Multiple Insecticide Resistances in the Disease Vector Culex p. Quinquefasciatus from Western Indian Ocean

Several mosquito-borne diseases affect the Western Indian Ocean islands. Culex pipiens quinquefasciatus is one of these vectors and transmits filariasis, Rift Valley and West Nile viruses and the Japanese encephalitis. To limit the impact of these diseases on public health, considerable vector control efforts have been implemented since the 50s, mainly through the use of neurotoxic insecticides belonging to Organochlorines (OC), Organophosphates (OP) and pyrethroids (PYR) families. However, mosquito control failures have been reported on site, and they were probably due to the selection of resistant individuals in response to insecticide exposure. In this study, we used different approaches to establish a first regional assessment of the levels and mechanisms of resistance to various insecticides. Bioassays were used to evaluate resistance to various insecticides, enzyme activity was measured to assess the presence of metabolic resistances through elevated detoxification, and molecular identification of known resistance alleles was investigated to determine the frequency of target-site mutations. These complementary approaches showed that resistance to the most used insecticides families (OC, OP and PYR) is widespread at a regional scale. However, the distribution of the different resistance genes is quite heterogeneous among the islands, some being found at high frequencies everywhere, others being frequent in some islands and absent in others. Moreover, two resistance alleles displayed clinal distributions in Mayotte and La Réunion, probably as a result of a heterogeneous selection due to local treatment practices. These widespread and diverse resistance mechanisms reduce the capacity of resistance management through classical strategies (e.g. insecticide rotation). In case of a disease outbreak, it could undermine the efforts of the vector control services, as only few compounds could be used. It thus becomes urgent to find alternatives to control populations of Cx. p. quinquefasciatus in the Indian Ocean.


Introduction
Several vector-borne diseases, transmitted mainly by mosquitoes, have affected the Western Indian Ocean islands, i.e. the Comoros, the Mascarene Archipelago and Madagascar ( Figure 1). The main ones are malaria transmitted by Anopheles species [1,2], dengue and chikungunya viruses transmitted by Aedes species [3][4][5]), and several filariasis transmitted by Culex pipiens quinquefasciatus [6]. This last species is also suspected to transmit the Rift Valley fever virus in the western part of the Indian Ocean [7][8][9] and is the vector of the West Nile virus and Japanese encephalitis at a worldwide scale [10]. Considerable efforts in vector control have therefore been carried out since the early 50s, in order to limit the impact of these diseases on public health [6,11].
In the western Indian Ocean islands, Cx. p. quinquefasciatus control was mainly implemented through the use of neurotoxic insecticides belonging to the Organochlorines (OC), the Organophosphates (OP) and the Pyrethroids (PYR) families [11][12][13]. The larvae of this species grow easily in breeding sites such as sewers or other wastewater collections [14], where in addition to insecticide treatments, they are also subject to a wide range of xenobiotics. In the field, mosquito control failures have been shown to result from resistant individuals, selected in response to insecticide exposure. However, the xenobiotics used for other purposes than mosquito control are present in Cx. p. quinquefasciatus breeding sites and may also have a role in the development of resistance, as suggested for other mosquito vector species [15][16][17][18]. Despite such repeated failures in mosquito control, very few data on insecticide resistance in Cx. p. quinquefasciatus are available for the Indian Ocean, with the recent exception of La Réunion Island [12].
The two main insecticide resistance mechanisms in mosquitoes are enzymatic detoxification (i.e. metabolic resistance) and target site modification (review in [19,20]). The major classes of enzymes involved in metabolic resistance are cytochrome P450 oxidases, esterases and glutathione-Stransferases. All classes are involved in resistance to different insecticides families, but oxidases play a major role in resistance to PYR, while esterases are mainly involved in resistance to OP. One esterase allele in particular, encoded by the Ester locus and named Ester 2 , is found all over the world in Cx. pipiens populations (in both pipiens and quinquefasciatus subspecies, [21]). Another class of enzymes -the DDTdehydrochlorinases (DDTases)-is particularly involved in resistance to DDT (DichloroDiphenylTrichloroethane, an OC). The second common resistance mechanism is the insecticide target modification. There are only few targets for neurotoxic insecticides, the main ones being the axonic voltage-gated sodium channels (Na-channels), the synaptic acetylcholinesterase (AChE1), encoded by the ace-1 gene, and the synaptic γ-aminobutyric acid receptor (GABA receptor), encoded by the Rdl gene. These different target proteins are highly constrained, and their variability limited, since they play a key role in the nervous system [22]. In Cx. p. quinquefasciatus, the most common target modifications are the L1014F mutation (kdr R allele) in the voltage-gated sodium channel gene, conferring resistance to PYR and DDT, the G119S ace-1 mutation (ace-1 R allele), conferring resistance to OP and carbamates, and the A302S Rdl mutation (Rdl R allele), conferring resistance to the OC dieldrin [12,23,24].
