About & Methodology

How Bay Area BirdPulse interprets data to track detection trends across the nine counties of the San Francisco Bay Area.

How can I interpret these plots?

Bay Area BirdPulse takes two of the richest bird-monitoring datasets in existence — the Christmas Bird Count (going back to the 1970s) and eBird (from 2007) — and asks a simple question of each species: across the nine counties of the San Francisco Bay Area, is this bird doing better, worse, or about the same as before?

The answer for each species is shown as a detection trend and a confidence in that trend:

Detection trend (the up-or-down axis): how fast the species appears to be changing, in percent per year. Down is decline, up is increase, near the middle is roughly stable.
Confidence (the left-to-right axis): how much we trust that number. A trend can be statistically strong (far right) or genuinely uncertain (far left) — because the data are thin, the two sources disagree, or the species is hard to count.

What you can read from a plot: a bird low and to the right is a decline we have good evidence for. A bird high and to the right is an increase we trust. A bird sitting far to the left — wherever it is vertically — is one where the data simply can't tell us much yet, and that's shown honestly rather than hidden.

What a plot does not tell you: it isn't a verdict on the species, and it isn't a measure of the bird's status across its whole range. It's a pointer — a place to look. The next section explains why that distinction matters.

What these trends are — and what they are not

These trends are pointers, not verdicts. They are a triage tool: a way to look across hundreds of species and thousands of data points and quickly see where attention may be needed. They are a window onto the data, not the final word on any bird.

There is some interpretation involved — we built a method to turn raw counts into trend categories, and reasonable people could draw the lines differently. But the categories exist to help you navigate the data and decide where to look more closely, not to deliver a final judgment.

A few honest limits worth knowing:

This is the Bay Area, not the whole range. Every trend reflects observations in the nine Bay Area counties only. A bird declining here may be stable or thriving across its full range — and the reverse is also true. Local and range-wide trends are different things.
We don't re-verify the underlying data. We run the eBird and Christmas Bird Count records through statistical analysis as they are. Known data-quality quirks in those sources aren't separately corrected here.
The sources sometimes disagree — and that's not a flaw. They cover different time spans, seasons, and observer patterns, so they can legitimately tell different stories about the same species. Where they conflict, we show it rather than paper over it.

Why no algorithm gets every species right

Birds are not data points. Ranges shift, newcomers arrive and establish, detections recover and then turn downward again, and a species that looks rare in one dataset may simply be hard to detect rather than truly scarce. No statistical method, however careful, captures all of that.

So we built uncertainty into the tool on purpose, and we made it visible:

The confidence axis shows how much to trust each trend.
The "Insufficient Data / Too Soon to Tell" category exists for species where the honest answer is "we can't say yet."
Flagged disagreements between sources are shown openly.
A "research warranted" marker flags species where there's a real declining signal, but our two data sources can't yet categorically pin down the trend. These birds still show our best-guess category — they're not increasing — but the flag is an honest signal that they need a closer look, and in some cases finer-grained monitoring data we're working to bring in.

These aren't gaps we failed to fill. They are the tool telling you exactly where the numbers run out and where local, on-the-ground knowledge becomes essential.

A real example — Mountain Quail

The long-term Christmas Bird Count data show a genuine decline since the 1970s. The recent eBird data wobble slightly upward — but at detection levels so low that the movement is mostly noise. Which story is right? We made our best call (a moderate decline, based on the long record), but the birders who actually walk these chaparral slopes know things the numbers can't show. That is precisely where you come in.

Another — Peregrine Falcon

Here's a species still in flux. After its remarkable recovery from the DDT era, the recent data show a decline — but is that a population genuinely shrinking, or a species that overshot during recovery and is now settling back toward a stable baseline? We honestly don't know what the future holds. So rather than pretend certainty, we show the declining signal and flag it for research.

We need your help — for the birds

Birders know these species in the field in ways no dataset ever will. You know which canyon still has Wrentits, where the swifts stopped nesting, which "rare" bird is actually just under-reported. That local knowledge is what turns a statistical triage tool into real conservation insight.

If a trend looks wrong to you, or right for a reason the data can't see — tell us. Your reviews, corrections, and observations are the missing ingredient. We genuinely need your eyes on this to get it right.

Help us put our finger on the pulse of Bay Area birds — for our feathered friends.

How the analysis works (methodology)

Plain-language summary; more detail available on request.

Acknowledgments & sources

Bay Area BirdPulse is a non-profit conservation and education project of the Golden Gate Bird Alliance. It would not be possible without the data and the volunteers behind it.

Christmas Bird Count

CBC data assets are provided by the National Audubon Society and through the generous efforts of Birds Canada and countless volunteers across the Western Hemisphere. We thank the National Audubon Society, Birds Canada, and the thousands of volunteer Christmas Bird Count compilers and participants who generously provide these data. (National Audubon Society. 2025. "Christmas Bird Count Results." Available at christmasbirdcount.org). See also audubon.org and birdscanada.org.

eBird and the Macaulay Library

We used the eBird Basic Dataset for our analysis and thank the Cornell Lab of Ornithology and the thousands of eBird volunteers who contributed the observations, as well as the photographers whose images from the Macaulay Library appear throughout this site. (Sullivan, B.L., C.L. Wood, M.J. Iliff, R.E. Bonney, D. Fink, and S. Kelling. 2009. eBird: a citizen-based bird observation network in the biological sciences. Biological Conservation 142: 2282–2292.)

Other sources

Colony-seabird trends draw on Point Blue Conservation Science's long-term Farallon censuses. Breeding information draws on the Bay Area Breeding Bird Atlases. Conservation-status designations draw on the IUCN Red List, Partners in Flight, and the Golden Gate Bird Alliance's regional Species of Conservation Concern.

We are also working to integrate finer-grained shorebird monitoring data, which will help refine the trends for several of the research-flagged species — part of our ongoing commitment to make this tool more accurate over time.

Trends shown here are derived by Bay Area BirdPulse from the underlying count data and are not official products of, nor endorsed by, the data providers named above. The data are used 'as is' for conservation and educational purposes.