2. MATERIALS AND METHODS

2.1 Study area

The study was undertaken in fourteen (14) localities in two bioclimatic zones in Burkina Faso (Fig. 1). The bioclimatic zones include: i) Sudan-sahelian (dry savannah) which extends throughout much of the central part of Burkina Faso and ii) Sudan-savannah (humid savannah) located in the south and west part of the country. Study area is characterized by a rainy period from May to September with average annual rainfall about 1200mm (Namountougou et al. 2019) .

2.2 Cotton pests control strategies

Overall, in the western and central Burkina Faso, cotton remains the main industrial crop and around 90% of total pesticides are applied for its protection. (Ouédraogo et al. 2011). Three strategies to control cotton pests are recommended in Burkina Faso and deployed in the different study sites: i) Conventional cotton cultivation (Banfora, Diébougou, Houndé, Gaoua, Kombissiri, Koubri, Orodara, Soumousso and Vallée du Kou) with high rates of insecticide treatment (N=9 sites). Insecticide applications were done throughout the cotton season (May to September) according to three treatment windows and depending on the abundance and pest species concerned (Table S1). These treatments are based on the temporal rotation of different insecticides including PY, OP/ Carbamates (CX) and neonicotinoids; ii) Organic cotton cultivation (Dano and Tiéfora) without chemical use (N=2 sites) still known as “coton bio”. It has been introduced in Burkina Faso since 2004 but its large-scale production has been supported since 2006 by some development partners who promote this technonology; organic cotton areas extend on limited areas (Table S2); iii) transgenic cotton (Bt ) cultivation (Fada N’Gourma, Koupéla and Manga) with low insecticide use (N= 3 sites). In 2008, as regards emergency and spread of insecticide resistance in cotton pest populations, transgenic cotton was adopted and implemented into Burkina Faso to control pest caterpillars such as Helicoverpa armigera (Raj et al. 2002, Chaturvedi 2007), a pest species that causes severe damages. In this technology, no PY and OP are used.

2.3 Mosquito sampling and rearing

Mosquito larvae were sampled during the rainy season of the years 2008, 2009, 2013 and 2014 from their natural breeding sites in the 14 localities throughout the three cotton cultivation areas of Burkina Faso. In each locality and each date of sampling, all instars of larvae were collected in at least 10 different breeding sites. Specimens collections were pooled per locality. Larvae were brought back to the insectarium of IRSS/Centre Muraz in Bobo-Dioulasso, reared under standard controlled conditions (temperature 25±2°C, relative Humidity: 80± 10% RH and12:12h Light-Dark) and fed with Tetramin baby® fish food every day until adultood. Upon emergence, we identified mosquitoes morphologically according to standard identification keys described by Gillies MT et Coetzee M (1987). Only, An. gambiae complex females were selected and maintained alive on 10% sugar solution until they were used for bioassays.

2.4 Insecticide susceptibility test

Bioassays were performed on 3–4 days old unfed female of An. gambiae complex mosquitoes using WHO test kits (WHO 1998, 2013). Insecticide-treated filter papers at the diagnostic dose as recommended by WHO (WHO 1998, 2016) were tested including: two pyrethroids (PY), Permethrin 0.75% and Deltamethrin 0.05%, one organochlorine, DDT 4%, one carbamate, Bendiocarb 0.1%, and one organophosphate, chlorpyriphos-methyl 0.4%. DDT and Permethrin were tested only in 2008 and 2009 while the three others insecticide were tested in 2008, 2009, 2013 and 2014. It is important to note that in Burkina Faso, permethrin was the reference insecticide until 2014 for long-lasting insecticide nets (LLINs) and after that date it was replaced by deltamethrin which is now the insecticide of preference for LLINs. DDT was chosen to verify the presence of cross-resistance to pyrethroids. The others insecticides belonging to the OP and CX families are used as reference insecticides for IRS. Four replicates of batches of 20–25 test mosquitoes were exposed for 1 hour to each insecticide-treated paper. The laboratory reference strain susceptible to all insecticides, An. gambiaeKisumu, has been tested on all insecticides impregnated papers as a control. Control mosquitoes (N=20–25 females per test population and the Kisumu strain) were tested for the same time to untreated filter papers. After exposure time, mosquitoes were transferred into insecticide-free observation tubes of the tests kit and maintained on 10% sucrose solution at 25 to 28°C. Final mortality in test and control mosquitoes was examined 24 h after exposure. For further molecular analysis, the samples of mosquitoes tested to different insecticides were kept individually at -20 °C according to their phenotypic state (dead or alive).

2.5 Molecular analysis 

Thirty females referring to An. gambiae complex per site were randomly chosen within the group of bioassays specimens for molecular tests. They were identified to species member of the An. gambiaecomplex according to PCR described by Santolamazza et al. ( 2008), and this sampling was considered as representative of the mosquito populations being tested in each locality (Chouaïbou et al. 2008). Their genotype at the kdr locus was determined using the diagnostic tests described by Martinez-Torres et al. (Martinez-Torres et al. 1998).

2.6 Statistical analyses

The threshold of susceptibility was fixed at 98% mortality rate for the six active molecules according to the WHO’s protocol. When the mortality rates were between 90 and 98%, the population was considered as “suspected resistant”. Mortality rates below 90% indicate the presence of insecticide resistance (WHO 2016). Genotypic frequencies at the kdr L1014F loci in An. gambiae populations were compared to Hardy-Weinberg expectations using the Fischer exact test procedure implemented in GenePop (ver.4.1.4) software (Raymond and Rousset 1995).
The impact of cotton pest management practices on kdr L1014Ffrequencies within An. gambiae complex populations was determined using generalized mixed effect models (GLMM). Models were fitted using the glmmTMB function run in the R software (The R Development Core Team 2008). Post-hoc Tukey method was used to do multiple comparison among modalities of the fixed terms using the ‘emmeans’ function of the ‘emmeans’ package (Russell et al. 2019). Also, the impact of cotton pest control practices on kdr L1014F frequencies in An. gambiaes.l. populations was assessed using a Kruskall-Wallis test and a Dunn’s post-hoc test for multiple comparisons (Dinno 2017). A difference is considered as significant when the p-value is less than 0.05.