We present GraphDiaries, an interface for temporal navigation in networks changing their topology over time. Navigation along time is necessary for the exploration of states of the network at different time points but in order to explore higher level changes, extended navigation techniques are required. With GraphDiaries we address temporal navigation methods which allow for fast navigation and better understanding of the network's evolution and introduce staged animated transitions with change highlighting between time steps. To that end, we extend conventional node link diagrams with staged transitions and small multiples organized along a timeline, as well as different layout adaption techniques. The design of GraphDiaries is grounded on a taxonomy that we introduce to classify tasks for the exploration of dynamic networks along three dimensions: place, time and network entities. Results of a user study that compared some of the techniques presented in this paper to existing graph navigation interfaces showed that the error rate decreases when exploring high-level changes with our interface. In order to adapt to as many tasks as possible, GraphDiaries is designed as an extension to other graph visualization systems and is independent from particular layout algorithms and other visualization techniques.