In this study, the resistance levels of Cx. p. quinquefasciatus to the three main insecticide families used in vector control (PYR, OP and OC) were evaluated for the first time in two of the western Indian Ocean islands. Resistance mechanisms were characterized in samples from ten populations distributed throughout Mayotte Island, four populations from Central Madagascar and four populations from Mauritius. Among them, the frequencies of four major resistance alleles were more particularly assessed: kdr R , ace-1 R , Ester 2 and Rdl R .
Mayotte is a small mountainous island, with a majority of coastal roads except for the two roads crossing from east to west. The Northeast is the most densely populated area ( Figure 1). Consequently this area is the main target for insecticide use. Thanks to a sampling scheme covering the totality of Mayotte, we were more specifically able to assess the impact of the heterogeneity of selective pressure and the role of migration on the distribution of insecticide resistance to various insecticides.
Finally, comparisons between Mayotte, Madagascar, Mauritius and a previous study from La Réunion Island [12] allowed a first regional assessment of insecticide resistance in the major disease vector Cx. p. quinquefasciatus in the western Indian Ocean.

2.1: Mosquitoes samples and strains
None of the samples in any location were collected in protected areas, and these field studies did not involve endangered or protected species. No specific permission was required to collect mosquito larvae in public areas, and when collected on private land or in private residences, the owners or residents gave permission for the study to be conducted on their land or in their residences.
Cx. p. quinquefasciatus larvae were collected in ten localities of Mayotte in 2011 (Figure 1), in various types of breeding sites (latrines, sewer…). Larvae were reared to adults in the laboratory and a sample was stored in liquid nitrogen for later analyses. For the Tsoundzou I sample (number 1 in Figure 1), some of the remaining adults were used in biochemical assays (see below) while the rest was bred to establish a laboratory strain (TZ1), and was thus maintained for several generations. Preliminary bioassays were carried out on the first generation (TZ1-F1) to identify the presence of any resistance in the field sample. This strain was then split in four replicates, each selected with a different insecticide to which TZ1-F1 showed resistance, in order to identify the responsible mechanism(s): TZ1-per, TZ1-tem and TZ1-chlor were respectively selected with permethrin (PYR), temephos (OP) and chlorpyrifos (OP) for six generations, and TZ1-diel was selected with dieldrin (OC) for seven generations.
Eight other samples were collected in 2010 from two other Indian Ocean islands, Madagascar and Mauritius (Figure 1; samples described in [25]). Adults were kept in liquid nitrogen for later analyses. One sample -collected at Les Salines in Mauritius (number 12 in Figure 1)-was maintained in the laboratory to establish a laboratory strain. The first generation (MAU-F1) was tested using preliminary bioassays to identify the presence of any resistance in the field sample. This strain was then split in two replicates, each selected with a different insecticide to which MAU-F1 showed some resistance, in order to identify the responsible mechanism(s): MAU-per and MAUchlor were thus respectively selected with permethrin (PYR) during eight generations (MAU-per) and with chlorpyrifos (OP) during nine generations (MAU-chlor).
Finally, two laboratory strains were used in this study. The strain Slab [26] was used as the susceptible reference strain. Slab is susceptible to all the insecticides tested in this study. The second strain, SGaba, shared the same genetic background as Slab but homozygous for the Rdl R allele. This strain was established through eleven backcrosses of 200 females from Montpellier area (France) and carrying the Rdl R allele on males from the Slab strain; at each generation, the progeny were selected using 0.025 ppm of dieldrin to kill the susceptible homozygotes. After these backcrosses, the individuals carrying Rdl R were allowed to mate for three generations, their progeny being selected as above. Crosses between Rdl R homozygotes allowed obtaining the SGaba strain.

2.2: Bioassays
Larval bioassays were performed as described by Raymond et al. [27], using ethanol solutions of permethrin (PYR), DDT (OC), temephos (OP), chlorpyrifos (OP) and dieldrin (OC) (all compounds were purchased from Dr Ehrenstorfer GmbH, Germany). They were conducted on sets of 20 early 4 th -instar larvae placed in a cup with 99 ml of water. One milliliter of the tested insecticide solution was then added in each cup. Assays of four to thirteen doses in a minimum of two replicates per dose were performed for each insecticide. Similar tests were performed in presence of different synergists: (i) the 1,1-bis-(pchlorophenyl) methyl carbinol (DMC, Dr Ehrenstorfer GmbH, Germany), a DDT dehydrochlorinases inhibitor (DDTases, [28,29]), (ii) the piperonyl butoxide (PBO, Dr Ehrenstorfer GmbH, Germany), an inhibitor of some P450 oxidases [30], and (iii) the S,S,S-tributyl-phosphorotrithioate (DEF, Dr Ehrenstorfer GmbH, Germany), an inhibitor of some esterases and some GST [31]. Larvae were exposed to classical sublethal doses of one synergist 4 hours before adding the insecticide (DMC: 2 mg.L -1 , PBO: 5 mg.L -1 , DEF: 0.08 g.L -1 ). In all assays, larval mortality was recorded after 24 hours of insecticide exposure.
