In a recent article in the Antennas and Propagation Society, they point out that loss is two parts when in water. A natural loss thru the water and antenna loading of the water.
Most antennas are resonant, current bounces back and forth on the wire. If you have water contacting that, it removes the resonance. Moving the water away from the antenna by insulating it with low loss material (plastic, castor oil, foam, air, etc.) can enable the antenna to be resonant again, a resonant antenna is efficient. Part of this explanation is my interpretation of the article, but I think it's correct.
Hence if you place a monopole in the center of a square ground plane and added a plastic rectangular or cube shaped housing around the circuit card and antenna, that would be nearly optimum. A monopole won't resonate when water touches it.
The reflection from air to the water is a problem, but at least your antenna isn't contacting the water. A wider bandwidth antenna would be the triangular shape circuit replacing the monopole.
One step better would be to use high dielectric from TransTech, (somewhat expensive) dielectric of 80 and create a dielectrically loaded antenna, probably with the triangle shaped or maybe a cone shaped antenna. This would eliminate the VSWR mismatch with the water.
It depends on just how much antenna performance you need. i.e. is two feet a difficult distance to send the signal to your receiver?
If this is in salt water, you could make a liquid/fresh water tunnel to talk between your transmitter and receiver. i.e just a bag filled with lower loss material approximately connecting the transmit and receive antennas. That would cut the loss down and easily transmit from one antenna to another.
What two items are connected by the rf energy.
kch