Optimal management of harvested population at the edge of extinction

TL;DR

This paper develops an optimal control framework for managing populations at risk of extinction, using PDE models with harvesting and reserve transportation, providing criteria for survival or extinction based on reduced models.

Contribution

It introduces a Galerkin approximation approach for PDE-based population control, deriving convergence, error estimates, and a critical reserve fraction for survival.

Findings

Nearly optimal solutions from reduced logistic ODE models

A critical reserve fraction for survival is identified

The approach bridges ODE and PDE optimal control problems

Abstract

Optimal control of harvested population at the edge of extinction in an unprotected area, is considered. The underlying population dynamics is governed by a Kolmogorov-Petrovsky-Piskunov equation with a harvesting term and space-dependent coefficients while the control consists of transporting individuals from a natural reserve. The nonlinear optimal control problem is approximated by means of a Galerkin scheme. Convergence result about the optimal controlled solutions and error estimates between the corresponding optimal controls, are derived. For certain parameter regimes, nearly optimal solutions are calculated from a simple logistic ordinary differential equation (ODE) with a harvesting term, obtained as a Galerkin approximation of the original partial differential equation (PDE) model. A critical allowable fraction $\underline{\alpha}$ of the reserve's population is inferred from the reduced logistic ODE with a harvesting term. This estimate obtained from the reduced model allows us to distinguish sharply between survival and extinction for the full PDE itself, and thus to declare whether a control strategy leads to success or failure for the corresponding rescue operation while ensuring survival in the reserve's population. In dynamical terms, this result illustrates that although continuous dependence on the forcing may hold on finite-time intervals, a high sensitivity in the system's response may occur in the asymptotic time. We believe that this work, by its generality, establishes bridges interesting to explore between optimal control problems of ODEs with a harvesting term and their PDE counterpart.