Mortality data were analyzed using the Priprobit software [32] based on Finney [33]. It allows testing the linearity of dosemortality response and computing its slope and standard deviation. It also calculates the dose of insecticide necessary to kill 50 % of the tested sample (Lethal Concentration 50, or LC 50 ) and the associated confidence intervals. Finally, it allows the comparison of two dose-mortality lines and the resistance ratios calculation, or RR (= LC 50 of field sample / LC 50 of the reference strain) and the synergism ratios, or SR (= LC 50 in absence of synergist / LC 50 in presence of synergist) and their 95 % confidence interval.

2.3: Metabolic resistance
Biochemical tests were performed on single 2-5 days-old females reared from 1 st -instar larvae from the TZ1 sample to evaluate the activity of the main families of detoxification enzymes. Protein activity was quantified in microplates using the method of Bradford [34], the quantity or activity of the different detoxifying enzymes being expressed per mg of protein present in the homogenate or quantity of molecules metabolized per minute, respectively. Cytochrome P450 monooxygenases (sometimes named Mixed Function Oxidases or MFO) were quantified indirectly by the peroxidase activity of the heme group with tetramethylbenzidine (note that all hemoproteins are thus quantified, not only MFO [35]), esterases by their ability to hydrolyze α-naphthyl and βnaphthyl acetates and GST by their ability to conjugate reduced glutathione and chlorodinitrobenzene [36].
Statistical comparisons of detoxification enzyme activity present in mosquitoes of the susceptible strain Slab and of the TZ1 sample were computed using Mann-Whitney tests with the Statistica software [37].
Over-produced esterases (Ester locus) were investigated in Mayotte samples using starch gel electrophoresis, according to Pasteur et al. [38]; thorax homogenates were used. Esterase activity was revealed using α-and β-naphthyl acetates (as substrates) and Fast Garnett salts as dye. The esterases encoded by the different Ester alleles were identified by their electrophoretic mobility. For Mauritius and Madagascar samples, the Ester locus was studied by PCR as described by Berticat et al. [39], after total DNA extraction of single mosquitoes using a CTAB protocol [40].
Statistical analyses to compare the phenotypic frequencies at the Ester locus between samples were performed using the R software (http://www.r-project.org/) through a generalized linear model (GLM).

2.4: Analyses of target-site modifications
The frequencies of the various phenotypes associated to the presence/absence of susceptible/resistant acetylcholinesterase-1 (AChE1), encoded by the ace-1 gene, were measured in Mayotte samples (except TZ1) using the TPP test described by Bourguet et al. [41]. For TZ1, Mauritius and Madagascar samples, the G119S mutation was investigated using the PCR-RFLP test described by Weill et al. [42], after total DNA extraction of single mosquitoes (CTAB protocol [40]). Both techniques provide the same information on the mosquito phenotypes ([RS] for heterozygotes, and [SS] or [RR] for susceptible and resistance allele homozygotes, respectively), so that their results are identical for a given individual. The choice on which method was used depended on whether the samples were conserved in liquid nitrogen (allowing the rapid TPP test on proteins) or in alcohol (where only the slower PCR test was usable).
For all samples from the different islands, genotyping of kdr and Rdl mutations was performed using a molecular test. Total DNA of single mosquitoes was extracted using the CTAB protocol [40]. The L1014F substitution causing resistance in the kdr gene was identified using the PASA method described in Martinez-Torres et al. [23]. The A302S substitution causing resistance in the Rdl gene was detected using the PCR-RFLP test described by Tantely et al. [12].
The frequency data from the ace-1, kdr and Rdl genes were analyzed using the Genepop software [43]. Hardy-Weinberg equilibrium was checked for each sample. Genotypic differentiation of the different Mayotte samples was computed by comparing each pair of samples with each locus. A p-value correction was applied using the sequential Bonferroni method to take multiple testing into account [44].

3.1: High resistance levels and several resistance mechanisms were identified by bioassays and detoxification enzyme activities
Our first goal was to identify the different mechanisms of resistance present in Mayotte and in Mauritius. To this aim, we analyzed two strains through bioassays and biochemical assays. We used a strain derived from one sample collected in Tsoundzou I (Mayotte) and named TZ1, and another strain derived from a sample collected in Les Salines (Mauritius) and named MAU.
3.1.1: TZ1 strain from Mayotte. Bioassays carried out on the first generation of the TZ1 strain (TZ1-F1) revealed resistance to the four tested insecticides when compared to the susceptible reference strain Slab, i.e. permethrin (PYR), chlorpyrifos and temephos (OP) and dieldrin (OC). Most assays suggested that the TZ1-F1 contained a mixture of susceptible and resistant individuals for the different insecticides tested (data not shown).
