the start date of expansion in the southern region by applying the rho statistic to all the high-frequency haplotypes and their descendents (Fig. 2), but excluding the mutational steps between the high-frequency haplotypes that pre-dated the southern expansion and the newer migrants represented by the singleton haplotypes, i.e. by dating the individual expansion clusters with the high-frequency haplotypes at their core and averaging the result. This gives an expansion start date of 15 592 ± 4684 (SE) years BP for leptokurtic dispersal and expansion into the southern region.

Possible factors underlying population history

Population expansion and speciation in the African malaria vector, A. gambiae, has been attributed to the increase in blood supply and suitable habitat resulting from human expansions (Ayala & Coluzzi, 2005). Parallel arguments could potentially be put forward for the expansions in A. dirus and A. baimaii as, like A. gambiae, they are highly anthropophilic malaria vectors. However, unlike A. gambiae, where the availability of larval habitat (sunlit, ground pools of water) has increased dramatically due to the farming and construction activities of humans, the opposite is true of A. dirus and A. baimaii. As the natural habitats of these species, tropical forests, are removed by agriculture and human-mediated deforestation, human expansions cannot have increased the available habitat for these species.

Given their high degree of anthropophilly, A. dirus and A. baimaii must have utilized the new source of blood meals resulting from the expansions of modern humans into Asia. However, like other species in the Leucosphyrus group to which they belong, these mosquitoes also feed on nonhuman primates (but not on domestic animals) (McWilson & Wharton, 1963; Scanlon & Sandhinand, 1965; Scanlon et al., 1968; Obsomer et al., 2007), e.g. on Langurs in a forested area of Bangladesh (Rosenberg, 1982). It is likely that prior to the presence of modern humans, earlier hominines that spread to Asia from Africa ca 2 Ma (Lewin, 1999), were also a food source for these mosquitoes. Overall, it is difficult to predict to what extent the increased availability of blood meals from modern humans could overcome the concomitant loss of forest habitat and their alternative nonhuman food sources.

The differences between the inferred A. dirus/A. baimaii and human demographies also do not support the hypothesis that mosquito expansions are dependent on modern human expansions. Archaeological evidence (Bellwood, 1992 and references therein) and more recent analysis of Southeast Asian human mtDNA (Macaulay et al., 2005) dates the time of modern human migration into Southeast Asia and associated demographic expansion at 40–65 kyr BP. These dates do not overlap any of the estimated confidence intervals of expansion times in mosquitoes (Table 4) which are either much older (probably > 300 kyr BP in the north) or much younger (ca 16 kyr BP in the south based on the arguments above). The rise of agriculture 5000–1000 years ago would also have increased human population density (Bellwood, 1992), but is too recent to be compatible with the mosquito expansions. Overall, the inferred relative datings reveal a history of expansions over a long time period that varies with geography, which is inconsistent with the relatively homogenous expansion of modern humans across the same area.

By contrast, the complex population history of these species correlates very well with our understanding of the effects of Pleistocene climatic change and the varying geographies in mainland Southeast Asia. Based on palaeoenvironmental reconstructions of the last glacial maximum (LGM) (Ray & Adams, 2001) forest refugia for species with distributions largely to the north of the Isthmus of Kra are likely to have been located in the more mountainous northern regions of mainland Southeast Asia or in mountain ranges running north–south along the Thai-Myanmar Border (the Tenasserim mountains) and through Vietnam (the Annamite mountains) (see Fig. 1). It has been hypothesized that mesic valleys in such mountainous habitat may provide the moisture necessary for pockets of forest to persist through climatic changes, with the altitudinal range allowing vegetation to track suitable conditions (Roy, 1997; Hewitt, 2000).

The north-western region, which is particularly wet and mountainous (see Fig. 1), appears to have provided continuous habitat for A. baimaii for at least the last 300 000 years indicating that this region may be a possible Pleistocene forest refugium. A transition from a glacial to a warm climate occurred at about 135 kyr BP suggesting that the expansions at the Thai-Myanmar border (dated to ca 135 kyr BP) may correspond to a time of increasing forest cover. A study of the disjunct distributions of Asian colobine monkeys indicated that the most recent deforestation period that occurred in Asia ca 80 to 17 kyr BP was less severe than the previous one ca 190 to 135 kyr BP (Brandon-Jones, 1996); so, possibly sufficient habitat remained during the LGM to sustain A. dirus in some of its range. Recently, it has been suggested that a Pleistocene population expansion in the Neotropical vector species A. darlingi may be attributed to Amazonian savannah contraction and re-expansion (Mirabello & Conn, 2006), but there are a limited number of phylogeographical studies in northern Southeast Asia with which to draw comparisons. However, another forest-associated mosquito species, Anopheles jeyporiensis, from southern China and northern Vietnam (Chen et al., 2004), a black fly species from Thailand (Pramual et al., 2005) and stone oaks sampled from Myanmar, Yunnan and northern Vietnam (Cannon & Manos, 2003), all show genetics signals of expansion using mtDNA or chloroplast DNA.

