This work considers the problem of distributed event-triggered consensus tracking control of multiple high-order uncertain nonlinear systems on a connected graph where the leader's reference information is sent to only a portion of the followers. The communication among the agents and the local control update are performed based on different event-triggering mechanisms which possess different triggering conditions and work independently. To facilitate consensus tracking control design, in each follower agent of the network, a group of distributed and cascaded estimators is constructed for estimating the reference trajectory and its high-order derivatives by event-triggered information exchange among the local neighbors on the communication graph. Theoretical analysis shows that the estimation errors of the reference states decay within a very short time to the neighborhood of the origin and that the Zeno behavior is excluded. Then by utilizing the estimated reference states, in each agent, a local event-triggered robust controller including a nonlinear robustifying term with a state signal dependent gain is proposed. The triggering condition is derived by taking the time-varying control gain into account, and the tracking performance of the local controller is rigorously analyzed. Furthermore, it is shown that the Zeno behavior is also excluded. Finally, simulation examples are provided to demonstrate the performance of the reference estimators and that of the consensus tracking controllers.