This heterogeneity was further investigated by analyzing (a) TZ1 field mosquitoes for the genes coding the target proteins of pyrethroids (kdr gene), organophosphates (ace-1 gene) and cyclodienes (Rdl gene) that can be identified by biochemical or molecular tests, (b) the global activity of different detoxifying enzymes (MFO, esterases and GST) on single mosquitoes of the TZ1-F1 strain, and (c) by re-analyzing dose-mortality responses in sub-strains derived from TZ1-F1 after six or seven generations of selection with permethrin (TZ1-per), temephos (TZ1-tem), chlorpyrifos (TZ1-chlor) and dieldrin (TZ1-diel).
All analyzed mosquitoes from the TZ1 sample were homozygous for the kdr R and the Rdl R alleles (N = 35 and 34, respectively), and 31 individuals out of 35 carried the ace-1 R allele (either homozygous or heterozygous). The distributions of esterases and GST global activity among individuals from TZ1 were significantly shifted towards higher values (p < 0.00005) compared to the distributions for Slab mosquitoes ( Figure 2). In contrast the global quantity of MFO was slightly lower for TZ1 than Slab (p = 0.002).
After six generations of permethrin selection of a replicate of TZ1-F1 (TZ1-per) resistance to this insecticide reached a resistance ratio (RR) of 199, compatible with previous studies [23]. Permethrin bioassays conducted with PBO did not show significant synergy effect (p > 0.05) between TZ1-per and Slab, suggesting that increased MFO detoxification was not involved in the observed permethrin resistance, in good agreement with the observed low global activity of MFO. In addition, TZ1-per displayed a strong cross-resistance to DDT (RR = 804; Table 1 and Figure S1B in supporting information), which was not synergized by DMC (a DDT-dehydrochlorinase inhibitor; Table  2), and which was thus probably due to the kdr mutation. These results indicated that permethrin resistance in TZ1 strains was probably mostly due to the presence of the resistant allele of the Na-channel gene (kdr R ), and that other resistance mechanisms (if present) had probably a very low frequency and a minor role.
Two subsets of TZ1-F1 were selected with temephos (TZ1tem) and chlorpyrifos (TZ1-chlor) during 6 generations. In TZ1tem, temephos resistance reached a relatively high level (RR = 86), and this resistance was synergized by DEF; however the DEF synergism ratio (SR) of Slab was higher than that of TZ1-tem (19 versus 9.9), and the temephos resistance observed in TZ1-tem could not be attributed to the increased esterase detoxification (Table 1 and Figure S1E in supporting information). In a manner similar to TZ1-chlor, chlorpyrifos resistance was particularly high (RR = 8070) but the addition of DEF yielded no effect (SR = 1.1 vs. SR = 275 for Slab; Table 1 and Figure S1F in supporting information), suggesting that detoxifying esterases were not involved in the observed resistance. These results were unexpected considering the high esterase activity observed in the TZ1 field sample with α-and β-naphthyl acetates, two substrates known to be hydrolyzed by overproduced esterases involved in OP resistance [36]. It must be noted that ace-1 R , which confers a <10-fold resistance to temephos and a ~20-fold resistance to chlorpyrifos, was probably also selected in the TZ1-tem and TZ1-chlor substrains; this resistance is not synergized by DEF. However, detoxifying esterases and ace-1 R when associated in a same mosquito provide a resistance that is mostly additive [45]: the extremely high chlorpyrifos resistance recorded in the TZ1-chlor strain is thus particularly difficult to explain. Studies on a Tunisian strain [46] reported a ~10,000 fold resistance to chlorpyrifos that, as in TZ1-chlor, was not synergized by DEF.
Finally, a subset of TZ1-F1 was selected with dieldrin (TZ1diel) during seven generations; resistance reached a high level (RR = 493 (419-574); Table 1). There is presently no known mechanism of dieldrin detoxification. To assess the level of resistance in a strain homozygous for the Rdl R allele of the synaptic GABA in absence of other resistance genes, we established the SGaba strain, a strain carrying this allele and sharing the same genetic background as Slab. Bioassays were conducted on this strain with dieldrin: while Slab LC 50 was 9.5 x10 -4 (6.8 x10 -4 -12.6 x10 -4 ), SGaba LC 50 reached 0.25 (0.21-0.30), thus RR = 264 (166-437), which is coherent to a previous study of the Rdl mutation (RR = 196, 176-216; [47]). This is not very different from the resistance level displayed by TZ1-diel, so that it is reasonable to assume that Rdl R , which was fixed in this sample (N = 34, Table 2), explains most of this resistance.
3.1.2: MAU strain from Mauritius. Bioassays conducted on the first generation of the MAU strain (MAU-F1) showed moderate resistance to chlorpyrifos (OP) and dieldrin (OC) and a large resistance to permethrin (PYR). In the MAU field sample, no ace-1 R and Rdl R allele was observed (N = 23 and 24, respectively) and kdr R was found to be present in 8 of the 24 mosquitoes analyzed. The detoxifying enzyme activity was not studied in the MAU-F1 strain before it was split in substrains for selection with the different insecticides.
