Mosquito survival from mark-recapture studies releasing at known age

The expected lifetime (EL) of mosquitoes has been the subject of a large body of historical research, including by mark-recapture (MR) methods. Previous researchers have presented collections of information informally, and the results of a large systematic search are available; however, a formal synthesis of information has not been carried out.

Mosquito studies were systematically selected and re-analysed, and the information pooled to characterise an age-dependent survival model. MR data were obtained from the subset of historical studies in which mosquitoes were released at known ages after emergence. Data were excluded if experiments held characteristics, established by simulation, in which age-dependent estimation would perform poorly. Analysis used the Cormack-Jolly-Seber (CJS) model with age-dependent (Weibull) survival and time-independent capture probabilities.

Shape parameter estimates in many studies showed evidence of increasing mortality with age (senescence, $\alpha >1$ ) but also decreasing mortality ( $\alpha <1$ ) in some, perhaps owing to high mortality early on. The expected lifetime (EL) of females was meta-analysed by the inverse-variance method taking into account parameter uncertainty, giving an overall estimate of 6.68 (95% confidence level [CL] 5.71-7.81) days and genus-specific estimates for Aedes (7.92, 95% CL 5.57-11.2; n = 5), Anopheles (3.61, 95% CL 2.51-5.18; n = 4) and Culex (7.22, 95% CL 6.26-8.33; n = 18).

The study synthesised EL for females overall and for three important mosquito genera, taking account of mosquito age, and systematically restricting to studies with the most acceptable characteristics. The estimated quantities for Aedes and Anopheles require further caution because the number of datasets, after exclusions, is small. A number of studies were excluded on the grounds of uncertain reliability when analysed independently, and it is suggested that a future integrated (Bayesian) analysis could offer an important advantage by utilising such studies, which include older release cohorts. Increased study availability could allow further adjustment for study characteristics, notably spatial configuration.