The expansion of A. baimaii in southern Thailand, assuming colonization by a subset of the northern haplotypes (see above), is very recent dating to ca 16 kyr BP which corresponds to the end of the LGM (18–21 kyr BP). The dating of this expansion is consistent with this area being dominated by savannah, or a mosaic of savannah and forest, until the present interglacial (Hope et al., 2004; White et al., 2004; Bird et al., 2005). A study of black flies in mainland Thailand also found that the genetic diversity in peninsular Thailand was a subset of the diversity from central/northern Thailand (Pramual et al., 2005) with more recent expansions in the south than the north. Together, this suggests that southern Thailand has experienced greater environmental fluctuations in response to climatic change than further north. A study using mtDNA and microsatellites in two species of forest-dependent fruit bats from peninsular Malaysia and southern Thailand found discrepant demographic signals with evidence for a recent bottleneck and expansion in one species and long-term stability in the other (Campbell et al., 2006). The nonconcordant histories were attributed to a bottleneck associated with speciation in the former species, and the a priori assumption of historical persistence of tropical forests in peninsular Malaysia in the latter species, Cynopterus brachyotis Forest (Campbell et al., 2006). Anopheles baimaii and the black fly species differ from C. brachyotis in that they have only spread southwards recently, presumably from northern refugia, whereas C. brachyotis has a more southerly distribution in peninsular Thailand and Malaysia (Campbell et al., 2004, 2006) and may have experienced very different effects from Pleistocene climatic change.

Gene flow and geography

As both species have experienced expansions, they are not at migration–drift equilibrium; so, gene flow cannot be inferred reliably from FST statistics (Nichols & Beaumont, 1996). However, the presence of genetic structure in A. baimaii and the genetic isolation of the A. dirus site 2 population indicate that there are limits to gene flow. The significance of the Mantel test in A. baimaii may indicate that an isolation-by-distance model is appropriate but this may also be a consequence of population history due to the gradient of decreasing diversity from the north-west to the south. Although the forest habitat of these species is now very fragmented (for example, a reduction from 75% to 25% of land cover in the last 90 years in Thailand; Wannitikul, 2005), until fairly recently, these mosquitoes enjoyed a fairly continuous habitat throughout their range. It is possible that the greater population structure in A. baimaii than A. dirus is due to greater physical barriers to gene flow as the area inhabited by A. baimaii is generally of greater elevation with more dramatic changes in altitude (see Fig. 1). However, as A. dirus and A. baimaii are closely related with similar ecology and behaviour (Baimai et al., 1988a), it seems more likely that the greater population structure in A. baimaii is due to the longer and more complex population history in A. baimaii rather than to differences in their ability to disperse.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

The patterns of diversity in A. baimaii and A. dirus reveal a complex population history, involving population expansions at different places and times, that is most consistent with the species having been influenced by past climatic and environmental change. The future use of nuclear loci would allow the alternative hypothesis of a selective sweep of mtDNA from A. baimaii into A. dirus following secondary contact to be tested and would provide independent data on the timing and nature of demographic expansions.

Although phylogeographic methods have made major contributions to our understanding of the effects of periodic glaciations on the genetic diversity of European and North American biota, there is a paucity of such studies in mainland Southeast Asia despite the fact that its rich tropical forests are under great pressure from human-mediated deforestation (Hewitt, 2000, 2004). Our findings indicate that the demographic history of A. baimaii and A. dirus is shaped by habitat, rather than host, availability and therefore these species may provide a useful starting point for studying the role of forest fragmentation (historical or more contemporary) in shaping the genetic diversity of forest-dependent taxa in Southeast Asia. Although mosquitoes have the advantage of the availability of good taxonomic and ecological information due to their role in malaria transmission, similar studies of other forest-dependent species are also needed to determine the generality of the patterns of genetic diversity found here and to identify if there are regions of particular conservation value. For instance, the high diversity and distinctiveness seen in Assam and Bangladesh suggests that these areas may be of special evolutionary interest.

