The weakest link: uncertainty and sensitivity analysis of extinction probability estimates for tsetse (Glossina spp) populations

Abstract

BackgroundA relatively simple life history allows us to derive an expression for the extinction probability of populations of tsetse, vectors of African sleeping sickness. We present the uncertainty and sensitivity analysis of extinction probability for tsetse population, to offer key insights into parameters in the control/eradication of tsetse populations. MethodsWe represent tsetse population growth as a branching process, and derive closed form estimates of population extinction from that model. Statistical and mathematical techniques are used to analyse the uncertainties in estimating extinction probability, and the sensitivity of the extinction probability to changes in input parameters representing the natural life history and vital dynamics of tsetse populations. ResultsFor fixed values of input parameters, the sensitivity of extinction probability depends on the baseline parameter values. For example, extinction probability is more sensitive to the probability that a female is inseminated by a fertile male when daily pupal mortality is low, whereas the extinction probability is more sensitive to daily mortality rate for adult females when daily pupal mortality, and extinction probabilities, are high. Global uncertainty and sensitivity analysis showed that daily mortality rate for adult females has the highest impact on the extinction probability. ConclusionsThe strong correlation between extinction probability and daily female adult mortality gives a strong argument that control techniques to increase daily female adult mortality may be the single most effective means of ensuring eradication of tsetse population. PLOS1/12 Author summary 1 Tsetse flies (Glossina spp) are vectors of Trypanosomiasis, a deadly disease commonly 2 called sleeping sickness in humans and nagana in livestock. The relatively simple life 3 history of tsetse enabled us to model its population growth as a stochastic branching 4process. We derived a closed-form expression for the probability that a population of 5 tsetse goes extinct, as a function of death, birth, development and insemination rates in 6 female tsetse. We analyzed the sensitivity of the extinction probability to the different 7 input parameters, in a bid to identify parameters with the highest impact on extinction 8 probability. This information can, potentially, inform policy direction for tsetse 9 control/elimination. In all the scenarios we considered, the daily mortality rate for adult 10 females has the greatest impact on the magnitude of extinction probability. Our 11 findings suggest that the mortality rate in the adult females is the weakest link in tsetse 12 life history, and this fact should be exploited in achieving tsetse population control, or 13 even elimination. 128will yield a 22% increase in θ, whereas, at θ = 0.96, a 10% decrease in will only yield 129 an 8.7% increase in θ.