MODELLING THE SPATIAL AND TEMPORAL DISTRIBUTION OF TSETSE FLIES FOR TARGETED CONTROL
Stella Gachoki is a PhD student in the department of Natural Resources. (Co)promotors are prof.dr. A.K. Skidmore, dr.ir. A. Vrieling from the Faculty of Geo-Information Science and Earth Observation and dr. D. Masiga from the International Centre of Insect Physiology and Ecology (ICIPE).
Efforts to eradicate African trypanosomiasis require precise information on tsetse fly distribution and abundance, the primary disease vector. However, field-based data on this is limited in both spatial and temporal dimensions. Species distribution models offer a solution by predicting occurrence and abundance using ground observation data correlated with environmental parameters derived from satellite imagery. In this thesis, these models were used to map tsetse fly occurrence (Chapters 2 and 3) and abundance (Chapters 4 and 5) across different geographic scales in Kenya and Rwanda. A summary of each chapter is detailed below.
Chapter 1 introduces the need to control tsetse flies, emphasizing their role as the exclusive biological vectors of African trypanosomiasis — a parasitic disease impacting both humans and livestock across sub-Saharan Africa. Although human trypanosomiasis is clinically well-managed, animal trypanosomiasis remains a significant constraint on livestock productivity in the region. Given the absence of vaccines and the reduced effectiveness of current trypanocide drugs, reducing tsetse populations to levels hindering disease transmission emerges as the most promising strategy. Therefore, proper identification of areas and periods when tsetse flies are a problem is needed for effective implementation of control strategies.
Chapter 2 focuses on the modelling of potential breeding and foraging sites for Glossina pallidipes, a widely distributed species of tsetse fly in Kenya, using satellite-derived information on environmental factors. The chapter integrates occurrence data of two adult life stages, teneral (recently emerged and unfed) and non-teneral (adults that have consumed a blood meal), with satellite-derived environmental data to predict potential breeding and foraging sites respectively. The model achieved a high predictive accuracy and identified woodland and cropland fractions as influential factors. The use of teneral flies as proxies for spatial mapping of tsetse breeding sites on a large scale, provides a practical alternative to labour-intensive manual searches for pupae.
Chapter 3 addresses the challenge of limited funding for extensive in-situ tsetse fly monitoring, proposing a cost-effective strategy involving the transferability of already developed habitat models. The study builds on the reliable models created in Chapter 2 and demonstrates that tsetse habitat models can effectively be transferred to regions with similar environmental characteristics. Location-specific models performed better but their development require ground data. This study marked the first attempt to transfer tsetse habitat models between regions, contributing to cost reduction in surveillance and control efforts.
Chapter 4 addresses the determination of optimal timing and locations for tsetse control in and around the Shimba Hills National Reserve in Kenya. The study utilizes time-series of in-situ tsetse abundance data and satellite-derived estimates of environmental and weather data. Continuous tsetse management within 1 km of the reserve boundary is deemed essential due to consistently high tsetse numbers. Beyond the 1 km mark, increased tsetse populations could be related to higher rainfall amounts one month prior to tsetse catches. Elevated tsetse numbers coincided with areas exhibiting high vegetation greenness, suggesting that controlling tsetse flies in these regions is more effective after increased rainfall and in areas with high vegetation.
Chapter 5 aimed to develop spatial predictive models of tsetse numbers at a national scale for Kenya utilizing limited tsetse trapping data and environmental factors identified in previous chapters. The model achieved an R2 of 0.41, with precipitation and tree cover identified as crucial variables explaining spatial variability in tsetse numbers. However, when extrapolated to entire Kenya, limitations arose due to incomplete in-situ tsetse data to represent all environmental conditions. While known tsetse hotspots were identified, high tsetse numbers were equally predicted in areas beyond the known tsetse belts, possibly indicating suitable habitats but not necessarily presence. This highlights the need for additional tsetse sampling in Kenya, particularly in the northern and eastern regions that currently lack sufficient in-situ data.
Chapter 6 serves as a comprehensive synthesis of the research's main discoveries. It provides an in-depth exploration of the possibilities and challenges associated with understanding the spatial and temporal behaviours of tsetse flies through the utilization of remotely sensed data. The chapter discusses the use of satellite data to address gaps identified in field-based samples through species distribution modelling. Furthermore, an analysis of the implications arising from the research results for targeted tsetse fly control is presented, along with an assessment of the potential environmental impacts associated with various methods of tsetse elimination. Additionally, the chapter outlines prospective avenues for future research and collaborative initiatives aimed at advancing our insights into tsetse flies and their global epidemiological importance.
