PhD Thesis
Rift Valley Fever (RVF) is climate-related arboviral disease of livestock and humans.
Rift Valley Fever epidemics are associated with dynamics of mosquito abundance.
Studies during inter-epidemic periods (IEP) where there is very little or no virus activity
pose challenges including where and how to effectively sample vectors and when should
the next outbreaks are expected. Entomological surveys were conducted on abundance
and distribution of potential RVF vectors in Ngorongoro district of northern Tanzania.
Mosquito sampling techniques and timing were also compared to effectively trap vectors.
Mosquitoes were sampled both outdoor and indoor using the CDC light traps and
Mosquito Magnets. Outdoor traps were placed in proximity with breeding sites and under
canopy in banana plantations in proximity to animals sleeping areas. After every three
hours, inspection was done on each trap to recover any trapped mosquito. Traps were set
repeatedly in each area for three consecutive days and nights during the study period. All
mosquitoes collected were sorted according to site of collection, type of trap and time of
collection. Mosquito species were identified morphologically using specific keys. After
morphological identification, mosquitoes were kept on ice during transportation to
laboratory. Data from this study was used in ecological niche modelling experiment using
maximum entropy (MaxEnt) to predict distributions of vectors (Aedes aegypti and Culex
pipiens complex) in relation to disease epidemics for the current and future climate
scenarios. A simulation model for mosquito vector population dynamics was developed
based on time-varying distributed delays (TVDD) and multi-way functional response
equations implemented in C++ programming language. These equations were
implemented to simulate mosquito vectors and hosts developmental stages and also to
establish interactions between stages and phases of mosquito vectors in relation to host
for infection introduction in compartmental phases. An open-source modelling platforms, Universal Simulator and Qt integrated development environment were used to develop
models in C++ programming language. Developed models include source codes for
mosquito fecundity, host fecundity, water level, mosquito infection, host infection,
interactions, and egg time. Extensible Mark-up Language (XML) files were used as
recipes to integrate source codes in Qt creator with Universal Simulator plug-in. A total of
1823 mosquitoes were collected, of which 87.11% were Culex pipiens complex, 12.40%
Aedes aegypti and 0.49% Anopheles species. About 36.4% of mosquitoes were collected
outdoors using Mosquito Magnets baited with Octenol as an attractant followed by indoor
trapping using unbaited CDC light traps (29.60%). Three-hour mosquito collections
showed differing patterns in activity, most Ae. aegypti species were collected primarily
during the first and last quarters of the day. Cx pipiens complex was active throughout the
night, early evening and early morning then decreased markedly during the daytime.
Ecological niche models predicted potential suitable areas with high success rates for both
species in the current and future climate scenarios. Model performance was statistically
significantly better than random for both species. Most suitable sites for the two vectors
were predicted in central and north-western Tanzania with records of previous disease
epidemics. Other important predicted risk areas include western Lake Victoria, northern
parts of Lake Nyasa, and the Rift Valley region in Kenya. During simulation modelling,
floodwater Aedines and Culicine populations fluctuated with temperature and water level
over simulation period. Simulated mosquito population showed sudden increase between
December 1997 and January 1998, a similar period when RVF outbreak occurred in
Ngorongoro district. Results provide insights into mosquito abundances and distribution
in the district while emphasizing the possibility of using Mosquito Magnets traps for
efficient sampling of day biting mosquitoes. Predicted distributions of vectors provide
guidance for selection of sampling areas for RVF vectors during IEP. Simulation model
results provide new opportunities for climate-driven RVF epidemic modelling.