Missing birds

I have long thought that birders tend to drastically overestimate the percentage of the continent's individual birds that we actually manage to see or hear. I know in the course of my daily routine I almost never come across other birders, and I am one. If I hardly ever come across them, it seems that a bird, going about its own, bird-thing-focused routine, would almost never happen to stumble within binocular range of a birder. So, indulge me in another of my little back-of-the-envelope calculations, and see if we can put some fast and loose numbers to this.

I have developed a general rule of thumb that in one hour spent in the field an experienced birder will find roughly the equivalent of all the birds in 0.1 square kilometer of land. This is very approximate, and of course not universal across species. Turkey Vultures are much more obvious than this, and LeConte's Sparrows are much better at staying hidden. But for the run-of-the-mill land bird in your average forest-field-and-park habitats, I find this rule of thumb to be a good rough estimate. For instance, I find it takes a couple of hours of criss-crossing my 16 ha farm to feel like I have seen the bulk of the birds on it (for the SI challenged, ha = hectare = 10,000 square meters; 100ha = 1 square km. Roughly 1 ha = a bit more than 2 acres). Or you can think of it this way: if you cover about 1 km of distance in an hour's birding, you have seen most of the birds within 50m either side of your path, and missed most of the birds farther away than this. Again, this is all very approximate, and there are infinite variations depending on whether you are sitting still, walking, bicycling, driving, in prairies, in deep forest, etc. etc. But we're doing that order-of-magnitude, back-of-the-envelope thing, so I'll take 10ha/hr as the generalized conversion factor for hours afield to area covered.**

So, how much land does an average, active birder cover in a week? I'm going to restrict this to the sort of birder who is experienced enough to identify rare birds consistently and accurately, AND who reports what s/he sees to the birding community as a whole, at least the rarities. In the birding world, if a rare bird sings in the forest, and the person who hears it doesn't tell the RBA, listserv, local field notes compiler, or rare bird committee about it, then that bird really didn't make a sound. Let's say this typical birder has a typical job and other responsibilities, and manages to get out for a few hours most weekends, say something like 3-5hrs. So, we'll estimate that this typical birder, in a typical week, sees something very roughly equivalent to all the birds in 30-50 ha. This may be generous; a lot of these sorts of birders can't swing this much time every week, all year. But we'll go forward with this.

OK, so how many of these birders are there? Let's just look at the contiguous 48 States. A number like 10,000 might be a good ballpark guess. There are about 50,000 Christmas Bird Count (CBC) participants in this area each year; there are something like 70,000 checklists submitted to the Great Backyard Bird Count annually. But both these numbers include double counting; they also include a lot of birders who are more casual and/or less likely to report their findings than the type of birder I described above. A total of 10,000 averages out to roughly 200 per State, or about one per 30,000 population (i.e. in a city of a million people, there would be thirty of these birders). These numbers seem to be of the right magnitude to me. If you look at the State listservs, or the regional reports in North American Birds, this is in line with the number of different names that you see popping up. So NOW.. we multiply our ballpark estimates:

Ten thousand birders covering 30-50ha per week gives us about 3,000 to 5,000 square-kilometers-worth of birds found during an average week, over the entire lower 48. But, the total land area of the Lower 48 is about 8,000,000 square km. So... all these birders, in an average week, find about 1/2000th of the birds here. For every rare bird that someone finds, there are THOUSANDS of others just like it than no birder has seen or ever will!

Now of course there are cases where this just plain isn't true. Big conspicuous birds of open country are probably found much more reliably than this, especially if they are prone to congregating in well-defined, identifiable areas. A much higher fraction of Trumpeter Swans almost certainly gets found, for instance. But Gyrfalcons? Not so sure. And stray passerines that are not of the feeder sitting sort, like MacGillivray's Warblers in the east? Or rare breeders in rural habitats? We find only a TINY fraction of these. There will also of course be a bias in favor of species that frequent more heavily-birded habitats, such as feeders, city parks, rural/suburban matrices close to major metro areas, coastal and interior wetland "hot spots," and the like; and a parallel bias against birds of underbirded habitats, like large continuous forest tracts, western rangelands, and high mountain habitats.

