The An. gambiae complex is composed of at least seven morphologically indistinguishable species [1, 25] throughout sub-Saharan Africa including neighbouring islands. Among them only An. gambiae s.s. and An. arabiensis are found in Burkina Faso. These species are sympatric in the major parts of the country but the relative frequency of the two species varies in rural and urban areas. Previously, Coluzzi and others  reported penetration of An. arabiensis into towns and cities of the rainy forest zone in southern Nigeria. Kristan and others  reported a similar trend among samples of An. gambiae s.l., in the urban localities of Aiyetoro and Lantoko of Nigeria where the majority of the vector population was identified as An. arabiensis. Although Lemasson and others , showed that An. arabiensis had a lower vectorial capacity than An. gambiae s.s. in Senegal, these results imply an extension/adaptation of this species into/to urban areas. Studies in more locations are needed to further confirm and understand what may be driving the expansion of An. arabiensis in West African cities. An. arabiensis is the most widespread species among the members of the An. gambiae complex and is the most adaptive in respect to feeding and resting choices [28, 29]. In 1986, Robert and others  studying malaria transmission in Bobo-Dioulasso city including Dioulassoba (the same site as the present study) identified only 3% of the mosquito malaria vector population as An. arabiensis. In 1999, Chandre and others  failed to identify any An. arabiensis in the same area. In 2002, Diabaté and others recorded that the malaria vector population in Dioulassoba was composed of 8.3% An. arabiensis. Our results show the advanced infiltration of An. arabiensis into this district where its proportion of the total vector composition now reaches 90% whatever the sampling period. The same situation has also been observed in Kodeni, a peripheral district of the city (>50% An. arabiensis vs 40% An. gambiae s.s.). According to a recent study, An. arabiensis was also reported as the dominant vector in the savannah around Bobo-Dioulasso city suggesting that this infiltration now extends beyond Bobo-Dioulasso . Indeed, while the frequency and distribution of An. arabiensis appears to be growing, the role of secondary vectors such as Anopheles nili which had previously played a local but important role in malaria transmission in rural areas surrounding Bobo-Dioulasso seems to be greatly reduced . The pattern of An. arabiensis expansion in this region could be explained by global ecological changes (such as climate change) or local human activities favouring the colonisation of this species, however, further investigation is needed to examine these two possibilities further. The trend identified in our study and those of others appears to be mirrored in the capital of Burkina Faso, Ouagadougou, where An. gambiae s.s. was formerly reported as the major vector species [32, 33]. However, recently An. arabiensis has been described as the predominant vector species in this city (55% An. arabiensis vs 45% for the An. gambiae M form) . In An. gambiae M and S forms adaptation to different ecological niches is associated with specific chromosomal inversions that appear to confer on the M form traits that allow exploitation of flooded and arid areas and make the S form significantly rain dependant [4, 34]. However, more investigation is needed to understand the genetic basis underlying the adaptation of An. arabiensis either to the humid meridian savannah or to polluted sites in urban areas as seems to be the case in our study.
Considering the changing pattern of vector bionomics in Bobo-Dioulasso over the last decade or so it might be assumed that the malaria transmission potential has altered. In 1986, a low annual entomological inoculation rate (EIR) of 0.19 infected bites per year (i/b/y) in Dioulassoba and 4.6 i/b/y in peripheral districts had been reported . In 2003 Diabaté (unpublished data) recorded higher inoculation rates reaching 57 i/b/y in Dioulassoba and 63 i/b/y in peripheral districts. In the current study our sampling technique did not allow the EIR to be estimated with any accuracy. However, taking into account mosquito infection rates alone it is possible that the transmission intensity may not have changed greatly from that formerly reported in 2003 by Diabaté (unpublished data).
To test the susceptibility of these urban vector populations to insecticides we exposed them to the most commonly used insecticides for public heath purposes and also to DDT. The frequency of two common resistance mechanisms in these populations, kdr and ace-1
, that confer resistance to pyrethroids/DDT and organophosphates/carbamates respectively was also examined. Taking the An. gambiae s.l. population as a whole, resistance was observed to pyrethroids, bendiocarb and most significantly to DDT at both sites. The An. gambiae S form showed the highest levels of resistance to all four insecticides and this was consistent with the high frequency of kdr/ace-1
observed in this molecular form, the former of which appears to be approaching fixation. Because of the relative rarity of the An. gambiae M form it was not possible to assess its resistance status to all insecticides, however, it showed clear resistance to DDT but was fully susceptible to bendiocarb, consistent with a frequency of ~0.5 for kdr and an absence of ace-1
. An. arabiensis populations also showed resistance to DDT, modest levels of resistance to pyrethroids, and were fully susceptible to bendiocarb. In contrast to the An. gambiae S form the L1014F kdr mutation was not found in An. arabiensis suggesting that other mechanism(s) underlie resistance to DDT/pyrethroids. It would therefore be interesting in future to investigate the role of detoxifying enzymes such as esterases, cytochrome P450s and glutathione-s-transferases in resistance. No An. arabiensis was found to carry the ace-1
mutation correlating with the results of bioassays showing that An. arabiensis remains susceptible to carbamate compounds.
The resistance level observed in An. gambiae s.s. at the two sites may be partly explained by the use of insecticides for crop protection in Kodeni where farmers apply large amounts of insecticides for vegetable production and the domestic use of insecticides such as aerosol sprays or coils in Dioulassoba. In 1993 in Bouaké, Côte d'Ivoire, the presence of permethrin resistance was attributed to widespread use of pyrethroids in households .
In this study An. gambiae s.s. mosquitoes carrying both kdr and ace-1
mutations were identified in the field. The presence of both these resistance mechanisms in An. gambiae s.s. from the West of Burkina Faso has been reported previously and would be expected to provide a level of protection to pyrethroids, carbamates and organophosphates . Presently, pyrethroid treated bednets alone or combined with indoor residual spraying remain the primary mechanism to control malaria vectors in these regions. An essential component of effective vector management strategies is the monitoring of vector populations for resistance to the insecticides used for control and the frequency and distribution of mechanism(s) underlying resistance. The data our study provides is therefore useful contemporary information for vector control programmes in Bobo-Dioulasso city.