Finally, there is evidence of restricted gene flow in these species, although we cannot determine the extent of this due to the confounding effects of population history. If a similar situation exists in A. gambiae, which shows similar genetic imprints of population expansion (Donnelly et al., 2001), this would have implications for the proposal to introduce malaria parasite refractory genes into A. gambiae and to drive them effectively through populations.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

We are indebted to the staff of the Office of Disease Prevention and Control in Chiang Mai, Thailand (in particular Ruksakul Kanthawong), the Centers of Malariaology, Parasitology and Entomology in Vientiane, Laos (under the direction of Dr Samlane Phompida) and Phnom Penh, Cambodia (under the direction of Dr Duong Socheat) and Gauhati University and the Malaria Research Centres in Assam, Bangladesh and Myanmar, without whom the collection of mosquitoes for this study would have been impossible. This work was funded by a NERC studentship to S. M. O’Loughlin and the Japan Society for Promotion of Sciences, grant no. 16300086.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information
  • Ayala, F.J. & Coluzzi, M. 2005. Chromosome speciation: humans, Drosophila, and mosquitoes. PNAS 102: 65356542.
  • Baimai, V., Andre, R.G., Harrison, B.A., Kijchalao, U. & Panthusiri, L. 1987. Crossing and chromosomal evidence for 2 additional sibling species within the taxon Anopheles dirus Peyton and Harrison (Diptera, Culicidae) in Thailand. Proc. Entomol. Soc. Wash. 89: 157166.
  • Baimai, V., Poopittayasataporn, A. & Kijchalao, U. 1988a. Cytological differences and chromosomal rearrangements in four members of the Anopheles dirus complex (Diptera, Culicidae) in Thailand. Genome 30: 372379.
  • Baimai, V., Kijchalao, U., Sawadwongporn, P. & Green, C.A. 1988b. Geographic distribution and biting behaviour of four species of the Anopheles dirus complex (Diptera: Culicidae) in Thailand. Southeast Asian J. Trop. Med. Public Health 19: 151161.
  • Baimai, V., Thu, M.M., Paing, M. & Maheswary, N.P. 1988c. Distribution and chromosomal polymorphism of the malaria vector Anopheles dirus species D. Southeast Asian J. Trop. Med. Public Health 19: 661665.
  • Bandelt, H.J., Forster, P. & Rohl, A. 1999. Median-joining networks for inferring intraspecific phylogenies. Mol. Biol. Evol. 16: 3748.
  • Beard, C.B., Hamm, D.M. & Collins, F.H. 1993. The mitochondrial genome of the mosquito Anopheles gambiae: DNA sequence, genome organization, and comparisons with mitochondrial sequences of other insects. Insect Mol. Biol. 2: 103124.
  • Bellwood, P. 1992. Southeast Asia before history. In: The Cambridge History of Southeast Asia, Vol. I (N.Tarling, ed.), pp. 55136. Cambridge University Press, Cambridge.
  • Bird, M.I., Taylor, D. & Hunt, C. 2005. Environments of insular Southeast Asia during the Last Glacial Period: a savanna corridor in Sundaland? Quaternary Sci. Rev. 24: 22282242.
  • Brandon-Jones, D. 1996. The Asian Colobinae (Mammalia: Cercopithecidae) as indicators of Quaternary climatic change. Biol. J. Linn. Soc. 59: 327350.
  • Brower, A.V. 1994. Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. PNAS 91: 64916495.
  • Campbell, P., Schneider, C.J., Adnan, A.M., Zubaid, A. & Kunz, T.H. 2004. Phylogeny and phylogeography of Old World fruit bats in the Cynopterus brachyotis complex. Mol. Phylogenet. Evol. 33: 764781.
  • Campbell, P., Schneider, C.J., Adnan, A.M., Zubaid, A. & Kunz, T.H. 2006. Comparative population structure of Cynopterus fruit bats in peninsular Malaysia and southern Thailand. Mol. Ecol. 15: 2947.
  • Cannon, C.H. & Manos, P.S. 2003. Phylogeography of the Southeast Asian stone oaks (Lithocarpus). J. Biogeogr. 30: 211226.
