If you're shooting at magnifications no greater than 1:1 there are many possibilities. I also shoot fish in aquaria with flash. Controlling reflections in this situation requires that the flash(es) axis be at a 45 degree angle to the tank fronts.
I have a rig that works well for fish in aquaria, also for flowers and insects. The range of magnifications I use is given by the lenses I use. With a 105/2.8 MicroNikkor, no greater than 1:1. I've gone digital and auto-ttl flash, previously used a similar rig with an FM2n, small manual flashes and Kodachrome 25. For flowers and insects, flashes on a Spiratone MacroDapter. For fish in tanks, bracketry as I now use, a straight bracket that holds my D810 with an AF 105/2.8 MicroNikkor in the center and an SB-600 at each end. A similar rig with just one flash would work nearly as well.
About using manual instead of auto-ttl flashes. I spent a roll of film shooting at likely apertures (f/11, f/16 and f/22 set, effective aperture smaller depending on magnification and aperture set) and magnifications (1:8, 1:4. 1:2 and 1:1). This with a pair of little fixed-output flashes on short tilting flash holders on a Spiratone MacroDapter. Same subject for all of the test shots. I examined the test shots, picked the best one for each combination of magnification and aperture set. This gave me a table of aperture to use given magnification. Linear interpolation between the test magnifications gave results that were better than good enough.
The MacroDapter makes a poor man's ring light, but with better modeling. It has 2 cold shoes, screws into the front of the taking lens. If you can get the flashes at the right distance from the lens' axis, a MacroDapter-based rig will give nearly constant aperture, given flash output, over 1:6 to 1:1. This works because changes in flash-to-subject distance with changes in magnification give changes in illumination on the subject that nearly exactly offset changes in illumination on the film plane due to changes in magnification. I hope this explanation is clear enough.
I did the same tests with a pair of Vivitar 283s and VP1s, found the right power settings given magnification and aperture set. One more look-up table. Worked a treat, but much heavier.