Resistance to permethrin reached a high level (RR = 641) in the MAU strain selected with permethrin for eight generations (MAU-per), compatible with previous studies [23]. As in Mayotte, MAU-per also presented a strong cross-resistance to DDT (RR = 605; Table 1), which was not synergized by the dehydrochlorinase synergist DMC. In addition, permethrin bioassays in presence of PBO synergist showed a significantly greater synergy in MAU-per than in Slab (SR = 15 and 2.8 respectively; Table 1 and Figure S1C   Even if only a low tolerance to chlorpyrifos was found in the MAU-F1 strain (RR = 5.2), selection with chlorpyrifos for nine generations has resulted in a sharp increase in resistance (RR = 6024 in MAU-chlor). Tests with synergists were not performed on this strain.

3.2: High heterogeneity of resistance genes in the Indian Ocean islands is revealed by biochemical and/or molecular identification tests
The polymorphism and distribution of four resistance genes were investigated, i.e. the three genes encoding target proteins (kdr, ace-1 and Rdl, Table 2) and a gene encoding detoxifying esterases (Ester, Table 3), in samples collected from 10 populations in Mayotte, 4 populations in Mauritius, and 4 other populations in Madagascar (Figure 1). Three of these four genes were also studied in La Réunion [12].
The kdr R mutation, identified using a PCR test, was observed in all the field samples of Mayotte where its frequency was high, ranging from 0.90 to 1 (mean frequency = 0.98 islandwide). In Mauritius, the resistance allele kdr R was present in all samples and had frequencies ranging from 0.05 to 0.39 (mean frequency = 0.18). Finally in Madagascar, the kdr R allele displayed frequencies ranging from 0.42 to 0.68 (mean frequency = 0.51). The distribution of kdr in La Réunion is unfortunately unknown, but bioassays and PCR-tests showed that it was present in the island [12]. The kdr R allele is thus widely distributed among the four islands where it can provide a strong resistance to PYR and DDT.
The Rdl R mutation, identified using a PCR test, showed a more restricted distribution than kdr R . It was observed in the nine studied samples collected in Mayotte. It was fixed in the TZ1 field sample and had frequencies ranging from 0.10 to 0.75 in the other Mayotte samples (mean frequency = 0.38, Table 2). Three samples (numbers 2, 6 and 10) showed a significant deficit of heterozygotes (p < 0.05), which was probably due to a Wahlund effect [48], i.e. a mixture of distinct sub-populations with reduced gene flow. The situation is quite similar to the one in La Réunion [12], where the frequency of Rdl R ranged from 0.08 to 1 (mean frequency = 0.56). Finally, the Rdl R allele was not found in any of the four field samples collected in Mauritius and was observed in a single mosquito (heterozygous) among the four studied samples from Madagascar.
The polymorphism of the ace-1 gene can be detected indifferently through a molecular PCR-RFLP test or through a biochemical assay, TPP. Both tests were used in this study. In Mayotte, the TZ1 field sample was analyzed with the PCR-RFLP test and ace-1 R was found to have a frequency of 0.61. The nine other samples were investigated using the TPP test.  The resistant allele ace-1 R was present throughout the island, with frequencies ranging from 0.15 to 0.61 (mean frequency = 0.39, Table 2). Among the ten Mayotte samples, four showed significant deviations from Hardy-Weinberg expectations due to an excess of heterozygotes (p < 0.05). Over all samples there was a close-to-significant excess of heterozygotes (p = 0.08). Such excesses of heterozygotes suggested the presence of duplicated haplotypes combining a susceptible and a resistant copy of the ace-1 gene (allele ace-1 D ) [49][50][51]. Such haplotypes have been identified in three samples of Mayotte by crossing experiments (as described in [50]) and are currently being further analyzed (unpublished data). phenotype was found in all tested samples, with frequencies ranging from 0.34 to 0.81, and a mean frequency of 0.59. In Mauritius, the resistant phenotype [Ester 2 ] was found in the four samples, with frequencies ranging from 0.63 to 0.88 (mean frequency = 0.76, Table 3). In Madagascar, the [Ester 2 ] phenotype was found in the four studied samples at very high frequencies (from 0.86 to 1, mean frequency = 0.96). Finally in La Réunion, the frequency of the [Ester 2 ] ranged from 0 to 0.88 (mean frequency = 0.18) [12].

3.3: Spatial distribution of resistance genes in Mayotte
Thanks to our sampling scheme all across the island, it was possible to analyze the distribution of the resistance alleles. Apart from the kdr R allele, which was close to fixation all over the island, the resistance alleles analyzed here displayed structured distributions (Figure 3).
Analysis of the Ester 2 phenotype distribution did not show any particular pattern (Figure 3). Samples were statistically grouped according to their frequency as follows: two groups contained samples presenting no significant differences (samples 2, 3, 5 and 8, and samples 4, 6 and 10), although there was a significant difference between these two groups. Samples 7 and 9 were different from all other samples with, respectively, the lowest and the highest frequency of the island (0.34 and 0.81). However, these differences displayed no clinal pattern across the island.
