Abstract: Grid infrastructures for e-Science projects are growing in magnitude terms. Improvements in data Grid replication algorithms may be critical in many of these infrastructures. This paper shows a decentralized replica optimization service, providing a general Emergent Artificial Intelligence (EAI) algorithm for the problem definition. Our aim is to set up a theoretical framework for emergent heuristics in Grid environments. Further, we describe two EAI approaches, the Particle Swarm Optimization PSO-Grid Multiswarm Federation and the Ant Colony Optimization ACO-Grid Asynchronous Colonies Optimization replica optimization algorithms, with some examples. We also present extended results with best performance and scalability features for PSO-Grid Multiswarrn Federation.