3.1 Experiment 1: pure glyphosate versus glyphosate-based herbicides
There was no effect of glyphosate type (pure or formulation) on larval mortality (model 1: χ²1= 0.88, p = 0.347) but glyphosate concentration (0, 0.05 and 0.1 mg/L) had a significant impact (model 1: χ²1= 6.65, p = 0.036). Contrast analyses showed a significant difference in mortality between the two mosquito groups exposed to glyphosate. Larval mortality was higher in mosquitoes reared in 0.05mg/L glyphosate solution than in mosquitoes reared in 0.1mg/L glyphosate solution (contrast analyses χ²1= 5.55, p = 0.018, larval mortality: 0 mg/L: 10.9%, 0.05 mg/L: 18.7%, and 0.1 mg/L: 10.3%).
Glyphosate concentration and glyphosate type did not have significant effect on both larval development time (model 2: glyphosate concentration: F= 1.93, p = 0.165, glyphosate type: F= 1.87, p = 0.154) and adult mosquito size (model 3: glyphosate concentration: F= 0.12, p = 0.732, glyphosate type: F= 2.08, p = 0.150). As expected an effect of sex was observed, males had a shorter development time (model 2: F= 629.38, p < 0.0001) and were smaller than females (model 3: F= 314.16, p < 0.0001).
While larvae exposed to the highest concentration of glyphosate (0.1mg/L) tended to take a smaller blood meal (0 mg/L: 28.6µg ± 1.9, 0.05mg/L: 29.6µg ± 1.4, 0.1mg/L: 26.7µg ± 1.1), the amount of blood ingested did not vary significantly between treatments (model 4: glyphosate concentration: χ²1= 2.76, p = 0.09; glyphosate type χ²1= 0.12, p = 0.724). No effect of glyphosate concentration and glyphosate type were observed on the number of laid eggs (model 5: χ²1= 2.19, p = 0.139; χ²1= 0.40, p = 0.526, respectively). A positive relationship was observed between blood meal size and the number of laid eggs (model 5: χ²1= 24.39, p < 0.0001).
Midgut dissection revealed that 100% of the mosquitoes fed on infected bird blood were infected with Plasmodium relictum.Glyphosate concentration and glyphosate type had no effect on the oocyst burden of mosquitoes (model 6: χ²1 = 0.11, p = 0.895, χ²1 = 0.02, p = 0.895, respectively). A positive relationship was observed between blood meal size and oocyst burden (model 6: χ²1= 25.56, p < 0.0001).