Classical approaches like process algebras or labelled transition systems deal with static composition to model non-trivial concurrent or distributed systems; this is not sufficient for systems with dynamic architecture and with variable number of components. We introduce a method to guide the modelling and the dynamic composition of processes to build large distributed systems with dynamic adhoc architecture. The modelling and the composition are based on an event-based approach that favour the decoupling of the system components. The composition uses the sharing of abstract communication channels. The method is appropriate to deal with evolving processes (with mobility, mutation). The event-B method is used for practical support. A fauna and its evolution are considered as a working system; this system presents some specificities, its behaviour is not foreseeable, it has an adhoc (not statically fixed) architecture.