Concerning ace-1, the resistant allele displayed a strong and heterogeneous spatial pattern, the average ace-1 R frequency decreased from east to west (Figure 3). Statistical analyses revealed four groups showing significant genotypic frequency differences. The first group, in the east of the island, was formed by samples 1, 2, 3, 9 and 10, with an average ace-1 R frequency of 0.54. The second and the third groups were formed by samples 3, 4, 8, 10 and samples 3, 4, 5, 6 and 8 with respective average ace-1 R frequencies of 0.39 and 0.32. The last group, formed by samples 5, 6, and 7, had an average ace-1 R frequency of 0.21. The four groups partially overlapped (leading to five frequency classes, see Figure 3) and ace-1 R frequency decreased as one moved away from Tsoundzou and Kaweni (samples 1 and 2, north-east) or from M'Tsamoudou (sample 9, south-east).
The Rdl alleles also displayed a marked variation in their spatial distribution over Mayotte (Figure 3). The resistant allele Rdl R frequency seemed to decrease as one moved away from Tsoundzou (sample 1), as shown by the negative correlation between Rdl R frequency and the distance from there (Pearson correlation: r = -0.89, p = 0.001). Rdl R frequencies ranged from complete fixation in Tsoundzou I to 0.10 in samples 4 and 9, respectively the most eastern and the most southern collection sites on the island. Statistically homogeneous but overlapping

Discussion
In the Indian Ocean, the mosquito Cx. p. quinquefasciatus is an important vector of several diseases, including filariasis, Rift Valley fever and West Nile viruses, and the Japanese encephalitis. In this study we investigated its status of resistance to the most commonly used insecticides to control its population densities (including the diseases it transmits). So far, there is no available data for the western Indian Ocean islands, except for La Réunion [12]. We investigated Madagascar and two of the main archipelagos -the Comoros, Mayotte and the Mascarenes (Mauritius)-in order to build the first regional assessment of insecticide resistance for this important vector.

a): Resistance to a large variety of insecticides is widespread in the western Indian Ocean
In the western Indian Ocean, Cx. p. quinquefasciatus presents resistances and/or resistance mechanisms to all the main insecticide families used so far in vector control, i.e. PYR (permethrin), OP (chlorpyrifos, temephos) and OC (dieldrin, DDT). The most common resistance mechanism to PYR is the kdr R mutation, which also confers resistance to DDT (OC); it was found through the whole region. In La Réunion and Mauritius, metabolic resistance due to an increased MFO detoxification was also present, but it was not found in Mayotte. The level of resistance to PYR in Mayotte and Mauritius is high, as expected from the presence of kdr R in this species [23].
OP resistance through esterase overexpression, especially the Ester 2 allele, is widespread and found at high frequencies in all the sampled western Indian Ocean islands. The ace-1 R mutation is also present in the area, although is less common (ex. not found in Mauritius). Our knowledge of the resistance to chlorpyrifos conferred by these two resistance genes does not explain the high resistance to chlorpyrifos (OP) observed after the selection of Mayotte and Mauritius field samples (TZ1 and MAU, > 6,000 folds after selection). Such an extremely high resistance to this insecticide has only been reported in Tunisia (> 10,000 folds; [46,52]) where it involved a new gene (named G) associated with resistant ace-1 R . It is possible that this gene is present in Mayotte and Mauritius, and possibly in other Indian Ocean islands, but further studies are needed to confirm it. Finally, dieldrin (OC) resistance through the Rdl R mutation has also been detected in some of the sampled islands, but not all.
In conclusion, in this study we used different and complementary approaches to describe the variety of resistance mechanisms in the Indian Ocean islands (bioassays, measures of enzyme activities, molecular identification of target-site mutations). Clearly, resistance to the most-used insecticides families is widespread at a regional scale; however, the distribution of these resistance mechanisms is quite heterogeneous among the islands.

b): Regional heterogeneity of resistance is probably due to vector control practices
There are indeed important differences among the western Indian Ocean islands for the frequencies of the different resistance mechanisms present. This is particularly the case for the kdr R allele: it is close to fixation in Mayotte (mean frequency = 0.98), but less frequent in Madagascar and Mauritius (0.51 and 0.18, respectively). In La Réunion, the kdr R frequency is unknown, but the allele was found in a strain selected with permethrin and DDT was used in the island for malaria control [12]. These differences are probably related to the insecticides used in vector control: in Mayotte, DDT was used from 1973 to 1984 [11], and then replaced by deltamethrin (PYR). This insecticide is still used for indoor residual spraying (IRS) and deltamethrin-treated nets have recently been distributed in the island (Zumbo, pers. com.). Forty years of such intense selection pressure on the sodium channel gene, the common target of PYR and DDT, explain the near-fixation of this allele in this island. The lower kdr R frequencies observed in Madagascar and Mauritius seems to indicate that the selection pressures on this gene, i.e. the intensities of PYR and DDT treatments, are certainly less important in these two islands. This seems surprising for Mauritius, as DDT has been continuously used from 1946 to 2011 for malaria control [13]. However, from 1990 on, the doses used could have been low enough to weaken the selection pressure intensity (ex. only 2 rounds per year of DDT spraying around the airport, [13]). The kdr R distribution could also be structured (in Mauritius, kdr R frequency ranged from 0.05 to 0.4, Table 2). Finally, kdr R may also have been selected in La Réunion and Mauritius by reinforced vector control of Aedes species following chikungunya and dengue outbreaks [12,53].
