Table 3.

Summarising the data and estimates for the laboratory experiments. For each laboratory strain, the number of raw data points (N data points) and mosquitoes (n mosquitoes tested) used to fit the base model is provided. Model estimates of the median lethal concentration (LC₅₀) at the strain level were summarised by generating the mean across all MCMC iterations and its 95% credible intervals (CIs).

Resistance status	Strain	N	n	Concentrations tested (% permethrin)	LC50 (% permethrin)		Heterogeneity (% permethrin)			Background mortality (%)		Mortality variability (%)
					Mean	95% CI	LC10	LC90	Difference	Mean	95% CI	Median	95% CI
Susceptible	Kisumu	6	196	0–0.0005	4.9 ​× ​10−5	3.4 ​× ​10−5–6.6 ​× ​10−5	1.1 ​× ​10−5	1.5 ​× ​10−4	1.4 ​× ​10−4	3	0–8	3.1	1.7–6.5
Resistant	Banfora	11	234	0–0.01	2.1 ​× ​10−3	1.2 ​× ​10−3–4 ​× ​10−3	8.4 ​× ​10−4	4.4 ​× ​10−3	3.5 ​× ​10−3	2	0–6	4	3.5–5.3
	FUMOZ-R	9	294	0–0.0032	1.3 ​× ​10−3	1.1 ​× ​10−3–1.5 ​× ​10−3	5.7 ​× ​10−4	2.3 ​× ​10−3	1.7 ​× ​10−3	2	0–6	4.4	3.8–5.8
	Tiassalé 13	8	238	0–1.97	3.9 ​× ​10−3	2.5 ​× ​10−3–5.9 ​× ​10−3	7.4 ​× ​10−4	1.5 ​× ​10−2	1.4 ​× ​10−2	4	0–10	2.3	0.9–5.1
	VK7 2014	8	277	0–0.05	4.8 ​× ​10−3	4.0 ​× ​10−3–5.9 ​× ​10−3	1.7 ​× ​10−3	1.1 ​× ​10−2	8.9 ​× ​10−3	9	0–22	8.5	7–13.5

Notes: Mosquito population heterogeneity is investigated by examining the range of concentrations at which 10% and 90% of the mosquitoes die (mean LC₁₀ and LC₉₀ of all model iterations for each strain) and the difference between these two doses. The amount of background mortality is quantified from model estimates, with the mean of all model iterations for each strain and 95% credible intervals provided above. The amount of variability in mortality is quantified from the average absolute distance of the data points to the best-fit line along the mortality axis, with the median of all model iterations for each strain and 95% credible intervals provided above.

Abbreviations: N, number of data points; n, number of mosquitoes tested.