The purpose of the beacon is the overriding factor in selecting the antenna to be used for the beacon. Beacons designed for very long range paths need as much gain as possible to maximize the ERP of the beacon system.
That high gain comes as a result of directivity, which may or may not be a bad thing. Most high gain antennas have narrow horizontal and vertical main lobes. While that maximizes gain in a single direction, it may reduce or prevent the signal from being heard by stations outside the main lobe. Clearly some trade-offs will need to be made to match the pattern in the desired directions with the gain desired.
The directivity is largely affected by the included angle of the receiving stations the beacon is beaming towards. If the receiving stations are tightly grouped together, or a path between two single points is being researched, a large yagi or arrays of yagis can be used. If the receiving stations are spread over a wide area, the beacon antenna needs a wide horizontal lobe. Stations on the ends of the trans-oceanic paths will benefit from these unidirectional antennas.
Beacons in mid-ocean locations face a more challenging antenna selection process. They will have potential paths in several or often opposite directions. They will need to choose between omnidirectional or bidirectional antennas. These inherently offer lower overall gain, and therefore lower ERP. Some locations may even choose to use multiple beacons with higher gain uni-directional antenna systems beaming in the different directions to maximize ERP. An alternative might be to use multiple antennas systems in the different directions, and have a means to switch between them at certain times or between subsequent transmissions.