These systems typically have a few electrodes on the tank that connect up to something (stepped resistors, I think) encased in an epoxy blob, which also has a couple wires going back to connect with the bargraph display panel.
Likely culprits in my estimation would be the connection to the lowermost electrode(s) in the tank, one of which is the common for all of them; connections between the blob and the control panel; and possibly a problem with the control panel itself.
It might be easily possible to temporarily swap the connections for one of the working tanks and the fresh water tank to see if it's the control panel circuit or the wiring or probe connections.