Reported collisions between American aircraft and American wildlife have risen eightfold since 1990. Five stricter ways of counting the same thing are carried alongside it, and every one of them rises less — the growth is almost entirely in the reports where nothing happened.
Every point on the top line is a wildlife strike somebody filled in a form about. The trouble with counting forms is that you are counting the forms. Filing one has been voluntary for civil operators for the whole period, and the effort put into collecting them has changed enormously — so a rising count is consistent with more strikes, with more reporting, or with both.
Two different checks are drawn over it. The obvious one is exposure: divide by how much anyone flew, and see whether the risk per departure moved. The less obvious one, and the one that does the real work here, is a severity threshold — keep re-counting the same database under definitions that are progressively harder to slip past. A strike that changed what the flight did leaves an operational trace whether or not anyone files. One that damaged the aircraft is harder to leave unreported than a smear on a radome. One that was substantial, or destroyed the aircraft, is nearly impossible to lose.
Order the six series by how hard each is to under-report and they line up almost perfectly by how fast they grew. Exposure alone barely dents it. Severity dissolves it.
The same database counted six ways, comparing the mean of the first five years on record with the mean of the most recent five complete ones. Read top to bottom. This is the whole argument on one screen.
One of those rows cannot be taken at face value, and the honest thing is to show you why rather than hope you do not check. The bottom row — substantial or destroyed — is the only series that falls, and a large part of that fall is the FAA changing how it grades damage, not aircraft being damaged less. Below is every damaging strike, split by the grade it was given.
If the rise is reporting rather than risk, the reporting should show its own fingerprints — and it does. The channel the reports arrive through has been rebuilt underneath the count: a paper form mailed in, replaced by one that can be filed from a phone at the gate before the aircraft has finished taxiing. Whatever else that changed, it did not change the number of birds.
Which animals matter is not the same question as which animals get hit. Every identified species is placed by how often it is struck and how often that strike damages the aircraft. The two are close to unrelated, and the species in the top-right corner are not the ones anyone would name first.
Everything above is a count, which makes it easy to forget what is being counted. These are the costliest strikes in the file and every strike in it that killed someone, in the words of whoever filled in the form.
Height above ground, where the report gives one. Almost everything happens in the first few hundred feet — which is the airport's problem, and therefore something an airport can do something about. The share that does damage moves the other way.
Reports are attributed to an airport where one is known. A big number here is not a dangerous airport — it is very often a well-run one, because a wildlife programme is exactly the thing that turns a strike nobody would have mentioned into a filed report.