Metabolic resistance to PYR is also contrasted between the islands, as MFO implication in resistance has been detected in Mauritius and La Réunion [12], but not in Mayotte (no available data for Madagascar). Considering the intensity of PYR used in Mayotte, the fact that no MFO-based metabolic resistance has been detected is surprising. One plausible explanation is that at the time of this study temephos (OP) was still intensively used in Mayotte (see below): temephos is bio-activated to temephosoxon (the toxic form) by oxidases [54]; if the same oxidases are implicated in both temephos activation and permethrin resistance, it might thus be possible that the intense use of temephos in Mayotte could have led to a counter-selection of oxidases. Further studies are required to establish this point.
Heavy uses of OP insecticides have been documented in most of these islands: temephos was used for vector control in Mayotte from 1973 to the end of 2010 ( [11]; Zumbo, pers. com.), until 2006 in La Réunion [12], and from 1975 to at least 2008 in Mauritius [13]; no information is available for Madagascar. In all four islands, Ester 2 is present at high frequencies, with some samples reaching frequencies of 0.8 in Mayotte, 0.9 in Mauritius and up to 1 in Madagascar and La Réunion (Table 3 and [12]). It suggests a relatively early spread of this resistance allele in the Indian Ocean, consistent with its highly invasive character [21,55], and appears a testimony of high OP selection pressure in all islands. However this selection may be due also to other OP and carbamates, for example those intended for agriculture and domestic usages.
Consequently, the more contrasted distribution of ace-1 appears surprising. The ace-1 R allele is indeed only present in La Réunion and Mayotte (plus one heterozygote in Madagascar), and at much lower frequencies than Ester 2 ( Table 2 and [12]). The absence of ace-1 in Mauritius is particularly puzzling: temephos is indeed used since 1975, many exchanges occur between the different Indian Ocean islands; furthermore all susceptible individuals tested (data not shown) displayed a 119 codon allowing the G119S mutation in one step [56]. One potential explanation is that, as ace-1 R provides low resistance to this OP (RR < 10, [27]), this limited advantage could, in certain treatment conditions, be unable to compensate its high fitness cost [57][58][59][60][61]. Another explanation for the discrepancies between Ester 2 and ace-1 R frequencies could be that Ester 2 would be selected by some other products (ex. agriculture), not necessarily used in vector control, and for which ace-1 is not the target. Esterases are indeed generalist detoxifying enzymes, able to provide protection against a large array of xenobiotics, including other insecticide families (ex. most PYR; [62,63]). Another observation is that the frequency of the ace-1 resistance allele is very different between La Réunion and Mayotte: it is much lower in the first than in the second island (from 0 to 0.29, mean = 0.05, and from 0.15 to 0.61, mean = 0.39, respectively, Table 2 and [12]). This may be due to the presence of ace-1 duplications in Mayotte, which were not found in La Réunion [12]. These duplicated alleles have been shown to provide resistance while reducing its fitness cost [50]. Although their frequencies still need to be evaluated, they may partly explain why ace-1 resistance is more frequent in Mayotte. An alternative but not exclusive explanation could be that ace-1 is currently invading the Indian Ocean from the northwest, i.e. recent importation from Eastern Africa or local mutation [12], which would explain why it is more frequent in Mayotte than in La Réunion, and so far absent or quasi-absent in Mauritius and Madagascar. Only long-term studies documenting the dynamics of the different resistance genes could help solve this issue.
Finally, the Rdl R allele conferring resistance to dieldrin exhibits a distribution very similar to that of ace-1 R : it is only found in Mayotte and La Réunion (and only one heterozygote in Madagascar) ( Table 2 and [12]). Before being banned in France, dieldrin was the only insecticide targeting the GABA receptor used for vector control in these French overseas departments: it has been used in Mayotte from 1952 to 1958 [11], but never in La Réunion [12]. As the dieldrin half-life is 7 years in the soil [64], it is nevertheless unlikely that this legally-used dieldrin persisted in the environment to explain the current resistance. However, other insecticides, such as lindane and fipronil, target the GABA receptor [65] and are respectively used by veterinarians and against termites [12]. Traces of these compounds have been reported in La Réunion coastal waters, as well as traces of dieldrin, probably from illegal uses [12]. This could explain the selection of Rdl R , although the presence of an unknown source cannot be excluded.

