Exploratory analysis of marine reserve site selection by Monte Carlo mixed integer programming

Abstract

Marine reserve site selection is a complex process that requires balancing ecological conservation, socio-economic considerations, and management feasibility. Traditional multi-objective optimization methods often rely on the weighted sum approach, which introduces subjectivity and uncertainty in weight estimation. To enhance transparency and robustness, this study introduces an exploratory framework that integrates Mixed Integer Programming (MIP), Monte Carlo (MC) experiments, and Decision Tree (DT) analysis. Using Xiao Liuqiu Island in Taiwan as a case study, we ran an MIP site selection model 10,000 times to systematically explore different weighting scenarios for 21 ecological, socio-economic, and management features. The MC results revealed that 90 % of simulations consistently identified a single high-priority conservation zone, which exhibits high ecological, low socio-economic, and high management values. Two alternative zones emerged under specific weight conditions. The DT analysis pinpointed the threshold conditions that drive site selection shifts towards these zones, highlighting the most influential features shaping the decision-making process. By reducing reliance on predefined weights, this approach ensures globally optimal marine reserve configurations while providing decision-makers with clearer insights. The findings contribute to marine spatial planning, adaptive conservation strategies, and stakeholder-driven decision-making, fostering more effective and informed marine protection efforts.