The use of nose breathing as a marker for the aerobic threshold came about after years of testing well trained XC skiers. I began to notice that the upper limits of comfortable nose breathing (upper limits of conversational-speaking in complete, long sentences) corresponded almost exactly with a blood lactate of 2mMol/L, a concentration often considered as indicative of ones AeT. This in turn corresponded well with what exercise science calls the First Ventilatory Threshold or VT1. VT1 shows up in the data collected during a gas exchange treadmill test. At VT1 the both the rate and depth of ventilation makes a noticeable jump (your lungs are pumping significantly more air).
When Steve and I started using the nose breathing/cpomversation test on less well endurance trained folks we discovered that the nice correlation I have noticed in my skiers broke down. We tested folks and heard from many more who could nose breath to crazy high heart rates. Since that time I have asked several sports scientists about this but still have not heard a good answer.
This is why we suggest the HR drift test as detailed in article by the same name. It takes the subjectivity out of determining AeT.