Wireless Sensor Networks (WSNs) are an important interface between physical and computational environments, where nodes clustering is a common technique to organize its traffic. Although current clustering protocols are focused on various kinds of dynamicity on the network, such as mobility or cluster-head rotations, few solutions consider the readings similarity management, which would provide benefits in terms of better use of compression techniques and reactive detection of anomalous events. This paper proposes a dynamic clustering protocol, called DDFC, that manages spatial similarity between nodes readings. Its operation is based on the biological principles of fireflies to ensure distributed synchronization of the cluster's similar readings aggregations, differentiating thus from the classic use of fireflies. Simulations show that DDFC is capable of maintaining the cluster's readings aggregation synchronized, hence clustering nodes dynamically according to their similar readings.