Several reasons that interlocked into a single viable option.

The design concept needed to be locked early in the design phase so that they could meet the demands of the mission profile. That meant, they had to go with the information and technology available at the time of the decision.

The information available was a fairly good understanding of the lower atmosphere, but not a lot of information about the composition of the extreme higher parts of the atmosphere and how it would interact with a manmade object.

They also didn't know how accurate the landing projection would be. They could easily be off by several miles. It was assumed to be a ballistic entry with no control on reentry. On land, a few miles could radically change the landing conditions. But, a water landing could easily meet certain criteria over large areas around the central aiming point. Water landings also allowed for a cushioned landing so they did not have to carry mass for struts and a suspension. The US had more than enough ships to easily deploy to any landing area.

For the US, the early rockets were very limited in terms of mass deliverable into orbit. Functionally, the heatshield and parachute demands were the same irrespective of a land or water based landing, so that was a constant. But, adding control mechanisms and all the needed materials to handle that would have greatly added to the mass of the vehicles. The Soviets had much larger ICBM's, so the mass they were able to launch into orbit was much greater than what the US could manage.

I don't know if you've ever seen a Mercury capsule in person, but it was tiny. It was 6 foot 10 inches (roughly 2 meters) in length and roughly the same at its widest point (it was not as quite as wide as tall and sharply tapered to very narrow at the point). There simply wasn't room in that design for any sort of controls on reentry. Functionally, they took the basic form of a nuclear warhead in an ICBM, then wrapped it around a person. The primary design constraint in the late 50's was speed as the US was visibly losing the Space War. They went with the fastest approach to accomplish it knowing full well that the technology was a dead end.

Gemini experimented with a rogallo wing design, but never took it into production. Spent something like $165 million on the design, but by this time, Gemini was being phased out at NASA and would be a USAF only operation for its manned lab. The USAF funding for that was later cut. They (NASA and the USAF) did a lot of interesting work on reentry in the late 60's. They could have built off that, but there's a whole other can-of-worms called the Shuttle Transportation System that would need to be covered for that.

The Apollo mission design was locked fairly early also, so they could not really roll lessons learned into its core design, but they were able to improve the ergonomics, equipment. and training based on earlier designs.