The Internet of Things is envisioned to provide connectivity and communication between various devices all over the world. Based on the devices and applications involved, the nature of the network formed differs. Thus, an intelligent and holistic ecosystem needs to be created wherein clients, data sources, smart objects, and services can all co-exist and interact with each other. We present Adaptive Ubiquitous Middleware for context-aware IoT ecosystems, which considers the situational context of the applications, devices, or people and the contexts of the network formed and accordingly adapts the behavior of the ecosystem. Adaptive Ubiquitous Middleware is a multi-agent, multi-communication protocol-facilitated middleware that acts as an integration point for applications to access relevant context, share it with other applications, and have relevant services made available via a multi-communication protocol bridge. We also present an optimal service allocation model for a single service class that utilizes available computing resources and achieves a minimum average response time. The system implementation has been evaluated with two use cases to demonstrate its applicability, effectiveness, and generality. The evaluation of optimal service allocation demonstrates the service response is much faster in the proposed model.