Modelling population dynamics to inform the evaluation of vector control tools in semi-field and field settings

Kipingu, Andrea Martin (2025) Modelling population dynamics to inform the evaluation of vector control tools in semi-field and field settings. PhD thesis, University of Glasgow.

Abstract

Malaria remains one of the most common life-threatening vector-borne diseases worldwide, with an estimated 263 million global cases and 597,000 deaths per year, 94% and 95% of which occur in sub-Saharan African countries, respectively. In sub-Saharan Africa, Tanzania accounts for approximately 3.3% and 4.3% of all malaria cases and deaths, placing it among the leading four countries responsible for just over half of global malaria deaths. Malaria is transmitted to humans through a bite by an infected female Anopheline mosquito. In Dar es Salaam, Tanzania, An. gambiae s.l. (i.e., An. gambiae s.s., An. arabiensis and An. merus) is the most important species in terms of malaria transmission, followed by An. funestus. Vector control remains the most effective strategy against malaria. The main malaria vector control interventions are insecticide-treated bed nets and indoor residual spraying; both were very successful but were not enough to eliminate transmission, so there is a continuous search for new tools and strategies for deployment.

Development of interventions typically starts in the laboratory and then moves to the semi-field system before going to the field; thus, we need robust ways to assess them at all these levels. Experiments for testing vector control interventions in semi-field systems serve as a cost-effective link between laboratory and field trials, enabling researchers to evaluate interventions or their combinations in controlled conditions. One way to achieve reliable outcomes is to design semi-field experiments with adequate statistical power. Evaluating power is crucial for determining necessary resources, including finances, time, and participants. However, power analysis is rarely done, possibly due to limitations in technical skills and the availability of tools such as software.

Furthermore, assessment of interventions in the field settings needs to not only determine the impact on population size but also regulatory processes (such as negative density dependence and positive density dependence known as Allee effect) that regulate populations. Negative density dependence is a regulatory process which typically operates in immature mosquitoes where growth rates decline at high densities, mainly caused by resource competition. Allee effect is another process operating in adult mosquitoes where population crash if density is low, mainly caused by mate limitation. Understanding the impacts on population dynamics and how low populations are regulated in the field settings could provide critical insights into how to improve vector control strategies. This is because at low densities, regulatory processes, particularly negative density dependence and Allee effects, have implications for vector suppression and elimination plans. However, the existence of Allee effects in the field settings with low mosquito population densities and their implications for vector control interventions is still unknown.

The main aim of this PhD thesis was to improve the evaluation of malaria vector control interventions. This was done through a combination of theoretical and statistical modelling approaches applied to both semi-field and field settings.There were three specific research aims: 1) how can vector control experimental designs be improved in semi-field systems? 2) what are the trade-offs between mosquito population regulatory mechanisms at low densities? and 3) do key mosquito population regulatory processes emerge from large-scale vector control?

To achieve aim 1, a simulation-based power analysis framework from a generalised linear mixed model was developed to assess how many chambers, sampling frequency and sampling size in semi-field systems would provide enough power to determine the impact of interaction between two tools, here pyriproxyfen autodissemination and the widespread insecticide-treated bed nets against malaria vector An. arabiensis across a range of commonly used semi-field experimental designs, such as single vs. combined interventions and short- vs. long-term experiments.

Results showed that the higher the effect sizes, the higher the power, but power also increased with the number of chambers, sampling frequency, and number of mosquitoes, while high variation between chambers reduced power. a generalisable power analysis framework was provided and can be used widely for other vector control tools, experimental scenarios and also other vectors.

For aim 2, a simulation model based on an age-structured population model was developed to quantify trade-offs between negative density dependence and the Allee effect and how these impact the outcomes of interventions.Results showed that while in isolation, these mechanisms are not able to drive the population into extinction, their co-existence can accelerate population extinction as populations become smaller. A combination of negative density dependence, the Allee effect, and sustained larvicidal intervention led to a decline in mosquito populations to levels from which they could not recover. Conversely, the combination of negative density dependence, the Allee effect, and short-term larvicidal applications did not decrease mosquito populations to lower levels enough to prevent a rebound. Understanding regulatory processes like Allee effects can support vector control by highlighting resilient and vulnerable aspects of the mosquito’s life cycle stages to interventions, and potentially accelerating malaria elimination.

To address aim 3, a population dynamics model was developed using the Bayesian state-space modelling approach. Initially, the model was fitted to simulated data to determine whether my framework would be able to quantify Allee effects if they exist in the wild. Results showed that the framework was indeed able to capture the life history traits, including negative density dependence and Allee effects. Subsequently, the model was fitted to female adult An. gambiae data from Dar es Salaam, Tanzania, to identify the presence of Allee effects in natural settings and quantify the impacts of a larvicide intervention. Results showed that there was no evidence of the Allee effect in the An. gambiae mosquito data from Dar es Salaam despite the larviciding having reduced the population by 60.92%. When planning for future malaria vector control strategies, it is essential to consider Allee effects, if they exist, fewer resources could result in better outcomes, similar to deploying more resources.

In conclusion, the methods and findings presented in this thesis will help future research to evaluate vector control interventions or their combinations in SFS and field settings. This thesis contributed to a general understanding of the trade-offs between negative density dependence and Allee effects and how they can contribute to vector control and accelerate malaria elimination. The Bayesian state-space modelling framework developed in this thesis will aid further research in identifying Allee effects in different settings with low mosquito population densities.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QR Microbiology; R Medicine > RA Public aspects of medicine > RA0421 Public health. Hygiene. Preventive Medicine
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Biodiversity, One Health & Veterinary Medicine
• Supervisor's Name: Viana, Dr. Mafalda, Johnson, Dr. Paul, Haydon, Professor Daniel and Kiware, Dr. Samson
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85104
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 02 May 2025 08:16
• Last Modified: 02 May 2025 08:19
• Thesis DOI: 10.5525/gla.thesis.85104
• URI: https://theses.gla.ac.uk/id/eprint/85104