Spatio-Temporal Dynamics of Aedes Mosquito Populations under Climatic Variability in Bhagalpur, Bihar: A Hierarchical Statistical Modelling Approach

Abstract

Populations of Aedes aegypti and Aedes albopictus are shaped by far more than weather alone. Their abundance reflects the combined influence of mosquito development and survival, availability of breeding habitats, characteristics of the built environment, human-modified surroundings, and climatic conditions. These factors operate at different spatial and temporal scales and often interact with one another. As a result, analyses based only on seasonal averages, simple pairwise correlations, or independent Poisson regression may provide an incomplete or overly precise picture of climate–mosquito relationships, particularly when mosquito counts are over dispersed and the same surveillance locations are observed repeatedly. We therefore propose a pilot statistical framework for prospective Aedes surveillance in Bhagalpur, Bihar, in which climatic exposures may act nonlinearly or after a time lag, while residual spatial and temporal variation is explicitly represented in the model. The proposed field design is based on weekly mosquito surveillance at geographically distributed sentinel sites using a standardized adult-trap or ovitrap protocol. Mosquito counts are recorded together with temperature, rainfall, relative humidity, sampling effort, and selected local environmental characteristics. The primary analytical approach uses a hierarchical negative-binomial model with a log link and an offset to account for variation in sampling effort. Climatic effects may be represented through quadratic or smooth functions, and lagged exposures are included where biologically plausible. Residual differences among surveillance sites are represented through a correlated Gaussian spatial effect, while temporal dependence is modelled using an autoregressive process. The statistical formulation includes the relevant probability mass functions, probability density functions, and cumulative distribution functions. A zero-inflated extension is considered only when the observed field data provide evidence for a distinct structural-zero process. The main inferential questions concern the overall contribution of climate, nonlinear temperature effects, delayed rainfall effects, residual spatial heterogeneity, and temporal persistence. Because a standardized longitudinal Aedes surveillance series is not yet available for Bhagalpur, the statistical framework was