Scalability in ad hoc networks is a problematic issue, most works presents experimental results for limited number of nodes (100-200) nodes in a field. Various "explicit" clustering techniques have been proposed to improve scalability obtaining successful sessions in fields of 400-800 nodes. However explicit clustering may damage the performances, e.g., sessions breaks due to fast movements of cluster heads and the overhead for the explicit partition to clusters. An alternative to explicit clustering is to use algorithms that are "naturally clustered", i.e., over time arrange the nodes in dynamic hierarchical structures obtaining a similar effect to that of explicit clustering. The explicit clustering is more adaptive than explicit clustering and basically comes without overhead as it does not require an additional protocol for explicit partition of the nodes to clusters and cluster heads. For example if a cluster head moves away from its group another node may replace it without updating its class member. In this work we study the effect of explicit clustering by comparing an advance version of the AODV (a core algorithm in ad hoc networks) with the MRA algorithm that has the naturally clustering property. We cover fundamental aspects of scalability and experimentally prove the superiority of explicit clustering over explicit clustering. In particular we consider heterogeneous theaters with several types of transmitters including personal, cars, helicopters and a GEO satellite. Naturally clustering is more effective in heterogeneous theaters as the more powerful transmitters (helicopters) serve as cluster heads.