Plant material and extraction
The leaves of A. pubescens Benth (Fig 1) were collected in July 2014 at Covè 7°28’25.2’’N latitude; 2°19’13.0’’E longitude) in Benin and authenticated at the National Herbarium of University of Abomey-Calavi (UAC) where it was kept under voucher AAC 188/HNB.
The leaves were shade dried at 25°C ±2°C for 72 hours. Three batches of 200 g of dried leaves were submitted to hydro-distillation in Clevenger apparatus at 100°C for 2 hours. The distilled oil was dried using anhydrous sodium sulphate and transferred into an airtight amber-coloured vial and stored at 4 °C until further use. The yields were averaged over the three experiments of the plant materials.
Chemical analysis of the essential oil of A. pubescens leaves.
Analysis by gas chromatography coupled with flame ionization detection (GC-FID)
The essential oil constituents were analysed by a capillary GC-FID equipped with a Supelco SPB-1 column (30 m×0.32 mm i.d, 0.25 µm film thickness). One µL of the essential oil diluted in chloroform were directly injected into the GC. Helium was used as carrier gas with the flow rate of 6 mL/min and the splitting ratio of 1/17. The inlet temperature was 250 °C/min, 200-310 °C at 20 °C/min, and then maintained at 310 °C for 2 min.
A capillary GC-MS was used on a TR-1MS column (30 m x 0.25 mm i.d., 0.25 µm film thickness). An electron impact was used with ionization energy of 70 eV. Helium was used as the carrier gas at a flow rate of 0.6 mL/min, and the splitting ratio 1/17. The temperature settings were as follows: 70-200 °C at 10 °C/min, 200-300 °C at 20 °C/min, and then maintained at 300°C for 1 min. Inlet and MS transfer line temperatures were set at 250 and 320 °C, respectively. All apparatus and accessories were from Thermo Scientific (Courtaboeuf, France) and software controlled data processing (Chromocard and XCalibur). The identification of the essential oil constituents was based on the comparison of their retention times and their Kovats retention indexes relative to (C8-C20) n-alkanes. Whenever possible, identifications were based on mass spectra of the authentic standard compounds. Otherwise, identifications were performed using published data  and comparison with the NIST mass spectral library.
Three Anopheles gambiae s.s (An. gambiae) laboratory strains that were regulary maintained at the insectarium of the laboratory of Vector-Borne Infectious Diseases at the Institut Régional de Santé Publique Alfred Quenum (IRSP-AQ) of the University of Abomey-Calavi in Ouidah (Benin) were used in this study. Kisumu strain originating from Kenya is a reference strain susceptible to all insecticides . Acerkis strain, which is resistant towards both organophosphate and carbamate based insecticides and is homozygous for (G119S) mutation . Kiskdr strain, which is homozygous for kdrR allele (L1014F) that confers resistance to pyrethroids and DDT . Both AcerKis and Kiskdr strains were supposed to share the same genetic background as the Kisumu strain but differ by the presence of resistance alleles.
The colonies of the three strains were maintained at the insectarium under optimum conditions (25-27°C temperature and 70-80% relative humidity). The third instar larvae, as well as adult females of 3-5 days old of each mosquito strains were used for the bioassays.
The larvicidal properties of the essential oil were conducted according to the standard method recommended by the World Health Organization  with slight modifications. Since the essential oil does not dissolve in water, six different concentrations (1000, 2000, 3000, 4000 and 5000 ppm) of the essential oil were prepared in ethanol 96%. Twenty-five third instar larvae of each strain were gently transferred into a plastic beaker containing 99 mL of water and 1 mL of each prepared concentration was added to obtain test solutions of 10, 20, 30, 40 and 50 ppm. During bioassays test, larvae were exposed for 24 hours at 26 ± 2°C (temperature measured using Waranet kit (Waranet Solutions SAS, Auch, France)) without any food. After exposure, the larval mortality was recorded. Larvae were considered dead when they were not able to move or swim actively when touched. For each strain, four replicates were performed for a total of 100 larvae per concentration. The control group consisted of batches of larvae exposed to water and the solvent alone (ethanol). In total, three different experiments were conducted on three different days.
Impregnation of mosquito nets with the essential oil
Fragments of insecticide free net (13 cm x 13 cm; 169 cm2) were coated with the essential oil.
