From the Forest Products Journal...not an easy read:

http://www.fpl.fs.fed.us/documnts/pdf1998/vick98b.pdf

And abstracts of some interesting discussion over the results:

Think we all can agree that resorcinol is our strongest wood glue, dry or wet, maybe not the easiest or most desirable to use in all situations, but still the strongest. That is why they used resorcinol as a comparison for polyurethane glues.

They used 2 woods, yellow birch and douglas fir.
On the d.f. in the dry test, the polyurethane had a sheer strength of 1200, while resorcinol had 900 lbs of sheer strength, poly had almost a 100% wood failure. In both cases more then resorcinol.

In the next 3 tests which all are wet tests including a boil test, it appears that poly falls way behind resorcinol. Resorcinol has about 2 times as much wood failure in d.f. and almost 4 times as much in yellow birch as does the poly.
Sounds convincing for the resorcinol, But hold it. Look at the sheer strength in the same tests, and this is where it gets interesting.

The poly failed at only about 10% less sheer strength then did the resorcinol.
So it would appear that poly is stronger then resorcinol in dry wood joints and about 90% as strong as resorcinol under wet wood conditions.
After all isn't the real test of a glue joint the actual lbs of sheer strength that it actually takes to rip the joint apart.

As for the greater difference in wood failure in wet wood for the resorcinol.When you think about it, resorcinol requires tight fits and tight clamping, and results in a 5 or 6 thousands glue line. Poly would have a 25 thousands glue line or about 5 times thicker glue line.It would be much easier for the glue line to fail with a much thicker glue line, but again bottom line is - what is the actual sheer strength to pull apart the glue line, and poly only has about 10% less strength then resorcinol in wet woods.

It would make sense that you use metal fasteners as well in wet conditions.

I looked in this site but, as of yet have not found a test comparing epoxy, it should be there somewhere and might be interesting as well to compare all 3 glues.

Durability, u.v., hot and cold resistance as well as life span and flexibility might be interesting as well. Resorcinol has a very long life. They claim the poly's do too.
1. The section on delamination suggests that poly is much more likely to delaminate under wet-dry cycling. Such wet-dry cycling is, of course, very common in boats.

2. In the tests I presume they used simple edge joints and that the glueline thickness was about the same for the various samples, rather than being thicker for poly.

3. I don't think you are correct in saying that metal fastenings would usually be used in wet situations. After all, this mast will almost certainly not have metal fastenings all over the place. More to the point (since a mast might not really be a wet situation in many cases), one of the most common uses for glue in marine situations is for laminating frames, stems, keels and other parts. In non-monocoque hulls especially, these parts may well have some fasteners in them (such as planking screws), but potentially not enough fastenings to fully hold them together. Even if there are enough, if they delaminate but don't disassemble that would still be highly undesirable. In a non-monocoque hull these parts are also very likely to see the regular wet-dry cycling that caused problems with the polyurethanes.