There are 3 pieces to this:
1. Recording/analog-to-digital conversion: This may be what you're thinking of, but there is a major caveat - once the noise floor of your analog signal hitting the converters is more than a little bit above the resolution of the converter, adding more gain does not give you any more resolution whatsoever. Note that with 24bit conversion you can have a huge range where this is the case and effectively makes this a non issue. It's much more important to avoid clipping here.
2. Inside the DAW: Once your signal enters Sonar (or most other DAWs), it is converted to 32bit floating point (or higher). Under floating point math the resolution is independent of signal level.
3. Exiting Sonar: When leaving Sonar either via an audio interface or exporting to a 16 or 24bit file, being reasonably close to the top of the available range can be advantageous, as much from an analog noise floor perspective as digital (because if the listener cranks up the volume in the analog domain they amplify the analog noise as well).
But if someone is thinking of using a limiter/maximizer to try to get more resolution they will end up distorting the loudest, easiest to hear part of the signal in order to supposedly get more resolution in the part of the signal that's almost always inaudible under typical listening conditions anyway. This is obviously not a sensible trade off to make.