Malaria remains one of the world's major health problems claiming at least one million deaths each year in Africa 1. In the forested areas of equatorial Africa, where malaria transmission occurs all year long, Anopheles moucheti mosquitoes can sustain malaria transmission intensities as high as 100–300 infected bites per man per year in villages located at the vicinity of large rivers and slow-moving streams where its larvae develop 2345. However, despite playing such an important epidemiological role in malaria transmission, this group of mosquitoes remains poorly known and insufficiently studied. Data on its bionomics and genetic structure are currently lacking although such data are of paramount importance for a comprehensive implementation and monitoring of malaria vector control in Central Africa 6.

Anopheles moucheti is a group of three morphological forms: A. moucheti moucheti, A. moucheti nigeriensis and A. moucheti bervoetsi, which can be distinguished by minor variations in the size and distribution of pale fringe spots and pale vein spots on the wings at the adult stage, and at the larval stage by the number of branches of the saddle hair (>5 branches for A. m. bervoetsi and <5 branches for the two others) 78. However, population genetics studies using allozyme markers revealed that these morphological variations were not segregating between the different taxonomic units that build up the A. moucheti group in Central Africa and were therefore of poor diagnostic value 9. More recently, DNA sequence differences were detected in the mitochondrial gene encoding the cytochrome B (CytB) and the ribosomal DNA Internal Transcribed Spacers (ITS) 1 and 2 between specimens of A. m. moucheti, A. m. nigeriensis and A. m. bervoetsi and a diagnostic PCR assay was subsequently developed allowing straightforward identification of the three taxonomic units within the A. moucheti group 10. This study further suggested that A. m. moucheti is widespread throughout the forested areas of Central Africa, whereas A. m. nigeriensis and A. m. bervoetsi were found only in their type localities in Nigeria and the Democratic Republic of Congo (DRC), respectively. Microsatellite DNA markers have been isolated from A. moucheti 11 and these were demonstrated to be suitable tools for population genetics studies within this group of mosquitoes 12. Very low levels of genetic differentiation (Fst<0.0275) were detected between A. moucheti populations situated 65–400 km apart in Cameroon, suggesting high levels of gene flow at this geographical scale 12.

Previous findings from Cameroon 12 are expanded through the inclusion of mosquitoes sampled in DRC and Uganda, to further explore the level of genetic structuring between populations of the A. moucheti group and to precise the taxonomic status of An. m. bervoetsi. Analytical methods, based on various aspects of the data, are used to provide insights into the role and relative importance of geographic distance, demographic parameters (eg effective population size and demographic instability) and natural barriers to gene flow such as habitat discontinuities and speciation in shaping the observed population structure.

In this study, six A. m. moucheti populations from different geographic locations and one A. m. bervoetsi population were compared for variation in polymorphism and allele distribution at 10 microsatellite loci. Successful amplification at each microsatellite locus was obtained for all A. m. moucheti specimens while only six loci could be amplified in the A. m. bervoetsi sample, one of which did not show any polymorphism, all specimens investigated showing the same single allele at a homozygous state. This result provides further support for speciation within the A. moucheti group of malaria vectors in Central Africa and reflects, for the first time, genome-wide differentiation between A. m. moucheti and A. m. bervoetsi. Indeed, although the exact cytological location of the microsatellite markers is not known yet, linkage disequilibrium analysis revealed no evidence for genetic linkage between loci, suggesting they provided independent replicates for genome-wide estimation of genetic differentiation between samples. Successful amplification of microsatellite alleles was demonstrated among closely related species such as humans and great apes 33 as well as between sibling species of wasps 34 and members of anophelines species complexes 3536. However, the proportion of loci developed for one species that can amplify in another decreases rapidly with increasing evolutionary distance 3738. These results are, therefore, in straight agreement with previous studies based on morphological 78 and molecular data (mtDNA CytB and rDNA ITS; 10), prompting for elevation of A. m. bervoetsi to full specific rank as a closely related sibling species of A. m. moucheti.

Linkage disequilibrium analysis further ruled out the hypothesis that the fairly high number of loci which were found out of HWE in several collections was indicative of inbreeding and/or population subdivision (within samples). If this was the case, genome-wide signatures of departure from HWE and high linkage disequilibrium between loci should be evidenced, because members of the different sub-populations would have different probabilities to carry certain combinations of alleles 30. Such trends were not observed in the dataset, suggesting null alleles, rather than population subdivision may be responsible for the deviations observed. Null alleles are a common finding in anophelines' population genetics studies 394041. Because the frequencies of such null alleles might differ between sub-populations, they contribute to the overall genetic differentiation between populations. Fst estimates between populations were therefore calculated using all the information available from all loci and all samples.

