You will quickly run out of the default allocation in the Lower memory pool with a scheme such as you describe, causing voices to be killed. Each reverb instance takes quite a bit of memory, even when bypassed. I encourage you to use the Wwise Profiler and check the Memory tab of the Advanced Profiler view to keep track of this.
Your scenario might be best implemented by using the Aux Sends system, with one Aux Bus per environment type. This way, Reverbs will be shared across all voices contributing to them, saving on memory and CPU.