a single conventional (deep groove, conrad-style) bearing can have a reasonable axial load rating, but IME the rating is often not made clear in the bearing specs and/or varies quite a lot from one manufacturer to another. I guess if you only have one bearing that takes the axial loads you don't have any choice but to use one such bearing.
If you want a higher load rating for the assembly then a simple approach (which may not be applicable in this case, see comments below) is to modify the parts so that you double up on the bearing that sees the load. This can work better than allowing both extant bearings to share the load, because that only works if the lengths are perfect and the axial stiffnesses of the shaft and housing match one another.
One special problem that trike bearings have is that the shaft flexes in bending and this (perhaps moreso than the axial loads per se) can very greatly shorten the life of the bearings. It might be prudent to specify a fit/tolerance for these bearings that is a bit looser than normal, so that the shaft can flex without the bearing seeing the most severe cantilever loads. The other thing that is worth doing is seeing if the shaft can be made stiffer. In extremis there are bearings which allow some articulation, but these have lower radial load ratings and are fiercely expensive.
It is a moot point now but regarding speed ratings and ball scuffing etc; bearing preload is important here, but in full complement bearings the ball-to-ball scuffing is deemed to be worse than the ball-to-clip scuffing in a standard bearing. So the speed rating is usually significantly reduced in a full complement bearing. This makes them unsuitable for some applications but bicycles ain't one of them; shaft speeds are normally still at least one order of magnitude lower than the speed rating of such bearings.
cheers
