A new dynamic multicast routing model was proposed in this paper. Specifically, we firstly considered two possible changes in integrated network: node movements and the change of link delay. Next, a mechanism called local rearrangement is used to handle changes in integrated networks. We designed an artificial immune algorithm based on clone process for optimizing the multicast sub-tree within the range of local rearrangement. In order to improve the effectiveness of our algorithm, we introduced gene library to meet the real-time requirement in online multicast routing fields. The simulation results indicated our algorithm balances better in three metrics compared with other two popular dynamic multicast routing algorithms.