Field Notes: Counting the Cormorant Roost at First Light

The drone doesn't fly until the birds have been counted. That's a rule at every Tidewright monitoring site, and it's the reason I spent a cold Tuesday in March sitting motionless on a headland at 05:40 with a thermos, a clicker and a pair of very good binoculars. Below me, on a whitewashed spine of rock, the cormorant roost at our Lumeriah Coast program was waking up — and the number of wings leaving that rock before sunrise is one of the cheapest, oldest and most informative data points we collect.

Why count birds by hand at all?

It's a fair question. Tidewright's Canopy Ledger platform maps forests from satellites and drones, runs biodiversity surveys by eDNA, and monitors fish communities with hydroacoustics. Against that arsenal, a human on a cliff with a clicker looks almost quaint. But each method answers a different question, and the dawn roost count answers one the machines handle poorly: how many birds are successfully feeding over this reef, consistently, across the whole season?

Acoustic fish surveys — we explain the method in our piece on acoustic fish surveys — tell us where the forage fish are at a moment in time. eDNA tells us what passed through the water recently. Satellite imagery can't see a bird at all. But a cormorant can't fake a commute. Every bird leaving the roost in the dark to fly to feeding water is a living integration of everything it found yesterday: small schooling fish, prey density, water clarity, competition. When the roost count climbs through a restoration season and holds, something in the water below is feeding those birds reliably. When it slides, we go looking underwater for the reason — often before our other instruments flag a change.

First light at the roost

The method is unglamorous and a little bit beautiful. You arrive in the dark, set up downwind and distant enough not to flush the birds, and wait. The roost at Lumeriah is a low ledge of dolerite that holds the drizzle-silver stains of decades of use. In early autumn the birds there are mostly the region's black cormorants and shags — the same guild of pursuit-divers profiled in our encyclopedia entry for the pelagic cormorant, which hunts the kelp edges of cold coasts across the Pacific.

The first departures come as singles: a bird unzipping from the rock into the dark, gone before you can be certain of the species. Then doubles, then small ragged squadrons. You count with the clicker for every bird heading seaward and subtract the returns. The count runs from civil twilight until the flow thins, usually forty minutes. It's cold, focused work — closer to listening than watching. By 06:45 I had my number, and the drone pilots on the team radio were cleared to launch their canopy-mapping run.

That day the count was 214 birds, up from a baseline near 60 when the program's baseline survey first walked this headland. One season's count means little on its own; the trend across seasons is the signal. But 214 fed the time series, and the time series is what feeds our biodiversity reporting.

What the roost told us that the drone couldn't

Two seasons back the roost count dipped sharply in late summer while the drone maps still showed healthy canopy overhead. That mismatch sent the dive teams down, and what they found was a recruitment gap: canopy at the surface, but very few juvenile kelp and noticeably fewer juvenile fish in the midwater — a quiet season of failed recruitment the canopy map couldn't see, because canopy is last year's news. The bird count wasn't wrong; it was early. That lag — canopy lags the food web — is exactly why we keep the low-tech metrics in the programme.

The roost data also feeds a different audience: it's one of the indicator species measures that underpinned Tidewright's first biodiversity credit issuance, where verifiers needed trend evidence that a human could defend in an audit room. A five-year, dawn-by-dawn roost series is easy to explain, hard to fake and cheap to sustain — three qualities expensive instruments don't always have.

Where people still matter in a monitoring stack

Nobody at Tidewright is sentimental about methods; if a satellite could replace the dawn count cheaply, we'd take the sleep. But some measurements carry properties that matter to restoration science:

• Integration. A predator's commute compresses a whole food web into one number.
• Low cost and long continuity. The same count protocol works with borrowed binoculars, which means five decades of comparable data is plausible.
• Audit legibility. Community rangers, school groups and verifiers can all learn a roost count in a morning and trust the result.
• Complementarity. Birds, acoustics, eDNA and canopy imagery disagree in instructive ways — the mismatches are often where the ecology is happening.

The encyclopedia's seabird group exists partly because of mornings like this one: the species profiles there are written from the same headlands we monitor, and the notes from those headlands, in turn, sharpen how we read the counts.

The count is the easy part

If there's a line I'd leave with anyone starting biodiversity monitoring, it's this: the count is ten percent of the job. The other ninety is showing up on the same rock, in the same light window, under the same protocol, year in and year out. A messy decade beats a brilliant month. The cormorants, thankfully, are punctual. The least we can do is be there when they clock in.

Key takeaways

• Dawn roost counts measure how consistently a reef is feeding its predators — an integration of the food web no single instrument captures.
• Bird counts can flag underwater change before canopy imagery does, because canopy lags the food web.
• Low-tech, high-continuity metrics are audit-friendly: they underpin trend evidence for biodiversity credits and community verification.
• The modern monitoring stack is additive, not substitutive — mismatches between instruments are often the informative signal.
• Protocol discipline over years matters more than any single count's cleverness.

FAQ

Do the monitoring drones disturb the birds?

If flown carelessly, yes — which is why the roost count happens first, before any drone launches. Flight plans are altitude- and distance-restricted near roosts and colonies, and the dawn count doubles as a welfare check before the day's flying begins.

Which species do you actually count at a roost?

Whatever uses the rock. At Lumeriah it's predominantly black cormorants and shags; observers record species where light allows and lump to "cormorant spp." where it doesn't. The protocol values consistency over resolution — a defensible group count every dawn beats an occasional perfect species split.

Can volunteers run roost counts?

Absolutely, and in several Reef Partners programs they do. The protocol is a one-page card: arrive before civil twilight, count departures minus returns, log weather and tide. Volunteer series run in parallel with staff counts for QA, and the overlap has been reassuringly tight.

How does a roost count become a creditable biodiversity metric?

By being part of a pre-registered indicator set with a documented baseline and QA. A roost count alone isn't a credit; it's one time series among several that, together, evidence a trend. The "field notes" of how our first biodiversity credit was assembled covers the whole chain.