S3 is used to store the source copy of your application version, so that if you need to scale-up, the new EC2 instance pulls a copy from S3. The cost for this is generally negligible.
You can use EBS, but you don't have to use EBS. EBS is on-instance storage, and is much faster than connecting to S3 if that sort of thing is important. EBS is also persistent storage, as opposed to ephemeral which goes away when the instance shuts down or is replaced.
Think of EBS as a hot-swappable hard drive that you can pull from one instance and pop into another that still has all of the data on it. This isn't as important for your application code (which is copied from S3 when a new instance is added to the environment), but may be important for other "hot" parts of your application. If you don't need this, then so be it.
RDS is its own thing. There's support for doing snapshots and/or regular backups of your data to S3/EBS (I don't recall which at the moment), but neither are strictly required for the day-to-day nature of RDS. Keeping regular backups of your database is highly recommended (for your own sake), but not required.
Make sense?