How many birds are in a flock? How to count a flock of birds

How many birds make a flock?

No set number of birds makes a flock. Wikipedia's article on flocks of birds defines one as “a gathering of individual birds to forage or travel collectively”: a description of behaviour, not of size. Two geese grazing together and a roost of thousands of starlings are both flocks, and neither that definition nor the counting schemes below sets a minimum.

So “how many birds are in a flock?” has no general answer, only a counted one for the flock in front of you. What changes with size is the method: individuals for a small group, a counted block for a large one, and a photograph for a flock that will not keep still.

The number you are trying to produce

Before any technique is useful, it is worth being clear what the count is for, because it decides the method. A checklist does not want the number of times you saw a bird. It wants the highest number of individuals present at one time, plus any clearly different individuals you are confident were not among them. Project FeederWatch's protocol, for one, reports only the highest number of each species seen simultaneously, because it is designed to prevent the same bird being counted more than once.

That definition does two jobs at once. It gives you a figure that can be compared with other people's counts of the same place, and it means a group that keeps moving around cannot inflate your number by being seen again.

eBird's own worked example is the clearest way to see the second half of the rule: if you see six female cardinals and three males, and later six males together, you have had at least twelve different cardinals — even though twelve were never in view at one moment. Sex, age and plumage differences are all legitimate grounds for adding birds you can tell apart.

Small groups: count them

Up to roughly twenty or thirty birds, count individuals. There is no technique to learn and no reason to estimate something you can simply count. Work left to right, or clockwise around a group on water, and keep your place with a tally counter rather than with your memory so that a distraction does not cost you the count.

The moment the flock stops being countable one by one — because it is too big, too dense, or moving too fast — switch deliberately to estimating rather than drifting into guessing.

Large flocks: the block method

This is the standard technique, and it is the one used by shorebird and waterfowl surveyors.

1. Count a block. Pick a small part of the flock at typical density and count the birds in it exactly — ten, twenty, twenty-five.
2. Fix the shape of that block in your eye. How much space those birds occupy is the unit you are about to use.
3. Superimpose it. Mentally lay that block over the rest of the flock and count how many times it fits.
4. Multiply. Blocks × birds per block is your estimate.

So a block of twenty that fits about twelve times is roughly two hundred and forty birds, and you report it as an estimate of around two hundred and fifty rather than as 243.

Counting by tens is the same idea in a simpler form, and it is the method eBird's own help article, How to Count Birds, gives for large flocks: “count the first ten birds in the flock, get a sense for what proportion of the flock they take up”, then step through the rest in tens. For very large or dense flocks the unit scales — blocks of a hundred, or of a thousand for something like a roost.

The same idea, an exactly counted unit multiplied across the whole, is how a standing crowd of people is estimated from its area and density; the guide to estimating crowd size works one through.

Choosing the block size

Two things go wrong with block size. Too large a block and you cannot count it accurately in the first place, so the error is multiplied through every repetition. Too small and you need so many repetitions that you lose track of how many you have counted. Ten to twenty-five is the usual advice for a first block because density varies and a small block is easier to judge honestly.

Density is the thing that catches people out. A block counted in a tightly packed part of the flock and then applied to a loose part will overestimate badly, and the reverse underestimates. Pick a block at typical density, not the densest part you can see.

Birds that will not hold still

A flock in flight is the hard case, and there is no trick that makes it easy. What helps:

• Count the flock as it passes a fixed point — a tree, a pylon, the edge of your binocular field — rather than trying to hold the whole shape at once.
• Photograph it if you can. A photograph converts an impossible moving count into a straightforward counting job at home, and it is the only method that lets you check yourself.
• Repeat and average. Where the flock stays put, estimate more than once at different moments and take the average. Surveyors do this routinely, and it is the cheapest improvement available.

Not counting the same bird twice

This is where a plain tally counter actively works against you. Tapping once per sighting turns the same six starlings, circling the field four times, into twenty-four birds — a fourfold overcount, and the one number a checklist specifically does not want.

The rule that prevents it is the one at the top: the most seen at one time. Count the group in front of you now; if a larger group appears later, replace the number; if the same group comes round again, change nothing. On a feeder the same applies — a bird that leaves and comes back on the same checklist is one bird. The bird counter keeps both numbers for each species, how many are in view now and the most seen at once, so a returning flock is never added a second time.

Across several checklists for the same place and period, the same caution applies at a larger scale: if you and someone else may have counted the same birds, say so to whoever is compiling, so the totals are not added together.

Being honest about precision

An estimate reported as an exact number is a small lie that survives into the data, and two habits prevent it.

Round to the precision you actually have. Blocks of twenty counted twelve times is “about 250”, not 243. The false decimal place in a bird count is a figure that looks counted when it was estimated.

Do not mix an exact count with an estimate for the same species. If you counted 40 gulls precisely and estimated another 300, the honest entry is an estimate of the whole, not 340 — because 340 reads as though every one of them was counted.

Both of these matter more than getting closer to the true number. A count that is roughly right and honestly labelled is usable; one that is precise-looking and wrong is not.

Birds that can be caught and ringed can be estimated another way, by mark and recapture, which gives a population size with a confidence interval rather than a count.