  • Chen, B., Harbach, R.E. & Butlin, R.K. 2004. Genetic variation and population structure of the mosquito Anopheles jeyporiensis in southern China. Mol. Ecol. 13: 30513056.
  • Collins, F.H., Kamau, L., Ranson, H.A. & Vulule, J.M. 2000. Molecular entomology and prospects for malaria control. Bull. World Health Organ. 78: 14121423.
  • Donnelly, M.J., Licht, M.C. & Lehmann, T. 2001. Evidence for recent population expansion in the evolutionary history of the malaria vectors Anopheles arabiensis and Anopheles gambiae. Mol. Biol. Evol. 18: 13531364.
  • Dupanloup, I., Schneider, S. & Excoffier, L. 2002. A simulated annealing approach to define the genetic structure of populations. Mol. Ecol. 11: 25712581.
  • Dutta, P., Bhattacharyya, D.R., Khan, S.A., Sharma, C.K. & Mahanta, J. 1996. Feeding patterns of Anopheles dirus, the major vector of forest malaria in north east India. Southeast Asian J. Trop. Med. Public Health 27: 378381.
  • Gasperi, G., Bonizzoni, M., Gomulski, L.M., Murelli, V., Torti, C., Malacrida, A.R. & Guglielmino, C.R. 2002. Genetic differentiation, gene flow and the origin of infestations of the medfly, Ceratitis capitata. Genetica 116: 125135.
  • Gathorne-Hardy, F.J., Syaukani, Davies, R.G., Eggleton, P. & Jones, D.T. 2002. Quaternary rainforest refugia in south-east Asia: using termites (Isoptera) as indicators. Biol. J. Linn. Soc. 75: 453466.
  • Gaunt, M.W. & Miles, M.A. 2002. An insect molecular clock dates the origin of the insects and accords with palaeontological and biogeographic landmarks. Mol. Biol. Evol. 19: 748761.
  • Green, C.A., Munstermann, L.E., Tan, S.G., Panyim, S. & Baimai, V. 1992. Population genetic evidence for species A, B, C and D of the Anopheles dirus complex in Thailand and enzyme electromorphs for their identification. Med. Vet. Entomol. 6: 2936.
  • Hewitt, G.M. 1999. Post-glacial re-colonization of European biota. Biol. J. Linn. Soc. 68: 87112.
  • Hewitt, G. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907913.
  • Hewitt, G.M. 2004. The structure of biodiversity – insights from molecular phylogeography. Front. Zool. 1: 116.
  • Hope, G., Kershaw, A.P., Kaars, Svd. et al. 2004. History of vegetation and habitat change in the Austral-Asian region. Quaternary Int. 118–119: 103126.
  • Ibrahim, K.M., Nichols, R.A. & Hewitt, G.M. 1996. Spatial patterns of genetic variation generated by different forms of dispersal during range expansion. Heredity 77: 282291.
  • Kitazoe, Y., Kishino, H., Okabayashi, T. et al. 2005. Multidimensional vector space representation for convergent evolution and molecular phylogeny. Mol. Biol. Evol. 22: 704715.
  • Kitazoe, Y., Kishino, H., Waddell, P. et al. 2007. Robust time estimation reconciles views of the antiquity of placental mammals. PLoS ONE 2: e384.
  • Krafsur, E.A., Cummings, M.A., Endsley, M.A., Marquez, J.G. & Nason, J.D. 2005. Geographic differentiation in the house fly estimated by microsatellite and mitochondrial variation. J. Hered. 96: 502512.
  • Lewin, R. 1999. Human Evolution: An Illustrated Introduction, 4th edn. Blackwell Science, Inc, Malden.
  • Macaulay, V., Hill, C., Achilli, A. et al. 2005. Single, rapid coastal settlement of Asia revealed by analysis of complete mitochondrial genomes. Science 308: 10341036.
  • McWilson, W. & Wharton, R.H. 1963. The vectors of simian malaria: identity, biology, and geographical distribution. J. Parasitol. 49: 892904.
  • Mirabello, L. & Conn, J.E. 2006. Molecular population genetics of the malaria vector Anopheles darlingi in Central and South America. Heredity 96: 311321.
  • Nichols, R.A. & Beaumont, M.A. 1996. Is it ancient or modern history that we can read in our genes? In: Aspects of the Genesis and Maintenance of Biological Diversity (M.E.Hochberg & R.Barbault, eds), pp. 6987. Oxford University Press, Oxford.
