Specific volume is not really an issue: a relatively strange looking intensity unit, Volume Flow per (unit of heat transferred) leads to the circuiting criteria for any of the refrigerant-in-tube heat exchangers, and that unit will be considerably different for R-717 as opposed to R-134 due to the large latent of R-717.
So an evap condenser designed for ammonia will have considerably different circuit total lengths than one designed for R-134, as would an air cooled.
The control topic with an evaporative can get quite involved, especially if liquid line pressure is a substantial criteria. Control on water cooled is pretty simple, the regulators are on the water side and therefore relatively serviceable, and to a large extent on a circuit that does not sprawl or have alternate diversions for discharge gas, there will be no need for a high pressure receiver or other tolerance for swing in active refrigerant charge.
Excluding freezing from the design criteria: An evaporative installation will be more compact and involve less piping and less pump and fan HP in total than a water cooled. At large loads, and barring a use for "tempered" air off an air cooled condenser, the evaporative will be more compact and involve less total piping than an air cooled, once you reach the point where more than (2) air cooled's are required to balance the load.
The water treatment issue is a good one, though on an air conditioning load, a desuperheater will reduce both treatment and consumption costs. For a one on one arrangement, a cooling tower is easier to maintain and of itself not as sensitive as an EC to the accumulation of small amounts of scale; but if descaling involves tube cleanout on 3 or 4 shell and tube condensers, the effectiveness nod goes back to the EC.