Speaker
Description
Many modelled biological systems are data-limited, due to the costs, time constraints, invasiveness or untrustworthiness of data collection. However, ecologists and biologists often possess valuable knowledge about how systems should behave, derived from theory, experiments, or expert understanding. We develop a new statistical framework for incorporating expert knowledge directly into mechanistic models via calibration. We calibrate population models with expert-elicited long-term population sizes, theoretical expectations of a stable coexisting equilibrium, and observed population responses to perturbations. Our results reveal that combining data with expert knowledge in calibration leads to enhanced predictions that better align with the underlying system, insightful inferences from the additional information, and an improved capacity to inform complex decisions.
