December 2020

eBird III: The Count

Quick! How many Bushtits are in that frenzied, fast-moving flock feeding in the shrub over there? According to the late Rich Stallcup, founder of the Point Reyes Bird Observatory, the right number is always "17". This tongue-in-check answer helps illustrate the levels of skill and experience needed to successfully count birds. And once species have been identified at a location, counting the individuals is in fact the main duty eBird-ers, Audubon Christmas Bird Count-ers, and FeederWatch-ers are tasked with. But it's not always this easy. In the current Year of Our Covid 2020, many CBC surveys will done by individuals using eBird, with compiling being an online aggregation activity. There won't be a Stallcup with every survey party, which makes each birder's counting skills all the more important.

Cornell Lab helpfully offers some tutorials, and given that the counting process is essentially the same with most citizen/community science endeavors, their guidelines are broadly applicable. Bird Counting 101[1] doesn't include Rick's rule, but it does explain the significance and value of estimating a number when no simple, precise count is possible. After all, if there's one bird in one tree there's no issue. But there are several techniques useful for large groups of a single species, and this tutorial introduces them well.

For mixed flocks and large numbers of moving birds, eBird's Bird Counting 201[2] illustrates both the difficulties and some solutions. Once again, estimation is our salvation, while avoiding "false precision". Rick's dictum might seem to be guilty of that, but it's actually a humorous examplar of useful approximation, as no two birders trying to make an "exact" count of a Bushtit flock are likely to agree. Remember that the point of the science here is the big picture, and that there's no way we're going to see every bird, much less count them all precisely. Fortunately, the estimates average out over a large enough data set.

However, even just approximating by trying to count a sample is a skill best practiced, and David Sibley offers an illuminating set of "Quizzes"[3] (this month's Favorite) that achieve that purpose well. Because the bulk of them are constructed with generic dots, the actual numbers are known, unlike his most recent set which is immensely challenging photographs. No hints offered, just good, entertaining fun. And rounded, imprecise, "correct" answers.

Quickly counting large flocks at a distance can only be done by smart sampling and rapid approximation, competencies which aerial bird surveyors have honed. Usefully for the Christmas Count and eBird, the US Fish and Wildlife Service provides a remarkable suite of Estimating[4] tutorial videos and an explanatory web page. Those visitors to the site who have access to a modern laptop or desktop browser can use the photograph-based Interactive Tests, which permit choice of skill level, flock size and duration of observation. With a 10-second maximum and instant feedback on each estimate and on an entire quiz set as well, it's truly instructional and engrossing. Each test session is generated on the fly (ho-ho) by the parameters the test-taker chooses.

There are several other counting and estimating resources, among which LycoBirds' Practice Tool[5] stands out for its uncluttered design and ease of use. It's a project of the Lycoming County (Pennsylvania) Audubon Society. Audubon members getting ready for the 2020 CBC should remember that while eBird still tolerates an "x" in the number column, for Audubon that species becomes literally uncounted for that survey, even if present, identified and documented!

But will any of these tips, tutorials, and practice quizzes even begin to help us with the Bushtit challenge offered at the outset of this discussion? No, here it's a question of deep experience and insightful expertise. The answer remains "17".

As a final note, how can someone legitimately report 3009 ducks? As Sibley points out in a blog post, adding 9 individuals that are seen and counted while joining a flock already estimated to be 3000 doesn't invalidate either number for any reasonable purpose. Unfortunately, neither eBird nor the CBC currently allow entering "3009 +/- 500", which would likely be more accurate.