nouseforaname,
Looking at the photo you posted of that bevel gear, the first thing that caught my attention was not the gear tooth that fractured across the roots. That should have been expected. Instead, what caught my attention was the gear's rim thickness. That's the thickness from the gear tooth root fillet to the adjacent rim surface. As a rule of thumb, good gear design dictates that this thickness should be equal to or greater than the gear tooth whole depth. Your example looks to be only about one-third of this.
The reason for maintaining this minimum gear rim thickness is so that any gear tooth root fracture will propagate through the tooth, instead of through the rim. A rim fracture would almost always be catastrophic, while a tooth fracture would be less so. In fact, if the gear mesh has a high enough contact ratio, it may even continue to operate with a missing tooth.
Your gear suffered an unusual tooth fracture. Looking closely at the photos, it appears the fracture started at the tooth root and then (logically) propagated towards the rim. But then it strangely turned around and finally broke out at the adjacent tooth root.
It's just my opinion, but I'd have a serious chat with the guys that designed that gear. Because that gear doesn't seem to have a safe rim thickness based on what I'm used to seeing on rotorcraft drivetrains. Here's a good reference from NASA:
NASA Technical Reports Server
Regards,
riff_raff