If you mean that you have one WAR with web services defined in it, and another separate WAR with the JSF stuff, I think it's really two separate projects each with their own Spring configuration.
The web service WAR will use either Spring web services or perhaps HTTP remoting to expose your service interfaces to clients via HTTP. This will have one set of application context configuration, either XML or annotations.
The JSF WAR will have the JSPs and controllers. The controllers will be injected with clients that will interact with the remote services to accomplish what you wish. That's all they need to know about the service WAR. There doesn't need to be any duplication of configuration at all.
It's actually a nice design, because it completely decouples the view from the rest of the problem.