ECOLOGICAL DIGITAL TWINS AND PREDICTIVE ARTIFICIAL INTELLIGENCE AGENTS: DYNAMIC DELIMITATION OF ENVIRONMENTAL PROTECTION ZONES IN AFRICAN TRANSPORT CORRIDORS

Environmental protection zones in Africa have historically been delimited as fixed perimeters, while species movements and transport pressures change over time. This article proposes and computationally evaluates the Ecological Digital Twin of Environmental Protection Zones (EDT-EPZ), which integrates transport infrastructure, artificial intelligence, distributed sensing, and dynamic protection polygons. A structured literature review, conceptual modelling, and three agent-based computational experiments were combined in a synthetic scenario parameterized to resemble the Lobito Corridor. The experiments used 1,500 animal agents, 200 replicates, 1,080 daily steps, and a fixed random seed. Mean trajectory coverage was 32.9% under static protection, 86.9% under daily dynamic delimitation, and 89.5% under the adaptive reinforcement-learning condition. The effectiveness index changed from 54 to 43 points in the static regime and from 60 to 77 points in the adaptive regime. With 1 km sensor spacing, modelled detection remained at or above 80% up to 1.57 km from the nearest sensor. Code, synthetic outputs, parameters, and verification hashes accompany the manuscript; two independent executions produced seven byte-identical data files. The findings demonstrate computational reproducibility and internal consistency under the stated assumptions. Empirical validation with field telemetry and sensor observations has not been performed and remains indispensable.