You are confusing BMI with body fat percentage.
I understand that they are different things. Unless these are bizarrely jacked ultra-marathoners, BMI should work just fine.
>> Mean (+/-SD) percent body fat values for men and women were 17+/-5 (range 5-35) and 21+/-6 (range 10-29) , and ranged from 6 to 14 and 14 to 27 among the fastest three men and women.
I would imagine the fastest finishers are not "bizarrely jacked" (they usually look pretty scrawny), but their BMI is not very well correlated with their body fat percentage.
> We conclude that despite wide variations in BMI and percent body fat among ultramarathon participants, the faster men have lower percent body fat values than the slower men, and finishers have lower percent body fat values than non-finishers.
So I do stand corrected that not all of them are significantly below 10% body fat, but some definitely were (I don't see a breakdown anywhere giving an indication of how many), including some of the top finishers. I'd also imagine the ones at higher percentages who performed well probably have higher floors to begin with for how little body fat they can reasonably reach.
In any case, it is not at all abnormal for BMI to have nothing to do with body fat percentage or how jacked someone looks, especially with athletes. BMI is probably one of the most meaningless health metrics in common use.
I am surprised no one has mentioned Ann Trason that won 13 (iirc) Western States and used to win many ultras outright.
It's a small sport with not much competition.