rich:tree needs data model not only to render data, but to do postback request processing, so data should be available. You can use Seam conversations or a4j:keepAlive to store your bean in order to avoid collisions.
Also you can set ajaxSubmitSelection="false" to switch off firing AJAX requests on selection changes.
I've tried to put my information into a ListDataModel, but after the recursive option doesn't work anymore, it doesn't found the propertie.
If I use the a4j:keepAlive, I must keep the ajaxSubmitSelection?
No, there's no need to keep it.
I personaly got a java.io.NotSerializableExceptio on my Bean.
I've added an implements Serializable;
When the object will be destroyed?
Is it exactly the same to have an @KeepAlive annotation on the class that have the JSP Tag inside my JSP?
Now I get some Injections problem, are this compatible with fields having an @EJB annotation?