It is tempting to just multiply this by 52 and conclude that in a year, this nation of birders will find about 1/40th of the birds in the continent. And this would be true if either the birds or the birders moved about randomly across the landscape. Of course, neither of us do. Birders tend to go back to the same places from week to week, and many birds tend to stay within small home ranges for weeks on end. To some degree, each week we keep seeing the same 1/2000th of our birds over and over again (and missing the same 1999/2000ths of them). There is also the matter that many birds are only in the area for a portion of the year. Some individual long-distance migrants might only spend a few weeks a year in the Lower 48, giving them few chances to be found. So the odds of detecting a single bird over the course of an entire year are not easy to estimate, even in the back-of-the-envelope fashion. Lets just say it is less than 1/40, in most cases probably MUCH less than that. Which means, of course, the great majority of individual birds are never seen or heard by a birder at all in the entire course of their lives.

It is interesting to apply this 10ha/hr rule of thumb to some of the large-scale bird survey projects. For instance, in most State Breeding Bird Atlas (BBA) projects, they had to accept about 10 hrs of coverage in an individual block as "complete." Even with the tireless efforts of the (mosty unpaid!) organizers and all us intrepid field workers, there was just too much land to cover. So in each 20 square km BBA block we saw about 1 square km worth of birds, or only 5% of the total. Given that most States also had to pare down coverage to only 1 out of 6 or even 1 out of 12 blocks, it looks like these State BBA projects probably found less than 1% of the individual breeding pairs of most species. Every single pair found represents 100 or more pairs that were not found; another way to look at it is that a species that had fewer than 100 pairs nesting in a State had a good chance of being missed entirely. A species with only a very few individual breeders (e.g. single digits) would have been quite unlikely to be detected at all.

At the other pole of the year, a typical Christmas Bird Count has on the order of 50-100 party hours of effort. Using the same rule of thumb, this works out to the equivalent 5-10 square km of area covered. A CBC circle has a total area of nearly 500 square km, so again we are looking at only 1-2% of the birds in the circle actually being counted. Even on the megacounts that have something like 400 party hours, this is still only about 10% coverage. It may feel like you are running out of area to cover at the end of the day, but you are in fact only running out of accessible area. All those backyards and roadless or inaccesible lands behind fences are hiding most of the birds. On a transcontiental scale, there are about 1500 CBCs, giving us about 10,000 square km worth of birds. This is about 0.1% of the total land area of the Lower 48. So once again, we find that even in the intensive CBC period, when birders go afield in massive quanities, for every bird counted there are about 1000 birds missed.

Another interesting implication of this is that a single individual bird who does not stay put within a small area is quite unlikely to be found again after it moves on. Exceptions, of course, for birds that are especially conspicuous and/or tied to especially well-covered habitats. But on the whole, even if two birds of the same species are found in two different places within a few days or weeks of each other, they still are probably not the same individual unless you are in extremely heavily birded territory. More likely is either that whatever circumstances led the first bird to stray brought some other birds of the same species with it, or the publicity about the first sighting got observers especially aware and alert so another bird was found that would have been overlooked otherwise (sort of a corollary to the "Patagonia Rest Area Effect").

I realize that these numbers are very crude estimates, and the generalizations are terribly broad. That's what the backs of envelopes are for. Even so, given the large orders of magnitude involved, if you somehow accounted for all these variabilities and powers of 10 one way or the other, the conclusion would stand up:

We see only a tiny fraction of the birds in our Nation. The vast majority of strays and low-density residents are never found by birders. Once lost track of, a single bird is unlikely to be relocated in its lifetime. And species present in extremely small numbers are likely to be entirely overlooked for long periods of time, even if they are ordinary birds with ordinary levels of detectability.


**Footnote: This number is actually just the ratio of a bird's true abundance (birds/ha) to its rate of detection (birds/hr) under a particular set of circumstances. Hence its units are ha/hr. It is just the conversion factor between detection rate and aerial abundance. So a value of 10ha/hr means that a bird with a density of 1 bird/ha will be detected at a rate of 10 birds/hr; similarly a bird that is detected at a rate of 1 bird/hr has a density of 0.1 birds/ha.