  • O’loughlin, S.M., Somboon, P. & Walton, C. 2007. High levels of population structure caused by habitat islands in the malarial vector Anopheles scanloni. Heredity 99: 3140.
  • Obsomer, V., Defourny, P. & Coosemans, M. 2007. The Anopheles dirus complex: spatial distribution and environmental drivers. Malar. J. 6: 26.
  • Okabayashi, T., Kitazoe, Y., Kishino, H., Watabe, T., Nakajima, N., Okuhara, Y., O’Loughlin, S. & Walton, C. 2006. Core set approach to reduce uncertainty of gene trees. BMC Evol. Biol. 6: 41.
  • Oo, T.T., Storch, E. & Becker, N. 2002. Studies on the bionomics of Anopheles dirus (Culicidae: Diptera) in Mudon, Mon State, Myanmar. J. Vector Ecol. 27: 4454.
  • Penny, D. 2001. A 40,000 year palynological record from north-east Thailand; implications for biogeography and palaeo-environmental reconstruction. Palaeogeogr. Palaeoecol. 171: 97128.
  • Posada, D. & Crandall, K.A. 1998. modeltest: testing the model of DNA substitution. Bioinformatics 14: 817818.
  • Prakash, A., Bhattacharyya, D.R., Mohapatra, P.K. & Mahanta, J. 1997. Breeding and day resting habitats of Anopheles dirus in Assam India. Southeast Asian J. Trop. Med. Public Health 28: 610614.
  • Pramual, P., Kuvangkadilok, C., Baimai, V. & Walton, C. 2005. Phylogeography of black fly Simulium tani (Diptera:Simuliidae) from Thailand inferred from mtDNA sequences. Mol. Ecol. 14: 39894001.
  • Pybus, O.G. & Rambaut, A. 2002. genie: estimating demographic history from molecular phylogenies. Bioinformatics 18: 14041405.
  • Rattanarithikul, R. & Panthusiri, P. 1994. Illustrated keys to the medically important mosquitoes of Thailand. Southeast Asian J. Trop. Med. Public Health 25: 166.
  • Ray, N. & Adams, J.M. 2001. A GIS-based vegetation map of the world at the last glacial maximum (25,000–15,000 BP). Internet Archeology. Available at: http://intarch.ac.uk/journal/issue11/.
  • Rogers, A.R. & Harpending, H. 1992. Population growth makes waves in the distribution of pairwise genetic differences. Mol. Biol. Evol. 9: 552569.
  • Rosenberg, R. 1982. Forest malaria in Bangladesh III. Breeding habits of Anopheles dirus. Am. J. Trop. Med. Hyg. 31: 192201.
  • Rosenberg, R. & Maheswary, N.P. 1982. Forest malaria in Bangladesh. II. Transmission by Anopheles dirus. Am. J. Trop. Med. Hyg. 31: 183191.
  • Rosenberg, R., Andre, R.G. & Somchit, L. 1990. Highly efficient dry season transmission of malaria in Thailand. Trans. R. Soc. Trop. Med. H 84: 2228.
  • Roy, M.S. 1997. Recent diversification in African Greenbuls (Pycnonotidae: Andropadus) supports a Montane speciation model. Proc. R. Soc. B Biol. Sci. 264: 13371344.
  • Saillard, J., Forster, P., Lynnerup, N., Bandelt, H.J. & Norby, S. 2000. mtDNA variation among Greenland Eskimos: the edge of the Beringian expansion. Am. J. Hum. Genet. 67: 718726.
  • Sallum, M.A., Peyton, E.L. & Wilkerson, R.C. 2005. Six new species of the Anopheles leucosphyrus group, reinterpretation of An. elegans and vector implications. Med. Vet. Entomol. 19: 158199.
  • Sambrook, J. 1989. Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.
  • Sanderson, M.J. 2003. r8s: inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics 19: 301302.
  • Scanlon, J.E. & Sandhinand, U. 1965. The distribution and biology of Anopheles balabacensis in Thailand. J. Med. Entomol. 2: 6169.
  • Scanlon, J.E., Reid, J.A. & Cheong, W.H. 1968. Ecology of Anopheles vectors of malaria in the oriental region. Cahiers ORSTOM Ser. Entomol. Med. Parasitol. 6: 237246.
  • Schneider, S. & Excoffier, L. 1999. Estimation of past demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: application to human mitochondrial DNA. Genetics 152: 10791089.
