EURO 2025 Leeds
Abstract Submission

283. Efficient Multi-Objective Optimization for Large-Scale Conservation Planning

Invited abstract in session MA-56: Computational Biology & Healthcare, stream Computational Biology, Bioinformatics and Medicine.

Monday, 8:30-10:00
Room: Liberty 1.11

Authors (first author is the speaker)

1. Aboozar Mohammadi
BIOPOLIS Program in Genomics, Biodiversity and Land Planning, CIBIO-InBIO, Campus of Vairão, University of Porto, Portugal
2. Jeffrey Hanson
Carleton University
3. Virgilio Hermoso
Universidad de Sevilla
4. Silvia B. Carvalho
BIOPOLIS Program in Genomics, Biodiversity and Land Planning, CIBIO-InBIO, Campus of Vairão, University of Porto

Abstract

Protected areas are essential for preventing species extinction and keeping biodiversity. They must preserve populations with high nucleotide diversity and populations from different genetic lineages to maximize resilience, given limited conservation funds.
Multi-objective optimization is well suited for finding solutions that can address multiple, often conflicting, goals. However, it has rarely been applied in conservation decision-making.
This work presents a novel multi-objective mixed-integer linear programming (MILP) model for conserving nucleotide diversity and genetic lineages for a set of species across available planning units in a cost-efficient manner. The constraints ensure the selected planning units satisfy the required targets while meeting a pre-defined budget.
To navigate trade-offs between objectives, we introduce a new multi-objective optimization algorithm based on the epsilon-constraint method to approximate the complete Pareto front by computing a list of non-dominated solutions. As advantages, this algorithm is efficiently applicable to other multi-objective MILPs and capable of approximating the complete Pareto front rather than a single solution. Additionally, special attention was given to computational complexity to ensure that the model and algorithm are well-suited for large datasets. Numerical results will be presented to demonstrate the algorithm's efficiency in approximating Pareto frontiers using both simulated and empirical datasets.

Keywords

Status: accepted

Back to the list of papers