Nuclear has the advantage of being lightweight, reliable and fairly compact. Also, a nuclear reactor can be dug in, and completely shielded. In fact, a small pocket reactor will outlive solar pannels about 20-40 year, depending on the type of reactor. Also, there's lot of uranium around.
Also, for those of you proposing open circuited systems(exposed circuitry), remember that any solar flare will kill the system stone death, unless you protect them. That'll cost lots of energy, reducing efficiency enormously.
As for cheap reentry systems. Remember paper planes. Scale that up. Apply origami. Have a compact, safe, lightweight and cheap reentry device. Won't work for Mars though. Air density is to low. You'd need retrorockets.
The main problem with Martian colonies is radiation. Solar flares especially. They won't kill anybody. Not immediatly. That's what all recent mars projects are proposing. Just don't account for the dangers. If it only kills the astronauts in 20 years, the mission has been a succes. They're sending people to their deaths, should a flare occur during the half a year journey, or during their stay on Mars. (Mars doesn't have a magnetic field).
The other main limiters are launch and design costs. Energy costs are not of the issue, at least not at the moment. Most likely it won't be for a long time. At the moment it costs 2000 dollar per kg to get in to LEO. Mars orbit is far more costly. Even for nuclear reactors, the largest part of the maintenance cost is safety. Fuel only amounts to less than 20% of the costs. With a space based reactor, you can drop most of that.
Also, uranium has a ridicously high energy density, far outranking any other fuel sources, barring fusion.
Where does the idea come from that renewable means reliable, or that fission supplies are limited. We have reserves for more than a hundred years. By that time fusion should be working. I don't think we'll ever use solar pannels as (sole) powersource on (manned) expeditions. They're just not reliable enough.