The masses of essential oil proportional to the net area (169 cm2) per concentration were determined: 9.3, 18.6 and 27.9 mg for the impregnation at 55, 110 and 165 µg/cm2 respectively after preliminary doses screening. A volume of 1.5 mL of ethanol HPLC grade was poured into a Petri dish containing the mass of essential oil corresponding to a given concentration. After complete dissolution, the fragment of the mosquito net was coated with the mixture. The impregnated fragment nets were left to dry at room temperature for 5 minutes to allow the essential oil to adhere to the mosquito net and to completely evaporate the ethanol. After drying, treated fragment nets were maintained in the dark to prevent likely reactions of the essential oil constituents with the light and were stored at 4°C for 2 to 4 hours, time to perform the cone tests. All coated fragment nets used during the day were treated in the morning at the same time. Different coated fragment nets were used in each replicate to avoid the essential oil concentration loss. The nets of the same size were also treated with 1.5 mL of ethanol and was used as control.
The cone test was used to assess the adulticidal activity of the essential oil on the adult mosquitoes. The cone test is an adaptation of the WHO cone bioassay , with the following modification: During the assay, the test operator holds a forearm behind the cone to provide a host for attraction (Fig 2).
Unfed 3-5 days old female mosquitoes of Kisumu, Acerkis and Kiskdr strains were used in the test. On the day of testing, female were starved for 4 hours before testing. Groups of five female mosquitoes were placed into plastic cups and moved into the testing room one hour before testing begins to allow the mosquitoes to acclimatise to room conditions. The fragment nets for test or control were placed over a dedicated hole on the Perspex boards and secured using a clear tape. A second Perspex board was laid on the first board creating a test/control net “sandwich” between the two boards. The cone was placed over the net and plugged above with a piece of parafilm. A batch of 5 mosquitoes was transferred into the cone with the operator’s forearm in position. Mosquitoes were then exposed to the fragments for three minutes. Ten replicates of batches of 5 mosquitoes of each strain were run per concentration of impregnated nets.
Monitoring of the lethal effect of mosquito exposure to the essential oil.
After exposure, mosquitoes were removed from the cone, transferred into a recovery cups and provided with 10% of honey solution soaked on a cotton pad. Mosquito knockdown was recorded at 60 minutes post-test. Mosquito mortality was then recorded every day until the death of the last female of each mosquito strain.
The analysis of dose-mortality responses in larval bioassays was performed using the BioRssay script version 6.2  in R software Version 3.0 . This script calculates the mortality-dose regression using a generalized linear model (GLM). To assess the adequacy of the model, a chi-square test between the observed dead numbers (data) and the dead numbers predicted by the regression is used. It also tests whether the mortality‐dose regressions are similar for the different strains, using a likelihood ratio test (LRT). If there are more than two strains test, it also computes the pairwise test, and corrects it using sequential Bonferroni correction (Hommel, 1988). Finally it computes the lethal concentrations inducing 50% (LC50) and 95% (LC95) mortality recorded in each strain and the associated confidence intervals; the resistance ratios, i.e. RR50 or RR95 (LC50 or LC95 in each strain divided respectively by the LC50 or LC95 of the reference strain) and their 95% confidence intervals. Susceptible or resistant status was defined according to Mazzarri & Georghiou  and Bisset et al.  criteria : RR50≤1 indicates susceptibility to the tested insecticide, while RR50>1 indicates insecticide resistance. For resistance levels, three categories were ranked as follow: low resistance for RR50<5, moderate resistance for 5≤RR50≤10 and high resistance for RR50>10 [34,35]. The times at which 50% or 95% of mosquitoes fell on their back or their side, i.e. knockdown time (KDT50 or KDT95) and their 95% confidence intervals were estimated after probit regression in R software using the package ‘ecotox’  based on the method described by Finney , the difference between two KDT50s was tested using the ratio test developed in Wheeler et al. 2006 . The mosquito survival after exposure to the essential oil impregnated net was analysed by Kaplan–Meier survival curves using GraphPad Prism 8.0.2 software (San Diego, California USA). The Log-rank test was performed to evaluate the difference in survival between the strains. All statistical analyses were set at a significance threshold of p ˂ 0.001.