c): Local gradients in resistance frequency reveal heterogeneous insecticide pressure
At a local scale, the distribution of kdr R , Rdl R , ace-1 R and Ester 2 were investigated in samples from 10 populations of Cx. p. quinquefasciatus throughout Mayotte. As a similar sampling scheme was performed in the previous study of La Réunion (except for kdr, [12]), we were able to compare the two islands and found that the distributions of resistance genes are particularly congruent.
Three of the genes present evidence of a strong structuration both in Mayotte and La Réunion, i.e. Rdl R , ace-1 R and Ester 2 ( Figure 3 in the present study and Figure 2 in [12]). No clear spatial pattern emerged for Ester 2 , either in La Réunion or in Mayotte (Figure 3): the gene was relatively frequent in both islands, with consequent variations between samples from different populations that probably reflect the heterogeneity of the selective agents in the environment. As discussed above, these selective agents may be the OP used in vector control, but also other xenobiotics, not used for vector control. Moreover, as Ester 2 can be relatively costly [57][58][59][60][61]66], the heterogeneity in its frequency distribution within an island could reflect a heterogeneity in the selective pressure intensity, i.e. the quantity of pesticide used.
Finally, both Rdl R and ace-1 R showed gradient frequency distributions: in Mayotte both decreased from east to west, while they decreased from northwest to southeast in La Réunion; in both islands this gradient reflected the decreasing human population density gradient ( Figure 3 in the present study and Figure 2 in [12]). While ace-1 R has been repetitively shown to be quite costly in absence of OP [57][58][59][60][61], few data exist on the potential cost of Rdl R in absence of dieldrin, although it has been shown to usually decline in absence of the insecticide [67]. Their clinal distributions are thus probably the result of a more intense selection in the most populated areas associated with a decline due to their cost in less treated/less populated areas, with migration redistributing the different alleles [57]. Again, the source(s) of this selection is(are) not clearly identified for Rdl R , while OPs or carbamates are the most probable cause for ace-1 R .

Conclusion
The status of Cx. p. quinquefasciatus insecticide resistance in the Western Indian Ocean is particularly worrying. Indeed, resistance mechanisms to all the most commonly used neurotoxic insecticide families (PYR, OC and OP) are found over the entire region. Both site mutations (kdr R , Rdl R , ace-1 R ) of their main targets (respectively, sodium channels, GABA receptor and AChE1) and metabolic resistance mechanisms (Ester 2 , MFO) are present at a regional scale, sometimes close to fixation in the natural populations of this mosquito. Even more, a not yet identified mechanism providing extreme resistance to chlorpyrifos in mosquitoes carrying ace-1 R , and duplicated alleles of the locus ace-1 are present. This type of multi-resistance is not uncommon and rather reflects the situation of many areas across the world for several mosquito species [68][69][70][71][72][73][74][75].
This resistance diversity gravely reduces the capacity of its management. Classical strategies indeed consist in insecticide family rotation, which would be very difficult at this stage considering the variety of mechanisms already present. It is even more worrying as such strategies rely on the existence of resistance fitness costs: unfortunately, resistances with reduced cost have already appeared (ex. ace-1 duplications) and different resistance mechanisms can act in synergy (ex. the presence of kdr R largely limits the cost of ace-1 R ; [61]). Moreover, large heterogeneities in the frequencies of the various resistance alleles were found, so that the control strategies should be precisely designed to adjust to the particular situation of each island.
In the case of Cx. p. quinquefasciatus in the Indian Ocean, the main risks are epidemics of Bancroftian filariasis [6,76,77] and the Rift Valley fever virus [7,9,78]. In case of an outbreak of either of these diseases, these already-established resistances could undermine the efforts of the vector control services. Temephos could still be used in emergency cases, but to do so the European legislation on this product should be changed, and the presence of resistance alleles could reduce its utility on a long-term basis.
In the meantime, alternative insecticides could also be potentially used to control an epidemic. Insect Growth Regulators (IGR) are efficient, but show very low persistence on Cx. p. quinquefasciatus at the currently used doses ( [79,80]; Pocquet et al., unpublished data). Bti toxins (extracted from Bacillus thuringiensis var. israelensis) could be a serious alternative, however their residual efficiency is relatively short, particularly in tropical environments and polluted water ( [81]; Pocquet et al., unpublished data).
Thus it becomes urgent to find alternatives to control populations of Cx. p. quinquefasciatus in the Indian Ocean. One of the most promising research paths is the development of Incompatible Insect Techniques (IIT). A first step in the development of these strategies has recently been performed: Atyame et al. [25] have introduced in a Cx. p. quinquefasciatus line a strain of Wolbachia incompatible with the strain present in the Indian Ocean. Cx. p. quinquefasciatus males of this new line could sterilize all females on most of the Indian Ocean islands. The development of such techniques would allow fighting effectively and specifically Cx. p. quinquefasciatus in this part of the world.