No Bird Apocalypse: Populations rise and fall with habitat change and guessing rules.

I was sad that songbirds were disappearing until I read this by a scientist who did bird counts.  It's complex.  There's good news.

First, here's an example of a bad news headline:  Three billion North American birds have vanished since 1970, surveys show.  

Counter-information you need to know:

-1 to 3 billion birds are killed each year by cats.. (Note the uncertainty.)

-About 1 billion birds die each year crashing into window reflections.  (Again uncertainty)

-Another third of a billion get whacked by cars and trucks each year. (Uncertainty)

Baby barn swallows, Knight's Inlet

-Almost half a billion of the loss is in three non-native species: House sparrows, pigeons and starlings.  

-Forests loving species are approaching pre-colonial levels as forest cover has rebounded.  

-Before spraying for spruce budworm in the US,  Tennessee warblers that specialized in eating those morsels exceeded 110 million. Their number has dropped to 30 million as the budworm count
dropped.

-COUNTING IS GUESSWORK.  The models used are no more accurate than the COVID-19 ones.

The models assume that each stop will account for all the birds within a 400‑meter radius. Because a crow is readily detected over that distance, no adjustments are made to the number of observed crows. But hummingbirds are not so easily detected. The earlier surveys assumed a hummingbird could only be detected if it was within an 80‑meter radius. So, to standardize the observations to an area with a 400‑meter radius, observations were multiplied by 25. Recent survey models now assume hummingbirds can only be detected within 50 meters, so their observations are now adjusted by multiplying by 64.  Thus, depending on their detection adjustments, one real observation could generate 50 or 128 virtual hummingbirds. That number is further scaled up to account for the time‑of‑day effects and the likely number of birds in the region’s un-surveyed landscapes.
Setting aside assumptions about the regional homogeneity of birds’ habitat, one very real problem with these adjustments that has yet to be addressed. If one bird is no longer observed at a roadside stop, the model assumes that the other 127 virtual birds also died.