Fst estimates recorded between A. m. bervoetsi and each of A. m. moucheti populations were very high and statistically significant (Fst>0.34, P < 0.001), falling in the upper range of values reported between well separated anophelines sibling species using various molecular markers 424344. This result, as well as results from the Bayesian analysis clearly identified A. m. bervoetsi as a genetically distinct entity within the A. moucheti group. Accordingly, it seems reasonable to consider this taxon as a full, independently evolving species within the A. moucheti group and, henceforth, to refer to this species as Anopheles bervoetsi. However, considering that A. bervoetsi has never been reported to occur in sympatry with A. moucheti s.s., nor outside of its type locality, this assertion should be validated through traditional crossing experiments, which are yet impossible to implement because members of the A. moucheti group have never been maintained successfully under insectary conditions. Preliminary analysis of 237 field-collected A. bervoetsi specimens after ELISA detected three females infected by Plasmodium falciparum (Antonio-Nkondjio C, Ndo C, Awono-Ambene HP and Simard F, unpublished). Although incrimination of this species as a malaria vector through dissection of its salivary gland still has to be processed, this points to a possible and previously unrecognized role of this mosquito in malaria transmission in Central Africa.

Significant isolation by distance was revealed between mainland A. m. moucheti populations from Cameroon and DRC, separated by distances >1,000 km (Figure 1), suggesting continuous habitat suitability for A. moucheti in these forested environments. Extrapolating the level of differentiation expected under the sole influence of geographical distance between mainland populations and the A. m. moucheti population collected on the island of Bufurima on Lake Victoria (Uganda) showed that the observed level of differentiation was three to nine folds higher than expected. Such high Fst estimates probably reflect the contribution of large water bodies separating this island population from mainland ones, acting as a barrier to gene flow by restricting opportunities for migration between populations, as was demonstrated for A. gambiae in this area 41 and elsewhere 364546. Moreover, significant differences in effective population sizes (Ne) were demonstrated, the island population of Bufurima showing significantly lower Ne than its mainland counterparts, and these might further increase Fst estimates 47. Although Ne estimates based on He are criticizable because they rely on a number of assumptions including correct estimation of microsatellite mutation rate and mutation model, populations at MDE and selective neutrality of the loci, the comparison of Neμ on a relative scale allowed relieving some of these assumptions. Lower effective population size on the islands of Lake Victoria was indeed demonstrated for A. gambiae compared to neighbouring mainland populations 41 and the results presented here suggest the same probably applies within the A. moucheti group. As mentioned above however, estimates of Ne derived from He are sensitive to deviation from MDE. No significant deviation from MDE was evidenced within A. m. moucheti populations, although, to some extent, trends for recent population expansion were revealed for the island population of Bufurima. Colonization of Lake Victoria islands by anthropophilic malaria vectors probably followed initial human settlements on these islands in the early 1900s 4148. Initial founder effect might have occurred at that time but experimental studies and simulations have shown such events are only detectable for a small number of generations before a new equilibrium is reached, especially when considering molecular markers with high mutation rates such as microsatellite loci 2649. Inferences suggesting population expansion are generally more robust but still need to be ascertained through in-depth investigations 25.

Clearly, a more comprehensive picture of the genetic structure and distribution of genetic diversity within and among natural populations of members of the A. moucheti group of malaria vectors would have been obtained with the inclusion in this study of specimens of A. m. nigeriensis. However, as mentioned above, collections conducted in and around the type locality of this species were unsuccessful. Earlier investigations allowed collection of a few representative specimens 10 but sample sizes were far too small to allow reliable microsatellite allelic frequencies assessment. The Lagos area has recently undergone significant levels of anthropogenic environmental reshaping and urban expansion 50 and this might have led to a significant drop in A. m. nigeriensis populations, as was observed for A. m. moucheti in areas of southern Cameroon 4. As formerly highlighted 6, the availability of PCR-based diagnostic tools and other molecular markers and their increased use in routine entomological surveys might allow more refined assessments of the diversity, geographic distribution ranges and relative epidemiological importance of the distinct anopheline species that constitute the extraordinary diverse and fluctuating malaria vector system in Africa. Such knowledge is of paramount importance for a comprehensive, efficient and sustainable implementation of vector control as a means to alleviate the malaria burden in Africa.

In conclusion, this study provides strong support for considering A. m. bervoetsi as a full-rank, genetically independent, species within the A. moucheti group of malaria vectors. The species should henceforth be named Anopheles bervoetsi. However, its epidemiological role as a vector of human malaria parasites still deserves further investigation because roughly nothing is known to date on its biology and behaviour. Isolation by distance seems to be the major factor shaping A. moucheti s.s. populations' genetic structure throughout its distribution range across forested areas of Central Africa but significant geographical barriers to gene flow exist, as evidenced from reduced effective population size and high levels of genetic differentiation observed in a population collected from an island on Lake Victoria. Such genetically isolated populations in a geographically confined environment might be of considerable interest for a safe assessment of new and innovative vector control strategies aiming at population suppression and/or replacement, such as those based on the release of sterile or otherwise genetically altered mosquitoes.

CAN was involved in the study design and implementation, conducted field sampling, microsatellite genotyping and data analysis, prepared and drafted the manuscript. FS supervised the study conception and design, contributed to data analysis and drafted the manuscript. CN participated to field sampling in Cameroon and microsatellite genotyping. LM organized and greatly contributed to sampling in Uganda. PK was involved in molecular analysis and helped with markers selection resources. PAA and DF were involved in the conception of the study and revised the manuscript.