Spatio-temporal modelling and prediction of Anopheles mosquito abundance in Tanga and Unguja, Tanzania: climatic drivers and insights for malaria early warning and vector control strategies

BACKGROUND: Anopheles mosquitoes, vectors of human malaria, are highly sensitive to environmental change. As climate alters temperature and precipitation patterns, mosquito populations may shift in sibling species composition, location and timing, altering transmission dynamics. Understanding these patterns is key for malaria control. This study explores links between meteorological factors and Anopheles abundance across a diversity of sites in Tanga and Unguja, Tanzania, to predict mosquito peaks and support the development of early warning systems for malaria outbreaks.

METHODS: Adult Anopheles mosquitoes were sampled monthly from September/October 2021 to December/September 2023 across 11 sites in Tanga and 4 shehias in Unguja. Spatio-temporal Generalized Additive Mixed Effects Models (GAMMs) were employed to assess the influence of meteorological factors on Anopheles abundance. Models were built and validated using mosquito counts alongside climate covariates obtained from Copernicus ERA5-Land and NASA's POWER platforms.

RESULTS: A total of 4312 adult Anopheles mosquitoes were sampled in Tanga and 1450 in Unguja. The GAMM revealed region-specific climatic drivers. In Tanga, Anopheles abundance increased with higher maximum temperatures but declined with higher minimum temperatures. In Unguja, abundance exhibited a non-linear relationship with mean temperature, peaking below 27.5 °C and decreasing thereafter. Precipitation in Tanga positively influenced Anopheles abundance both concurrently and with a two-month lag, whereas in Unguja only the two-month lag effect was significant. Relative humidity exhibited a non-linear effect in both regions, with higher humidity associated with increased abundance. The GAMMs demonstrated strong predictive performance as evidenced by low MAE and RMSE, Theil's U < 1, and correlation exceeding 0.8 between observed and predicted values. Importantly, the models accurately forecasted Anopheles abundance peaks in Unguja in November 2023, preceding the reported malaria surge in Zanzibar in late 2023 and early 2024, highlighting its potential as a proxy for malaria risk and a scalable early warning system to support proactive targeted vector control.

CONCLUSION: The study highlights the importance of integrating meteorological variability into mosquito surveillance and control. The spatio-temporal GAMM captured weather-driven mosquito dynamics and predicted surges in Anopheles abundance prior to the Zanzibar malaria outbreak in late 2023. These insights can guide targeted interventions across diverse eco-climatic regions, enhancing malaria vector control.