Don't know whether I got you correctly. I have never seen an admin who wanted his clusters to start up at os boot. In case you are new to HACMP you might not be in an ideal situation for making such a cluster design decision. You can configure HACMP to start up with the os but if you can take some piece of advice: don't do it. IMHO the potential problems by far outweigh the potential adavantages.
HACMP is very flexible here, it lets you script everything and you can control when a command is run by pre or post events or by the start/stop scripts. You could create an ssh key with empty password distribute it accordingly to the nodes you want to access and then script the start of the other clusters. Following your description a good place for the startup command might be at the bottom of the RG start script. If you mean to power on the servers themselves you need to script HMC or CSM commands to name two possibilities.
So you need information about HACMP monitors and how to combine them with a run-time policy. A monitor watches out for the existance of a certain process and uses your scripts to react if the process is not there/disappears. Then there are the following run-time Resource Group dependencies available from HACMP 5.3 on:
Bring parent online before or after child.
Bring parent online on same node / different node / same site.
However from the information you supply it would be hard to give detailed information how to setup the RGs. I suggest you check with the cluster docs about how to configure process monitors and custom monitors. Secondly read about the difference between startup monitoring mode and long running mode or the combined monitor that will check both start and runtime. Only monitors and scripts you define/write will ensure that the DB/app will start up in the order you wish. High Availability Cluster Multi-Processing (HACMP) publications - IBM e(logo)Server Cluster Information Center