State-Dependent Resource Harvesting with Lagged Information about System States

Area of Science:

• Environmental science
• Ecology
• Decision science

Background:

• Markov decision processes (MDPs) are used for sequential decision-making in dynamic systems.
• Natural resource management often involves delayed information about system states.

Purpose of the Study:

• To adapt MDPs for situations with significant time lags between state observation and action.
• To develop optimal decision policies under uncertainty with lagged states.

Main Methods:

• Modified traditional MDPs to incorporate lagged system state observations and previous actions.
• Developed methods to solve these MDPs with both known and uncertain state transition structures.
• Focused on uncertain transition structures, conditioning actions on prior states, actions, and model probabilities.

Main Results:

• Demonstrated a framework for MDPs with lagged states.
• Calculated and simulated optimal, adaptive policies for mallard harvest regulation.
• Showed that using lagged state information improved management performance.

Conclusions:

• The developed MDP framework effectively handles lagged state information in natural resource management.
• Adaptive policies using lagged data can maintain management performance even with uncertain system dynamics.
• This approach is applicable to real-world problems like wildlife harvest regulation.