public and/or private IPs used in "Limiting network bandwidth"

The "Limiting network bandwidth" feature allows a user to define a bandwidth limit policy that includes a set of clients based on an IP range or CIDR block.  It also allows for the specification of an exclusion range/CIDR block (a set of clients that should be excluded from the limit). 

The examples shown use 192.168.0.X IP addresses (private RFC1918 addresses). 

My question: Are bandwidth limits applied/enforced by individual clients based on their LAN (private) ip addresses? 

Can bandwidth limits also be set on a set of clients that are behind shared nat gateway by specifying the common single wan/public IP (used by the nat gateway) on the limit definition?