  • Schneider, S., Roessli, D. & Excoffier, L. 2000. arlequin: A Software for Population Genetics Data Analysis. Ver 2.000. Genetics and Biometry Laboratory, Department of Anthropology, University of Geneva, Geneva.
  • Sharpe, R.G., Harbach, R.E. & Butlin, R.K. 2000. Molecular variation and phylogeny of members of the Minimus Group of Anopheles subgenus Cellia (Diptera: Culicidae). Syst. Entomol. 25: 263272.
  • Strimmer, K. & Pybus, O.G. 2001. Exploring the demographic history of DNA sequences using the generalized skyline plot. Mol. Biol. Evol. 18: 22982305.
  • Swofford, D.L. 2002. paup* Phylogenetic Analysis Using Parsimony (* and Other Methods). ver. 4.0b10. Sinauer Associates, Sunderland, MA.
  • Templeton, A.R., Routman, E. & Phillips, C.A. 1995. Separating population structure from population history: a cladistic analysis of the geographical distribution of mitochondrial DNA haplotypes in the tiger salamander, Ambystoma tigrinum. Genetics 140: 767782.
  • Trung, H.D., Van Bortel, W., Sochantha, T., Keokenchanh, K., Briët, O.J.T. & Coosemans, M. 2005. Behavioural heterogeneity of Anopheles species in ecologically different localities in Southeast Asia: a challenge for vector control. Trop. Med. Int. Health 10: 251262.
  • Tun-Lin, W., Htay-Aung, M.-M., Sebastian, A. & Myo-Paing, M.-M.-T. 1987. Some environmental factors influencing the breeding of Anopeheles balabacensis complex (dirus) in domestic wells in Burma. J. Commun. Dis. 19: 291299.
  • Tun-Lin, W., Thu, M.M., Than, S.M. & Mya, M.M. 1995. Hyperendemic malaria in a forested, hilly Myanmar village. J. Am. Mosq. Contr. 11: 401407.
  • Walton, C., Handley, J.M., Kuvangkadilok, C. et al. 1999a. Identification of five species of the Anopheles dirus complex from Thailand, using allele-specific polymerase chain reaction. Med. Vet. Entomol. 13: 2432.
  • Walton, C., Sharpe, R.G., Pritchard, S.J., Thelwell, N.J. & Butlin, R.K. 1999b. Molecular identification of mosquito species. Biol. J. Linn. Soc. 68: 241256.
  • Walton, C., Handley, J.M., Tun-Lin, W. et al. 2000. Population structure and population history of Anopheles dirus mosquitoes in Southeast Asia. Mol. Biol. Evol. 17: 962974.
  • Walton, C., Handley, J.M., Collins, F.H. et al. 2001. Genetic population structure and introgression in Anopheles dirus mosquitoes in South-east Asia. Mol. Ecol. 10: 569580.
  • Wannitikul, G. 2005. Deforestation in northeast Thailand, 1975–91: results of a general statistical model. Singapore J. Trop. Geogr. 26: 102118.
  • White, J.C., Penny, D., Kealhofer, L. & Maloney, B. 2004. Vegetation changes from the late Pleistocene through the Holocene from three areas of archaeological significance in Thailand. Quaternary Int. 113: 111132.
  • Wilkinson, R.N., Gould, D.J., Boonyakanist, P. & Segal, H.E. 1978. Observations on Anopheles balabacensis (Diptera: Culicidae) in Thailand. J. Med. Entomol. 14: 666671.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Figure S1 Median joining network for Anopheles dirus and Anopheles baimaii from network 4.1.0.7 software. Branch lengths are proportional to the number of mutations on the branch. Circles represent haplotypes and are proportional to frequency. Unsampled intermediate haplotypes are not shown.

Table S1 Results of parametric model testing in genie. The demographic models are described in Pybus & Rambaut (2002).

Table S2 Results of log-likelihood ratio tests for nested parametric models (log-likelihood values are given in Supporting Information Table  S1).

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JEB_1606_sm_TabS1.doc58KSupporting info item
JEB_1606_sm_TabS2.doc81KSupporting info item
JEB_1606_sm_FigS1.pdf1758KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.