Usage: two approaches
Apply reverb directly to the original track
The more obvious approach of applying reverb directly to a track works badly with GVerb's default settings, but there's an excellent reason for this. Specifically, the "dry" level (no effect applied) is set to zero. Why? This is based on the assumption that you are going to duplicate your track and convert the duplicate to reverb-only by applying GVerb to it, leaving the original track untouched. Then you control the amount of reverb by mixing in different volume levels of the dry and reverb-only tracks. This is a professional yet conservative studio-like approach. A professional sound engineer would pick up the original signal from the board, send it through a reverb unit and return it through a separate input channel of his mixing board.
There are some advantages to applying reverb directly. It means less disc space and CPU usage, and lessens the possibility of playback problems on slower machines. It can give you a quick, reasonably useful reverb, if you know suitable instant reverb settings that you can apply.
Duplicate and mix together
The alternative "duplicate and mix" approach has two major advantages:
Quicker to get exactly the effect required
There are a very wide variety of sounds possible on a reverb, and it can be difficult and time-consuming to determine the parameters required for a particular sound, instrument or project. Reverb processing is not real-time: even on fast machine it may take 30 seconds or more to process a four-minute stereo track in 96000 Hz, 24 bit quality. Audacity does have a preview for effects (limited to 3 seconds in Audacity 1.2.x, but with configurable length in 1.3.x). Even so, the reverb parameters cannot be modified during preview, so unless you already know exactly what you want from previous experience, you may spend a lot of time experimenting, processing, listening, perhaps undoing, and changing parameters. To get the exact sound you want more quickly, it is actually very useful to utilize the sound engineer's method of duplicate track reverb layering. Once a chosen reverb has been processed into the duplicated track, you can mix in the amount of reverb "wet" signal in real time using just the -....+ gain control on the reverb track.
Non-destructive, and expandable
You always keep the base "dry" track as it was originally. As well as being important in its own right, this gives you the option of working with multiple reverb tracks, without destroying either the dry track or the individual reverb tracks. As a simple example, you can further manipulate the reverb track with compression, equalization or noise gating (via a third party plug-in such as Dr.Expander ), without affecting the "dry" track.
Working with more than one reverb track opens up several possibilities. You could make the reverb amount change over time by duplicating the reverb track, applying a more extreme reverb to one of them, then use Audacity's envelope tool to gradually adjust the volume of each reverb track. Or you could prepare two different reverb tracks, (say one with the Early Reflection only and a second with the Tail only). This way, you could mix both types of reverb with the dry original in real time and really fine-tune the exact sound you want with immediate feedback. Going even further with this approach, you can try different sets of parameters on different sets of tracks, muting and unmuting them to compare one set with another in A/B fashion, again with immediate feedback in real time.