Different papers use different methods to estimate predation risk in various locations, for instance. Some base it on actual predation incidents, some on landscape factors (which may model risk perception as opposed to actual risk).
There's also discussion about the way active predators such as wolves impact an area, which is different from lie-in-wait predators. And there's an idea that a constant background threat comes to be ignored, as opposed to a sort of pulsing threat, where there are surges of predation.
Also worth noting that various studies take place in different areas of the park, and in different seasons. Not to mention there's three main varieties of trees under discussion (aspens, cottonwoods, willows), which have all seen different recovery rates in different areas.
It's not a simple question, and from what I've seen in my sort of casual reading on the topic, it's not at all settled.