Since early 1950s, humans have controlled the populations of many agricultural or medical arthropod pests, mostly with chemical insecticides. After years of success, evolutionary adaptations to these new conditions began to occur and resistance spread rapidly; more than 500 species are now resistant to at least one insecticide [1]. Insecticide resistance is a rapid and recent evolutionary phenomenon, providing insight into the processes of adaptation through natural selection, but it has serious economic and public health implications. In the arms race between arthropods and humans, the mosquito Anopheles gambiae, the main vector of malaria, seems to have just moved up a gear with the emergence of a resistant duplicated allele of the gene encoding acetylcholinesterase 1 (AChE1).

AChE1 is a critical enzyme in nerve transmission and the target of two of the most commonly used types of insecticides (organophosphates, OPs, and carbamates, CXs). Like several other mosquito species (including Culex pipiens, the well-studied vector of West Nile virus), An. gambiae displays resistance due to a single amino-acid substitution, from a glycine to a serine at the position 119, in the AChE1 catalytic site (G119S)[2]. In C. pipiens, there is direct and indirect evidence that the resistance allele (ace-1 ^R ) entails a large fitness cost, probably due to the mutated AChE1 having a much lower level of activity. Homozygous ace-1 ^R mosquitoes survive in the presence of insecticide, but are rapidly outcompeted in the absence of insecticide (see review in [3]). Heterozygotes are subject to smaller costs than resistant homozygotes in the absence of insecticide. In treated areas, they survive better than susceptible homozygotes, but are less resistant than ace-1 ^R homozygotes. Due to the patchy nature of mosquito control, the generalist heterozygote is advantaged across treated and non-treated areas, although the more specialist resistant and susceptible homozygotes are locally selected in treated and non-treated environments respectively. Moreover, heterozygotes cannot invade due to the segregation burden leading to the loss of the advantage in half of their progeny.

Several duplicated alleles (ace-1 ^D ) have recently appeared, which link a susceptible and a resistant copy of the ace-1 gene on the same chromosome [4]. Duplication thus creates a "permanent heterozygote" allele. The first case of ace-1 gene duplication was recently discovered in An. gambiae [5]. Molecular analysis showed this duplicated allele (Ag-ace-1 ^D ) to be present at several sites and to have probably spread among the two molecular forms S and M of An. gambiae s.s, by introgression.

Unfortunately, it is not possible to design a simple test for studying the frequency of Ag-ace-1 ^D due to the lack of features specific to this duplication, as with available genotyping methods carriers of this duplicated allele cannot be distinguished from classical heterozygotes. Thus an indirect method previously developed for C. pipiens was used to estimate Ag-ace-1 ^D frequency in the field [6]. The results of this analysis and the potential consequences for An. gambiae population management and on malaria control are discussed.

The frequency of the recently discovered duplicated allele of the ace-1 gene in An. gambiae, Ag-ace-1 ^D , was investigated in natural populations from West Africa by considering the departure from Hardy-Weinberg proportions caused by its presence [6].

Figure 1 shows the predicted spatial distribution of Ag-ace-1 ^D in the S and M forms of An. gambiae, as shown by previous molecular investigations and analyses of heterozygote excess. The probability of Ag-ace-1 ^D being present was significant in nine samples (five after Bonferroni correction) from Ivory Coast (four samples) and Burkina Faso (five samples), in both M (two samples) and S (seven samples) molecular forms of An. gambiae (Figure 1 and Table 2). In these samples, the frequency of Ag-ace-1 ^D was up to 0.65, with the lowest significant frequency being 0.24, consistent with the expected highly conservative output of the method used. Indeed, this method will detect low frequencies only in large samples; for example, Ag-ace-1 ^D was not detected with this method in one of the analysed populations (Boromo, population #26, Table 2), whereas molecular methods proved this duplication to be present [5]. The frequency and the geographic distribution of this duplication are therefore probably underestimated. For example, the analysis of each Benin population independently did not provide any indication supporting the presence of the duplication in this country (Figure 1 and Table 2). Nevertheless, the pooled analysis yields a P-value of 0.052, which points toward the potential presence of Ag-ace-1 ^D as this method underestimate the excess of heterozygotes and thus its frequency. However, more data are required to confirm the presence of the duplicated allele in Benin (Table 2). The complete lack of variation of the Ag-ace-1 ^D sequence over several countries [5] indicates that this allele was generated by a single genetic event and its current distribution suggests that it is probably spreading.

Unfortunately, the spread of this new resistance allele poses a potential major threat to public health, as An. gambiae is the main vector of malaria. Indeed, several studies of a similar allele in C. pipiens have indicated that the duplication entails a lower fitness cost than the single-copy resistance gene, ace-1 ^R [4, 6] (but see [16]). This is probably also the case for An. gambiae, as the mutated AChE1 gene is also associated with a strong decrease in enzyme activity [17]. The presence and spread of the Ag-ace-1 ^D allele may greatly impede An. gambiae control strategies designed to maintain resistance alleles at low frequencies through the use of different insecticides with no cross-resistance in a mosaic or rotation strategy. It has been clearly demonstrated [18, 19] that the efficiency of such strategies increases with the fitness cost of resistance.

Insecticides for controlling malaria vectors are a major weapon in the battle between humans and malaria. Unfortunately, these insecticides exert strong selection pressure on vector populations, causing the spread of resistance genes, such as the resistance allele observed at the ace-1 locus in An. gambiae. The long-term use of an insecticide promotes the selection of new resistant variants, with a high risk of selecting a low (or null)-cost variant. The ace-1 duplicated allele recently appeared in An. gambiae is probably an example of such a low-cost variant. It is shown here that the presence of this duplicated allele, known from the molecular analysis of a few mosquitoes in some samples from Burkina Faso and Ivory Coast [5] is largely distributed in several countries of Western Africa, sometimes at high frequencies, and that it is probably spreading.

To prevent such spreads of resistance genes, it is crucial to develop the largest possible number of complementary means of control (e.g. larval insecticides, mosquito nets, biological agents, etc.) and to use them wisely. However, the emergence of ace-1 duplication in natural populations of An. gambiae, has just given mosquitoes the edge in this particular battle, seriously undermining our efforts to control vector populations and increasing the risk of malaria outbreaks.