According to "The Making of the Atomic Bomb" by Richard Rhodes there were several scenarios apart from a completely successful mission:

1. B-29 returns to base with bomb (this was explicitly specified and non-obvious).
2. Jettison Fat Man in shallow water (not possible with Little Boy - water would moderate neutrons and risk detonation).
3. Conventional explosives (the "explosive lens") go off and the bomb is utterly destroyed.
4. Nuclear fizzle - a smaller critical mass than planned, but bomb is still utterly destroyed.

I think the consequences of all charges failing to go off are also described somewhere as being a small problem, but I can't find the exact quote at this time. Feel free to look for it - it's a great book.

The big fears that project leaders expressed was that if the bombs landed in such a way that the fissile material was still easily usable in another bomb. That is, they weren't planning on retrieving the material (it would be quite an expensive loss, but such is life), but they wanted to scuttle it, to make sure it could not be used back against them. They expressed this as early as May 1943, and it was part of one of the first discussions on the record of what they might do with the bomb once they had it:

The point of use of the first bomb was discussed and the general view appeared to be that its best point of use would be on a Japanese fleet concentration in the Harbor of Truk. General Styer suggested Tokio but it was pointed out that the bomb should be used where, if it failed to go off, it would land in water of sufficient depth to prevent easy salvage. The Japanese were selected as they would not be so apt to secure knowledge from it as would the Germans.

It was a long time before they would have specific bomb designs to worry about, though.

For the Little Boy bomb the salvage question was not an issue, since it was going to be destroyed no matter what happened to it, and would probably have a nuclear yield as long as it wasn't dropped into water. If it was dropped into water, it would still probably have a chain reaction of some sort (water would act as a moderator) and probably damage itself or become radioactive. Its design was very simple and it contained a ridiculous amount of fissile material, so it was pretty much ready to go.

For the Fat Man bomb, the implosion design, they equipped the firing case with four pressure-sensitive ("plunger") switches that would detonate the high explosives if the bomb hit the ground. This would not likely lead to a nuclear yield (though it wasn't impossible for a small nuclear yield — the Fat Man bomb was not "one point safe" to use a later terminology about bomb safety), but it would definitely scatter the plutonium in an unrecoverable form. You can easily see two of the switches in this photograph — they are the equally-spaced cylinders on the front of the bomb casing, around the nose.

It isn't clear what they would have done if the Fat Man bomb had been dropped in water and the switches didn't detonate it. They may have tried to recover it. But to my knowledge they made no explicit plans along these lines. A lot of what they were doing in terms of the operational aspects of the bomb were still very ad hoc, being figured out on the spot. It should perhaps be noted that when they drop-tested Fat Man casings, they got pretty destroyed.

The best source I have seen on the specifics of these bombs and how they were fielded is John Coster-Mullen's Atomic Bombs. It is self-published but I don't know of any serious bomb nerds or historians these days who don't have a copy. I have two.