The changing animal foraging behavior due to climate variability caused by global environmental
change has significant implications for the persistence of the population and stability in the ecosystem.
The non-stationary and stochastic climatic forcing makes it difficult to predict these responses. The
given research builds a hybrid model, agent-based and machine learning (ABM-ML), in order to
simulate and predict the foraging behavior of animals in a climate fluctuation scenario. Animal agents
are a heterogeneous set of behavioral characteristics that control movement, patch choice, sensitivity
to risk, and energetic state, and the environment as a spatially explicit climate-driven resource field.
In order to enhance predictive accuracy in varying environmental conditions, machine learning is
employed to calibrate important behavioral parameters with the use of empirical movement and
foraging data. Benchmark dataset validation of the model with migratory birds and marine predators
shows high levels of performance with net energy intake coefficients of determination of 0.82 and
patch residence time coefficients of determination of 0.76. The models used to predict the future
climate show considerable drops in the foraging efficiency by the middle of the century and the
average falls of about 18 percent under the RCP 4.5 and up to 37 percent under the RCP 8.5. Findings
also indicate that behaviorally flexible foragers are more affected by efficiency losses to a
disproportionate degree, which implies that plasticity can make them more vulnerable to severe
climate variability. The proposed model will offer an assessment instrument of climate effects on
foraging behavior that is scalable and transparent, and suggests ecological tipping points and adaptive
conservation and management strategies.