Mechanistic‐Statistical Inference of Mosquito Dynamics From Mark‐Release‐Recapture Data

ABSTRACT Biological control strategies against mosquito‐borne diseases require reliable field estimates of dispersal and survival of released males. We present a mechanistic‐statistical framework for mark‐release‐recapture (MRR) data that links (i) an individual‐based 2D Itô diffusion model with stochastic death and capture, and (ii) its reaction–diffusion approximation yielding expected densities and trap‐specific captures. Inference is achieved by solving the reaction–diffusion system within a Poisson observation model for daily trap counts, with uncertainty assessed via parametric bootstrap. We first validate parameter identifiability using simulated data that closely mimic the experimental data. We then analyze an urban MRR campaign in El Cano (Havana, Cuba) comprising four weekly releases of differentially marked sterile Aedes aegypti males and a network of indoor BG‐Sentinel traps. A homogeneous mobility model and an urban‐nonresidential heterogeneous alternative gave very similar results for the present dataset. Estimates indicate a postrelease life expectancy of about 5 days (8 days total adult age) and a typical displacement of 180 m after 5 days, with trap efficiency jointly identified alongside movement and survival. Unlike empirical analyzes that summarize trap returns (e.g., exponential fits of daily survival or mean recapture distance), the mechanistic formulation couples movement, mortality, and capture explicitly and reduces biases from trap layout—enabling joint estimation and biologically interpretable parameters. The approach delivers a calibrated, computationally efficient procedure for extracting biologically interpretable parameters from sparse MRR data and offers a principled alternative to empirical summaries for the design and evaluation of sterile insect technique (SIT)–style interventions in urban settings.