A coordination protocol for systems of unmanned marine vehicles is proposed for protection against asymmetric threats. The problem is first modelled in a game theoretic framework, as a potential game. Then an extension of existing learning algorithms is proposed to address the problem of tracking the possibly moving threat. The approach is evaluated in scenarios of different geometric complexity such as open sea, bay, and harbours. Performance of the approach is evaluated in terms of a security index that will allow us to obtain a tool for team sizing. The tool provides the minimum number of marine vehicles to be used in the system, given a desired security level to be guaranteed and the maximum threat velocity.