This paper deals with the dynamic modeling and identification of an electrically driven underwater robot manipulator. The proposed study includes the dynamic modeling of the actuators of the arm as well as the identification of the parameters of the model. The proposed method deals with the specific case of heterogeneously actuated arms, namely arms with actuators behaving differently for each joint, being considered at the kinematic level. Indeed, we show how to estimate the arms parameters when some of their revolute joints are directly actuated by geared motors, while the others are actuated by linear actuators. A minimum set of identifiable parameters is determined, and adequate excitation trajectories are generated and used in the identification procedure. Real-time experimental validation on the manipulator arms of Ifremer's HROV (Hybrid Remotely Operated Vehicle) Ariane underwater vehicle demonstrates that the proposed method improves the estimation of the dynamic model.