Photosynthesis is a key driver of the global carbon cycle, and inaccuracies in its simulation can lead to substantial uncertainties in Earth system predictions. It is governed by multiple physiological parameters, many of which exhibit sensitivity to environmental conditions, reflecting plant acclimation over time. In parallel, plant hydraulics, which regulates water transport from roots to leaves and directly influences stomatal behavior and gas exchange, plays a crucial role in shaping photosynthetic responses under varying environmental stress. In this study, we developed a physics-informed machine learning framework that integrates photosynthesis and plant hydraulic processes while learning from multivariate observational datasets. By explicitly incorporating environmental dependencies, the model captures acclimation effects and improves the generalizability of photosynthetic predictions.